diff options
371 files changed, 15281 insertions, 22818 deletions
diff --git a/.ci/scripts/format/exec.sh b/.ci/scripts/format/exec.sh index 5d6393b38..e9e9d2e17 100644 --- a/.ci/scripts/format/exec.sh +++ b/.ci/scripts/format/exec.sh @@ -1,4 +1,7 @@ #!/bin/bash -ex chmod a+x ./.ci/scripts/format/docker.sh +# the UID for the container yuzu user is 1027 +sudo chown -R 1027 ./ docker run -v $(pwd):/yuzu yuzuemu/build-environments:linux-clang-format /bin/bash -ex /yuzu/.ci/scripts/format/docker.sh +sudo chown -R $UID ./ diff --git a/.ci/scripts/linux/exec.sh b/.ci/scripts/linux/exec.sh index 9fafa9208..a7deddeb3 100644 --- a/.ci/scripts/linux/exec.sh +++ b/.ci/scripts/linux/exec.sh @@ -2,4 +2,7 @@ mkdir -p "ccache" || true chmod a+x ./.ci/scripts/linux/docker.sh +# the UID for the container yuzu user is 1027 +sudo chown -R 1027 ./ docker run -e ENABLE_COMPATIBILITY_REPORTING -e CCACHE_DIR=/yuzu/ccache -v $(pwd):/yuzu yuzuemu/build-environments:linux-fresh /bin/bash /yuzu/.ci/scripts/linux/docker.sh $1 +sudo chown -R $UID ./ diff --git a/.ci/scripts/windows/docker.sh b/.ci/scripts/windows/docker.sh index 5d5650d3b..d53281741 100755 --- a/.ci/scripts/windows/docker.sh +++ b/.ci/scripts/windows/docker.sh @@ -4,22 +4,10 @@ cd /yuzu ccache -s -# Dirty hack to trick unicorn makefile into believing we are in a MINGW system -mv /bin/uname /bin/uname1 && echo -e '#!/bin/sh\necho MINGW64' >> /bin/uname -chmod +x /bin/uname - -# Dirty hack to trick unicorn makefile into believing we have cmd -echo '' >> /bin/cmd -chmod +x /bin/cmd - mkdir build || true && cd build cmake .. -G Ninja -DDISPLAY_VERSION=$1 -DCMAKE_TOOLCHAIN_FILE="$(pwd)/../CMakeModules/MinGWCross.cmake" -DUSE_CCACHE=ON -DYUZU_USE_BUNDLED_UNICORN=ON -DENABLE_COMPATIBILITY_LIST_DOWNLOAD=ON -DCMAKE_BUILD_TYPE=Release ninja -# Clean up the dirty hacks -rm /bin/uname && mv /bin/uname1 /bin/uname -rm /bin/cmd - ccache -s echo "Tests skipped" diff --git a/.ci/scripts/windows/exec.sh b/.ci/scripts/windows/exec.sh index 4155ed5fc..f904544bd 100644 --- a/.ci/scripts/windows/exec.sh +++ b/.ci/scripts/windows/exec.sh @@ -2,4 +2,7 @@ mkdir -p "ccache" || true chmod a+x ./.ci/scripts/windows/docker.sh +# the UID for the container yuzu user is 1027 +sudo chown -R 1027 ./ docker run -e CCACHE_DIR=/yuzu/ccache -v $(pwd):/yuzu yuzuemu/build-environments:linux-mingw /bin/bash -ex /yuzu/.ci/scripts/windows/docker.sh $1 +sudo chown -R $UID ./ diff --git a/.ci/templates/build-msvc.yml b/.ci/templates/build-msvc.yml index adcecf0ec..d85a949aa 100644 --- a/.ci/templates/build-msvc.yml +++ b/.ci/templates/build-msvc.yml @@ -4,7 +4,9 @@ parameters: version: '' steps: -- script: mkdir build && cd build && cmake -G "Visual Studio 16 2019" -A x64 --config Release -DYUZU_USE_BUNDLED_QT=1 -DYUZU_USE_BUNDLED_SDL2=1 -DYUZU_USE_BUNDLED_UNICORN=1 -DYUZU_USE_QT_WEB_ENGINE=ON -DENABLE_COMPATIBILITY_LIST_DOWNLOAD=ON -DYUZU_ENABLE_COMPATIBILITY_REPORTING=${COMPAT} -DUSE_DISCORD_PRESENCE=ON -DDISPLAY_VERSION=${{ parameters['version'] }} .. && cd .. +- script: python -m pip install --upgrade pip conan + displayName: 'Install conan' +- script: mkdir build && cd build && cmake -G "Visual Studio 16 2019" -A x64 --config Release -DYUZU_USE_BUNDLED_QT=1 -DYUZU_USE_BUNDLED_UNICORN=1 -DYUZU_USE_QT_WEB_ENGINE=ON -DENABLE_COMPATIBILITY_LIST_DOWNLOAD=ON -DYUZU_ENABLE_COMPATIBILITY_REPORTING=${COMPAT} -DUSE_DISCORD_PRESENCE=ON -DDISPLAY_VERSION=${{ parameters['version'] }} .. && cd .. displayName: 'Configure CMake' - task: MSBuild@1 displayName: 'Build' diff --git a/.ci/yuzu-patreon-step2.yml b/.ci/yuzu-patreon-step2.yml index 26e287257..41eccd973 100644 --- a/.ci/yuzu-patreon-step2.yml +++ b/.ci/yuzu-patreon-step2.yml @@ -5,18 +5,7 @@ variables: DisplayVersion: $[counter(variables['DisplayPrefix'], 1)] stages: -- stage: format - displayName: 'format' - jobs: - - job: format - displayName: 'clang' - continueOnError: true - pool: - vmImage: ubuntu-latest - steps: - - template: ./templates/format-check.yml - stage: build - dependsOn: format displayName: 'build' jobs: - job: build diff --git a/.gitmodules b/.gitmodules index 63bf2cda0..9ba8fe207 100644 --- a/.gitmodules +++ b/.gitmodules @@ -1,33 +1,15 @@ [submodule "inih"] path = externals/inih/inih url = https://github.com/svn2github/inih -[submodule "boost"] - path = externals/boost - url = https://github.com/yuzu-emu/ext-boost.git -[submodule "catch"] - path = externals/catch - url = https://github.com/philsquared/Catch.git [submodule "cubeb"] path = externals/cubeb url = https://github.com/kinetiknz/cubeb.git [submodule "dynarmic"] path = externals/dynarmic url = https://github.com/MerryMage/dynarmic.git -[submodule "fmt"] - path = externals/fmt - url = https://github.com/fmtlib/fmt.git -[submodule "lz4"] - path = externals/lz4 - url = https://github.com/lz4/lz4.git [submodule "unicorn"] path = externals/unicorn url = https://github.com/yuzu-emu/unicorn -[submodule "mbedtls"] - path = externals/mbedtls - url = https://github.com/DarkLordZach/mbedtls -[submodule "opus"] - path = externals/opus/opus - url = https://github.com/xiph/opus.git [submodule "soundtouch"] path = externals/soundtouch url = https://github.com/citra-emu/ext-soundtouch.git @@ -40,15 +22,15 @@ [submodule "Vulkan-Headers"] path = externals/Vulkan-Headers url = https://github.com/KhronosGroup/Vulkan-Headers.git -[submodule "externals/zstd"] - path = externals/zstd - url = https://github.com/facebook/zstd [submodule "sirit"] path = externals/sirit url = https://github.com/ReinUsesLisp/sirit +[submodule "mbedtls"] + path = externals/mbedtls + url = https://github.com/DarkLordZach/mbedtls [submodule "libzip"] path = externals/libzip/libzip url = https://github.com/nih-at/libzip.git -[submodule "zlib"] - path = externals/zlib/zlib - url = https://github.com/madler/zlib.git +[submodule "xbyak"] + path = externals/xbyak + url = https://github.com/herumi/xbyak.git diff --git a/CMakeLists.txt b/CMakeLists.txt index c906c5a50..b71071271 100644 --- a/CMakeLists.txt +++ b/CMakeLists.txt @@ -1,7 +1,8 @@ -cmake_minimum_required(VERSION 3.7) +cmake_minimum_required(VERSION 3.15) list(APPEND CMAKE_MODULE_PATH "${CMAKE_CURRENT_SOURCE_DIR}/CMakeModules") list(APPEND CMAKE_MODULE_PATH "${CMAKE_CURRENT_SOURCE_DIR}/externals/cmake-modules") +list(APPEND CMAKE_MODULE_PATH "${CMAKE_CURRENT_SOURCE_DIR}/externals/find-modules") include(DownloadExternals) include(CMakeDependentOption) @@ -10,7 +11,6 @@ project(yuzu) # Set bundled sdl2/qt as dependent options. # OFF by default, but if ENABLE_SDL2 and MSVC are true then ON option(ENABLE_SDL2 "Enable the SDL2 frontend" ON) -CMAKE_DEPENDENT_OPTION(YUZU_USE_BUNDLED_SDL2 "Download bundled SDL2 binaries" ON "ENABLE_SDL2;MSVC" OFF) option(ENABLE_QT "Enable the Qt frontend" ON) CMAKE_DEPENDENT_OPTION(YUZU_USE_BUNDLED_QT "Download bundled Qt binaries" ON "ENABLE_QT;MSVC" OFF) @@ -29,6 +29,13 @@ option(ENABLE_VULKAN "Enables Vulkan backend" ON) option(USE_DISCORD_PRESENCE "Enables Discord Rich Presence" OFF) +# Default to a Release build +get_property(IS_MULTI_CONFIG GLOBAL PROPERTY GENERATOR_IS_MULTI_CONFIG) +if (NOT IS_MULTI_CONFIG AND NOT CMAKE_BUILD_TYPE) + set(CMAKE_BUILD_TYPE "Release" CACHE STRING "Choose the type of build, options are: Debug Release RelWithDebInfo MinSizeRel." FORCE) + message(STATUS "Defaulting to a Release build") +endif() + if(EXISTS ${PROJECT_SOURCE_DIR}/hooks/pre-commit AND NOT EXISTS ${PROJECT_SOURCE_DIR}/.git/hooks/pre-commit) message(STATUS "Copying pre-commit hook") file(COPY hooks/pre-commit @@ -53,7 +60,6 @@ endfunction() if(EXISTS ${PROJECT_SOURCE_DIR}/.gitmodules) check_submodules_present() endif() - configure_file(${PROJECT_SOURCE_DIR}/dist/compatibility_list/compatibility_list.qrc ${PROJECT_BINARY_DIR}/dist/compatibility_list/compatibility_list.qrc COPYONLY) @@ -115,63 +121,217 @@ message(STATUS "Target architecture: ${ARCHITECTURE}") set(CMAKE_CXX_STANDARD 17) set(CMAKE_CXX_STANDARD_REQUIRED ON) -# System imported libraries -# ====================== - -find_package(Boost 1.66.0 QUIET) -if (NOT Boost_FOUND) - message(STATUS "Boost 1.66.0 or newer not found, falling back to externals") +# Output binaries to bin/ +set(CMAKE_RUNTIME_OUTPUT_DIRECTORY ${PROJECT_BINARY_DIR}/bin) - set(BOOST_ROOT "${PROJECT_SOURCE_DIR}/externals/boost") - set(Boost_NO_SYSTEM_PATHS OFF) - find_package(Boost QUIET REQUIRED) +# System imported libraries +# If not found, download any missing through Conan +# ======================================================================= +set(CONAN_CMAKE_SILENT_OUTPUT TRUE) +set(CMAKE_FIND_PACKAGE_PREFER_CONFIG TRUE) +if (YUZU_CONAN_INSTALLED) + if (IS_MULTI_CONFIG) + include(${CMAKE_BINARY_DIR}/conanbuildinfo_multi.cmake) + else() + include(${CMAKE_BINARY_DIR}/conanbuildinfo.cmake) + endif() + list(APPEND CMAKE_MODULE_PATH "${CMAKE_BINARY_DIR}") + list(APPEND CMAKE_PREFIX_PATH "${CMAKE_BINARY_DIR}") + conan_basic_setup() + message(STATUS "Adding conan installed libraries to the search path") endif() -# Output binaries to bin/ -set(CMAKE_RUNTIME_OUTPUT_DIRECTORY ${PROJECT_BINARY_DIR}/bin) +macro(yuzu_find_packages) + set(options FORCE_REQUIRED) + cmake_parse_arguments(FN "${options}" "" "" ${ARGN}) + + # Cmake has a *serious* lack of 2D array or associative array... + # Capitalization matters here. We need the naming to match the generated paths from Conan + set(REQUIRED_LIBS + # Cmake Pkg Prefix Version Conan Pkg + "Boost 1.71 boost/1.72.0" + "Catch2 2.11 catch2/2.11.0" + "fmt 6.2 fmt/6.2.0" + # can't use until https://github.com/bincrafters/community/issues/1173 + #"libzip 1.5 libzip/1.5.2@bincrafters/stable" + "lz4 1.8 lz4/1.9.2" + "nlohmann_json 3.7 nlohmann_json/3.7.3" + # we need to be careful as the version check might be broken https://github.com/xiph/opus/issues/110 + "opus 1.3 opus/1.3.1" + "ZLIB 1.2 zlib/1.2.11" + "zstd 1.4 zstd/1.4.4" + ) + + foreach(PACKAGE ${REQUIRED_LIBS}) + string(REGEX REPLACE "[ \t\r\n]+" ";" PACKAGE_SPLIT ${PACKAGE}) + list(GET PACKAGE_SPLIT 0 PACKAGE_PREFIX) + list(GET PACKAGE_SPLIT 1 PACKAGE_VERSION) + list(GET PACKAGE_SPLIT 2 PACKAGE_CONAN) + # This function is called twice, once to check if the packages exist on the system already + # and a second time to check if conan installed them properly. The second check passes in FORCE_REQUIRED + if (NOT ${PACKAGE_PREFIX}_FOUND) + if (FN_FORCE_REQUIRED) + find_package(${PACKAGE_PREFIX} ${PACKAGE_VERSION} REQUIRED) + else() + find_package(${PACKAGE_PREFIX} ${PACKAGE_VERSION}) + endif() + endif() + if (NOT ${PACKAGE_PREFIX}_FOUND) + list(APPEND CONAN_REQUIRED_LIBS ${PACKAGE_CONAN}) + else() + # Set a legacy findPackage.cmake style PACKAGE_LIBRARIES variable for subprojects that rely on this + set(${PACKAGE_PREFIX}_LIBRARIES "${PACKAGE_PREFIX}::${PACKAGE_PREFIX}") + endif() + endforeach() + unset(FN_FORCE_REQUIRED) +endmacro() -# Prefer the -pthread flag on Linux. -set(THREADS_PREFER_PTHREAD_FLAG ON) -find_package(Threads REQUIRED) +# Attempt to locate any packages that are required and report the missing ones in CONAN_REQUIRED_LIBS +yuzu_find_packages() -if (ENABLE_SDL2) - if (YUZU_USE_BUNDLED_SDL2) - # Detect toolchain and platform +# Qt5 requires that we find components, so it doesn't fit our pretty little find package function +if(ENABLE_QT) + # We want to load the generated conan qt config so that we get the QT_ROOT var so that we can use the official + # Qt5Config inside the root folder instead of the conan generated one. + if(EXISTS ${CMAKE_BINARY_DIR}/qtConfig.cmake) + include(${CMAKE_BINARY_DIR}/qtConfig.cmake) + list(APPEND CMAKE_MODULE_PATH "${CONAN_QT_ROOT_RELEASE}") + list(APPEND CMAKE_PREFIX_PATH "${CONAN_QT_ROOT_RELEASE}") + endif() + # Workaround for an issue where conan tries to build Qt from scratch instead of download prebuilt binaries + set(QT_PREFIX_HINT) + if(YUZU_USE_BUNDLED_QT) if ((MSVC_VERSION GREATER_EQUAL 1910 AND MSVC_VERSION LESS 1930) AND ARCHITECTURE_x86_64) - set(SDL2_VER "SDL2-2.0.8") + set(QT_VER qt-5.12.0-msvc2017_64) else() - message(FATAL_ERROR "No bundled SDL2 binaries for your toolchain. Disable YUZU_USE_BUNDLED_SDL2 and provide your own.") + message(FATAL_ERROR "No bundled Qt binaries for your toolchain. Disable YUZU_USE_BUNDLED_QT and provide your own.") endif() - if (DEFINED SDL2_VER) - download_bundled_external("sdl2/" ${SDL2_VER} SDL2_PREFIX) + if (DEFINED QT_VER) + download_bundled_external("qt/" ${QT_VER} QT_PREFIX) endif() - set(SDL2_FOUND YES) - set(SDL2_INCLUDE_DIR "${SDL2_PREFIX}/include" CACHE PATH "Path to SDL2 headers") - set(SDL2_LIBRARY "${SDL2_PREFIX}/lib/x64/SDL2.lib" CACHE PATH "Path to SDL2 library") - set(SDL2_DLL_DIR "${SDL2_PREFIX}/lib/x64/" CACHE PATH "Path to SDL2.dll") + set(QT_PREFIX_HINT HINTS "${QT_PREFIX}") + endif() + find_package(Qt5 5.9 COMPONENTS Widgets OpenGL ${QT_PREFIX_HINT}) + if (NOT Qt5_FOUND) + list(APPEND CONAN_REQUIRED_LIBS "qt/5.14.1@bincrafters/stable") + endif() +endif() +# find SDL2 exports a bunch of variables that are needed, so its easier to do this outside of the yuzu_find_package +if(ENABLE_SDL2) + if(EXISTS ${CMAKE_BINARY_DIR}/sdl2Config.cmake) + include(${CMAKE_BINARY_DIR}/sdl2Config.cmake) + list(APPEND CMAKE_MODULE_PATH "${CONAN_SDL2_ROOT_RELEASE}") + list(APPEND CMAKE_PREFIX_PATH "${CONAN_SDL2_ROOT_RELEASE}") + endif() + find_package(SDL2) + if (NOT SDL2_FOUND) + # otherwise add this to the list of libraries to install + list(APPEND CONAN_REQUIRED_LIBS "sdl2/2.0.12@bincrafters/stable") + endif() +endif() - add_library(SDL2 INTERFACE) - target_link_libraries(SDL2 INTERFACE "${SDL2_LIBRARY}") - target_include_directories(SDL2 INTERFACE "${SDL2_INCLUDE_DIR}") +# Install any missing dependencies with conan install +if (CONAN_REQUIRED_LIBS) + message(STATUS "Packages ${CONAN_REQUIRED_LIBS} not found!") + # Use Conan to fetch the libraries that aren't found + # Download conan.cmake automatically, you can also just copy the conan.cmake file + if(NOT EXISTS "${CMAKE_BINARY_DIR}/conan.cmake") + message(STATUS "Downloading conan.cmake from https://github.com/conan-io/cmake-conan") + file(DOWNLOAD "https://github.com/conan-io/cmake-conan/raw/v0.15/conan.cmake" + "${CMAKE_BINARY_DIR}/conan.cmake") + endif() + include(${CMAKE_BINARY_DIR}/conan.cmake) + + set(CONAN_LIB_OPTIONS + libzip:with_openssl=False + libzip:enable_windows_crypto=False + ) + conan_check(VERSION 1.24.0 REQUIRED) + # Add the bincrafters remote + conan_add_remote(NAME bincrafters + URL https://api.bintray.com/conan/bincrafters/public-conan) + + # Manually add iconv to fix a dep conflict between qt and sdl2 + # We don't need to add it through find_package or anything since the other two can find it just fine + if ("${CONAN_REQUIRED_LIBS}" MATCHES "qt" AND "${CONAN_REQUIRED_LIBS}" MATCHES "sdl") + list(APPEND CONAN_REQUIRED_LIBS "libiconv/1.16") + endif() + if (IS_MULTI_CONFIG) + conan_cmake_run(REQUIRES ${CONAN_REQUIRED_LIBS} + OPTIONS ${CONAN_LIB_OPTIONS} + BUILD missing + CONFIGURATION_TYPES "Release;Debug" + GENERATORS cmake_multi cmake_find_package_multi) + include(${CMAKE_BINARY_DIR}/conanbuildinfo_multi.cmake) else() + conan_cmake_run(REQUIRES ${CONAN_REQUIRED_LIBS} + OPTIONS ${CONAN_LIB_OPTIONS} + BUILD missing + GENERATORS cmake cmake_find_package_multi) + include(${CMAKE_BINARY_DIR}/conanbuildinfo.cmake) + endif() + list(APPEND CMAKE_MODULE_PATH "${CMAKE_BINARY_DIR}") + list(APPEND CMAKE_PREFIX_PATH "${CMAKE_BINARY_DIR}") + conan_basic_setup() + + set(YUZU_CONAN_INSTALLED TRUE CACHE BOOL "If true, the following builds will add conan to the lib search path" FORCE) + + # Now that we've installed what we are missing, try to locate them again, + # this time with required, so we bail if its not found. + yuzu_find_packages(FORCE_REQUIRED) + + # Due to issues with variable scopes in functions, we need to also find_package(qt5) outside of the function + if(ENABLE_QT) + list(APPEND CMAKE_MODULE_PATH "${CONAN_QT_ROOT_RELEASE}") + list(APPEND CMAKE_PREFIX_PATH "${CONAN_QT_ROOT_RELEASE}") + find_package(Qt5 5.9 REQUIRED COMPONENTS Widgets OpenGL) + if (YUZU_USE_QT_WEB_ENGINE) + find_package(Qt5 REQUIRED COMPONENTS WebEngineCore WebEngineWidgets) + endif() + endif() + if(ENABLE_SDL2) + list(APPEND CMAKE_MODULE_PATH "${CONAN_SDL2_ROOT_RELEASE}") + list(APPEND CMAKE_PREFIX_PATH "${CONAN_SDL2_ROOT_RELEASE}") find_package(SDL2 REQUIRED) + endif() - # Some installations don't set SDL2_LIBRARIES - if("${SDL2_LIBRARIES}" STREQUAL "") - message(WARNING "SDL2_LIBRARIES wasn't set, manually setting to SDL2::SDL2") - set(SDL2_LIBRARIES "SDL2::SDL2") - endif() +endif() - include_directories(SYSTEM ${SDL2_INCLUDE_DIRS}) - add_library(SDL2 INTERFACE) - target_link_libraries(SDL2 INTERFACE "${SDL2_LIBRARIES}") +# Reexport some targets that are named differently when using the upstream CmakeConfig vs the generated Conan config +# In order to ALIAS targets to a new name, they first need to be IMPORTED_GLOBAL +# Dynarmic checks for target `boost` and so we want to make sure it can find it through our system instead of using their external +if (TARGET Boost::Boost) + set_target_properties(Boost::Boost PROPERTIES IMPORTED_GLOBAL TRUE) + add_library(Boost::boost ALIAS Boost::Boost) + add_library(boost ALIAS Boost::Boost) +elseif (TARGET Boost::boost) + set_target_properties(Boost::boost PROPERTIES IMPORTED_GLOBAL TRUE) + add_library(boost ALIAS Boost::boost) +endif() + +if (TARGET sdl2::sdl2) + # imported from the conan generated sdl2Config.cmake + set_target_properties(sdl2::sdl2 PROPERTIES IMPORTED_GLOBAL TRUE) + add_library(SDL2 ALIAS sdl2::sdl2) +elseif(SDL2_FOUND) + # found through the system package manager + # Some installations don't set SDL2_LIBRARIES + if("${SDL2_LIBRARIES}" STREQUAL "") + message(WARNING "SDL2_LIBRARIES wasn't set, manually setting to SDL2::SDL2") + set(SDL2_LIBRARIES "SDL2::SDL2") endif() -else() - set(SDL2_FOUND NO) + + include_directories(SYSTEM ${SDL2_INCLUDE_DIRS}) + add_library(SDL2 INTERFACE) + target_link_libraries(SDL2 INTERFACE "${SDL2_LIBRARIES}") endif() +# Prefer the -pthread flag on Linux. +set(THREADS_PREFER_PTHREAD_FLAG ON) +find_package(Threads REQUIRED) + # If unicorn isn't found, msvc -> download bundled unicorn; everyone else -> build external if (YUZU_USE_BUNDLED_UNICORN) if (MSVC) @@ -212,8 +372,12 @@ if (YUZU_USE_BUNDLED_UNICORN) find_package(PythonInterp 2.7 REQUIRED) if (MINGW) + # Intentionally call the unicorn makefile directly instead of using make.sh so that we can override the + # UNAME_S makefile variable to MINGW. This way we don't have to hack at the uname binary to build + # Additionally, overriding DO_WINDOWS_EXPORT prevents unicorn from patching the static unicorn.a by using msvc and cmd, + # which are both things we don't have in a mingw cross compiling environment. add_custom_command(OUTPUT ${LIBUNICORN_LIBRARY} - COMMAND ${CMAKE_COMMAND} -E env UNICORN_ARCHS="aarch64" PYTHON="${PYTHON_EXECUTABLE}" /bin/sh make.sh cross-win64 + COMMAND ${CMAKE_COMMAND} -E env UNICORN_ARCHS="aarch64" PYTHON="${PYTHON_EXECUTABLE}" CC=x86_64-w64-mingw32-gcc AR=x86_64-w64-mingw32-gcc-ar RANLIB=x86_64-w64-mingw32-gcc-ranlib make UNAME_S=MINGW DO_WINDOWS_EXPORT=0 WORKING_DIRECTORY ${UNICORN_PREFIX} ) else() @@ -243,32 +407,6 @@ else() message(FATAL_ERROR "Could not find or build unicorn which is required.") endif() -if (ENABLE_QT) - if (YUZU_USE_BUNDLED_QT) - if ((MSVC_VERSION GREATER_EQUAL 1910 AND MSVC_VERSION LESS 1930) AND ARCHITECTURE_x86_64) - set(QT_VER qt-5.12.0-msvc2017_64) - else() - message(FATAL_ERROR "No bundled Qt binaries for your toolchain. Disable YUZU_USE_BUNDLED_QT and provide your own.") - endif() - - if (DEFINED QT_VER) - download_bundled_external("qt/" ${QT_VER} QT_PREFIX) - endif() - - set(QT_PREFIX_HINT HINTS "${QT_PREFIX}") - else() - # Passing an empty HINTS seems to cause default system paths to get ignored in CMake 2.8 so - # make sure to not pass anything if we don't have one. - set(QT_PREFIX_HINT) - endif() - - find_package(Qt5 REQUIRED COMPONENTS Widgets OpenGL ${QT_PREFIX_HINT}) - - if (YUZU_USE_QT_WEB_ENGINE) - find_package(Qt5 REQUIRED COMPONENTS WebEngineCore WebEngineWidgets ${QT_PREFIX_HINT}) - endif () -endif() - # Platform-specific library requirements # ====================================== @@ -299,7 +437,7 @@ find_program(CLANG_FORMAT PATHS ${PROJECT_BINARY_DIR}/externals) # if find_program doesn't find it, try to download from externals if (NOT CLANG_FORMAT) - if (WIN32) + if (WIN32 AND NOT CMAKE_CROSSCOMPILING) message(STATUS "Clang format not found! Downloading...") set(CLANG_FORMAT "${PROJECT_BINARY_DIR}/externals/clang-format${CLANG_FORMAT_POSTFIX}.exe") file(DOWNLOAD diff --git a/CMakeModules/GenerateSCMRev.cmake b/CMakeModules/GenerateSCMRev.cmake index 83e4e9df2..311ba1c2e 100644 --- a/CMakeModules/GenerateSCMRev.cmake +++ b/CMakeModules/GenerateSCMRev.cmake @@ -51,6 +51,8 @@ endif() # The variable SRC_DIR must be passed into the script (since it uses the current build directory for all values of CMAKE_*_DIR) set(VIDEO_CORE "${SRC_DIR}/src/video_core") set(HASH_FILES + "${VIDEO_CORE}/renderer_opengl/gl_arb_decompiler.cpp" + "${VIDEO_CORE}/renderer_opengl/gl_arb_decompiler.h" "${VIDEO_CORE}/renderer_opengl/gl_shader_cache.cpp" "${VIDEO_CORE}/renderer_opengl/gl_shader_cache.h" "${VIDEO_CORE}/renderer_opengl/gl_shader_decompiler.cpp" diff --git a/dist/qt_themes/qdarkstyle/style.qss b/dist/qt_themes/qdarkstyle/style.qss index 7d088a719..2d5c9761f 100644 --- a/dist/qt_themes/qdarkstyle/style.qss +++ b/dist/qt_themes/qdarkstyle/style.qss @@ -673,10 +673,6 @@ QTabWidget::pane { border-bottom-left-radius: 2px; } -QTabWidget::tab-bar { - overflow: visible; -} - QTabBar { qproperty-drawBase: 0; border-radius: 3px; diff --git a/dist/yuzu.manifest b/dist/yuzu.manifest index fd30b656f..038edff23 100644 --- a/dist/yuzu.manifest +++ b/dist/yuzu.manifest @@ -1,24 +1,58 @@ <?xml version="1.0" encoding="UTF-8" standalone="yes"?> -<assembly xmlns="urn:schemas-microsoft-com:asm.v1" manifestVersion="1.0"> - <trustInfo xmlns="urn:schemas-microsoft-com:asm.v3"> - <security> - <requestedPrivileges> - <requestedExecutionLevel level="asInvoker" uiAccess="false"/> - </requestedPrivileges> - </security> - </trustInfo> - <application xmlns="urn:schemas-microsoft-com:asm.v3"> - <windowsSettings> - <dpiAware xmlns="http://schemas.microsoft.com/SMI/2005/WindowsSettings">True/PM</dpiAware> - <longPathAware xmlns="http://schemas.microsoft.com/SMI/2016/WindowsSettings">true</longPathAware> - </windowsSettings> - </application> - <compatibility xmlns="urn:schemas-microsoft-com:compatibility.v1"> - <application> - <supportedOS Id="{35138b9a-5d96-4fbd-8e2d-a2440225f93a}"/> - <supportedOS Id="{4a2f28e3-53b9-4441-ba9c-d69d4a4a6e38}"/> - <supportedOS Id="{1f676c76-80e1-4239-95bb-83d0f6d0da78}"/> - <supportedOS Id="{8e0f7a12-bfb3-4fe8-b9a5-48fd50a15a9a}"/> - </application> - </compatibility> -</assembly>
\ No newline at end of file +<assembly manifestVersion="1.0" + xmlns="urn:schemas-microsoft-com:asm.v1" + xmlns:asmv3="urn:schemas-microsoft-com:asm.v3"> + <asmv3:application> + <asmv3:windowsSettings> + <!-- Windows 7/8/8.1/10 --> + <dpiAware + xmlns="http://schemas.microsoft.com/SMI/2005/WindowsSettings"> + true/pm + </dpiAware> + <!-- Windows 10, version 1607 or later --> + <dpiAwareness + xmlns="http://schemas.microsoft.com/SMI/2016/WindowsSettings"> + PerMonitorV2 + </dpiAwareness> + <!-- Windows 10, version 1703 or later --> + <gdiScaling + xmlns="http://schemas.microsoft.com/SMI/2017/WindowsSettings"> + true + </gdiScaling> + <ws2:longPathAware + xmlns:ws3="http://schemas.microsoft.com/SMI/2016/WindowsSettings"> + true + </ws2:longPathAware> + </asmv3:windowsSettings> + </asmv3:application> + <compatibility + xmlns="urn:schemas-microsoft-com:compatibility.v1"> + <application> + <!-- Windows 10 --> + <supportedOS Id="{8e0f7a12-bfb3-4fe8-b9a5-48fd50a15a9a}"/> + <!-- Windows 8.1 --> + <supportedOS Id="{1f676c76-80e1-4239-95bb-83d0f6d0da78}"/> + <!-- Windows 8 --> + <supportedOS Id="{4a2f28e3-53b9-4441-ba9c-d69d4a4a6e38}"/> + <!-- Windows 7 --> + <supportedOS Id="{35138b9a-5d96-4fbd-8e2d-a2440225f93a}"/> + </application> + </compatibility> + <trustInfo + xmlns="urn:schemas-microsoft-com:asm.v3"> + <security> + <requestedPrivileges> + <!-- + UAC settings: + - app should run at same integrity level as calling process + - app does not need to manipulate windows belonging to + higher-integrity-level processes + --> + <requestedExecutionLevel + level="asInvoker" + uiAccess="false" + /> + </requestedPrivileges> + </security> + </trustInfo> +</assembly> diff --git a/externals/CMakeLists.txt b/externals/CMakeLists.txt index 653c64f47..b80b27605 100644 --- a/externals/CMakeLists.txt +++ b/externals/CMakeLists.txt @@ -1,30 +1,20 @@ # Definitions for all external bundled libraries -set(CMAKE_MODULE_PATH ${CMAKE_MODULE_PATH} ${PROJECT_SOURCE_DIR}/CMakeModules) +list(APPEND CMAKE_MODULE_PATH "${PROJECT_SOURCE_DIR}/CMakeModules") +list(APPEND CMAKE_MODULE_PATH "${PROJECT_SOURCE_DIR}/externals/find-modules") include(DownloadExternals) -# pkgconfig -- it is used to find shared libraries without cmake modules on linux systems -find_package(PkgConfig) -if (NOT PkgConfig_FOUND) - function(pkg_check_modules) - # STUB - endfunction() +# xbyak +if (ARCHITECTURE_x86 OR ARCHITECTURE_x86_64) + add_library(xbyak INTERFACE) + target_include_directories(xbyak SYSTEM INTERFACE ./xbyak/xbyak) + target_compile_definitions(xbyak INTERFACE XBYAK_NO_OP_NAMES) endif() # Catch add_library(catch-single-include INTERFACE) target_include_directories(catch-single-include INTERFACE catch/single_include) -# libfmt -pkg_check_modules(FMT IMPORTED_TARGET GLOBAL fmt>=6.2.0) -if (FMT_FOUND) - add_library(fmt::fmt ALIAS PkgConfig::FMT) -else() - message(STATUS "fmt 6.2.0 or newer not found, falling back to externals") - add_subdirectory(fmt) - add_library(fmt::fmt ALIAS fmt) -endif() - # Dynarmic if (ARCHITECTURE_x86_64) set(DYNARMIC_TESTS OFF) @@ -43,17 +33,6 @@ add_subdirectory(glad) # inih add_subdirectory(inih) -# lz4 -pkg_check_modules(LIBLZ4 IMPORTED_TARGET GLOBAL liblz4>=1.8.0) -if (LIBLZ4_FOUND) - add_library(lz4_static ALIAS PkgConfig::LIBLZ4) -else() - message(STATUS "liblz4 1.8.0 or newer not found, falling back to externals") - set(LZ4_BUNDLED_MODE ON) - add_subdirectory(lz4/contrib/cmake_unofficial EXCLUDE_FROM_ALL) - target_include_directories(lz4_static INTERFACE ./lz4/lib) -endif() - # mbedtls add_subdirectory(mbedtls EXCLUDE_FROM_ALL) target_include_directories(mbedtls PUBLIC ./mbedtls/include) @@ -66,29 +45,9 @@ target_include_directories(microprofile INTERFACE ./microprofile) add_library(unicorn-headers INTERFACE) target_include_directories(unicorn-headers INTERFACE ./unicorn/include) -# Zstandard -pkg_check_modules(LIBZSTD IMPORTED_TARGET GLOBAL libzstd>=1.3.8) -if (LIBZSTD_FOUND) - add_library(libzstd_static ALIAS PkgConfig::LIBZSTD) -else() - message(STATUS "libzstd 1.3.8 or newer not found, falling back to externals") - add_subdirectory(zstd/build/cmake EXCLUDE_FROM_ALL) - target_include_directories(libzstd_static INTERFACE ./zstd/lib) -endif() - # SoundTouch add_subdirectory(soundtouch) -# Opus -pkg_check_modules(OPUS IMPORTED_TARGET GLOBAL opus>=1.3.1) -if (OPUS_FOUND) - add_library(opus ALIAS PkgConfig::OPUS) -else() - message(STATUS "opus 1.3.1 or newer not found, falling back to externals") - add_subdirectory(opus) - target_include_directories(opus INTERFACE ./opus/include) -endif() - # Cubeb if(ENABLE_CUBEB) set(BUILD_TESTS OFF CACHE BOOL "") @@ -106,36 +65,22 @@ if (ENABLE_VULKAN) add_subdirectory(sirit) endif() -# zlib -find_package(ZLIB 1.2.11) -if (NOT ZLIB_FOUND) - message(STATUS "zlib 1.2.11 or newer not found, falling back to externals") - add_subdirectory(zlib EXCLUDE_FROM_ALL) - set(ZLIB_LIBRARIES z) -endif() - # libzip -pkg_check_modules(LIBZIP IMPORTED_TARGET GLOBAL libzip>=1.5.3) -if (LIBZIP_FOUND) - add_library(zip ALIAS PkgConfig::LIBZIP) -else() - message(STATUS "libzip 1.5.3 or newer not found, falling back to externals") +find_package(Libzip 1.5) +if (NOT LIBZIP_FOUND) + message(STATUS "libzip 1.5 or newer not found, falling back to externals") add_subdirectory(libzip EXCLUDE_FROM_ALL) endif() if (ENABLE_WEB_SERVICE) # LibreSSL - find_package(OpenSSL COMPONENTS Crypto SSL) - if (NOT OpenSSL_FOUND) - message(STATUS "OpenSSL not found, falling back to externals") - set(LIBRESSL_SKIP_INSTALL ON CACHE BOOL "") - add_subdirectory(libressl EXCLUDE_FROM_ALL) - target_include_directories(ssl INTERFACE ./libressl/include) - target_compile_definitions(ssl PRIVATE -DHAVE_INET_NTOP) - get_directory_property(OPENSSL_LIBRARIES - DIRECTORY libressl - DEFINITION OPENSSL_LIBS) - endif() + set(LIBRESSL_SKIP_INSTALL ON CACHE BOOL "") + add_subdirectory(libressl EXCLUDE_FROM_ALL) + target_include_directories(ssl INTERFACE ./libressl/include) + target_compile_definitions(ssl PRIVATE -DHAVE_INET_NTOP) + get_directory_property(OPENSSL_LIBRARIES + DIRECTORY libressl + DEFINITION OPENSSL_LIBS) # lurlparser add_subdirectory(lurlparser EXCLUDE_FROM_ALL) @@ -145,8 +90,4 @@ if (ENABLE_WEB_SERVICE) target_include_directories(httplib INTERFACE ./httplib) target_compile_definitions(httplib INTERFACE -DCPPHTTPLIB_OPENSSL_SUPPORT) target_link_libraries(httplib INTERFACE ${OPENSSL_LIBRARIES}) - - # JSON - add_library(json-headers INTERFACE) - target_include_directories(json-headers INTERFACE ./json) endif() diff --git a/externals/Vulkan-Headers b/externals/Vulkan-Headers -Subproject 0e78ffd1dcfc3e9f14a966b9660dbc59bd967c5 +Subproject 9250d5ae8f50202005233dc0512a1d460c8b483 diff --git a/externals/boost b/externals/boost deleted file mode 160000 -Subproject 5e8300b76a627f3a1ba215304e04ead33b5bc23 diff --git a/externals/catch b/externals/catch deleted file mode 160000 -Subproject 15cf3caaceb21172ea42a24e595a2eb58c3ec96 diff --git a/externals/dynarmic b/externals/dynarmic -Subproject e7166e8ba74d7b9c85e87afc0aaf667e7e84cfe +Subproject 4f967387c07365b7ea35d2fa3e19b7df8872a09 diff --git a/externals/find-modules/FindCatch2.cmake b/externals/find-modules/FindCatch2.cmake new file mode 100644 index 000000000..ce1d40bae --- /dev/null +++ b/externals/find-modules/FindCatch2.cmake @@ -0,0 +1,49 @@ + +find_package(PkgConfig QUIET) +pkg_check_modules(PC_Catch2 QUIET Catch2) + +find_path(Catch2_INCLUDE_DIR + NAMES catch.hpp + PATHS ${PC_Catch2_INCLUDE_DIRS} ${CONAN_CATCH2_ROOT} + PATH_SUFFIXES catch2 +) + +if(Catch2_INCLUDE_DIR) + file(STRINGS "${Catch2_INCLUDE_DIR}/catch.hpp" _Catch2_version_lines + REGEX "#define[ \t]+CATCH_VERSION_(MAJOR|MINOR|PATCH)") + string(REGEX REPLACE ".*CATCH_VERSION_MAJOR +\([0-9]+\).*" "\\1" _Catch2_version_major "${_Catch2_version_lines}") + string(REGEX REPLACE ".*CATCH_VERSION_MINOR +\([0-9]+\).*" "\\1" _Catch2_version_minor "${_Catch2_version_lines}") + string(REGEX REPLACE ".*CATCH_VERSION_PATCH +\([0-9]+\).*" "\\1" _Catch2_version_patch "${_Catch2_version_lines}") + set(Catch2_VERSION "${_Catch2_version_major}.${_Catch2_version_minor}.${_Catch2_version_patch}") + unset(_Catch2_version_major) + unset(_Catch2_version_minor) + unset(_Catch2_version_patch) + unset(_Catch2_version_lines) +endif() + +include(FindPackageHandleStandardArgs) +find_package_handle_standard_args(Catch2 + FOUND_VAR Catch2_FOUND + REQUIRED_VARS + Catch2_INCLUDE_DIR + Catch2_VERSION + VERSION_VAR Catch2_VERSION +) + +if(Catch2_FOUND) + set(Catch2_INCLUDE_DIRS ${Catch2_INCLUDE_DIR}) + set(Catch2_DEFINITIONS ${PC_Catch2_CFLAGS_OTHER}) +endif() + +if(Catch2_FOUND AND NOT TARGET Catch2::Catch2) + add_library(Catch2::Catch2 UNKNOWN IMPORTED) + set_target_properties(Catch2::Catch2 PROPERTIES + IMPORTED_LOCATION "${Catch2_LIBRARY}" + INTERFACE_COMPILE_OPTIONS "${PC_Catch2_CFLAGS_OTHER}" + INTERFACE_INCLUDE_DIRECTORIES "${Catch2_INCLUDE_DIR}" + ) +endif() + +mark_as_advanced( + Catch2_INCLUDE_DIR +) diff --git a/externals/find-modules/FindLibzip.cmake b/externals/find-modules/FindLibzip.cmake new file mode 100644 index 000000000..f36b1687a --- /dev/null +++ b/externals/find-modules/FindLibzip.cmake @@ -0,0 +1,72 @@ + +find_package(PkgConfig QUIET) +pkg_check_modules(PC_LIBZIP QUIET libzip) + +find_path(LIBZIP_INCLUDE_DIR + NAMES zip.h + PATHS ${PC_LIBZIP_INCLUDE_DIRS} + "$ENV{LIB_DIR}/include" + "$ENV{INCLUDE}" + /usr/local/include + /usr/include +) +find_path(LIBZIP_INCLUDE_DIR_ZIPCONF + NAMES zipconf.h + HINTS ${PC_LIBZIP_INCLUDE_DIRS} + "$ENV{LIB_DIR}/include" + "$ENV{LIB_DIR}/lib/libzip/include" + "$ENV{LIB}/lib/libzip/include" + /usr/local/lib/libzip/include + /usr/lib/libzip/include + /usr/local/include + /usr/include + "$ENV{INCLUDE}" +) +find_library(LIBZIP_LIBRARY + NAMES zip + PATHS ${PC_LIBZIP_LIBRARY_DIRS} + "$ENV{LIB_DIR}/lib" "$ENV{LIB}" /usr/local/lib /usr/lib +) + +if (LIBZIP_INCLUDE_DIR_ZIPCONF) + FILE(READ "${LIBZIP_INCLUDE_DIR_ZIPCONF}/zipconf.h" _LIBZIP_VERSION_CONTENTS) + if (_LIBZIP_VERSION_CONTENTS) + STRING(REGEX REPLACE ".*#define LIBZIP_VERSION \"([0-9.]+)\".*" "\\1" LIBZIP_VERSION "${_LIBZIP_VERSION_CONTENTS}") + endif() + unset(_LIBZIP_VERSION_CONTENTS) +endif() + +set(LIBZIP_VERSION ${LIBZIP_VERSION} CACHE STRING "Version number of libzip") + +include(FindPackageHandleStandardArgs) +find_package_handle_standard_args(Libzip + FOUND_VAR LIBZIP_FOUND + REQUIRED_VARS + LIBZIP_LIBRARY + LIBZIP_INCLUDE_DIR + LIBZIP_INCLUDE_DIR_ZIPCONF + LIBZIP_VERSION + VERSION_VAR LIBZIP_VERSION +) + +if(LIBZIP_FOUND) + set(LIBZIP_LIBRARIES ${LIBZIP_LIBRARY}) + set(LIBZIP_INCLUDE_DIRS ${LIBZIP_INCLUDE_DIR}) + set(LIBZIP_DEFINITIONS ${PC_LIBZIP_CFLAGS_OTHER}) +endif() + +if(LIBZIP_FOUND AND NOT TARGET libzip::libzip) + add_library(libzip::libzip UNKNOWN IMPORTED) + set_target_properties(libzip::libzip PROPERTIES + IMPORTED_LOCATION "${LIBZIP_LIBRARY}" + INTERFACE_COMPILE_OPTIONS "${PC_LIBZIP_CFLAGS_OTHER}" + INTERFACE_INCLUDE_DIRECTORIES "${LIBZIP_INCLUDE_DIR}" + ) +endif() + +mark_as_advanced( + LIBZIP_INCLUDE_DIR + LIBZIP_INCLUDE_DIR_ZIPCONF + LIBZIP_LIBRARY + LIBZIP_VERSION +) diff --git a/externals/cmake-modules/FindUnicorn.cmake b/externals/find-modules/FindUnicorn.cmake index a0f2a71f6..a0f2a71f6 100644 --- a/externals/cmake-modules/FindUnicorn.cmake +++ b/externals/find-modules/FindUnicorn.cmake diff --git a/externals/find-modules/Findfmt.cmake b/externals/find-modules/Findfmt.cmake new file mode 100644 index 000000000..8ba51cbea --- /dev/null +++ b/externals/find-modules/Findfmt.cmake @@ -0,0 +1,69 @@ + +find_package(PkgConfig QUIET) +pkg_check_modules(PC_fmt QUIET fmt) + +find_path(fmt_INCLUDE_DIR + NAMES format.h + PATHS ${PC_fmt_INCLUDE_DIRS} ${CONAN_INCLUDE_DIRS_fmt} + PATH_SUFFIXES fmt +) + +find_library(fmt_LIBRARY + NAMES fmt + PATHS ${PC_fmt_LIBRARY_DIRS} ${CONAN_LIB_DIRS_fmt} +) + +if(fmt_INCLUDE_DIR) + set(_fmt_version_file "${fmt_INCLUDE_DIR}/core.h") + if(NOT EXISTS "${_fmt_version_file}") + set(_fmt_version_file "${fmt_INCLUDE_DIR}/format.h") + endif() + if(EXISTS "${_fmt_version_file}") + # parse "#define FMT_VERSION 60200" to 6.2.0 + file(STRINGS "${_fmt_version_file}" fmt_VERSION_LINE + REGEX "^#define[ \t]+FMT_VERSION[ \t]+[0-9]+$") + string(REGEX REPLACE "^#define[ \t]+FMT_VERSION[ \t]+([0-9]+)$" + "\\1" fmt_VERSION "${fmt_VERSION_LINE}") + foreach(ver "fmt_VERSION_PATCH" "fmt_VERSION_MINOR" "fmt_VERSION_MAJOR") + math(EXPR ${ver} "${fmt_VERSION} % 100") + math(EXPR fmt_VERSION "(${fmt_VERSION} - ${${ver}}) / 100") + endforeach() + set(fmt_VERSION + "${fmt_VERSION_MAJOR}.${fmt_VERSION_MINOR}.${fmt_VERSION_PATCH}") + endif() + unset(_fmt_version_file) + unset(fmt_VERSION_LINE) + unset(fmt_VERSION_MAJOR) + unset(fmt_VERSION_MINOR) + unset(fmt_VERSION_PATCH) +endif() + +include(FindPackageHandleStandardArgs) +find_package_handle_standard_args(fmt + FOUND_VAR fmt_FOUND + REQUIRED_VARS + fmt_LIBRARY + fmt_INCLUDE_DIR + fmt_VERSION + VERSION_VAR fmt_VERSION +) + +if(fmt_FOUND) + set(fmt_LIBRARIES ${fmt_LIBRARY}) + set(fmt_INCLUDE_DIRS ${fmt_INCLUDE_DIR}) + set(fmt_DEFINITIONS ${PC_fmt_CFLAGS_OTHER}) +endif() + +if(fmt_FOUND AND NOT TARGET fmt::fmt) + add_library(fmt::fmt UNKNOWN IMPORTED) + set_target_properties(fmt::fmt PROPERTIES + IMPORTED_LOCATION "${fmt_LIBRARY}" + INTERFACE_COMPILE_OPTIONS "${PC_fmt_CFLAGS_OTHER}" + INTERFACE_INCLUDE_DIRECTORIES "${fmt_INCLUDE_DIR}" + ) +endif() + +mark_as_advanced( + fmt_INCLUDE_DIR + fmt_LIBRARY +) diff --git a/externals/find-modules/Findlz4.cmake b/externals/find-modules/Findlz4.cmake new file mode 100644 index 000000000..6279854c0 --- /dev/null +++ b/externals/find-modules/Findlz4.cmake @@ -0,0 +1,54 @@ + +find_package(PkgConfig QUIET) +pkg_check_modules(PC_lz4 QUIET lz4) + +find_path(lz4_INCLUDE_DIR + NAMES lz4.h + PATHS ${PC_lz4_INCLUDE_DIRS} +) +find_library(lz4_LIBRARY + NAMES lz4 + PATHS ${PC_lz4_LIBRARY_DIRS} +) + +if(lz4_INCLUDE_DIR) + file(STRINGS "${lz4_INCLUDE_DIR}/lz4.h" _lz4_version_lines + REGEX "#define[ \t]+LZ4_VERSION_(MAJOR|MINOR|RELEASE)") + string(REGEX REPLACE ".*LZ4_VERSION_MAJOR *\([0-9]*\).*" "\\1" _lz4_version_major "${_lz4_version_lines}") + string(REGEX REPLACE ".*LZ4_VERSION_MINOR *\([0-9]*\).*" "\\1" _lz4_version_minor "${_lz4_version_lines}") + string(REGEX REPLACE ".*LZ4_VERSION_RELEASE *\([0-9]*\).*" "\\1" _lz4_version_release "${_lz4_version_lines}") + set(lz4_VERSION "${_lz4_version_major}.${_lz4_version_minor}.${_lz4_version_release}") + unset(_lz4_version_major) + unset(_lz4_version_minor) + unset(_lz4_version_release) + unset(_lz4_version_lines) +endif() + +include(FindPackageHandleStandardArgs) +find_package_handle_standard_args(lz4 + FOUND_VAR lz4_FOUND + REQUIRED_VARS + lz4_LIBRARY + lz4_INCLUDE_DIR + VERSION_VAR lz4_VERSION +) + +if(lz4_FOUND) + set(lz4_LIBRARIES ${lz4_LIBRARY}) + set(lz4_INCLUDE_DIRS ${lz4_INCLUDE_DIR}) + set(lz4_DEFINITIONS ${PC_lz4_CFLAGS_OTHER}) +endif() + +if(lz4_FOUND AND NOT TARGET lz4::lz4) + add_library(lz4::lz4 UNKNOWN IMPORTED) + set_target_properties(lz4::lz4 PROPERTIES + IMPORTED_LOCATION "${lz4_LIBRARY}" + INTERFACE_COMPILE_OPTIONS "${PC_lz4_CFLAGS_OTHER}" + INTERFACE_INCLUDE_DIRECTORIES "${lz4_INCLUDE_DIR}" + ) +endif() + +mark_as_advanced( + lz4_INCLUDE_DIR + lz4_LIBRARY +) diff --git a/externals/find-modules/Findnlohmann_json.cmake b/externals/find-modules/Findnlohmann_json.cmake new file mode 100644 index 000000000..b0c5b3e4e --- /dev/null +++ b/externals/find-modules/Findnlohmann_json.cmake @@ -0,0 +1,49 @@ + +find_package(PkgConfig QUIET) +pkg_check_modules(PC_nlohmann_json QUIET nlohmann_json) + +find_path(nlohmann_json_INCLUDE_DIR + NAMES json.hpp + PATHS ${PC_nlohmann_json_INCLUDE_DIRS} + PATH_SUFFIXES nlohmann +) + +if(nlohmann_json_INCLUDE_DIR) + file(STRINGS "${nlohmann_json_INCLUDE_DIR}/json.hpp" _nlohmann_json_version_lines + REGEX "#define[ \t]+NLOHMANN_JSON_VERSION_(MAJOR|MINOR|PATCH)") + string(REGEX REPLACE ".*NLOHMANN_JSON_VERSION_MAJOR +\([0-9]+\).*" "\\1" _nlohmann_json_version_major "${_nlohmann_json_version_lines}") + string(REGEX REPLACE ".*NLOHMANN_JSON_VERSION_MINOR +\([0-9]+\).*" "\\1" _nlohmann_json_version_minor "${_nlohmann_json_version_lines}") + string(REGEX REPLACE ".*NLOHMANN_JSON_VERSION_PATCH +\([0-9]+\).*" "\\1" _nlohmann_json_version_patch "${_nlohmann_json_version_lines}") + set(nlohmann_json_VERSION "${_nlohmann_json_version_major}.${_nlohmann_json_version_minor}.${_nlohmann_json_version_patch}") + unset(_nlohmann_json_version_major) + unset(_nlohmann_json_version_minor) + unset(_nlohmann_json_version_patch) + unset(_nlohmann_json_version_lines) +endif() + +include(FindPackageHandleStandardArgs) +find_package_handle_standard_args(nlohmann_json + FOUND_VAR nlohmann_json_FOUND + REQUIRED_VARS + nlohmann_json_INCLUDE_DIR + nlohmann_json_VERSION + VERSION_VAR nlohmann_json_VERSION +) + +if(nlohmann_json_FOUND) + set(nlohmann_json_INCLUDE_DIRS ${nlohmann_json_INCLUDE_DIR}) + set(nlohmann_json_DEFINITIONS ${PC_nlohmann_json_CFLAGS_OTHER}) +endif() + +if(nlohmann_json_FOUND AND NOT TARGET nlohmann_json::nlohmann_json) + add_library(nlohmann_json::nlohmann_json UNKNOWN IMPORTED) + set_target_properties(nlohmann_json::nlohmann_json PROPERTIES + IMPORTED_LOCATION "${nlohmann_json_LIBRARY}" + INTERFACE_COMPILE_OPTIONS "${PC_nlohmann_json_CFLAGS_OTHER}" + INTERFACE_INCLUDE_DIRECTORIES "${nlohmann_json_INCLUDE_DIR}" + ) +endif() + +mark_as_advanced( + nlohmann_json_INCLUDE_DIR +) diff --git a/externals/find-modules/Findopus.cmake b/externals/find-modules/Findopus.cmake new file mode 100644 index 000000000..de84bd995 --- /dev/null +++ b/externals/find-modules/Findopus.cmake @@ -0,0 +1,42 @@ + +find_package(PkgConfig QUIET) +pkg_check_modules(PC_opus QUIET opus) + +find_path(opus_INCLUDE_DIR + NAMES opus.h + PATHS ${PC_opus_INCLUDE_DIRS} + PATH_SUFFIXES opus +) +find_library(opus_LIBRARY + NAMES opus + PATHS ${PC_opus_LIBRARY_DIRS} +) + +include(FindPackageHandleStandardArgs) +find_package_handle_standard_args(opus + FOUND_VAR opus_FOUND + REQUIRED_VARS + opus_LIBRARY + opus_INCLUDE_DIR + VERSION_VAR opus_VERSION +) + +if(opus_FOUND) + set(Opus_LIBRARIES ${opus_LIBRARY}) + set(Opus_INCLUDE_DIRS ${opus_INCLUDE_DIR}) + set(Opus_DEFINITIONS ${PC_opus_CFLAGS_OTHER}) +endif() + +if(opus_FOUND AND NOT TARGET Opus::Opus) + add_library(Opus::Opus UNKNOWN IMPORTED) + set_target_properties(Opus::Opus PROPERTIES + IMPORTED_LOCATION "${opus_LIBRARY}" + INTERFACE_COMPILE_OPTIONS "${PC_opus_CFLAGS_OTHER}" + INTERFACE_INCLUDE_DIRECTORIES "${opus_INCLUDE_DIR}" + ) +endif() + +mark_as_advanced( + opus_INCLUDE_DIR + opus_LIBRARY +) diff --git a/externals/find-modules/Findzstd.cmake b/externals/find-modules/Findzstd.cmake new file mode 100644 index 000000000..539abbafc --- /dev/null +++ b/externals/find-modules/Findzstd.cmake @@ -0,0 +1,55 @@ + +find_package(PkgConfig QUIET) +pkg_check_modules(PC_zstd QUIET libzstd) + +find_path(zstd_INCLUDE_DIR + NAMES zstd.h + PATHS ${PC_zstd_INCLUDE_DIRS} +) +find_library(zstd_LIBRARY + NAMES zstd + PATHS ${PC_zstd_LIBRARY_DIRS} +) + +if(zstd_INCLUDE_DIR) + file(STRINGS "${zstd_INCLUDE_DIR}/zstd.h" _zstd_version_lines + REGEX "#define[ \t]+ZSTD_VERSION_(MAJOR|MINOR|RELEASE)") + string(REGEX REPLACE ".*ZSTD_VERSION_MAJOR *\([0-9]*\).*" "\\1" _zstd_version_major "${_zstd_version_lines}") + string(REGEX REPLACE ".*ZSTD_VERSION_MINOR *\([0-9]*\).*" "\\1" _zstd_version_minor "${_zstd_version_lines}") + string(REGEX REPLACE ".*ZSTD_VERSION_RELEASE *\([0-9]*\).*" "\\1" _zstd_version_release "${_zstd_version_lines}") + set(zstd_VERSION "${_zstd_version_major}.${_zstd_version_minor}.${_zstd_version_release}") + unset(_zstd_version_major) + unset(_zstd_version_minor) + unset(_zstd_version_release) + unset(_zstd_version_lines) +endif() + +include(FindPackageHandleStandardArgs) +find_package_handle_standard_args(zstd + FOUND_VAR zstd_FOUND + REQUIRED_VARS + zstd_LIBRARY + zstd_INCLUDE_DIR + zstd_VERSION + VERSION_VAR zstd_VERSION +) + +if(zstd_FOUND) + set(zstd_LIBRARIES ${zstd_LIBRARY}) + set(zstd_INCLUDE_DIRS ${zstd_INCLUDE_DIR}) + set(zstd_DEFINITIONS ${PC_zstd_CFLAGS_OTHER}) +endif() + +if(zstd_FOUND AND NOT TARGET zstd::zstd) + add_library(zstd::zstd UNKNOWN IMPORTED) + set_target_properties(zstd::zstd PROPERTIES + IMPORTED_LOCATION "${zstd_LIBRARY}" + INTERFACE_COMPILE_OPTIONS "${PC_zstd_CFLAGS_OTHER}" + INTERFACE_INCLUDE_DIRECTORIES "${zstd_INCLUDE_DIR}" + ) +endif() + +mark_as_advanced( + zstd_INCLUDE_DIR + zstd_LIBRARY +) diff --git a/externals/fmt b/externals/fmt deleted file mode 160000 -Subproject 9bdd1596cef1b57b9556f8bef32dc4a32322ef3 diff --git a/externals/json/README.md b/externals/json/README.md deleted file mode 100644 index 2db68ffde..000000000 --- a/externals/json/README.md +++ /dev/null @@ -1,9 +0,0 @@ -JSON for Modern C++ -=================== - -v3.1.2 - -This is a mirror providing the single required header file. - -The original repository can be found at: -https://github.com/nlohmann/json/commit/d2dd27dc3b8472dbaa7d66f83619b3ebcd9185fe diff --git a/externals/json/json.hpp b/externals/json/json.hpp deleted file mode 100644 index 6b6655af8..000000000 --- a/externals/json/json.hpp +++ /dev/null @@ -1,17300 +0,0 @@ -/* - __ _____ _____ _____ - __| | __| | | | JSON for Modern C++ -| | |__ | | | | | | version 3.1.2 -|_____|_____|_____|_|___| https://github.com/nlohmann/json - -Licensed under the MIT License <http://opensource.org/licenses/MIT>. -Copyright (c) 2013-2018 Niels Lohmann <http://nlohmann.me>. - -Permission is hereby granted, free of charge, to any person obtaining a copy -of this software and associated documentation files (the "Software"), to deal -in the Software without restriction, including without limitation the rights -to use, copy, modify, merge, publish, distribute, sublicense, and/or sell -copies of the Software, and to permit persons to whom the Software is -furnished to do so, subject to the following conditions: - -The above copyright notice and this permission notice shall be included in all -copies or substantial portions of the Software. - -THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE -AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER -LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, -OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE -SOFTWARE. -*/ - -#ifndef NLOHMANN_JSON_HPP -#define NLOHMANN_JSON_HPP - -#define NLOHMANN_JSON_VERSION_MAJOR 3 -#define NLOHMANN_JSON_VERSION_MINOR 1 -#define NLOHMANN_JSON_VERSION_PATCH 2 - -#include <algorithm> // all_of, find, for_each -#include <cassert> // assert -#include <ciso646> // and, not, or -#include <cstddef> // nullptr_t, ptrdiff_t, size_t -#include <functional> // hash, less -#include <initializer_list> // initializer_list -#include <iosfwd> // istream, ostream -#include <iterator> // iterator_traits, random_access_iterator_tag -#include <numeric> // accumulate -#include <string> // string, stoi, to_string -#include <utility> // declval, forward, move, pair, swap - -// #include <nlohmann/json_fwd.hpp> -#ifndef NLOHMANN_JSON_FWD_HPP -#define NLOHMANN_JSON_FWD_HPP - -#include <cstdint> // int64_t, uint64_t -#include <map> // map -#include <memory> // allocator -#include <string> // string -#include <vector> // vector - -/*! -@brief namespace for Niels Lohmann -@see https://github.com/nlohmann -@since version 1.0.0 -*/ -namespace nlohmann -{ -/*! -@brief default JSONSerializer template argument - -This serializer ignores the template arguments and uses ADL -([argument-dependent lookup](http://en.cppreference.com/w/cpp/language/adl)) -for serialization. -*/ -template<typename = void, typename = void> -struct adl_serializer; - -template<template<typename U, typename V, typename... Args> class ObjectType = - std::map, - template<typename U, typename... Args> class ArrayType = std::vector, - class StringType = std::string, class BooleanType = bool, - class NumberIntegerType = std::int64_t, - class NumberUnsignedType = std::uint64_t, - class NumberFloatType = double, - template<typename U> class AllocatorType = std::allocator, - template<typename T, typename SFINAE = void> class JSONSerializer = - adl_serializer> -class basic_json; - -/*! -@brief JSON Pointer - -A JSON pointer defines a string syntax for identifying a specific value -within a JSON document. It can be used with functions `at` and -`operator[]`. Furthermore, JSON pointers are the base for JSON patches. - -@sa [RFC 6901](https://tools.ietf.org/html/rfc6901) - -@since version 2.0.0 -*/ -template<typename BasicJsonType> -class json_pointer; - -/*! -@brief default JSON class - -This type is the default specialization of the @ref basic_json class which -uses the standard template types. - -@since version 1.0.0 -*/ -using json = basic_json<>; -} - -#endif - -// #include <nlohmann/detail/macro_scope.hpp> - - -// This file contains all internal macro definitions -// You MUST include macro_unscope.hpp at the end of json.hpp to undef all of them - -// exclude unsupported compilers -#if defined(__clang__) - #if (__clang_major__ * 10000 + __clang_minor__ * 100 + __clang_patchlevel__) < 30400 - #error "unsupported Clang version - see https://github.com/nlohmann/json#supported-compilers" - #endif -#elif defined(__GNUC__) && !(defined(__ICC) || defined(__INTEL_COMPILER)) - #if (__GNUC__ * 10000 + __GNUC_MINOR__ * 100 + __GNUC_PATCHLEVEL__) < 40900 - #error "unsupported GCC version - see https://github.com/nlohmann/json#supported-compilers" - #endif -#endif - -// disable float-equal warnings on GCC/clang -#if defined(__clang__) || defined(__GNUC__) || defined(__GNUG__) - #pragma GCC diagnostic push - #pragma GCC diagnostic ignored "-Wfloat-equal" -#endif - -// disable documentation warnings on clang -#if defined(__clang__) - #pragma GCC diagnostic push - #pragma GCC diagnostic ignored "-Wdocumentation" -#endif - -// allow for portable deprecation warnings -#if defined(__clang__) || defined(__GNUC__) || defined(__GNUG__) - #define JSON_DEPRECATED __attribute__((deprecated)) -#elif defined(_MSC_VER) - #define JSON_DEPRECATED __declspec(deprecated) -#else - #define JSON_DEPRECATED -#endif - -// allow to disable exceptions -#if (defined(__cpp_exceptions) || defined(__EXCEPTIONS) || defined(_CPPUNWIND)) && !defined(JSON_NOEXCEPTION) - #define JSON_THROW(exception) throw exception - #define JSON_TRY try - #define JSON_CATCH(exception) catch(exception) -#else - #define JSON_THROW(exception) std::abort() - #define JSON_TRY if(true) - #define JSON_CATCH(exception) if(false) -#endif - -// override exception macros -#if defined(JSON_THROW_USER) - #undef JSON_THROW - #define JSON_THROW JSON_THROW_USER -#endif -#if defined(JSON_TRY_USER) - #undef JSON_TRY - #define JSON_TRY JSON_TRY_USER -#endif -#if defined(JSON_CATCH_USER) - #undef JSON_CATCH - #define JSON_CATCH JSON_CATCH_USER -#endif - -// manual branch prediction -#if defined(__clang__) || defined(__GNUC__) || defined(__GNUG__) - #define JSON_LIKELY(x) __builtin_expect(!!(x), 1) - #define JSON_UNLIKELY(x) __builtin_expect(!!(x), 0) -#else - #define JSON_LIKELY(x) x - #define JSON_UNLIKELY(x) x -#endif - -// C++ language standard detection -#if (defined(__cplusplus) && __cplusplus >= 201703L) || (defined(_HAS_CXX17) && _HAS_CXX17 == 1) // fix for issue #464 - #define JSON_HAS_CPP_17 - #define JSON_HAS_CPP_14 -#elif (defined(__cplusplus) && __cplusplus >= 201402L) || (defined(_HAS_CXX14) && _HAS_CXX14 == 1) - #define JSON_HAS_CPP_14 -#endif - -// Ugly macros to avoid uglier copy-paste when specializing basic_json. They -// may be removed in the future once the class is split. - -#define NLOHMANN_BASIC_JSON_TPL_DECLARATION \ - template<template<typename, typename, typename...> class ObjectType, \ - template<typename, typename...> class ArrayType, \ - class StringType, class BooleanType, class NumberIntegerType, \ - class NumberUnsignedType, class NumberFloatType, \ - template<typename> class AllocatorType, \ - template<typename, typename = void> class JSONSerializer> - -#define NLOHMANN_BASIC_JSON_TPL \ - basic_json<ObjectType, ArrayType, StringType, BooleanType, \ - NumberIntegerType, NumberUnsignedType, NumberFloatType, \ - AllocatorType, JSONSerializer> - -/*! -@brief Helper to determine whether there's a key_type for T. - -This helper is used to tell associative containers apart from other containers -such as sequence containers. For instance, `std::map` passes the test as it -contains a `mapped_type`, whereas `std::vector` fails the test. - -@sa http://stackoverflow.com/a/7728728/266378 -@since version 1.0.0, overworked in version 2.0.6 -*/ -#define NLOHMANN_JSON_HAS_HELPER(type) \ - template<typename T> struct has_##type { \ - private: \ - template<typename U, typename = typename U::type> \ - static int detect(U &&); \ - static void detect(...); \ - public: \ - static constexpr bool value = \ - std::is_integral<decltype(detect(std::declval<T>()))>::value; \ - } - -// #include <nlohmann/detail/meta.hpp> - - -#include <ciso646> // not -#include <cstddef> // size_t -#include <limits> // numeric_limits -#include <type_traits> // conditional, enable_if, false_type, integral_constant, is_constructible, is_integral, is_same, remove_cv, remove_reference, true_type -#include <utility> // declval - -// #include <nlohmann/json_fwd.hpp> - -// #include <nlohmann/detail/macro_scope.hpp> - - -namespace nlohmann -{ -/*! -@brief detail namespace with internal helper functions - -This namespace collects functions that should not be exposed, -implementations of some @ref basic_json methods, and meta-programming helpers. - -@since version 2.1.0 -*/ -namespace detail -{ -///////////// -// helpers // -///////////// - -template<typename> struct is_basic_json : std::false_type {}; - -NLOHMANN_BASIC_JSON_TPL_DECLARATION -struct is_basic_json<NLOHMANN_BASIC_JSON_TPL> : std::true_type {}; - -// alias templates to reduce boilerplate -template<bool B, typename T = void> -using enable_if_t = typename std::enable_if<B, T>::type; - -template<typename T> -using uncvref_t = typename std::remove_cv<typename std::remove_reference<T>::type>::type; - -// implementation of C++14 index_sequence and affiliates -// source: https://stackoverflow.com/a/32223343 -template<std::size_t... Ints> -struct index_sequence -{ - using type = index_sequence; - using value_type = std::size_t; - static constexpr std::size_t size() noexcept - { - return sizeof...(Ints); - } -}; - -template<class Sequence1, class Sequence2> -struct merge_and_renumber; - -template<std::size_t... I1, std::size_t... I2> -struct merge_and_renumber<index_sequence<I1...>, index_sequence<I2...>> - : index_sequence < I1..., (sizeof...(I1) + I2)... > {}; - -template<std::size_t N> -struct make_index_sequence - : merge_and_renumber < typename make_index_sequence < N / 2 >::type, - typename make_index_sequence < N - N / 2 >::type > {}; - -template<> struct make_index_sequence<0> : index_sequence<> {}; -template<> struct make_index_sequence<1> : index_sequence<0> {}; - -template<typename... Ts> -using index_sequence_for = make_index_sequence<sizeof...(Ts)>; - -/* -Implementation of two C++17 constructs: conjunction, negation. This is needed -to avoid evaluating all the traits in a condition - -For example: not std::is_same<void, T>::value and has_value_type<T>::value -will not compile when T = void (on MSVC at least). Whereas -conjunction<negation<std::is_same<void, T>>, has_value_type<T>>::value will -stop evaluating if negation<...>::value == false - -Please note that those constructs must be used with caution, since symbols can -become very long quickly (which can slow down compilation and cause MSVC -internal compiler errors). Only use it when you have to (see example ahead). -*/ -template<class...> struct conjunction : std::true_type {}; -template<class B1> struct conjunction<B1> : B1 {}; -template<class B1, class... Bn> -struct conjunction<B1, Bn...> : std::conditional<bool(B1::value), conjunction<Bn...>, B1>::type {}; - -template<class B> struct negation : std::integral_constant<bool, not B::value> {}; - -// dispatch utility (taken from ranges-v3) -template<unsigned N> struct priority_tag : priority_tag < N - 1 > {}; -template<> struct priority_tag<0> {}; - -//////////////////////// -// has_/is_ functions // -//////////////////////// - -// source: https://stackoverflow.com/a/37193089/4116453 - -template <typename T, typename = void> -struct is_complete_type : std::false_type {}; - -template <typename T> -struct is_complete_type<T, decltype(void(sizeof(T)))> : std::true_type {}; - -NLOHMANN_JSON_HAS_HELPER(mapped_type); -NLOHMANN_JSON_HAS_HELPER(key_type); -NLOHMANN_JSON_HAS_HELPER(value_type); -NLOHMANN_JSON_HAS_HELPER(iterator); - -template<bool B, class RealType, class CompatibleObjectType> -struct is_compatible_object_type_impl : std::false_type {}; - -template<class RealType, class CompatibleObjectType> -struct is_compatible_object_type_impl<true, RealType, CompatibleObjectType> -{ - static constexpr auto value = - std::is_constructible<typename RealType::key_type, typename CompatibleObjectType::key_type>::value and - std::is_constructible<typename RealType::mapped_type, typename CompatibleObjectType::mapped_type>::value; -}; - -template<class BasicJsonType, class CompatibleObjectType> -struct is_compatible_object_type -{ - static auto constexpr value = is_compatible_object_type_impl < - conjunction<negation<std::is_same<void, CompatibleObjectType>>, - has_mapped_type<CompatibleObjectType>, - has_key_type<CompatibleObjectType>>::value, - typename BasicJsonType::object_t, CompatibleObjectType >::value; -}; - -template<typename BasicJsonType, typename T> -struct is_basic_json_nested_type -{ - static auto constexpr value = std::is_same<T, typename BasicJsonType::iterator>::value or - std::is_same<T, typename BasicJsonType::const_iterator>::value or - std::is_same<T, typename BasicJsonType::reverse_iterator>::value or - std::is_same<T, typename BasicJsonType::const_reverse_iterator>::value; -}; - -template<class BasicJsonType, class CompatibleArrayType> -struct is_compatible_array_type -{ - static auto constexpr value = - conjunction<negation<std::is_same<void, CompatibleArrayType>>, - negation<is_compatible_object_type< - BasicJsonType, CompatibleArrayType>>, - negation<std::is_constructible<typename BasicJsonType::string_t, - CompatibleArrayType>>, - negation<is_basic_json_nested_type<BasicJsonType, CompatibleArrayType>>, - has_value_type<CompatibleArrayType>, - has_iterator<CompatibleArrayType>>::value; -}; - -template<bool, typename, typename> -struct is_compatible_integer_type_impl : std::false_type {}; - -template<typename RealIntegerType, typename CompatibleNumberIntegerType> -struct is_compatible_integer_type_impl<true, RealIntegerType, CompatibleNumberIntegerType> -{ - // is there an assert somewhere on overflows? - using RealLimits = std::numeric_limits<RealIntegerType>; - using CompatibleLimits = std::numeric_limits<CompatibleNumberIntegerType>; - - static constexpr auto value = - std::is_constructible<RealIntegerType, CompatibleNumberIntegerType>::value and - CompatibleLimits::is_integer and - RealLimits::is_signed == CompatibleLimits::is_signed; -}; - -template<typename RealIntegerType, typename CompatibleNumberIntegerType> -struct is_compatible_integer_type -{ - static constexpr auto value = - is_compatible_integer_type_impl < - std::is_integral<CompatibleNumberIntegerType>::value and - not std::is_same<bool, CompatibleNumberIntegerType>::value, - RealIntegerType, CompatibleNumberIntegerType > ::value; -}; - -// trait checking if JSONSerializer<T>::from_json(json const&, udt&) exists -template<typename BasicJsonType, typename T> -struct has_from_json -{ - private: - // also check the return type of from_json - template<typename U, typename = enable_if_t<std::is_same<void, decltype(uncvref_t<U>::from_json( - std::declval<BasicJsonType>(), std::declval<T&>()))>::value>> - static int detect(U&&); - static void detect(...); - - public: - static constexpr bool value = std::is_integral<decltype( - detect(std::declval<typename BasicJsonType::template json_serializer<T, void>>()))>::value; -}; - -// This trait checks if JSONSerializer<T>::from_json(json const&) exists -// this overload is used for non-default-constructible user-defined-types -template<typename BasicJsonType, typename T> -struct has_non_default_from_json -{ - private: - template < - typename U, - typename = enable_if_t<std::is_same< - T, decltype(uncvref_t<U>::from_json(std::declval<BasicJsonType>()))>::value >> - static int detect(U&&); - static void detect(...); - - public: - static constexpr bool value = std::is_integral<decltype(detect( - std::declval<typename BasicJsonType::template json_serializer<T, void>>()))>::value; -}; - -// This trait checks if BasicJsonType::json_serializer<T>::to_json exists -template<typename BasicJsonType, typename T> -struct has_to_json -{ - private: - template<typename U, typename = decltype(uncvref_t<U>::to_json( - std::declval<BasicJsonType&>(), std::declval<T>()))> - static int detect(U&&); - static void detect(...); - - public: - static constexpr bool value = std::is_integral<decltype(detect( - std::declval<typename BasicJsonType::template json_serializer<T, void>>()))>::value; -}; - -template <typename BasicJsonType, typename CompatibleCompleteType> -struct is_compatible_complete_type -{ - static constexpr bool value = - not std::is_base_of<std::istream, CompatibleCompleteType>::value and - not is_basic_json<CompatibleCompleteType>::value and - not is_basic_json_nested_type<BasicJsonType, CompatibleCompleteType>::value and - has_to_json<BasicJsonType, CompatibleCompleteType>::value; -}; - -template <typename BasicJsonType, typename CompatibleType> -struct is_compatible_type - : conjunction<is_complete_type<CompatibleType>, - is_compatible_complete_type<BasicJsonType, CompatibleType>> -{ -}; - -// taken from ranges-v3 -template<typename T> -struct static_const -{ - static constexpr T value{}; -}; - -template<typename T> -constexpr T static_const<T>::value; -} -} - -// #include <nlohmann/detail/exceptions.hpp> - - -#include <exception> // exception -#include <stdexcept> // runtime_error -#include <string> // to_string - -namespace nlohmann -{ -namespace detail -{ -//////////////// -// exceptions // -//////////////// - -/*! -@brief general exception of the @ref basic_json class - -This class is an extension of `std::exception` objects with a member @a id for -exception ids. It is used as the base class for all exceptions thrown by the -@ref basic_json class. This class can hence be used as "wildcard" to catch -exceptions. - -Subclasses: -- @ref parse_error for exceptions indicating a parse error -- @ref invalid_iterator for exceptions indicating errors with iterators -- @ref type_error for exceptions indicating executing a member function with - a wrong type -- @ref out_of_range for exceptions indicating access out of the defined range -- @ref other_error for exceptions indicating other library errors - -@internal -@note To have nothrow-copy-constructible exceptions, we internally use - `std::runtime_error` which can cope with arbitrary-length error messages. - Intermediate strings are built with static functions and then passed to - the actual constructor. -@endinternal - -@liveexample{The following code shows how arbitrary library exceptions can be -caught.,exception} - -@since version 3.0.0 -*/ -class exception : public std::exception -{ - public: - /// returns the explanatory string - const char* what() const noexcept override - { - return m.what(); - } - - /// the id of the exception - const int id; - - protected: - exception(int id_, const char* what_arg) : id(id_), m(what_arg) {} - - static std::string name(const std::string& ename, int id_) - { - return "[json.exception." + ename + "." + std::to_string(id_) + "] "; - } - - private: - /// an exception object as storage for error messages - std::runtime_error m; -}; - -/*! -@brief exception indicating a parse error - -This exception is thrown by the library when a parse error occurs. Parse errors -can occur during the deserialization of JSON text, CBOR, MessagePack, as well -as when using JSON Patch. - -Member @a byte holds the byte index of the last read character in the input -file. - -Exceptions have ids 1xx. - -name / id | example message | description ------------------------------- | --------------- | ------------------------- -json.exception.parse_error.101 | parse error at 2: unexpected end of input; expected string literal | This error indicates a syntax error while deserializing a JSON text. The error message describes that an unexpected token (character) was encountered, and the member @a byte indicates the error position. -json.exception.parse_error.102 | parse error at 14: missing or wrong low surrogate | JSON uses the `\uxxxx` format to describe Unicode characters. Code points above above 0xFFFF are split into two `\uxxxx` entries ("surrogate pairs"). This error indicates that the surrogate pair is incomplete or contains an invalid code point. -json.exception.parse_error.103 | parse error: code points above 0x10FFFF are invalid | Unicode supports code points up to 0x10FFFF. Code points above 0x10FFFF are invalid. -json.exception.parse_error.104 | parse error: JSON patch must be an array of objects | [RFC 6902](https://tools.ietf.org/html/rfc6902) requires a JSON Patch document to be a JSON document that represents an array of objects. -json.exception.parse_error.105 | parse error: operation must have string member 'op' | An operation of a JSON Patch document must contain exactly one "op" member, whose value indicates the operation to perform. Its value must be one of "add", "remove", "replace", "move", "copy", or "test"; other values are errors. -json.exception.parse_error.106 | parse error: array index '01' must not begin with '0' | An array index in a JSON Pointer ([RFC 6901](https://tools.ietf.org/html/rfc6901)) may be `0` or any number without a leading `0`. -json.exception.parse_error.107 | parse error: JSON pointer must be empty or begin with '/' - was: 'foo' | A JSON Pointer must be a Unicode string containing a sequence of zero or more reference tokens, each prefixed by a `/` character. -json.exception.parse_error.108 | parse error: escape character '~' must be followed with '0' or '1' | In a JSON Pointer, only `~0` and `~1` are valid escape sequences. -json.exception.parse_error.109 | parse error: array index 'one' is not a number | A JSON Pointer array index must be a number. -json.exception.parse_error.110 | parse error at 1: cannot read 2 bytes from vector | When parsing CBOR or MessagePack, the byte vector ends before the complete value has been read. -json.exception.parse_error.112 | parse error at 1: error reading CBOR; last byte: 0xF8 | Not all types of CBOR or MessagePack are supported. This exception occurs if an unsupported byte was read. -json.exception.parse_error.113 | parse error at 2: expected a CBOR string; last byte: 0x98 | While parsing a map key, a value that is not a string has been read. - -@note For an input with n bytes, 1 is the index of the first character and n+1 - is the index of the terminating null byte or the end of file. This also - holds true when reading a byte vector (CBOR or MessagePack). - -@liveexample{The following code shows how a `parse_error` exception can be -caught.,parse_error} - -@sa @ref exception for the base class of the library exceptions -@sa @ref invalid_iterator for exceptions indicating errors with iterators -@sa @ref type_error for exceptions indicating executing a member function with - a wrong type -@sa @ref out_of_range for exceptions indicating access out of the defined range -@sa @ref other_error for exceptions indicating other library errors - -@since version 3.0.0 -*/ -class parse_error : public exception -{ - public: - /*! - @brief create a parse error exception - @param[in] id_ the id of the exception - @param[in] byte_ the byte index where the error occurred (or 0 if the - position cannot be determined) - @param[in] what_arg the explanatory string - @return parse_error object - */ - static parse_error create(int id_, std::size_t byte_, const std::string& what_arg) - { - std::string w = exception::name("parse_error", id_) + "parse error" + - (byte_ != 0 ? (" at " + std::to_string(byte_)) : "") + - ": " + what_arg; - return parse_error(id_, byte_, w.c_str()); - } - - /*! - @brief byte index of the parse error - - The byte index of the last read character in the input file. - - @note For an input with n bytes, 1 is the index of the first character and - n+1 is the index of the terminating null byte or the end of file. - This also holds true when reading a byte vector (CBOR or MessagePack). - */ - const std::size_t byte; - - private: - parse_error(int id_, std::size_t byte_, const char* what_arg) - : exception(id_, what_arg), byte(byte_) {} -}; - -/*! -@brief exception indicating errors with iterators - -This exception is thrown if iterators passed to a library function do not match -the expected semantics. - -Exceptions have ids 2xx. - -name / id | example message | description ------------------------------------ | --------------- | ------------------------- -json.exception.invalid_iterator.201 | iterators are not compatible | The iterators passed to constructor @ref basic_json(InputIT first, InputIT last) are not compatible, meaning they do not belong to the same container. Therefore, the range (@a first, @a last) is invalid. -json.exception.invalid_iterator.202 | iterator does not fit current value | In an erase or insert function, the passed iterator @a pos does not belong to the JSON value for which the function was called. It hence does not define a valid position for the deletion/insertion. -json.exception.invalid_iterator.203 | iterators do not fit current value | Either iterator passed to function @ref erase(IteratorType first, IteratorType last) does not belong to the JSON value from which values shall be erased. It hence does not define a valid range to delete values from. -json.exception.invalid_iterator.204 | iterators out of range | When an iterator range for a primitive type (number, boolean, or string) is passed to a constructor or an erase function, this range has to be exactly (@ref begin(), @ref end()), because this is the only way the single stored value is expressed. All other ranges are invalid. -json.exception.invalid_iterator.205 | iterator out of range | When an iterator for a primitive type (number, boolean, or string) is passed to an erase function, the iterator has to be the @ref begin() iterator, because it is the only way to address the stored value. All other iterators are invalid. -json.exception.invalid_iterator.206 | cannot construct with iterators from null | The iterators passed to constructor @ref basic_json(InputIT first, InputIT last) belong to a JSON null value and hence to not define a valid range. -json.exception.invalid_iterator.207 | cannot use key() for non-object iterators | The key() member function can only be used on iterators belonging to a JSON object, because other types do not have a concept of a key. -json.exception.invalid_iterator.208 | cannot use operator[] for object iterators | The operator[] to specify a concrete offset cannot be used on iterators belonging to a JSON object, because JSON objects are unordered. -json.exception.invalid_iterator.209 | cannot use offsets with object iterators | The offset operators (+, -, +=, -=) cannot be used on iterators belonging to a JSON object, because JSON objects are unordered. -json.exception.invalid_iterator.210 | iterators do not fit | The iterator range passed to the insert function are not compatible, meaning they do not belong to the same container. Therefore, the range (@a first, @a last) is invalid. -json.exception.invalid_iterator.211 | passed iterators may not belong to container | The iterator range passed to the insert function must not be a subrange of the container to insert to. -json.exception.invalid_iterator.212 | cannot compare iterators of different containers | When two iterators are compared, they must belong to the same container. -json.exception.invalid_iterator.213 | cannot compare order of object iterators | The order of object iterators cannot be compared, because JSON objects are unordered. -json.exception.invalid_iterator.214 | cannot get value | Cannot get value for iterator: Either the iterator belongs to a null value or it is an iterator to a primitive type (number, boolean, or string), but the iterator is different to @ref begin(). - -@liveexample{The following code shows how an `invalid_iterator` exception can be -caught.,invalid_iterator} - -@sa @ref exception for the base class of the library exceptions -@sa @ref parse_error for exceptions indicating a parse error -@sa @ref type_error for exceptions indicating executing a member function with - a wrong type -@sa @ref out_of_range for exceptions indicating access out of the defined range -@sa @ref other_error for exceptions indicating other library errors - -@since version 3.0.0 -*/ -class invalid_iterator : public exception -{ - public: - static invalid_iterator create(int id_, const std::string& what_arg) - { - std::string w = exception::name("invalid_iterator", id_) + what_arg; - return invalid_iterator(id_, w.c_str()); - } - - private: - invalid_iterator(int id_, const char* what_arg) - : exception(id_, what_arg) {} -}; - -/*! -@brief exception indicating executing a member function with a wrong type - -This exception is thrown in case of a type error; that is, a library function is -executed on a JSON value whose type does not match the expected semantics. - -Exceptions have ids 3xx. - -name / id | example message | description ------------------------------ | --------------- | ------------------------- -json.exception.type_error.301 | cannot create object from initializer list | To create an object from an initializer list, the initializer list must consist only of a list of pairs whose first element is a string. When this constraint is violated, an array is created instead. -json.exception.type_error.302 | type must be object, but is array | During implicit or explicit value conversion, the JSON type must be compatible to the target type. For instance, a JSON string can only be converted into string types, but not into numbers or boolean types. -json.exception.type_error.303 | incompatible ReferenceType for get_ref, actual type is object | To retrieve a reference to a value stored in a @ref basic_json object with @ref get_ref, the type of the reference must match the value type. For instance, for a JSON array, the @a ReferenceType must be @ref array_t&. -json.exception.type_error.304 | cannot use at() with string | The @ref at() member functions can only be executed for certain JSON types. -json.exception.type_error.305 | cannot use operator[] with string | The @ref operator[] member functions can only be executed for certain JSON types. -json.exception.type_error.306 | cannot use value() with string | The @ref value() member functions can only be executed for certain JSON types. -json.exception.type_error.307 | cannot use erase() with string | The @ref erase() member functions can only be executed for certain JSON types. -json.exception.type_error.308 | cannot use push_back() with string | The @ref push_back() and @ref operator+= member functions can only be executed for certain JSON types. -json.exception.type_error.309 | cannot use insert() with | The @ref insert() member functions can only be executed for certain JSON types. -json.exception.type_error.310 | cannot use swap() with number | The @ref swap() member functions can only be executed for certain JSON types. -json.exception.type_error.311 | cannot use emplace_back() with string | The @ref emplace_back() member function can only be executed for certain JSON types. -json.exception.type_error.312 | cannot use update() with string | The @ref update() member functions can only be executed for certain JSON types. -json.exception.type_error.313 | invalid value to unflatten | The @ref unflatten function converts an object whose keys are JSON Pointers back into an arbitrary nested JSON value. The JSON Pointers must not overlap, because then the resulting value would not be well defined. -json.exception.type_error.314 | only objects can be unflattened | The @ref unflatten function only works for an object whose keys are JSON Pointers. -json.exception.type_error.315 | values in object must be primitive | The @ref unflatten function only works for an object whose keys are JSON Pointers and whose values are primitive. -json.exception.type_error.316 | invalid UTF-8 byte at index 10: 0x7E | The @ref dump function only works with UTF-8 encoded strings; that is, if you assign a `std::string` to a JSON value, make sure it is UTF-8 encoded. | - -@liveexample{The following code shows how a `type_error` exception can be -caught.,type_error} - -@sa @ref exception for the base class of the library exceptions -@sa @ref parse_error for exceptions indicating a parse error -@sa @ref invalid_iterator for exceptions indicating errors with iterators -@sa @ref out_of_range for exceptions indicating access out of the defined range -@sa @ref other_error for exceptions indicating other library errors - -@since version 3.0.0 -*/ -class type_error : public exception -{ - public: - static type_error create(int id_, const std::string& what_arg) - { - std::string w = exception::name("type_error", id_) + what_arg; - return type_error(id_, w.c_str()); - } - - private: - type_error(int id_, const char* what_arg) : exception(id_, what_arg) {} -}; - -/*! -@brief exception indicating access out of the defined range - -This exception is thrown in case a library function is called on an input -parameter that exceeds the expected range, for instance in case of array -indices or nonexisting object keys. - -Exceptions have ids 4xx. - -name / id | example message | description -------------------------------- | --------------- | ------------------------- -json.exception.out_of_range.401 | array index 3 is out of range | The provided array index @a i is larger than @a size-1. -json.exception.out_of_range.402 | array index '-' (3) is out of range | The special array index `-` in a JSON Pointer never describes a valid element of the array, but the index past the end. That is, it can only be used to add elements at this position, but not to read it. -json.exception.out_of_range.403 | key 'foo' not found | The provided key was not found in the JSON object. -json.exception.out_of_range.404 | unresolved reference token 'foo' | A reference token in a JSON Pointer could not be resolved. -json.exception.out_of_range.405 | JSON pointer has no parent | The JSON Patch operations 'remove' and 'add' can not be applied to the root element of the JSON value. -json.exception.out_of_range.406 | number overflow parsing '10E1000' | A parsed number could not be stored as without changing it to NaN or INF. -json.exception.out_of_range.407 | number overflow serializing '9223372036854775808' | UBJSON only supports integers numbers up to 9223372036854775807. | -json.exception.out_of_range.408 | excessive array size: 8658170730974374167 | The size (following `#`) of an UBJSON array or object exceeds the maximal capacity. | - -@liveexample{The following code shows how an `out_of_range` exception can be -caught.,out_of_range} - -@sa @ref exception for the base class of the library exceptions -@sa @ref parse_error for exceptions indicating a parse error -@sa @ref invalid_iterator for exceptions indicating errors with iterators -@sa @ref type_error for exceptions indicating executing a member function with - a wrong type -@sa @ref other_error for exceptions indicating other library errors - -@since version 3.0.0 -*/ -class out_of_range : public exception -{ - public: - static out_of_range create(int id_, const std::string& what_arg) - { - std::string w = exception::name("out_of_range", id_) + what_arg; - return out_of_range(id_, w.c_str()); - } - - private: - out_of_range(int id_, const char* what_arg) : exception(id_, what_arg) {} -}; - -/*! -@brief exception indicating other library errors - -This exception is thrown in case of errors that cannot be classified with the -other exception types. - -Exceptions have ids 5xx. - -name / id | example message | description ------------------------------- | --------------- | ------------------------- -json.exception.other_error.501 | unsuccessful: {"op":"test","path":"/baz", "value":"bar"} | A JSON Patch operation 'test' failed. The unsuccessful operation is also printed. - -@sa @ref exception for the base class of the library exceptions -@sa @ref parse_error for exceptions indicating a parse error -@sa @ref invalid_iterator for exceptions indicating errors with iterators -@sa @ref type_error for exceptions indicating executing a member function with - a wrong type -@sa @ref out_of_range for exceptions indicating access out of the defined range - -@liveexample{The following code shows how an `other_error` exception can be -caught.,other_error} - -@since version 3.0.0 -*/ -class other_error : public exception -{ - public: - static other_error create(int id_, const std::string& what_arg) - { - std::string w = exception::name("other_error", id_) + what_arg; - return other_error(id_, w.c_str()); - } - - private: - other_error(int id_, const char* what_arg) : exception(id_, what_arg) {} -}; -} -} - -// #include <nlohmann/detail/value_t.hpp> - - -#include <array> // array -#include <ciso646> // and -#include <cstddef> // size_t -#include <cstdint> // uint8_t - -namespace nlohmann -{ -namespace detail -{ -/////////////////////////// -// JSON type enumeration // -/////////////////////////// - -/*! -@brief the JSON type enumeration - -This enumeration collects the different JSON types. It is internally used to -distinguish the stored values, and the functions @ref basic_json::is_null(), -@ref basic_json::is_object(), @ref basic_json::is_array(), -@ref basic_json::is_string(), @ref basic_json::is_boolean(), -@ref basic_json::is_number() (with @ref basic_json::is_number_integer(), -@ref basic_json::is_number_unsigned(), and @ref basic_json::is_number_float()), -@ref basic_json::is_discarded(), @ref basic_json::is_primitive(), and -@ref basic_json::is_structured() rely on it. - -@note There are three enumeration entries (number_integer, number_unsigned, and -number_float), because the library distinguishes these three types for numbers: -@ref basic_json::number_unsigned_t is used for unsigned integers, -@ref basic_json::number_integer_t is used for signed integers, and -@ref basic_json::number_float_t is used for floating-point numbers or to -approximate integers which do not fit in the limits of their respective type. - -@sa @ref basic_json::basic_json(const value_t value_type) -- create a JSON -value with the default value for a given type - -@since version 1.0.0 -*/ -enum class value_t : std::uint8_t -{ - null, ///< null value - object, ///< object (unordered set of name/value pairs) - array, ///< array (ordered collection of values) - string, ///< string value - boolean, ///< boolean value - number_integer, ///< number value (signed integer) - number_unsigned, ///< number value (unsigned integer) - number_float, ///< number value (floating-point) - discarded ///< discarded by the the parser callback function -}; - -/*! -@brief comparison operator for JSON types - -Returns an ordering that is similar to Python: -- order: null < boolean < number < object < array < string -- furthermore, each type is not smaller than itself -- discarded values are not comparable - -@since version 1.0.0 -*/ -inline bool operator<(const value_t lhs, const value_t rhs) noexcept -{ - static constexpr std::array<std::uint8_t, 8> order = {{ - 0 /* null */, 3 /* object */, 4 /* array */, 5 /* string */, - 1 /* boolean */, 2 /* integer */, 2 /* unsigned */, 2 /* float */ - } - }; - - const auto l_index = static_cast<std::size_t>(lhs); - const auto r_index = static_cast<std::size_t>(rhs); - return l_index < order.size() and r_index < order.size() and order[l_index] < order[r_index]; -} -} -} - -// #include <nlohmann/detail/conversions/from_json.hpp> - - -#include <algorithm> // transform -#include <array> // array -#include <ciso646> // and, not -#include <forward_list> // forward_list -#include <iterator> // inserter, front_inserter, end -#include <string> // string -#include <tuple> // tuple, make_tuple -#include <type_traits> // is_arithmetic, is_same, is_enum, underlying_type, is_convertible -#include <utility> // pair, declval -#include <valarray> // valarray - -// #include <nlohmann/detail/exceptions.hpp> - -// #include <nlohmann/detail/macro_scope.hpp> - -// #include <nlohmann/detail/meta.hpp> - -// #include <nlohmann/detail/value_t.hpp> - - -namespace nlohmann -{ -namespace detail -{ -// overloads for basic_json template parameters -template<typename BasicJsonType, typename ArithmeticType, - enable_if_t<std::is_arithmetic<ArithmeticType>::value and - not std::is_same<ArithmeticType, typename BasicJsonType::boolean_t>::value, - int> = 0> -void get_arithmetic_value(const BasicJsonType& j, ArithmeticType& val) -{ - switch (static_cast<value_t>(j)) - { - case value_t::number_unsigned: - { - val = static_cast<ArithmeticType>(*j.template get_ptr<const typename BasicJsonType::number_unsigned_t*>()); - break; - } - case value_t::number_integer: - { - val = static_cast<ArithmeticType>(*j.template get_ptr<const typename BasicJsonType::number_integer_t*>()); - break; - } - case value_t::number_float: - { - val = static_cast<ArithmeticType>(*j.template get_ptr<const typename BasicJsonType::number_float_t*>()); - break; - } - - default: - JSON_THROW(type_error::create(302, "type must be number, but is " + std::string(j.type_name()))); - } -} - -template<typename BasicJsonType> -void from_json(const BasicJsonType& j, typename BasicJsonType::boolean_t& b) -{ - if (JSON_UNLIKELY(not j.is_boolean())) - { - JSON_THROW(type_error::create(302, "type must be boolean, but is " + std::string(j.type_name()))); - } - b = *j.template get_ptr<const typename BasicJsonType::boolean_t*>(); -} - -template<typename BasicJsonType> -void from_json(const BasicJsonType& j, typename BasicJsonType::string_t& s) -{ - if (JSON_UNLIKELY(not j.is_string())) - { - JSON_THROW(type_error::create(302, "type must be string, but is " + std::string(j.type_name()))); - } - s = *j.template get_ptr<const typename BasicJsonType::string_t*>(); -} - -template<typename BasicJsonType> -void from_json(const BasicJsonType& j, typename BasicJsonType::number_float_t& val) -{ - get_arithmetic_value(j, val); -} - -template<typename BasicJsonType> -void from_json(const BasicJsonType& j, typename BasicJsonType::number_unsigned_t& val) -{ - get_arithmetic_value(j, val); -} - -template<typename BasicJsonType> -void from_json(const BasicJsonType& j, typename BasicJsonType::number_integer_t& val) -{ - get_arithmetic_value(j, val); -} - -template<typename BasicJsonType, typename EnumType, - enable_if_t<std::is_enum<EnumType>::value, int> = 0> -void from_json(const BasicJsonType& j, EnumType& e) -{ - typename std::underlying_type<EnumType>::type val; - get_arithmetic_value(j, val); - e = static_cast<EnumType>(val); -} - -template<typename BasicJsonType> -void from_json(const BasicJsonType& j, typename BasicJsonType::array_t& arr) -{ - if (JSON_UNLIKELY(not j.is_array())) - { - JSON_THROW(type_error::create(302, "type must be array, but is " + std::string(j.type_name()))); - } - arr = *j.template get_ptr<const typename BasicJsonType::array_t*>(); -} - -// forward_list doesn't have an insert method -template<typename BasicJsonType, typename T, typename Allocator, - enable_if_t<std::is_convertible<BasicJsonType, T>::value, int> = 0> -void from_json(const BasicJsonType& j, std::forward_list<T, Allocator>& l) -{ - if (JSON_UNLIKELY(not j.is_array())) - { - JSON_THROW(type_error::create(302, "type must be array, but is " + std::string(j.type_name()))); - } - std::transform(j.rbegin(), j.rend(), - std::front_inserter(l), [](const BasicJsonType & i) - { - return i.template get<T>(); - }); -} - -// valarray doesn't have an insert method -template<typename BasicJsonType, typename T, - enable_if_t<std::is_convertible<BasicJsonType, T>::value, int> = 0> -void from_json(const BasicJsonType& j, std::valarray<T>& l) -{ - if (JSON_UNLIKELY(not j.is_array())) - { - JSON_THROW(type_error::create(302, "type must be array, but is " + std::string(j.type_name()))); - } - l.resize(j.size()); - std::copy(j.m_value.array->begin(), j.m_value.array->end(), std::begin(l)); -} - -template<typename BasicJsonType, typename CompatibleArrayType> -void from_json_array_impl(const BasicJsonType& j, CompatibleArrayType& arr, priority_tag<0> /*unused*/) -{ - using std::end; - - std::transform(j.begin(), j.end(), - std::inserter(arr, end(arr)), [](const BasicJsonType & i) - { - // get<BasicJsonType>() returns *this, this won't call a from_json - // method when value_type is BasicJsonType - return i.template get<typename CompatibleArrayType::value_type>(); - }); -} - -template<typename BasicJsonType, typename CompatibleArrayType> -auto from_json_array_impl(const BasicJsonType& j, CompatibleArrayType& arr, priority_tag<1> /*unused*/) --> decltype( - arr.reserve(std::declval<typename CompatibleArrayType::size_type>()), - void()) -{ - using std::end; - - arr.reserve(j.size()); - std::transform(j.begin(), j.end(), - std::inserter(arr, end(arr)), [](const BasicJsonType & i) - { - // get<BasicJsonType>() returns *this, this won't call a from_json - // method when value_type is BasicJsonType - return i.template get<typename CompatibleArrayType::value_type>(); - }); -} - -template<typename BasicJsonType, typename T, std::size_t N> -void from_json_array_impl(const BasicJsonType& j, std::array<T, N>& arr, priority_tag<2> /*unused*/) -{ - for (std::size_t i = 0; i < N; ++i) - { - arr[i] = j.at(i).template get<T>(); - } -} - -template < - typename BasicJsonType, typename CompatibleArrayType, - enable_if_t < - is_compatible_array_type<BasicJsonType, CompatibleArrayType>::value and - not std::is_same<typename BasicJsonType::array_t, - CompatibleArrayType>::value and - std::is_constructible < - BasicJsonType, typename CompatibleArrayType::value_type >::value, - int > = 0 > -void from_json(const BasicJsonType& j, CompatibleArrayType& arr) -{ - if (JSON_UNLIKELY(not j.is_array())) - { - JSON_THROW(type_error::create(302, "type must be array, but is " + - std::string(j.type_name()))); - } - - from_json_array_impl(j, arr, priority_tag<2> {}); -} - -template<typename BasicJsonType, typename CompatibleObjectType, - enable_if_t<is_compatible_object_type<BasicJsonType, CompatibleObjectType>::value, int> = 0> -void from_json(const BasicJsonType& j, CompatibleObjectType& obj) -{ - if (JSON_UNLIKELY(not j.is_object())) - { - JSON_THROW(type_error::create(302, "type must be object, but is " + std::string(j.type_name()))); - } - - auto inner_object = j.template get_ptr<const typename BasicJsonType::object_t*>(); - using value_type = typename CompatibleObjectType::value_type; - std::transform( - inner_object->begin(), inner_object->end(), - std::inserter(obj, obj.begin()), - [](typename BasicJsonType::object_t::value_type const & p) - { - return value_type(p.first, p.second.template get<typename CompatibleObjectType::mapped_type>()); - }); -} - -// overload for arithmetic types, not chosen for basic_json template arguments -// (BooleanType, etc..); note: Is it really necessary to provide explicit -// overloads for boolean_t etc. in case of a custom BooleanType which is not -// an arithmetic type? -template<typename BasicJsonType, typename ArithmeticType, - enable_if_t < - std::is_arithmetic<ArithmeticType>::value and - not std::is_same<ArithmeticType, typename BasicJsonType::number_unsigned_t>::value and - not std::is_same<ArithmeticType, typename BasicJsonType::number_integer_t>::value and - not std::is_same<ArithmeticType, typename BasicJsonType::number_float_t>::value and - not std::is_same<ArithmeticType, typename BasicJsonType::boolean_t>::value, - int> = 0> -void from_json(const BasicJsonType& j, ArithmeticType& val) -{ - switch (static_cast<value_t>(j)) - { - case value_t::number_unsigned: - { - val = static_cast<ArithmeticType>(*j.template get_ptr<const typename BasicJsonType::number_unsigned_t*>()); - break; - } - case value_t::number_integer: - { - val = static_cast<ArithmeticType>(*j.template get_ptr<const typename BasicJsonType::number_integer_t*>()); - break; - } - case value_t::number_float: - { - val = static_cast<ArithmeticType>(*j.template get_ptr<const typename BasicJsonType::number_float_t*>()); - break; - } - case value_t::boolean: - { - val = static_cast<ArithmeticType>(*j.template get_ptr<const typename BasicJsonType::boolean_t*>()); - break; - } - - default: - JSON_THROW(type_error::create(302, "type must be number, but is " + std::string(j.type_name()))); - } -} - -template<typename BasicJsonType, typename A1, typename A2> -void from_json(const BasicJsonType& j, std::pair<A1, A2>& p) -{ - p = {j.at(0).template get<A1>(), j.at(1).template get<A2>()}; -} - -template<typename BasicJsonType, typename Tuple, std::size_t... Idx> -void from_json_tuple_impl(const BasicJsonType& j, Tuple& t, index_sequence<Idx...>) -{ - t = std::make_tuple(j.at(Idx).template get<typename std::tuple_element<Idx, Tuple>::type>()...); -} - -template<typename BasicJsonType, typename... Args> -void from_json(const BasicJsonType& j, std::tuple<Args...>& t) -{ - from_json_tuple_impl(j, t, index_sequence_for<Args...> {}); -} - -struct from_json_fn -{ - private: - template<typename BasicJsonType, typename T> - auto call(const BasicJsonType& j, T& val, priority_tag<1> /*unused*/) const - noexcept(noexcept(from_json(j, val))) - -> decltype(from_json(j, val), void()) - { - return from_json(j, val); - } - - template<typename BasicJsonType, typename T> - void call(const BasicJsonType& /*unused*/, T& /*unused*/, priority_tag<0> /*unused*/) const noexcept - { - static_assert(sizeof(BasicJsonType) == 0, - "could not find from_json() method in T's namespace"); -#ifdef _MSC_VER - // MSVC does not show a stacktrace for the above assert - using decayed = uncvref_t<T>; - static_assert(sizeof(typename decayed::force_msvc_stacktrace) == 0, - "forcing MSVC stacktrace to show which T we're talking about."); -#endif - } - - public: - template<typename BasicJsonType, typename T> - void operator()(const BasicJsonType& j, T& val) const - noexcept(noexcept(std::declval<from_json_fn>().call(j, val, priority_tag<1> {}))) - { - return call(j, val, priority_tag<1> {}); - } -}; -} - -/// namespace to hold default `from_json` function -/// to see why this is required: -/// http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2015/n4381.html -namespace -{ -constexpr const auto& from_json = detail::static_const<detail::from_json_fn>::value; -} -} - -// #include <nlohmann/detail/conversions/to_json.hpp> - - -#include <ciso646> // or, and, not -#include <iterator> // begin, end -#include <tuple> // tuple, get -#include <type_traits> // is_same, is_constructible, is_floating_point, is_enum, underlying_type -#include <utility> // move, forward, declval, pair -#include <valarray> // valarray -#include <vector> // vector - -// #include <nlohmann/detail/meta.hpp> - -// #include <nlohmann/detail/value_t.hpp> - - -namespace nlohmann -{ -namespace detail -{ -////////////////// -// constructors // -////////////////// - -template<value_t> struct external_constructor; - -template<> -struct external_constructor<value_t::boolean> -{ - template<typename BasicJsonType> - static void construct(BasicJsonType& j, typename BasicJsonType::boolean_t b) noexcept - { - j.m_type = value_t::boolean; - j.m_value = b; - j.assert_invariant(); - } -}; - -template<> -struct external_constructor<value_t::string> -{ - template<typename BasicJsonType> - static void construct(BasicJsonType& j, const typename BasicJsonType::string_t& s) - { - j.m_type = value_t::string; - j.m_value = s; - j.assert_invariant(); - } - - template<typename BasicJsonType> - static void construct(BasicJsonType& j, typename BasicJsonType::string_t&& s) - { - j.m_type = value_t::string; - j.m_value = std::move(s); - j.assert_invariant(); - } -}; - -template<> -struct external_constructor<value_t::number_float> -{ - template<typename BasicJsonType> - static void construct(BasicJsonType& j, typename BasicJsonType::number_float_t val) noexcept - { - j.m_type = value_t::number_float; - j.m_value = val; - j.assert_invariant(); - } -}; - -template<> -struct external_constructor<value_t::number_unsigned> -{ - template<typename BasicJsonType> - static void construct(BasicJsonType& j, typename BasicJsonType::number_unsigned_t val) noexcept - { - j.m_type = value_t::number_unsigned; - j.m_value = val; - j.assert_invariant(); - } -}; - -template<> -struct external_constructor<value_t::number_integer> -{ - template<typename BasicJsonType> - static void construct(BasicJsonType& j, typename BasicJsonType::number_integer_t val) noexcept - { - j.m_type = value_t::number_integer; - j.m_value = val; - j.assert_invariant(); - } -}; - -template<> -struct external_constructor<value_t::array> -{ - template<typename BasicJsonType> - static void construct(BasicJsonType& j, const typename BasicJsonType::array_t& arr) - { - j.m_type = value_t::array; - j.m_value = arr; - j.assert_invariant(); - } - - template<typename BasicJsonType> - static void construct(BasicJsonType& j, typename BasicJsonType::array_t&& arr) - { - j.m_type = value_t::array; - j.m_value = std::move(arr); - j.assert_invariant(); - } - - template<typename BasicJsonType, typename CompatibleArrayType, - enable_if_t<not std::is_same<CompatibleArrayType, typename BasicJsonType::array_t>::value, - int> = 0> - static void construct(BasicJsonType& j, const CompatibleArrayType& arr) - { - using std::begin; - using std::end; - j.m_type = value_t::array; - j.m_value.array = j.template create<typename BasicJsonType::array_t>(begin(arr), end(arr)); - j.assert_invariant(); - } - - template<typename BasicJsonType> - static void construct(BasicJsonType& j, const std::vector<bool>& arr) - { - j.m_type = value_t::array; - j.m_value = value_t::array; - j.m_value.array->reserve(arr.size()); - for (const bool x : arr) - { - j.m_value.array->push_back(x); - } - j.assert_invariant(); - } - - template<typename BasicJsonType, typename T, - enable_if_t<std::is_convertible<T, BasicJsonType>::value, int> = 0> - static void construct(BasicJsonType& j, const std::valarray<T>& arr) - { - j.m_type = value_t::array; - j.m_value = value_t::array; - j.m_value.array->resize(arr.size()); - std::copy(std::begin(arr), std::end(arr), j.m_value.array->begin()); - j.assert_invariant(); - } -}; - -template<> -struct external_constructor<value_t::object> -{ - template<typename BasicJsonType> - static void construct(BasicJsonType& j, const typename BasicJsonType::object_t& obj) - { - j.m_type = value_t::object; - j.m_value = obj; - j.assert_invariant(); - } - - template<typename BasicJsonType> - static void construct(BasicJsonType& j, typename BasicJsonType::object_t&& obj) - { - j.m_type = value_t::object; - j.m_value = std::move(obj); - j.assert_invariant(); - } - - template<typename BasicJsonType, typename CompatibleObjectType, - enable_if_t<not std::is_same<CompatibleObjectType, typename BasicJsonType::object_t>::value, int> = 0> - static void construct(BasicJsonType& j, const CompatibleObjectType& obj) - { - using std::begin; - using std::end; - - j.m_type = value_t::object; - j.m_value.object = j.template create<typename BasicJsonType::object_t>(begin(obj), end(obj)); - j.assert_invariant(); - } -}; - -///////////// -// to_json // -///////////// - -template<typename BasicJsonType, typename T, - enable_if_t<std::is_same<T, typename BasicJsonType::boolean_t>::value, int> = 0> -void to_json(BasicJsonType& j, T b) noexcept -{ - external_constructor<value_t::boolean>::construct(j, b); -} - -template<typename BasicJsonType, typename CompatibleString, - enable_if_t<std::is_constructible<typename BasicJsonType::string_t, CompatibleString>::value, int> = 0> -void to_json(BasicJsonType& j, const CompatibleString& s) -{ - external_constructor<value_t::string>::construct(j, s); -} - -template<typename BasicJsonType> -void to_json(BasicJsonType& j, typename BasicJsonType::string_t&& s) -{ - external_constructor<value_t::string>::construct(j, std::move(s)); -} - -template<typename BasicJsonType, typename FloatType, - enable_if_t<std::is_floating_point<FloatType>::value, int> = 0> -void to_json(BasicJsonType& j, FloatType val) noexcept -{ - external_constructor<value_t::number_float>::construct(j, static_cast<typename BasicJsonType::number_float_t>(val)); -} - -template<typename BasicJsonType, typename CompatibleNumberUnsignedType, - enable_if_t<is_compatible_integer_type<typename BasicJsonType::number_unsigned_t, CompatibleNumberUnsignedType>::value, int> = 0> -void to_json(BasicJsonType& j, CompatibleNumberUnsignedType val) noexcept -{ - external_constructor<value_t::number_unsigned>::construct(j, static_cast<typename BasicJsonType::number_unsigned_t>(val)); -} - -template<typename BasicJsonType, typename CompatibleNumberIntegerType, - enable_if_t<is_compatible_integer_type<typename BasicJsonType::number_integer_t, CompatibleNumberIntegerType>::value, int> = 0> -void to_json(BasicJsonType& j, CompatibleNumberIntegerType val) noexcept -{ - external_constructor<value_t::number_integer>::construct(j, static_cast<typename BasicJsonType::number_integer_t>(val)); -} - -template<typename BasicJsonType, typename EnumType, - enable_if_t<std::is_enum<EnumType>::value, int> = 0> -void to_json(BasicJsonType& j, EnumType e) noexcept -{ - using underlying_type = typename std::underlying_type<EnumType>::type; - external_constructor<value_t::number_integer>::construct(j, static_cast<underlying_type>(e)); -} - -template<typename BasicJsonType> -void to_json(BasicJsonType& j, const std::vector<bool>& e) -{ - external_constructor<value_t::array>::construct(j, e); -} - -template<typename BasicJsonType, typename CompatibleArrayType, - enable_if_t<is_compatible_array_type<BasicJsonType, CompatibleArrayType>::value or - std::is_same<typename BasicJsonType::array_t, CompatibleArrayType>::value, - int> = 0> -void to_json(BasicJsonType& j, const CompatibleArrayType& arr) -{ - external_constructor<value_t::array>::construct(j, arr); -} - -template<typename BasicJsonType, typename T, - enable_if_t<std::is_convertible<T, BasicJsonType>::value, int> = 0> -void to_json(BasicJsonType& j, std::valarray<T> arr) -{ - external_constructor<value_t::array>::construct(j, std::move(arr)); -} - -template<typename BasicJsonType> -void to_json(BasicJsonType& j, typename BasicJsonType::array_t&& arr) -{ - external_constructor<value_t::array>::construct(j, std::move(arr)); -} - -template<typename BasicJsonType, typename CompatibleObjectType, - enable_if_t<is_compatible_object_type<BasicJsonType, CompatibleObjectType>::value, int> = 0> -void to_json(BasicJsonType& j, const CompatibleObjectType& obj) -{ - external_constructor<value_t::object>::construct(j, obj); -} - -template<typename BasicJsonType> -void to_json(BasicJsonType& j, typename BasicJsonType::object_t&& obj) -{ - external_constructor<value_t::object>::construct(j, std::move(obj)); -} - -template<typename BasicJsonType, typename T, std::size_t N, - enable_if_t<not std::is_constructible<typename BasicJsonType::string_t, T (&)[N]>::value, int> = 0> -void to_json(BasicJsonType& j, T (&arr)[N]) -{ - external_constructor<value_t::array>::construct(j, arr); -} - -template<typename BasicJsonType, typename... Args> -void to_json(BasicJsonType& j, const std::pair<Args...>& p) -{ - j = {p.first, p.second}; -} - -template<typename BasicJsonType, typename Tuple, std::size_t... Idx> -void to_json_tuple_impl(BasicJsonType& j, const Tuple& t, index_sequence<Idx...>) -{ - j = {std::get<Idx>(t)...}; -} - -template<typename BasicJsonType, typename... Args> -void to_json(BasicJsonType& j, const std::tuple<Args...>& t) -{ - to_json_tuple_impl(j, t, index_sequence_for<Args...> {}); -} - -struct to_json_fn -{ - private: - template<typename BasicJsonType, typename T> - auto call(BasicJsonType& j, T&& val, priority_tag<1> /*unused*/) const noexcept(noexcept(to_json(j, std::forward<T>(val)))) - -> decltype(to_json(j, std::forward<T>(val)), void()) - { - return to_json(j, std::forward<T>(val)); - } - - template<typename BasicJsonType, typename T> - void call(BasicJsonType& /*unused*/, T&& /*unused*/, priority_tag<0> /*unused*/) const noexcept - { - static_assert(sizeof(BasicJsonType) == 0, - "could not find to_json() method in T's namespace"); - -#ifdef _MSC_VER - // MSVC does not show a stacktrace for the above assert - using decayed = uncvref_t<T>; - static_assert(sizeof(typename decayed::force_msvc_stacktrace) == 0, - "forcing MSVC stacktrace to show which T we're talking about."); -#endif - } - - public: - template<typename BasicJsonType, typename T> - void operator()(BasicJsonType& j, T&& val) const - noexcept(noexcept(std::declval<to_json_fn>().call(j, std::forward<T>(val), priority_tag<1> {}))) - { - return call(j, std::forward<T>(val), priority_tag<1> {}); - } -}; -} - -/// namespace to hold default `to_json` function -namespace -{ -constexpr const auto& to_json = detail::static_const<detail::to_json_fn>::value; -} -} - -// #include <nlohmann/detail/input/input_adapters.hpp> - - -#include <algorithm> // min -#include <array> // array -#include <cassert> // assert -#include <cstddef> // size_t -#include <cstring> // strlen -#include <ios> // streamsize, streamoff, streampos -#include <istream> // istream -#include <iterator> // begin, end, iterator_traits, random_access_iterator_tag, distance, next -#include <memory> // shared_ptr, make_shared, addressof -#include <numeric> // accumulate -#include <string> // string, char_traits -#include <type_traits> // enable_if, is_base_of, is_pointer, is_integral, remove_pointer -#include <utility> // pair, declval - -// #include <nlohmann/detail/macro_scope.hpp> - - -namespace nlohmann -{ -namespace detail -{ -//////////////////// -// input adapters // -//////////////////// - -/*! -@brief abstract input adapter interface - -Produces a stream of std::char_traits<char>::int_type characters from a -std::istream, a buffer, or some other input type. Accepts the return of exactly -one non-EOF character for future input. The int_type characters returned -consist of all valid char values as positive values (typically unsigned char), -plus an EOF value outside that range, specified by the value of the function -std::char_traits<char>::eof(). This value is typically -1, but could be any -arbitrary value which is not a valid char value. -*/ -struct input_adapter_protocol -{ - /// get a character [0,255] or std::char_traits<char>::eof(). - virtual std::char_traits<char>::int_type get_character() = 0; - /// restore the last non-eof() character to input - virtual void unget_character() = 0; - virtual ~input_adapter_protocol() = default; -}; - -/// a type to simplify interfaces -using input_adapter_t = std::shared_ptr<input_adapter_protocol>; - -/*! -Input adapter for a (caching) istream. Ignores a UFT Byte Order Mark at -beginning of input. Does not support changing the underlying std::streambuf -in mid-input. Maintains underlying std::istream and std::streambuf to support -subsequent use of standard std::istream operations to process any input -characters following those used in parsing the JSON input. Clears the -std::istream flags; any input errors (e.g., EOF) will be detected by the first -subsequent call for input from the std::istream. -*/ -class input_stream_adapter : public input_adapter_protocol -{ - public: - ~input_stream_adapter() override - { - // clear stream flags; we use underlying streambuf I/O, do not - // maintain ifstream flags - is.clear(); - } - - explicit input_stream_adapter(std::istream& i) - : is(i), sb(*i.rdbuf()) - { - // skip byte order mark - std::char_traits<char>::int_type c; - if ((c = get_character()) == 0xEF) - { - if ((c = get_character()) == 0xBB) - { - if ((c = get_character()) == 0xBF) - { - return; // Ignore BOM - } - else if (c != std::char_traits<char>::eof()) - { - is.unget(); - } - is.putback('\xBB'); - } - else if (c != std::char_traits<char>::eof()) - { - is.unget(); - } - is.putback('\xEF'); - } - else if (c != std::char_traits<char>::eof()) - { - is.unget(); // no byte order mark; process as usual - } - } - - // delete because of pointer members - input_stream_adapter(const input_stream_adapter&) = delete; - input_stream_adapter& operator=(input_stream_adapter&) = delete; - - // std::istream/std::streambuf use std::char_traits<char>::to_int_type, to - // ensure that std::char_traits<char>::eof() and the character 0xFF do not - // end up as the same value, eg. 0xFFFFFFFF. - std::char_traits<char>::int_type get_character() override - { - return sb.sbumpc(); - } - - void unget_character() override - { - sb.sungetc(); // is.unget() avoided for performance - } - - private: - /// the associated input stream - std::istream& is; - std::streambuf& sb; -}; - -/// input adapter for buffer input -class input_buffer_adapter : public input_adapter_protocol -{ - public: - input_buffer_adapter(const char* b, const std::size_t l) - : cursor(b), limit(b + l), start(b) - { - // skip byte order mark - if (l >= 3 and b[0] == '\xEF' and b[1] == '\xBB' and b[2] == '\xBF') - { - cursor += 3; - } - } - - // delete because of pointer members - input_buffer_adapter(const input_buffer_adapter&) = delete; - input_buffer_adapter& operator=(input_buffer_adapter&) = delete; - - std::char_traits<char>::int_type get_character() noexcept override - { - if (JSON_LIKELY(cursor < limit)) - { - return std::char_traits<char>::to_int_type(*(cursor++)); - } - - return std::char_traits<char>::eof(); - } - - void unget_character() noexcept override - { - if (JSON_LIKELY(cursor > start)) - { - --cursor; - } - } - - private: - /// pointer to the current character - const char* cursor; - /// pointer past the last character - const char* limit; - /// pointer to the first character - const char* start; -}; - -class input_adapter -{ - public: - // native support - - /// input adapter for input stream - input_adapter(std::istream& i) - : ia(std::make_shared<input_stream_adapter>(i)) {} - - /// input adapter for input stream - input_adapter(std::istream&& i) - : ia(std::make_shared<input_stream_adapter>(i)) {} - - /// input adapter for buffer - template<typename CharT, - typename std::enable_if< - std::is_pointer<CharT>::value and - std::is_integral<typename std::remove_pointer<CharT>::type>::value and - sizeof(typename std::remove_pointer<CharT>::type) == 1, - int>::type = 0> - input_adapter(CharT b, std::size_t l) - : ia(std::make_shared<input_buffer_adapter>(reinterpret_cast<const char*>(b), l)) {} - - // derived support - - /// input adapter for string literal - template<typename CharT, - typename std::enable_if< - std::is_pointer<CharT>::value and - std::is_integral<typename std::remove_pointer<CharT>::type>::value and - sizeof(typename std::remove_pointer<CharT>::type) == 1, - int>::type = 0> - input_adapter(CharT b) - : input_adapter(reinterpret_cast<const char*>(b), - std::strlen(reinterpret_cast<const char*>(b))) {} - - /// input adapter for iterator range with contiguous storage - template<class IteratorType, - typename std::enable_if< - std::is_same<typename std::iterator_traits<IteratorType>::iterator_category, std::random_access_iterator_tag>::value, - int>::type = 0> - input_adapter(IteratorType first, IteratorType last) - { - // assertion to check that the iterator range is indeed contiguous, - // see http://stackoverflow.com/a/35008842/266378 for more discussion - assert(std::accumulate( - first, last, std::pair<bool, int>(true, 0), - [&first](std::pair<bool, int> res, decltype(*first) val) - { - res.first &= (val == *(std::next(std::addressof(*first), res.second++))); - return res; - }).first); - - // assertion to check that each element is 1 byte long - static_assert( - sizeof(typename std::iterator_traits<IteratorType>::value_type) == 1, - "each element in the iterator range must have the size of 1 byte"); - - const auto len = static_cast<size_t>(std::distance(first, last)); - if (JSON_LIKELY(len > 0)) - { - // there is at least one element: use the address of first - ia = std::make_shared<input_buffer_adapter>(reinterpret_cast<const char*>(&(*first)), len); - } - else - { - // the address of first cannot be used: use nullptr - ia = std::make_shared<input_buffer_adapter>(nullptr, len); - } - } - - /// input adapter for array - template<class T, std::size_t N> - input_adapter(T (&array)[N]) - : input_adapter(std::begin(array), std::end(array)) {} - - /// input adapter for contiguous container - template<class ContiguousContainer, typename - std::enable_if<not std::is_pointer<ContiguousContainer>::value and - std::is_base_of<std::random_access_iterator_tag, typename std::iterator_traits<decltype(std::begin(std::declval<ContiguousContainer const>()))>::iterator_category>::value, - int>::type = 0> - input_adapter(const ContiguousContainer& c) - : input_adapter(std::begin(c), std::end(c)) {} - - operator input_adapter_t() - { - return ia; - } - - private: - /// the actual adapter - input_adapter_t ia = nullptr; -}; -} -} - -// #include <nlohmann/detail/input/lexer.hpp> - - -#include <clocale> // localeconv -#include <cstddef> // size_t -#include <cstdlib> // strtof, strtod, strtold, strtoll, strtoull -#include <initializer_list> // initializer_list -#include <ios> // hex, uppercase -#include <iomanip> // setw, setfill -#include <sstream> // stringstream -#include <string> // char_traits, string -#include <vector> // vector - -// #include <nlohmann/detail/macro_scope.hpp> - -// #include <nlohmann/detail/input/input_adapters.hpp> - - -namespace nlohmann -{ -namespace detail -{ -/////////// -// lexer // -/////////// - -/*! -@brief lexical analysis - -This class organizes the lexical analysis during JSON deserialization. -*/ -template<typename BasicJsonType> -class lexer -{ - using number_integer_t = typename BasicJsonType::number_integer_t; - using number_unsigned_t = typename BasicJsonType::number_unsigned_t; - using number_float_t = typename BasicJsonType::number_float_t; - using string_t = typename BasicJsonType::string_t; - - public: - /// token types for the parser - enum class token_type - { - uninitialized, ///< indicating the scanner is uninitialized - literal_true, ///< the `true` literal - literal_false, ///< the `false` literal - literal_null, ///< the `null` literal - value_string, ///< a string -- use get_string() for actual value - value_unsigned, ///< an unsigned integer -- use get_number_unsigned() for actual value - value_integer, ///< a signed integer -- use get_number_integer() for actual value - value_float, ///< an floating point number -- use get_number_float() for actual value - begin_array, ///< the character for array begin `[` - begin_object, ///< the character for object begin `{` - end_array, ///< the character for array end `]` - end_object, ///< the character for object end `}` - name_separator, ///< the name separator `:` - value_separator, ///< the value separator `,` - parse_error, ///< indicating a parse error - end_of_input, ///< indicating the end of the input buffer - literal_or_value ///< a literal or the begin of a value (only for diagnostics) - }; - - /// return name of values of type token_type (only used for errors) - static const char* token_type_name(const token_type t) noexcept - { - switch (t) - { - case token_type::uninitialized: - return "<uninitialized>"; - case token_type::literal_true: - return "true literal"; - case token_type::literal_false: - return "false literal"; - case token_type::literal_null: - return "null literal"; - case token_type::value_string: - return "string literal"; - case lexer::token_type::value_unsigned: - case lexer::token_type::value_integer: - case lexer::token_type::value_float: - return "number literal"; - case token_type::begin_array: - return "'['"; - case token_type::begin_object: - return "'{'"; - case token_type::end_array: - return "']'"; - case token_type::end_object: - return "'}'"; - case token_type::name_separator: - return "':'"; - case token_type::value_separator: - return "','"; - case token_type::parse_error: - return "<parse error>"; - case token_type::end_of_input: - return "end of input"; - case token_type::literal_or_value: - return "'[', '{', or a literal"; - default: // catch non-enum values - return "unknown token"; // LCOV_EXCL_LINE - } - } - - explicit lexer(detail::input_adapter_t adapter) - : ia(std::move(adapter)), decimal_point_char(get_decimal_point()) {} - - // delete because of pointer members - lexer(const lexer&) = delete; - lexer& operator=(lexer&) = delete; - - private: - ///////////////////// - // locales - ///////////////////// - - /// return the locale-dependent decimal point - static char get_decimal_point() noexcept - { - const auto loc = localeconv(); - assert(loc != nullptr); - return (loc->decimal_point == nullptr) ? '.' : *(loc->decimal_point); - } - - ///////////////////// - // scan functions - ///////////////////// - - /*! - @brief get codepoint from 4 hex characters following `\u` - - For input "\u c1 c2 c3 c4" the codepoint is: - (c1 * 0x1000) + (c2 * 0x0100) + (c3 * 0x0010) + c4 - = (c1 << 12) + (c2 << 8) + (c3 << 4) + (c4 << 0) - - Furthermore, the possible characters '0'..'9', 'A'..'F', and 'a'..'f' - must be converted to the integers 0x0..0x9, 0xA..0xF, 0xA..0xF, resp. The - conversion is done by subtracting the offset (0x30, 0x37, and 0x57) - between the ASCII value of the character and the desired integer value. - - @return codepoint (0x0000..0xFFFF) or -1 in case of an error (e.g. EOF or - non-hex character) - */ - int get_codepoint() - { - // this function only makes sense after reading `\u` - assert(current == 'u'); - int codepoint = 0; - - const auto factors = { 12, 8, 4, 0 }; - for (const auto factor : factors) - { - get(); - - if (current >= '0' and current <= '9') - { - codepoint += ((current - 0x30) << factor); - } - else if (current >= 'A' and current <= 'F') - { - codepoint += ((current - 0x37) << factor); - } - else if (current >= 'a' and current <= 'f') - { - codepoint += ((current - 0x57) << factor); - } - else - { - return -1; - } - } - - assert(0x0000 <= codepoint and codepoint <= 0xFFFF); - return codepoint; - } - - /*! - @brief check if the next byte(s) are inside a given range - - Adds the current byte and, for each passed range, reads a new byte and - checks if it is inside the range. If a violation was detected, set up an - error message and return false. Otherwise, return true. - - @param[in] ranges list of integers; interpreted as list of pairs of - inclusive lower and upper bound, respectively - - @pre The passed list @a ranges must have 2, 4, or 6 elements; that is, - 1, 2, or 3 pairs. This precondition is enforced by an assertion. - - @return true if and only if no range violation was detected - */ - bool next_byte_in_range(std::initializer_list<int> ranges) - { - assert(ranges.size() == 2 or ranges.size() == 4 or ranges.size() == 6); - add(current); - - for (auto range = ranges.begin(); range != ranges.end(); ++range) - { - get(); - if (JSON_LIKELY(*range <= current and current <= *(++range))) - { - add(current); - } - else - { - error_message = "invalid string: ill-formed UTF-8 byte"; - return false; - } - } - - return true; - } - - /*! - @brief scan a string literal - - This function scans a string according to Sect. 7 of RFC 7159. While - scanning, bytes are escaped and copied into buffer token_buffer. Then the - function returns successfully, token_buffer is *not* null-terminated (as it - may contain \0 bytes), and token_buffer.size() is the number of bytes in the - string. - - @return token_type::value_string if string could be successfully scanned, - token_type::parse_error otherwise - - @note In case of errors, variable error_message contains a textual - description. - */ - token_type scan_string() - { - // reset token_buffer (ignore opening quote) - reset(); - - // we entered the function by reading an open quote - assert(current == '\"'); - - while (true) - { - // get next character - switch (get()) - { - // end of file while parsing string - case std::char_traits<char>::eof(): - { - error_message = "invalid string: missing closing quote"; - return token_type::parse_error; - } - - // closing quote - case '\"': - { - return token_type::value_string; - } - - // escapes - case '\\': - { - switch (get()) - { - // quotation mark - case '\"': - add('\"'); - break; - // reverse solidus - case '\\': - add('\\'); - break; - // solidus - case '/': - add('/'); - break; - // backspace - case 'b': - add('\b'); - break; - // form feed - case 'f': - add('\f'); - break; - // line feed - case 'n': - add('\n'); - break; - // carriage return - case 'r': - add('\r'); - break; - // tab - case 't': - add('\t'); - break; - - // unicode escapes - case 'u': - { - const int codepoint1 = get_codepoint(); - int codepoint = codepoint1; // start with codepoint1 - - if (JSON_UNLIKELY(codepoint1 == -1)) - { - error_message = "invalid string: '\\u' must be followed by 4 hex digits"; - return token_type::parse_error; - } - - // check if code point is a high surrogate - if (0xD800 <= codepoint1 and codepoint1 <= 0xDBFF) - { - // expect next \uxxxx entry - if (JSON_LIKELY(get() == '\\' and get() == 'u')) - { - const int codepoint2 = get_codepoint(); - - if (JSON_UNLIKELY(codepoint2 == -1)) - { - error_message = "invalid string: '\\u' must be followed by 4 hex digits"; - return token_type::parse_error; - } - - // check if codepoint2 is a low surrogate - if (JSON_LIKELY(0xDC00 <= codepoint2 and codepoint2 <= 0xDFFF)) - { - // overwrite codepoint - codepoint = - // high surrogate occupies the most significant 22 bits - (codepoint1 << 10) - // low surrogate occupies the least significant 15 bits - + codepoint2 - // there is still the 0xD800, 0xDC00 and 0x10000 noise - // in the result so we have to subtract with: - // (0xD800 << 10) + DC00 - 0x10000 = 0x35FDC00 - - 0x35FDC00; - } - else - { - error_message = "invalid string: surrogate U+DC00..U+DFFF must be followed by U+DC00..U+DFFF"; - return token_type::parse_error; - } - } - else - { - error_message = "invalid string: surrogate U+DC00..U+DFFF must be followed by U+DC00..U+DFFF"; - return token_type::parse_error; - } - } - else - { - if (JSON_UNLIKELY(0xDC00 <= codepoint1 and codepoint1 <= 0xDFFF)) - { - error_message = "invalid string: surrogate U+DC00..U+DFFF must follow U+D800..U+DBFF"; - return token_type::parse_error; - } - } - - // result of the above calculation yields a proper codepoint - assert(0x00 <= codepoint and codepoint <= 0x10FFFF); - - // translate codepoint into bytes - if (codepoint < 0x80) - { - // 1-byte characters: 0xxxxxxx (ASCII) - add(codepoint); - } - else if (codepoint <= 0x7FF) - { - // 2-byte characters: 110xxxxx 10xxxxxx - add(0xC0 | (codepoint >> 6)); - add(0x80 | (codepoint & 0x3F)); - } - else if (codepoint <= 0xFFFF) - { - // 3-byte characters: 1110xxxx 10xxxxxx 10xxxxxx - add(0xE0 | (codepoint >> 12)); - add(0x80 | ((codepoint >> 6) & 0x3F)); - add(0x80 | (codepoint & 0x3F)); - } - else - { - // 4-byte characters: 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx - add(0xF0 | (codepoint >> 18)); - add(0x80 | ((codepoint >> 12) & 0x3F)); - add(0x80 | ((codepoint >> 6) & 0x3F)); - add(0x80 | (codepoint & 0x3F)); - } - - break; - } - - // other characters after escape - default: - error_message = "invalid string: forbidden character after backslash"; - return token_type::parse_error; - } - - break; - } - - // invalid control characters - case 0x00: - case 0x01: - case 0x02: - case 0x03: - case 0x04: - case 0x05: - case 0x06: - case 0x07: - case 0x08: - case 0x09: - case 0x0A: - case 0x0B: - case 0x0C: - case 0x0D: - case 0x0E: - case 0x0F: - case 0x10: - case 0x11: - case 0x12: - case 0x13: - case 0x14: - case 0x15: - case 0x16: - case 0x17: - case 0x18: - case 0x19: - case 0x1A: - case 0x1B: - case 0x1C: - case 0x1D: - case 0x1E: - case 0x1F: - { - error_message = "invalid string: control character must be escaped"; - return token_type::parse_error; - } - - // U+0020..U+007F (except U+0022 (quote) and U+005C (backspace)) - case 0x20: - case 0x21: - case 0x23: - case 0x24: - case 0x25: - case 0x26: - case 0x27: - case 0x28: - case 0x29: - case 0x2A: - case 0x2B: - case 0x2C: - case 0x2D: - case 0x2E: - case 0x2F: - case 0x30: - case 0x31: - case 0x32: - case 0x33: - case 0x34: - case 0x35: - case 0x36: - case 0x37: - case 0x38: - case 0x39: - case 0x3A: - case 0x3B: - case 0x3C: - case 0x3D: - case 0x3E: - case 0x3F: - case 0x40: - case 0x41: - case 0x42: - case 0x43: - case 0x44: - case 0x45: - case 0x46: - case 0x47: - case 0x48: - case 0x49: - case 0x4A: - case 0x4B: - case 0x4C: - case 0x4D: - case 0x4E: - case 0x4F: - case 0x50: - case 0x51: - case 0x52: - case 0x53: - case 0x54: - case 0x55: - case 0x56: - case 0x57: - case 0x58: - case 0x59: - case 0x5A: - case 0x5B: - case 0x5D: - case 0x5E: - case 0x5F: - case 0x60: - case 0x61: - case 0x62: - case 0x63: - case 0x64: - case 0x65: - case 0x66: - case 0x67: - case 0x68: - case 0x69: - case 0x6A: - case 0x6B: - case 0x6C: - case 0x6D: - case 0x6E: - case 0x6F: - case 0x70: - case 0x71: - case 0x72: - case 0x73: - case 0x74: - case 0x75: - case 0x76: - case 0x77: - case 0x78: - case 0x79: - case 0x7A: - case 0x7B: - case 0x7C: - case 0x7D: - case 0x7E: - case 0x7F: - { - add(current); - break; - } - - // U+0080..U+07FF: bytes C2..DF 80..BF - case 0xC2: - case 0xC3: - case 0xC4: - case 0xC5: - case 0xC6: - case 0xC7: - case 0xC8: - case 0xC9: - case 0xCA: - case 0xCB: - case 0xCC: - case 0xCD: - case 0xCE: - case 0xCF: - case 0xD0: - case 0xD1: - case 0xD2: - case 0xD3: - case 0xD4: - case 0xD5: - case 0xD6: - case 0xD7: - case 0xD8: - case 0xD9: - case 0xDA: - case 0xDB: - case 0xDC: - case 0xDD: - case 0xDE: - case 0xDF: - { - if (JSON_UNLIKELY(not next_byte_in_range({0x80, 0xBF}))) - { - return token_type::parse_error; - } - break; - } - - // U+0800..U+0FFF: bytes E0 A0..BF 80..BF - case 0xE0: - { - if (JSON_UNLIKELY(not (next_byte_in_range({0xA0, 0xBF, 0x80, 0xBF})))) - { - return token_type::parse_error; - } - break; - } - - // U+1000..U+CFFF: bytes E1..EC 80..BF 80..BF - // U+E000..U+FFFF: bytes EE..EF 80..BF 80..BF - case 0xE1: - case 0xE2: - case 0xE3: - case 0xE4: - case 0xE5: - case 0xE6: - case 0xE7: - case 0xE8: - case 0xE9: - case 0xEA: - case 0xEB: - case 0xEC: - case 0xEE: - case 0xEF: - { - if (JSON_UNLIKELY(not (next_byte_in_range({0x80, 0xBF, 0x80, 0xBF})))) - { - return token_type::parse_error; - } - break; - } - - // U+D000..U+D7FF: bytes ED 80..9F 80..BF - case 0xED: - { - if (JSON_UNLIKELY(not (next_byte_in_range({0x80, 0x9F, 0x80, 0xBF})))) - { - return token_type::parse_error; - } - break; - } - - // U+10000..U+3FFFF F0 90..BF 80..BF 80..BF - case 0xF0: - { - if (JSON_UNLIKELY(not (next_byte_in_range({0x90, 0xBF, 0x80, 0xBF, 0x80, 0xBF})))) - { - return token_type::parse_error; - } - break; - } - - // U+40000..U+FFFFF F1..F3 80..BF 80..BF 80..BF - case 0xF1: - case 0xF2: - case 0xF3: - { - if (JSON_UNLIKELY(not (next_byte_in_range({0x80, 0xBF, 0x80, 0xBF, 0x80, 0xBF})))) - { - return token_type::parse_error; - } - break; - } - - // U+100000..U+10FFFF F4 80..8F 80..BF 80..BF - case 0xF4: - { - if (JSON_UNLIKELY(not (next_byte_in_range({0x80, 0x8F, 0x80, 0xBF, 0x80, 0xBF})))) - { - return token_type::parse_error; - } - break; - } - - // remaining bytes (80..C1 and F5..FF) are ill-formed - default: - { - error_message = "invalid string: ill-formed UTF-8 byte"; - return token_type::parse_error; - } - } - } - } - - static void strtof(float& f, const char* str, char** endptr) noexcept - { - f = std::strtof(str, endptr); - } - - static void strtof(double& f, const char* str, char** endptr) noexcept - { - f = std::strtod(str, endptr); - } - - static void strtof(long double& f, const char* str, char** endptr) noexcept - { - f = std::strtold(str, endptr); - } - - /*! - @brief scan a number literal - - This function scans a string according to Sect. 6 of RFC 7159. - - The function is realized with a deterministic finite state machine derived - from the grammar described in RFC 7159. Starting in state "init", the - input is read and used to determined the next state. Only state "done" - accepts the number. State "error" is a trap state to model errors. In the - table below, "anything" means any character but the ones listed before. - - state | 0 | 1-9 | e E | + | - | . | anything - ---------|----------|----------|----------|---------|---------|----------|----------- - init | zero | any1 | [error] | [error] | minus | [error] | [error] - minus | zero | any1 | [error] | [error] | [error] | [error] | [error] - zero | done | done | exponent | done | done | decimal1 | done - any1 | any1 | any1 | exponent | done | done | decimal1 | done - decimal1 | decimal2 | [error] | [error] | [error] | [error] | [error] | [error] - decimal2 | decimal2 | decimal2 | exponent | done | done | done | done - exponent | any2 | any2 | [error] | sign | sign | [error] | [error] - sign | any2 | any2 | [error] | [error] | [error] | [error] | [error] - any2 | any2 | any2 | done | done | done | done | done - - The state machine is realized with one label per state (prefixed with - "scan_number_") and `goto` statements between them. The state machine - contains cycles, but any cycle can be left when EOF is read. Therefore, - the function is guaranteed to terminate. - - During scanning, the read bytes are stored in token_buffer. This string is - then converted to a signed integer, an unsigned integer, or a - floating-point number. - - @return token_type::value_unsigned, token_type::value_integer, or - token_type::value_float if number could be successfully scanned, - token_type::parse_error otherwise - - @note The scanner is independent of the current locale. Internally, the - locale's decimal point is used instead of `.` to work with the - locale-dependent converters. - */ - token_type scan_number() - { - // reset token_buffer to store the number's bytes - reset(); - - // the type of the parsed number; initially set to unsigned; will be - // changed if minus sign, decimal point or exponent is read - token_type number_type = token_type::value_unsigned; - - // state (init): we just found out we need to scan a number - switch (current) - { - case '-': - { - add(current); - goto scan_number_minus; - } - - case '0': - { - add(current); - goto scan_number_zero; - } - - case '1': - case '2': - case '3': - case '4': - case '5': - case '6': - case '7': - case '8': - case '9': - { - add(current); - goto scan_number_any1; - } - - default: - { - // all other characters are rejected outside scan_number() - assert(false); // LCOV_EXCL_LINE - } - } - -scan_number_minus: - // state: we just parsed a leading minus sign - number_type = token_type::value_integer; - switch (get()) - { - case '0': - { - add(current); - goto scan_number_zero; - } - - case '1': - case '2': - case '3': - case '4': - case '5': - case '6': - case '7': - case '8': - case '9': - { - add(current); - goto scan_number_any1; - } - - default: - { - error_message = "invalid number; expected digit after '-'"; - return token_type::parse_error; - } - } - -scan_number_zero: - // state: we just parse a zero (maybe with a leading minus sign) - switch (get()) - { - case '.': - { - add(decimal_point_char); - goto scan_number_decimal1; - } - - case 'e': - case 'E': - { - add(current); - goto scan_number_exponent; - } - - default: - goto scan_number_done; - } - -scan_number_any1: - // state: we just parsed a number 0-9 (maybe with a leading minus sign) - switch (get()) - { - case '0': - case '1': - case '2': - case '3': - case '4': - case '5': - case '6': - case '7': - case '8': - case '9': - { - add(current); - goto scan_number_any1; - } - - case '.': - { - add(decimal_point_char); - goto scan_number_decimal1; - } - - case 'e': - case 'E': - { - add(current); - goto scan_number_exponent; - } - - default: - goto scan_number_done; - } - -scan_number_decimal1: - // state: we just parsed a decimal point - number_type = token_type::value_float; - switch (get()) - { - case '0': - case '1': - case '2': - case '3': - case '4': - case '5': - case '6': - case '7': - case '8': - case '9': - { - add(current); - goto scan_number_decimal2; - } - - default: - { - error_message = "invalid number; expected digit after '.'"; - return token_type::parse_error; - } - } - -scan_number_decimal2: - // we just parsed at least one number after a decimal point - switch (get()) - { - case '0': - case '1': - case '2': - case '3': - case '4': - case '5': - case '6': - case '7': - case '8': - case '9': - { - add(current); - goto scan_number_decimal2; - } - - case 'e': - case 'E': - { - add(current); - goto scan_number_exponent; - } - - default: - goto scan_number_done; - } - -scan_number_exponent: - // we just parsed an exponent - number_type = token_type::value_float; - switch (get()) - { - case '+': - case '-': - { - add(current); - goto scan_number_sign; - } - - case '0': - case '1': - case '2': - case '3': - case '4': - case '5': - case '6': - case '7': - case '8': - case '9': - { - add(current); - goto scan_number_any2; - } - - default: - { - error_message = - "invalid number; expected '+', '-', or digit after exponent"; - return token_type::parse_error; - } - } - -scan_number_sign: - // we just parsed an exponent sign - switch (get()) - { - case '0': - case '1': - case '2': - case '3': - case '4': - case '5': - case '6': - case '7': - case '8': - case '9': - { - add(current); - goto scan_number_any2; - } - - default: - { - error_message = "invalid number; expected digit after exponent sign"; - return token_type::parse_error; - } - } - -scan_number_any2: - // we just parsed a number after the exponent or exponent sign - switch (get()) - { - case '0': - case '1': - case '2': - case '3': - case '4': - case '5': - case '6': - case '7': - case '8': - case '9': - { - add(current); - goto scan_number_any2; - } - - default: - goto scan_number_done; - } - -scan_number_done: - // unget the character after the number (we only read it to know that - // we are done scanning a number) - unget(); - - char* endptr = nullptr; - errno = 0; - - // try to parse integers first and fall back to floats - if (number_type == token_type::value_unsigned) - { - const auto x = std::strtoull(token_buffer.data(), &endptr, 10); - - // we checked the number format before - assert(endptr == token_buffer.data() + token_buffer.size()); - - if (errno == 0) - { - value_unsigned = static_cast<number_unsigned_t>(x); - if (value_unsigned == x) - { - return token_type::value_unsigned; - } - } - } - else if (number_type == token_type::value_integer) - { - const auto x = std::strtoll(token_buffer.data(), &endptr, 10); - - // we checked the number format before - assert(endptr == token_buffer.data() + token_buffer.size()); - - if (errno == 0) - { - value_integer = static_cast<number_integer_t>(x); - if (value_integer == x) - { - return token_type::value_integer; - } - } - } - - // this code is reached if we parse a floating-point number or if an - // integer conversion above failed - strtof(value_float, token_buffer.data(), &endptr); - - // we checked the number format before - assert(endptr == token_buffer.data() + token_buffer.size()); - - return token_type::value_float; - } - - /*! - @param[in] literal_text the literal text to expect - @param[in] length the length of the passed literal text - @param[in] return_type the token type to return on success - */ - token_type scan_literal(const char* literal_text, const std::size_t length, - token_type return_type) - { - assert(current == literal_text[0]); - for (std::size_t i = 1; i < length; ++i) - { - if (JSON_UNLIKELY(get() != literal_text[i])) - { - error_message = "invalid literal"; - return token_type::parse_error; - } - } - return return_type; - } - - ///////////////////// - // input management - ///////////////////// - - /// reset token_buffer; current character is beginning of token - void reset() noexcept - { - token_buffer.clear(); - token_string.clear(); - token_string.push_back(std::char_traits<char>::to_char_type(current)); - } - - /* - @brief get next character from the input - - This function provides the interface to the used input adapter. It does - not throw in case the input reached EOF, but returns a - `std::char_traits<char>::eof()` in that case. Stores the scanned characters - for use in error messages. - - @return character read from the input - */ - std::char_traits<char>::int_type get() - { - ++chars_read; - current = ia->get_character(); - if (JSON_LIKELY(current != std::char_traits<char>::eof())) - { - token_string.push_back(std::char_traits<char>::to_char_type(current)); - } - return current; - } - - /// unget current character (return it again on next get) - void unget() - { - --chars_read; - if (JSON_LIKELY(current != std::char_traits<char>::eof())) - { - ia->unget_character(); - assert(token_string.size() != 0); - token_string.pop_back(); - } - } - - /// add a character to token_buffer - void add(int c) - { - token_buffer.push_back(std::char_traits<char>::to_char_type(c)); - } - - public: - ///////////////////// - // value getters - ///////////////////// - - /// return integer value - constexpr number_integer_t get_number_integer() const noexcept - { - return value_integer; - } - - /// return unsigned integer value - constexpr number_unsigned_t get_number_unsigned() const noexcept - { - return value_unsigned; - } - - /// return floating-point value - constexpr number_float_t get_number_float() const noexcept - { - return value_float; - } - - /// return current string value (implicitly resets the token; useful only once) - string_t&& move_string() - { - return std::move(token_buffer); - } - - ///////////////////// - // diagnostics - ///////////////////// - - /// return position of last read token - constexpr std::size_t get_position() const noexcept - { - return chars_read; - } - - /// return the last read token (for errors only). Will never contain EOF - /// (an arbitrary value that is not a valid char value, often -1), because - /// 255 may legitimately occur. May contain NUL, which should be escaped. - std::string get_token_string() const - { - // escape control characters - std::string result; - for (const auto c : token_string) - { - if ('\x00' <= c and c <= '\x1F') - { - // escape control characters - std::stringstream ss; - ss << "<U+" << std::setw(4) << std::uppercase << std::setfill('0') - << std::hex << static_cast<int>(c) << ">"; - result += ss.str(); - } - else - { - // add character as is - result.push_back(c); - } - } - - return result; - } - - /// return syntax error message - constexpr const char* get_error_message() const noexcept - { - return error_message; - } - - ///////////////////// - // actual scanner - ///////////////////// - - token_type scan() - { - // read next character and ignore whitespace - do - { - get(); - } - while (current == ' ' or current == '\t' or current == '\n' or current == '\r'); - - switch (current) - { - // structural characters - case '[': - return token_type::begin_array; - case ']': - return token_type::end_array; - case '{': - return token_type::begin_object; - case '}': - return token_type::end_object; - case ':': - return token_type::name_separator; - case ',': - return token_type::value_separator; - - // literals - case 't': - return scan_literal("true", 4, token_type::literal_true); - case 'f': - return scan_literal("false", 5, token_type::literal_false); - case 'n': - return scan_literal("null", 4, token_type::literal_null); - - // string - case '\"': - return scan_string(); - - // number - case '-': - case '0': - case '1': - case '2': - case '3': - case '4': - case '5': - case '6': - case '7': - case '8': - case '9': - return scan_number(); - - // end of input (the null byte is needed when parsing from - // string literals) - case '\0': - case std::char_traits<char>::eof(): - return token_type::end_of_input; - - // error - default: - error_message = "invalid literal"; - return token_type::parse_error; - } - } - - private: - /// input adapter - detail::input_adapter_t ia = nullptr; - - /// the current character - std::char_traits<char>::int_type current = std::char_traits<char>::eof(); - - /// the number of characters read - std::size_t chars_read = 0; - - /// raw input token string (for error messages) - std::vector<char> token_string {}; - - /// buffer for variable-length tokens (numbers, strings) - string_t token_buffer {}; - - /// a description of occurred lexer errors - const char* error_message = ""; - - // number values - number_integer_t value_integer = 0; - number_unsigned_t value_unsigned = 0; - number_float_t value_float = 0; - - /// the decimal point - const char decimal_point_char = '.'; -}; -} -} - -// #include <nlohmann/detail/input/parser.hpp> - - -#include <cassert> // assert -#include <cmath> // isfinite -#include <cstdint> // uint8_t -#include <functional> // function -#include <string> // string -#include <utility> // move - -// #include <nlohmann/detail/exceptions.hpp> - -// #include <nlohmann/detail/macro_scope.hpp> - -// #include <nlohmann/detail/input/input_adapters.hpp> - -// #include <nlohmann/detail/input/lexer.hpp> - -// #include <nlohmann/detail/value_t.hpp> - - -namespace nlohmann -{ -namespace detail -{ -//////////// -// parser // -//////////// - -/*! -@brief syntax analysis - -This class implements a recursive decent parser. -*/ -template<typename BasicJsonType> -class parser -{ - using number_integer_t = typename BasicJsonType::number_integer_t; - using number_unsigned_t = typename BasicJsonType::number_unsigned_t; - using number_float_t = typename BasicJsonType::number_float_t; - using string_t = typename BasicJsonType::string_t; - using lexer_t = lexer<BasicJsonType>; - using token_type = typename lexer_t::token_type; - - public: - enum class parse_event_t : uint8_t - { - /// the parser read `{` and started to process a JSON object - object_start, - /// the parser read `}` and finished processing a JSON object - object_end, - /// the parser read `[` and started to process a JSON array - array_start, - /// the parser read `]` and finished processing a JSON array - array_end, - /// the parser read a key of a value in an object - key, - /// the parser finished reading a JSON value - value - }; - - using parser_callback_t = - std::function<bool(int depth, parse_event_t event, BasicJsonType& parsed)>; - - /// a parser reading from an input adapter - explicit parser(detail::input_adapter_t adapter, - const parser_callback_t cb = nullptr, - const bool allow_exceptions_ = true) - : callback(cb), m_lexer(adapter), allow_exceptions(allow_exceptions_) - {} - - /*! - @brief public parser interface - - @param[in] strict whether to expect the last token to be EOF - @param[in,out] result parsed JSON value - - @throw parse_error.101 in case of an unexpected token - @throw parse_error.102 if to_unicode fails or surrogate error - @throw parse_error.103 if to_unicode fails - */ - void parse(const bool strict, BasicJsonType& result) - { - // read first token - get_token(); - - parse_internal(true, result); - result.assert_invariant(); - - // in strict mode, input must be completely read - if (strict) - { - get_token(); - expect(token_type::end_of_input); - } - - // in case of an error, return discarded value - if (errored) - { - result = value_t::discarded; - return; - } - - // set top-level value to null if it was discarded by the callback - // function - if (result.is_discarded()) - { - result = nullptr; - } - } - - /*! - @brief public accept interface - - @param[in] strict whether to expect the last token to be EOF - @return whether the input is a proper JSON text - */ - bool accept(const bool strict = true) - { - // read first token - get_token(); - - if (not accept_internal()) - { - return false; - } - - // strict => last token must be EOF - return not strict or (get_token() == token_type::end_of_input); - } - - private: - /*! - @brief the actual parser - @throw parse_error.101 in case of an unexpected token - @throw parse_error.102 if to_unicode fails or surrogate error - @throw parse_error.103 if to_unicode fails - */ - void parse_internal(bool keep, BasicJsonType& result) - { - // never parse after a parse error was detected - assert(not errored); - - // start with a discarded value - if (not result.is_discarded()) - { - result.m_value.destroy(result.m_type); - result.m_type = value_t::discarded; - } - - switch (last_token) - { - case token_type::begin_object: - { - if (keep) - { - if (callback) - { - keep = callback(depth++, parse_event_t::object_start, result); - } - - if (not callback or keep) - { - // explicitly set result to object to cope with {} - result.m_type = value_t::object; - result.m_value = value_t::object; - } - } - - // read next token - get_token(); - - // closing } -> we are done - if (last_token == token_type::end_object) - { - if (keep and callback and not callback(--depth, parse_event_t::object_end, result)) - { - result.m_value.destroy(result.m_type); - result.m_type = value_t::discarded; - } - break; - } - - // parse values - string_t key; - BasicJsonType value; - while (true) - { - // store key - if (not expect(token_type::value_string)) - { - return; - } - key = m_lexer.move_string(); - - bool keep_tag = false; - if (keep) - { - if (callback) - { - BasicJsonType k(key); - keep_tag = callback(depth, parse_event_t::key, k); - } - else - { - keep_tag = true; - } - } - - // parse separator (:) - get_token(); - if (not expect(token_type::name_separator)) - { - return; - } - - // parse and add value - get_token(); - value.m_value.destroy(value.m_type); - value.m_type = value_t::discarded; - parse_internal(keep, value); - - if (JSON_UNLIKELY(errored)) - { - return; - } - - if (keep and keep_tag and not value.is_discarded()) - { - result.m_value.object->emplace(std::move(key), std::move(value)); - } - - // comma -> next value - get_token(); - if (last_token == token_type::value_separator) - { - get_token(); - continue; - } - - // closing } - if (not expect(token_type::end_object)) - { - return; - } - break; - } - - if (keep and callback and not callback(--depth, parse_event_t::object_end, result)) - { - result.m_value.destroy(result.m_type); - result.m_type = value_t::discarded; - } - break; - } - - case token_type::begin_array: - { - if (keep) - { - if (callback) - { - keep = callback(depth++, parse_event_t::array_start, result); - } - - if (not callback or keep) - { - // explicitly set result to array to cope with [] - result.m_type = value_t::array; - result.m_value = value_t::array; - } - } - - // read next token - get_token(); - - // closing ] -> we are done - if (last_token == token_type::end_array) - { - if (callback and not callback(--depth, parse_event_t::array_end, result)) - { - result.m_value.destroy(result.m_type); - result.m_type = value_t::discarded; - } - break; - } - - // parse values - BasicJsonType value; - while (true) - { - // parse value - value.m_value.destroy(value.m_type); - value.m_type = value_t::discarded; - parse_internal(keep, value); - - if (JSON_UNLIKELY(errored)) - { - return; - } - - if (keep and not value.is_discarded()) - { - result.m_value.array->push_back(std::move(value)); - } - - // comma -> next value - get_token(); - if (last_token == token_type::value_separator) - { - get_token(); - continue; - } - - // closing ] - if (not expect(token_type::end_array)) - { - return; - } - break; - } - - if (keep and callback and not callback(--depth, parse_event_t::array_end, result)) - { - result.m_value.destroy(result.m_type); - result.m_type = value_t::discarded; - } - break; - } - - case token_type::literal_null: - { - result.m_type = value_t::null; - break; - } - - case token_type::value_string: - { - result.m_type = value_t::string; - result.m_value = m_lexer.move_string(); - break; - } - - case token_type::literal_true: - { - result.m_type = value_t::boolean; - result.m_value = true; - break; - } - - case token_type::literal_false: - { - result.m_type = value_t::boolean; - result.m_value = false; - break; - } - - case token_type::value_unsigned: - { - result.m_type = value_t::number_unsigned; - result.m_value = m_lexer.get_number_unsigned(); - break; - } - - case token_type::value_integer: - { - result.m_type = value_t::number_integer; - result.m_value = m_lexer.get_number_integer(); - break; - } - - case token_type::value_float: - { - result.m_type = value_t::number_float; - result.m_value = m_lexer.get_number_float(); - - // throw in case of infinity or NAN - if (JSON_UNLIKELY(not std::isfinite(result.m_value.number_float))) - { - if (allow_exceptions) - { - JSON_THROW(out_of_range::create(406, "number overflow parsing '" + - m_lexer.get_token_string() + "'")); - } - expect(token_type::uninitialized); - } - break; - } - - case token_type::parse_error: - { - // using "uninitialized" to avoid "expected" message - if (not expect(token_type::uninitialized)) - { - return; - } - break; // LCOV_EXCL_LINE - } - - default: - { - // the last token was unexpected; we expected a value - if (not expect(token_type::literal_or_value)) - { - return; - } - break; // LCOV_EXCL_LINE - } - } - - if (keep and callback and not callback(depth, parse_event_t::value, result)) - { - result.m_value.destroy(result.m_type); - result.m_type = value_t::discarded; - } - } - - /*! - @brief the actual acceptor - - @invariant 1. The last token is not yet processed. Therefore, the caller - of this function must make sure a token has been read. - 2. When this function returns, the last token is processed. - That is, the last read character was already considered. - - This invariant makes sure that no token needs to be "unput". - */ - bool accept_internal() - { - switch (last_token) - { - case token_type::begin_object: - { - // read next token - get_token(); - - // closing } -> we are done - if (last_token == token_type::end_object) - { - return true; - } - - // parse values - while (true) - { - // parse key - if (last_token != token_type::value_string) - { - return false; - } - - // parse separator (:) - get_token(); - if (last_token != token_type::name_separator) - { - return false; - } - - // parse value - get_token(); - if (not accept_internal()) - { - return false; - } - - // comma -> next value - get_token(); - if (last_token == token_type::value_separator) - { - get_token(); - continue; - } - - // closing } - return (last_token == token_type::end_object); - } - } - - case token_type::begin_array: - { - // read next token - get_token(); - - // closing ] -> we are done - if (last_token == token_type::end_array) - { - return true; - } - - // parse values - while (true) - { - // parse value - if (not accept_internal()) - { - return false; - } - - // comma -> next value - get_token(); - if (last_token == token_type::value_separator) - { - get_token(); - continue; - } - - // closing ] - return (last_token == token_type::end_array); - } - } - - case token_type::value_float: - { - // reject infinity or NAN - return std::isfinite(m_lexer.get_number_float()); - } - - case token_type::literal_false: - case token_type::literal_null: - case token_type::literal_true: - case token_type::value_integer: - case token_type::value_string: - case token_type::value_unsigned: - return true; - - default: // the last token was unexpected - return false; - } - } - - /// get next token from lexer - token_type get_token() - { - return (last_token = m_lexer.scan()); - } - - /*! - @throw parse_error.101 if expected token did not occur - */ - bool expect(token_type t) - { - if (JSON_UNLIKELY(t != last_token)) - { - errored = true; - expected = t; - if (allow_exceptions) - { - throw_exception(); - } - else - { - return false; - } - } - - return true; - } - - [[noreturn]] void throw_exception() const - { - std::string error_msg = "syntax error - "; - if (last_token == token_type::parse_error) - { - error_msg += std::string(m_lexer.get_error_message()) + "; last read: '" + - m_lexer.get_token_string() + "'"; - } - else - { - error_msg += "unexpected " + std::string(lexer_t::token_type_name(last_token)); - } - - if (expected != token_type::uninitialized) - { - error_msg += "; expected " + std::string(lexer_t::token_type_name(expected)); - } - - JSON_THROW(parse_error::create(101, m_lexer.get_position(), error_msg)); - } - - private: - /// current level of recursion - int depth = 0; - /// callback function - const parser_callback_t callback = nullptr; - /// the type of the last read token - token_type last_token = token_type::uninitialized; - /// the lexer - lexer_t m_lexer; - /// whether a syntax error occurred - bool errored = false; - /// possible reason for the syntax error - token_type expected = token_type::uninitialized; - /// whether to throw exceptions in case of errors - const bool allow_exceptions = true; -}; -} -} - -// #include <nlohmann/detail/iterators/primitive_iterator.hpp> - - -#include <cstddef> // ptrdiff_t -#include <limits> // numeric_limits - -namespace nlohmann -{ -namespace detail -{ -/* -@brief an iterator for primitive JSON types - -This class models an iterator for primitive JSON types (boolean, number, -string). It's only purpose is to allow the iterator/const_iterator classes -to "iterate" over primitive values. Internally, the iterator is modeled by -a `difference_type` variable. Value begin_value (`0`) models the begin, -end_value (`1`) models past the end. -*/ -class primitive_iterator_t -{ - private: - using difference_type = std::ptrdiff_t; - static constexpr difference_type begin_value = 0; - static constexpr difference_type end_value = begin_value + 1; - - /// iterator as signed integer type - difference_type m_it = (std::numeric_limits<std::ptrdiff_t>::min)(); - - public: - constexpr difference_type get_value() const noexcept - { - return m_it; - } - - /// set iterator to a defined beginning - void set_begin() noexcept - { - m_it = begin_value; - } - - /// set iterator to a defined past the end - void set_end() noexcept - { - m_it = end_value; - } - - /// return whether the iterator can be dereferenced - constexpr bool is_begin() const noexcept - { - return m_it == begin_value; - } - - /// return whether the iterator is at end - constexpr bool is_end() const noexcept - { - return m_it == end_value; - } - - friend constexpr bool operator==(primitive_iterator_t lhs, primitive_iterator_t rhs) noexcept - { - return lhs.m_it == rhs.m_it; - } - - friend constexpr bool operator<(primitive_iterator_t lhs, primitive_iterator_t rhs) noexcept - { - return lhs.m_it < rhs.m_it; - } - - primitive_iterator_t operator+(difference_type n) noexcept - { - auto result = *this; - result += n; - return result; - } - - friend constexpr difference_type operator-(primitive_iterator_t lhs, primitive_iterator_t rhs) noexcept - { - return lhs.m_it - rhs.m_it; - } - - primitive_iterator_t& operator++() noexcept - { - ++m_it; - return *this; - } - - primitive_iterator_t const operator++(int) noexcept - { - auto result = *this; - m_it++; - return result; - } - - primitive_iterator_t& operator--() noexcept - { - --m_it; - return *this; - } - - primitive_iterator_t const operator--(int) noexcept - { - auto result = *this; - m_it--; - return result; - } - - primitive_iterator_t& operator+=(difference_type n) noexcept - { - m_it += n; - return *this; - } - - primitive_iterator_t& operator-=(difference_type n) noexcept - { - m_it -= n; - return *this; - } -}; -} -} - -// #include <nlohmann/detail/iterators/internal_iterator.hpp> - - -// #include <nlohmann/detail/iterators/primitive_iterator.hpp> - - -namespace nlohmann -{ -namespace detail -{ -/*! -@brief an iterator value - -@note This structure could easily be a union, but MSVC currently does not allow -unions members with complex constructors, see https://github.com/nlohmann/json/pull/105. -*/ -template<typename BasicJsonType> struct internal_iterator -{ - /// iterator for JSON objects - typename BasicJsonType::object_t::iterator object_iterator {}; - /// iterator for JSON arrays - typename BasicJsonType::array_t::iterator array_iterator {}; - /// generic iterator for all other types - primitive_iterator_t primitive_iterator {}; -}; -} -} - -// #include <nlohmann/detail/iterators/iter_impl.hpp> - - -#include <ciso646> // not -#include <iterator> // iterator, random_access_iterator_tag, bidirectional_iterator_tag, advance, next -#include <type_traits> // conditional, is_const, remove_const - -// #include <nlohmann/detail/exceptions.hpp> - -// #include <nlohmann/detail/iterators/internal_iterator.hpp> - -// #include <nlohmann/detail/iterators/primitive_iterator.hpp> - -// #include <nlohmann/detail/macro_scope.hpp> - -// #include <nlohmann/detail/meta.hpp> - -// #include <nlohmann/detail/value_t.hpp> - - -namespace nlohmann -{ -namespace detail -{ -// forward declare, to be able to friend it later on -template<typename IteratorType> class iteration_proxy; - -/*! -@brief a template for a bidirectional iterator for the @ref basic_json class - -This class implements a both iterators (iterator and const_iterator) for the -@ref basic_json class. - -@note An iterator is called *initialized* when a pointer to a JSON value has - been set (e.g., by a constructor or a copy assignment). If the iterator is - default-constructed, it is *uninitialized* and most methods are undefined. - **The library uses assertions to detect calls on uninitialized iterators.** - -@requirement The class satisfies the following concept requirements: -- -[BidirectionalIterator](http://en.cppreference.com/w/cpp/concept/BidirectionalIterator): - The iterator that can be moved can be moved in both directions (i.e. - incremented and decremented). - -@since version 1.0.0, simplified in version 2.0.9, change to bidirectional - iterators in version 3.0.0 (see https://github.com/nlohmann/json/issues/593) -*/ -template<typename BasicJsonType> -class iter_impl -{ - /// allow basic_json to access private members - friend iter_impl<typename std::conditional<std::is_const<BasicJsonType>::value, typename std::remove_const<BasicJsonType>::type, const BasicJsonType>::type>; - friend BasicJsonType; - friend iteration_proxy<iter_impl>; - - using object_t = typename BasicJsonType::object_t; - using array_t = typename BasicJsonType::array_t; - // make sure BasicJsonType is basic_json or const basic_json - static_assert(is_basic_json<typename std::remove_const<BasicJsonType>::type>::value, - "iter_impl only accepts (const) basic_json"); - - public: - - /// The std::iterator class template (used as a base class to provide typedefs) is deprecated in C++17. - /// The C++ Standard has never required user-defined iterators to derive from std::iterator. - /// A user-defined iterator should provide publicly accessible typedefs named - /// iterator_category, value_type, difference_type, pointer, and reference. - /// Note that value_type is required to be non-const, even for constant iterators. - using iterator_category = std::bidirectional_iterator_tag; - - /// the type of the values when the iterator is dereferenced - using value_type = typename BasicJsonType::value_type; - /// a type to represent differences between iterators - using difference_type = typename BasicJsonType::difference_type; - /// defines a pointer to the type iterated over (value_type) - using pointer = typename std::conditional<std::is_const<BasicJsonType>::value, - typename BasicJsonType::const_pointer, - typename BasicJsonType::pointer>::type; - /// defines a reference to the type iterated over (value_type) - using reference = - typename std::conditional<std::is_const<BasicJsonType>::value, - typename BasicJsonType::const_reference, - typename BasicJsonType::reference>::type; - - /// default constructor - iter_impl() = default; - - /*! - @brief constructor for a given JSON instance - @param[in] object pointer to a JSON object for this iterator - @pre object != nullptr - @post The iterator is initialized; i.e. `m_object != nullptr`. - */ - explicit iter_impl(pointer object) noexcept : m_object(object) - { - assert(m_object != nullptr); - - switch (m_object->m_type) - { - case value_t::object: - { - m_it.object_iterator = typename object_t::iterator(); - break; - } - - case value_t::array: - { - m_it.array_iterator = typename array_t::iterator(); - break; - } - - default: - { - m_it.primitive_iterator = primitive_iterator_t(); - break; - } - } - } - - /*! - @note The conventional copy constructor and copy assignment are implicitly - defined. Combined with the following converting constructor and - assignment, they support: (1) copy from iterator to iterator, (2) - copy from const iterator to const iterator, and (3) conversion from - iterator to const iterator. However conversion from const iterator - to iterator is not defined. - */ - - /*! - @brief converting constructor - @param[in] other non-const iterator to copy from - @note It is not checked whether @a other is initialized. - */ - iter_impl(const iter_impl<typename std::remove_const<BasicJsonType>::type>& other) noexcept - : m_object(other.m_object), m_it(other.m_it) {} - - /*! - @brief converting assignment - @param[in,out] other non-const iterator to copy from - @return const/non-const iterator - @note It is not checked whether @a other is initialized. - */ - iter_impl& operator=(const iter_impl<typename std::remove_const<BasicJsonType>::type>& other) noexcept - { - m_object = other.m_object; - m_it = other.m_it; - return *this; - } - - private: - /*! - @brief set the iterator to the first value - @pre The iterator is initialized; i.e. `m_object != nullptr`. - */ - void set_begin() noexcept - { - assert(m_object != nullptr); - - switch (m_object->m_type) - { - case value_t::object: - { - m_it.object_iterator = m_object->m_value.object->begin(); - break; - } - - case value_t::array: - { - m_it.array_iterator = m_object->m_value.array->begin(); - break; - } - - case value_t::null: - { - // set to end so begin()==end() is true: null is empty - m_it.primitive_iterator.set_end(); - break; - } - - default: - { - m_it.primitive_iterator.set_begin(); - break; - } - } - } - - /*! - @brief set the iterator past the last value - @pre The iterator is initialized; i.e. `m_object != nullptr`. - */ - void set_end() noexcept - { - assert(m_object != nullptr); - - switch (m_object->m_type) - { - case value_t::object: - { - m_it.object_iterator = m_object->m_value.object->end(); - break; - } - - case value_t::array: - { - m_it.array_iterator = m_object->m_value.array->end(); - break; - } - - default: - { - m_it.primitive_iterator.set_end(); - break; - } - } - } - - public: - /*! - @brief return a reference to the value pointed to by the iterator - @pre The iterator is initialized; i.e. `m_object != nullptr`. - */ - reference operator*() const - { - assert(m_object != nullptr); - - switch (m_object->m_type) - { - case value_t::object: - { - assert(m_it.object_iterator != m_object->m_value.object->end()); - return m_it.object_iterator->second; - } - - case value_t::array: - { - assert(m_it.array_iterator != m_object->m_value.array->end()); - return *m_it.array_iterator; - } - - case value_t::null: - JSON_THROW(invalid_iterator::create(214, "cannot get value")); - - default: - { - if (JSON_LIKELY(m_it.primitive_iterator.is_begin())) - { - return *m_object; - } - - JSON_THROW(invalid_iterator::create(214, "cannot get value")); - } - } - } - - /*! - @brief dereference the iterator - @pre The iterator is initialized; i.e. `m_object != nullptr`. - */ - pointer operator->() const - { - assert(m_object != nullptr); - - switch (m_object->m_type) - { - case value_t::object: - { - assert(m_it.object_iterator != m_object->m_value.object->end()); - return &(m_it.object_iterator->second); - } - - case value_t::array: - { - assert(m_it.array_iterator != m_object->m_value.array->end()); - return &*m_it.array_iterator; - } - - default: - { - if (JSON_LIKELY(m_it.primitive_iterator.is_begin())) - { - return m_object; - } - - JSON_THROW(invalid_iterator::create(214, "cannot get value")); - } - } - } - - /*! - @brief post-increment (it++) - @pre The iterator is initialized; i.e. `m_object != nullptr`. - */ - iter_impl const operator++(int) - { - auto result = *this; - ++(*this); - return result; - } - - /*! - @brief pre-increment (++it) - @pre The iterator is initialized; i.e. `m_object != nullptr`. - */ - iter_impl& operator++() - { - assert(m_object != nullptr); - - switch (m_object->m_type) - { - case value_t::object: - { - std::advance(m_it.object_iterator, 1); - break; - } - - case value_t::array: - { - std::advance(m_it.array_iterator, 1); - break; - } - - default: - { - ++m_it.primitive_iterator; - break; - } - } - - return *this; - } - - /*! - @brief post-decrement (it--) - @pre The iterator is initialized; i.e. `m_object != nullptr`. - */ - iter_impl const operator--(int) - { - auto result = *this; - --(*this); - return result; - } - - /*! - @brief pre-decrement (--it) - @pre The iterator is initialized; i.e. `m_object != nullptr`. - */ - iter_impl& operator--() - { - assert(m_object != nullptr); - - switch (m_object->m_type) - { - case value_t::object: - { - std::advance(m_it.object_iterator, -1); - break; - } - - case value_t::array: - { - std::advance(m_it.array_iterator, -1); - break; - } - - default: - { - --m_it.primitive_iterator; - break; - } - } - - return *this; - } - - /*! - @brief comparison: equal - @pre The iterator is initialized; i.e. `m_object != nullptr`. - */ - bool operator==(const iter_impl& other) const - { - // if objects are not the same, the comparison is undefined - if (JSON_UNLIKELY(m_object != other.m_object)) - { - JSON_THROW(invalid_iterator::create(212, "cannot compare iterators of different containers")); - } - - assert(m_object != nullptr); - - switch (m_object->m_type) - { - case value_t::object: - return (m_it.object_iterator == other.m_it.object_iterator); - - case value_t::array: - return (m_it.array_iterator == other.m_it.array_iterator); - - default: - return (m_it.primitive_iterator == other.m_it.primitive_iterator); - } - } - - /*! - @brief comparison: not equal - @pre The iterator is initialized; i.e. `m_object != nullptr`. - */ - bool operator!=(const iter_impl& other) const - { - return not operator==(other); - } - - /*! - @brief comparison: smaller - @pre The iterator is initialized; i.e. `m_object != nullptr`. - */ - bool operator<(const iter_impl& other) const - { - // if objects are not the same, the comparison is undefined - if (JSON_UNLIKELY(m_object != other.m_object)) - { - JSON_THROW(invalid_iterator::create(212, "cannot compare iterators of different containers")); - } - - assert(m_object != nullptr); - - switch (m_object->m_type) - { - case value_t::object: - JSON_THROW(invalid_iterator::create(213, "cannot compare order of object iterators")); - - case value_t::array: - return (m_it.array_iterator < other.m_it.array_iterator); - - default: - return (m_it.primitive_iterator < other.m_it.primitive_iterator); - } - } - - /*! - @brief comparison: less than or equal - @pre The iterator is initialized; i.e. `m_object != nullptr`. - */ - bool operator<=(const iter_impl& other) const - { - return not other.operator < (*this); - } - - /*! - @brief comparison: greater than - @pre The iterator is initialized; i.e. `m_object != nullptr`. - */ - bool operator>(const iter_impl& other) const - { - return not operator<=(other); - } - - /*! - @brief comparison: greater than or equal - @pre The iterator is initialized; i.e. `m_object != nullptr`. - */ - bool operator>=(const iter_impl& other) const - { - return not operator<(other); - } - - /*! - @brief add to iterator - @pre The iterator is initialized; i.e. `m_object != nullptr`. - */ - iter_impl& operator+=(difference_type i) - { - assert(m_object != nullptr); - - switch (m_object->m_type) - { - case value_t::object: - JSON_THROW(invalid_iterator::create(209, "cannot use offsets with object iterators")); - - case value_t::array: - { - std::advance(m_it.array_iterator, i); - break; - } - - default: - { - m_it.primitive_iterator += i; - break; - } - } - - return *this; - } - - /*! - @brief subtract from iterator - @pre The iterator is initialized; i.e. `m_object != nullptr`. - */ - iter_impl& operator-=(difference_type i) - { - return operator+=(-i); - } - - /*! - @brief add to iterator - @pre The iterator is initialized; i.e. `m_object != nullptr`. - */ - iter_impl operator+(difference_type i) const - { - auto result = *this; - result += i; - return result; - } - - /*! - @brief addition of distance and iterator - @pre The iterator is initialized; i.e. `m_object != nullptr`. - */ - friend iter_impl operator+(difference_type i, const iter_impl& it) - { - auto result = it; - result += i; - return result; - } - - /*! - @brief subtract from iterator - @pre The iterator is initialized; i.e. `m_object != nullptr`. - */ - iter_impl operator-(difference_type i) const - { - auto result = *this; - result -= i; - return result; - } - - /*! - @brief return difference - @pre The iterator is initialized; i.e. `m_object != nullptr`. - */ - difference_type operator-(const iter_impl& other) const - { - assert(m_object != nullptr); - - switch (m_object->m_type) - { - case value_t::object: - JSON_THROW(invalid_iterator::create(209, "cannot use offsets with object iterators")); - - case value_t::array: - return m_it.array_iterator - other.m_it.array_iterator; - - default: - return m_it.primitive_iterator - other.m_it.primitive_iterator; - } - } - - /*! - @brief access to successor - @pre The iterator is initialized; i.e. `m_object != nullptr`. - */ - reference operator[](difference_type n) const - { - assert(m_object != nullptr); - - switch (m_object->m_type) - { - case value_t::object: - JSON_THROW(invalid_iterator::create(208, "cannot use operator[] for object iterators")); - - case value_t::array: - return *std::next(m_it.array_iterator, n); - - case value_t::null: - JSON_THROW(invalid_iterator::create(214, "cannot get value")); - - default: - { - if (JSON_LIKELY(m_it.primitive_iterator.get_value() == -n)) - { - return *m_object; - } - - JSON_THROW(invalid_iterator::create(214, "cannot get value")); - } - } - } - - /*! - @brief return the key of an object iterator - @pre The iterator is initialized; i.e. `m_object != nullptr`. - */ - typename object_t::key_type key() const - { - assert(m_object != nullptr); - - if (JSON_LIKELY(m_object->is_object())) - { - return m_it.object_iterator->first; - } - - JSON_THROW(invalid_iterator::create(207, "cannot use key() for non-object iterators")); - } - - /*! - @brief return the value of an iterator - @pre The iterator is initialized; i.e. `m_object != nullptr`. - */ - reference value() const - { - return operator*(); - } - - private: - /// associated JSON instance - pointer m_object = nullptr; - /// the actual iterator of the associated instance - internal_iterator<typename std::remove_const<BasicJsonType>::type> m_it; -}; -} -} - -// #include <nlohmann/detail/iterators/iteration_proxy.hpp> - - -#include <cstddef> // size_t -#include <string> // string, to_string - -// #include <nlohmann/detail/value_t.hpp> - - -namespace nlohmann -{ -namespace detail -{ -/// proxy class for the items() function -template<typename IteratorType> class iteration_proxy -{ - private: - /// helper class for iteration - class iteration_proxy_internal - { - private: - /// the iterator - IteratorType anchor; - /// an index for arrays (used to create key names) - std::size_t array_index = 0; - - public: - explicit iteration_proxy_internal(IteratorType it) noexcept : anchor(it) {} - - /// dereference operator (needed for range-based for) - iteration_proxy_internal& operator*() - { - return *this; - } - - /// increment operator (needed for range-based for) - iteration_proxy_internal& operator++() - { - ++anchor; - ++array_index; - - return *this; - } - - /// inequality operator (needed for range-based for) - bool operator!=(const iteration_proxy_internal& o) const noexcept - { - return anchor != o.anchor; - } - - /// return key of the iterator - std::string key() const - { - assert(anchor.m_object != nullptr); - - switch (anchor.m_object->type()) - { - // use integer array index as key - case value_t::array: - return std::to_string(array_index); - - // use key from the object - case value_t::object: - return anchor.key(); - - // use an empty key for all primitive types - default: - return ""; - } - } - - /// return value of the iterator - typename IteratorType::reference value() const - { - return anchor.value(); - } - }; - - /// the container to iterate - typename IteratorType::reference container; - - public: - /// construct iteration proxy from a container - explicit iteration_proxy(typename IteratorType::reference cont) noexcept - : container(cont) {} - - /// return iterator begin (needed for range-based for) - iteration_proxy_internal begin() noexcept - { - return iteration_proxy_internal(container.begin()); - } - - /// return iterator end (needed for range-based for) - iteration_proxy_internal end() noexcept - { - return iteration_proxy_internal(container.end()); - } -}; -} -} - -// #include <nlohmann/detail/iterators/json_reverse_iterator.hpp> - - -#include <cstddef> // ptrdiff_t -#include <iterator> // reverse_iterator -#include <utility> // declval - -namespace nlohmann -{ -namespace detail -{ -////////////////////// -// reverse_iterator // -////////////////////// - -/*! -@brief a template for a reverse iterator class - -@tparam Base the base iterator type to reverse. Valid types are @ref -iterator (to create @ref reverse_iterator) and @ref const_iterator (to -create @ref const_reverse_iterator). - -@requirement The class satisfies the following concept requirements: -- -[BidirectionalIterator](http://en.cppreference.com/w/cpp/concept/BidirectionalIterator): - The iterator that can be moved can be moved in both directions (i.e. - incremented and decremented). -- [OutputIterator](http://en.cppreference.com/w/cpp/concept/OutputIterator): - It is possible to write to the pointed-to element (only if @a Base is - @ref iterator). - -@since version 1.0.0 -*/ -template<typename Base> -class json_reverse_iterator : public std::reverse_iterator<Base> -{ - public: - using difference_type = std::ptrdiff_t; - /// shortcut to the reverse iterator adapter - using base_iterator = std::reverse_iterator<Base>; - /// the reference type for the pointed-to element - using reference = typename Base::reference; - - /// create reverse iterator from iterator - json_reverse_iterator(const typename base_iterator::iterator_type& it) noexcept - : base_iterator(it) {} - - /// create reverse iterator from base class - json_reverse_iterator(const base_iterator& it) noexcept : base_iterator(it) {} - - /// post-increment (it++) - json_reverse_iterator const operator++(int) - { - return static_cast<json_reverse_iterator>(base_iterator::operator++(1)); - } - - /// pre-increment (++it) - json_reverse_iterator& operator++() - { - return static_cast<json_reverse_iterator&>(base_iterator::operator++()); - } - - /// post-decrement (it--) - json_reverse_iterator const operator--(int) - { - return static_cast<json_reverse_iterator>(base_iterator::operator--(1)); - } - - /// pre-decrement (--it) - json_reverse_iterator& operator--() - { - return static_cast<json_reverse_iterator&>(base_iterator::operator--()); - } - - /// add to iterator - json_reverse_iterator& operator+=(difference_type i) - { - return static_cast<json_reverse_iterator&>(base_iterator::operator+=(i)); - } - - /// add to iterator - json_reverse_iterator operator+(difference_type i) const - { - return static_cast<json_reverse_iterator>(base_iterator::operator+(i)); - } - - /// subtract from iterator - json_reverse_iterator operator-(difference_type i) const - { - return static_cast<json_reverse_iterator>(base_iterator::operator-(i)); - } - - /// return difference - difference_type operator-(const json_reverse_iterator& other) const - { - return base_iterator(*this) - base_iterator(other); - } - - /// access to successor - reference operator[](difference_type n) const - { - return *(this->operator+(n)); - } - - /// return the key of an object iterator - auto key() const -> decltype(std::declval<Base>().key()) - { - auto it = --this->base(); - return it.key(); - } - - /// return the value of an iterator - reference value() const - { - auto it = --this->base(); - return it.operator * (); - } -}; -} -} - -// #include <nlohmann/detail/output/output_adapters.hpp> - - -#include <algorithm> // copy -#include <cstddef> // size_t -#include <ios> // streamsize -#include <iterator> // back_inserter -#include <memory> // shared_ptr, make_shared -#include <ostream> // basic_ostream -#include <string> // basic_string -#include <vector> // vector - -namespace nlohmann -{ -namespace detail -{ -/// abstract output adapter interface -template<typename CharType> struct output_adapter_protocol -{ - virtual void write_character(CharType c) = 0; - virtual void write_characters(const CharType* s, std::size_t length) = 0; - virtual ~output_adapter_protocol() = default; -}; - -/// a type to simplify interfaces -template<typename CharType> -using output_adapter_t = std::shared_ptr<output_adapter_protocol<CharType>>; - -/// output adapter for byte vectors -template<typename CharType> -class output_vector_adapter : public output_adapter_protocol<CharType> -{ - public: - explicit output_vector_adapter(std::vector<CharType>& vec) : v(vec) {} - - void write_character(CharType c) override - { - v.push_back(c); - } - - void write_characters(const CharType* s, std::size_t length) override - { - std::copy(s, s + length, std::back_inserter(v)); - } - - private: - std::vector<CharType>& v; -}; - -/// output adapter for output streams -template<typename CharType> -class output_stream_adapter : public output_adapter_protocol<CharType> -{ - public: - explicit output_stream_adapter(std::basic_ostream<CharType>& s) : stream(s) {} - - void write_character(CharType c) override - { - stream.put(c); - } - - void write_characters(const CharType* s, std::size_t length) override - { - stream.write(s, static_cast<std::streamsize>(length)); - } - - private: - std::basic_ostream<CharType>& stream; -}; - -/// output adapter for basic_string -template<typename CharType, typename StringType = std::basic_string<CharType>> -class output_string_adapter : public output_adapter_protocol<CharType> -{ - public: - explicit output_string_adapter(StringType& s) : str(s) {} - - void write_character(CharType c) override - { - str.push_back(c); - } - - void write_characters(const CharType* s, std::size_t length) override - { - str.append(s, length); - } - - private: - StringType& str; -}; - -template<typename CharType, typename StringType = std::basic_string<CharType>> -class output_adapter -{ - public: - output_adapter(std::vector<CharType>& vec) - : oa(std::make_shared<output_vector_adapter<CharType>>(vec)) {} - - output_adapter(std::basic_ostream<CharType>& s) - : oa(std::make_shared<output_stream_adapter<CharType>>(s)) {} - - output_adapter(StringType& s) - : oa(std::make_shared<output_string_adapter<CharType, StringType>>(s)) {} - - operator output_adapter_t<CharType>() - { - return oa; - } - - private: - output_adapter_t<CharType> oa = nullptr; -}; -} -} - -// #include <nlohmann/detail/input/binary_reader.hpp> - - -#include <algorithm> // generate_n -#include <array> // array -#include <cassert> // assert -#include <cmath> // ldexp -#include <cstddef> // size_t -#include <cstdint> // uint8_t, uint16_t, uint32_t, uint64_t -#include <cstring> // memcpy -#include <iomanip> // setw, setfill -#include <ios> // hex -#include <iterator> // back_inserter -#include <limits> // numeric_limits -#include <sstream> // stringstream -#include <string> // char_traits, string -#include <utility> // make_pair, move - -// #include <nlohmann/detail/input/input_adapters.hpp> - -// #include <nlohmann/detail/exceptions.hpp> - -// #include <nlohmann/detail/macro_scope.hpp> - -// #include <nlohmann/detail/value_t.hpp> - - -namespace nlohmann -{ -namespace detail -{ -/////////////////// -// binary reader // -/////////////////// - -/*! -@brief deserialization of CBOR and MessagePack values -*/ -template<typename BasicJsonType> -class binary_reader -{ - using number_integer_t = typename BasicJsonType::number_integer_t; - using number_unsigned_t = typename BasicJsonType::number_unsigned_t; - using string_t = typename BasicJsonType::string_t; - - public: - /*! - @brief create a binary reader - - @param[in] adapter input adapter to read from - */ - explicit binary_reader(input_adapter_t adapter) : ia(std::move(adapter)) - { - assert(ia); - } - - /*! - @brief create a JSON value from CBOR input - - @param[in] strict whether to expect the input to be consumed completed - @return JSON value created from CBOR input - - @throw parse_error.110 if input ended unexpectedly or the end of file was - not reached when @a strict was set to true - @throw parse_error.112 if unsupported byte was read - */ - BasicJsonType parse_cbor(const bool strict) - { - const auto res = parse_cbor_internal(); - if (strict) - { - get(); - expect_eof(); - } - return res; - } - - /*! - @brief create a JSON value from MessagePack input - - @param[in] strict whether to expect the input to be consumed completed - @return JSON value created from MessagePack input - - @throw parse_error.110 if input ended unexpectedly or the end of file was - not reached when @a strict was set to true - @throw parse_error.112 if unsupported byte was read - */ - BasicJsonType parse_msgpack(const bool strict) - { - const auto res = parse_msgpack_internal(); - if (strict) - { - get(); - expect_eof(); - } - return res; - } - - /*! - @brief create a JSON value from UBJSON input - - @param[in] strict whether to expect the input to be consumed completed - @return JSON value created from UBJSON input - - @throw parse_error.110 if input ended unexpectedly or the end of file was - not reached when @a strict was set to true - @throw parse_error.112 if unsupported byte was read - */ - BasicJsonType parse_ubjson(const bool strict) - { - const auto res = parse_ubjson_internal(); - if (strict) - { - get_ignore_noop(); - expect_eof(); - } - return res; - } - - /*! - @brief determine system byte order - - @return true if and only if system's byte order is little endian - - @note from http://stackoverflow.com/a/1001328/266378 - */ - static constexpr bool little_endianess(int num = 1) noexcept - { - return (*reinterpret_cast<char*>(&num) == 1); - } - - private: - /*! - @param[in] get_char whether a new character should be retrieved from the - input (true, default) or whether the last read - character should be considered instead - */ - BasicJsonType parse_cbor_internal(const bool get_char = true) - { - switch (get_char ? get() : current) - { - // EOF - case std::char_traits<char>::eof(): - JSON_THROW(parse_error::create(110, chars_read, "unexpected end of input")); - - // Integer 0x00..0x17 (0..23) - case 0x00: - case 0x01: - case 0x02: - case 0x03: - case 0x04: - case 0x05: - case 0x06: - case 0x07: - case 0x08: - case 0x09: - case 0x0A: - case 0x0B: - case 0x0C: - case 0x0D: - case 0x0E: - case 0x0F: - case 0x10: - case 0x11: - case 0x12: - case 0x13: - case 0x14: - case 0x15: - case 0x16: - case 0x17: - return static_cast<number_unsigned_t>(current); - - case 0x18: // Unsigned integer (one-byte uint8_t follows) - return get_number<uint8_t>(); - - case 0x19: // Unsigned integer (two-byte uint16_t follows) - return get_number<uint16_t>(); - - case 0x1A: // Unsigned integer (four-byte uint32_t follows) - return get_number<uint32_t>(); - - case 0x1B: // Unsigned integer (eight-byte uint64_t follows) - return get_number<uint64_t>(); - - // Negative integer -1-0x00..-1-0x17 (-1..-24) - case 0x20: - case 0x21: - case 0x22: - case 0x23: - case 0x24: - case 0x25: - case 0x26: - case 0x27: - case 0x28: - case 0x29: - case 0x2A: - case 0x2B: - case 0x2C: - case 0x2D: - case 0x2E: - case 0x2F: - case 0x30: - case 0x31: - case 0x32: - case 0x33: - case 0x34: - case 0x35: - case 0x36: - case 0x37: - return static_cast<int8_t>(0x20 - 1 - current); - - case 0x38: // Negative integer (one-byte uint8_t follows) - { - return static_cast<number_integer_t>(-1) - get_number<uint8_t>(); - } - - case 0x39: // Negative integer -1-n (two-byte uint16_t follows) - { - return static_cast<number_integer_t>(-1) - get_number<uint16_t>(); - } - - case 0x3A: // Negative integer -1-n (four-byte uint32_t follows) - { - return static_cast<number_integer_t>(-1) - get_number<uint32_t>(); - } - - case 0x3B: // Negative integer -1-n (eight-byte uint64_t follows) - { - return static_cast<number_integer_t>(-1) - - static_cast<number_integer_t>(get_number<uint64_t>()); - } - - // UTF-8 string (0x00..0x17 bytes follow) - case 0x60: - case 0x61: - case 0x62: - case 0x63: - case 0x64: - case 0x65: - case 0x66: - case 0x67: - case 0x68: - case 0x69: - case 0x6A: - case 0x6B: - case 0x6C: - case 0x6D: - case 0x6E: - case 0x6F: - case 0x70: - case 0x71: - case 0x72: - case 0x73: - case 0x74: - case 0x75: - case 0x76: - case 0x77: - case 0x78: // UTF-8 string (one-byte uint8_t for n follows) - case 0x79: // UTF-8 string (two-byte uint16_t for n follow) - case 0x7A: // UTF-8 string (four-byte uint32_t for n follow) - case 0x7B: // UTF-8 string (eight-byte uint64_t for n follow) - case 0x7F: // UTF-8 string (indefinite length) - { - return get_cbor_string(); - } - - // array (0x00..0x17 data items follow) - case 0x80: - case 0x81: - case 0x82: - case 0x83: - case 0x84: - case 0x85: - case 0x86: - case 0x87: - case 0x88: - case 0x89: - case 0x8A: - case 0x8B: - case 0x8C: - case 0x8D: - case 0x8E: - case 0x8F: - case 0x90: - case 0x91: - case 0x92: - case 0x93: - case 0x94: - case 0x95: - case 0x96: - case 0x97: - { - return get_cbor_array(current & 0x1F); - } - - case 0x98: // array (one-byte uint8_t for n follows) - { - return get_cbor_array(get_number<uint8_t>()); - } - - case 0x99: // array (two-byte uint16_t for n follow) - { - return get_cbor_array(get_number<uint16_t>()); - } - - case 0x9A: // array (four-byte uint32_t for n follow) - { - return get_cbor_array(get_number<uint32_t>()); - } - - case 0x9B: // array (eight-byte uint64_t for n follow) - { - return get_cbor_array(get_number<uint64_t>()); - } - - case 0x9F: // array (indefinite length) - { - BasicJsonType result = value_t::array; - while (get() != 0xFF) - { - result.push_back(parse_cbor_internal(false)); - } - return result; - } - - // map (0x00..0x17 pairs of data items follow) - case 0xA0: - case 0xA1: - case 0xA2: - case 0xA3: - case 0xA4: - case 0xA5: - case 0xA6: - case 0xA7: - case 0xA8: - case 0xA9: - case 0xAA: - case 0xAB: - case 0xAC: - case 0xAD: - case 0xAE: - case 0xAF: - case 0xB0: - case 0xB1: - case 0xB2: - case 0xB3: - case 0xB4: - case 0xB5: - case 0xB6: - case 0xB7: - { - return get_cbor_object(current & 0x1F); - } - - case 0xB8: // map (one-byte uint8_t for n follows) - { - return get_cbor_object(get_number<uint8_t>()); - } - - case 0xB9: // map (two-byte uint16_t for n follow) - { - return get_cbor_object(get_number<uint16_t>()); - } - - case 0xBA: // map (four-byte uint32_t for n follow) - { - return get_cbor_object(get_number<uint32_t>()); - } - - case 0xBB: // map (eight-byte uint64_t for n follow) - { - return get_cbor_object(get_number<uint64_t>()); - } - - case 0xBF: // map (indefinite length) - { - BasicJsonType result = value_t::object; - while (get() != 0xFF) - { - auto key = get_cbor_string(); - result[key] = parse_cbor_internal(); - } - return result; - } - - case 0xF4: // false - { - return false; - } - - case 0xF5: // true - { - return true; - } - - case 0xF6: // null - { - return value_t::null; - } - - case 0xF9: // Half-Precision Float (two-byte IEEE 754) - { - const int byte1 = get(); - unexpect_eof(); - const int byte2 = get(); - unexpect_eof(); - - // code from RFC 7049, Appendix D, Figure 3: - // As half-precision floating-point numbers were only added - // to IEEE 754 in 2008, today's programming platforms often - // still only have limited support for them. It is very - // easy to include at least decoding support for them even - // without such support. An example of a small decoder for - // half-precision floating-point numbers in the C language - // is shown in Fig. 3. - const int half = (byte1 << 8) + byte2; - const int exp = (half >> 10) & 0x1F; - const int mant = half & 0x3FF; - double val; - if (exp == 0) - { - val = std::ldexp(mant, -24); - } - else if (exp != 31) - { - val = std::ldexp(mant + 1024, exp - 25); - } - else - { - val = (mant == 0) ? std::numeric_limits<double>::infinity() - : std::numeric_limits<double>::quiet_NaN(); - } - return (half & 0x8000) != 0 ? -val : val; - } - - case 0xFA: // Single-Precision Float (four-byte IEEE 754) - { - return get_number<float>(); - } - - case 0xFB: // Double-Precision Float (eight-byte IEEE 754) - { - return get_number<double>(); - } - - default: // anything else (0xFF is handled inside the other types) - { - std::stringstream ss; - ss << std::setw(2) << std::uppercase << std::setfill('0') << std::hex << current; - JSON_THROW(parse_error::create(112, chars_read, "error reading CBOR; last byte: 0x" + ss.str())); - } - } - } - - BasicJsonType parse_msgpack_internal() - { - switch (get()) - { - // EOF - case std::char_traits<char>::eof(): - JSON_THROW(parse_error::create(110, chars_read, "unexpected end of input")); - - // positive fixint - case 0x00: - case 0x01: - case 0x02: - case 0x03: - case 0x04: - case 0x05: - case 0x06: - case 0x07: - case 0x08: - case 0x09: - case 0x0A: - case 0x0B: - case 0x0C: - case 0x0D: - case 0x0E: - case 0x0F: - case 0x10: - case 0x11: - case 0x12: - case 0x13: - case 0x14: - case 0x15: - case 0x16: - case 0x17: - case 0x18: - case 0x19: - case 0x1A: - case 0x1B: - case 0x1C: - case 0x1D: - case 0x1E: - case 0x1F: - case 0x20: - case 0x21: - case 0x22: - case 0x23: - case 0x24: - case 0x25: - case 0x26: - case 0x27: - case 0x28: - case 0x29: - case 0x2A: - case 0x2B: - case 0x2C: - case 0x2D: - case 0x2E: - case 0x2F: - case 0x30: - case 0x31: - case 0x32: - case 0x33: - case 0x34: - case 0x35: - case 0x36: - case 0x37: - case 0x38: - case 0x39: - case 0x3A: - case 0x3B: - case 0x3C: - case 0x3D: - case 0x3E: - case 0x3F: - case 0x40: - case 0x41: - case 0x42: - case 0x43: - case 0x44: - case 0x45: - case 0x46: - case 0x47: - case 0x48: - case 0x49: - case 0x4A: - case 0x4B: - case 0x4C: - case 0x4D: - case 0x4E: - case 0x4F: - case 0x50: - case 0x51: - case 0x52: - case 0x53: - case 0x54: - case 0x55: - case 0x56: - case 0x57: - case 0x58: - case 0x59: - case 0x5A: - case 0x5B: - case 0x5C: - case 0x5D: - case 0x5E: - case 0x5F: - case 0x60: - case 0x61: - case 0x62: - case 0x63: - case 0x64: - case 0x65: - case 0x66: - case 0x67: - case 0x68: - case 0x69: - case 0x6A: - case 0x6B: - case 0x6C: - case 0x6D: - case 0x6E: - case 0x6F: - case 0x70: - case 0x71: - case 0x72: - case 0x73: - case 0x74: - case 0x75: - case 0x76: - case 0x77: - case 0x78: - case 0x79: - case 0x7A: - case 0x7B: - case 0x7C: - case 0x7D: - case 0x7E: - case 0x7F: - return static_cast<number_unsigned_t>(current); - - // fixmap - case 0x80: - case 0x81: - case 0x82: - case 0x83: - case 0x84: - case 0x85: - case 0x86: - case 0x87: - case 0x88: - case 0x89: - case 0x8A: - case 0x8B: - case 0x8C: - case 0x8D: - case 0x8E: - case 0x8F: - { - return get_msgpack_object(current & 0x0F); - } - - // fixarray - case 0x90: - case 0x91: - case 0x92: - case 0x93: - case 0x94: - case 0x95: - case 0x96: - case 0x97: - case 0x98: - case 0x99: - case 0x9A: - case 0x9B: - case 0x9C: - case 0x9D: - case 0x9E: - case 0x9F: - { - return get_msgpack_array(current & 0x0F); - } - - // fixstr - case 0xA0: - case 0xA1: - case 0xA2: - case 0xA3: - case 0xA4: - case 0xA5: - case 0xA6: - case 0xA7: - case 0xA8: - case 0xA9: - case 0xAA: - case 0xAB: - case 0xAC: - case 0xAD: - case 0xAE: - case 0xAF: - case 0xB0: - case 0xB1: - case 0xB2: - case 0xB3: - case 0xB4: - case 0xB5: - case 0xB6: - case 0xB7: - case 0xB8: - case 0xB9: - case 0xBA: - case 0xBB: - case 0xBC: - case 0xBD: - case 0xBE: - case 0xBF: - return get_msgpack_string(); - - case 0xC0: // nil - return value_t::null; - - case 0xC2: // false - return false; - - case 0xC3: // true - return true; - - case 0xCA: // float 32 - return get_number<float>(); - - case 0xCB: // float 64 - return get_number<double>(); - - case 0xCC: // uint 8 - return get_number<uint8_t>(); - - case 0xCD: // uint 16 - return get_number<uint16_t>(); - - case 0xCE: // uint 32 - return get_number<uint32_t>(); - - case 0xCF: // uint 64 - return get_number<uint64_t>(); - - case 0xD0: // int 8 - return get_number<int8_t>(); - - case 0xD1: // int 16 - return get_number<int16_t>(); - - case 0xD2: // int 32 - return get_number<int32_t>(); - - case 0xD3: // int 64 - return get_number<int64_t>(); - - case 0xD9: // str 8 - case 0xDA: // str 16 - case 0xDB: // str 32 - return get_msgpack_string(); - - case 0xDC: // array 16 - { - return get_msgpack_array(get_number<uint16_t>()); - } - - case 0xDD: // array 32 - { - return get_msgpack_array(get_number<uint32_t>()); - } - - case 0xDE: // map 16 - { - return get_msgpack_object(get_number<uint16_t>()); - } - - case 0xDF: // map 32 - { - return get_msgpack_object(get_number<uint32_t>()); - } - - // positive fixint - case 0xE0: - case 0xE1: - case 0xE2: - case 0xE3: - case 0xE4: - case 0xE5: - case 0xE6: - case 0xE7: - case 0xE8: - case 0xE9: - case 0xEA: - case 0xEB: - case 0xEC: - case 0xED: - case 0xEE: - case 0xEF: - case 0xF0: - case 0xF1: - case 0xF2: - case 0xF3: - case 0xF4: - case 0xF5: - case 0xF6: - case 0xF7: - case 0xF8: - case 0xF9: - case 0xFA: - case 0xFB: - case 0xFC: - case 0xFD: - case 0xFE: - case 0xFF: - return static_cast<int8_t>(current); - - default: // anything else - { - std::stringstream ss; - ss << std::setw(2) << std::uppercase << std::setfill('0') << std::hex << current; - JSON_THROW(parse_error::create(112, chars_read, - "error reading MessagePack; last byte: 0x" + ss.str())); - } - } - } - - /*! - @param[in] get_char whether a new character should be retrieved from the - input (true, default) or whether the last read - character should be considered instead - */ - BasicJsonType parse_ubjson_internal(const bool get_char = true) - { - return get_ubjson_value(get_char ? get_ignore_noop() : current); - } - - /*! - @brief get next character from the input - - This function provides the interface to the used input adapter. It does - not throw in case the input reached EOF, but returns a -'ve valued - `std::char_traits<char>::eof()` in that case. - - @return character read from the input - */ - int get() - { - ++chars_read; - return (current = ia->get_character()); - } - - /*! - @return character read from the input after ignoring all 'N' entries - */ - int get_ignore_noop() - { - do - { - get(); - } - while (current == 'N'); - - return current; - } - - /* - @brief read a number from the input - - @tparam NumberType the type of the number - - @return number of type @a NumberType - - @note This function needs to respect the system's endianess, because - bytes in CBOR and MessagePack are stored in network order (big - endian) and therefore need reordering on little endian systems. - - @throw parse_error.110 if input has less than `sizeof(NumberType)` bytes - */ - template<typename NumberType> NumberType get_number() - { - // step 1: read input into array with system's byte order - std::array<uint8_t, sizeof(NumberType)> vec; - for (std::size_t i = 0; i < sizeof(NumberType); ++i) - { - get(); - unexpect_eof(); - - // reverse byte order prior to conversion if necessary - if (is_little_endian) - { - vec[sizeof(NumberType) - i - 1] = static_cast<uint8_t>(current); - } - else - { - vec[i] = static_cast<uint8_t>(current); // LCOV_EXCL_LINE - } - } - - // step 2: convert array into number of type T and return - NumberType result; - std::memcpy(&result, vec.data(), sizeof(NumberType)); - return result; - } - - /*! - @brief create a string by reading characters from the input - - @param[in] len number of bytes to read - - @note We can not reserve @a len bytes for the result, because @a len - may be too large. Usually, @ref unexpect_eof() detects the end of - the input before we run out of string memory. - - @return string created by reading @a len bytes - - @throw parse_error.110 if input has less than @a len bytes - */ - template<typename NumberType> - string_t get_string(const NumberType len) - { - string_t result; - std::generate_n(std::back_inserter(result), len, [this]() - { - get(); - unexpect_eof(); - return static_cast<char>(current); - }); - return result; - } - - /*! - @brief reads a CBOR string - - This function first reads starting bytes to determine the expected - string length and then copies this number of bytes into a string. - Additionally, CBOR's strings with indefinite lengths are supported. - - @return string - - @throw parse_error.110 if input ended - @throw parse_error.113 if an unexpected byte is read - */ - string_t get_cbor_string() - { - unexpect_eof(); - - switch (current) - { - // UTF-8 string (0x00..0x17 bytes follow) - case 0x60: - case 0x61: - case 0x62: - case 0x63: - case 0x64: - case 0x65: - case 0x66: - case 0x67: - case 0x68: - case 0x69: - case 0x6A: - case 0x6B: - case 0x6C: - case 0x6D: - case 0x6E: - case 0x6F: - case 0x70: - case 0x71: - case 0x72: - case 0x73: - case 0x74: - case 0x75: - case 0x76: - case 0x77: - { - return get_string(current & 0x1F); - } - - case 0x78: // UTF-8 string (one-byte uint8_t for n follows) - { - return get_string(get_number<uint8_t>()); - } - - case 0x79: // UTF-8 string (two-byte uint16_t for n follow) - { - return get_string(get_number<uint16_t>()); - } - - case 0x7A: // UTF-8 string (four-byte uint32_t for n follow) - { - return get_string(get_number<uint32_t>()); - } - - case 0x7B: // UTF-8 string (eight-byte uint64_t for n follow) - { - return get_string(get_number<uint64_t>()); - } - - case 0x7F: // UTF-8 string (indefinite length) - { - string_t result; - while (get() != 0xFF) - { - result.append(get_cbor_string()); - } - return result; - } - - default: - { - std::stringstream ss; - ss << std::setw(2) << std::uppercase << std::setfill('0') << std::hex << current; - JSON_THROW(parse_error::create(113, chars_read, "expected a CBOR string; last byte: 0x" + ss.str())); - } - } - } - - template<typename NumberType> - BasicJsonType get_cbor_array(const NumberType len) - { - BasicJsonType result = value_t::array; - std::generate_n(std::back_inserter(*result.m_value.array), len, [this]() - { - return parse_cbor_internal(); - }); - return result; - } - - template<typename NumberType> - BasicJsonType get_cbor_object(const NumberType len) - { - BasicJsonType result = value_t::object; - std::generate_n(std::inserter(*result.m_value.object, - result.m_value.object->end()), - len, [this]() - { - get(); - auto key = get_cbor_string(); - auto val = parse_cbor_internal(); - return std::make_pair(std::move(key), std::move(val)); - }); - return result; - } - - /*! - @brief reads a MessagePack string - - This function first reads starting bytes to determine the expected - string length and then copies this number of bytes into a string. - - @return string - - @throw parse_error.110 if input ended - @throw parse_error.113 if an unexpected byte is read - */ - string_t get_msgpack_string() - { - unexpect_eof(); - - switch (current) - { - // fixstr - case 0xA0: - case 0xA1: - case 0xA2: - case 0xA3: - case 0xA4: - case 0xA5: - case 0xA6: - case 0xA7: - case 0xA8: - case 0xA9: - case 0xAA: - case 0xAB: - case 0xAC: - case 0xAD: - case 0xAE: - case 0xAF: - case 0xB0: - case 0xB1: - case 0xB2: - case 0xB3: - case 0xB4: - case 0xB5: - case 0xB6: - case 0xB7: - case 0xB8: - case 0xB9: - case 0xBA: - case 0xBB: - case 0xBC: - case 0xBD: - case 0xBE: - case 0xBF: - { - return get_string(current & 0x1F); - } - - case 0xD9: // str 8 - { - return get_string(get_number<uint8_t>()); - } - - case 0xDA: // str 16 - { - return get_string(get_number<uint16_t>()); - } - - case 0xDB: // str 32 - { - return get_string(get_number<uint32_t>()); - } - - default: - { - std::stringstream ss; - ss << std::setw(2) << std::uppercase << std::setfill('0') << std::hex << current; - JSON_THROW(parse_error::create(113, chars_read, - "expected a MessagePack string; last byte: 0x" + ss.str())); - } - } - } - - template<typename NumberType> - BasicJsonType get_msgpack_array(const NumberType len) - { - BasicJsonType result = value_t::array; - std::generate_n(std::back_inserter(*result.m_value.array), len, [this]() - { - return parse_msgpack_internal(); - }); - return result; - } - - template<typename NumberType> - BasicJsonType get_msgpack_object(const NumberType len) - { - BasicJsonType result = value_t::object; - std::generate_n(std::inserter(*result.m_value.object, - result.m_value.object->end()), - len, [this]() - { - get(); - auto key = get_msgpack_string(); - auto val = parse_msgpack_internal(); - return std::make_pair(std::move(key), std::move(val)); - }); - return result; - } - - /*! - @brief reads a UBJSON string - - This function is either called after reading the 'S' byte explicitly - indicating a string, or in case of an object key where the 'S' byte can be - left out. - - @param[in] get_char whether a new character should be retrieved from the - input (true, default) or whether the last read - character should be considered instead - - @return string - - @throw parse_error.110 if input ended - @throw parse_error.113 if an unexpected byte is read - */ - string_t get_ubjson_string(const bool get_char = true) - { - if (get_char) - { - get(); // TODO: may we ignore N here? - } - - unexpect_eof(); - - switch (current) - { - case 'U': - return get_string(get_number<uint8_t>()); - case 'i': - return get_string(get_number<int8_t>()); - case 'I': - return get_string(get_number<int16_t>()); - case 'l': - return get_string(get_number<int32_t>()); - case 'L': - return get_string(get_number<int64_t>()); - default: - std::stringstream ss; - ss << std::setw(2) << std::uppercase << std::setfill('0') << std::hex << current; - JSON_THROW(parse_error::create(113, chars_read, - "expected a UBJSON string; last byte: 0x" + ss.str())); - } - } - - /*! - @brief determine the type and size for a container - - In the optimized UBJSON format, a type and a size can be provided to allow - for a more compact representation. - - @return pair of the size and the type - */ - std::pair<std::size_t, int> get_ubjson_size_type() - { - std::size_t sz = string_t::npos; - int tc = 0; - - get_ignore_noop(); - - if (current == '$') - { - tc = get(); // must not ignore 'N', because 'N' maybe the type - unexpect_eof(); - - get_ignore_noop(); - if (current != '#') - { - std::stringstream ss; - ss << std::setw(2) << std::uppercase << std::setfill('0') << std::hex << current; - JSON_THROW(parse_error::create(112, chars_read, - "expected '#' after UBJSON type information; last byte: 0x" + ss.str())); - } - sz = parse_ubjson_internal(); - } - else if (current == '#') - { - sz = parse_ubjson_internal(); - } - - return std::make_pair(sz, tc); - } - - BasicJsonType get_ubjson_value(const int prefix) - { - switch (prefix) - { - case std::char_traits<char>::eof(): // EOF - JSON_THROW(parse_error::create(110, chars_read, "unexpected end of input")); - - case 'T': // true - return true; - case 'F': // false - return false; - - case 'Z': // null - return nullptr; - - case 'U': - return get_number<uint8_t>(); - case 'i': - return get_number<int8_t>(); - case 'I': - return get_number<int16_t>(); - case 'l': - return get_number<int32_t>(); - case 'L': - return get_number<int64_t>(); - case 'd': - return get_number<float>(); - case 'D': - return get_number<double>(); - - case 'C': // char - { - get(); - unexpect_eof(); - if (JSON_UNLIKELY(current > 127)) - { - std::stringstream ss; - ss << std::setw(2) << std::uppercase << std::setfill('0') << std::hex << current; - JSON_THROW(parse_error::create(113, chars_read, - "byte after 'C' must be in range 0x00..0x7F; last byte: 0x" + ss.str())); - } - return string_t(1, static_cast<char>(current)); - } - - case 'S': // string - return get_ubjson_string(); - - case '[': // array - return get_ubjson_array(); - - case '{': // object - return get_ubjson_object(); - - default: // anything else - std::stringstream ss; - ss << std::setw(2) << std::uppercase << std::setfill('0') << std::hex << current; - JSON_THROW(parse_error::create(112, chars_read, - "error reading UBJSON; last byte: 0x" + ss.str())); - } - } - - BasicJsonType get_ubjson_array() - { - BasicJsonType result = value_t::array; - const auto size_and_type = get_ubjson_size_type(); - - if (size_and_type.first != string_t::npos) - { - if (JSON_UNLIKELY(size_and_type.first > result.max_size())) - { - JSON_THROW(out_of_range::create(408, - "excessive array size: " + std::to_string(size_and_type.first))); - } - - if (size_and_type.second != 0) - { - if (size_and_type.second != 'N') - { - std::generate_n(std::back_inserter(*result.m_value.array), - size_and_type.first, [this, size_and_type]() - { - return get_ubjson_value(size_and_type.second); - }); - } - } - else - { - std::generate_n(std::back_inserter(*result.m_value.array), - size_and_type.first, [this]() - { - return parse_ubjson_internal(); - }); - } - } - else - { - while (current != ']') - { - result.push_back(parse_ubjson_internal(false)); - get_ignore_noop(); - } - } - - return result; - } - - BasicJsonType get_ubjson_object() - { - BasicJsonType result = value_t::object; - const auto size_and_type = get_ubjson_size_type(); - - if (size_and_type.first != string_t::npos) - { - if (JSON_UNLIKELY(size_and_type.first > result.max_size())) - { - JSON_THROW(out_of_range::create(408, - "excessive object size: " + std::to_string(size_and_type.first))); - } - - if (size_and_type.second != 0) - { - std::generate_n(std::inserter(*result.m_value.object, - result.m_value.object->end()), - size_and_type.first, [this, size_and_type]() - { - auto key = get_ubjson_string(); - auto val = get_ubjson_value(size_and_type.second); - return std::make_pair(std::move(key), std::move(val)); - }); - } - else - { - std::generate_n(std::inserter(*result.m_value.object, - result.m_value.object->end()), - size_and_type.first, [this]() - { - auto key = get_ubjson_string(); - auto val = parse_ubjson_internal(); - return std::make_pair(std::move(key), std::move(val)); - }); - } - } - else - { - while (current != '}') - { - auto key = get_ubjson_string(false); - result[std::move(key)] = parse_ubjson_internal(); - get_ignore_noop(); - } - } - - return result; - } - - /*! - @brief throw if end of input is not reached - @throw parse_error.110 if input not ended - */ - void expect_eof() const - { - if (JSON_UNLIKELY(current != std::char_traits<char>::eof())) - { - JSON_THROW(parse_error::create(110, chars_read, "expected end of input")); - } - } - - /*! - @briefthrow if end of input is reached - @throw parse_error.110 if input ended - */ - void unexpect_eof() const - { - if (JSON_UNLIKELY(current == std::char_traits<char>::eof())) - { - JSON_THROW(parse_error::create(110, chars_read, "unexpected end of input")); - } - } - - private: - /// input adapter - input_adapter_t ia = nullptr; - - /// the current character - int current = std::char_traits<char>::eof(); - - /// the number of characters read - std::size_t chars_read = 0; - - /// whether we can assume little endianess - const bool is_little_endian = little_endianess(); -}; -} -} - -// #include <nlohmann/detail/output/binary_writer.hpp> - - -#include <algorithm> // reverse -#include <array> // array -#include <cstdint> // uint8_t, uint16_t, uint32_t, uint64_t -#include <cstring> // memcpy -#include <limits> // numeric_limits - -// #include <nlohmann/detail/input/binary_reader.hpp> - -// #include <nlohmann/detail/output/output_adapters.hpp> - - -namespace nlohmann -{ -namespace detail -{ -/////////////////// -// binary writer // -/////////////////// - -/*! -@brief serialization to CBOR and MessagePack values -*/ -template<typename BasicJsonType, typename CharType> -class binary_writer -{ - public: - /*! - @brief create a binary writer - - @param[in] adapter output adapter to write to - */ - explicit binary_writer(output_adapter_t<CharType> adapter) : oa(adapter) - { - assert(oa); - } - - /*! - @brief[in] j JSON value to serialize - */ - void write_cbor(const BasicJsonType& j) - { - switch (j.type()) - { - case value_t::null: - { - oa->write_character(static_cast<CharType>(0xF6)); - break; - } - - case value_t::boolean: - { - oa->write_character(j.m_value.boolean - ? static_cast<CharType>(0xF5) - : static_cast<CharType>(0xF4)); - break; - } - - case value_t::number_integer: - { - if (j.m_value.number_integer >= 0) - { - // CBOR does not differentiate between positive signed - // integers and unsigned integers. Therefore, we used the - // code from the value_t::number_unsigned case here. - if (j.m_value.number_integer <= 0x17) - { - write_number(static_cast<uint8_t>(j.m_value.number_integer)); - } - else if (j.m_value.number_integer <= (std::numeric_limits<uint8_t>::max)()) - { - oa->write_character(static_cast<CharType>(0x18)); - write_number(static_cast<uint8_t>(j.m_value.number_integer)); - } - else if (j.m_value.number_integer <= (std::numeric_limits<uint16_t>::max)()) - { - oa->write_character(static_cast<CharType>(0x19)); - write_number(static_cast<uint16_t>(j.m_value.number_integer)); - } - else if (j.m_value.number_integer <= (std::numeric_limits<uint32_t>::max)()) - { - oa->write_character(static_cast<CharType>(0x1A)); - write_number(static_cast<uint32_t>(j.m_value.number_integer)); - } - else - { - oa->write_character(static_cast<CharType>(0x1B)); - write_number(static_cast<uint64_t>(j.m_value.number_integer)); - } - } - else - { - // The conversions below encode the sign in the first - // byte, and the value is converted to a positive number. - const auto positive_number = -1 - j.m_value.number_integer; - if (j.m_value.number_integer >= -24) - { - write_number(static_cast<uint8_t>(0x20 + positive_number)); - } - else if (positive_number <= (std::numeric_limits<uint8_t>::max)()) - { - oa->write_character(static_cast<CharType>(0x38)); - write_number(static_cast<uint8_t>(positive_number)); - } - else if (positive_number <= (std::numeric_limits<uint16_t>::max)()) - { - oa->write_character(static_cast<CharType>(0x39)); - write_number(static_cast<uint16_t>(positive_number)); - } - else if (positive_number <= (std::numeric_limits<uint32_t>::max)()) - { - oa->write_character(static_cast<CharType>(0x3A)); - write_number(static_cast<uint32_t>(positive_number)); - } - else - { - oa->write_character(static_cast<CharType>(0x3B)); - write_number(static_cast<uint64_t>(positive_number)); - } - } - break; - } - - case value_t::number_unsigned: - { - if (j.m_value.number_unsigned <= 0x17) - { - write_number(static_cast<uint8_t>(j.m_value.number_unsigned)); - } - else if (j.m_value.number_unsigned <= (std::numeric_limits<uint8_t>::max)()) - { - oa->write_character(static_cast<CharType>(0x18)); - write_number(static_cast<uint8_t>(j.m_value.number_unsigned)); - } - else if (j.m_value.number_unsigned <= (std::numeric_limits<uint16_t>::max)()) - { - oa->write_character(static_cast<CharType>(0x19)); - write_number(static_cast<uint16_t>(j.m_value.number_unsigned)); - } - else if (j.m_value.number_unsigned <= (std::numeric_limits<uint32_t>::max)()) - { - oa->write_character(static_cast<CharType>(0x1A)); - write_number(static_cast<uint32_t>(j.m_value.number_unsigned)); - } - else - { - oa->write_character(static_cast<CharType>(0x1B)); - write_number(static_cast<uint64_t>(j.m_value.number_unsigned)); - } - break; - } - - case value_t::number_float: // Double-Precision Float - { - oa->write_character(static_cast<CharType>(0xFB)); - write_number(j.m_value.number_float); - break; - } - - case value_t::string: - { - // step 1: write control byte and the string length - const auto N = j.m_value.string->size(); - if (N <= 0x17) - { - write_number(static_cast<uint8_t>(0x60 + N)); - } - else if (N <= (std::numeric_limits<uint8_t>::max)()) - { - oa->write_character(static_cast<CharType>(0x78)); - write_number(static_cast<uint8_t>(N)); - } - else if (N <= (std::numeric_limits<uint16_t>::max)()) - { - oa->write_character(static_cast<CharType>(0x79)); - write_number(static_cast<uint16_t>(N)); - } - else if (N <= (std::numeric_limits<uint32_t>::max)()) - { - oa->write_character(static_cast<CharType>(0x7A)); - write_number(static_cast<uint32_t>(N)); - } - // LCOV_EXCL_START - else if (N <= (std::numeric_limits<uint64_t>::max)()) - { - oa->write_character(static_cast<CharType>(0x7B)); - write_number(static_cast<uint64_t>(N)); - } - // LCOV_EXCL_STOP - - // step 2: write the string - oa->write_characters( - reinterpret_cast<const CharType*>(j.m_value.string->c_str()), - j.m_value.string->size()); - break; - } - - case value_t::array: - { - // step 1: write control byte and the array size - const auto N = j.m_value.array->size(); - if (N <= 0x17) - { - write_number(static_cast<uint8_t>(0x80 + N)); - } - else if (N <= (std::numeric_limits<uint8_t>::max)()) - { - oa->write_character(static_cast<CharType>(0x98)); - write_number(static_cast<uint8_t>(N)); - } - else if (N <= (std::numeric_limits<uint16_t>::max)()) - { - oa->write_character(static_cast<CharType>(0x99)); - write_number(static_cast<uint16_t>(N)); - } - else if (N <= (std::numeric_limits<uint32_t>::max)()) - { - oa->write_character(static_cast<CharType>(0x9A)); - write_number(static_cast<uint32_t>(N)); - } - // LCOV_EXCL_START - else if (N <= (std::numeric_limits<uint64_t>::max)()) - { - oa->write_character(static_cast<CharType>(0x9B)); - write_number(static_cast<uint64_t>(N)); - } - // LCOV_EXCL_STOP - - // step 2: write each element - for (const auto& el : *j.m_value.array) - { - write_cbor(el); - } - break; - } - - case value_t::object: - { - // step 1: write control byte and the object size - const auto N = j.m_value.object->size(); - if (N <= 0x17) - { - write_number(static_cast<uint8_t>(0xA0 + N)); - } - else if (N <= (std::numeric_limits<uint8_t>::max)()) - { - oa->write_character(static_cast<CharType>(0xB8)); - write_number(static_cast<uint8_t>(N)); - } - else if (N <= (std::numeric_limits<uint16_t>::max)()) - { - oa->write_character(static_cast<CharType>(0xB9)); - write_number(static_cast<uint16_t>(N)); - } - else if (N <= (std::numeric_limits<uint32_t>::max)()) - { - oa->write_character(static_cast<CharType>(0xBA)); - write_number(static_cast<uint32_t>(N)); - } - // LCOV_EXCL_START - else if (N <= (std::numeric_limits<uint64_t>::max)()) - { - oa->write_character(static_cast<CharType>(0xBB)); - write_number(static_cast<uint64_t>(N)); - } - // LCOV_EXCL_STOP - - // step 2: write each element - for (const auto& el : *j.m_value.object) - { - write_cbor(el.first); - write_cbor(el.second); - } - break; - } - - default: - break; - } - } - - /*! - @brief[in] j JSON value to serialize - */ - void write_msgpack(const BasicJsonType& j) - { - switch (j.type()) - { - case value_t::null: // nil - { - oa->write_character(static_cast<CharType>(0xC0)); - break; - } - - case value_t::boolean: // true and false - { - oa->write_character(j.m_value.boolean - ? static_cast<CharType>(0xC3) - : static_cast<CharType>(0xC2)); - break; - } - - case value_t::number_integer: - { - if (j.m_value.number_integer >= 0) - { - // MessagePack does not differentiate between positive - // signed integers and unsigned integers. Therefore, we used - // the code from the value_t::number_unsigned case here. - if (j.m_value.number_unsigned < 128) - { - // positive fixnum - write_number(static_cast<uint8_t>(j.m_value.number_integer)); - } - else if (j.m_value.number_unsigned <= (std::numeric_limits<uint8_t>::max)()) - { - // uint 8 - oa->write_character(static_cast<CharType>(0xCC)); - write_number(static_cast<uint8_t>(j.m_value.number_integer)); - } - else if (j.m_value.number_unsigned <= (std::numeric_limits<uint16_t>::max)()) - { - // uint 16 - oa->write_character(static_cast<CharType>(0xCD)); - write_number(static_cast<uint16_t>(j.m_value.number_integer)); - } - else if (j.m_value.number_unsigned <= (std::numeric_limits<uint32_t>::max)()) - { - // uint 32 - oa->write_character(static_cast<CharType>(0xCE)); - write_number(static_cast<uint32_t>(j.m_value.number_integer)); - } - else if (j.m_value.number_unsigned <= (std::numeric_limits<uint64_t>::max)()) - { - // uint 64 - oa->write_character(static_cast<CharType>(0xCF)); - write_number(static_cast<uint64_t>(j.m_value.number_integer)); - } - } - else - { - if (j.m_value.number_integer >= -32) - { - // negative fixnum - write_number(static_cast<int8_t>(j.m_value.number_integer)); - } - else if (j.m_value.number_integer >= (std::numeric_limits<int8_t>::min)() and - j.m_value.number_integer <= (std::numeric_limits<int8_t>::max)()) - { - // int 8 - oa->write_character(static_cast<CharType>(0xD0)); - write_number(static_cast<int8_t>(j.m_value.number_integer)); - } - else if (j.m_value.number_integer >= (std::numeric_limits<int16_t>::min)() and - j.m_value.number_integer <= (std::numeric_limits<int16_t>::max)()) - { - // int 16 - oa->write_character(static_cast<CharType>(0xD1)); - write_number(static_cast<int16_t>(j.m_value.number_integer)); - } - else if (j.m_value.number_integer >= (std::numeric_limits<int32_t>::min)() and - j.m_value.number_integer <= (std::numeric_limits<int32_t>::max)()) - { - // int 32 - oa->write_character(static_cast<CharType>(0xD2)); - write_number(static_cast<int32_t>(j.m_value.number_integer)); - } - else if (j.m_value.number_integer >= (std::numeric_limits<int64_t>::min)() and - j.m_value.number_integer <= (std::numeric_limits<int64_t>::max)()) - { - // int 64 - oa->write_character(static_cast<CharType>(0xD3)); - write_number(static_cast<int64_t>(j.m_value.number_integer)); - } - } - break; - } - - case value_t::number_unsigned: - { - if (j.m_value.number_unsigned < 128) - { - // positive fixnum - write_number(static_cast<uint8_t>(j.m_value.number_integer)); - } - else if (j.m_value.number_unsigned <= (std::numeric_limits<uint8_t>::max)()) - { - // uint 8 - oa->write_character(static_cast<CharType>(0xCC)); - write_number(static_cast<uint8_t>(j.m_value.number_integer)); - } - else if (j.m_value.number_unsigned <= (std::numeric_limits<uint16_t>::max)()) - { - // uint 16 - oa->write_character(static_cast<CharType>(0xCD)); - write_number(static_cast<uint16_t>(j.m_value.number_integer)); - } - else if (j.m_value.number_unsigned <= (std::numeric_limits<uint32_t>::max)()) - { - // uint 32 - oa->write_character(static_cast<CharType>(0xCE)); - write_number(static_cast<uint32_t>(j.m_value.number_integer)); - } - else if (j.m_value.number_unsigned <= (std::numeric_limits<uint64_t>::max)()) - { - // uint 64 - oa->write_character(static_cast<CharType>(0xCF)); - write_number(static_cast<uint64_t>(j.m_value.number_integer)); - } - break; - } - - case value_t::number_float: // float 64 - { - oa->write_character(static_cast<CharType>(0xCB)); - write_number(j.m_value.number_float); - break; - } - - case value_t::string: - { - // step 1: write control byte and the string length - const auto N = j.m_value.string->size(); - if (N <= 31) - { - // fixstr - write_number(static_cast<uint8_t>(0xA0 | N)); - } - else if (N <= (std::numeric_limits<uint8_t>::max)()) - { - // str 8 - oa->write_character(static_cast<CharType>(0xD9)); - write_number(static_cast<uint8_t>(N)); - } - else if (N <= (std::numeric_limits<uint16_t>::max)()) - { - // str 16 - oa->write_character(static_cast<CharType>(0xDA)); - write_number(static_cast<uint16_t>(N)); - } - else if (N <= (std::numeric_limits<uint32_t>::max)()) - { - // str 32 - oa->write_character(static_cast<CharType>(0xDB)); - write_number(static_cast<uint32_t>(N)); - } - - // step 2: write the string - oa->write_characters( - reinterpret_cast<const CharType*>(j.m_value.string->c_str()), - j.m_value.string->size()); - break; - } - - case value_t::array: - { - // step 1: write control byte and the array size - const auto N = j.m_value.array->size(); - if (N <= 15) - { - // fixarray - write_number(static_cast<uint8_t>(0x90 | N)); - } - else if (N <= (std::numeric_limits<uint16_t>::max)()) - { - // array 16 - oa->write_character(static_cast<CharType>(0xDC)); - write_number(static_cast<uint16_t>(N)); - } - else if (N <= (std::numeric_limits<uint32_t>::max)()) - { - // array 32 - oa->write_character(static_cast<CharType>(0xDD)); - write_number(static_cast<uint32_t>(N)); - } - - // step 2: write each element - for (const auto& el : *j.m_value.array) - { - write_msgpack(el); - } - break; - } - - case value_t::object: - { - // step 1: write control byte and the object size - const auto N = j.m_value.object->size(); - if (N <= 15) - { - // fixmap - write_number(static_cast<uint8_t>(0x80 | (N & 0xF))); - } - else if (N <= (std::numeric_limits<uint16_t>::max)()) - { - // map 16 - oa->write_character(static_cast<CharType>(0xDE)); - write_number(static_cast<uint16_t>(N)); - } - else if (N <= (std::numeric_limits<uint32_t>::max)()) - { - // map 32 - oa->write_character(static_cast<CharType>(0xDF)); - write_number(static_cast<uint32_t>(N)); - } - - // step 2: write each element - for (const auto& el : *j.m_value.object) - { - write_msgpack(el.first); - write_msgpack(el.second); - } - break; - } - - default: - break; - } - } - - /*! - @param[in] j JSON value to serialize - @param[in] use_count whether to use '#' prefixes (optimized format) - @param[in] use_type whether to use '$' prefixes (optimized format) - @param[in] add_prefix whether prefixes need to be used for this value - */ - void write_ubjson(const BasicJsonType& j, const bool use_count, - const bool use_type, const bool add_prefix = true) - { - switch (j.type()) - { - case value_t::null: - { - if (add_prefix) - { - oa->write_character(static_cast<CharType>('Z')); - } - break; - } - - case value_t::boolean: - { - if (add_prefix) - oa->write_character(j.m_value.boolean - ? static_cast<CharType>('T') - : static_cast<CharType>('F')); - break; - } - - case value_t::number_integer: - { - write_number_with_ubjson_prefix(j.m_value.number_integer, add_prefix); - break; - } - - case value_t::number_unsigned: - { - write_number_with_ubjson_prefix(j.m_value.number_unsigned, add_prefix); - break; - } - - case value_t::number_float: - { - write_number_with_ubjson_prefix(j.m_value.number_float, add_prefix); - break; - } - - case value_t::string: - { - if (add_prefix) - { - oa->write_character(static_cast<CharType>('S')); - } - write_number_with_ubjson_prefix(j.m_value.string->size(), true); - oa->write_characters( - reinterpret_cast<const CharType*>(j.m_value.string->c_str()), - j.m_value.string->size()); - break; - } - - case value_t::array: - { - if (add_prefix) - { - oa->write_character(static_cast<CharType>('[')); - } - - bool prefix_required = true; - if (use_type and not j.m_value.array->empty()) - { - assert(use_count); - const char first_prefix = ubjson_prefix(j.front()); - const bool same_prefix = std::all_of(j.begin() + 1, j.end(), - [this, first_prefix](const BasicJsonType & v) - { - return ubjson_prefix(v) == first_prefix; - }); - - if (same_prefix) - { - prefix_required = false; - oa->write_character(static_cast<CharType>('$')); - oa->write_character(static_cast<CharType>(first_prefix)); - } - } - - if (use_count) - { - oa->write_character(static_cast<CharType>('#')); - write_number_with_ubjson_prefix(j.m_value.array->size(), true); - } - - for (const auto& el : *j.m_value.array) - { - write_ubjson(el, use_count, use_type, prefix_required); - } - - if (not use_count) - { - oa->write_character(static_cast<CharType>(']')); - } - - break; - } - - case value_t::object: - { - if (add_prefix) - { - oa->write_character(static_cast<CharType>('{')); - } - - bool prefix_required = true; - if (use_type and not j.m_value.object->empty()) - { - assert(use_count); - const char first_prefix = ubjson_prefix(j.front()); - const bool same_prefix = std::all_of(j.begin(), j.end(), - [this, first_prefix](const BasicJsonType & v) - { - return ubjson_prefix(v) == first_prefix; - }); - - if (same_prefix) - { - prefix_required = false; - oa->write_character(static_cast<CharType>('$')); - oa->write_character(static_cast<CharType>(first_prefix)); - } - } - - if (use_count) - { - oa->write_character(static_cast<CharType>('#')); - write_number_with_ubjson_prefix(j.m_value.object->size(), true); - } - - for (const auto& el : *j.m_value.object) - { - write_number_with_ubjson_prefix(el.first.size(), true); - oa->write_characters( - reinterpret_cast<const CharType*>(el.first.c_str()), - el.first.size()); - write_ubjson(el.second, use_count, use_type, prefix_required); - } - - if (not use_count) - { - oa->write_character(static_cast<CharType>('}')); - } - - break; - } - - default: - break; - } - } - - private: - /* - @brief write a number to output input - - @param[in] n number of type @a NumberType - @tparam NumberType the type of the number - - @note This function needs to respect the system's endianess, because bytes - in CBOR, MessagePack, and UBJSON are stored in network order (big - endian) and therefore need reordering on little endian systems. - */ - template<typename NumberType> - void write_number(const NumberType n) - { - // step 1: write number to array of length NumberType - std::array<CharType, sizeof(NumberType)> vec; - std::memcpy(vec.data(), &n, sizeof(NumberType)); - - // step 2: write array to output (with possible reordering) - if (is_little_endian) - { - // reverse byte order prior to conversion if necessary - std::reverse(vec.begin(), vec.end()); - } - - oa->write_characters(vec.data(), sizeof(NumberType)); - } - - // UBJSON: write number (floating point) - template<typename NumberType, typename std::enable_if< - std::is_floating_point<NumberType>::value, int>::type = 0> - void write_number_with_ubjson_prefix(const NumberType n, - const bool add_prefix) - { - if (add_prefix) - { - oa->write_character(static_cast<CharType>('D')); // float64 - } - write_number(n); - } - - // UBJSON: write number (unsigned integer) - template<typename NumberType, typename std::enable_if< - std::is_unsigned<NumberType>::value, int>::type = 0> - void write_number_with_ubjson_prefix(const NumberType n, - const bool add_prefix) - { - if (n <= static_cast<uint64_t>((std::numeric_limits<int8_t>::max)())) - { - if (add_prefix) - { - oa->write_character(static_cast<CharType>('i')); // int8 - } - write_number(static_cast<uint8_t>(n)); - } - else if (n <= (std::numeric_limits<uint8_t>::max)()) - { - if (add_prefix) - { - oa->write_character(static_cast<CharType>('U')); // uint8 - } - write_number(static_cast<uint8_t>(n)); - } - else if (n <= static_cast<uint64_t>((std::numeric_limits<int16_t>::max)())) - { - if (add_prefix) - { - oa->write_character(static_cast<CharType>('I')); // int16 - } - write_number(static_cast<int16_t>(n)); - } - else if (n <= static_cast<uint64_t>((std::numeric_limits<int32_t>::max)())) - { - if (add_prefix) - { - oa->write_character(static_cast<CharType>('l')); // int32 - } - write_number(static_cast<int32_t>(n)); - } - else if (n <= static_cast<uint64_t>((std::numeric_limits<int64_t>::max)())) - { - if (add_prefix) - { - oa->write_character(static_cast<CharType>('L')); // int64 - } - write_number(static_cast<int64_t>(n)); - } - else - { - JSON_THROW(out_of_range::create(407, "number overflow serializing " + std::to_string(n))); - } - } - - // UBJSON: write number (signed integer) - template<typename NumberType, typename std::enable_if< - std::is_signed<NumberType>::value and - not std::is_floating_point<NumberType>::value, int>::type = 0> - void write_number_with_ubjson_prefix(const NumberType n, - const bool add_prefix) - { - if ((std::numeric_limits<int8_t>::min)() <= n and n <= (std::numeric_limits<int8_t>::max)()) - { - if (add_prefix) - { - oa->write_character(static_cast<CharType>('i')); // int8 - } - write_number(static_cast<int8_t>(n)); - } - else if (static_cast<int64_t>((std::numeric_limits<uint8_t>::min)()) <= n and n <= static_cast<int64_t>((std::numeric_limits<uint8_t>::max)())) - { - if (add_prefix) - { - oa->write_character(static_cast<CharType>('U')); // uint8 - } - write_number(static_cast<uint8_t>(n)); - } - else if ((std::numeric_limits<int16_t>::min)() <= n and n <= (std::numeric_limits<int16_t>::max)()) - { - if (add_prefix) - { - oa->write_character(static_cast<CharType>('I')); // int16 - } - write_number(static_cast<int16_t>(n)); - } - else if ((std::numeric_limits<int32_t>::min)() <= n and n <= (std::numeric_limits<int32_t>::max)()) - { - if (add_prefix) - { - oa->write_character(static_cast<CharType>('l')); // int32 - } - write_number(static_cast<int32_t>(n)); - } - else if ((std::numeric_limits<int64_t>::min)() <= n and n <= (std::numeric_limits<int64_t>::max)()) - { - if (add_prefix) - { - oa->write_character(static_cast<CharType>('L')); // int64 - } - write_number(static_cast<int64_t>(n)); - } - // LCOV_EXCL_START - else - { - JSON_THROW(out_of_range::create(407, "number overflow serializing " + std::to_string(n))); - } - // LCOV_EXCL_STOP - } - - /*! - @brief determine the type prefix of container values - - @note This function does not need to be 100% accurate when it comes to - integer limits. In case a number exceeds the limits of int64_t, - this will be detected by a later call to function - write_number_with_ubjson_prefix. Therefore, we return 'L' for any - value that does not fit the previous limits. - */ - char ubjson_prefix(const BasicJsonType& j) const noexcept - { - switch (j.type()) - { - case value_t::null: - return 'Z'; - - case value_t::boolean: - return j.m_value.boolean ? 'T' : 'F'; - - case value_t::number_integer: - { - if ((std::numeric_limits<int8_t>::min)() <= j.m_value.number_integer and j.m_value.number_integer <= (std::numeric_limits<int8_t>::max)()) - { - return 'i'; - } - else if ((std::numeric_limits<uint8_t>::min)() <= j.m_value.number_integer and j.m_value.number_integer <= (std::numeric_limits<uint8_t>::max)()) - { - return 'U'; - } - else if ((std::numeric_limits<int16_t>::min)() <= j.m_value.number_integer and j.m_value.number_integer <= (std::numeric_limits<int16_t>::max)()) - { - return 'I'; - } - else if ((std::numeric_limits<int32_t>::min)() <= j.m_value.number_integer and j.m_value.number_integer <= (std::numeric_limits<int32_t>::max)()) - { - return 'l'; - } - else // no check and assume int64_t (see note above) - { - return 'L'; - } - } - - case value_t::number_unsigned: - { - if (j.m_value.number_unsigned <= (std::numeric_limits<int8_t>::max)()) - { - return 'i'; - } - else if (j.m_value.number_unsigned <= (std::numeric_limits<uint8_t>::max)()) - { - return 'U'; - } - else if (j.m_value.number_unsigned <= (std::numeric_limits<int16_t>::max)()) - { - return 'I'; - } - else if (j.m_value.number_unsigned <= (std::numeric_limits<int32_t>::max)()) - { - return 'l'; - } - else // no check and assume int64_t (see note above) - { - return 'L'; - } - } - - case value_t::number_float: - return 'D'; - - case value_t::string: - return 'S'; - - case value_t::array: - return '['; - - case value_t::object: - return '{'; - - default: // discarded values - return 'N'; - } - } - - private: - /// whether we can assume little endianess - const bool is_little_endian = binary_reader<BasicJsonType>::little_endianess(); - - /// the output - output_adapter_t<CharType> oa = nullptr; -}; -} -} - -// #include <nlohmann/detail/output/serializer.hpp> - - -#include <algorithm> // reverse, remove, fill, find, none_of -#include <array> // array -#include <cassert> // assert -#include <ciso646> // and, or -#include <clocale> // localeconv, lconv -#include <cmath> // labs, isfinite, isnan, signbit -#include <cstddef> // size_t, ptrdiff_t -#include <cstdint> // uint8_t -#include <cstdio> // snprintf -#include <iomanip> // setfill -#include <iterator> // next -#include <limits> // numeric_limits -#include <string> // string -#include <sstream> // stringstream -#include <type_traits> // is_same - -// #include <nlohmann/detail/exceptions.hpp> - -// #include <nlohmann/detail/conversions/to_chars.hpp> - - -#include <cassert> // assert -#include <ciso646> // or, and, not -#include <cmath> // signbit, isfinite -#include <cstdint> // intN_t, uintN_t -#include <cstring> // memcpy, memmove - -namespace nlohmann -{ -namespace detail -{ - -/*! -@brief implements the Grisu2 algorithm for binary to decimal floating-point -conversion. - -This implementation is a slightly modified version of the reference -implementation which may be obtained from -http://florian.loitsch.com/publications (bench.tar.gz). - -The code is distributed under the MIT license, Copyright (c) 2009 Florian Loitsch. - -For a detailed description of the algorithm see: - -[1] Loitsch, "Printing Floating-Point Numbers Quickly and Accurately with - Integers", Proceedings of the ACM SIGPLAN 2010 Conference on Programming - Language Design and Implementation, PLDI 2010 -[2] Burger, Dybvig, "Printing Floating-Point Numbers Quickly and Accurately", - Proceedings of the ACM SIGPLAN 1996 Conference on Programming Language - Design and Implementation, PLDI 1996 -*/ -namespace dtoa_impl -{ - -template <typename Target, typename Source> -Target reinterpret_bits(const Source source) -{ - static_assert(sizeof(Target) == sizeof(Source), "size mismatch"); - - Target target; - std::memcpy(&target, &source, sizeof(Source)); - return target; -} - -struct diyfp // f * 2^e -{ - static constexpr int kPrecision = 64; // = q - - uint64_t f; - int e; - - constexpr diyfp() noexcept : f(0), e(0) {} - constexpr diyfp(uint64_t f_, int e_) noexcept : f(f_), e(e_) {} - - /*! - @brief returns x - y - @pre x.e == y.e and x.f >= y.f - */ - static diyfp sub(const diyfp& x, const diyfp& y) noexcept - { - assert(x.e == y.e); - assert(x.f >= y.f); - - return diyfp(x.f - y.f, x.e); - } - - /*! - @brief returns x * y - @note The result is rounded. (Only the upper q bits are returned.) - */ - static diyfp mul(const diyfp& x, const diyfp& y) noexcept - { - static_assert(kPrecision == 64, "internal error"); - - // Computes: - // f = round((x.f * y.f) / 2^q) - // e = x.e + y.e + q - - // Emulate the 64-bit * 64-bit multiplication: - // - // p = u * v - // = (u_lo + 2^32 u_hi) (v_lo + 2^32 v_hi) - // = (u_lo v_lo ) + 2^32 ((u_lo v_hi ) + (u_hi v_lo )) + 2^64 (u_hi v_hi ) - // = (p0 ) + 2^32 ((p1 ) + (p2 )) + 2^64 (p3 ) - // = (p0_lo + 2^32 p0_hi) + 2^32 ((p1_lo + 2^32 p1_hi) + (p2_lo + 2^32 p2_hi)) + 2^64 (p3 ) - // = (p0_lo ) + 2^32 (p0_hi + p1_lo + p2_lo ) + 2^64 (p1_hi + p2_hi + p3) - // = (p0_lo ) + 2^32 (Q ) + 2^64 (H ) - // = (p0_lo ) + 2^32 (Q_lo + 2^32 Q_hi ) + 2^64 (H ) - // - // (Since Q might be larger than 2^32 - 1) - // - // = (p0_lo + 2^32 Q_lo) + 2^64 (Q_hi + H) - // - // (Q_hi + H does not overflow a 64-bit int) - // - // = p_lo + 2^64 p_hi - - const uint64_t u_lo = x.f & 0xFFFFFFFF; - const uint64_t u_hi = x.f >> 32; - const uint64_t v_lo = y.f & 0xFFFFFFFF; - const uint64_t v_hi = y.f >> 32; - - const uint64_t p0 = u_lo * v_lo; - const uint64_t p1 = u_lo * v_hi; - const uint64_t p2 = u_hi * v_lo; - const uint64_t p3 = u_hi * v_hi; - - const uint64_t p0_hi = p0 >> 32; - const uint64_t p1_lo = p1 & 0xFFFFFFFF; - const uint64_t p1_hi = p1 >> 32; - const uint64_t p2_lo = p2 & 0xFFFFFFFF; - const uint64_t p2_hi = p2 >> 32; - - uint64_t Q = p0_hi + p1_lo + p2_lo; - - // The full product might now be computed as - // - // p_hi = p3 + p2_hi + p1_hi + (Q >> 32) - // p_lo = p0_lo + (Q << 32) - // - // But in this particular case here, the full p_lo is not required. - // Effectively we only need to add the highest bit in p_lo to p_hi (and - // Q_hi + 1 does not overflow). - - Q += uint64_t{1} << (64 - 32 - 1); // round, ties up - - const uint64_t h = p3 + p2_hi + p1_hi + (Q >> 32); - - return diyfp(h, x.e + y.e + 64); - } - - /*! - @brief normalize x such that the significand is >= 2^(q-1) - @pre x.f != 0 - */ - static diyfp normalize(diyfp x) noexcept - { - assert(x.f != 0); - - while ((x.f >> 63) == 0) - { - x.f <<= 1; - x.e--; - } - - return x; - } - - /*! - @brief normalize x such that the result has the exponent E - @pre e >= x.e and the upper e - x.e bits of x.f must be zero. - */ - static diyfp normalize_to(const diyfp& x, const int target_exponent) noexcept - { - const int delta = x.e - target_exponent; - - assert(delta >= 0); - assert(((x.f << delta) >> delta) == x.f); - - return diyfp(x.f << delta, target_exponent); - } -}; - -struct boundaries -{ - diyfp w; - diyfp minus; - diyfp plus; -}; - -/*! -Compute the (normalized) diyfp representing the input number 'value' and its -boundaries. - -@pre value must be finite and positive -*/ -template <typename FloatType> -boundaries compute_boundaries(FloatType value) -{ - assert(std::isfinite(value)); - assert(value > 0); - - // Convert the IEEE representation into a diyfp. - // - // If v is denormal: - // value = 0.F * 2^(1 - bias) = ( F) * 2^(1 - bias - (p-1)) - // If v is normalized: - // value = 1.F * 2^(E - bias) = (2^(p-1) + F) * 2^(E - bias - (p-1)) - - static_assert(std::numeric_limits<FloatType>::is_iec559, - "internal error: dtoa_short requires an IEEE-754 floating-point implementation"); - - constexpr int kPrecision = std::numeric_limits<FloatType>::digits; // = p (includes the hidden bit) - constexpr int kBias = std::numeric_limits<FloatType>::max_exponent - 1 + (kPrecision - 1); - constexpr int kMinExp = 1 - kBias; - constexpr uint64_t kHiddenBit = uint64_t{1} << (kPrecision - 1); // = 2^(p-1) - - using bits_type = typename std::conditional< kPrecision == 24, uint32_t, uint64_t >::type; - - const uint64_t bits = reinterpret_bits<bits_type>(value); - const uint64_t E = bits >> (kPrecision - 1); - const uint64_t F = bits & (kHiddenBit - 1); - - const bool is_denormal = (E == 0); - const diyfp v = is_denormal - ? diyfp(F, kMinExp) - : diyfp(F + kHiddenBit, static_cast<int>(E) - kBias); - - // Compute the boundaries m- and m+ of the floating-point value - // v = f * 2^e. - // - // Determine v- and v+, the floating-point predecessor and successor if v, - // respectively. - // - // v- = v - 2^e if f != 2^(p-1) or e == e_min (A) - // = v - 2^(e-1) if f == 2^(p-1) and e > e_min (B) - // - // v+ = v + 2^e - // - // Let m- = (v- + v) / 2 and m+ = (v + v+) / 2. All real numbers _strictly_ - // between m- and m+ round to v, regardless of how the input rounding - // algorithm breaks ties. - // - // ---+-------------+-------------+-------------+-------------+--- (A) - // v- m- v m+ v+ - // - // -----------------+------+------+-------------+-------------+--- (B) - // v- m- v m+ v+ - - const bool lower_boundary_is_closer = (F == 0 and E > 1); - const diyfp m_plus = diyfp(2 * v.f + 1, v.e - 1); - const diyfp m_minus = lower_boundary_is_closer - ? diyfp(4 * v.f - 1, v.e - 2) // (B) - : diyfp(2 * v.f - 1, v.e - 1); // (A) - - // Determine the normalized w+ = m+. - const diyfp w_plus = diyfp::normalize(m_plus); - - // Determine w- = m- such that e_(w-) = e_(w+). - const diyfp w_minus = diyfp::normalize_to(m_minus, w_plus.e); - - return {diyfp::normalize(v), w_minus, w_plus}; -} - -// Given normalized diyfp w, Grisu needs to find a (normalized) cached -// power-of-ten c, such that the exponent of the product c * w = f * 2^e lies -// within a certain range [alpha, gamma] (Definition 3.2 from [1]) -// -// alpha <= e = e_c + e_w + q <= gamma -// -// or -// -// f_c * f_w * 2^alpha <= f_c 2^(e_c) * f_w 2^(e_w) * 2^q -// <= f_c * f_w * 2^gamma -// -// Since c and w are normalized, i.e. 2^(q-1) <= f < 2^q, this implies -// -// 2^(q-1) * 2^(q-1) * 2^alpha <= c * w * 2^q < 2^q * 2^q * 2^gamma -// -// or -// -// 2^(q - 2 + alpha) <= c * w < 2^(q + gamma) -// -// The choice of (alpha,gamma) determines the size of the table and the form of -// the digit generation procedure. Using (alpha,gamma)=(-60,-32) works out well -// in practice: -// -// The idea is to cut the number c * w = f * 2^e into two parts, which can be -// processed independently: An integral part p1, and a fractional part p2: -// -// f * 2^e = ( (f div 2^-e) * 2^-e + (f mod 2^-e) ) * 2^e -// = (f div 2^-e) + (f mod 2^-e) * 2^e -// = p1 + p2 * 2^e -// -// The conversion of p1 into decimal form requires a series of divisions and -// modulos by (a power of) 10. These operations are faster for 32-bit than for -// 64-bit integers, so p1 should ideally fit into a 32-bit integer. This can be -// achieved by choosing -// -// -e >= 32 or e <= -32 := gamma -// -// In order to convert the fractional part -// -// p2 * 2^e = p2 / 2^-e = d[-1] / 10^1 + d[-2] / 10^2 + ... -// -// into decimal form, the fraction is repeatedly multiplied by 10 and the digits -// d[-i] are extracted in order: -// -// (10 * p2) div 2^-e = d[-1] -// (10 * p2) mod 2^-e = d[-2] / 10^1 + ... -// -// The multiplication by 10 must not overflow. It is sufficient to choose -// -// 10 * p2 < 16 * p2 = 2^4 * p2 <= 2^64. -// -// Since p2 = f mod 2^-e < 2^-e, -// -// -e <= 60 or e >= -60 := alpha - -constexpr int kAlpha = -60; -constexpr int kGamma = -32; - -struct cached_power // c = f * 2^e ~= 10^k -{ - uint64_t f; - int e; - int k; -}; - -/*! -For a normalized diyfp w = f * 2^e, this function returns a (normalized) cached -power-of-ten c = f_c * 2^e_c, such that the exponent of the product w * c -satisfies (Definition 3.2 from [1]) - - alpha <= e_c + e + q <= gamma. -*/ -inline cached_power get_cached_power_for_binary_exponent(int e) -{ - // Now - // - // alpha <= e_c + e + q <= gamma (1) - // ==> f_c * 2^alpha <= c * 2^e * 2^q - // - // and since the c's are normalized, 2^(q-1) <= f_c, - // - // ==> 2^(q - 1 + alpha) <= c * 2^(e + q) - // ==> 2^(alpha - e - 1) <= c - // - // If c were an exakt power of ten, i.e. c = 10^k, one may determine k as - // - // k = ceil( log_10( 2^(alpha - e - 1) ) ) - // = ceil( (alpha - e - 1) * log_10(2) ) - // - // From the paper: - // "In theory the result of the procedure could be wrong since c is rounded, - // and the computation itself is approximated [...]. In practice, however, - // this simple function is sufficient." - // - // For IEEE double precision floating-point numbers converted into - // normalized diyfp's w = f * 2^e, with q = 64, - // - // e >= -1022 (min IEEE exponent) - // -52 (p - 1) - // -52 (p - 1, possibly normalize denormal IEEE numbers) - // -11 (normalize the diyfp) - // = -1137 - // - // and - // - // e <= +1023 (max IEEE exponent) - // -52 (p - 1) - // -11 (normalize the diyfp) - // = 960 - // - // This binary exponent range [-1137,960] results in a decimal exponent - // range [-307,324]. One does not need to store a cached power for each - // k in this range. For each such k it suffices to find a cached power - // such that the exponent of the product lies in [alpha,gamma]. - // This implies that the difference of the decimal exponents of adjacent - // table entries must be less than or equal to - // - // floor( (gamma - alpha) * log_10(2) ) = 8. - // - // (A smaller distance gamma-alpha would require a larger table.) - - // NB: - // Actually this function returns c, such that -60 <= e_c + e + 64 <= -34. - - constexpr int kCachedPowersSize = 79; - constexpr int kCachedPowersMinDecExp = -300; - constexpr int kCachedPowersDecStep = 8; - - static constexpr cached_power kCachedPowers[] = - { - { 0xAB70FE17C79AC6CA, -1060, -300 }, - { 0xFF77B1FCBEBCDC4F, -1034, -292 }, - { 0xBE5691EF416BD60C, -1007, -284 }, - { 0x8DD01FAD907FFC3C, -980, -276 }, - { 0xD3515C2831559A83, -954, -268 }, - { 0x9D71AC8FADA6C9B5, -927, -260 }, - { 0xEA9C227723EE8BCB, -901, -252 }, - { 0xAECC49914078536D, -874, -244 }, - { 0x823C12795DB6CE57, -847, -236 }, - { 0xC21094364DFB5637, -821, -228 }, - { 0x9096EA6F3848984F, -794, -220 }, - { 0xD77485CB25823AC7, -768, -212 }, - { 0xA086CFCD97BF97F4, -741, -204 }, - { 0xEF340A98172AACE5, -715, -196 }, - { 0xB23867FB2A35B28E, -688, -188 }, - { 0x84C8D4DFD2C63F3B, -661, -180 }, - { 0xC5DD44271AD3CDBA, -635, -172 }, - { 0x936B9FCEBB25C996, -608, -164 }, - { 0xDBAC6C247D62A584, -582, -156 }, - { 0xA3AB66580D5FDAF6, -555, -148 }, - { 0xF3E2F893DEC3F126, -529, -140 }, - { 0xB5B5ADA8AAFF80B8, -502, -132 }, - { 0x87625F056C7C4A8B, -475, -124 }, - { 0xC9BCFF6034C13053, -449, -116 }, - { 0x964E858C91BA2655, -422, -108 }, - { 0xDFF9772470297EBD, -396, -100 }, - { 0xA6DFBD9FB8E5B88F, -369, -92 }, - { 0xF8A95FCF88747D94, -343, -84 }, - { 0xB94470938FA89BCF, -316, -76 }, - { 0x8A08F0F8BF0F156B, -289, -68 }, - { 0xCDB02555653131B6, -263, -60 }, - { 0x993FE2C6D07B7FAC, -236, -52 }, - { 0xE45C10C42A2B3B06, -210, -44 }, - { 0xAA242499697392D3, -183, -36 }, - { 0xFD87B5F28300CA0E, -157, -28 }, - { 0xBCE5086492111AEB, -130, -20 }, - { 0x8CBCCC096F5088CC, -103, -12 }, - { 0xD1B71758E219652C, -77, -4 }, - { 0x9C40000000000000, -50, 4 }, - { 0xE8D4A51000000000, -24, 12 }, - { 0xAD78EBC5AC620000, 3, 20 }, - { 0x813F3978F8940984, 30, 28 }, - { 0xC097CE7BC90715B3, 56, 36 }, - { 0x8F7E32CE7BEA5C70, 83, 44 }, - { 0xD5D238A4ABE98068, 109, 52 }, - { 0x9F4F2726179A2245, 136, 60 }, - { 0xED63A231D4C4FB27, 162, 68 }, - { 0xB0DE65388CC8ADA8, 189, 76 }, - { 0x83C7088E1AAB65DB, 216, 84 }, - { 0xC45D1DF942711D9A, 242, 92 }, - { 0x924D692CA61BE758, 269, 100 }, - { 0xDA01EE641A708DEA, 295, 108 }, - { 0xA26DA3999AEF774A, 322, 116 }, - { 0xF209787BB47D6B85, 348, 124 }, - { 0xB454E4A179DD1877, 375, 132 }, - { 0x865B86925B9BC5C2, 402, 140 }, - { 0xC83553C5C8965D3D, 428, 148 }, - { 0x952AB45CFA97A0B3, 455, 156 }, - { 0xDE469FBD99A05FE3, 481, 164 }, - { 0xA59BC234DB398C25, 508, 172 }, - { 0xF6C69A72A3989F5C, 534, 180 }, - { 0xB7DCBF5354E9BECE, 561, 188 }, - { 0x88FCF317F22241E2, 588, 196 }, - { 0xCC20CE9BD35C78A5, 614, 204 }, - { 0x98165AF37B2153DF, 641, 212 }, - { 0xE2A0B5DC971F303A, 667, 220 }, - { 0xA8D9D1535CE3B396, 694, 228 }, - { 0xFB9B7CD9A4A7443C, 720, 236 }, - { 0xBB764C4CA7A44410, 747, 244 }, - { 0x8BAB8EEFB6409C1A, 774, 252 }, - { 0xD01FEF10A657842C, 800, 260 }, - { 0x9B10A4E5E9913129, 827, 268 }, - { 0xE7109BFBA19C0C9D, 853, 276 }, - { 0xAC2820D9623BF429, 880, 284 }, - { 0x80444B5E7AA7CF85, 907, 292 }, - { 0xBF21E44003ACDD2D, 933, 300 }, - { 0x8E679C2F5E44FF8F, 960, 308 }, - { 0xD433179D9C8CB841, 986, 316 }, - { 0x9E19DB92B4E31BA9, 1013, 324 }, - }; - - // This computation gives exactly the same results for k as - // k = ceil((kAlpha - e - 1) * 0.30102999566398114) - // for |e| <= 1500, but doesn't require floating-point operations. - // NB: log_10(2) ~= 78913 / 2^18 - assert(e >= -1500); - assert(e <= 1500); - const int f = kAlpha - e - 1; - const int k = (f * 78913) / (1 << 18) + (f > 0); - - const int index = (-kCachedPowersMinDecExp + k + (kCachedPowersDecStep - 1)) / kCachedPowersDecStep; - assert(index >= 0); - assert(index < kCachedPowersSize); - static_cast<void>(kCachedPowersSize); // Fix warning. - - const cached_power cached = kCachedPowers[index]; - assert(kAlpha <= cached.e + e + 64); - assert(kGamma >= cached.e + e + 64); - - return cached; -} - -/*! -For n != 0, returns k, such that pow10 := 10^(k-1) <= n < 10^k. -For n == 0, returns 1 and sets pow10 := 1. -*/ -inline int find_largest_pow10(const uint32_t n, uint32_t& pow10) -{ - // LCOV_EXCL_START - if (n >= 1000000000) - { - pow10 = 1000000000; - return 10; - } - // LCOV_EXCL_STOP - else if (n >= 100000000) - { - pow10 = 100000000; - return 9; - } - else if (n >= 10000000) - { - pow10 = 10000000; - return 8; - } - else if (n >= 1000000) - { - pow10 = 1000000; - return 7; - } - else if (n >= 100000) - { - pow10 = 100000; - return 6; - } - else if (n >= 10000) - { - pow10 = 10000; - return 5; - } - else if (n >= 1000) - { - pow10 = 1000; - return 4; - } - else if (n >= 100) - { - pow10 = 100; - return 3; - } - else if (n >= 10) - { - pow10 = 10; - return 2; - } - else - { - pow10 = 1; - return 1; - } -} - -inline void grisu2_round(char* buf, int len, uint64_t dist, uint64_t delta, - uint64_t rest, uint64_t ten_k) -{ - assert(len >= 1); - assert(dist <= delta); - assert(rest <= delta); - assert(ten_k > 0); - - // <--------------------------- delta ----> - // <---- dist ---------> - // --------------[------------------+-------------------]-------------- - // M- w M+ - // - // ten_k - // <------> - // <---- rest ----> - // --------------[------------------+----+--------------]-------------- - // w V - // = buf * 10^k - // - // ten_k represents a unit-in-the-last-place in the decimal representation - // stored in buf. - // Decrement buf by ten_k while this takes buf closer to w. - - // The tests are written in this order to avoid overflow in unsigned - // integer arithmetic. - - while (rest < dist - and delta - rest >= ten_k - and (rest + ten_k < dist or dist - rest > rest + ten_k - dist)) - { - assert(buf[len - 1] != '0'); - buf[len - 1]--; - rest += ten_k; - } -} - -/*! -Generates V = buffer * 10^decimal_exponent, such that M- <= V <= M+. -M- and M+ must be normalized and share the same exponent -60 <= e <= -32. -*/ -inline void grisu2_digit_gen(char* buffer, int& length, int& decimal_exponent, - diyfp M_minus, diyfp w, diyfp M_plus) -{ - static_assert(kAlpha >= -60, "internal error"); - static_assert(kGamma <= -32, "internal error"); - - // Generates the digits (and the exponent) of a decimal floating-point - // number V = buffer * 10^decimal_exponent in the range [M-, M+]. The diyfp's - // w, M- and M+ share the same exponent e, which satisfies alpha <= e <= gamma. - // - // <--------------------------- delta ----> - // <---- dist ---------> - // --------------[------------------+-------------------]-------------- - // M- w M+ - // - // Grisu2 generates the digits of M+ from left to right and stops as soon as - // V is in [M-,M+]. - - assert(M_plus.e >= kAlpha); - assert(M_plus.e <= kGamma); - - uint64_t delta = diyfp::sub(M_plus, M_minus).f; // (significand of (M+ - M-), implicit exponent is e) - uint64_t dist = diyfp::sub(M_plus, w ).f; // (significand of (M+ - w ), implicit exponent is e) - - // Split M+ = f * 2^e into two parts p1 and p2 (note: e < 0): - // - // M+ = f * 2^e - // = ((f div 2^-e) * 2^-e + (f mod 2^-e)) * 2^e - // = ((p1 ) * 2^-e + (p2 )) * 2^e - // = p1 + p2 * 2^e - - const diyfp one(uint64_t{1} << -M_plus.e, M_plus.e); - - uint32_t p1 = static_cast<uint32_t>(M_plus.f >> -one.e); // p1 = f div 2^-e (Since -e >= 32, p1 fits into a 32-bit int.) - uint64_t p2 = M_plus.f & (one.f - 1); // p2 = f mod 2^-e - - // 1) - // - // Generate the digits of the integral part p1 = d[n-1]...d[1]d[0] - - assert(p1 > 0); - - uint32_t pow10; - const int k = find_largest_pow10(p1, pow10); - - // 10^(k-1) <= p1 < 10^k, pow10 = 10^(k-1) - // - // p1 = (p1 div 10^(k-1)) * 10^(k-1) + (p1 mod 10^(k-1)) - // = (d[k-1] ) * 10^(k-1) + (p1 mod 10^(k-1)) - // - // M+ = p1 + p2 * 2^e - // = d[k-1] * 10^(k-1) + (p1 mod 10^(k-1)) + p2 * 2^e - // = d[k-1] * 10^(k-1) + ((p1 mod 10^(k-1)) * 2^-e + p2) * 2^e - // = d[k-1] * 10^(k-1) + ( rest) * 2^e - // - // Now generate the digits d[n] of p1 from left to right (n = k-1,...,0) - // - // p1 = d[k-1]...d[n] * 10^n + d[n-1]...d[0] - // - // but stop as soon as - // - // rest * 2^e = (d[n-1]...d[0] * 2^-e + p2) * 2^e <= delta * 2^e - - int n = k; - while (n > 0) - { - // Invariants: - // M+ = buffer * 10^n + (p1 + p2 * 2^e) (buffer = 0 for n = k) - // pow10 = 10^(n-1) <= p1 < 10^n - // - const uint32_t d = p1 / pow10; // d = p1 div 10^(n-1) - const uint32_t r = p1 % pow10; // r = p1 mod 10^(n-1) - // - // M+ = buffer * 10^n + (d * 10^(n-1) + r) + p2 * 2^e - // = (buffer * 10 + d) * 10^(n-1) + (r + p2 * 2^e) - // - assert(d <= 9); - buffer[length++] = static_cast<char>('0' + d); // buffer := buffer * 10 + d - // - // M+ = buffer * 10^(n-1) + (r + p2 * 2^e) - // - p1 = r; - n--; - // - // M+ = buffer * 10^n + (p1 + p2 * 2^e) - // pow10 = 10^n - // - - // Now check if enough digits have been generated. - // Compute - // - // p1 + p2 * 2^e = (p1 * 2^-e + p2) * 2^e = rest * 2^e - // - // Note: - // Since rest and delta share the same exponent e, it suffices to - // compare the significands. - const uint64_t rest = (uint64_t{p1} << -one.e) + p2; - if (rest <= delta) - { - // V = buffer * 10^n, with M- <= V <= M+. - - decimal_exponent += n; - - // We may now just stop. But instead look if the buffer could be - // decremented to bring V closer to w. - // - // pow10 = 10^n is now 1 ulp in the decimal representation V. - // The rounding procedure works with diyfp's with an implicit - // exponent of e. - // - // 10^n = (10^n * 2^-e) * 2^e = ulp * 2^e - // - const uint64_t ten_n = uint64_t{pow10} << -one.e; - grisu2_round(buffer, length, dist, delta, rest, ten_n); - - return; - } - - pow10 /= 10; - // - // pow10 = 10^(n-1) <= p1 < 10^n - // Invariants restored. - } - - // 2) - // - // The digits of the integral part have been generated: - // - // M+ = d[k-1]...d[1]d[0] + p2 * 2^e - // = buffer + p2 * 2^e - // - // Now generate the digits of the fractional part p2 * 2^e. - // - // Note: - // No decimal point is generated: the exponent is adjusted instead. - // - // p2 actually represents the fraction - // - // p2 * 2^e - // = p2 / 2^-e - // = d[-1] / 10^1 + d[-2] / 10^2 + ... - // - // Now generate the digits d[-m] of p1 from left to right (m = 1,2,...) - // - // p2 * 2^e = d[-1]d[-2]...d[-m] * 10^-m - // + 10^-m * (d[-m-1] / 10^1 + d[-m-2] / 10^2 + ...) - // - // using - // - // 10^m * p2 = ((10^m * p2) div 2^-e) * 2^-e + ((10^m * p2) mod 2^-e) - // = ( d) * 2^-e + ( r) - // - // or - // 10^m * p2 * 2^e = d + r * 2^e - // - // i.e. - // - // M+ = buffer + p2 * 2^e - // = buffer + 10^-m * (d + r * 2^e) - // = (buffer * 10^m + d) * 10^-m + 10^-m * r * 2^e - // - // and stop as soon as 10^-m * r * 2^e <= delta * 2^e - - assert(p2 > delta); - - int m = 0; - for (;;) - { - // Invariant: - // M+ = buffer * 10^-m + 10^-m * (d[-m-1] / 10 + d[-m-2] / 10^2 + ...) * 2^e - // = buffer * 10^-m + 10^-m * (p2 ) * 2^e - // = buffer * 10^-m + 10^-m * (1/10 * (10 * p2) ) * 2^e - // = buffer * 10^-m + 10^-m * (1/10 * ((10*p2 div 2^-e) * 2^-e + (10*p2 mod 2^-e)) * 2^e - // - assert(p2 <= UINT64_MAX / 10); - p2 *= 10; - const uint64_t d = p2 >> -one.e; // d = (10 * p2) div 2^-e - const uint64_t r = p2 & (one.f - 1); // r = (10 * p2) mod 2^-e - // - // M+ = buffer * 10^-m + 10^-m * (1/10 * (d * 2^-e + r) * 2^e - // = buffer * 10^-m + 10^-m * (1/10 * (d + r * 2^e)) - // = (buffer * 10 + d) * 10^(-m-1) + 10^(-m-1) * r * 2^e - // - assert(d <= 9); - buffer[length++] = static_cast<char>('0' + d); // buffer := buffer * 10 + d - // - // M+ = buffer * 10^(-m-1) + 10^(-m-1) * r * 2^e - // - p2 = r; - m++; - // - // M+ = buffer * 10^-m + 10^-m * p2 * 2^e - // Invariant restored. - - // Check if enough digits have been generated. - // - // 10^-m * p2 * 2^e <= delta * 2^e - // p2 * 2^e <= 10^m * delta * 2^e - // p2 <= 10^m * delta - delta *= 10; - dist *= 10; - if (p2 <= delta) - { - break; - } - } - - // V = buffer * 10^-m, with M- <= V <= M+. - - decimal_exponent -= m; - - // 1 ulp in the decimal representation is now 10^-m. - // Since delta and dist are now scaled by 10^m, we need to do the - // same with ulp in order to keep the units in sync. - // - // 10^m * 10^-m = 1 = 2^-e * 2^e = ten_m * 2^e - // - const uint64_t ten_m = one.f; - grisu2_round(buffer, length, dist, delta, p2, ten_m); - - // By construction this algorithm generates the shortest possible decimal - // number (Loitsch, Theorem 6.2) which rounds back to w. - // For an input number of precision p, at least - // - // N = 1 + ceil(p * log_10(2)) - // - // decimal digits are sufficient to identify all binary floating-point - // numbers (Matula, "In-and-Out conversions"). - // This implies that the algorithm does not produce more than N decimal - // digits. - // - // N = 17 for p = 53 (IEEE double precision) - // N = 9 for p = 24 (IEEE single precision) -} - -/*! -v = buf * 10^decimal_exponent -len is the length of the buffer (number of decimal digits) -The buffer must be large enough, i.e. >= max_digits10. -*/ -inline void grisu2(char* buf, int& len, int& decimal_exponent, - diyfp m_minus, diyfp v, diyfp m_plus) -{ - assert(m_plus.e == m_minus.e); - assert(m_plus.e == v.e); - - // --------(-----------------------+-----------------------)-------- (A) - // m- v m+ - // - // --------------------(-----------+-----------------------)-------- (B) - // m- v m+ - // - // First scale v (and m- and m+) such that the exponent is in the range - // [alpha, gamma]. - - const cached_power cached = get_cached_power_for_binary_exponent(m_plus.e); - - const diyfp c_minus_k(cached.f, cached.e); // = c ~= 10^-k - - // The exponent of the products is = v.e + c_minus_k.e + q and is in the range [alpha,gamma] - const diyfp w = diyfp::mul(v, c_minus_k); - const diyfp w_minus = diyfp::mul(m_minus, c_minus_k); - const diyfp w_plus = diyfp::mul(m_plus, c_minus_k); - - // ----(---+---)---------------(---+---)---------------(---+---)---- - // w- w w+ - // = c*m- = c*v = c*m+ - // - // diyfp::mul rounds its result and c_minus_k is approximated too. w, w- and - // w+ are now off by a small amount. - // In fact: - // - // w - v * 10^k < 1 ulp - // - // To account for this inaccuracy, add resp. subtract 1 ulp. - // - // --------+---[---------------(---+---)---------------]---+-------- - // w- M- w M+ w+ - // - // Now any number in [M-, M+] (bounds included) will round to w when input, - // regardless of how the input rounding algorithm breaks ties. - // - // And digit_gen generates the shortest possible such number in [M-, M+]. - // Note that this does not mean that Grisu2 always generates the shortest - // possible number in the interval (m-, m+). - const diyfp M_minus(w_minus.f + 1, w_minus.e); - const diyfp M_plus (w_plus.f - 1, w_plus.e ); - - decimal_exponent = -cached.k; // = -(-k) = k - - grisu2_digit_gen(buf, len, decimal_exponent, M_minus, w, M_plus); -} - -/*! -v = buf * 10^decimal_exponent -len is the length of the buffer (number of decimal digits) -The buffer must be large enough, i.e. >= max_digits10. -*/ -template <typename FloatType> -void grisu2(char* buf, int& len, int& decimal_exponent, FloatType value) -{ - static_assert(diyfp::kPrecision >= std::numeric_limits<FloatType>::digits + 3, - "internal error: not enough precision"); - - assert(std::isfinite(value)); - assert(value > 0); - - // If the neighbors (and boundaries) of 'value' are always computed for double-precision - // numbers, all float's can be recovered using strtod (and strtof). However, the resulting - // decimal representations are not exactly "short". - // - // The documentation for 'std::to_chars' (http://en.cppreference.com/w/cpp/utility/to_chars) - // says "value is converted to a string as if by std::sprintf in the default ("C") locale" - // and since sprintf promotes float's to double's, I think this is exactly what 'std::to_chars' - // does. - // On the other hand, the documentation for 'std::to_chars' requires that "parsing the - // representation using the corresponding std::from_chars function recovers value exactly". That - // indicates that single precision floating-point numbers should be recovered using - // 'std::strtof'. - // - // NB: If the neighbors are computed for single-precision numbers, there is a single float - // (7.0385307e-26f) which can't be recovered using strtod. The resulting double precision - // value is off by 1 ulp. -#if 0 - const boundaries w = compute_boundaries(static_cast<double>(value)); -#else - const boundaries w = compute_boundaries(value); -#endif - - grisu2(buf, len, decimal_exponent, w.minus, w.w, w.plus); -} - -/*! -@brief appends a decimal representation of e to buf -@return a pointer to the element following the exponent. -@pre -1000 < e < 1000 -*/ -inline char* append_exponent(char* buf, int e) -{ - assert(e > -1000); - assert(e < 1000); - - if (e < 0) - { - e = -e; - *buf++ = '-'; - } - else - { - *buf++ = '+'; - } - - uint32_t k = static_cast<uint32_t>(e); - if (k < 10) - { - // Always print at least two digits in the exponent. - // This is for compatibility with printf("%g"). - *buf++ = '0'; - *buf++ = static_cast<char>('0' + k); - } - else if (k < 100) - { - *buf++ = static_cast<char>('0' + k / 10); - k %= 10; - *buf++ = static_cast<char>('0' + k); - } - else - { - *buf++ = static_cast<char>('0' + k / 100); - k %= 100; - *buf++ = static_cast<char>('0' + k / 10); - k %= 10; - *buf++ = static_cast<char>('0' + k); - } - - return buf; -} - -/*! -@brief prettify v = buf * 10^decimal_exponent - -If v is in the range [10^min_exp, 10^max_exp) it will be printed in fixed-point -notation. Otherwise it will be printed in exponential notation. - -@pre min_exp < 0 -@pre max_exp > 0 -*/ -inline char* format_buffer(char* buf, int len, int decimal_exponent, - int min_exp, int max_exp) -{ - assert(min_exp < 0); - assert(max_exp > 0); - - const int k = len; - const int n = len + decimal_exponent; - - // v = buf * 10^(n-k) - // k is the length of the buffer (number of decimal digits) - // n is the position of the decimal point relative to the start of the buffer. - - if (k <= n and n <= max_exp) - { - // digits[000] - // len <= max_exp + 2 - - std::memset(buf + k, '0', static_cast<size_t>(n - k)); - // Make it look like a floating-point number (#362, #378) - buf[n + 0] = '.'; - buf[n + 1] = '0'; - return buf + (n + 2); - } - - if (0 < n and n <= max_exp) - { - // dig.its - // len <= max_digits10 + 1 - - assert(k > n); - - std::memmove(buf + (n + 1), buf + n, static_cast<size_t>(k - n)); - buf[n] = '.'; - return buf + (k + 1); - } - - if (min_exp < n and n <= 0) - { - // 0.[000]digits - // len <= 2 + (-min_exp - 1) + max_digits10 - - std::memmove(buf + (2 + -n), buf, static_cast<size_t>(k)); - buf[0] = '0'; - buf[1] = '.'; - std::memset(buf + 2, '0', static_cast<size_t>(-n)); - return buf + (2 + (-n) + k); - } - - if (k == 1) - { - // dE+123 - // len <= 1 + 5 - - buf += 1; - } - else - { - // d.igitsE+123 - // len <= max_digits10 + 1 + 5 - - std::memmove(buf + 2, buf + 1, static_cast<size_t>(k - 1)); - buf[1] = '.'; - buf += 1 + k; - } - - *buf++ = 'e'; - return append_exponent(buf, n - 1); -} - -} // namespace dtoa_impl - -/*! -@brief generates a decimal representation of the floating-point number value in [first, last). - -The format of the resulting decimal representation is similar to printf's %g -format. Returns an iterator pointing past-the-end of the decimal representation. - -@note The input number must be finite, i.e. NaN's and Inf's are not supported. -@note The buffer must be large enough. -@note The result is NOT null-terminated. -*/ -template <typename FloatType> -char* to_chars(char* first, char* last, FloatType value) -{ - static_cast<void>(last); // maybe unused - fix warning - assert(std::isfinite(value)); - - // Use signbit(value) instead of (value < 0) since signbit works for -0. - if (std::signbit(value)) - { - value = -value; - *first++ = '-'; - } - - if (value == 0) // +-0 - { - *first++ = '0'; - // Make it look like a floating-point number (#362, #378) - *first++ = '.'; - *first++ = '0'; - return first; - } - - assert(last - first >= std::numeric_limits<FloatType>::max_digits10); - - // Compute v = buffer * 10^decimal_exponent. - // The decimal digits are stored in the buffer, which needs to be interpreted - // as an unsigned decimal integer. - // len is the length of the buffer, i.e. the number of decimal digits. - int len = 0; - int decimal_exponent = 0; - dtoa_impl::grisu2(first, len, decimal_exponent, value); - - assert(len <= std::numeric_limits<FloatType>::max_digits10); - - // Format the buffer like printf("%.*g", prec, value) - constexpr int kMinExp = -4; - // Use digits10 here to increase compatibility with version 2. - constexpr int kMaxExp = std::numeric_limits<FloatType>::digits10; - - assert(last - first >= kMaxExp + 2); - assert(last - first >= 2 + (-kMinExp - 1) + std::numeric_limits<FloatType>::max_digits10); - assert(last - first >= std::numeric_limits<FloatType>::max_digits10 + 6); - - return dtoa_impl::format_buffer(first, len, decimal_exponent, kMinExp, kMaxExp); -} - -} // namespace detail -} // namespace nlohmann - -// #include <nlohmann/detail/macro_scope.hpp> - -// #include <nlohmann/detail/meta.hpp> - -// #include <nlohmann/detail/output/output_adapters.hpp> - -// #include <nlohmann/detail/value_t.hpp> - - -namespace nlohmann -{ -namespace detail -{ -/////////////////// -// serialization // -/////////////////// - -template<typename BasicJsonType> -class serializer -{ - using string_t = typename BasicJsonType::string_t; - using number_float_t = typename BasicJsonType::number_float_t; - using number_integer_t = typename BasicJsonType::number_integer_t; - using number_unsigned_t = typename BasicJsonType::number_unsigned_t; - static constexpr uint8_t UTF8_ACCEPT = 0; - static constexpr uint8_t UTF8_REJECT = 1; - - public: - /*! - @param[in] s output stream to serialize to - @param[in] ichar indentation character to use - */ - serializer(output_adapter_t<char> s, const char ichar) - : o(std::move(s)), loc(std::localeconv()), - thousands_sep(loc->thousands_sep == nullptr ? '\0' : * (loc->thousands_sep)), - decimal_point(loc->decimal_point == nullptr ? '\0' : * (loc->decimal_point)), - indent_char(ichar), indent_string(512, indent_char) - {} - - // delete because of pointer members - serializer(const serializer&) = delete; - serializer& operator=(const serializer&) = delete; - - /*! - @brief internal implementation of the serialization function - - This function is called by the public member function dump and organizes - the serialization internally. The indentation level is propagated as - additional parameter. In case of arrays and objects, the function is - called recursively. - - - strings and object keys are escaped using `escape_string()` - - integer numbers are converted implicitly via `operator<<` - - floating-point numbers are converted to a string using `"%g"` format - - @param[in] val value to serialize - @param[in] pretty_print whether the output shall be pretty-printed - @param[in] indent_step the indent level - @param[in] current_indent the current indent level (only used internally) - */ - void dump(const BasicJsonType& val, const bool pretty_print, - const bool ensure_ascii, - const unsigned int indent_step, - const unsigned int current_indent = 0) - { - switch (val.m_type) - { - case value_t::object: - { - if (val.m_value.object->empty()) - { - o->write_characters("{}", 2); - return; - } - - if (pretty_print) - { - o->write_characters("{\n", 2); - - // variable to hold indentation for recursive calls - const auto new_indent = current_indent + indent_step; - if (JSON_UNLIKELY(indent_string.size() < new_indent)) - { - indent_string.resize(indent_string.size() * 2, ' '); - } - - // first n-1 elements - auto i = val.m_value.object->cbegin(); - for (std::size_t cnt = 0; cnt < val.m_value.object->size() - 1; ++cnt, ++i) - { - o->write_characters(indent_string.c_str(), new_indent); - o->write_character('\"'); - dump_escaped(i->first, ensure_ascii); - o->write_characters("\": ", 3); - dump(i->second, true, ensure_ascii, indent_step, new_indent); - o->write_characters(",\n", 2); - } - - // last element - assert(i != val.m_value.object->cend()); - assert(std::next(i) == val.m_value.object->cend()); - o->write_characters(indent_string.c_str(), new_indent); - o->write_character('\"'); - dump_escaped(i->first, ensure_ascii); - o->write_characters("\": ", 3); - dump(i->second, true, ensure_ascii, indent_step, new_indent); - - o->write_character('\n'); - o->write_characters(indent_string.c_str(), current_indent); - o->write_character('}'); - } - else - { - o->write_character('{'); - - // first n-1 elements - auto i = val.m_value.object->cbegin(); - for (std::size_t cnt = 0; cnt < val.m_value.object->size() - 1; ++cnt, ++i) - { - o->write_character('\"'); - dump_escaped(i->first, ensure_ascii); - o->write_characters("\":", 2); - dump(i->second, false, ensure_ascii, indent_step, current_indent); - o->write_character(','); - } - - // last element - assert(i != val.m_value.object->cend()); - assert(std::next(i) == val.m_value.object->cend()); - o->write_character('\"'); - dump_escaped(i->first, ensure_ascii); - o->write_characters("\":", 2); - dump(i->second, false, ensure_ascii, indent_step, current_indent); - - o->write_character('}'); - } - - return; - } - - case value_t::array: - { - if (val.m_value.array->empty()) - { - o->write_characters("[]", 2); - return; - } - - if (pretty_print) - { - o->write_characters("[\n", 2); - - // variable to hold indentation for recursive calls - const auto new_indent = current_indent + indent_step; - if (JSON_UNLIKELY(indent_string.size() < new_indent)) - { - indent_string.resize(indent_string.size() * 2, ' '); - } - - // first n-1 elements - for (auto i = val.m_value.array->cbegin(); - i != val.m_value.array->cend() - 1; ++i) - { - o->write_characters(indent_string.c_str(), new_indent); - dump(*i, true, ensure_ascii, indent_step, new_indent); - o->write_characters(",\n", 2); - } - - // last element - assert(not val.m_value.array->empty()); - o->write_characters(indent_string.c_str(), new_indent); - dump(val.m_value.array->back(), true, ensure_ascii, indent_step, new_indent); - - o->write_character('\n'); - o->write_characters(indent_string.c_str(), current_indent); - o->write_character(']'); - } - else - { - o->write_character('['); - - // first n-1 elements - for (auto i = val.m_value.array->cbegin(); - i != val.m_value.array->cend() - 1; ++i) - { - dump(*i, false, ensure_ascii, indent_step, current_indent); - o->write_character(','); - } - - // last element - assert(not val.m_value.array->empty()); - dump(val.m_value.array->back(), false, ensure_ascii, indent_step, current_indent); - - o->write_character(']'); - } - - return; - } - - case value_t::string: - { - o->write_character('\"'); - dump_escaped(*val.m_value.string, ensure_ascii); - o->write_character('\"'); - return; - } - - case value_t::boolean: - { - if (val.m_value.boolean) - { - o->write_characters("true", 4); - } - else - { - o->write_characters("false", 5); - } - return; - } - - case value_t::number_integer: - { - dump_integer(val.m_value.number_integer); - return; - } - - case value_t::number_unsigned: - { - dump_integer(val.m_value.number_unsigned); - return; - } - - case value_t::number_float: - { - dump_float(val.m_value.number_float); - return; - } - - case value_t::discarded: - { - o->write_characters("<discarded>", 11); - return; - } - - case value_t::null: - { - o->write_characters("null", 4); - return; - } - } - } - - private: - /*! - @brief dump escaped string - - Escape a string by replacing certain special characters by a sequence of an - escape character (backslash) and another character and other control - characters by a sequence of "\u" followed by a four-digit hex - representation. The escaped string is written to output stream @a o. - - @param[in] s the string to escape - @param[in] ensure_ascii whether to escape non-ASCII characters with - \uXXXX sequences - - @complexity Linear in the length of string @a s. - */ - void dump_escaped(const string_t& s, const bool ensure_ascii) - { - uint32_t codepoint; - uint8_t state = UTF8_ACCEPT; - std::size_t bytes = 0; // number of bytes written to string_buffer - - for (std::size_t i = 0; i < s.size(); ++i) - { - const auto byte = static_cast<uint8_t>(s[i]); - - switch (decode(state, codepoint, byte)) - { - case UTF8_ACCEPT: // decode found a new code point - { - switch (codepoint) - { - case 0x08: // backspace - { - string_buffer[bytes++] = '\\'; - string_buffer[bytes++] = 'b'; - break; - } - - case 0x09: // horizontal tab - { - string_buffer[bytes++] = '\\'; - string_buffer[bytes++] = 't'; - break; - } - - case 0x0A: // newline - { - string_buffer[bytes++] = '\\'; - string_buffer[bytes++] = 'n'; - break; - } - - case 0x0C: // formfeed - { - string_buffer[bytes++] = '\\'; - string_buffer[bytes++] = 'f'; - break; - } - - case 0x0D: // carriage return - { - string_buffer[bytes++] = '\\'; - string_buffer[bytes++] = 'r'; - break; - } - - case 0x22: // quotation mark - { - string_buffer[bytes++] = '\\'; - string_buffer[bytes++] = '\"'; - break; - } - - case 0x5C: // reverse solidus - { - string_buffer[bytes++] = '\\'; - string_buffer[bytes++] = '\\'; - break; - } - - default: - { - // escape control characters (0x00..0x1F) or, if - // ensure_ascii parameter is used, non-ASCII characters - if ((codepoint <= 0x1F) or (ensure_ascii and (codepoint >= 0x7F))) - { - if (codepoint <= 0xFFFF) - { - std::snprintf(string_buffer.data() + bytes, 7, "\\u%04x", - static_cast<uint16_t>(codepoint)); - bytes += 6; - } - else - { - std::snprintf(string_buffer.data() + bytes, 13, "\\u%04x\\u%04x", - static_cast<uint16_t>(0xD7C0 + (codepoint >> 10)), - static_cast<uint16_t>(0xDC00 + (codepoint & 0x3FF))); - bytes += 12; - } - } - else - { - // copy byte to buffer (all previous bytes - // been copied have in default case above) - string_buffer[bytes++] = s[i]; - } - break; - } - } - - // write buffer and reset index; there must be 13 bytes - // left, as this is the maximal number of bytes to be - // written ("\uxxxx\uxxxx\0") for one code point - if (string_buffer.size() - bytes < 13) - { - o->write_characters(string_buffer.data(), bytes); - bytes = 0; - } - break; - } - - case UTF8_REJECT: // decode found invalid UTF-8 byte - { - std::stringstream ss; - ss << std::setw(2) << std::uppercase << std::setfill('0') << std::hex << static_cast<int>(byte); - JSON_THROW(type_error::create(316, "invalid UTF-8 byte at index " + std::to_string(i) + ": 0x" + ss.str())); - } - - default: // decode found yet incomplete multi-byte code point - { - if (not ensure_ascii) - { - // code point will not be escaped - copy byte to buffer - string_buffer[bytes++] = s[i]; - } - break; - } - } - } - - if (JSON_LIKELY(state == UTF8_ACCEPT)) - { - // write buffer - if (bytes > 0) - { - o->write_characters(string_buffer.data(), bytes); - } - } - else - { - // we finish reading, but do not accept: string was incomplete - std::stringstream ss; - ss << std::setw(2) << std::uppercase << std::setfill('0') << std::hex << static_cast<int>(static_cast<uint8_t>(s.back())); - JSON_THROW(type_error::create(316, "incomplete UTF-8 string; last byte: 0x" + ss.str())); - } - } - - /*! - @brief dump an integer - - Dump a given integer to output stream @a o. Works internally with - @a number_buffer. - - @param[in] x integer number (signed or unsigned) to dump - @tparam NumberType either @a number_integer_t or @a number_unsigned_t - */ - template<typename NumberType, detail::enable_if_t< - std::is_same<NumberType, number_unsigned_t>::value or - std::is_same<NumberType, number_integer_t>::value, - int> = 0> - void dump_integer(NumberType x) - { - // special case for "0" - if (x == 0) - { - o->write_character('0'); - return; - } - - const bool is_negative = (x <= 0) and (x != 0); // see issue #755 - std::size_t i = 0; - - while (x != 0) - { - // spare 1 byte for '\0' - assert(i < number_buffer.size() - 1); - - const auto digit = std::labs(static_cast<long>(x % 10)); - number_buffer[i++] = static_cast<char>('0' + digit); - x /= 10; - } - - if (is_negative) - { - // make sure there is capacity for the '-' - assert(i < number_buffer.size() - 2); - number_buffer[i++] = '-'; - } - - std::reverse(number_buffer.begin(), number_buffer.begin() + i); - o->write_characters(number_buffer.data(), i); - } - - /*! - @brief dump a floating-point number - - Dump a given floating-point number to output stream @a o. Works internally - with @a number_buffer. - - @param[in] x floating-point number to dump - */ - void dump_float(number_float_t x) - { - // NaN / inf - if (not std::isfinite(x)) - { - o->write_characters("null", 4); - return; - } - - // If number_float_t is an IEEE-754 single or double precision number, - // use the Grisu2 algorithm to produce short numbers which are - // guaranteed to round-trip, using strtof and strtod, resp. - // - // NB: The test below works if <long double> == <double>. - static constexpr bool is_ieee_single_or_double - = (std::numeric_limits<number_float_t>::is_iec559 and std::numeric_limits<number_float_t>::digits == 24 and std::numeric_limits<number_float_t>::max_exponent == 128) or - (std::numeric_limits<number_float_t>::is_iec559 and std::numeric_limits<number_float_t>::digits == 53 and std::numeric_limits<number_float_t>::max_exponent == 1024); - - dump_float(x, std::integral_constant<bool, is_ieee_single_or_double>()); - } - - void dump_float(number_float_t x, std::true_type /*is_ieee_single_or_double*/) - { - char* begin = number_buffer.data(); - char* end = ::nlohmann::detail::to_chars(begin, begin + number_buffer.size(), x); - - o->write_characters(begin, static_cast<size_t>(end - begin)); - } - - void dump_float(number_float_t x, std::false_type /*is_ieee_single_or_double*/) - { - // get number of digits for a float -> text -> float round-trip - static constexpr auto d = std::numeric_limits<number_float_t>::max_digits10; - - // the actual conversion - std::ptrdiff_t len = snprintf(number_buffer.data(), number_buffer.size(), "%.*g", d, x); - - // negative value indicates an error - assert(len > 0); - // check if buffer was large enough - assert(static_cast<std::size_t>(len) < number_buffer.size()); - - // erase thousands separator - if (thousands_sep != '\0') - { - const auto end = std::remove(number_buffer.begin(), - number_buffer.begin() + len, thousands_sep); - std::fill(end, number_buffer.end(), '\0'); - assert((end - number_buffer.begin()) <= len); - len = (end - number_buffer.begin()); - } - - // convert decimal point to '.' - if (decimal_point != '\0' and decimal_point != '.') - { - const auto dec_pos = std::find(number_buffer.begin(), number_buffer.end(), decimal_point); - if (dec_pos != number_buffer.end()) - { - *dec_pos = '.'; - } - } - - o->write_characters(number_buffer.data(), static_cast<std::size_t>(len)); - - // determine if need to append ".0" - const bool value_is_int_like = - std::none_of(number_buffer.begin(), number_buffer.begin() + len + 1, - [](char c) - { - return (c == '.' or c == 'e'); - }); - - if (value_is_int_like) - { - o->write_characters(".0", 2); - } - } - - /*! - @brief check whether a string is UTF-8 encoded - - The function checks each byte of a string whether it is UTF-8 encoded. The - result of the check is stored in the @a state parameter. The function must - be called initially with state 0 (accept). State 1 means the string must - be rejected, because the current byte is not allowed. If the string is - completely processed, but the state is non-zero, the string ended - prematurely; that is, the last byte indicated more bytes should have - followed. - - @param[in,out] state the state of the decoding - @param[in,out] codep codepoint (valid only if resulting state is UTF8_ACCEPT) - @param[in] byte next byte to decode - @return new state - - @note The function has been edited: a std::array is used. - - @copyright Copyright (c) 2008-2009 Bjoern Hoehrmann <bjoern@hoehrmann.de> - @sa http://bjoern.hoehrmann.de/utf-8/decoder/dfa/ - */ - static uint8_t decode(uint8_t& state, uint32_t& codep, const uint8_t byte) noexcept - { - static const std::array<uint8_t, 400> utf8d = - { - { - 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // 00..1F - 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // 20..3F - 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // 40..5F - 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // 60..7F - 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, // 80..9F - 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, // A0..BF - 8, 8, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, // C0..DF - 0xA, 0x3, 0x3, 0x3, 0x3, 0x3, 0x3, 0x3, 0x3, 0x3, 0x3, 0x3, 0x3, 0x4, 0x3, 0x3, // E0..EF - 0xB, 0x6, 0x6, 0x6, 0x5, 0x8, 0x8, 0x8, 0x8, 0x8, 0x8, 0x8, 0x8, 0x8, 0x8, 0x8, // F0..FF - 0x0, 0x1, 0x2, 0x3, 0x5, 0x8, 0x7, 0x1, 0x1, 0x1, 0x4, 0x6, 0x1, 0x1, 0x1, 0x1, // s0..s0 - 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 0, 1, 0, 1, 1, 1, 1, 1, 1, // s1..s2 - 1, 2, 1, 1, 1, 1, 1, 2, 1, 2, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 1, 1, 1, 1, 1, 1, 1, 1, // s3..s4 - 1, 2, 1, 1, 1, 1, 1, 1, 1, 2, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 3, 1, 3, 1, 1, 1, 1, 1, 1, // s5..s6 - 1, 3, 1, 1, 1, 1, 1, 3, 1, 3, 1, 1, 1, 1, 1, 1, 1, 3, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1 // s7..s8 - } - }; - - const uint8_t type = utf8d[byte]; - - codep = (state != UTF8_ACCEPT) - ? (byte & 0x3fu) | (codep << 6) - : static_cast<uint32_t>(0xff >> type) & (byte); - - state = utf8d[256u + state * 16u + type]; - return state; - } - - private: - /// the output of the serializer - output_adapter_t<char> o = nullptr; - - /// a (hopefully) large enough character buffer - std::array<char, 64> number_buffer{{}}; - - /// the locale - const std::lconv* loc = nullptr; - /// the locale's thousand separator character - const char thousands_sep = '\0'; - /// the locale's decimal point character - const char decimal_point = '\0'; - - /// string buffer - std::array<char, 512> string_buffer{{}}; - - /// the indentation character - const char indent_char; - /// the indentation string - string_t indent_string; -}; -} -} - -// #include <nlohmann/detail/json_ref.hpp> - - -#include <initializer_list> -#include <utility> - -namespace nlohmann -{ -namespace detail -{ -template<typename BasicJsonType> -class json_ref -{ - public: - using value_type = BasicJsonType; - - json_ref(value_type&& value) - : owned_value(std::move(value)), value_ref(&owned_value), is_rvalue(true) - {} - - json_ref(const value_type& value) - : value_ref(const_cast<value_type*>(&value)), is_rvalue(false) - {} - - json_ref(std::initializer_list<json_ref> init) - : owned_value(init), value_ref(&owned_value), is_rvalue(true) - {} - - template<class... Args> - json_ref(Args&& ... args) - : owned_value(std::forward<Args>(args)...), value_ref(&owned_value), is_rvalue(true) - {} - - // class should be movable only - json_ref(json_ref&&) = default; - json_ref(const json_ref&) = delete; - json_ref& operator=(const json_ref&) = delete; - - value_type moved_or_copied() const - { - if (is_rvalue) - { - return std::move(*value_ref); - } - return *value_ref; - } - - value_type const& operator*() const - { - return *static_cast<value_type const*>(value_ref); - } - - value_type const* operator->() const - { - return static_cast<value_type const*>(value_ref); - } - - private: - mutable value_type owned_value = nullptr; - value_type* value_ref = nullptr; - const bool is_rvalue; -}; -} -} - -// #include <nlohmann/detail/json_pointer.hpp> - - -#include <cassert> // assert -#include <numeric> // accumulate -#include <string> // string -#include <vector> // vector - -// #include <nlohmann/detail/macro_scope.hpp> - -// #include <nlohmann/detail/exceptions.hpp> - -// #include <nlohmann/detail/value_t.hpp> - - -namespace nlohmann -{ -template<typename BasicJsonType> -class json_pointer -{ - // allow basic_json to access private members - NLOHMANN_BASIC_JSON_TPL_DECLARATION - friend class basic_json; - - public: - /*! - @brief create JSON pointer - - Create a JSON pointer according to the syntax described in - [Section 3 of RFC6901](https://tools.ietf.org/html/rfc6901#section-3). - - @param[in] s string representing the JSON pointer; if omitted, the empty - string is assumed which references the whole JSON value - - @throw parse_error.107 if the given JSON pointer @a s is nonempty and does - not begin with a slash (`/`); see example below - - @throw parse_error.108 if a tilde (`~`) in the given JSON pointer @a s is - not followed by `0` (representing `~`) or `1` (representing `/`); see - example below - - @liveexample{The example shows the construction several valid JSON pointers - as well as the exceptional behavior.,json_pointer} - - @since version 2.0.0 - */ - explicit json_pointer(const std::string& s = "") - : reference_tokens(split(s)) - {} - - /*! - @brief return a string representation of the JSON pointer - - @invariant For each JSON pointer `ptr`, it holds: - @code {.cpp} - ptr == json_pointer(ptr.to_string()); - @endcode - - @return a string representation of the JSON pointer - - @liveexample{The example shows the result of `to_string`., - json_pointer__to_string} - - @since version 2.0.0 - */ - std::string to_string() const noexcept - { - return std::accumulate(reference_tokens.begin(), reference_tokens.end(), - std::string{}, - [](const std::string & a, const std::string & b) - { - return a + "/" + escape(b); - }); - } - - /// @copydoc to_string() - operator std::string() const - { - return to_string(); - } - - /*! - @param[in] s reference token to be converted into an array index - - @return integer representation of @a s - - @throw out_of_range.404 if string @a s could not be converted to an integer - */ - static int array_index(const std::string& s) - { - std::size_t processed_chars = 0; - const int res = std::stoi(s, &processed_chars); - - // check if the string was completely read - if (JSON_UNLIKELY(processed_chars != s.size())) - { - JSON_THROW(detail::out_of_range::create(404, "unresolved reference token '" + s + "'")); - } - - return res; - } - - private: - /*! - @brief remove and return last reference pointer - @throw out_of_range.405 if JSON pointer has no parent - */ - std::string pop_back() - { - if (JSON_UNLIKELY(is_root())) - { - JSON_THROW(detail::out_of_range::create(405, "JSON pointer has no parent")); - } - - auto last = reference_tokens.back(); - reference_tokens.pop_back(); - return last; - } - - /// return whether pointer points to the root document - bool is_root() const - { - return reference_tokens.empty(); - } - - json_pointer top() const - { - if (JSON_UNLIKELY(is_root())) - { - JSON_THROW(detail::out_of_range::create(405, "JSON pointer has no parent")); - } - - json_pointer result = *this; - result.reference_tokens = {reference_tokens[0]}; - return result; - } - - /*! - @brief create and return a reference to the pointed to value - - @complexity Linear in the number of reference tokens. - - @throw parse_error.109 if array index is not a number - @throw type_error.313 if value cannot be unflattened - */ - BasicJsonType& get_and_create(BasicJsonType& j) const - { - using size_type = typename BasicJsonType::size_type; - auto result = &j; - - // in case no reference tokens exist, return a reference to the JSON value - // j which will be overwritten by a primitive value - for (const auto& reference_token : reference_tokens) - { - switch (result->m_type) - { - case detail::value_t::null: - { - if (reference_token == "0") - { - // start a new array if reference token is 0 - result = &result->operator[](0); - } - else - { - // start a new object otherwise - result = &result->operator[](reference_token); - } - break; - } - - case detail::value_t::object: - { - // create an entry in the object - result = &result->operator[](reference_token); - break; - } - - case detail::value_t::array: - { - // create an entry in the array - JSON_TRY - { - result = &result->operator[](static_cast<size_type>(array_index(reference_token))); - } - JSON_CATCH(std::invalid_argument&) - { - JSON_THROW(detail::parse_error::create(109, 0, "array index '" + reference_token + "' is not a number")); - } - break; - } - - /* - The following code is only reached if there exists a reference - token _and_ the current value is primitive. In this case, we have - an error situation, because primitive values may only occur as - single value; that is, with an empty list of reference tokens. - */ - default: - JSON_THROW(detail::type_error::create(313, "invalid value to unflatten")); - } - } - - return *result; - } - - /*! - @brief return a reference to the pointed to value - - @note This version does not throw if a value is not present, but tries to - create nested values instead. For instance, calling this function - with pointer `"/this/that"` on a null value is equivalent to calling - `operator[]("this").operator[]("that")` on that value, effectively - changing the null value to an object. - - @param[in] ptr a JSON value - - @return reference to the JSON value pointed to by the JSON pointer - - @complexity Linear in the length of the JSON pointer. - - @throw parse_error.106 if an array index begins with '0' - @throw parse_error.109 if an array index was not a number - @throw out_of_range.404 if the JSON pointer can not be resolved - */ - BasicJsonType& get_unchecked(BasicJsonType* ptr) const - { - using size_type = typename BasicJsonType::size_type; - for (const auto& reference_token : reference_tokens) - { - // convert null values to arrays or objects before continuing - if (ptr->m_type == detail::value_t::null) - { - // check if reference token is a number - const bool nums = - std::all_of(reference_token.begin(), reference_token.end(), - [](const char x) - { - return (x >= '0' and x <= '9'); - }); - - // change value to array for numbers or "-" or to object otherwise - *ptr = (nums or reference_token == "-") - ? detail::value_t::array - : detail::value_t::object; - } - - switch (ptr->m_type) - { - case detail::value_t::object: - { - // use unchecked object access - ptr = &ptr->operator[](reference_token); - break; - } - - case detail::value_t::array: - { - // error condition (cf. RFC 6901, Sect. 4) - if (JSON_UNLIKELY(reference_token.size() > 1 and reference_token[0] == '0')) - { - JSON_THROW(detail::parse_error::create(106, 0, - "array index '" + reference_token + - "' must not begin with '0'")); - } - - if (reference_token == "-") - { - // explicitly treat "-" as index beyond the end - ptr = &ptr->operator[](ptr->m_value.array->size()); - } - else - { - // convert array index to number; unchecked access - JSON_TRY - { - ptr = &ptr->operator[]( - static_cast<size_type>(array_index(reference_token))); - } - JSON_CATCH(std::invalid_argument&) - { - JSON_THROW(detail::parse_error::create(109, 0, "array index '" + reference_token + "' is not a number")); - } - } - break; - } - - default: - JSON_THROW(detail::out_of_range::create(404, "unresolved reference token '" + reference_token + "'")); - } - } - - return *ptr; - } - - /*! - @throw parse_error.106 if an array index begins with '0' - @throw parse_error.109 if an array index was not a number - @throw out_of_range.402 if the array index '-' is used - @throw out_of_range.404 if the JSON pointer can not be resolved - */ - BasicJsonType& get_checked(BasicJsonType* ptr) const - { - using size_type = typename BasicJsonType::size_type; - for (const auto& reference_token : reference_tokens) - { - switch (ptr->m_type) - { - case detail::value_t::object: - { - // note: at performs range check - ptr = &ptr->at(reference_token); - break; - } - - case detail::value_t::array: - { - if (JSON_UNLIKELY(reference_token == "-")) - { - // "-" always fails the range check - JSON_THROW(detail::out_of_range::create(402, - "array index '-' (" + std::to_string(ptr->m_value.array->size()) + - ") is out of range")); - } - - // error condition (cf. RFC 6901, Sect. 4) - if (JSON_UNLIKELY(reference_token.size() > 1 and reference_token[0] == '0')) - { - JSON_THROW(detail::parse_error::create(106, 0, - "array index '" + reference_token + - "' must not begin with '0'")); - } - - // note: at performs range check - JSON_TRY - { - ptr = &ptr->at(static_cast<size_type>(array_index(reference_token))); - } - JSON_CATCH(std::invalid_argument&) - { - JSON_THROW(detail::parse_error::create(109, 0, "array index '" + reference_token + "' is not a number")); - } - break; - } - - default: - JSON_THROW(detail::out_of_range::create(404, "unresolved reference token '" + reference_token + "'")); - } - } - - return *ptr; - } - - /*! - @brief return a const reference to the pointed to value - - @param[in] ptr a JSON value - - @return const reference to the JSON value pointed to by the JSON - pointer - - @throw parse_error.106 if an array index begins with '0' - @throw parse_error.109 if an array index was not a number - @throw out_of_range.402 if the array index '-' is used - @throw out_of_range.404 if the JSON pointer can not be resolved - */ - const BasicJsonType& get_unchecked(const BasicJsonType* ptr) const - { - using size_type = typename BasicJsonType::size_type; - for (const auto& reference_token : reference_tokens) - { - switch (ptr->m_type) - { - case detail::value_t::object: - { - // use unchecked object access - ptr = &ptr->operator[](reference_token); - break; - } - - case detail::value_t::array: - { - if (JSON_UNLIKELY(reference_token == "-")) - { - // "-" cannot be used for const access - JSON_THROW(detail::out_of_range::create(402, - "array index '-' (" + std::to_string(ptr->m_value.array->size()) + - ") is out of range")); - } - - // error condition (cf. RFC 6901, Sect. 4) - if (JSON_UNLIKELY(reference_token.size() > 1 and reference_token[0] == '0')) - { - JSON_THROW(detail::parse_error::create(106, 0, - "array index '" + reference_token + - "' must not begin with '0'")); - } - - // use unchecked array access - JSON_TRY - { - ptr = &ptr->operator[]( - static_cast<size_type>(array_index(reference_token))); - } - JSON_CATCH(std::invalid_argument&) - { - JSON_THROW(detail::parse_error::create(109, 0, "array index '" + reference_token + "' is not a number")); - } - break; - } - - default: - JSON_THROW(detail::out_of_range::create(404, "unresolved reference token '" + reference_token + "'")); - } - } - - return *ptr; - } - - /*! - @throw parse_error.106 if an array index begins with '0' - @throw parse_error.109 if an array index was not a number - @throw out_of_range.402 if the array index '-' is used - @throw out_of_range.404 if the JSON pointer can not be resolved - */ - const BasicJsonType& get_checked(const BasicJsonType* ptr) const - { - using size_type = typename BasicJsonType::size_type; - for (const auto& reference_token : reference_tokens) - { - switch (ptr->m_type) - { - case detail::value_t::object: - { - // note: at performs range check - ptr = &ptr->at(reference_token); - break; - } - - case detail::value_t::array: - { - if (JSON_UNLIKELY(reference_token == "-")) - { - // "-" always fails the range check - JSON_THROW(detail::out_of_range::create(402, - "array index '-' (" + std::to_string(ptr->m_value.array->size()) + - ") is out of range")); - } - - // error condition (cf. RFC 6901, Sect. 4) - if (JSON_UNLIKELY(reference_token.size() > 1 and reference_token[0] == '0')) - { - JSON_THROW(detail::parse_error::create(106, 0, - "array index '" + reference_token + - "' must not begin with '0'")); - } - - // note: at performs range check - JSON_TRY - { - ptr = &ptr->at(static_cast<size_type>(array_index(reference_token))); - } - JSON_CATCH(std::invalid_argument&) - { - JSON_THROW(detail::parse_error::create(109, 0, "array index '" + reference_token + "' is not a number")); - } - break; - } - - default: - JSON_THROW(detail::out_of_range::create(404, "unresolved reference token '" + reference_token + "'")); - } - } - - return *ptr; - } - - /*! - @brief split the string input to reference tokens - - @note This function is only called by the json_pointer constructor. - All exceptions below are documented there. - - @throw parse_error.107 if the pointer is not empty or begins with '/' - @throw parse_error.108 if character '~' is not followed by '0' or '1' - */ - static std::vector<std::string> split(const std::string& reference_string) - { - std::vector<std::string> result; - - // special case: empty reference string -> no reference tokens - if (reference_string.empty()) - { - return result; - } - - // check if nonempty reference string begins with slash - if (JSON_UNLIKELY(reference_string[0] != '/')) - { - JSON_THROW(detail::parse_error::create(107, 1, - "JSON pointer must be empty or begin with '/' - was: '" + - reference_string + "'")); - } - - // extract the reference tokens: - // - slash: position of the last read slash (or end of string) - // - start: position after the previous slash - for ( - // search for the first slash after the first character - std::size_t slash = reference_string.find_first_of('/', 1), - // set the beginning of the first reference token - start = 1; - // we can stop if start == string::npos+1 = 0 - start != 0; - // set the beginning of the next reference token - // (will eventually be 0 if slash == std::string::npos) - start = slash + 1, - // find next slash - slash = reference_string.find_first_of('/', start)) - { - // use the text between the beginning of the reference token - // (start) and the last slash (slash). - auto reference_token = reference_string.substr(start, slash - start); - - // check reference tokens are properly escaped - for (std::size_t pos = reference_token.find_first_of('~'); - pos != std::string::npos; - pos = reference_token.find_first_of('~', pos + 1)) - { - assert(reference_token[pos] == '~'); - - // ~ must be followed by 0 or 1 - if (JSON_UNLIKELY(pos == reference_token.size() - 1 or - (reference_token[pos + 1] != '0' and - reference_token[pos + 1] != '1'))) - { - JSON_THROW(detail::parse_error::create(108, 0, "escape character '~' must be followed with '0' or '1'")); - } - } - - // finally, store the reference token - unescape(reference_token); - result.push_back(reference_token); - } - - return result; - } - - /*! - @brief replace all occurrences of a substring by another string - - @param[in,out] s the string to manipulate; changed so that all - occurrences of @a f are replaced with @a t - @param[in] f the substring to replace with @a t - @param[in] t the string to replace @a f - - @pre The search string @a f must not be empty. **This precondition is - enforced with an assertion.** - - @since version 2.0.0 - */ - static void replace_substring(std::string& s, const std::string& f, - const std::string& t) - { - assert(not f.empty()); - for (auto pos = s.find(f); // find first occurrence of f - pos != std::string::npos; // make sure f was found - s.replace(pos, f.size(), t), // replace with t, and - pos = s.find(f, pos + t.size())) // find next occurrence of f - {} - } - - /// escape "~"" to "~0" and "/" to "~1" - static std::string escape(std::string s) - { - replace_substring(s, "~", "~0"); - replace_substring(s, "/", "~1"); - return s; - } - - /// unescape "~1" to tilde and "~0" to slash (order is important!) - static void unescape(std::string& s) - { - replace_substring(s, "~1", "/"); - replace_substring(s, "~0", "~"); - } - - /*! - @param[in] reference_string the reference string to the current value - @param[in] value the value to consider - @param[in,out] result the result object to insert values to - - @note Empty objects or arrays are flattened to `null`. - */ - static void flatten(const std::string& reference_string, - const BasicJsonType& value, - BasicJsonType& result) - { - switch (value.m_type) - { - case detail::value_t::array: - { - if (value.m_value.array->empty()) - { - // flatten empty array as null - result[reference_string] = nullptr; - } - else - { - // iterate array and use index as reference string - for (std::size_t i = 0; i < value.m_value.array->size(); ++i) - { - flatten(reference_string + "/" + std::to_string(i), - value.m_value.array->operator[](i), result); - } - } - break; - } - - case detail::value_t::object: - { - if (value.m_value.object->empty()) - { - // flatten empty object as null - result[reference_string] = nullptr; - } - else - { - // iterate object and use keys as reference string - for (const auto& element : *value.m_value.object) - { - flatten(reference_string + "/" + escape(element.first), element.second, result); - } - } - break; - } - - default: - { - // add primitive value with its reference string - result[reference_string] = value; - break; - } - } - } - - /*! - @param[in] value flattened JSON - - @return unflattened JSON - - @throw parse_error.109 if array index is not a number - @throw type_error.314 if value is not an object - @throw type_error.315 if object values are not primitive - @throw type_error.313 if value cannot be unflattened - */ - static BasicJsonType - unflatten(const BasicJsonType& value) - { - if (JSON_UNLIKELY(not value.is_object())) - { - JSON_THROW(detail::type_error::create(314, "only objects can be unflattened")); - } - - BasicJsonType result; - - // iterate the JSON object values - for (const auto& element : *value.m_value.object) - { - if (JSON_UNLIKELY(not element.second.is_primitive())) - { - JSON_THROW(detail::type_error::create(315, "values in object must be primitive")); - } - - // assign value to reference pointed to by JSON pointer; Note that if - // the JSON pointer is "" (i.e., points to the whole value), function - // get_and_create returns a reference to result itself. An assignment - // will then create a primitive value. - json_pointer(element.first).get_and_create(result) = element.second; - } - - return result; - } - - friend bool operator==(json_pointer const& lhs, - json_pointer const& rhs) noexcept - { - return (lhs.reference_tokens == rhs.reference_tokens); - } - - friend bool operator!=(json_pointer const& lhs, - json_pointer const& rhs) noexcept - { - return not (lhs == rhs); - } - - /// the reference tokens - std::vector<std::string> reference_tokens; -}; -} - -// #include <nlohmann/adl_serializer.hpp> - - -#include <utility> - -// #include <nlohmann/detail/conversions/from_json.hpp> - -// #include <nlohmann/detail/conversions/to_json.hpp> - - -namespace nlohmann -{ -template<typename, typename> -struct adl_serializer -{ - /*! - @brief convert a JSON value to any value type - - This function is usually called by the `get()` function of the - @ref basic_json class (either explicit or via conversion operators). - - @param[in] j JSON value to read from - @param[in,out] val value to write to - */ - template<typename BasicJsonType, typename ValueType> - static void from_json(BasicJsonType&& j, ValueType& val) noexcept( - noexcept(::nlohmann::from_json(std::forward<BasicJsonType>(j), val))) - { - ::nlohmann::from_json(std::forward<BasicJsonType>(j), val); - } - - /*! - @brief convert any value type to a JSON value - - This function is usually called by the constructors of the @ref basic_json - class. - - @param[in,out] j JSON value to write to - @param[in] val value to read from - */ - template<typename BasicJsonType, typename ValueType> - static void to_json(BasicJsonType& j, ValueType&& val) noexcept( - noexcept(::nlohmann::to_json(j, std::forward<ValueType>(val)))) - { - ::nlohmann::to_json(j, std::forward<ValueType>(val)); - } -}; -} - - -/*! -@brief namespace for Niels Lohmann -@see https://github.com/nlohmann -@since version 1.0.0 -*/ -namespace nlohmann -{ - -/*! -@brief a class to store JSON values - -@tparam ObjectType type for JSON objects (`std::map` by default; will be used -in @ref object_t) -@tparam ArrayType type for JSON arrays (`std::vector` by default; will be used -in @ref array_t) -@tparam StringType type for JSON strings and object keys (`std::string` by -default; will be used in @ref string_t) -@tparam BooleanType type for JSON booleans (`bool` by default; will be used -in @ref boolean_t) -@tparam NumberIntegerType type for JSON integer numbers (`int64_t` by -default; will be used in @ref number_integer_t) -@tparam NumberUnsignedType type for JSON unsigned integer numbers (@c -`uint64_t` by default; will be used in @ref number_unsigned_t) -@tparam NumberFloatType type for JSON floating-point numbers (`double` by -default; will be used in @ref number_float_t) -@tparam AllocatorType type of the allocator to use (`std::allocator` by -default) -@tparam JSONSerializer the serializer to resolve internal calls to `to_json()` -and `from_json()` (@ref adl_serializer by default) - -@requirement The class satisfies the following concept requirements: -- Basic - - [DefaultConstructible](http://en.cppreference.com/w/cpp/concept/DefaultConstructible): - JSON values can be default constructed. The result will be a JSON null - value. - - [MoveConstructible](http://en.cppreference.com/w/cpp/concept/MoveConstructible): - A JSON value can be constructed from an rvalue argument. - - [CopyConstructible](http://en.cppreference.com/w/cpp/concept/CopyConstructible): - A JSON value can be copy-constructed from an lvalue expression. - - [MoveAssignable](http://en.cppreference.com/w/cpp/concept/MoveAssignable): - A JSON value van be assigned from an rvalue argument. - - [CopyAssignable](http://en.cppreference.com/w/cpp/concept/CopyAssignable): - A JSON value can be copy-assigned from an lvalue expression. - - [Destructible](http://en.cppreference.com/w/cpp/concept/Destructible): - JSON values can be destructed. -- Layout - - [StandardLayoutType](http://en.cppreference.com/w/cpp/concept/StandardLayoutType): - JSON values have - [standard layout](http://en.cppreference.com/w/cpp/language/data_members#Standard_layout): - All non-static data members are private and standard layout types, the - class has no virtual functions or (virtual) base classes. -- Library-wide - - [EqualityComparable](http://en.cppreference.com/w/cpp/concept/EqualityComparable): - JSON values can be compared with `==`, see @ref - operator==(const_reference,const_reference). - - [LessThanComparable](http://en.cppreference.com/w/cpp/concept/LessThanComparable): - JSON values can be compared with `<`, see @ref - operator<(const_reference,const_reference). - - [Swappable](http://en.cppreference.com/w/cpp/concept/Swappable): - Any JSON lvalue or rvalue of can be swapped with any lvalue or rvalue of - other compatible types, using unqualified function call @ref swap(). - - [NullablePointer](http://en.cppreference.com/w/cpp/concept/NullablePointer): - JSON values can be compared against `std::nullptr_t` objects which are used - to model the `null` value. -- Container - - [Container](http://en.cppreference.com/w/cpp/concept/Container): - JSON values can be used like STL containers and provide iterator access. - - [ReversibleContainer](http://en.cppreference.com/w/cpp/concept/ReversibleContainer); - JSON values can be used like STL containers and provide reverse iterator - access. - -@invariant The member variables @a m_value and @a m_type have the following -relationship: -- If `m_type == value_t::object`, then `m_value.object != nullptr`. -- If `m_type == value_t::array`, then `m_value.array != nullptr`. -- If `m_type == value_t::string`, then `m_value.string != nullptr`. -The invariants are checked by member function assert_invariant(). - -@internal -@note ObjectType trick from http://stackoverflow.com/a/9860911 -@endinternal - -@see [RFC 7159: The JavaScript Object Notation (JSON) Data Interchange -Format](http://rfc7159.net/rfc7159) - -@since version 1.0.0 - -@nosubgrouping -*/ -NLOHMANN_BASIC_JSON_TPL_DECLARATION -class basic_json -{ - private: - template<detail::value_t> friend struct detail::external_constructor; - friend ::nlohmann::json_pointer<basic_json>; - friend ::nlohmann::detail::parser<basic_json>; - friend ::nlohmann::detail::serializer<basic_json>; - template<typename BasicJsonType> - friend class ::nlohmann::detail::iter_impl; - template<typename BasicJsonType, typename CharType> - friend class ::nlohmann::detail::binary_writer; - template<typename BasicJsonType> - friend class ::nlohmann::detail::binary_reader; - - /// workaround type for MSVC - using basic_json_t = NLOHMANN_BASIC_JSON_TPL; - - // convenience aliases for types residing in namespace detail; - using lexer = ::nlohmann::detail::lexer<basic_json>; - using parser = ::nlohmann::detail::parser<basic_json>; - - using primitive_iterator_t = ::nlohmann::detail::primitive_iterator_t; - template<typename BasicJsonType> - using internal_iterator = ::nlohmann::detail::internal_iterator<BasicJsonType>; - template<typename BasicJsonType> - using iter_impl = ::nlohmann::detail::iter_impl<BasicJsonType>; - template<typename Iterator> - using iteration_proxy = ::nlohmann::detail::iteration_proxy<Iterator>; - template<typename Base> using json_reverse_iterator = ::nlohmann::detail::json_reverse_iterator<Base>; - - template<typename CharType> - using output_adapter_t = ::nlohmann::detail::output_adapter_t<CharType>; - - using binary_reader = ::nlohmann::detail::binary_reader<basic_json>; - template<typename CharType> using binary_writer = ::nlohmann::detail::binary_writer<basic_json, CharType>; - - using serializer = ::nlohmann::detail::serializer<basic_json>; - - public: - using value_t = detail::value_t; - /// @copydoc nlohmann::json_pointer - using json_pointer = ::nlohmann::json_pointer<basic_json>; - template<typename T, typename SFINAE> - using json_serializer = JSONSerializer<T, SFINAE>; - /// helper type for initializer lists of basic_json values - using initializer_list_t = std::initializer_list<detail::json_ref<basic_json>>; - - //////////////// - // exceptions // - //////////////// - - /// @name exceptions - /// Classes to implement user-defined exceptions. - /// @{ - - /// @copydoc detail::exception - using exception = detail::exception; - /// @copydoc detail::parse_error - using parse_error = detail::parse_error; - /// @copydoc detail::invalid_iterator - using invalid_iterator = detail::invalid_iterator; - /// @copydoc detail::type_error - using type_error = detail::type_error; - /// @copydoc detail::out_of_range - using out_of_range = detail::out_of_range; - /// @copydoc detail::other_error - using other_error = detail::other_error; - - /// @} - - - ///////////////////// - // container types // - ///////////////////// - - /// @name container types - /// The canonic container types to use @ref basic_json like any other STL - /// container. - /// @{ - - /// the type of elements in a basic_json container - using value_type = basic_json; - - /// the type of an element reference - using reference = value_type&; - /// the type of an element const reference - using const_reference = const value_type&; - - /// a type to represent differences between iterators - using difference_type = std::ptrdiff_t; - /// a type to represent container sizes - using size_type = std::size_t; - - /// the allocator type - using allocator_type = AllocatorType<basic_json>; - - /// the type of an element pointer - using pointer = typename std::allocator_traits<allocator_type>::pointer; - /// the type of an element const pointer - using const_pointer = typename std::allocator_traits<allocator_type>::const_pointer; - - /// an iterator for a basic_json container - using iterator = iter_impl<basic_json>; - /// a const iterator for a basic_json container - using const_iterator = iter_impl<const basic_json>; - /// a reverse iterator for a basic_json container - using reverse_iterator = json_reverse_iterator<typename basic_json::iterator>; - /// a const reverse iterator for a basic_json container - using const_reverse_iterator = json_reverse_iterator<typename basic_json::const_iterator>; - - /// @} - - - /*! - @brief returns the allocator associated with the container - */ - static allocator_type get_allocator() - { - return allocator_type(); - } - - /*! - @brief returns version information on the library - - This function returns a JSON object with information about the library, - including the version number and information on the platform and compiler. - - @return JSON object holding version information - key | description - ----------- | --------------- - `compiler` | Information on the used compiler. It is an object with the following keys: `c++` (the used C++ standard), `family` (the compiler family; possible values are `clang`, `icc`, `gcc`, `ilecpp`, `msvc`, `pgcpp`, `sunpro`, and `unknown`), and `version` (the compiler version). - `copyright` | The copyright line for the library as string. - `name` | The name of the library as string. - `platform` | The used platform as string. Possible values are `win32`, `linux`, `apple`, `unix`, and `unknown`. - `url` | The URL of the project as string. - `version` | The version of the library. It is an object with the following keys: `major`, `minor`, and `patch` as defined by [Semantic Versioning](http://semver.org), and `string` (the version string). - - @liveexample{The following code shows an example output of the `meta()` - function.,meta} - - @exceptionsafety Strong guarantee: if an exception is thrown, there are no - changes to any JSON value. - - @complexity Constant. - - @since 2.1.0 - */ - static basic_json meta() - { - basic_json result; - - result["copyright"] = "(C) 2013-2017 Niels Lohmann"; - result["name"] = "JSON for Modern C++"; - result["url"] = "https://github.com/nlohmann/json"; - result["version"]["string"] = - std::to_string(NLOHMANN_JSON_VERSION_MAJOR) + "." + - std::to_string(NLOHMANN_JSON_VERSION_MINOR) + "." + - std::to_string(NLOHMANN_JSON_VERSION_PATCH); - result["version"]["major"] = NLOHMANN_JSON_VERSION_MAJOR; - result["version"]["minor"] = NLOHMANN_JSON_VERSION_MINOR; - result["version"]["patch"] = NLOHMANN_JSON_VERSION_PATCH; - -#ifdef _WIN32 - result["platform"] = "win32"; -#elif defined __linux__ - result["platform"] = "linux"; -#elif defined __APPLE__ - result["platform"] = "apple"; -#elif defined __unix__ - result["platform"] = "unix"; -#else - result["platform"] = "unknown"; -#endif - -#if defined(__ICC) || defined(__INTEL_COMPILER) - result["compiler"] = {{"family", "icc"}, {"version", __INTEL_COMPILER}}; -#elif defined(__clang__) - result["compiler"] = {{"family", "clang"}, {"version", __clang_version__}}; -#elif defined(__GNUC__) || defined(__GNUG__) - result["compiler"] = {{"family", "gcc"}, {"version", std::to_string(__GNUC__) + "." + std::to_string(__GNUC_MINOR__) + "." + std::to_string(__GNUC_PATCHLEVEL__)}}; -#elif defined(__HP_cc) || defined(__HP_aCC) - result["compiler"] = "hp" -#elif defined(__IBMCPP__) - result["compiler"] = {{"family", "ilecpp"}, {"version", __IBMCPP__}}; -#elif defined(_MSC_VER) - result["compiler"] = {{"family", "msvc"}, {"version", _MSC_VER}}; -#elif defined(__PGI) - result["compiler"] = {{"family", "pgcpp"}, {"version", __PGI}}; -#elif defined(__SUNPRO_CC) - result["compiler"] = {{"family", "sunpro"}, {"version", __SUNPRO_CC}}; -#else - result["compiler"] = {{"family", "unknown"}, {"version", "unknown"}}; -#endif - -#ifdef __cplusplus - result["compiler"]["c++"] = std::to_string(__cplusplus); -#else - result["compiler"]["c++"] = "unknown"; -#endif - return result; - } - - - /////////////////////////// - // JSON value data types // - /////////////////////////// - - /// @name JSON value data types - /// The data types to store a JSON value. These types are derived from - /// the template arguments passed to class @ref basic_json. - /// @{ - -#if defined(JSON_HAS_CPP_14) - // Use transparent comparator if possible, combined with perfect forwarding - // on find() and count() calls prevents unnecessary string construction. - using object_comparator_t = std::less<>; -#else - using object_comparator_t = std::less<StringType>; -#endif - - /*! - @brief a type for an object - - [RFC 7159](http://rfc7159.net/rfc7159) describes JSON objects as follows: - > An object is an unordered collection of zero or more name/value pairs, - > where a name is a string and a value is a string, number, boolean, null, - > object, or array. - - To store objects in C++, a type is defined by the template parameters - described below. - - @tparam ObjectType the container to store objects (e.g., `std::map` or - `std::unordered_map`) - @tparam StringType the type of the keys or names (e.g., `std::string`). - The comparison function `std::less<StringType>` is used to order elements - inside the container. - @tparam AllocatorType the allocator to use for objects (e.g., - `std::allocator`) - - #### Default type - - With the default values for @a ObjectType (`std::map`), @a StringType - (`std::string`), and @a AllocatorType (`std::allocator`), the default - value for @a object_t is: - - @code {.cpp} - std::map< - std::string, // key_type - basic_json, // value_type - std::less<std::string>, // key_compare - std::allocator<std::pair<const std::string, basic_json>> // allocator_type - > - @endcode - - #### Behavior - - The choice of @a object_t influences the behavior of the JSON class. With - the default type, objects have the following behavior: - - - When all names are unique, objects will be interoperable in the sense - that all software implementations receiving that object will agree on - the name-value mappings. - - When the names within an object are not unique, it is unspecified which - one of the values for a given key will be chosen. For instance, - `{"key": 2, "key": 1}` could be equal to either `{"key": 1}` or - `{"key": 2}`. - - Internally, name/value pairs are stored in lexicographical order of the - names. Objects will also be serialized (see @ref dump) in this order. - For instance, `{"b": 1, "a": 2}` and `{"a": 2, "b": 1}` will be stored - and serialized as `{"a": 2, "b": 1}`. - - When comparing objects, the order of the name/value pairs is irrelevant. - This makes objects interoperable in the sense that they will not be - affected by these differences. For instance, `{"b": 1, "a": 2}` and - `{"a": 2, "b": 1}` will be treated as equal. - - #### Limits - - [RFC 7159](http://rfc7159.net/rfc7159) specifies: - > An implementation may set limits on the maximum depth of nesting. - - In this class, the object's limit of nesting is not explicitly constrained. - However, a maximum depth of nesting may be introduced by the compiler or - runtime environment. A theoretical limit can be queried by calling the - @ref max_size function of a JSON object. - - #### Storage - - Objects are stored as pointers in a @ref basic_json type. That is, for any - access to object values, a pointer of type `object_t*` must be - dereferenced. - - @sa @ref array_t -- type for an array value - - @since version 1.0.0 - - @note The order name/value pairs are added to the object is *not* - preserved by the library. Therefore, iterating an object may return - name/value pairs in a different order than they were originally stored. In - fact, keys will be traversed in alphabetical order as `std::map` with - `std::less` is used by default. Please note this behavior conforms to [RFC - 7159](http://rfc7159.net/rfc7159), because any order implements the - specified "unordered" nature of JSON objects. - */ - using object_t = ObjectType<StringType, - basic_json, - object_comparator_t, - AllocatorType<std::pair<const StringType, - basic_json>>>; - - /*! - @brief a type for an array - - [RFC 7159](http://rfc7159.net/rfc7159) describes JSON arrays as follows: - > An array is an ordered sequence of zero or more values. - - To store objects in C++, a type is defined by the template parameters - explained below. - - @tparam ArrayType container type to store arrays (e.g., `std::vector` or - `std::list`) - @tparam AllocatorType allocator to use for arrays (e.g., `std::allocator`) - - #### Default type - - With the default values for @a ArrayType (`std::vector`) and @a - AllocatorType (`std::allocator`), the default value for @a array_t is: - - @code {.cpp} - std::vector< - basic_json, // value_type - std::allocator<basic_json> // allocator_type - > - @endcode - - #### Limits - - [RFC 7159](http://rfc7159.net/rfc7159) specifies: - > An implementation may set limits on the maximum depth of nesting. - - In this class, the array's limit of nesting is not explicitly constrained. - However, a maximum depth of nesting may be introduced by the compiler or - runtime environment. A theoretical limit can be queried by calling the - @ref max_size function of a JSON array. - - #### Storage - - Arrays are stored as pointers in a @ref basic_json type. That is, for any - access to array values, a pointer of type `array_t*` must be dereferenced. - - @sa @ref object_t -- type for an object value - - @since version 1.0.0 - */ - using array_t = ArrayType<basic_json, AllocatorType<basic_json>>; - - /*! - @brief a type for a string - - [RFC 7159](http://rfc7159.net/rfc7159) describes JSON strings as follows: - > A string is a sequence of zero or more Unicode characters. - - To store objects in C++, a type is defined by the template parameter - described below. Unicode values are split by the JSON class into - byte-sized characters during deserialization. - - @tparam StringType the container to store strings (e.g., `std::string`). - Note this container is used for keys/names in objects, see @ref object_t. - - #### Default type - - With the default values for @a StringType (`std::string`), the default - value for @a string_t is: - - @code {.cpp} - std::string - @endcode - - #### Encoding - - Strings are stored in UTF-8 encoding. Therefore, functions like - `std::string::size()` or `std::string::length()` return the number of - bytes in the string rather than the number of characters or glyphs. - - #### String comparison - - [RFC 7159](http://rfc7159.net/rfc7159) states: - > Software implementations are typically required to test names of object - > members for equality. Implementations that transform the textual - > representation into sequences of Unicode code units and then perform the - > comparison numerically, code unit by code unit, are interoperable in the - > sense that implementations will agree in all cases on equality or - > inequality of two strings. For example, implementations that compare - > strings with escaped characters unconverted may incorrectly find that - > `"a\\b"` and `"a\u005Cb"` are not equal. - - This implementation is interoperable as it does compare strings code unit - by code unit. - - #### Storage - - String values are stored as pointers in a @ref basic_json type. That is, - for any access to string values, a pointer of type `string_t*` must be - dereferenced. - - @since version 1.0.0 - */ - using string_t = StringType; - - /*! - @brief a type for a boolean - - [RFC 7159](http://rfc7159.net/rfc7159) implicitly describes a boolean as a - type which differentiates the two literals `true` and `false`. - - To store objects in C++, a type is defined by the template parameter @a - BooleanType which chooses the type to use. - - #### Default type - - With the default values for @a BooleanType (`bool`), the default value for - @a boolean_t is: - - @code {.cpp} - bool - @endcode - - #### Storage - - Boolean values are stored directly inside a @ref basic_json type. - - @since version 1.0.0 - */ - using boolean_t = BooleanType; - - /*! - @brief a type for a number (integer) - - [RFC 7159](http://rfc7159.net/rfc7159) describes numbers as follows: - > The representation of numbers is similar to that used in most - > programming languages. A number is represented in base 10 using decimal - > digits. It contains an integer component that may be prefixed with an - > optional minus sign, which may be followed by a fraction part and/or an - > exponent part. Leading zeros are not allowed. (...) Numeric values that - > cannot be represented in the grammar below (such as Infinity and NaN) - > are not permitted. - - This description includes both integer and floating-point numbers. - However, C++ allows more precise storage if it is known whether the number - is a signed integer, an unsigned integer or a floating-point number. - Therefore, three different types, @ref number_integer_t, @ref - number_unsigned_t and @ref number_float_t are used. - - To store integer numbers in C++, a type is defined by the template - parameter @a NumberIntegerType which chooses the type to use. - - #### Default type - - With the default values for @a NumberIntegerType (`int64_t`), the default - value for @a number_integer_t is: - - @code {.cpp} - int64_t - @endcode - - #### Default behavior - - - The restrictions about leading zeros is not enforced in C++. Instead, - leading zeros in integer literals lead to an interpretation as octal - number. Internally, the value will be stored as decimal number. For - instance, the C++ integer literal `010` will be serialized to `8`. - During deserialization, leading zeros yield an error. - - Not-a-number (NaN) values will be serialized to `null`. - - #### Limits - - [RFC 7159](http://rfc7159.net/rfc7159) specifies: - > An implementation may set limits on the range and precision of numbers. - - When the default type is used, the maximal integer number that can be - stored is `9223372036854775807` (INT64_MAX) and the minimal integer number - that can be stored is `-9223372036854775808` (INT64_MIN). Integer numbers - that are out of range will yield over/underflow when used in a - constructor. During deserialization, too large or small integer numbers - will be automatically be stored as @ref number_unsigned_t or @ref - number_float_t. - - [RFC 7159](http://rfc7159.net/rfc7159) further states: - > Note that when such software is used, numbers that are integers and are - > in the range \f$[-2^{53}+1, 2^{53}-1]\f$ are interoperable in the sense - > that implementations will agree exactly on their numeric values. - - As this range is a subrange of the exactly supported range [INT64_MIN, - INT64_MAX], this class's integer type is interoperable. - - #### Storage - - Integer number values are stored directly inside a @ref basic_json type. - - @sa @ref number_float_t -- type for number values (floating-point) - - @sa @ref number_unsigned_t -- type for number values (unsigned integer) - - @since version 1.0.0 - */ - using number_integer_t = NumberIntegerType; - - /*! - @brief a type for a number (unsigned) - - [RFC 7159](http://rfc7159.net/rfc7159) describes numbers as follows: - > The representation of numbers is similar to that used in most - > programming languages. A number is represented in base 10 using decimal - > digits. It contains an integer component that may be prefixed with an - > optional minus sign, which may be followed by a fraction part and/or an - > exponent part. Leading zeros are not allowed. (...) Numeric values that - > cannot be represented in the grammar below (such as Infinity and NaN) - > are not permitted. - - This description includes both integer and floating-point numbers. - However, C++ allows more precise storage if it is known whether the number - is a signed integer, an unsigned integer or a floating-point number. - Therefore, three different types, @ref number_integer_t, @ref - number_unsigned_t and @ref number_float_t are used. - - To store unsigned integer numbers in C++, a type is defined by the - template parameter @a NumberUnsignedType which chooses the type to use. - - #### Default type - - With the default values for @a NumberUnsignedType (`uint64_t`), the - default value for @a number_unsigned_t is: - - @code {.cpp} - uint64_t - @endcode - - #### Default behavior - - - The restrictions about leading zeros is not enforced in C++. Instead, - leading zeros in integer literals lead to an interpretation as octal - number. Internally, the value will be stored as decimal number. For - instance, the C++ integer literal `010` will be serialized to `8`. - During deserialization, leading zeros yield an error. - - Not-a-number (NaN) values will be serialized to `null`. - - #### Limits - - [RFC 7159](http://rfc7159.net/rfc7159) specifies: - > An implementation may set limits on the range and precision of numbers. - - When the default type is used, the maximal integer number that can be - stored is `18446744073709551615` (UINT64_MAX) and the minimal integer - number that can be stored is `0`. Integer numbers that are out of range - will yield over/underflow when used in a constructor. During - deserialization, too large or small integer numbers will be automatically - be stored as @ref number_integer_t or @ref number_float_t. - - [RFC 7159](http://rfc7159.net/rfc7159) further states: - > Note that when such software is used, numbers that are integers and are - > in the range \f$[-2^{53}+1, 2^{53}-1]\f$ are interoperable in the sense - > that implementations will agree exactly on their numeric values. - - As this range is a subrange (when considered in conjunction with the - number_integer_t type) of the exactly supported range [0, UINT64_MAX], - this class's integer type is interoperable. - - #### Storage - - Integer number values are stored directly inside a @ref basic_json type. - - @sa @ref number_float_t -- type for number values (floating-point) - @sa @ref number_integer_t -- type for number values (integer) - - @since version 2.0.0 - */ - using number_unsigned_t = NumberUnsignedType; - - /*! - @brief a type for a number (floating-point) - - [RFC 7159](http://rfc7159.net/rfc7159) describes numbers as follows: - > The representation of numbers is similar to that used in most - > programming languages. A number is represented in base 10 using decimal - > digits. It contains an integer component that may be prefixed with an - > optional minus sign, which may be followed by a fraction part and/or an - > exponent part. Leading zeros are not allowed. (...) Numeric values that - > cannot be represented in the grammar below (such as Infinity and NaN) - > are not permitted. - - This description includes both integer and floating-point numbers. - However, C++ allows more precise storage if it is known whether the number - is a signed integer, an unsigned integer or a floating-point number. - Therefore, three different types, @ref number_integer_t, @ref - number_unsigned_t and @ref number_float_t are used. - - To store floating-point numbers in C++, a type is defined by the template - parameter @a NumberFloatType which chooses the type to use. - - #### Default type - - With the default values for @a NumberFloatType (`double`), the default - value for @a number_float_t is: - - @code {.cpp} - double - @endcode - - #### Default behavior - - - The restrictions about leading zeros is not enforced in C++. Instead, - leading zeros in floating-point literals will be ignored. Internally, - the value will be stored as decimal number. For instance, the C++ - floating-point literal `01.2` will be serialized to `1.2`. During - deserialization, leading zeros yield an error. - - Not-a-number (NaN) values will be serialized to `null`. - - #### Limits - - [RFC 7159](http://rfc7159.net/rfc7159) states: - > This specification allows implementations to set limits on the range and - > precision of numbers accepted. Since software that implements IEEE - > 754-2008 binary64 (double precision) numbers is generally available and - > widely used, good interoperability can be achieved by implementations - > that expect no more precision or range than these provide, in the sense - > that implementations will approximate JSON numbers within the expected - > precision. - - This implementation does exactly follow this approach, as it uses double - precision floating-point numbers. Note values smaller than - `-1.79769313486232e+308` and values greater than `1.79769313486232e+308` - will be stored as NaN internally and be serialized to `null`. - - #### Storage - - Floating-point number values are stored directly inside a @ref basic_json - type. - - @sa @ref number_integer_t -- type for number values (integer) - - @sa @ref number_unsigned_t -- type for number values (unsigned integer) - - @since version 1.0.0 - */ - using number_float_t = NumberFloatType; - - /// @} - - private: - - /// helper for exception-safe object creation - template<typename T, typename... Args> - static T* create(Args&& ... args) - { - AllocatorType<T> alloc; - using AllocatorTraits = std::allocator_traits<AllocatorType<T>>; - - auto deleter = [&](T * object) - { - AllocatorTraits::deallocate(alloc, object, 1); - }; - std::unique_ptr<T, decltype(deleter)> object(AllocatorTraits::allocate(alloc, 1), deleter); - AllocatorTraits::construct(alloc, object.get(), std::forward<Args>(args)...); - assert(object != nullptr); - return object.release(); - } - - //////////////////////// - // JSON value storage // - //////////////////////// - - /*! - @brief a JSON value - - The actual storage for a JSON value of the @ref basic_json class. This - union combines the different storage types for the JSON value types - defined in @ref value_t. - - JSON type | value_t type | used type - --------- | --------------- | ------------------------ - object | object | pointer to @ref object_t - array | array | pointer to @ref array_t - string | string | pointer to @ref string_t - boolean | boolean | @ref boolean_t - number | number_integer | @ref number_integer_t - number | number_unsigned | @ref number_unsigned_t - number | number_float | @ref number_float_t - null | null | *no value is stored* - - @note Variable-length types (objects, arrays, and strings) are stored as - pointers. The size of the union should not exceed 64 bits if the default - value types are used. - - @since version 1.0.0 - */ - union json_value - { - /// object (stored with pointer to save storage) - object_t* object; - /// array (stored with pointer to save storage) - array_t* array; - /// string (stored with pointer to save storage) - string_t* string; - /// boolean - boolean_t boolean; - /// number (integer) - number_integer_t number_integer; - /// number (unsigned integer) - number_unsigned_t number_unsigned; - /// number (floating-point) - number_float_t number_float; - - /// default constructor (for null values) - json_value() = default; - /// constructor for booleans - json_value(boolean_t v) noexcept : boolean(v) {} - /// constructor for numbers (integer) - json_value(number_integer_t v) noexcept : number_integer(v) {} - /// constructor for numbers (unsigned) - json_value(number_unsigned_t v) noexcept : number_unsigned(v) {} - /// constructor for numbers (floating-point) - json_value(number_float_t v) noexcept : number_float(v) {} - /// constructor for empty values of a given type - json_value(value_t t) - { - switch (t) - { - case value_t::object: - { - object = create<object_t>(); - break; - } - - case value_t::array: - { - array = create<array_t>(); - break; - } - - case value_t::string: - { - string = create<string_t>(""); - break; - } - - case value_t::boolean: - { - boolean = boolean_t(false); - break; - } - - case value_t::number_integer: - { - number_integer = number_integer_t(0); - break; - } - - case value_t::number_unsigned: - { - number_unsigned = number_unsigned_t(0); - break; - } - - case value_t::number_float: - { - number_float = number_float_t(0.0); - break; - } - - case value_t::null: - { - object = nullptr; // silence warning, see #821 - break; - } - - default: - { - object = nullptr; // silence warning, see #821 - if (JSON_UNLIKELY(t == value_t::null)) - { - JSON_THROW(other_error::create(500, "961c151d2e87f2686a955a9be24d316f1362bf21 3.1.2")); // LCOV_EXCL_LINE - } - break; - } - } - } - - /// constructor for strings - json_value(const string_t& value) - { - string = create<string_t>(value); - } - - /// constructor for rvalue strings - json_value(string_t&& value) - { - string = create<string_t>(std::move(value)); - } - - /// constructor for objects - json_value(const object_t& value) - { - object = create<object_t>(value); - } - - /// constructor for rvalue objects - json_value(object_t&& value) - { - object = create<object_t>(std::move(value)); - } - - /// constructor for arrays - json_value(const array_t& value) - { - array = create<array_t>(value); - } - - /// constructor for rvalue arrays - json_value(array_t&& value) - { - array = create<array_t>(std::move(value)); - } - - void destroy(value_t t) noexcept - { - switch (t) - { - case value_t::object: - { - AllocatorType<object_t> alloc; - std::allocator_traits<decltype(alloc)>::destroy(alloc, object); - std::allocator_traits<decltype(alloc)>::deallocate(alloc, object, 1); - break; - } - - case value_t::array: - { - AllocatorType<array_t> alloc; - std::allocator_traits<decltype(alloc)>::destroy(alloc, array); - std::allocator_traits<decltype(alloc)>::deallocate(alloc, array, 1); - break; - } - - case value_t::string: - { - AllocatorType<string_t> alloc; - std::allocator_traits<decltype(alloc)>::destroy(alloc, string); - std::allocator_traits<decltype(alloc)>::deallocate(alloc, string, 1); - break; - } - - default: - { - break; - } - } - } - }; - - /*! - @brief checks the class invariants - - This function asserts the class invariants. It needs to be called at the - end of every constructor to make sure that created objects respect the - invariant. Furthermore, it has to be called each time the type of a JSON - value is changed, because the invariant expresses a relationship between - @a m_type and @a m_value. - */ - void assert_invariant() const noexcept - { - assert(m_type != value_t::object or m_value.object != nullptr); - assert(m_type != value_t::array or m_value.array != nullptr); - assert(m_type != value_t::string or m_value.string != nullptr); - } - - public: - ////////////////////////// - // JSON parser callback // - ////////////////////////// - - /*! - @brief parser event types - - The parser callback distinguishes the following events: - - `object_start`: the parser read `{` and started to process a JSON object - - `key`: the parser read a key of a value in an object - - `object_end`: the parser read `}` and finished processing a JSON object - - `array_start`: the parser read `[` and started to process a JSON array - - `array_end`: the parser read `]` and finished processing a JSON array - - `value`: the parser finished reading a JSON value - - @image html callback_events.png "Example when certain parse events are triggered" - - @sa @ref parser_callback_t for more information and examples - */ - using parse_event_t = typename parser::parse_event_t; - - /*! - @brief per-element parser callback type - - With a parser callback function, the result of parsing a JSON text can be - influenced. When passed to @ref parse, it is called on certain events - (passed as @ref parse_event_t via parameter @a event) with a set recursion - depth @a depth and context JSON value @a parsed. The return value of the - callback function is a boolean indicating whether the element that emitted - the callback shall be kept or not. - - We distinguish six scenarios (determined by the event type) in which the - callback function can be called. The following table describes the values - of the parameters @a depth, @a event, and @a parsed. - - parameter @a event | description | parameter @a depth | parameter @a parsed - ------------------ | ----------- | ------------------ | ------------------- - parse_event_t::object_start | the parser read `{` and started to process a JSON object | depth of the parent of the JSON object | a JSON value with type discarded - parse_event_t::key | the parser read a key of a value in an object | depth of the currently parsed JSON object | a JSON string containing the key - parse_event_t::object_end | the parser read `}` and finished processing a JSON object | depth of the parent of the JSON object | the parsed JSON object - parse_event_t::array_start | the parser read `[` and started to process a JSON array | depth of the parent of the JSON array | a JSON value with type discarded - parse_event_t::array_end | the parser read `]` and finished processing a JSON array | depth of the parent of the JSON array | the parsed JSON array - parse_event_t::value | the parser finished reading a JSON value | depth of the value | the parsed JSON value - - @image html callback_events.png "Example when certain parse events are triggered" - - Discarding a value (i.e., returning `false`) has different effects - depending on the context in which function was called: - - - Discarded values in structured types are skipped. That is, the parser - will behave as if the discarded value was never read. - - In case a value outside a structured type is skipped, it is replaced - with `null`. This case happens if the top-level element is skipped. - - @param[in] depth the depth of the recursion during parsing - - @param[in] event an event of type parse_event_t indicating the context in - the callback function has been called - - @param[in,out] parsed the current intermediate parse result; note that - writing to this value has no effect for parse_event_t::key events - - @return Whether the JSON value which called the function during parsing - should be kept (`true`) or not (`false`). In the latter case, it is either - skipped completely or replaced by an empty discarded object. - - @sa @ref parse for examples - - @since version 1.0.0 - */ - using parser_callback_t = typename parser::parser_callback_t; - - - ////////////////// - // constructors // - ////////////////// - - /// @name constructors and destructors - /// Constructors of class @ref basic_json, copy/move constructor, copy - /// assignment, static functions creating objects, and the destructor. - /// @{ - - /*! - @brief create an empty value with a given type - - Create an empty JSON value with a given type. The value will be default - initialized with an empty value which depends on the type: - - Value type | initial value - ----------- | ------------- - null | `null` - boolean | `false` - string | `""` - number | `0` - object | `{}` - array | `[]` - - @param[in] v the type of the value to create - - @complexity Constant. - - @exceptionsafety Strong guarantee: if an exception is thrown, there are no - changes to any JSON value. - - @liveexample{The following code shows the constructor for different @ref - value_t values,basic_json__value_t} - - @sa @ref clear() -- restores the postcondition of this constructor - - @since version 1.0.0 - */ - basic_json(const value_t v) - : m_type(v), m_value(v) - { - assert_invariant(); - } - - /*! - @brief create a null object - - Create a `null` JSON value. It either takes a null pointer as parameter - (explicitly creating `null`) or no parameter (implicitly creating `null`). - The passed null pointer itself is not read -- it is only used to choose - the right constructor. - - @complexity Constant. - - @exceptionsafety No-throw guarantee: this constructor never throws - exceptions. - - @liveexample{The following code shows the constructor with and without a - null pointer parameter.,basic_json__nullptr_t} - - @since version 1.0.0 - */ - basic_json(std::nullptr_t = nullptr) noexcept - : basic_json(value_t::null) - { - assert_invariant(); - } - - /*! - @brief create a JSON value - - This is a "catch all" constructor for all compatible JSON types; that is, - types for which a `to_json()` method exists. The constructor forwards the - parameter @a val to that method (to `json_serializer<U>::to_json` method - with `U = uncvref_t<CompatibleType>`, to be exact). - - Template type @a CompatibleType includes, but is not limited to, the - following types: - - **arrays**: @ref array_t and all kinds of compatible containers such as - `std::vector`, `std::deque`, `std::list`, `std::forward_list`, - `std::array`, `std::valarray`, `std::set`, `std::unordered_set`, - `std::multiset`, and `std::unordered_multiset` with a `value_type` from - which a @ref basic_json value can be constructed. - - **objects**: @ref object_t and all kinds of compatible associative - containers such as `std::map`, `std::unordered_map`, `std::multimap`, - and `std::unordered_multimap` with a `key_type` compatible to - @ref string_t and a `value_type` from which a @ref basic_json value can - be constructed. - - **strings**: @ref string_t, string literals, and all compatible string - containers can be used. - - **numbers**: @ref number_integer_t, @ref number_unsigned_t, - @ref number_float_t, and all convertible number types such as `int`, - `size_t`, `int64_t`, `float` or `double` can be used. - - **boolean**: @ref boolean_t / `bool` can be used. - - See the examples below. - - @tparam CompatibleType a type such that: - - @a CompatibleType is not derived from `std::istream`, - - @a CompatibleType is not @ref basic_json (to avoid hijacking copy/move - constructors), - - @a CompatibleType is not a different @ref basic_json type (i.e. with different template arguments) - - @a CompatibleType is not a @ref basic_json nested type (e.g., - @ref json_pointer, @ref iterator, etc ...) - - @ref @ref json_serializer<U> has a - `to_json(basic_json_t&, CompatibleType&&)` method - - @tparam U = `uncvref_t<CompatibleType>` - - @param[in] val the value to be forwarded to the respective constructor - - @complexity Usually linear in the size of the passed @a val, also - depending on the implementation of the called `to_json()` - method. - - @exceptionsafety Depends on the called constructor. For types directly - supported by the library (i.e., all types for which no `to_json()` function - was provided), strong guarantee holds: if an exception is thrown, there are - no changes to any JSON value. - - @liveexample{The following code shows the constructor with several - compatible types.,basic_json__CompatibleType} - - @since version 2.1.0 - */ - template <typename CompatibleType, - typename U = detail::uncvref_t<CompatibleType>, - detail::enable_if_t< - detail::is_compatible_type<basic_json_t, U>::value, int> = 0> - basic_json(CompatibleType && val) noexcept(noexcept( - JSONSerializer<U>::to_json(std::declval<basic_json_t&>(), - std::forward<CompatibleType>(val)))) - { - JSONSerializer<U>::to_json(*this, std::forward<CompatibleType>(val)); - assert_invariant(); - } - - /*! - @brief create a JSON value from an existing one - - This is a constructor for existing @ref basic_json types. - It does not hijack copy/move constructors, since the parameter has different - template arguments than the current ones. - - The constructor tries to convert the internal @ref m_value of the parameter. - - @tparam BasicJsonType a type such that: - - @a BasicJsonType is a @ref basic_json type. - - @a BasicJsonType has different template arguments than @ref basic_json_t. - - @param[in] val the @ref basic_json value to be converted. - - @complexity Usually linear in the size of the passed @a val, also - depending on the implementation of the called `to_json()` - method. - - @exceptionsafety Depends on the called constructor. For types directly - supported by the library (i.e., all types for which no `to_json()` function - was provided), strong guarantee holds: if an exception is thrown, there are - no changes to any JSON value. - - @since version 3.1.2 - */ - template <typename BasicJsonType, - detail::enable_if_t< - detail::is_basic_json<BasicJsonType>::value and not std::is_same<basic_json, BasicJsonType>::value, int> = 0> - basic_json(const BasicJsonType& val) - { - using other_boolean_t = typename BasicJsonType::boolean_t; - using other_number_float_t = typename BasicJsonType::number_float_t; - using other_number_integer_t = typename BasicJsonType::number_integer_t; - using other_number_unsigned_t = typename BasicJsonType::number_unsigned_t; - using other_string_t = typename BasicJsonType::string_t; - using other_object_t = typename BasicJsonType::object_t; - using other_array_t = typename BasicJsonType::array_t; - - switch (val.type()) - { - case value_t::boolean: - JSONSerializer<other_boolean_t>::to_json(*this, val.template get<other_boolean_t>()); - break; - case value_t::number_float: - JSONSerializer<other_number_float_t>::to_json(*this, val.template get<other_number_float_t>()); - break; - case value_t::number_integer: - JSONSerializer<other_number_integer_t>::to_json(*this, val.template get<other_number_integer_t>()); - break; - case value_t::number_unsigned: - JSONSerializer<other_number_unsigned_t>::to_json(*this, val.template get<other_number_unsigned_t>()); - break; - case value_t::string: - JSONSerializer<other_string_t>::to_json(*this, val.template get_ref<const other_string_t&>()); - break; - case value_t::object: - JSONSerializer<other_object_t>::to_json(*this, val.template get_ref<const other_object_t&>()); - break; - case value_t::array: - JSONSerializer<other_array_t>::to_json(*this, val.template get_ref<const other_array_t&>()); - break; - case value_t::null: - *this = nullptr; - break; - case value_t::discarded: - m_type = value_t::discarded; - break; - } - assert_invariant(); - } - - /*! - @brief create a container (array or object) from an initializer list - - Creates a JSON value of type array or object from the passed initializer - list @a init. In case @a type_deduction is `true` (default), the type of - the JSON value to be created is deducted from the initializer list @a init - according to the following rules: - - 1. If the list is empty, an empty JSON object value `{}` is created. - 2. If the list consists of pairs whose first element is a string, a JSON - object value is created where the first elements of the pairs are - treated as keys and the second elements are as values. - 3. In all other cases, an array is created. - - The rules aim to create the best fit between a C++ initializer list and - JSON values. The rationale is as follows: - - 1. The empty initializer list is written as `{}` which is exactly an empty - JSON object. - 2. C++ has no way of describing mapped types other than to list a list of - pairs. As JSON requires that keys must be of type string, rule 2 is the - weakest constraint one can pose on initializer lists to interpret them - as an object. - 3. In all other cases, the initializer list could not be interpreted as - JSON object type, so interpreting it as JSON array type is safe. - - With the rules described above, the following JSON values cannot be - expressed by an initializer list: - - - the empty array (`[]`): use @ref array(initializer_list_t) - with an empty initializer list in this case - - arrays whose elements satisfy rule 2: use @ref - array(initializer_list_t) with the same initializer list - in this case - - @note When used without parentheses around an empty initializer list, @ref - basic_json() is called instead of this function, yielding the JSON null - value. - - @param[in] init initializer list with JSON values - - @param[in] type_deduction internal parameter; when set to `true`, the type - of the JSON value is deducted from the initializer list @a init; when set - to `false`, the type provided via @a manual_type is forced. This mode is - used by the functions @ref array(initializer_list_t) and - @ref object(initializer_list_t). - - @param[in] manual_type internal parameter; when @a type_deduction is set - to `false`, the created JSON value will use the provided type (only @ref - value_t::array and @ref value_t::object are valid); when @a type_deduction - is set to `true`, this parameter has no effect - - @throw type_error.301 if @a type_deduction is `false`, @a manual_type is - `value_t::object`, but @a init contains an element which is not a pair - whose first element is a string. In this case, the constructor could not - create an object. If @a type_deduction would have be `true`, an array - would have been created. See @ref object(initializer_list_t) - for an example. - - @complexity Linear in the size of the initializer list @a init. - - @exceptionsafety Strong guarantee: if an exception is thrown, there are no - changes to any JSON value. - - @liveexample{The example below shows how JSON values are created from - initializer lists.,basic_json__list_init_t} - - @sa @ref array(initializer_list_t) -- create a JSON array - value from an initializer list - @sa @ref object(initializer_list_t) -- create a JSON object - value from an initializer list - - @since version 1.0.0 - */ - basic_json(initializer_list_t init, - bool type_deduction = true, - value_t manual_type = value_t::array) - { - // check if each element is an array with two elements whose first - // element is a string - bool is_an_object = std::all_of(init.begin(), init.end(), - [](const detail::json_ref<basic_json>& element_ref) - { - return (element_ref->is_array() and element_ref->size() == 2 and (*element_ref)[0].is_string()); - }); - - // adjust type if type deduction is not wanted - if (not type_deduction) - { - // if array is wanted, do not create an object though possible - if (manual_type == value_t::array) - { - is_an_object = false; - } - - // if object is wanted but impossible, throw an exception - if (JSON_UNLIKELY(manual_type == value_t::object and not is_an_object)) - { - JSON_THROW(type_error::create(301, "cannot create object from initializer list")); - } - } - - if (is_an_object) - { - // the initializer list is a list of pairs -> create object - m_type = value_t::object; - m_value = value_t::object; - - std::for_each(init.begin(), init.end(), [this](const detail::json_ref<basic_json>& element_ref) - { - auto element = element_ref.moved_or_copied(); - m_value.object->emplace( - std::move(*((*element.m_value.array)[0].m_value.string)), - std::move((*element.m_value.array)[1])); - }); - } - else - { - // the initializer list describes an array -> create array - m_type = value_t::array; - m_value.array = create<array_t>(init.begin(), init.end()); - } - - assert_invariant(); - } - - /*! - @brief explicitly create an array from an initializer list - - Creates a JSON array value from a given initializer list. That is, given a - list of values `a, b, c`, creates the JSON value `[a, b, c]`. If the - initializer list is empty, the empty array `[]` is created. - - @note This function is only needed to express two edge cases that cannot - be realized with the initializer list constructor (@ref - basic_json(initializer_list_t, bool, value_t)). These cases - are: - 1. creating an array whose elements are all pairs whose first element is a - string -- in this case, the initializer list constructor would create an - object, taking the first elements as keys - 2. creating an empty array -- passing the empty initializer list to the - initializer list constructor yields an empty object - - @param[in] init initializer list with JSON values to create an array from - (optional) - - @return JSON array value - - @complexity Linear in the size of @a init. - - @exceptionsafety Strong guarantee: if an exception is thrown, there are no - changes to any JSON value. - - @liveexample{The following code shows an example for the `array` - function.,array} - - @sa @ref basic_json(initializer_list_t, bool, value_t) -- - create a JSON value from an initializer list - @sa @ref object(initializer_list_t) -- create a JSON object - value from an initializer list - - @since version 1.0.0 - */ - static basic_json array(initializer_list_t init = {}) - { - return basic_json(init, false, value_t::array); - } - - /*! - @brief explicitly create an object from an initializer list - - Creates a JSON object value from a given initializer list. The initializer - lists elements must be pairs, and their first elements must be strings. If - the initializer list is empty, the empty object `{}` is created. - - @note This function is only added for symmetry reasons. In contrast to the - related function @ref array(initializer_list_t), there are - no cases which can only be expressed by this function. That is, any - initializer list @a init can also be passed to the initializer list - constructor @ref basic_json(initializer_list_t, bool, value_t). - - @param[in] init initializer list to create an object from (optional) - - @return JSON object value - - @throw type_error.301 if @a init is not a list of pairs whose first - elements are strings. In this case, no object can be created. When such a - value is passed to @ref basic_json(initializer_list_t, bool, value_t), - an array would have been created from the passed initializer list @a init. - See example below. - - @complexity Linear in the size of @a init. - - @exceptionsafety Strong guarantee: if an exception is thrown, there are no - changes to any JSON value. - - @liveexample{The following code shows an example for the `object` - function.,object} - - @sa @ref basic_json(initializer_list_t, bool, value_t) -- - create a JSON value from an initializer list - @sa @ref array(initializer_list_t) -- create a JSON array - value from an initializer list - - @since version 1.0.0 - */ - static basic_json object(initializer_list_t init = {}) - { - return basic_json(init, false, value_t::object); - } - - /*! - @brief construct an array with count copies of given value - - Constructs a JSON array value by creating @a cnt copies of a passed value. - In case @a cnt is `0`, an empty array is created. - - @param[in] cnt the number of JSON copies of @a val to create - @param[in] val the JSON value to copy - - @post `std::distance(begin(),end()) == cnt` holds. - - @complexity Linear in @a cnt. - - @exceptionsafety Strong guarantee: if an exception is thrown, there are no - changes to any JSON value. - - @liveexample{The following code shows examples for the @ref - basic_json(size_type\, const basic_json&) - constructor.,basic_json__size_type_basic_json} - - @since version 1.0.0 - */ - basic_json(size_type cnt, const basic_json& val) - : m_type(value_t::array) - { - m_value.array = create<array_t>(cnt, val); - assert_invariant(); - } - - /*! - @brief construct a JSON container given an iterator range - - Constructs the JSON value with the contents of the range `[first, last)`. - The semantics depends on the different types a JSON value can have: - - In case of a null type, invalid_iterator.206 is thrown. - - In case of other primitive types (number, boolean, or string), @a first - must be `begin()` and @a last must be `end()`. In this case, the value is - copied. Otherwise, invalid_iterator.204 is thrown. - - In case of structured types (array, object), the constructor behaves as - similar versions for `std::vector` or `std::map`; that is, a JSON array - or object is constructed from the values in the range. - - @tparam InputIT an input iterator type (@ref iterator or @ref - const_iterator) - - @param[in] first begin of the range to copy from (included) - @param[in] last end of the range to copy from (excluded) - - @pre Iterators @a first and @a last must be initialized. **This - precondition is enforced with an assertion (see warning).** If - assertions are switched off, a violation of this precondition yields - undefined behavior. - - @pre Range `[first, last)` is valid. Usually, this precondition cannot be - checked efficiently. Only certain edge cases are detected; see the - description of the exceptions below. A violation of this precondition - yields undefined behavior. - - @warning A precondition is enforced with a runtime assertion that will - result in calling `std::abort` if this precondition is not met. - Assertions can be disabled by defining `NDEBUG` at compile time. - See http://en.cppreference.com/w/cpp/error/assert for more - information. - - @throw invalid_iterator.201 if iterators @a first and @a last are not - compatible (i.e., do not belong to the same JSON value). In this case, - the range `[first, last)` is undefined. - @throw invalid_iterator.204 if iterators @a first and @a last belong to a - primitive type (number, boolean, or string), but @a first does not point - to the first element any more. In this case, the range `[first, last)` is - undefined. See example code below. - @throw invalid_iterator.206 if iterators @a first and @a last belong to a - null value. In this case, the range `[first, last)` is undefined. - - @complexity Linear in distance between @a first and @a last. - - @exceptionsafety Strong guarantee: if an exception is thrown, there are no - changes to any JSON value. - - @liveexample{The example below shows several ways to create JSON values by - specifying a subrange with iterators.,basic_json__InputIt_InputIt} - - @since version 1.0.0 - */ - template<class InputIT, typename std::enable_if< - std::is_same<InputIT, typename basic_json_t::iterator>::value or - std::is_same<InputIT, typename basic_json_t::const_iterator>::value, int>::type = 0> - basic_json(InputIT first, InputIT last) - { - assert(first.m_object != nullptr); - assert(last.m_object != nullptr); - - // make sure iterator fits the current value - if (JSON_UNLIKELY(first.m_object != last.m_object)) - { - JSON_THROW(invalid_iterator::create(201, "iterators are not compatible")); - } - - // copy type from first iterator - m_type = first.m_object->m_type; - - // check if iterator range is complete for primitive values - switch (m_type) - { - case value_t::boolean: - case value_t::number_float: - case value_t::number_integer: - case value_t::number_unsigned: - case value_t::string: - { - if (JSON_UNLIKELY(not first.m_it.primitive_iterator.is_begin() - or not last.m_it.primitive_iterator.is_end())) - { - JSON_THROW(invalid_iterator::create(204, "iterators out of range")); - } - break; - } - - default: - break; - } - - switch (m_type) - { - case value_t::number_integer: - { - m_value.number_integer = first.m_object->m_value.number_integer; - break; - } - - case value_t::number_unsigned: - { - m_value.number_unsigned = first.m_object->m_value.number_unsigned; - break; - } - - case value_t::number_float: - { - m_value.number_float = first.m_object->m_value.number_float; - break; - } - - case value_t::boolean: - { - m_value.boolean = first.m_object->m_value.boolean; - break; - } - - case value_t::string: - { - m_value = *first.m_object->m_value.string; - break; - } - - case value_t::object: - { - m_value.object = create<object_t>(first.m_it.object_iterator, - last.m_it.object_iterator); - break; - } - - case value_t::array: - { - m_value.array = create<array_t>(first.m_it.array_iterator, - last.m_it.array_iterator); - break; - } - - default: - JSON_THROW(invalid_iterator::create(206, "cannot construct with iterators from " + - std::string(first.m_object->type_name()))); - } - - assert_invariant(); - } - - - /////////////////////////////////////// - // other constructors and destructor // - /////////////////////////////////////// - - /// @private - basic_json(const detail::json_ref<basic_json>& ref) - : basic_json(ref.moved_or_copied()) - {} - - /*! - @brief copy constructor - - Creates a copy of a given JSON value. - - @param[in] other the JSON value to copy - - @post `*this == other` - - @complexity Linear in the size of @a other. - - @exceptionsafety Strong guarantee: if an exception is thrown, there are no - changes to any JSON value. - - @requirement This function helps `basic_json` satisfying the - [Container](http://en.cppreference.com/w/cpp/concept/Container) - requirements: - - The complexity is linear. - - As postcondition, it holds: `other == basic_json(other)`. - - @liveexample{The following code shows an example for the copy - constructor.,basic_json__basic_json} - - @since version 1.0.0 - */ - basic_json(const basic_json& other) - : m_type(other.m_type) - { - // check of passed value is valid - other.assert_invariant(); - - switch (m_type) - { - case value_t::object: - { - m_value = *other.m_value.object; - break; - } - - case value_t::array: - { - m_value = *other.m_value.array; - break; - } - - case value_t::string: - { - m_value = *other.m_value.string; - break; - } - - case value_t::boolean: - { - m_value = other.m_value.boolean; - break; - } - - case value_t::number_integer: - { - m_value = other.m_value.number_integer; - break; - } - - case value_t::number_unsigned: - { - m_value = other.m_value.number_unsigned; - break; - } - - case value_t::number_float: - { - m_value = other.m_value.number_float; - break; - } - - default: - break; - } - - assert_invariant(); - } - - /*! - @brief move constructor - - Move constructor. Constructs a JSON value with the contents of the given - value @a other using move semantics. It "steals" the resources from @a - other and leaves it as JSON null value. - - @param[in,out] other value to move to this object - - @post `*this` has the same value as @a other before the call. - @post @a other is a JSON null value. - - @complexity Constant. - - @exceptionsafety No-throw guarantee: this constructor never throws - exceptions. - - @requirement This function helps `basic_json` satisfying the - [MoveConstructible](http://en.cppreference.com/w/cpp/concept/MoveConstructible) - requirements. - - @liveexample{The code below shows the move constructor explicitly called - via std::move.,basic_json__moveconstructor} - - @since version 1.0.0 - */ - basic_json(basic_json&& other) noexcept - : m_type(std::move(other.m_type)), - m_value(std::move(other.m_value)) - { - // check that passed value is valid - other.assert_invariant(); - - // invalidate payload - other.m_type = value_t::null; - other.m_value = {}; - - assert_invariant(); - } - - /*! - @brief copy assignment - - Copy assignment operator. Copies a JSON value via the "copy and swap" - strategy: It is expressed in terms of the copy constructor, destructor, - and the `swap()` member function. - - @param[in] other value to copy from - - @complexity Linear. - - @requirement This function helps `basic_json` satisfying the - [Container](http://en.cppreference.com/w/cpp/concept/Container) - requirements: - - The complexity is linear. - - @liveexample{The code below shows and example for the copy assignment. It - creates a copy of value `a` which is then swapped with `b`. Finally\, the - copy of `a` (which is the null value after the swap) is - destroyed.,basic_json__copyassignment} - - @since version 1.0.0 - */ - reference& operator=(basic_json other) noexcept ( - std::is_nothrow_move_constructible<value_t>::value and - std::is_nothrow_move_assignable<value_t>::value and - std::is_nothrow_move_constructible<json_value>::value and - std::is_nothrow_move_assignable<json_value>::value - ) - { - // check that passed value is valid - other.assert_invariant(); - - using std::swap; - swap(m_type, other.m_type); - swap(m_value, other.m_value); - - assert_invariant(); - return *this; - } - - /*! - @brief destructor - - Destroys the JSON value and frees all allocated memory. - - @complexity Linear. - - @requirement This function helps `basic_json` satisfying the - [Container](http://en.cppreference.com/w/cpp/concept/Container) - requirements: - - The complexity is linear. - - All stored elements are destroyed and all memory is freed. - - @since version 1.0.0 - */ - ~basic_json() noexcept - { - assert_invariant(); - m_value.destroy(m_type); - } - - /// @} - - public: - /////////////////////// - // object inspection // - /////////////////////// - - /// @name object inspection - /// Functions to inspect the type of a JSON value. - /// @{ - - /*! - @brief serialization - - Serialization function for JSON values. The function tries to mimic - Python's `json.dumps()` function, and currently supports its @a indent - and @a ensure_ascii parameters. - - @param[in] indent If indent is nonnegative, then array elements and object - members will be pretty-printed with that indent level. An indent level of - `0` will only insert newlines. `-1` (the default) selects the most compact - representation. - @param[in] indent_char The character to use for indentation if @a indent is - greater than `0`. The default is ` ` (space). - @param[in] ensure_ascii If @a ensure_ascii is true, all non-ASCII characters - in the output are escaped with `\uXXXX` sequences, and the result consists - of ASCII characters only. - - @return string containing the serialization of the JSON value - - @throw type_error.316 if a string stored inside the JSON value is not - UTF-8 encoded - - @complexity Linear. - - @exceptionsafety Strong guarantee: if an exception is thrown, there are no - changes in the JSON value. - - @liveexample{The following example shows the effect of different @a indent\, - @a indent_char\, and @a ensure_ascii parameters to the result of the - serialization.,dump} - - @see https://docs.python.org/2/library/json.html#json.dump - - @since version 1.0.0; indentation character @a indent_char, option - @a ensure_ascii and exceptions added in version 3.0.0 - */ - string_t dump(const int indent = -1, const char indent_char = ' ', - const bool ensure_ascii = false) const - { - string_t result; - serializer s(detail::output_adapter<char, string_t>(result), indent_char); - - if (indent >= 0) - { - s.dump(*this, true, ensure_ascii, static_cast<unsigned int>(indent)); - } - else - { - s.dump(*this, false, ensure_ascii, 0); - } - - return result; - } - - /*! - @brief return the type of the JSON value (explicit) - - Return the type of the JSON value as a value from the @ref value_t - enumeration. - - @return the type of the JSON value - Value type | return value - ------------------------- | ------------------------- - null | value_t::null - boolean | value_t::boolean - string | value_t::string - number (integer) | value_t::number_integer - number (unsigned integer) | value_t::number_unsigned - number (floating-point) | value_t::number_float - object | value_t::object - array | value_t::array - discarded | value_t::discarded - - @complexity Constant. - - @exceptionsafety No-throw guarantee: this member function never throws - exceptions. - - @liveexample{The following code exemplifies `type()` for all JSON - types.,type} - - @sa @ref operator value_t() -- return the type of the JSON value (implicit) - @sa @ref type_name() -- return the type as string - - @since version 1.0.0 - */ - constexpr value_t type() const noexcept - { - return m_type; - } - - /*! - @brief return whether type is primitive - - This function returns true if and only if the JSON type is primitive - (string, number, boolean, or null). - - @return `true` if type is primitive (string, number, boolean, or null), - `false` otherwise. - - @complexity Constant. - - @exceptionsafety No-throw guarantee: this member function never throws - exceptions. - - @liveexample{The following code exemplifies `is_primitive()` for all JSON - types.,is_primitive} - - @sa @ref is_structured() -- returns whether JSON value is structured - @sa @ref is_null() -- returns whether JSON value is `null` - @sa @ref is_string() -- returns whether JSON value is a string - @sa @ref is_boolean() -- returns whether JSON value is a boolean - @sa @ref is_number() -- returns whether JSON value is a number - - @since version 1.0.0 - */ - constexpr bool is_primitive() const noexcept - { - return is_null() or is_string() or is_boolean() or is_number(); - } - - /*! - @brief return whether type is structured - - This function returns true if and only if the JSON type is structured - (array or object). - - @return `true` if type is structured (array or object), `false` otherwise. - - @complexity Constant. - - @exceptionsafety No-throw guarantee: this member function never throws - exceptions. - - @liveexample{The following code exemplifies `is_structured()` for all JSON - types.,is_structured} - - @sa @ref is_primitive() -- returns whether value is primitive - @sa @ref is_array() -- returns whether value is an array - @sa @ref is_object() -- returns whether value is an object - - @since version 1.0.0 - */ - constexpr bool is_structured() const noexcept - { - return is_array() or is_object(); - } - - /*! - @brief return whether value is null - - This function returns true if and only if the JSON value is null. - - @return `true` if type is null, `false` otherwise. - - @complexity Constant. - - @exceptionsafety No-throw guarantee: this member function never throws - exceptions. - - @liveexample{The following code exemplifies `is_null()` for all JSON - types.,is_null} - - @since version 1.0.0 - */ - constexpr bool is_null() const noexcept - { - return (m_type == value_t::null); - } - - /*! - @brief return whether value is a boolean - - This function returns true if and only if the JSON value is a boolean. - - @return `true` if type is boolean, `false` otherwise. - - @complexity Constant. - - @exceptionsafety No-throw guarantee: this member function never throws - exceptions. - - @liveexample{The following code exemplifies `is_boolean()` for all JSON - types.,is_boolean} - - @since version 1.0.0 - */ - constexpr bool is_boolean() const noexcept - { - return (m_type == value_t::boolean); - } - - /*! - @brief return whether value is a number - - This function returns true if and only if the JSON value is a number. This - includes both integer (signed and unsigned) and floating-point values. - - @return `true` if type is number (regardless whether integer, unsigned - integer or floating-type), `false` otherwise. - - @complexity Constant. - - @exceptionsafety No-throw guarantee: this member function never throws - exceptions. - - @liveexample{The following code exemplifies `is_number()` for all JSON - types.,is_number} - - @sa @ref is_number_integer() -- check if value is an integer or unsigned - integer number - @sa @ref is_number_unsigned() -- check if value is an unsigned integer - number - @sa @ref is_number_float() -- check if value is a floating-point number - - @since version 1.0.0 - */ - constexpr bool is_number() const noexcept - { - return is_number_integer() or is_number_float(); - } - - /*! - @brief return whether value is an integer number - - This function returns true if and only if the JSON value is a signed or - unsigned integer number. This excludes floating-point values. - - @return `true` if type is an integer or unsigned integer number, `false` - otherwise. - - @complexity Constant. - - @exceptionsafety No-throw guarantee: this member function never throws - exceptions. - - @liveexample{The following code exemplifies `is_number_integer()` for all - JSON types.,is_number_integer} - - @sa @ref is_number() -- check if value is a number - @sa @ref is_number_unsigned() -- check if value is an unsigned integer - number - @sa @ref is_number_float() -- check if value is a floating-point number - - @since version 1.0.0 - */ - constexpr bool is_number_integer() const noexcept - { - return (m_type == value_t::number_integer or m_type == value_t::number_unsigned); - } - - /*! - @brief return whether value is an unsigned integer number - - This function returns true if and only if the JSON value is an unsigned - integer number. This excludes floating-point and signed integer values. - - @return `true` if type is an unsigned integer number, `false` otherwise. - - @complexity Constant. - - @exceptionsafety No-throw guarantee: this member function never throws - exceptions. - - @liveexample{The following code exemplifies `is_number_unsigned()` for all - JSON types.,is_number_unsigned} - - @sa @ref is_number() -- check if value is a number - @sa @ref is_number_integer() -- check if value is an integer or unsigned - integer number - @sa @ref is_number_float() -- check if value is a floating-point number - - @since version 2.0.0 - */ - constexpr bool is_number_unsigned() const noexcept - { - return (m_type == value_t::number_unsigned); - } - - /*! - @brief return whether value is a floating-point number - - This function returns true if and only if the JSON value is a - floating-point number. This excludes signed and unsigned integer values. - - @return `true` if type is a floating-point number, `false` otherwise. - - @complexity Constant. - - @exceptionsafety No-throw guarantee: this member function never throws - exceptions. - - @liveexample{The following code exemplifies `is_number_float()` for all - JSON types.,is_number_float} - - @sa @ref is_number() -- check if value is number - @sa @ref is_number_integer() -- check if value is an integer number - @sa @ref is_number_unsigned() -- check if value is an unsigned integer - number - - @since version 1.0.0 - */ - constexpr bool is_number_float() const noexcept - { - return (m_type == value_t::number_float); - } - - /*! - @brief return whether value is an object - - This function returns true if and only if the JSON value is an object. - - @return `true` if type is object, `false` otherwise. - - @complexity Constant. - - @exceptionsafety No-throw guarantee: this member function never throws - exceptions. - - @liveexample{The following code exemplifies `is_object()` for all JSON - types.,is_object} - - @since version 1.0.0 - */ - constexpr bool is_object() const noexcept - { - return (m_type == value_t::object); - } - - /*! - @brief return whether value is an array - - This function returns true if and only if the JSON value is an array. - - @return `true` if type is array, `false` otherwise. - - @complexity Constant. - - @exceptionsafety No-throw guarantee: this member function never throws - exceptions. - - @liveexample{The following code exemplifies `is_array()` for all JSON - types.,is_array} - - @since version 1.0.0 - */ - constexpr bool is_array() const noexcept - { - return (m_type == value_t::array); - } - - /*! - @brief return whether value is a string - - This function returns true if and only if the JSON value is a string. - - @return `true` if type is string, `false` otherwise. - - @complexity Constant. - - @exceptionsafety No-throw guarantee: this member function never throws - exceptions. - - @liveexample{The following code exemplifies `is_string()` for all JSON - types.,is_string} - - @since version 1.0.0 - */ - constexpr bool is_string() const noexcept - { - return (m_type == value_t::string); - } - - /*! - @brief return whether value is discarded - - This function returns true if and only if the JSON value was discarded - during parsing with a callback function (see @ref parser_callback_t). - - @note This function will always be `false` for JSON values after parsing. - That is, discarded values can only occur during parsing, but will be - removed when inside a structured value or replaced by null in other cases. - - @return `true` if type is discarded, `false` otherwise. - - @complexity Constant. - - @exceptionsafety No-throw guarantee: this member function never throws - exceptions. - - @liveexample{The following code exemplifies `is_discarded()` for all JSON - types.,is_discarded} - - @since version 1.0.0 - */ - constexpr bool is_discarded() const noexcept - { - return (m_type == value_t::discarded); - } - - /*! - @brief return the type of the JSON value (implicit) - - Implicitly return the type of the JSON value as a value from the @ref - value_t enumeration. - - @return the type of the JSON value - - @complexity Constant. - - @exceptionsafety No-throw guarantee: this member function never throws - exceptions. - - @liveexample{The following code exemplifies the @ref value_t operator for - all JSON types.,operator__value_t} - - @sa @ref type() -- return the type of the JSON value (explicit) - @sa @ref type_name() -- return the type as string - - @since version 1.0.0 - */ - constexpr operator value_t() const noexcept - { - return m_type; - } - - /// @} - - private: - ////////////////// - // value access // - ////////////////// - - /// get a boolean (explicit) - boolean_t get_impl(boolean_t* /*unused*/) const - { - if (JSON_LIKELY(is_boolean())) - { - return m_value.boolean; - } - - JSON_THROW(type_error::create(302, "type must be boolean, but is " + std::string(type_name()))); - } - - /// get a pointer to the value (object) - object_t* get_impl_ptr(object_t* /*unused*/) noexcept - { - return is_object() ? m_value.object : nullptr; - } - - /// get a pointer to the value (object) - constexpr const object_t* get_impl_ptr(const object_t* /*unused*/) const noexcept - { - return is_object() ? m_value.object : nullptr; - } - - /// get a pointer to the value (array) - array_t* get_impl_ptr(array_t* /*unused*/) noexcept - { - return is_array() ? m_value.array : nullptr; - } - - /// get a pointer to the value (array) - constexpr const array_t* get_impl_ptr(const array_t* /*unused*/) const noexcept - { - return is_array() ? m_value.array : nullptr; - } - - /// get a pointer to the value (string) - string_t* get_impl_ptr(string_t* /*unused*/) noexcept - { - return is_string() ? m_value.string : nullptr; - } - - /// get a pointer to the value (string) - constexpr const string_t* get_impl_ptr(const string_t* /*unused*/) const noexcept - { - return is_string() ? m_value.string : nullptr; - } - - /// get a pointer to the value (boolean) - boolean_t* get_impl_ptr(boolean_t* /*unused*/) noexcept - { - return is_boolean() ? &m_value.boolean : nullptr; - } - - /// get a pointer to the value (boolean) - constexpr const boolean_t* get_impl_ptr(const boolean_t* /*unused*/) const noexcept - { - return is_boolean() ? &m_value.boolean : nullptr; - } - - /// get a pointer to the value (integer number) - number_integer_t* get_impl_ptr(number_integer_t* /*unused*/) noexcept - { - return is_number_integer() ? &m_value.number_integer : nullptr; - } - - /// get a pointer to the value (integer number) - constexpr const number_integer_t* get_impl_ptr(const number_integer_t* /*unused*/) const noexcept - { - return is_number_integer() ? &m_value.number_integer : nullptr; - } - - /// get a pointer to the value (unsigned number) - number_unsigned_t* get_impl_ptr(number_unsigned_t* /*unused*/) noexcept - { - return is_number_unsigned() ? &m_value.number_unsigned : nullptr; - } - - /// get a pointer to the value (unsigned number) - constexpr const number_unsigned_t* get_impl_ptr(const number_unsigned_t* /*unused*/) const noexcept - { - return is_number_unsigned() ? &m_value.number_unsigned : nullptr; - } - - /// get a pointer to the value (floating-point number) - number_float_t* get_impl_ptr(number_float_t* /*unused*/) noexcept - { - return is_number_float() ? &m_value.number_float : nullptr; - } - - /// get a pointer to the value (floating-point number) - constexpr const number_float_t* get_impl_ptr(const number_float_t* /*unused*/) const noexcept - { - return is_number_float() ? &m_value.number_float : nullptr; - } - - /*! - @brief helper function to implement get_ref() - - This function helps to implement get_ref() without code duplication for - const and non-const overloads - - @tparam ThisType will be deduced as `basic_json` or `const basic_json` - - @throw type_error.303 if ReferenceType does not match underlying value - type of the current JSON - */ - template<typename ReferenceType, typename ThisType> - static ReferenceType get_ref_impl(ThisType& obj) - { - // delegate the call to get_ptr<>() - auto ptr = obj.template get_ptr<typename std::add_pointer<ReferenceType>::type>(); - - if (JSON_LIKELY(ptr != nullptr)) - { - return *ptr; - } - - JSON_THROW(type_error::create(303, "incompatible ReferenceType for get_ref, actual type is " + std::string(obj.type_name()))); - } - - public: - /// @name value access - /// Direct access to the stored value of a JSON value. - /// @{ - - /*! - @brief get special-case overload - - This overloads avoids a lot of template boilerplate, it can be seen as the - identity method - - @tparam BasicJsonType == @ref basic_json - - @return a copy of *this - - @complexity Constant. - - @since version 2.1.0 - */ - template<typename BasicJsonType, detail::enable_if_t< - std::is_same<typename std::remove_const<BasicJsonType>::type, basic_json_t>::value, - int> = 0> - basic_json get() const - { - return *this; - } - - /*! - @brief get special-case overload - - This overloads converts the current @ref basic_json in a different - @ref basic_json type - - @tparam BasicJsonType == @ref basic_json - - @return a copy of *this, converted into @tparam BasicJsonType - - @complexity Depending on the implementation of the called `from_json()` - method. - - @since version 3.1.2 - */ - template<typename BasicJsonType, detail::enable_if_t< - not std::is_same<BasicJsonType, basic_json>::value and - detail::is_basic_json<BasicJsonType>::value, int> = 0> - BasicJsonType get() const - { - return *this; - } - - /*! - @brief get a value (explicit) - - Explicit type conversion between the JSON value and a compatible value - which is [CopyConstructible](http://en.cppreference.com/w/cpp/concept/CopyConstructible) - and [DefaultConstructible](http://en.cppreference.com/w/cpp/concept/DefaultConstructible). - The value is converted by calling the @ref json_serializer<ValueType> - `from_json()` method. - - The function is equivalent to executing - @code {.cpp} - ValueType ret; - JSONSerializer<ValueType>::from_json(*this, ret); - return ret; - @endcode - - This overloads is chosen if: - - @a ValueType is not @ref basic_json, - - @ref json_serializer<ValueType> has a `from_json()` method of the form - `void from_json(const basic_json&, ValueType&)`, and - - @ref json_serializer<ValueType> does not have a `from_json()` method of - the form `ValueType from_json(const basic_json&)` - - @tparam ValueTypeCV the provided value type - @tparam ValueType the returned value type - - @return copy of the JSON value, converted to @a ValueType - - @throw what @ref json_serializer<ValueType> `from_json()` method throws - - @liveexample{The example below shows several conversions from JSON values - to other types. There a few things to note: (1) Floating-point numbers can - be converted to integers\, (2) A JSON array can be converted to a standard - `std::vector<short>`\, (3) A JSON object can be converted to C++ - associative containers such as `std::unordered_map<std::string\, - json>`.,get__ValueType_const} - - @since version 2.1.0 - */ - template<typename ValueTypeCV, typename ValueType = detail::uncvref_t<ValueTypeCV>, - detail::enable_if_t < - not detail::is_basic_json<ValueType>::value and - detail::has_from_json<basic_json_t, ValueType>::value and - not detail::has_non_default_from_json<basic_json_t, ValueType>::value, - int> = 0> - ValueType get() const noexcept(noexcept( - JSONSerializer<ValueType>::from_json(std::declval<const basic_json_t&>(), std::declval<ValueType&>()))) - { - // we cannot static_assert on ValueTypeCV being non-const, because - // there is support for get<const basic_json_t>(), which is why we - // still need the uncvref - static_assert(not std::is_reference<ValueTypeCV>::value, - "get() cannot be used with reference types, you might want to use get_ref()"); - static_assert(std::is_default_constructible<ValueType>::value, - "types must be DefaultConstructible when used with get()"); - - ValueType ret; - JSONSerializer<ValueType>::from_json(*this, ret); - return ret; - } - - /*! - @brief get a value (explicit); special case - - Explicit type conversion between the JSON value and a compatible value - which is **not** [CopyConstructible](http://en.cppreference.com/w/cpp/concept/CopyConstructible) - and **not** [DefaultConstructible](http://en.cppreference.com/w/cpp/concept/DefaultConstructible). - The value is converted by calling the @ref json_serializer<ValueType> - `from_json()` method. - - The function is equivalent to executing - @code {.cpp} - return JSONSerializer<ValueTypeCV>::from_json(*this); - @endcode - - This overloads is chosen if: - - @a ValueType is not @ref basic_json and - - @ref json_serializer<ValueType> has a `from_json()` method of the form - `ValueType from_json(const basic_json&)` - - @note If @ref json_serializer<ValueType> has both overloads of - `from_json()`, this one is chosen. - - @tparam ValueTypeCV the provided value type - @tparam ValueType the returned value type - - @return copy of the JSON value, converted to @a ValueType - - @throw what @ref json_serializer<ValueType> `from_json()` method throws - - @since version 2.1.0 - */ - template<typename ValueTypeCV, typename ValueType = detail::uncvref_t<ValueTypeCV>, - detail::enable_if_t<not std::is_same<basic_json_t, ValueType>::value and - detail::has_non_default_from_json<basic_json_t, ValueType>::value, - int> = 0> - ValueType get() const noexcept(noexcept( - JSONSerializer<ValueTypeCV>::from_json(std::declval<const basic_json_t&>()))) - { - static_assert(not std::is_reference<ValueTypeCV>::value, - "get() cannot be used with reference types, you might want to use get_ref()"); - return JSONSerializer<ValueTypeCV>::from_json(*this); - } - - /*! - @brief get a pointer value (explicit) - - Explicit pointer access to the internally stored JSON value. No copies are - made. - - @warning The pointer becomes invalid if the underlying JSON object - changes. - - @tparam PointerType pointer type; must be a pointer to @ref array_t, @ref - object_t, @ref string_t, @ref boolean_t, @ref number_integer_t, - @ref number_unsigned_t, or @ref number_float_t. - - @return pointer to the internally stored JSON value if the requested - pointer type @a PointerType fits to the JSON value; `nullptr` otherwise - - @complexity Constant. - - @liveexample{The example below shows how pointers to internal values of a - JSON value can be requested. Note that no type conversions are made and a - `nullptr` is returned if the value and the requested pointer type does not - match.,get__PointerType} - - @sa @ref get_ptr() for explicit pointer-member access - - @since version 1.0.0 - */ - template<typename PointerType, typename std::enable_if< - std::is_pointer<PointerType>::value, int>::type = 0> - PointerType get() noexcept - { - // delegate the call to get_ptr - return get_ptr<PointerType>(); - } - - /*! - @brief get a pointer value (explicit) - @copydoc get() - */ - template<typename PointerType, typename std::enable_if< - std::is_pointer<PointerType>::value, int>::type = 0> - constexpr const PointerType get() const noexcept - { - // delegate the call to get_ptr - return get_ptr<PointerType>(); - } - - /*! - @brief get a pointer value (implicit) - - Implicit pointer access to the internally stored JSON value. No copies are - made. - - @warning Writing data to the pointee of the result yields an undefined - state. - - @tparam PointerType pointer type; must be a pointer to @ref array_t, @ref - object_t, @ref string_t, @ref boolean_t, @ref number_integer_t, - @ref number_unsigned_t, or @ref number_float_t. Enforced by a static - assertion. - - @return pointer to the internally stored JSON value if the requested - pointer type @a PointerType fits to the JSON value; `nullptr` otherwise - - @complexity Constant. - - @liveexample{The example below shows how pointers to internal values of a - JSON value can be requested. Note that no type conversions are made and a - `nullptr` is returned if the value and the requested pointer type does not - match.,get_ptr} - - @since version 1.0.0 - */ - template<typename PointerType, typename std::enable_if< - std::is_pointer<PointerType>::value, int>::type = 0> - PointerType get_ptr() noexcept - { - // get the type of the PointerType (remove pointer and const) - using pointee_t = typename std::remove_const<typename - std::remove_pointer<typename - std::remove_const<PointerType>::type>::type>::type; - // make sure the type matches the allowed types - static_assert( - std::is_same<object_t, pointee_t>::value - or std::is_same<array_t, pointee_t>::value - or std::is_same<string_t, pointee_t>::value - or std::is_same<boolean_t, pointee_t>::value - or std::is_same<number_integer_t, pointee_t>::value - or std::is_same<number_unsigned_t, pointee_t>::value - or std::is_same<number_float_t, pointee_t>::value - , "incompatible pointer type"); - - // delegate the call to get_impl_ptr<>() - return get_impl_ptr(static_cast<PointerType>(nullptr)); - } - - /*! - @brief get a pointer value (implicit) - @copydoc get_ptr() - */ - template<typename PointerType, typename std::enable_if< - std::is_pointer<PointerType>::value and - std::is_const<typename std::remove_pointer<PointerType>::type>::value, int>::type = 0> - constexpr const PointerType get_ptr() const noexcept - { - // get the type of the PointerType (remove pointer and const) - using pointee_t = typename std::remove_const<typename - std::remove_pointer<typename - std::remove_const<PointerType>::type>::type>::type; - // make sure the type matches the allowed types - static_assert( - std::is_same<object_t, pointee_t>::value - or std::is_same<array_t, pointee_t>::value - or std::is_same<string_t, pointee_t>::value - or std::is_same<boolean_t, pointee_t>::value - or std::is_same<number_integer_t, pointee_t>::value - or std::is_same<number_unsigned_t, pointee_t>::value - or std::is_same<number_float_t, pointee_t>::value - , "incompatible pointer type"); - - // delegate the call to get_impl_ptr<>() const - return get_impl_ptr(static_cast<PointerType>(nullptr)); - } - - /*! - @brief get a reference value (implicit) - - Implicit reference access to the internally stored JSON value. No copies - are made. - - @warning Writing data to the referee of the result yields an undefined - state. - - @tparam ReferenceType reference type; must be a reference to @ref array_t, - @ref object_t, @ref string_t, @ref boolean_t, @ref number_integer_t, or - @ref number_float_t. Enforced by static assertion. - - @return reference to the internally stored JSON value if the requested - reference type @a ReferenceType fits to the JSON value; throws - type_error.303 otherwise - - @throw type_error.303 in case passed type @a ReferenceType is incompatible - with the stored JSON value; see example below - - @complexity Constant. - - @liveexample{The example shows several calls to `get_ref()`.,get_ref} - - @since version 1.1.0 - */ - template<typename ReferenceType, typename std::enable_if< - std::is_reference<ReferenceType>::value, int>::type = 0> - ReferenceType get_ref() - { - // delegate call to get_ref_impl - return get_ref_impl<ReferenceType>(*this); - } - - /*! - @brief get a reference value (implicit) - @copydoc get_ref() - */ - template<typename ReferenceType, typename std::enable_if< - std::is_reference<ReferenceType>::value and - std::is_const<typename std::remove_reference<ReferenceType>::type>::value, int>::type = 0> - ReferenceType get_ref() const - { - // delegate call to get_ref_impl - return get_ref_impl<ReferenceType>(*this); - } - - /*! - @brief get a value (implicit) - - Implicit type conversion between the JSON value and a compatible value. - The call is realized by calling @ref get() const. - - @tparam ValueType non-pointer type compatible to the JSON value, for - instance `int` for JSON integer numbers, `bool` for JSON booleans, or - `std::vector` types for JSON arrays. The character type of @ref string_t - as well as an initializer list of this type is excluded to avoid - ambiguities as these types implicitly convert to `std::string`. - - @return copy of the JSON value, converted to type @a ValueType - - @throw type_error.302 in case passed type @a ValueType is incompatible - to the JSON value type (e.g., the JSON value is of type boolean, but a - string is requested); see example below - - @complexity Linear in the size of the JSON value. - - @liveexample{The example below shows several conversions from JSON values - to other types. There a few things to note: (1) Floating-point numbers can - be converted to integers\, (2) A JSON array can be converted to a standard - `std::vector<short>`\, (3) A JSON object can be converted to C++ - associative containers such as `std::unordered_map<std::string\, - json>`.,operator__ValueType} - - @since version 1.0.0 - */ - template < typename ValueType, typename std::enable_if < - not std::is_pointer<ValueType>::value and - not std::is_same<ValueType, detail::json_ref<basic_json>>::value and - not std::is_same<ValueType, typename string_t::value_type>::value and - not detail::is_basic_json<ValueType>::value -#ifndef _MSC_VER // fix for issue #167 operator<< ambiguity under VS2015 - and not std::is_same<ValueType, std::initializer_list<typename string_t::value_type>>::value -#endif -#if defined(JSON_HAS_CPP_17) - and not std::is_same<ValueType, typename std::string_view>::value -#endif - , int >::type = 0 > - operator ValueType() const - { - // delegate the call to get<>() const - return get<ValueType>(); - } - - /// @} - - - //////////////////// - // element access // - //////////////////// - - /// @name element access - /// Access to the JSON value. - /// @{ - - /*! - @brief access specified array element with bounds checking - - Returns a reference to the element at specified location @a idx, with - bounds checking. - - @param[in] idx index of the element to access - - @return reference to the element at index @a idx - - @throw type_error.304 if the JSON value is not an array; in this case, - calling `at` with an index makes no sense. See example below. - @throw out_of_range.401 if the index @a idx is out of range of the array; - that is, `idx >= size()`. See example below. - - @exceptionsafety Strong guarantee: if an exception is thrown, there are no - changes in the JSON value. - - @complexity Constant. - - @since version 1.0.0 - - @liveexample{The example below shows how array elements can be read and - written using `at()`. It also demonstrates the different exceptions that - can be thrown.,at__size_type} - */ - reference at(size_type idx) - { - // at only works for arrays - if (JSON_LIKELY(is_array())) - { - JSON_TRY - { - return m_value.array->at(idx); - } - JSON_CATCH (std::out_of_range&) - { - // create better exception explanation - JSON_THROW(out_of_range::create(401, "array index " + std::to_string(idx) + " is out of range")); - } - } - else - { - JSON_THROW(type_error::create(304, "cannot use at() with " + std::string(type_name()))); - } - } - - /*! - @brief access specified array element with bounds checking - - Returns a const reference to the element at specified location @a idx, - with bounds checking. - - @param[in] idx index of the element to access - - @return const reference to the element at index @a idx - - @throw type_error.304 if the JSON value is not an array; in this case, - calling `at` with an index makes no sense. See example below. - @throw out_of_range.401 if the index @a idx is out of range of the array; - that is, `idx >= size()`. See example below. - - @exceptionsafety Strong guarantee: if an exception is thrown, there are no - changes in the JSON value. - - @complexity Constant. - - @since version 1.0.0 - - @liveexample{The example below shows how array elements can be read using - `at()`. It also demonstrates the different exceptions that can be thrown., - at__size_type_const} - */ - const_reference at(size_type idx) const - { - // at only works for arrays - if (JSON_LIKELY(is_array())) - { - JSON_TRY - { - return m_value.array->at(idx); - } - JSON_CATCH (std::out_of_range&) - { - // create better exception explanation - JSON_THROW(out_of_range::create(401, "array index " + std::to_string(idx) + " is out of range")); - } - } - else - { - JSON_THROW(type_error::create(304, "cannot use at() with " + std::string(type_name()))); - } - } - - /*! - @brief access specified object element with bounds checking - - Returns a reference to the element at with specified key @a key, with - bounds checking. - - @param[in] key key of the element to access - - @return reference to the element at key @a key - - @throw type_error.304 if the JSON value is not an object; in this case, - calling `at` with a key makes no sense. See example below. - @throw out_of_range.403 if the key @a key is is not stored in the object; - that is, `find(key) == end()`. See example below. - - @exceptionsafety Strong guarantee: if an exception is thrown, there are no - changes in the JSON value. - - @complexity Logarithmic in the size of the container. - - @sa @ref operator[](const typename object_t::key_type&) for unchecked - access by reference - @sa @ref value() for access by value with a default value - - @since version 1.0.0 - - @liveexample{The example below shows how object elements can be read and - written using `at()`. It also demonstrates the different exceptions that - can be thrown.,at__object_t_key_type} - */ - reference at(const typename object_t::key_type& key) - { - // at only works for objects - if (JSON_LIKELY(is_object())) - { - JSON_TRY - { - return m_value.object->at(key); - } - JSON_CATCH (std::out_of_range&) - { - // create better exception explanation - JSON_THROW(out_of_range::create(403, "key '" + key + "' not found")); - } - } - else - { - JSON_THROW(type_error::create(304, "cannot use at() with " + std::string(type_name()))); - } - } - - /*! - @brief access specified object element with bounds checking - - Returns a const reference to the element at with specified key @a key, - with bounds checking. - - @param[in] key key of the element to access - - @return const reference to the element at key @a key - - @throw type_error.304 if the JSON value is not an object; in this case, - calling `at` with a key makes no sense. See example below. - @throw out_of_range.403 if the key @a key is is not stored in the object; - that is, `find(key) == end()`. See example below. - - @exceptionsafety Strong guarantee: if an exception is thrown, there are no - changes in the JSON value. - - @complexity Logarithmic in the size of the container. - - @sa @ref operator[](const typename object_t::key_type&) for unchecked - access by reference - @sa @ref value() for access by value with a default value - - @since version 1.0.0 - - @liveexample{The example below shows how object elements can be read using - `at()`. It also demonstrates the different exceptions that can be thrown., - at__object_t_key_type_const} - */ - const_reference at(const typename object_t::key_type& key) const - { - // at only works for objects - if (JSON_LIKELY(is_object())) - { - JSON_TRY - { - return m_value.object->at(key); - } - JSON_CATCH (std::out_of_range&) - { - // create better exception explanation - JSON_THROW(out_of_range::create(403, "key '" + key + "' not found")); - } - } - else - { - JSON_THROW(type_error::create(304, "cannot use at() with " + std::string(type_name()))); - } - } - - /*! - @brief access specified array element - - Returns a reference to the element at specified location @a idx. - - @note If @a idx is beyond the range of the array (i.e., `idx >= size()`), - then the array is silently filled up with `null` values to make `idx` a - valid reference to the last stored element. - - @param[in] idx index of the element to access - - @return reference to the element at index @a idx - - @throw type_error.305 if the JSON value is not an array or null; in that - cases, using the [] operator with an index makes no sense. - - @complexity Constant if @a idx is in the range of the array. Otherwise - linear in `idx - size()`. - - @liveexample{The example below shows how array elements can be read and - written using `[]` operator. Note the addition of `null` - values.,operatorarray__size_type} - - @since version 1.0.0 - */ - reference operator[](size_type idx) - { - // implicitly convert null value to an empty array - if (is_null()) - { - m_type = value_t::array; - m_value.array = create<array_t>(); - assert_invariant(); - } - - // operator[] only works for arrays - if (JSON_LIKELY(is_array())) - { - // fill up array with null values if given idx is outside range - if (idx >= m_value.array->size()) - { - m_value.array->insert(m_value.array->end(), - idx - m_value.array->size() + 1, - basic_json()); - } - - return m_value.array->operator[](idx); - } - - JSON_THROW(type_error::create(305, "cannot use operator[] with " + std::string(type_name()))); - } - - /*! - @brief access specified array element - - Returns a const reference to the element at specified location @a idx. - - @param[in] idx index of the element to access - - @return const reference to the element at index @a idx - - @throw type_error.305 if the JSON value is not an array; in that case, - using the [] operator with an index makes no sense. - - @complexity Constant. - - @liveexample{The example below shows how array elements can be read using - the `[]` operator.,operatorarray__size_type_const} - - @since version 1.0.0 - */ - const_reference operator[](size_type idx) const - { - // const operator[] only works for arrays - if (JSON_LIKELY(is_array())) - { - return m_value.array->operator[](idx); - } - - JSON_THROW(type_error::create(305, "cannot use operator[] with " + std::string(type_name()))); - } - - /*! - @brief access specified object element - - Returns a reference to the element at with specified key @a key. - - @note If @a key is not found in the object, then it is silently added to - the object and filled with a `null` value to make `key` a valid reference. - In case the value was `null` before, it is converted to an object. - - @param[in] key key of the element to access - - @return reference to the element at key @a key - - @throw type_error.305 if the JSON value is not an object or null; in that - cases, using the [] operator with a key makes no sense. - - @complexity Logarithmic in the size of the container. - - @liveexample{The example below shows how object elements can be read and - written using the `[]` operator.,operatorarray__key_type} - - @sa @ref at(const typename object_t::key_type&) for access by reference - with range checking - @sa @ref value() for access by value with a default value - - @since version 1.0.0 - */ - reference operator[](const typename object_t::key_type& key) - { - // implicitly convert null value to an empty object - if (is_null()) - { - m_type = value_t::object; - m_value.object = create<object_t>(); - assert_invariant(); - } - - // operator[] only works for objects - if (JSON_LIKELY(is_object())) - { - return m_value.object->operator[](key); - } - - JSON_THROW(type_error::create(305, "cannot use operator[] with " + std::string(type_name()))); - } - - /*! - @brief read-only access specified object element - - Returns a const reference to the element at with specified key @a key. No - bounds checking is performed. - - @warning If the element with key @a key does not exist, the behavior is - undefined. - - @param[in] key key of the element to access - - @return const reference to the element at key @a key - - @pre The element with key @a key must exist. **This precondition is - enforced with an assertion.** - - @throw type_error.305 if the JSON value is not an object; in that case, - using the [] operator with a key makes no sense. - - @complexity Logarithmic in the size of the container. - - @liveexample{The example below shows how object elements can be read using - the `[]` operator.,operatorarray__key_type_const} - - @sa @ref at(const typename object_t::key_type&) for access by reference - with range checking - @sa @ref value() for access by value with a default value - - @since version 1.0.0 - */ - const_reference operator[](const typename object_t::key_type& key) const - { - // const operator[] only works for objects - if (JSON_LIKELY(is_object())) - { - assert(m_value.object->find(key) != m_value.object->end()); - return m_value.object->find(key)->second; - } - - JSON_THROW(type_error::create(305, "cannot use operator[] with " + std::string(type_name()))); - } - - /*! - @brief access specified object element - - Returns a reference to the element at with specified key @a key. - - @note If @a key is not found in the object, then it is silently added to - the object and filled with a `null` value to make `key` a valid reference. - In case the value was `null` before, it is converted to an object. - - @param[in] key key of the element to access - - @return reference to the element at key @a key - - @throw type_error.305 if the JSON value is not an object or null; in that - cases, using the [] operator with a key makes no sense. - - @complexity Logarithmic in the size of the container. - - @liveexample{The example below shows how object elements can be read and - written using the `[]` operator.,operatorarray__key_type} - - @sa @ref at(const typename object_t::key_type&) for access by reference - with range checking - @sa @ref value() for access by value with a default value - - @since version 1.1.0 - */ - template<typename T> - reference operator[](T* key) - { - // implicitly convert null to object - if (is_null()) - { - m_type = value_t::object; - m_value = value_t::object; - assert_invariant(); - } - - // at only works for objects - if (JSON_LIKELY(is_object())) - { - return m_value.object->operator[](key); - } - - JSON_THROW(type_error::create(305, "cannot use operator[] with " + std::string(type_name()))); - } - - /*! - @brief read-only access specified object element - - Returns a const reference to the element at with specified key @a key. No - bounds checking is performed. - - @warning If the element with key @a key does not exist, the behavior is - undefined. - - @param[in] key key of the element to access - - @return const reference to the element at key @a key - - @pre The element with key @a key must exist. **This precondition is - enforced with an assertion.** - - @throw type_error.305 if the JSON value is not an object; in that case, - using the [] operator with a key makes no sense. - - @complexity Logarithmic in the size of the container. - - @liveexample{The example below shows how object elements can be read using - the `[]` operator.,operatorarray__key_type_const} - - @sa @ref at(const typename object_t::key_type&) for access by reference - with range checking - @sa @ref value() for access by value with a default value - - @since version 1.1.0 - */ - template<typename T> - const_reference operator[](T* key) const - { - // at only works for objects - if (JSON_LIKELY(is_object())) - { - assert(m_value.object->find(key) != m_value.object->end()); - return m_value.object->find(key)->second; - } - - JSON_THROW(type_error::create(305, "cannot use operator[] with " + std::string(type_name()))); - } - - /*! - @brief access specified object element with default value - - Returns either a copy of an object's element at the specified key @a key - or a given default value if no element with key @a key exists. - - The function is basically equivalent to executing - @code {.cpp} - try { - return at(key); - } catch(out_of_range) { - return default_value; - } - @endcode - - @note Unlike @ref at(const typename object_t::key_type&), this function - does not throw if the given key @a key was not found. - - @note Unlike @ref operator[](const typename object_t::key_type& key), this - function does not implicitly add an element to the position defined by @a - key. This function is furthermore also applicable to const objects. - - @param[in] key key of the element to access - @param[in] default_value the value to return if @a key is not found - - @tparam ValueType type compatible to JSON values, for instance `int` for - JSON integer numbers, `bool` for JSON booleans, or `std::vector` types for - JSON arrays. Note the type of the expected value at @a key and the default - value @a default_value must be compatible. - - @return copy of the element at key @a key or @a default_value if @a key - is not found - - @throw type_error.306 if the JSON value is not an object; in that case, - using `value()` with a key makes no sense. - - @complexity Logarithmic in the size of the container. - - @liveexample{The example below shows how object elements can be queried - with a default value.,basic_json__value} - - @sa @ref at(const typename object_t::key_type&) for access by reference - with range checking - @sa @ref operator[](const typename object_t::key_type&) for unchecked - access by reference - - @since version 1.0.0 - */ - template<class ValueType, typename std::enable_if< - std::is_convertible<basic_json_t, ValueType>::value, int>::type = 0> - ValueType value(const typename object_t::key_type& key, const ValueType& default_value) const - { - // at only works for objects - if (JSON_LIKELY(is_object())) - { - // if key is found, return value and given default value otherwise - const auto it = find(key); - if (it != end()) - { - return *it; - } - - return default_value; - } - - JSON_THROW(type_error::create(306, "cannot use value() with " + std::string(type_name()))); - } - - /*! - @brief overload for a default value of type const char* - @copydoc basic_json::value(const typename object_t::key_type&, ValueType) const - */ - string_t value(const typename object_t::key_type& key, const char* default_value) const - { - return value(key, string_t(default_value)); - } - - /*! - @brief access specified object element via JSON Pointer with default value - - Returns either a copy of an object's element at the specified key @a key - or a given default value if no element with key @a key exists. - - The function is basically equivalent to executing - @code {.cpp} - try { - return at(ptr); - } catch(out_of_range) { - return default_value; - } - @endcode - - @note Unlike @ref at(const json_pointer&), this function does not throw - if the given key @a key was not found. - - @param[in] ptr a JSON pointer to the element to access - @param[in] default_value the value to return if @a ptr found no value - - @tparam ValueType type compatible to JSON values, for instance `int` for - JSON integer numbers, `bool` for JSON booleans, or `std::vector` types for - JSON arrays. Note the type of the expected value at @a key and the default - value @a default_value must be compatible. - - @return copy of the element at key @a key or @a default_value if @a key - is not found - - @throw type_error.306 if the JSON value is not an objec; in that case, - using `value()` with a key makes no sense. - - @complexity Logarithmic in the size of the container. - - @liveexample{The example below shows how object elements can be queried - with a default value.,basic_json__value_ptr} - - @sa @ref operator[](const json_pointer&) for unchecked access by reference - - @since version 2.0.2 - */ - template<class ValueType, typename std::enable_if< - std::is_convertible<basic_json_t, ValueType>::value, int>::type = 0> - ValueType value(const json_pointer& ptr, const ValueType& default_value) const - { - // at only works for objects - if (JSON_LIKELY(is_object())) - { - // if pointer resolves a value, return it or use default value - JSON_TRY - { - return ptr.get_checked(this); - } - JSON_CATCH (out_of_range&) - { - return default_value; - } - } - - JSON_THROW(type_error::create(306, "cannot use value() with " + std::string(type_name()))); - } - - /*! - @brief overload for a default value of type const char* - @copydoc basic_json::value(const json_pointer&, ValueType) const - */ - string_t value(const json_pointer& ptr, const char* default_value) const - { - return value(ptr, string_t(default_value)); - } - - /*! - @brief access the first element - - Returns a reference to the first element in the container. For a JSON - container `c`, the expression `c.front()` is equivalent to `*c.begin()`. - - @return In case of a structured type (array or object), a reference to the - first element is returned. In case of number, string, or boolean values, a - reference to the value is returned. - - @complexity Constant. - - @pre The JSON value must not be `null` (would throw `std::out_of_range`) - or an empty array or object (undefined behavior, **guarded by - assertions**). - @post The JSON value remains unchanged. - - @throw invalid_iterator.214 when called on `null` value - - @liveexample{The following code shows an example for `front()`.,front} - - @sa @ref back() -- access the last element - - @since version 1.0.0 - */ - reference front() - { - return *begin(); - } - - /*! - @copydoc basic_json::front() - */ - const_reference front() const - { - return *cbegin(); - } - - /*! - @brief access the last element - - Returns a reference to the last element in the container. For a JSON - container `c`, the expression `c.back()` is equivalent to - @code {.cpp} - auto tmp = c.end(); - --tmp; - return *tmp; - @endcode - - @return In case of a structured type (array or object), a reference to the - last element is returned. In case of number, string, or boolean values, a - reference to the value is returned. - - @complexity Constant. - - @pre The JSON value must not be `null` (would throw `std::out_of_range`) - or an empty array or object (undefined behavior, **guarded by - assertions**). - @post The JSON value remains unchanged. - - @throw invalid_iterator.214 when called on a `null` value. See example - below. - - @liveexample{The following code shows an example for `back()`.,back} - - @sa @ref front() -- access the first element - - @since version 1.0.0 - */ - reference back() - { - auto tmp = end(); - --tmp; - return *tmp; - } - - /*! - @copydoc basic_json::back() - */ - const_reference back() const - { - auto tmp = cend(); - --tmp; - return *tmp; - } - - /*! - @brief remove element given an iterator - - Removes the element specified by iterator @a pos. The iterator @a pos must - be valid and dereferenceable. Thus the `end()` iterator (which is valid, - but is not dereferenceable) cannot be used as a value for @a pos. - - If called on a primitive type other than `null`, the resulting JSON value - will be `null`. - - @param[in] pos iterator to the element to remove - @return Iterator following the last removed element. If the iterator @a - pos refers to the last element, the `end()` iterator is returned. - - @tparam IteratorType an @ref iterator or @ref const_iterator - - @post Invalidates iterators and references at or after the point of the - erase, including the `end()` iterator. - - @throw type_error.307 if called on a `null` value; example: `"cannot use - erase() with null"` - @throw invalid_iterator.202 if called on an iterator which does not belong - to the current JSON value; example: `"iterator does not fit current - value"` - @throw invalid_iterator.205 if called on a primitive type with invalid - iterator (i.e., any iterator which is not `begin()`); example: `"iterator - out of range"` - - @complexity The complexity depends on the type: - - objects: amortized constant - - arrays: linear in distance between @a pos and the end of the container - - strings: linear in the length of the string - - other types: constant - - @liveexample{The example shows the result of `erase()` for different JSON - types.,erase__IteratorType} - - @sa @ref erase(IteratorType, IteratorType) -- removes the elements in - the given range - @sa @ref erase(const typename object_t::key_type&) -- removes the element - from an object at the given key - @sa @ref erase(const size_type) -- removes the element from an array at - the given index - - @since version 1.0.0 - */ - template<class IteratorType, typename std::enable_if< - std::is_same<IteratorType, typename basic_json_t::iterator>::value or - std::is_same<IteratorType, typename basic_json_t::const_iterator>::value, int>::type - = 0> - IteratorType erase(IteratorType pos) - { - // make sure iterator fits the current value - if (JSON_UNLIKELY(this != pos.m_object)) - { - JSON_THROW(invalid_iterator::create(202, "iterator does not fit current value")); - } - - IteratorType result = end(); - - switch (m_type) - { - case value_t::boolean: - case value_t::number_float: - case value_t::number_integer: - case value_t::number_unsigned: - case value_t::string: - { - if (JSON_UNLIKELY(not pos.m_it.primitive_iterator.is_begin())) - { - JSON_THROW(invalid_iterator::create(205, "iterator out of range")); - } - - if (is_string()) - { - AllocatorType<string_t> alloc; - std::allocator_traits<decltype(alloc)>::destroy(alloc, m_value.string); - std::allocator_traits<decltype(alloc)>::deallocate(alloc, m_value.string, 1); - m_value.string = nullptr; - } - - m_type = value_t::null; - assert_invariant(); - break; - } - - case value_t::object: - { - result.m_it.object_iterator = m_value.object->erase(pos.m_it.object_iterator); - break; - } - - case value_t::array: - { - result.m_it.array_iterator = m_value.array->erase(pos.m_it.array_iterator); - break; - } - - default: - JSON_THROW(type_error::create(307, "cannot use erase() with " + std::string(type_name()))); - } - - return result; - } - - /*! - @brief remove elements given an iterator range - - Removes the element specified by the range `[first; last)`. The iterator - @a first does not need to be dereferenceable if `first == last`: erasing - an empty range is a no-op. - - If called on a primitive type other than `null`, the resulting JSON value - will be `null`. - - @param[in] first iterator to the beginning of the range to remove - @param[in] last iterator past the end of the range to remove - @return Iterator following the last removed element. If the iterator @a - second refers to the last element, the `end()` iterator is returned. - - @tparam IteratorType an @ref iterator or @ref const_iterator - - @post Invalidates iterators and references at or after the point of the - erase, including the `end()` iterator. - - @throw type_error.307 if called on a `null` value; example: `"cannot use - erase() with null"` - @throw invalid_iterator.203 if called on iterators which does not belong - to the current JSON value; example: `"iterators do not fit current value"` - @throw invalid_iterator.204 if called on a primitive type with invalid - iterators (i.e., if `first != begin()` and `last != end()`); example: - `"iterators out of range"` - - @complexity The complexity depends on the type: - - objects: `log(size()) + std::distance(first, last)` - - arrays: linear in the distance between @a first and @a last, plus linear - in the distance between @a last and end of the container - - strings: linear in the length of the string - - other types: constant - - @liveexample{The example shows the result of `erase()` for different JSON - types.,erase__IteratorType_IteratorType} - - @sa @ref erase(IteratorType) -- removes the element at a given position - @sa @ref erase(const typename object_t::key_type&) -- removes the element - from an object at the given key - @sa @ref erase(const size_type) -- removes the element from an array at - the given index - - @since version 1.0.0 - */ - template<class IteratorType, typename std::enable_if< - std::is_same<IteratorType, typename basic_json_t::iterator>::value or - std::is_same<IteratorType, typename basic_json_t::const_iterator>::value, int>::type - = 0> - IteratorType erase(IteratorType first, IteratorType last) - { - // make sure iterator fits the current value - if (JSON_UNLIKELY(this != first.m_object or this != last.m_object)) - { - JSON_THROW(invalid_iterator::create(203, "iterators do not fit current value")); - } - - IteratorType result = end(); - - switch (m_type) - { - case value_t::boolean: - case value_t::number_float: - case value_t::number_integer: - case value_t::number_unsigned: - case value_t::string: - { - if (JSON_LIKELY(not first.m_it.primitive_iterator.is_begin() - or not last.m_it.primitive_iterator.is_end())) - { - JSON_THROW(invalid_iterator::create(204, "iterators out of range")); - } - - if (is_string()) - { - AllocatorType<string_t> alloc; - std::allocator_traits<decltype(alloc)>::destroy(alloc, m_value.string); - std::allocator_traits<decltype(alloc)>::deallocate(alloc, m_value.string, 1); - m_value.string = nullptr; - } - - m_type = value_t::null; - assert_invariant(); - break; - } - - case value_t::object: - { - result.m_it.object_iterator = m_value.object->erase(first.m_it.object_iterator, - last.m_it.object_iterator); - break; - } - - case value_t::array: - { - result.m_it.array_iterator = m_value.array->erase(first.m_it.array_iterator, - last.m_it.array_iterator); - break; - } - - default: - JSON_THROW(type_error::create(307, "cannot use erase() with " + std::string(type_name()))); - } - - return result; - } - - /*! - @brief remove element from a JSON object given a key - - Removes elements from a JSON object with the key value @a key. - - @param[in] key value of the elements to remove - - @return Number of elements removed. If @a ObjectType is the default - `std::map` type, the return value will always be `0` (@a key was not - found) or `1` (@a key was found). - - @post References and iterators to the erased elements are invalidated. - Other references and iterators are not affected. - - @throw type_error.307 when called on a type other than JSON object; - example: `"cannot use erase() with null"` - - @complexity `log(size()) + count(key)` - - @liveexample{The example shows the effect of `erase()`.,erase__key_type} - - @sa @ref erase(IteratorType) -- removes the element at a given position - @sa @ref erase(IteratorType, IteratorType) -- removes the elements in - the given range - @sa @ref erase(const size_type) -- removes the element from an array at - the given index - - @since version 1.0.0 - */ - size_type erase(const typename object_t::key_type& key) - { - // this erase only works for objects - if (JSON_LIKELY(is_object())) - { - return m_value.object->erase(key); - } - - JSON_THROW(type_error::create(307, "cannot use erase() with " + std::string(type_name()))); - } - - /*! - @brief remove element from a JSON array given an index - - Removes element from a JSON array at the index @a idx. - - @param[in] idx index of the element to remove - - @throw type_error.307 when called on a type other than JSON object; - example: `"cannot use erase() with null"` - @throw out_of_range.401 when `idx >= size()`; example: `"array index 17 - is out of range"` - - @complexity Linear in distance between @a idx and the end of the container. - - @liveexample{The example shows the effect of `erase()`.,erase__size_type} - - @sa @ref erase(IteratorType) -- removes the element at a given position - @sa @ref erase(IteratorType, IteratorType) -- removes the elements in - the given range - @sa @ref erase(const typename object_t::key_type&) -- removes the element - from an object at the given key - - @since version 1.0.0 - */ - void erase(const size_type idx) - { - // this erase only works for arrays - if (JSON_LIKELY(is_array())) - { - if (JSON_UNLIKELY(idx >= size())) - { - JSON_THROW(out_of_range::create(401, "array index " + std::to_string(idx) + " is out of range")); - } - - m_value.array->erase(m_value.array->begin() + static_cast<difference_type>(idx)); - } - else - { - JSON_THROW(type_error::create(307, "cannot use erase() with " + std::string(type_name()))); - } - } - - /// @} - - - //////////// - // lookup // - //////////// - - /// @name lookup - /// @{ - - /*! - @brief find an element in a JSON object - - Finds an element in a JSON object with key equivalent to @a key. If the - element is not found or the JSON value is not an object, end() is - returned. - - @note This method always returns @ref end() when executed on a JSON type - that is not an object. - - @param[in] key key value of the element to search for. - - @return Iterator to an element with key equivalent to @a key. If no such - element is found or the JSON value is not an object, past-the-end (see - @ref end()) iterator is returned. - - @complexity Logarithmic in the size of the JSON object. - - @liveexample{The example shows how `find()` is used.,find__key_type} - - @since version 1.0.0 - */ - template<typename KeyT> - iterator find(KeyT&& key) - { - auto result = end(); - - if (is_object()) - { - result.m_it.object_iterator = m_value.object->find(std::forward<KeyT>(key)); - } - - return result; - } - - /*! - @brief find an element in a JSON object - @copydoc find(KeyT&&) - */ - template<typename KeyT> - const_iterator find(KeyT&& key) const - { - auto result = cend(); - - if (is_object()) - { - result.m_it.object_iterator = m_value.object->find(std::forward<KeyT>(key)); - } - - return result; - } - - /*! - @brief returns the number of occurrences of a key in a JSON object - - Returns the number of elements with key @a key. If ObjectType is the - default `std::map` type, the return value will always be `0` (@a key was - not found) or `1` (@a key was found). - - @note This method always returns `0` when executed on a JSON type that is - not an object. - - @param[in] key key value of the element to count - - @return Number of elements with key @a key. If the JSON value is not an - object, the return value will be `0`. - - @complexity Logarithmic in the size of the JSON object. - - @liveexample{The example shows how `count()` is used.,count} - - @since version 1.0.0 - */ - template<typename KeyT> - size_type count(KeyT&& key) const - { - // return 0 for all nonobject types - return is_object() ? m_value.object->count(std::forward<KeyT>(key)) : 0; - } - - /// @} - - - /////////////// - // iterators // - /////////////// - - /// @name iterators - /// @{ - - /*! - @brief returns an iterator to the first element - - Returns an iterator to the first element. - - @image html range-begin-end.svg "Illustration from cppreference.com" - - @return iterator to the first element - - @complexity Constant. - - @requirement This function helps `basic_json` satisfying the - [Container](http://en.cppreference.com/w/cpp/concept/Container) - requirements: - - The complexity is constant. - - @liveexample{The following code shows an example for `begin()`.,begin} - - @sa @ref cbegin() -- returns a const iterator to the beginning - @sa @ref end() -- returns an iterator to the end - @sa @ref cend() -- returns a const iterator to the end - - @since version 1.0.0 - */ - iterator begin() noexcept - { - iterator result(this); - result.set_begin(); - return result; - } - - /*! - @copydoc basic_json::cbegin() - */ - const_iterator begin() const noexcept - { - return cbegin(); - } - - /*! - @brief returns a const iterator to the first element - - Returns a const iterator to the first element. - - @image html range-begin-end.svg "Illustration from cppreference.com" - - @return const iterator to the first element - - @complexity Constant. - - @requirement This function helps `basic_json` satisfying the - [Container](http://en.cppreference.com/w/cpp/concept/Container) - requirements: - - The complexity is constant. - - Has the semantics of `const_cast<const basic_json&>(*this).begin()`. - - @liveexample{The following code shows an example for `cbegin()`.,cbegin} - - @sa @ref begin() -- returns an iterator to the beginning - @sa @ref end() -- returns an iterator to the end - @sa @ref cend() -- returns a const iterator to the end - - @since version 1.0.0 - */ - const_iterator cbegin() const noexcept - { - const_iterator result(this); - result.set_begin(); - return result; - } - - /*! - @brief returns an iterator to one past the last element - - Returns an iterator to one past the last element. - - @image html range-begin-end.svg "Illustration from cppreference.com" - - @return iterator one past the last element - - @complexity Constant. - - @requirement This function helps `basic_json` satisfying the - [Container](http://en.cppreference.com/w/cpp/concept/Container) - requirements: - - The complexity is constant. - - @liveexample{The following code shows an example for `end()`.,end} - - @sa @ref cend() -- returns a const iterator to the end - @sa @ref begin() -- returns an iterator to the beginning - @sa @ref cbegin() -- returns a const iterator to the beginning - - @since version 1.0.0 - */ - iterator end() noexcept - { - iterator result(this); - result.set_end(); - return result; - } - - /*! - @copydoc basic_json::cend() - */ - const_iterator end() const noexcept - { - return cend(); - } - - /*! - @brief returns a const iterator to one past the last element - - Returns a const iterator to one past the last element. - - @image html range-begin-end.svg "Illustration from cppreference.com" - - @return const iterator one past the last element - - @complexity Constant. - - @requirement This function helps `basic_json` satisfying the - [Container](http://en.cppreference.com/w/cpp/concept/Container) - requirements: - - The complexity is constant. - - Has the semantics of `const_cast<const basic_json&>(*this).end()`. - - @liveexample{The following code shows an example for `cend()`.,cend} - - @sa @ref end() -- returns an iterator to the end - @sa @ref begin() -- returns an iterator to the beginning - @sa @ref cbegin() -- returns a const iterator to the beginning - - @since version 1.0.0 - */ - const_iterator cend() const noexcept - { - const_iterator result(this); - result.set_end(); - return result; - } - - /*! - @brief returns an iterator to the reverse-beginning - - Returns an iterator to the reverse-beginning; that is, the last element. - - @image html range-rbegin-rend.svg "Illustration from cppreference.com" - - @complexity Constant. - - @requirement This function helps `basic_json` satisfying the - [ReversibleContainer](http://en.cppreference.com/w/cpp/concept/ReversibleContainer) - requirements: - - The complexity is constant. - - Has the semantics of `reverse_iterator(end())`. - - @liveexample{The following code shows an example for `rbegin()`.,rbegin} - - @sa @ref crbegin() -- returns a const reverse iterator to the beginning - @sa @ref rend() -- returns a reverse iterator to the end - @sa @ref crend() -- returns a const reverse iterator to the end - - @since version 1.0.0 - */ - reverse_iterator rbegin() noexcept - { - return reverse_iterator(end()); - } - - /*! - @copydoc basic_json::crbegin() - */ - const_reverse_iterator rbegin() const noexcept - { - return crbegin(); - } - - /*! - @brief returns an iterator to the reverse-end - - Returns an iterator to the reverse-end; that is, one before the first - element. - - @image html range-rbegin-rend.svg "Illustration from cppreference.com" - - @complexity Constant. - - @requirement This function helps `basic_json` satisfying the - [ReversibleContainer](http://en.cppreference.com/w/cpp/concept/ReversibleContainer) - requirements: - - The complexity is constant. - - Has the semantics of `reverse_iterator(begin())`. - - @liveexample{The following code shows an example for `rend()`.,rend} - - @sa @ref crend() -- returns a const reverse iterator to the end - @sa @ref rbegin() -- returns a reverse iterator to the beginning - @sa @ref crbegin() -- returns a const reverse iterator to the beginning - - @since version 1.0.0 - */ - reverse_iterator rend() noexcept - { - return reverse_iterator(begin()); - } - - /*! - @copydoc basic_json::crend() - */ - const_reverse_iterator rend() const noexcept - { - return crend(); - } - - /*! - @brief returns a const reverse iterator to the last element - - Returns a const iterator to the reverse-beginning; that is, the last - element. - - @image html range-rbegin-rend.svg "Illustration from cppreference.com" - - @complexity Constant. - - @requirement This function helps `basic_json` satisfying the - [ReversibleContainer](http://en.cppreference.com/w/cpp/concept/ReversibleContainer) - requirements: - - The complexity is constant. - - Has the semantics of `const_cast<const basic_json&>(*this).rbegin()`. - - @liveexample{The following code shows an example for `crbegin()`.,crbegin} - - @sa @ref rbegin() -- returns a reverse iterator to the beginning - @sa @ref rend() -- returns a reverse iterator to the end - @sa @ref crend() -- returns a const reverse iterator to the end - - @since version 1.0.0 - */ - const_reverse_iterator crbegin() const noexcept - { - return const_reverse_iterator(cend()); - } - - /*! - @brief returns a const reverse iterator to one before the first - - Returns a const reverse iterator to the reverse-end; that is, one before - the first element. - - @image html range-rbegin-rend.svg "Illustration from cppreference.com" - - @complexity Constant. - - @requirement This function helps `basic_json` satisfying the - [ReversibleContainer](http://en.cppreference.com/w/cpp/concept/ReversibleContainer) - requirements: - - The complexity is constant. - - Has the semantics of `const_cast<const basic_json&>(*this).rend()`. - - @liveexample{The following code shows an example for `crend()`.,crend} - - @sa @ref rend() -- returns a reverse iterator to the end - @sa @ref rbegin() -- returns a reverse iterator to the beginning - @sa @ref crbegin() -- returns a const reverse iterator to the beginning - - @since version 1.0.0 - */ - const_reverse_iterator crend() const noexcept - { - return const_reverse_iterator(cbegin()); - } - - public: - /*! - @brief wrapper to access iterator member functions in range-based for - - This function allows to access @ref iterator::key() and @ref - iterator::value() during range-based for loops. In these loops, a - reference to the JSON values is returned, so there is no access to the - underlying iterator. - - For loop without iterator_wrapper: - - @code{cpp} - for (auto it = j_object.begin(); it != j_object.end(); ++it) - { - std::cout << "key: " << it.key() << ", value:" << it.value() << '\n'; - } - @endcode - - Range-based for loop without iterator proxy: - - @code{cpp} - for (auto it : j_object) - { - // "it" is of type json::reference and has no key() member - std::cout << "value: " << it << '\n'; - } - @endcode - - Range-based for loop with iterator proxy: - - @code{cpp} - for (auto it : json::iterator_wrapper(j_object)) - { - std::cout << "key: " << it.key() << ", value:" << it.value() << '\n'; - } - @endcode - - @note When iterating over an array, `key()` will return the index of the - element as string (see example). - - @param[in] ref reference to a JSON value - @return iteration proxy object wrapping @a ref with an interface to use in - range-based for loops - - @liveexample{The following code shows how the wrapper is used,iterator_wrapper} - - @exceptionsafety Strong guarantee: if an exception is thrown, there are no - changes in the JSON value. - - @complexity Constant. - - @note The name of this function is not yet final and may change in the - future. - - @deprecated This stream operator is deprecated and will be removed in - future 4.0.0 of the library. Please use @ref items() instead; - that is, replace `json::iterator_wrapper(j)` with `j.items()`. - */ - JSON_DEPRECATED - static iteration_proxy<iterator> iterator_wrapper(reference ref) noexcept - { - return ref.items(); - } - - /*! - @copydoc iterator_wrapper(reference) - */ - JSON_DEPRECATED - static iteration_proxy<const_iterator> iterator_wrapper(const_reference ref) noexcept - { - return ref.items(); - } - - /*! - @brief helper to access iterator member functions in range-based for - - This function allows to access @ref iterator::key() and @ref - iterator::value() during range-based for loops. In these loops, a - reference to the JSON values is returned, so there is no access to the - underlying iterator. - - For loop without `items()` function: - - @code{cpp} - for (auto it = j_object.begin(); it != j_object.end(); ++it) - { - std::cout << "key: " << it.key() << ", value:" << it.value() << '\n'; - } - @endcode - - Range-based for loop without `items()` function: - - @code{cpp} - for (auto it : j_object) - { - // "it" is of type json::reference and has no key() member - std::cout << "value: " << it << '\n'; - } - @endcode - - Range-based for loop with `items()` function: - - @code{cpp} - for (auto it : j_object.items()) - { - std::cout << "key: " << it.key() << ", value:" << it.value() << '\n'; - } - @endcode - - @note When iterating over an array, `key()` will return the index of the - element as string (see example). For primitive types (e.g., numbers), - `key()` returns an empty string. - - @return iteration proxy object wrapping @a ref with an interface to use in - range-based for loops - - @liveexample{The following code shows how the function is used.,items} - - @exceptionsafety Strong guarantee: if an exception is thrown, there are no - changes in the JSON value. - - @complexity Constant. - - @since version 3.x.x. - */ - iteration_proxy<iterator> items() noexcept - { - return iteration_proxy<iterator>(*this); - } - - /*! - @copydoc items() - */ - iteration_proxy<const_iterator> items() const noexcept - { - return iteration_proxy<const_iterator>(*this); - } - - /// @} - - - ////////////// - // capacity // - ////////////// - - /// @name capacity - /// @{ - - /*! - @brief checks whether the container is empty. - - Checks if a JSON value has no elements (i.e. whether its @ref size is `0`). - - @return The return value depends on the different types and is - defined as follows: - Value type | return value - ----------- | ------------- - null | `true` - boolean | `false` - string | `false` - number | `false` - object | result of function `object_t::empty()` - array | result of function `array_t::empty()` - - @liveexample{The following code uses `empty()` to check if a JSON - object contains any elements.,empty} - - @complexity Constant, as long as @ref array_t and @ref object_t satisfy - the Container concept; that is, their `empty()` functions have constant - complexity. - - @iterators No changes. - - @exceptionsafety No-throw guarantee: this function never throws exceptions. - - @note This function does not return whether a string stored as JSON value - is empty - it returns whether the JSON container itself is empty which is - false in the case of a string. - - @requirement This function helps `basic_json` satisfying the - [Container](http://en.cppreference.com/w/cpp/concept/Container) - requirements: - - The complexity is constant. - - Has the semantics of `begin() == end()`. - - @sa @ref size() -- returns the number of elements - - @since version 1.0.0 - */ - bool empty() const noexcept - { - switch (m_type) - { - case value_t::null: - { - // null values are empty - return true; - } - - case value_t::array: - { - // delegate call to array_t::empty() - return m_value.array->empty(); - } - - case value_t::object: - { - // delegate call to object_t::empty() - return m_value.object->empty(); - } - - default: - { - // all other types are nonempty - return false; - } - } - } - - /*! - @brief returns the number of elements - - Returns the number of elements in a JSON value. - - @return The return value depends on the different types and is - defined as follows: - Value type | return value - ----------- | ------------- - null | `0` - boolean | `1` - string | `1` - number | `1` - object | result of function object_t::size() - array | result of function array_t::size() - - @liveexample{The following code calls `size()` on the different value - types.,size} - - @complexity Constant, as long as @ref array_t and @ref object_t satisfy - the Container concept; that is, their size() functions have constant - complexity. - - @iterators No changes. - - @exceptionsafety No-throw guarantee: this function never throws exceptions. - - @note This function does not return the length of a string stored as JSON - value - it returns the number of elements in the JSON value which is 1 in - the case of a string. - - @requirement This function helps `basic_json` satisfying the - [Container](http://en.cppreference.com/w/cpp/concept/Container) - requirements: - - The complexity is constant. - - Has the semantics of `std::distance(begin(), end())`. - - @sa @ref empty() -- checks whether the container is empty - @sa @ref max_size() -- returns the maximal number of elements - - @since version 1.0.0 - */ - size_type size() const noexcept - { - switch (m_type) - { - case value_t::null: - { - // null values are empty - return 0; - } - - case value_t::array: - { - // delegate call to array_t::size() - return m_value.array->size(); - } - - case value_t::object: - { - // delegate call to object_t::size() - return m_value.object->size(); - } - - default: - { - // all other types have size 1 - return 1; - } - } - } - - /*! - @brief returns the maximum possible number of elements - - Returns the maximum number of elements a JSON value is able to hold due to - system or library implementation limitations, i.e. `std::distance(begin(), - end())` for the JSON value. - - @return The return value depends on the different types and is - defined as follows: - Value type | return value - ----------- | ------------- - null | `0` (same as `size()`) - boolean | `1` (same as `size()`) - string | `1` (same as `size()`) - number | `1` (same as `size()`) - object | result of function `object_t::max_size()` - array | result of function `array_t::max_size()` - - @liveexample{The following code calls `max_size()` on the different value - types. Note the output is implementation specific.,max_size} - - @complexity Constant, as long as @ref array_t and @ref object_t satisfy - the Container concept; that is, their `max_size()` functions have constant - complexity. - - @iterators No changes. - - @exceptionsafety No-throw guarantee: this function never throws exceptions. - - @requirement This function helps `basic_json` satisfying the - [Container](http://en.cppreference.com/w/cpp/concept/Container) - requirements: - - The complexity is constant. - - Has the semantics of returning `b.size()` where `b` is the largest - possible JSON value. - - @sa @ref size() -- returns the number of elements - - @since version 1.0.0 - */ - size_type max_size() const noexcept - { - switch (m_type) - { - case value_t::array: - { - // delegate call to array_t::max_size() - return m_value.array->max_size(); - } - - case value_t::object: - { - // delegate call to object_t::max_size() - return m_value.object->max_size(); - } - - default: - { - // all other types have max_size() == size() - return size(); - } - } - } - - /// @} - - - /////////////// - // modifiers // - /////////////// - - /// @name modifiers - /// @{ - - /*! - @brief clears the contents - - Clears the content of a JSON value and resets it to the default value as - if @ref basic_json(value_t) would have been called with the current value - type from @ref type(): - - Value type | initial value - ----------- | ------------- - null | `null` - boolean | `false` - string | `""` - number | `0` - object | `{}` - array | `[]` - - @post Has the same effect as calling - @code {.cpp} - *this = basic_json(type()); - @endcode - - @liveexample{The example below shows the effect of `clear()` to different - JSON types.,clear} - - @complexity Linear in the size of the JSON value. - - @iterators All iterators, pointers and references related to this container - are invalidated. - - @exceptionsafety No-throw guarantee: this function never throws exceptions. - - @sa @ref basic_json(value_t) -- constructor that creates an object with the - same value than calling `clear()` - - @since version 1.0.0 - */ - void clear() noexcept - { - switch (m_type) - { - case value_t::number_integer: - { - m_value.number_integer = 0; - break; - } - - case value_t::number_unsigned: - { - m_value.number_unsigned = 0; - break; - } - - case value_t::number_float: - { - m_value.number_float = 0.0; - break; - } - - case value_t::boolean: - { - m_value.boolean = false; - break; - } - - case value_t::string: - { - m_value.string->clear(); - break; - } - - case value_t::array: - { - m_value.array->clear(); - break; - } - - case value_t::object: - { - m_value.object->clear(); - break; - } - - default: - break; - } - } - - /*! - @brief add an object to an array - - Appends the given element @a val to the end of the JSON value. If the - function is called on a JSON null value, an empty array is created before - appending @a val. - - @param[in] val the value to add to the JSON array - - @throw type_error.308 when called on a type other than JSON array or - null; example: `"cannot use push_back() with number"` - - @complexity Amortized constant. - - @liveexample{The example shows how `push_back()` and `+=` can be used to - add elements to a JSON array. Note how the `null` value was silently - converted to a JSON array.,push_back} - - @since version 1.0.0 - */ - void push_back(basic_json&& val) - { - // push_back only works for null objects or arrays - if (JSON_UNLIKELY(not(is_null() or is_array()))) - { - JSON_THROW(type_error::create(308, "cannot use push_back() with " + std::string(type_name()))); - } - - // transform null object into an array - if (is_null()) - { - m_type = value_t::array; - m_value = value_t::array; - assert_invariant(); - } - - // add element to array (move semantics) - m_value.array->push_back(std::move(val)); - // invalidate object - val.m_type = value_t::null; - } - - /*! - @brief add an object to an array - @copydoc push_back(basic_json&&) - */ - reference operator+=(basic_json&& val) - { - push_back(std::move(val)); - return *this; - } - - /*! - @brief add an object to an array - @copydoc push_back(basic_json&&) - */ - void push_back(const basic_json& val) - { - // push_back only works for null objects or arrays - if (JSON_UNLIKELY(not(is_null() or is_array()))) - { - JSON_THROW(type_error::create(308, "cannot use push_back() with " + std::string(type_name()))); - } - - // transform null object into an array - if (is_null()) - { - m_type = value_t::array; - m_value = value_t::array; - assert_invariant(); - } - - // add element to array - m_value.array->push_back(val); - } - - /*! - @brief add an object to an array - @copydoc push_back(basic_json&&) - */ - reference operator+=(const basic_json& val) - { - push_back(val); - return *this; - } - - /*! - @brief add an object to an object - - Inserts the given element @a val to the JSON object. If the function is - called on a JSON null value, an empty object is created before inserting - @a val. - - @param[in] val the value to add to the JSON object - - @throw type_error.308 when called on a type other than JSON object or - null; example: `"cannot use push_back() with number"` - - @complexity Logarithmic in the size of the container, O(log(`size()`)). - - @liveexample{The example shows how `push_back()` and `+=` can be used to - add elements to a JSON object. Note how the `null` value was silently - converted to a JSON object.,push_back__object_t__value} - - @since version 1.0.0 - */ - void push_back(const typename object_t::value_type& val) - { - // push_back only works for null objects or objects - if (JSON_UNLIKELY(not(is_null() or is_object()))) - { - JSON_THROW(type_error::create(308, "cannot use push_back() with " + std::string(type_name()))); - } - - // transform null object into an object - if (is_null()) - { - m_type = value_t::object; - m_value = value_t::object; - assert_invariant(); - } - - // add element to array - m_value.object->insert(val); - } - - /*! - @brief add an object to an object - @copydoc push_back(const typename object_t::value_type&) - */ - reference operator+=(const typename object_t::value_type& val) - { - push_back(val); - return *this; - } - - /*! - @brief add an object to an object - - This function allows to use `push_back` with an initializer list. In case - - 1. the current value is an object, - 2. the initializer list @a init contains only two elements, and - 3. the first element of @a init is a string, - - @a init is converted into an object element and added using - @ref push_back(const typename object_t::value_type&). Otherwise, @a init - is converted to a JSON value and added using @ref push_back(basic_json&&). - - @param[in] init an initializer list - - @complexity Linear in the size of the initializer list @a init. - - @note This function is required to resolve an ambiguous overload error, - because pairs like `{"key", "value"}` can be both interpreted as - `object_t::value_type` or `std::initializer_list<basic_json>`, see - https://github.com/nlohmann/json/issues/235 for more information. - - @liveexample{The example shows how initializer lists are treated as - objects when possible.,push_back__initializer_list} - */ - void push_back(initializer_list_t init) - { - if (is_object() and init.size() == 2 and (*init.begin())->is_string()) - { - basic_json&& key = init.begin()->moved_or_copied(); - push_back(typename object_t::value_type( - std::move(key.get_ref<string_t&>()), (init.begin() + 1)->moved_or_copied())); - } - else - { - push_back(basic_json(init)); - } - } - - /*! - @brief add an object to an object - @copydoc push_back(initializer_list_t) - */ - reference operator+=(initializer_list_t init) - { - push_back(init); - return *this; - } - - /*! - @brief add an object to an array - - Creates a JSON value from the passed parameters @a args to the end of the - JSON value. If the function is called on a JSON null value, an empty array - is created before appending the value created from @a args. - - @param[in] args arguments to forward to a constructor of @ref basic_json - @tparam Args compatible types to create a @ref basic_json object - - @throw type_error.311 when called on a type other than JSON array or - null; example: `"cannot use emplace_back() with number"` - - @complexity Amortized constant. - - @liveexample{The example shows how `push_back()` can be used to add - elements to a JSON array. Note how the `null` value was silently converted - to a JSON array.,emplace_back} - - @since version 2.0.8 - */ - template<class... Args> - void emplace_back(Args&& ... args) - { - // emplace_back only works for null objects or arrays - if (JSON_UNLIKELY(not(is_null() or is_array()))) - { - JSON_THROW(type_error::create(311, "cannot use emplace_back() with " + std::string(type_name()))); - } - - // transform null object into an array - if (is_null()) - { - m_type = value_t::array; - m_value = value_t::array; - assert_invariant(); - } - - // add element to array (perfect forwarding) - m_value.array->emplace_back(std::forward<Args>(args)...); - } - - /*! - @brief add an object to an object if key does not exist - - Inserts a new element into a JSON object constructed in-place with the - given @a args if there is no element with the key in the container. If the - function is called on a JSON null value, an empty object is created before - appending the value created from @a args. - - @param[in] args arguments to forward to a constructor of @ref basic_json - @tparam Args compatible types to create a @ref basic_json object - - @return a pair consisting of an iterator to the inserted element, or the - already-existing element if no insertion happened, and a bool - denoting whether the insertion took place. - - @throw type_error.311 when called on a type other than JSON object or - null; example: `"cannot use emplace() with number"` - - @complexity Logarithmic in the size of the container, O(log(`size()`)). - - @liveexample{The example shows how `emplace()` can be used to add elements - to a JSON object. Note how the `null` value was silently converted to a - JSON object. Further note how no value is added if there was already one - value stored with the same key.,emplace} - - @since version 2.0.8 - */ - template<class... Args> - std::pair<iterator, bool> emplace(Args&& ... args) - { - // emplace only works for null objects or arrays - if (JSON_UNLIKELY(not(is_null() or is_object()))) - { - JSON_THROW(type_error::create(311, "cannot use emplace() with " + std::string(type_name()))); - } - - // transform null object into an object - if (is_null()) - { - m_type = value_t::object; - m_value = value_t::object; - assert_invariant(); - } - - // add element to array (perfect forwarding) - auto res = m_value.object->emplace(std::forward<Args>(args)...); - // create result iterator and set iterator to the result of emplace - auto it = begin(); - it.m_it.object_iterator = res.first; - - // return pair of iterator and boolean - return {it, res.second}; - } - - /*! - @brief inserts element - - Inserts element @a val before iterator @a pos. - - @param[in] pos iterator before which the content will be inserted; may be - the end() iterator - @param[in] val element to insert - @return iterator pointing to the inserted @a val. - - @throw type_error.309 if called on JSON values other than arrays; - example: `"cannot use insert() with string"` - @throw invalid_iterator.202 if @a pos is not an iterator of *this; - example: `"iterator does not fit current value"` - - @complexity Constant plus linear in the distance between @a pos and end of - the container. - - @liveexample{The example shows how `insert()` is used.,insert} - - @since version 1.0.0 - */ - iterator insert(const_iterator pos, const basic_json& val) - { - // insert only works for arrays - if (JSON_LIKELY(is_array())) - { - // check if iterator pos fits to this JSON value - if (JSON_UNLIKELY(pos.m_object != this)) - { - JSON_THROW(invalid_iterator::create(202, "iterator does not fit current value")); - } - - // insert to array and return iterator - iterator result(this); - result.m_it.array_iterator = m_value.array->insert(pos.m_it.array_iterator, val); - return result; - } - - JSON_THROW(type_error::create(309, "cannot use insert() with " + std::string(type_name()))); - } - - /*! - @brief inserts element - @copydoc insert(const_iterator, const basic_json&) - */ - iterator insert(const_iterator pos, basic_json&& val) - { - return insert(pos, val); - } - - /*! - @brief inserts elements - - Inserts @a cnt copies of @a val before iterator @a pos. - - @param[in] pos iterator before which the content will be inserted; may be - the end() iterator - @param[in] cnt number of copies of @a val to insert - @param[in] val element to insert - @return iterator pointing to the first element inserted, or @a pos if - `cnt==0` - - @throw type_error.309 if called on JSON values other than arrays; example: - `"cannot use insert() with string"` - @throw invalid_iterator.202 if @a pos is not an iterator of *this; - example: `"iterator does not fit current value"` - - @complexity Linear in @a cnt plus linear in the distance between @a pos - and end of the container. - - @liveexample{The example shows how `insert()` is used.,insert__count} - - @since version 1.0.0 - */ - iterator insert(const_iterator pos, size_type cnt, const basic_json& val) - { - // insert only works for arrays - if (JSON_LIKELY(is_array())) - { - // check if iterator pos fits to this JSON value - if (JSON_UNLIKELY(pos.m_object != this)) - { - JSON_THROW(invalid_iterator::create(202, "iterator does not fit current value")); - } - - // insert to array and return iterator - iterator result(this); - result.m_it.array_iterator = m_value.array->insert(pos.m_it.array_iterator, cnt, val); - return result; - } - - JSON_THROW(type_error::create(309, "cannot use insert() with " + std::string(type_name()))); - } - - /*! - @brief inserts elements - - Inserts elements from range `[first, last)` before iterator @a pos. - - @param[in] pos iterator before which the content will be inserted; may be - the end() iterator - @param[in] first begin of the range of elements to insert - @param[in] last end of the range of elements to insert - - @throw type_error.309 if called on JSON values other than arrays; example: - `"cannot use insert() with string"` - @throw invalid_iterator.202 if @a pos is not an iterator of *this; - example: `"iterator does not fit current value"` - @throw invalid_iterator.210 if @a first and @a last do not belong to the - same JSON value; example: `"iterators do not fit"` - @throw invalid_iterator.211 if @a first or @a last are iterators into - container for which insert is called; example: `"passed iterators may not - belong to container"` - - @return iterator pointing to the first element inserted, or @a pos if - `first==last` - - @complexity Linear in `std::distance(first, last)` plus linear in the - distance between @a pos and end of the container. - - @liveexample{The example shows how `insert()` is used.,insert__range} - - @since version 1.0.0 - */ - iterator insert(const_iterator pos, const_iterator first, const_iterator last) - { - // insert only works for arrays - if (JSON_UNLIKELY(not is_array())) - { - JSON_THROW(type_error::create(309, "cannot use insert() with " + std::string(type_name()))); - } - - // check if iterator pos fits to this JSON value - if (JSON_UNLIKELY(pos.m_object != this)) - { - JSON_THROW(invalid_iterator::create(202, "iterator does not fit current value")); - } - - // check if range iterators belong to the same JSON object - if (JSON_UNLIKELY(first.m_object != last.m_object)) - { - JSON_THROW(invalid_iterator::create(210, "iterators do not fit")); - } - - if (JSON_UNLIKELY(first.m_object == this)) - { - JSON_THROW(invalid_iterator::create(211, "passed iterators may not belong to container")); - } - - // insert to array and return iterator - iterator result(this); - result.m_it.array_iterator = m_value.array->insert( - pos.m_it.array_iterator, - first.m_it.array_iterator, - last.m_it.array_iterator); - return result; - } - - /*! - @brief inserts elements - - Inserts elements from initializer list @a ilist before iterator @a pos. - - @param[in] pos iterator before which the content will be inserted; may be - the end() iterator - @param[in] ilist initializer list to insert the values from - - @throw type_error.309 if called on JSON values other than arrays; example: - `"cannot use insert() with string"` - @throw invalid_iterator.202 if @a pos is not an iterator of *this; - example: `"iterator does not fit current value"` - - @return iterator pointing to the first element inserted, or @a pos if - `ilist` is empty - - @complexity Linear in `ilist.size()` plus linear in the distance between - @a pos and end of the container. - - @liveexample{The example shows how `insert()` is used.,insert__ilist} - - @since version 1.0.0 - */ - iterator insert(const_iterator pos, initializer_list_t ilist) - { - // insert only works for arrays - if (JSON_UNLIKELY(not is_array())) - { - JSON_THROW(type_error::create(309, "cannot use insert() with " + std::string(type_name()))); - } - - // check if iterator pos fits to this JSON value - if (JSON_UNLIKELY(pos.m_object != this)) - { - JSON_THROW(invalid_iterator::create(202, "iterator does not fit current value")); - } - - // insert to array and return iterator - iterator result(this); - result.m_it.array_iterator = m_value.array->insert(pos.m_it.array_iterator, ilist.begin(), ilist.end()); - return result; - } - - /*! - @brief inserts elements - - Inserts elements from range `[first, last)`. - - @param[in] first begin of the range of elements to insert - @param[in] last end of the range of elements to insert - - @throw type_error.309 if called on JSON values other than objects; example: - `"cannot use insert() with string"` - @throw invalid_iterator.202 if iterator @a first or @a last does does not - point to an object; example: `"iterators first and last must point to - objects"` - @throw invalid_iterator.210 if @a first and @a last do not belong to the - same JSON value; example: `"iterators do not fit"` - - @complexity Logarithmic: `O(N*log(size() + N))`, where `N` is the number - of elements to insert. - - @liveexample{The example shows how `insert()` is used.,insert__range_object} - - @since version 3.0.0 - */ - void insert(const_iterator first, const_iterator last) - { - // insert only works for objects - if (JSON_UNLIKELY(not is_object())) - { - JSON_THROW(type_error::create(309, "cannot use insert() with " + std::string(type_name()))); - } - - // check if range iterators belong to the same JSON object - if (JSON_UNLIKELY(first.m_object != last.m_object)) - { - JSON_THROW(invalid_iterator::create(210, "iterators do not fit")); - } - - // passed iterators must belong to objects - if (JSON_UNLIKELY(not first.m_object->is_object())) - { - JSON_THROW(invalid_iterator::create(202, "iterators first and last must point to objects")); - } - - m_value.object->insert(first.m_it.object_iterator, last.m_it.object_iterator); - } - - /*! - @brief updates a JSON object from another object, overwriting existing keys - - Inserts all values from JSON object @a j and overwrites existing keys. - - @param[in] j JSON object to read values from - - @throw type_error.312 if called on JSON values other than objects; example: - `"cannot use update() with string"` - - @complexity O(N*log(size() + N)), where N is the number of elements to - insert. - - @liveexample{The example shows how `update()` is used.,update} - - @sa https://docs.python.org/3.6/library/stdtypes.html#dict.update - - @since version 3.0.0 - */ - void update(const_reference j) - { - // implicitly convert null value to an empty object - if (is_null()) - { - m_type = value_t::object; - m_value.object = create<object_t>(); - assert_invariant(); - } - - if (JSON_UNLIKELY(not is_object())) - { - JSON_THROW(type_error::create(312, "cannot use update() with " + std::string(type_name()))); - } - if (JSON_UNLIKELY(not j.is_object())) - { - JSON_THROW(type_error::create(312, "cannot use update() with " + std::string(j.type_name()))); - } - - for (auto it = j.cbegin(); it != j.cend(); ++it) - { - m_value.object->operator[](it.key()) = it.value(); - } - } - - /*! - @brief updates a JSON object from another object, overwriting existing keys - - Inserts all values from from range `[first, last)` and overwrites existing - keys. - - @param[in] first begin of the range of elements to insert - @param[in] last end of the range of elements to insert - - @throw type_error.312 if called on JSON values other than objects; example: - `"cannot use update() with string"` - @throw invalid_iterator.202 if iterator @a first or @a last does does not - point to an object; example: `"iterators first and last must point to - objects"` - @throw invalid_iterator.210 if @a first and @a last do not belong to the - same JSON value; example: `"iterators do not fit"` - - @complexity O(N*log(size() + N)), where N is the number of elements to - insert. - - @liveexample{The example shows how `update()` is used__range.,update} - - @sa https://docs.python.org/3.6/library/stdtypes.html#dict.update - - @since version 3.0.0 - */ - void update(const_iterator first, const_iterator last) - { - // implicitly convert null value to an empty object - if (is_null()) - { - m_type = value_t::object; - m_value.object = create<object_t>(); - assert_invariant(); - } - - if (JSON_UNLIKELY(not is_object())) - { - JSON_THROW(type_error::create(312, "cannot use update() with " + std::string(type_name()))); - } - - // check if range iterators belong to the same JSON object - if (JSON_UNLIKELY(first.m_object != last.m_object)) - { - JSON_THROW(invalid_iterator::create(210, "iterators do not fit")); - } - - // passed iterators must belong to objects - if (JSON_UNLIKELY(not first.m_object->is_object() - or not last.m_object->is_object())) - { - JSON_THROW(invalid_iterator::create(202, "iterators first and last must point to objects")); - } - - for (auto it = first; it != last; ++it) - { - m_value.object->operator[](it.key()) = it.value(); - } - } - - /*! - @brief exchanges the values - - Exchanges the contents of the JSON value with those of @a other. Does not - invoke any move, copy, or swap operations on individual elements. All - iterators and references remain valid. The past-the-end iterator is - invalidated. - - @param[in,out] other JSON value to exchange the contents with - - @complexity Constant. - - @liveexample{The example below shows how JSON values can be swapped with - `swap()`.,swap__reference} - - @since version 1.0.0 - */ - void swap(reference other) noexcept ( - std::is_nothrow_move_constructible<value_t>::value and - std::is_nothrow_move_assignable<value_t>::value and - std::is_nothrow_move_constructible<json_value>::value and - std::is_nothrow_move_assignable<json_value>::value - ) - { - std::swap(m_type, other.m_type); - std::swap(m_value, other.m_value); - assert_invariant(); - } - - /*! - @brief exchanges the values - - Exchanges the contents of a JSON array with those of @a other. Does not - invoke any move, copy, or swap operations on individual elements. All - iterators and references remain valid. The past-the-end iterator is - invalidated. - - @param[in,out] other array to exchange the contents with - - @throw type_error.310 when JSON value is not an array; example: `"cannot - use swap() with string"` - - @complexity Constant. - - @liveexample{The example below shows how arrays can be swapped with - `swap()`.,swap__array_t} - - @since version 1.0.0 - */ - void swap(array_t& other) - { - // swap only works for arrays - if (JSON_LIKELY(is_array())) - { - std::swap(*(m_value.array), other); - } - else - { - JSON_THROW(type_error::create(310, "cannot use swap() with " + std::string(type_name()))); - } - } - - /*! - @brief exchanges the values - - Exchanges the contents of a JSON object with those of @a other. Does not - invoke any move, copy, or swap operations on individual elements. All - iterators and references remain valid. The past-the-end iterator is - invalidated. - - @param[in,out] other object to exchange the contents with - - @throw type_error.310 when JSON value is not an object; example: - `"cannot use swap() with string"` - - @complexity Constant. - - @liveexample{The example below shows how objects can be swapped with - `swap()`.,swap__object_t} - - @since version 1.0.0 - */ - void swap(object_t& other) - { - // swap only works for objects - if (JSON_LIKELY(is_object())) - { - std::swap(*(m_value.object), other); - } - else - { - JSON_THROW(type_error::create(310, "cannot use swap() with " + std::string(type_name()))); - } - } - - /*! - @brief exchanges the values - - Exchanges the contents of a JSON string with those of @a other. Does not - invoke any move, copy, or swap operations on individual elements. All - iterators and references remain valid. The past-the-end iterator is - invalidated. - - @param[in,out] other string to exchange the contents with - - @throw type_error.310 when JSON value is not a string; example: `"cannot - use swap() with boolean"` - - @complexity Constant. - - @liveexample{The example below shows how strings can be swapped with - `swap()`.,swap__string_t} - - @since version 1.0.0 - */ - void swap(string_t& other) - { - // swap only works for strings - if (JSON_LIKELY(is_string())) - { - std::swap(*(m_value.string), other); - } - else - { - JSON_THROW(type_error::create(310, "cannot use swap() with " + std::string(type_name()))); - } - } - - /// @} - - public: - ////////////////////////////////////////// - // lexicographical comparison operators // - ////////////////////////////////////////// - - /// @name lexicographical comparison operators - /// @{ - - /*! - @brief comparison: equal - - Compares two JSON values for equality according to the following rules: - - Two JSON values are equal if (1) they are from the same type and (2) - their stored values are the same according to their respective - `operator==`. - - Integer and floating-point numbers are automatically converted before - comparison. Note than two NaN values are always treated as unequal. - - Two JSON null values are equal. - - @note Floating-point inside JSON values numbers are compared with - `json::number_float_t::operator==` which is `double::operator==` by - default. To compare floating-point while respecting an epsilon, an alternative - [comparison function](https://github.com/mariokonrad/marnav/blob/master/src/marnav/math/floatingpoint.hpp#L34-#L39) - could be used, for instance - @code {.cpp} - template<typename T, typename = typename std::enable_if<std::is_floating_point<T>::value, T>::type> - inline bool is_same(T a, T b, T epsilon = std::numeric_limits<T>::epsilon()) noexcept - { - return std::abs(a - b) <= epsilon; - } - @endcode - - @note NaN values never compare equal to themselves or to other NaN values. - - @param[in] lhs first JSON value to consider - @param[in] rhs second JSON value to consider - @return whether the values @a lhs and @a rhs are equal - - @exceptionsafety No-throw guarantee: this function never throws exceptions. - - @complexity Linear. - - @liveexample{The example demonstrates comparing several JSON - types.,operator__equal} - - @since version 1.0.0 - */ - friend bool operator==(const_reference lhs, const_reference rhs) noexcept - { - const auto lhs_type = lhs.type(); - const auto rhs_type = rhs.type(); - - if (lhs_type == rhs_type) - { - switch (lhs_type) - { - case value_t::array: - return (*lhs.m_value.array == *rhs.m_value.array); - - case value_t::object: - return (*lhs.m_value.object == *rhs.m_value.object); - - case value_t::null: - return true; - - case value_t::string: - return (*lhs.m_value.string == *rhs.m_value.string); - - case value_t::boolean: - return (lhs.m_value.boolean == rhs.m_value.boolean); - - case value_t::number_integer: - return (lhs.m_value.number_integer == rhs.m_value.number_integer); - - case value_t::number_unsigned: - return (lhs.m_value.number_unsigned == rhs.m_value.number_unsigned); - - case value_t::number_float: - return (lhs.m_value.number_float == rhs.m_value.number_float); - - default: - return false; - } - } - else if (lhs_type == value_t::number_integer and rhs_type == value_t::number_float) - { - return (static_cast<number_float_t>(lhs.m_value.number_integer) == rhs.m_value.number_float); - } - else if (lhs_type == value_t::number_float and rhs_type == value_t::number_integer) - { - return (lhs.m_value.number_float == static_cast<number_float_t>(rhs.m_value.number_integer)); - } - else if (lhs_type == value_t::number_unsigned and rhs_type == value_t::number_float) - { - return (static_cast<number_float_t>(lhs.m_value.number_unsigned) == rhs.m_value.number_float); - } - else if (lhs_type == value_t::number_float and rhs_type == value_t::number_unsigned) - { - return (lhs.m_value.number_float == static_cast<number_float_t>(rhs.m_value.number_unsigned)); - } - else if (lhs_type == value_t::number_unsigned and rhs_type == value_t::number_integer) - { - return (static_cast<number_integer_t>(lhs.m_value.number_unsigned) == rhs.m_value.number_integer); - } - else if (lhs_type == value_t::number_integer and rhs_type == value_t::number_unsigned) - { - return (lhs.m_value.number_integer == static_cast<number_integer_t>(rhs.m_value.number_unsigned)); - } - - return false; - } - - /*! - @brief comparison: equal - @copydoc operator==(const_reference, const_reference) - */ - template<typename ScalarType, typename std::enable_if< - std::is_scalar<ScalarType>::value, int>::type = 0> - friend bool operator==(const_reference lhs, const ScalarType rhs) noexcept - { - return (lhs == basic_json(rhs)); - } - - /*! - @brief comparison: equal - @copydoc operator==(const_reference, const_reference) - */ - template<typename ScalarType, typename std::enable_if< - std::is_scalar<ScalarType>::value, int>::type = 0> - friend bool operator==(const ScalarType lhs, const_reference rhs) noexcept - { - return (basic_json(lhs) == rhs); - } - - /*! - @brief comparison: not equal - - Compares two JSON values for inequality by calculating `not (lhs == rhs)`. - - @param[in] lhs first JSON value to consider - @param[in] rhs second JSON value to consider - @return whether the values @a lhs and @a rhs are not equal - - @complexity Linear. - - @exceptionsafety No-throw guarantee: this function never throws exceptions. - - @liveexample{The example demonstrates comparing several JSON - types.,operator__notequal} - - @since version 1.0.0 - */ - friend bool operator!=(const_reference lhs, const_reference rhs) noexcept - { - return not (lhs == rhs); - } - - /*! - @brief comparison: not equal - @copydoc operator!=(const_reference, const_reference) - */ - template<typename ScalarType, typename std::enable_if< - std::is_scalar<ScalarType>::value, int>::type = 0> - friend bool operator!=(const_reference lhs, const ScalarType rhs) noexcept - { - return (lhs != basic_json(rhs)); - } - - /*! - @brief comparison: not equal - @copydoc operator!=(const_reference, const_reference) - */ - template<typename ScalarType, typename std::enable_if< - std::is_scalar<ScalarType>::value, int>::type = 0> - friend bool operator!=(const ScalarType lhs, const_reference rhs) noexcept - { - return (basic_json(lhs) != rhs); - } - - /*! - @brief comparison: less than - - Compares whether one JSON value @a lhs is less than another JSON value @a - rhs according to the following rules: - - If @a lhs and @a rhs have the same type, the values are compared using - the default `<` operator. - - Integer and floating-point numbers are automatically converted before - comparison - - In case @a lhs and @a rhs have different types, the values are ignored - and the order of the types is considered, see - @ref operator<(const value_t, const value_t). - - @param[in] lhs first JSON value to consider - @param[in] rhs second JSON value to consider - @return whether @a lhs is less than @a rhs - - @complexity Linear. - - @exceptionsafety No-throw guarantee: this function never throws exceptions. - - @liveexample{The example demonstrates comparing several JSON - types.,operator__less} - - @since version 1.0.0 - */ - friend bool operator<(const_reference lhs, const_reference rhs) noexcept - { - const auto lhs_type = lhs.type(); - const auto rhs_type = rhs.type(); - - if (lhs_type == rhs_type) - { - switch (lhs_type) - { - case value_t::array: - return (*lhs.m_value.array) < (*rhs.m_value.array); - - case value_t::object: - return *lhs.m_value.object < *rhs.m_value.object; - - case value_t::null: - return false; - - case value_t::string: - return *lhs.m_value.string < *rhs.m_value.string; - - case value_t::boolean: - return lhs.m_value.boolean < rhs.m_value.boolean; - - case value_t::number_integer: - return lhs.m_value.number_integer < rhs.m_value.number_integer; - - case value_t::number_unsigned: - return lhs.m_value.number_unsigned < rhs.m_value.number_unsigned; - - case value_t::number_float: - return lhs.m_value.number_float < rhs.m_value.number_float; - - default: - return false; - } - } - else if (lhs_type == value_t::number_integer and rhs_type == value_t::number_float) - { - return static_cast<number_float_t>(lhs.m_value.number_integer) < rhs.m_value.number_float; - } - else if (lhs_type == value_t::number_float and rhs_type == value_t::number_integer) - { - return lhs.m_value.number_float < static_cast<number_float_t>(rhs.m_value.number_integer); - } - else if (lhs_type == value_t::number_unsigned and rhs_type == value_t::number_float) - { - return static_cast<number_float_t>(lhs.m_value.number_unsigned) < rhs.m_value.number_float; - } - else if (lhs_type == value_t::number_float and rhs_type == value_t::number_unsigned) - { - return lhs.m_value.number_float < static_cast<number_float_t>(rhs.m_value.number_unsigned); - } - else if (lhs_type == value_t::number_integer and rhs_type == value_t::number_unsigned) - { - return lhs.m_value.number_integer < static_cast<number_integer_t>(rhs.m_value.number_unsigned); - } - else if (lhs_type == value_t::number_unsigned and rhs_type == value_t::number_integer) - { - return static_cast<number_integer_t>(lhs.m_value.number_unsigned) < rhs.m_value.number_integer; - } - - // We only reach this line if we cannot compare values. In that case, - // we compare types. Note we have to call the operator explicitly, - // because MSVC has problems otherwise. - return operator<(lhs_type, rhs_type); - } - - /*! - @brief comparison: less than - @copydoc operator<(const_reference, const_reference) - */ - template<typename ScalarType, typename std::enable_if< - std::is_scalar<ScalarType>::value, int>::type = 0> - friend bool operator<(const_reference lhs, const ScalarType rhs) noexcept - { - return (lhs < basic_json(rhs)); - } - - /*! - @brief comparison: less than - @copydoc operator<(const_reference, const_reference) - */ - template<typename ScalarType, typename std::enable_if< - std::is_scalar<ScalarType>::value, int>::type = 0> - friend bool operator<(const ScalarType lhs, const_reference rhs) noexcept - { - return (basic_json(lhs) < rhs); - } - - /*! - @brief comparison: less than or equal - - Compares whether one JSON value @a lhs is less than or equal to another - JSON value by calculating `not (rhs < lhs)`. - - @param[in] lhs first JSON value to consider - @param[in] rhs second JSON value to consider - @return whether @a lhs is less than or equal to @a rhs - - @complexity Linear. - - @exceptionsafety No-throw guarantee: this function never throws exceptions. - - @liveexample{The example demonstrates comparing several JSON - types.,operator__greater} - - @since version 1.0.0 - */ - friend bool operator<=(const_reference lhs, const_reference rhs) noexcept - { - return not (rhs < lhs); - } - - /*! - @brief comparison: less than or equal - @copydoc operator<=(const_reference, const_reference) - */ - template<typename ScalarType, typename std::enable_if< - std::is_scalar<ScalarType>::value, int>::type = 0> - friend bool operator<=(const_reference lhs, const ScalarType rhs) noexcept - { - return (lhs <= basic_json(rhs)); - } - - /*! - @brief comparison: less than or equal - @copydoc operator<=(const_reference, const_reference) - */ - template<typename ScalarType, typename std::enable_if< - std::is_scalar<ScalarType>::value, int>::type = 0> - friend bool operator<=(const ScalarType lhs, const_reference rhs) noexcept - { - return (basic_json(lhs) <= rhs); - } - - /*! - @brief comparison: greater than - - Compares whether one JSON value @a lhs is greater than another - JSON value by calculating `not (lhs <= rhs)`. - - @param[in] lhs first JSON value to consider - @param[in] rhs second JSON value to consider - @return whether @a lhs is greater than to @a rhs - - @complexity Linear. - - @exceptionsafety No-throw guarantee: this function never throws exceptions. - - @liveexample{The example demonstrates comparing several JSON - types.,operator__lessequal} - - @since version 1.0.0 - */ - friend bool operator>(const_reference lhs, const_reference rhs) noexcept - { - return not (lhs <= rhs); - } - - /*! - @brief comparison: greater than - @copydoc operator>(const_reference, const_reference) - */ - template<typename ScalarType, typename std::enable_if< - std::is_scalar<ScalarType>::value, int>::type = 0> - friend bool operator>(const_reference lhs, const ScalarType rhs) noexcept - { - return (lhs > basic_json(rhs)); - } - - /*! - @brief comparison: greater than - @copydoc operator>(const_reference, const_reference) - */ - template<typename ScalarType, typename std::enable_if< - std::is_scalar<ScalarType>::value, int>::type = 0> - friend bool operator>(const ScalarType lhs, const_reference rhs) noexcept - { - return (basic_json(lhs) > rhs); - } - - /*! - @brief comparison: greater than or equal - - Compares whether one JSON value @a lhs is greater than or equal to another - JSON value by calculating `not (lhs < rhs)`. - - @param[in] lhs first JSON value to consider - @param[in] rhs second JSON value to consider - @return whether @a lhs is greater than or equal to @a rhs - - @complexity Linear. - - @exceptionsafety No-throw guarantee: this function never throws exceptions. - - @liveexample{The example demonstrates comparing several JSON - types.,operator__greaterequal} - - @since version 1.0.0 - */ - friend bool operator>=(const_reference lhs, const_reference rhs) noexcept - { - return not (lhs < rhs); - } - - /*! - @brief comparison: greater than or equal - @copydoc operator>=(const_reference, const_reference) - */ - template<typename ScalarType, typename std::enable_if< - std::is_scalar<ScalarType>::value, int>::type = 0> - friend bool operator>=(const_reference lhs, const ScalarType rhs) noexcept - { - return (lhs >= basic_json(rhs)); - } - - /*! - @brief comparison: greater than or equal - @copydoc operator>=(const_reference, const_reference) - */ - template<typename ScalarType, typename std::enable_if< - std::is_scalar<ScalarType>::value, int>::type = 0> - friend bool operator>=(const ScalarType lhs, const_reference rhs) noexcept - { - return (basic_json(lhs) >= rhs); - } - - /// @} - - /////////////////// - // serialization // - /////////////////// - - /// @name serialization - /// @{ - - /*! - @brief serialize to stream - - Serialize the given JSON value @a j to the output stream @a o. The JSON - value will be serialized using the @ref dump member function. - - - The indentation of the output can be controlled with the member variable - `width` of the output stream @a o. For instance, using the manipulator - `std::setw(4)` on @a o sets the indentation level to `4` and the - serialization result is the same as calling `dump(4)`. - - - The indentation character can be controlled with the member variable - `fill` of the output stream @a o. For instance, the manipulator - `std::setfill('\\t')` sets indentation to use a tab character rather than - the default space character. - - @param[in,out] o stream to serialize to - @param[in] j JSON value to serialize - - @return the stream @a o - - @throw type_error.316 if a string stored inside the JSON value is not - UTF-8 encoded - - @complexity Linear. - - @liveexample{The example below shows the serialization with different - parameters to `width` to adjust the indentation level.,operator_serialize} - - @since version 1.0.0; indentation character added in version 3.0.0 - */ - friend std::ostream& operator<<(std::ostream& o, const basic_json& j) - { - // read width member and use it as indentation parameter if nonzero - const bool pretty_print = (o.width() > 0); - const auto indentation = (pretty_print ? o.width() : 0); - - // reset width to 0 for subsequent calls to this stream - o.width(0); - - // do the actual serialization - serializer s(detail::output_adapter<char>(o), o.fill()); - s.dump(j, pretty_print, false, static_cast<unsigned int>(indentation)); - return o; - } - - /*! - @brief serialize to stream - @deprecated This stream operator is deprecated and will be removed in - future 4.0.0 of the library. Please use - @ref operator<<(std::ostream&, const basic_json&) - instead; that is, replace calls like `j >> o;` with `o << j;`. - @since version 1.0.0; deprecated since version 3.0.0 - */ - JSON_DEPRECATED - friend std::ostream& operator>>(const basic_json& j, std::ostream& o) - { - return o << j; - } - - /// @} - - - ///////////////////// - // deserialization // - ///////////////////// - - /// @name deserialization - /// @{ - - /*! - @brief deserialize from a compatible input - - This function reads from a compatible input. Examples are: - - an array of 1-byte values - - strings with character/literal type with size of 1 byte - - input streams - - container with contiguous storage of 1-byte values. Compatible container - types include `std::vector`, `std::string`, `std::array`, - `std::valarray`, and `std::initializer_list`. Furthermore, C-style - arrays can be used with `std::begin()`/`std::end()`. User-defined - containers can be used as long as they implement random-access iterators - and a contiguous storage. - - @pre Each element of the container has a size of 1 byte. Violating this - precondition yields undefined behavior. **This precondition is enforced - with a static assertion.** - - @pre The container storage is contiguous. Violating this precondition - yields undefined behavior. **This precondition is enforced with an - assertion.** - @pre Each element of the container has a size of 1 byte. Violating this - precondition yields undefined behavior. **This precondition is enforced - with a static assertion.** - - @warning There is no way to enforce all preconditions at compile-time. If - the function is called with a noncompliant container and with - assertions switched off, the behavior is undefined and will most - likely yield segmentation violation. - - @param[in] i input to read from - @param[in] cb a parser callback function of type @ref parser_callback_t - which is used to control the deserialization by filtering unwanted values - (optional) - - @return result of the deserialization - - @throw parse_error.101 if a parse error occurs; example: `""unexpected end - of input; expected string literal""` - @throw parse_error.102 if to_unicode fails or surrogate error - @throw parse_error.103 if to_unicode fails - - @complexity Linear in the length of the input. The parser is a predictive - LL(1) parser. The complexity can be higher if the parser callback function - @a cb has a super-linear complexity. - - @note A UTF-8 byte order mark is silently ignored. - - @liveexample{The example below demonstrates the `parse()` function reading - from an array.,parse__array__parser_callback_t} - - @liveexample{The example below demonstrates the `parse()` function with - and without callback function.,parse__string__parser_callback_t} - - @liveexample{The example below demonstrates the `parse()` function with - and without callback function.,parse__istream__parser_callback_t} - - @liveexample{The example below demonstrates the `parse()` function reading - from a contiguous container.,parse__contiguouscontainer__parser_callback_t} - - @since version 2.0.3 (contiguous containers) - */ - static basic_json parse(detail::input_adapter i, - const parser_callback_t cb = nullptr, - const bool allow_exceptions = true) - { - basic_json result; - parser(i, cb, allow_exceptions).parse(true, result); - return result; - } - - /*! - @copydoc basic_json parse(detail::input_adapter, const parser_callback_t) - */ - static basic_json parse(detail::input_adapter& i, - const parser_callback_t cb = nullptr, - const bool allow_exceptions = true) - { - basic_json result; - parser(i, cb, allow_exceptions).parse(true, result); - return result; - } - - static bool accept(detail::input_adapter i) - { - return parser(i).accept(true); - } - - static bool accept(detail::input_adapter& i) - { - return parser(i).accept(true); - } - - /*! - @brief deserialize from an iterator range with contiguous storage - - This function reads from an iterator range of a container with contiguous - storage of 1-byte values. Compatible container types include - `std::vector`, `std::string`, `std::array`, `std::valarray`, and - `std::initializer_list`. Furthermore, C-style arrays can be used with - `std::begin()`/`std::end()`. User-defined containers can be used as long - as they implement random-access iterators and a contiguous storage. - - @pre The iterator range is contiguous. Violating this precondition yields - undefined behavior. **This precondition is enforced with an assertion.** - @pre Each element in the range has a size of 1 byte. Violating this - precondition yields undefined behavior. **This precondition is enforced - with a static assertion.** - - @warning There is no way to enforce all preconditions at compile-time. If - the function is called with noncompliant iterators and with - assertions switched off, the behavior is undefined and will most - likely yield segmentation violation. - - @tparam IteratorType iterator of container with contiguous storage - @param[in] first begin of the range to parse (included) - @param[in] last end of the range to parse (excluded) - @param[in] cb a parser callback function of type @ref parser_callback_t - which is used to control the deserialization by filtering unwanted values - (optional) - @param[in] allow_exceptions whether to throw exceptions in case of a - parse error (optional, true by default) - - @return result of the deserialization - - @throw parse_error.101 in case of an unexpected token - @throw parse_error.102 if to_unicode fails or surrogate error - @throw parse_error.103 if to_unicode fails - - @complexity Linear in the length of the input. The parser is a predictive - LL(1) parser. The complexity can be higher if the parser callback function - @a cb has a super-linear complexity. - - @note A UTF-8 byte order mark is silently ignored. - - @liveexample{The example below demonstrates the `parse()` function reading - from an iterator range.,parse__iteratortype__parser_callback_t} - - @since version 2.0.3 - */ - template<class IteratorType, typename std::enable_if< - std::is_base_of< - std::random_access_iterator_tag, - typename std::iterator_traits<IteratorType>::iterator_category>::value, int>::type = 0> - static basic_json parse(IteratorType first, IteratorType last, - const parser_callback_t cb = nullptr, - const bool allow_exceptions = true) - { - basic_json result; - parser(detail::input_adapter(first, last), cb, allow_exceptions).parse(true, result); - return result; - } - - template<class IteratorType, typename std::enable_if< - std::is_base_of< - std::random_access_iterator_tag, - typename std::iterator_traits<IteratorType>::iterator_category>::value, int>::type = 0> - static bool accept(IteratorType first, IteratorType last) - { - return parser(detail::input_adapter(first, last)).accept(true); - } - - /*! - @brief deserialize from stream - @deprecated This stream operator is deprecated and will be removed in - version 4.0.0 of the library. Please use - @ref operator>>(std::istream&, basic_json&) - instead; that is, replace calls like `j << i;` with `i >> j;`. - @since version 1.0.0; deprecated since version 3.0.0 - */ - JSON_DEPRECATED - friend std::istream& operator<<(basic_json& j, std::istream& i) - { - return operator>>(i, j); - } - - /*! - @brief deserialize from stream - - Deserializes an input stream to a JSON value. - - @param[in,out] i input stream to read a serialized JSON value from - @param[in,out] j JSON value to write the deserialized input to - - @throw parse_error.101 in case of an unexpected token - @throw parse_error.102 if to_unicode fails or surrogate error - @throw parse_error.103 if to_unicode fails - - @complexity Linear in the length of the input. The parser is a predictive - LL(1) parser. - - @note A UTF-8 byte order mark is silently ignored. - - @liveexample{The example below shows how a JSON value is constructed by - reading a serialization from a stream.,operator_deserialize} - - @sa parse(std::istream&, const parser_callback_t) for a variant with a - parser callback function to filter values while parsing - - @since version 1.0.0 - */ - friend std::istream& operator>>(std::istream& i, basic_json& j) - { - parser(detail::input_adapter(i)).parse(false, j); - return i; - } - - /// @} - - /////////////////////////// - // convenience functions // - /////////////////////////// - - /*! - @brief return the type as string - - Returns the type name as string to be used in error messages - usually to - indicate that a function was called on a wrong JSON type. - - @return a string representation of a the @a m_type member: - Value type | return value - ----------- | ------------- - null | `"null"` - boolean | `"boolean"` - string | `"string"` - number | `"number"` (for all number types) - object | `"object"` - array | `"array"` - discarded | `"discarded"` - - @exceptionsafety No-throw guarantee: this function never throws exceptions. - - @complexity Constant. - - @liveexample{The following code exemplifies `type_name()` for all JSON - types.,type_name} - - @sa @ref type() -- return the type of the JSON value - @sa @ref operator value_t() -- return the type of the JSON value (implicit) - - @since version 1.0.0, public since 2.1.0, `const char*` and `noexcept` - since 3.0.0 - */ - const char* type_name() const noexcept - { - { - switch (m_type) - { - case value_t::null: - return "null"; - case value_t::object: - return "object"; - case value_t::array: - return "array"; - case value_t::string: - return "string"; - case value_t::boolean: - return "boolean"; - case value_t::discarded: - return "discarded"; - default: - return "number"; - } - } - } - - - private: - ////////////////////// - // member variables // - ////////////////////// - - /// the type of the current element - value_t m_type = value_t::null; - - /// the value of the current element - json_value m_value = {}; - - ////////////////////////////////////////// - // binary serialization/deserialization // - ////////////////////////////////////////// - - /// @name binary serialization/deserialization support - /// @{ - - public: - /*! - @brief create a CBOR serialization of a given JSON value - - Serializes a given JSON value @a j to a byte vector using the CBOR (Concise - Binary Object Representation) serialization format. CBOR is a binary - serialization format which aims to be more compact than JSON itself, yet - more efficient to parse. - - The library uses the following mapping from JSON values types to - CBOR types according to the CBOR specification (RFC 7049): - - JSON value type | value/range | CBOR type | first byte - --------------- | ------------------------------------------ | ---------------------------------- | --------------- - null | `null` | Null | 0xF6 - boolean | `true` | True | 0xF5 - boolean | `false` | False | 0xF4 - number_integer | -9223372036854775808..-2147483649 | Negative integer (8 bytes follow) | 0x3B - number_integer | -2147483648..-32769 | Negative integer (4 bytes follow) | 0x3A - number_integer | -32768..-129 | Negative integer (2 bytes follow) | 0x39 - number_integer | -128..-25 | Negative integer (1 byte follow) | 0x38 - number_integer | -24..-1 | Negative integer | 0x20..0x37 - number_integer | 0..23 | Integer | 0x00..0x17 - number_integer | 24..255 | Unsigned integer (1 byte follow) | 0x18 - number_integer | 256..65535 | Unsigned integer (2 bytes follow) | 0x19 - number_integer | 65536..4294967295 | Unsigned integer (4 bytes follow) | 0x1A - number_integer | 4294967296..18446744073709551615 | Unsigned integer (8 bytes follow) | 0x1B - number_unsigned | 0..23 | Integer | 0x00..0x17 - number_unsigned | 24..255 | Unsigned integer (1 byte follow) | 0x18 - number_unsigned | 256..65535 | Unsigned integer (2 bytes follow) | 0x19 - number_unsigned | 65536..4294967295 | Unsigned integer (4 bytes follow) | 0x1A - number_unsigned | 4294967296..18446744073709551615 | Unsigned integer (8 bytes follow) | 0x1B - number_float | *any value* | Double-Precision Float | 0xFB - string | *length*: 0..23 | UTF-8 string | 0x60..0x77 - string | *length*: 23..255 | UTF-8 string (1 byte follow) | 0x78 - string | *length*: 256..65535 | UTF-8 string (2 bytes follow) | 0x79 - string | *length*: 65536..4294967295 | UTF-8 string (4 bytes follow) | 0x7A - string | *length*: 4294967296..18446744073709551615 | UTF-8 string (8 bytes follow) | 0x7B - array | *size*: 0..23 | array | 0x80..0x97 - array | *size*: 23..255 | array (1 byte follow) | 0x98 - array | *size*: 256..65535 | array (2 bytes follow) | 0x99 - array | *size*: 65536..4294967295 | array (4 bytes follow) | 0x9A - array | *size*: 4294967296..18446744073709551615 | array (8 bytes follow) | 0x9B - object | *size*: 0..23 | map | 0xA0..0xB7 - object | *size*: 23..255 | map (1 byte follow) | 0xB8 - object | *size*: 256..65535 | map (2 bytes follow) | 0xB9 - object | *size*: 65536..4294967295 | map (4 bytes follow) | 0xBA - object | *size*: 4294967296..18446744073709551615 | map (8 bytes follow) | 0xBB - - @note The mapping is **complete** in the sense that any JSON value type - can be converted to a CBOR value. - - @note If NaN or Infinity are stored inside a JSON number, they are - serialized properly. This behavior differs from the @ref dump() - function which serializes NaN or Infinity to `null`. - - @note The following CBOR types are not used in the conversion: - - byte strings (0x40..0x5F) - - UTF-8 strings terminated by "break" (0x7F) - - arrays terminated by "break" (0x9F) - - maps terminated by "break" (0xBF) - - date/time (0xC0..0xC1) - - bignum (0xC2..0xC3) - - decimal fraction (0xC4) - - bigfloat (0xC5) - - tagged items (0xC6..0xD4, 0xD8..0xDB) - - expected conversions (0xD5..0xD7) - - simple values (0xE0..0xF3, 0xF8) - - undefined (0xF7) - - half and single-precision floats (0xF9-0xFA) - - break (0xFF) - - @param[in] j JSON value to serialize - @return MessagePack serialization as byte vector - - @complexity Linear in the size of the JSON value @a j. - - @liveexample{The example shows the serialization of a JSON value to a byte - vector in CBOR format.,to_cbor} - - @sa http://cbor.io - @sa @ref from_cbor(detail::input_adapter, const bool strict) for the - analogous deserialization - @sa @ref to_msgpack(const basic_json&) for the related MessagePack format - @sa @ref to_ubjson(const basic_json&, const bool, const bool) for the - related UBJSON format - - @since version 2.0.9 - */ - static std::vector<uint8_t> to_cbor(const basic_json& j) - { - std::vector<uint8_t> result; - to_cbor(j, result); - return result; - } - - static void to_cbor(const basic_json& j, detail::output_adapter<uint8_t> o) - { - binary_writer<uint8_t>(o).write_cbor(j); - } - - static void to_cbor(const basic_json& j, detail::output_adapter<char> o) - { - binary_writer<char>(o).write_cbor(j); - } - - /*! - @brief create a MessagePack serialization of a given JSON value - - Serializes a given JSON value @a j to a byte vector using the MessagePack - serialization format. MessagePack is a binary serialization format which - aims to be more compact than JSON itself, yet more efficient to parse. - - The library uses the following mapping from JSON values types to - MessagePack types according to the MessagePack specification: - - JSON value type | value/range | MessagePack type | first byte - --------------- | --------------------------------- | ---------------- | ---------- - null | `null` | nil | 0xC0 - boolean | `true` | true | 0xC3 - boolean | `false` | false | 0xC2 - number_integer | -9223372036854775808..-2147483649 | int64 | 0xD3 - number_integer | -2147483648..-32769 | int32 | 0xD2 - number_integer | -32768..-129 | int16 | 0xD1 - number_integer | -128..-33 | int8 | 0xD0 - number_integer | -32..-1 | negative fixint | 0xE0..0xFF - number_integer | 0..127 | positive fixint | 0x00..0x7F - number_integer | 128..255 | uint 8 | 0xCC - number_integer | 256..65535 | uint 16 | 0xCD - number_integer | 65536..4294967295 | uint 32 | 0xCE - number_integer | 4294967296..18446744073709551615 | uint 64 | 0xCF - number_unsigned | 0..127 | positive fixint | 0x00..0x7F - number_unsigned | 128..255 | uint 8 | 0xCC - number_unsigned | 256..65535 | uint 16 | 0xCD - number_unsigned | 65536..4294967295 | uint 32 | 0xCE - number_unsigned | 4294967296..18446744073709551615 | uint 64 | 0xCF - number_float | *any value* | float 64 | 0xCB - string | *length*: 0..31 | fixstr | 0xA0..0xBF - string | *length*: 32..255 | str 8 | 0xD9 - string | *length*: 256..65535 | str 16 | 0xDA - string | *length*: 65536..4294967295 | str 32 | 0xDB - array | *size*: 0..15 | fixarray | 0x90..0x9F - array | *size*: 16..65535 | array 16 | 0xDC - array | *size*: 65536..4294967295 | array 32 | 0xDD - object | *size*: 0..15 | fix map | 0x80..0x8F - object | *size*: 16..65535 | map 16 | 0xDE - object | *size*: 65536..4294967295 | map 32 | 0xDF - - @note The mapping is **complete** in the sense that any JSON value type - can be converted to a MessagePack value. - - @note The following values can **not** be converted to a MessagePack value: - - strings with more than 4294967295 bytes - - arrays with more than 4294967295 elements - - objects with more than 4294967295 elements - - @note The following MessagePack types are not used in the conversion: - - bin 8 - bin 32 (0xC4..0xC6) - - ext 8 - ext 32 (0xC7..0xC9) - - float 32 (0xCA) - - fixext 1 - fixext 16 (0xD4..0xD8) - - @note Any MessagePack output created @ref to_msgpack can be successfully - parsed by @ref from_msgpack. - - @note If NaN or Infinity are stored inside a JSON number, they are - serialized properly. This behavior differs from the @ref dump() - function which serializes NaN or Infinity to `null`. - - @param[in] j JSON value to serialize - @return MessagePack serialization as byte vector - - @complexity Linear in the size of the JSON value @a j. - - @liveexample{The example shows the serialization of a JSON value to a byte - vector in MessagePack format.,to_msgpack} - - @sa http://msgpack.org - @sa @ref from_msgpack(const std::vector<uint8_t>&, const size_t) for the - analogous deserialization - @sa @ref to_cbor(const basic_json& for the related CBOR format - @sa @ref to_ubjson(const basic_json&, const bool, const bool) for the - related UBJSON format - - @since version 2.0.9 - */ - static std::vector<uint8_t> to_msgpack(const basic_json& j) - { - std::vector<uint8_t> result; - to_msgpack(j, result); - return result; - } - - static void to_msgpack(const basic_json& j, detail::output_adapter<uint8_t> o) - { - binary_writer<uint8_t>(o).write_msgpack(j); - } - - static void to_msgpack(const basic_json& j, detail::output_adapter<char> o) - { - binary_writer<char>(o).write_msgpack(j); - } - - /*! - @brief create a UBJSON serialization of a given JSON value - - Serializes a given JSON value @a j to a byte vector using the UBJSON - (Universal Binary JSON) serialization format. UBJSON aims to be more compact - than JSON itself, yet more efficient to parse. - - The library uses the following mapping from JSON values types to - UBJSON types according to the UBJSON specification: - - JSON value type | value/range | UBJSON type | marker - --------------- | --------------------------------- | ----------- | ------ - null | `null` | null | `Z` - boolean | `true` | true | `T` - boolean | `false` | false | `F` - number_integer | -9223372036854775808..-2147483649 | int64 | `L` - number_integer | -2147483648..-32769 | int32 | `l` - number_integer | -32768..-129 | int16 | `I` - number_integer | -128..127 | int8 | `i` - number_integer | 128..255 | uint8 | `U` - number_integer | 256..32767 | int16 | `I` - number_integer | 32768..2147483647 | int32 | `l` - number_integer | 2147483648..9223372036854775807 | int64 | `L` - number_unsigned | 0..127 | int8 | `i` - number_unsigned | 128..255 | uint8 | `U` - number_unsigned | 256..32767 | int16 | `I` - number_unsigned | 32768..2147483647 | int32 | `l` - number_unsigned | 2147483648..9223372036854775807 | int64 | `L` - number_float | *any value* | float64 | `D` - string | *with shortest length indicator* | string | `S` - array | *see notes on optimized format* | array | `[` - object | *see notes on optimized format* | map | `{` - - @note The mapping is **complete** in the sense that any JSON value type - can be converted to a UBJSON value. - - @note The following values can **not** be converted to a UBJSON value: - - strings with more than 9223372036854775807 bytes (theoretical) - - unsigned integer numbers above 9223372036854775807 - - @note The following markers are not used in the conversion: - - `Z`: no-op values are not created. - - `C`: single-byte strings are serialized with `S` markers. - - @note Any UBJSON output created @ref to_ubjson can be successfully parsed - by @ref from_ubjson. - - @note If NaN or Infinity are stored inside a JSON number, they are - serialized properly. This behavior differs from the @ref dump() - function which serializes NaN or Infinity to `null`. - - @note The optimized formats for containers are supported: Parameter - @a use_size adds size information to the beginning of a container and - removes the closing marker. Parameter @a use_type further checks - whether all elements of a container have the same type and adds the - type marker to the beginning of the container. The @a use_type - parameter must only be used together with @a use_size = true. Note - that @a use_size = true alone may result in larger representations - - the benefit of this parameter is that the receiving side is - immediately informed on the number of elements of the container. - - @param[in] j JSON value to serialize - @param[in] use_size whether to add size annotations to container types - @param[in] use_type whether to add type annotations to container types - (must be combined with @a use_size = true) - @return UBJSON serialization as byte vector - - @complexity Linear in the size of the JSON value @a j. - - @liveexample{The example shows the serialization of a JSON value to a byte - vector in UBJSON format.,to_ubjson} - - @sa http://ubjson.org - @sa @ref from_ubjson(detail::input_adapter, const bool strict) for the - analogous deserialization - @sa @ref to_cbor(const basic_json& for the related CBOR format - @sa @ref to_msgpack(const basic_json&) for the related MessagePack format - - @since version 3.1.0 - */ - static std::vector<uint8_t> to_ubjson(const basic_json& j, - const bool use_size = false, - const bool use_type = false) - { - std::vector<uint8_t> result; - to_ubjson(j, result, use_size, use_type); - return result; - } - - static void to_ubjson(const basic_json& j, detail::output_adapter<uint8_t> o, - const bool use_size = false, const bool use_type = false) - { - binary_writer<uint8_t>(o).write_ubjson(j, use_size, use_type); - } - - static void to_ubjson(const basic_json& j, detail::output_adapter<char> o, - const bool use_size = false, const bool use_type = false) - { - binary_writer<char>(o).write_ubjson(j, use_size, use_type); - } - - /*! - @brief create a JSON value from an input in CBOR format - - Deserializes a given input @a i to a JSON value using the CBOR (Concise - Binary Object Representation) serialization format. - - The library maps CBOR types to JSON value types as follows: - - CBOR type | JSON value type | first byte - ---------------------- | --------------- | ---------- - Integer | number_unsigned | 0x00..0x17 - Unsigned integer | number_unsigned | 0x18 - Unsigned integer | number_unsigned | 0x19 - Unsigned integer | number_unsigned | 0x1A - Unsigned integer | number_unsigned | 0x1B - Negative integer | number_integer | 0x20..0x37 - Negative integer | number_integer | 0x38 - Negative integer | number_integer | 0x39 - Negative integer | number_integer | 0x3A - Negative integer | number_integer | 0x3B - Negative integer | number_integer | 0x40..0x57 - UTF-8 string | string | 0x60..0x77 - UTF-8 string | string | 0x78 - UTF-8 string | string | 0x79 - UTF-8 string | string | 0x7A - UTF-8 string | string | 0x7B - UTF-8 string | string | 0x7F - array | array | 0x80..0x97 - array | array | 0x98 - array | array | 0x99 - array | array | 0x9A - array | array | 0x9B - array | array | 0x9F - map | object | 0xA0..0xB7 - map | object | 0xB8 - map | object | 0xB9 - map | object | 0xBA - map | object | 0xBB - map | object | 0xBF - False | `false` | 0xF4 - True | `true` | 0xF5 - Nill | `null` | 0xF6 - Half-Precision Float | number_float | 0xF9 - Single-Precision Float | number_float | 0xFA - Double-Precision Float | number_float | 0xFB - - @warning The mapping is **incomplete** in the sense that not all CBOR - types can be converted to a JSON value. The following CBOR types - are not supported and will yield parse errors (parse_error.112): - - byte strings (0x40..0x5F) - - date/time (0xC0..0xC1) - - bignum (0xC2..0xC3) - - decimal fraction (0xC4) - - bigfloat (0xC5) - - tagged items (0xC6..0xD4, 0xD8..0xDB) - - expected conversions (0xD5..0xD7) - - simple values (0xE0..0xF3, 0xF8) - - undefined (0xF7) - - @warning CBOR allows map keys of any type, whereas JSON only allows - strings as keys in object values. Therefore, CBOR maps with keys - other than UTF-8 strings are rejected (parse_error.113). - - @note Any CBOR output created @ref to_cbor can be successfully parsed by - @ref from_cbor. - - @param[in] i an input in CBOR format convertible to an input adapter - @param[in] strict whether to expect the input to be consumed until EOF - (true by default) - @return deserialized JSON value - - @throw parse_error.110 if the given input ends prematurely or the end of - file was not reached when @a strict was set to true - @throw parse_error.112 if unsupported features from CBOR were - used in the given input @a v or if the input is not valid CBOR - @throw parse_error.113 if a string was expected as map key, but not found - - @complexity Linear in the size of the input @a i. - - @liveexample{The example shows the deserialization of a byte vector in CBOR - format to a JSON value.,from_cbor} - - @sa http://cbor.io - @sa @ref to_cbor(const basic_json&) for the analogous serialization - @sa @ref from_msgpack(detail::input_adapter, const bool) for the - related MessagePack format - @sa @ref from_ubjson(detail::input_adapter, const bool) for the related - UBJSON format - - @since version 2.0.9; parameter @a start_index since 2.1.1; changed to - consume input adapters, removed start_index parameter, and added - @a strict parameter since 3.0.0 - */ - static basic_json from_cbor(detail::input_adapter i, - const bool strict = true) - { - return binary_reader(i).parse_cbor(strict); - } - - /*! - @copydoc from_cbor(detail::input_adapter, const bool) - */ - template<typename A1, typename A2, - detail::enable_if_t<std::is_constructible<detail::input_adapter, A1, A2>::value, int> = 0> - static basic_json from_cbor(A1 && a1, A2 && a2, const bool strict = true) - { - return binary_reader(detail::input_adapter(std::forward<A1>(a1), std::forward<A2>(a2))).parse_cbor(strict); - } - - /*! - @brief create a JSON value from an input in MessagePack format - - Deserializes a given input @a i to a JSON value using the MessagePack - serialization format. - - The library maps MessagePack types to JSON value types as follows: - - MessagePack type | JSON value type | first byte - ---------------- | --------------- | ---------- - positive fixint | number_unsigned | 0x00..0x7F - fixmap | object | 0x80..0x8F - fixarray | array | 0x90..0x9F - fixstr | string | 0xA0..0xBF - nil | `null` | 0xC0 - false | `false` | 0xC2 - true | `true` | 0xC3 - float 32 | number_float | 0xCA - float 64 | number_float | 0xCB - uint 8 | number_unsigned | 0xCC - uint 16 | number_unsigned | 0xCD - uint 32 | number_unsigned | 0xCE - uint 64 | number_unsigned | 0xCF - int 8 | number_integer | 0xD0 - int 16 | number_integer | 0xD1 - int 32 | number_integer | 0xD2 - int 64 | number_integer | 0xD3 - str 8 | string | 0xD9 - str 16 | string | 0xDA - str 32 | string | 0xDB - array 16 | array | 0xDC - array 32 | array | 0xDD - map 16 | object | 0xDE - map 32 | object | 0xDF - negative fixint | number_integer | 0xE0-0xFF - - @warning The mapping is **incomplete** in the sense that not all - MessagePack types can be converted to a JSON value. The following - MessagePack types are not supported and will yield parse errors: - - bin 8 - bin 32 (0xC4..0xC6) - - ext 8 - ext 32 (0xC7..0xC9) - - fixext 1 - fixext 16 (0xD4..0xD8) - - @note Any MessagePack output created @ref to_msgpack can be successfully - parsed by @ref from_msgpack. - - @param[in] i an input in MessagePack format convertible to an input - adapter - @param[in] strict whether to expect the input to be consumed until EOF - (true by default) - - @throw parse_error.110 if the given input ends prematurely or the end of - file was not reached when @a strict was set to true - @throw parse_error.112 if unsupported features from MessagePack were - used in the given input @a i or if the input is not valid MessagePack - @throw parse_error.113 if a string was expected as map key, but not found - - @complexity Linear in the size of the input @a i. - - @liveexample{The example shows the deserialization of a byte vector in - MessagePack format to a JSON value.,from_msgpack} - - @sa http://msgpack.org - @sa @ref to_msgpack(const basic_json&) for the analogous serialization - @sa @ref from_cbor(detail::input_adapter, const bool) for the related CBOR - format - @sa @ref from_ubjson(detail::input_adapter, const bool) for the related - UBJSON format - - @since version 2.0.9; parameter @a start_index since 2.1.1; changed to - consume input adapters, removed start_index parameter, and added - @a strict parameter since 3.0.0 - */ - static basic_json from_msgpack(detail::input_adapter i, - const bool strict = true) - { - return binary_reader(i).parse_msgpack(strict); - } - - /*! - @copydoc from_msgpack(detail::input_adapter, const bool) - */ - template<typename A1, typename A2, - detail::enable_if_t<std::is_constructible<detail::input_adapter, A1, A2>::value, int> = 0> - static basic_json from_msgpack(A1 && a1, A2 && a2, const bool strict = true) - { - return binary_reader(detail::input_adapter(std::forward<A1>(a1), std::forward<A2>(a2))).parse_msgpack(strict); - } - - /*! - @brief create a JSON value from an input in UBJSON format - - Deserializes a given input @a i to a JSON value using the UBJSON (Universal - Binary JSON) serialization format. - - The library maps UBJSON types to JSON value types as follows: - - UBJSON type | JSON value type | marker - ----------- | --------------------------------------- | ------ - no-op | *no value, next value is read* | `N` - null | `null` | `Z` - false | `false` | `F` - true | `true` | `T` - float32 | number_float | `d` - float64 | number_float | `D` - uint8 | number_unsigned | `U` - int8 | number_integer | `i` - int16 | number_integer | `I` - int32 | number_integer | `l` - int64 | number_integer | `L` - string | string | `S` - char | string | `C` - array | array (optimized values are supported) | `[` - object | object (optimized values are supported) | `{` - - @note The mapping is **complete** in the sense that any UBJSON value can - be converted to a JSON value. - - @param[in] i an input in UBJSON format convertible to an input adapter - @param[in] strict whether to expect the input to be consumed until EOF - (true by default) - - @throw parse_error.110 if the given input ends prematurely or the end of - file was not reached when @a strict was set to true - @throw parse_error.112 if a parse error occurs - @throw parse_error.113 if a string could not be parsed successfully - - @complexity Linear in the size of the input @a i. - - @liveexample{The example shows the deserialization of a byte vector in - UBJSON format to a JSON value.,from_ubjson} - - @sa http://ubjson.org - @sa @ref to_ubjson(const basic_json&, const bool, const bool) for the - analogous serialization - @sa @ref from_cbor(detail::input_adapter, const bool) for the related CBOR - format - @sa @ref from_msgpack(detail::input_adapter, const bool) for the related - MessagePack format - - @since version 3.1.0 - */ - static basic_json from_ubjson(detail::input_adapter i, - const bool strict = true) - { - return binary_reader(i).parse_ubjson(strict); - } - - template<typename A1, typename A2, - detail::enable_if_t<std::is_constructible<detail::input_adapter, A1, A2>::value, int> = 0> - static basic_json from_ubjson(A1 && a1, A2 && a2, const bool strict = true) - { - return binary_reader(detail::input_adapter(std::forward<A1>(a1), std::forward<A2>(a2))).parse_ubjson(strict); - } - - /// @} - - ////////////////////////// - // JSON Pointer support // - ////////////////////////// - - /// @name JSON Pointer functions - /// @{ - - /*! - @brief access specified element via JSON Pointer - - Uses a JSON pointer to retrieve a reference to the respective JSON value. - No bound checking is performed. Similar to @ref operator[](const typename - object_t::key_type&), `null` values are created in arrays and objects if - necessary. - - In particular: - - If the JSON pointer points to an object key that does not exist, it - is created an filled with a `null` value before a reference to it - is returned. - - If the JSON pointer points to an array index that does not exist, it - is created an filled with a `null` value before a reference to it - is returned. All indices between the current maximum and the given - index are also filled with `null`. - - The special value `-` is treated as a synonym for the index past the - end. - - @param[in] ptr a JSON pointer - - @return reference to the element pointed to by @a ptr - - @complexity Constant. - - @throw parse_error.106 if an array index begins with '0' - @throw parse_error.109 if an array index was not a number - @throw out_of_range.404 if the JSON pointer can not be resolved - - @liveexample{The behavior is shown in the example.,operatorjson_pointer} - - @since version 2.0.0 - */ - reference operator[](const json_pointer& ptr) - { - return ptr.get_unchecked(this); - } - - /*! - @brief access specified element via JSON Pointer - - Uses a JSON pointer to retrieve a reference to the respective JSON value. - No bound checking is performed. The function does not change the JSON - value; no `null` values are created. In particular, the the special value - `-` yields an exception. - - @param[in] ptr JSON pointer to the desired element - - @return const reference to the element pointed to by @a ptr - - @complexity Constant. - - @throw parse_error.106 if an array index begins with '0' - @throw parse_error.109 if an array index was not a number - @throw out_of_range.402 if the array index '-' is used - @throw out_of_range.404 if the JSON pointer can not be resolved - - @liveexample{The behavior is shown in the example.,operatorjson_pointer_const} - - @since version 2.0.0 - */ - const_reference operator[](const json_pointer& ptr) const - { - return ptr.get_unchecked(this); - } - - /*! - @brief access specified element via JSON Pointer - - Returns a reference to the element at with specified JSON pointer @a ptr, - with bounds checking. - - @param[in] ptr JSON pointer to the desired element - - @return reference to the element pointed to by @a ptr - - @throw parse_error.106 if an array index in the passed JSON pointer @a ptr - begins with '0'. See example below. - - @throw parse_error.109 if an array index in the passed JSON pointer @a ptr - is not a number. See example below. - - @throw out_of_range.401 if an array index in the passed JSON pointer @a ptr - is out of range. See example below. - - @throw out_of_range.402 if the array index '-' is used in the passed JSON - pointer @a ptr. As `at` provides checked access (and no elements are - implicitly inserted), the index '-' is always invalid. See example below. - - @throw out_of_range.403 if the JSON pointer describes a key of an object - which cannot be found. See example below. - - @throw out_of_range.404 if the JSON pointer @a ptr can not be resolved. - See example below. - - @exceptionsafety Strong guarantee: if an exception is thrown, there are no - changes in the JSON value. - - @complexity Constant. - - @since version 2.0.0 - - @liveexample{The behavior is shown in the example.,at_json_pointer} - */ - reference at(const json_pointer& ptr) - { - return ptr.get_checked(this); - } - - /*! - @brief access specified element via JSON Pointer - - Returns a const reference to the element at with specified JSON pointer @a - ptr, with bounds checking. - - @param[in] ptr JSON pointer to the desired element - - @return reference to the element pointed to by @a ptr - - @throw parse_error.106 if an array index in the passed JSON pointer @a ptr - begins with '0'. See example below. - - @throw parse_error.109 if an array index in the passed JSON pointer @a ptr - is not a number. See example below. - - @throw out_of_range.401 if an array index in the passed JSON pointer @a ptr - is out of range. See example below. - - @throw out_of_range.402 if the array index '-' is used in the passed JSON - pointer @a ptr. As `at` provides checked access (and no elements are - implicitly inserted), the index '-' is always invalid. See example below. - - @throw out_of_range.403 if the JSON pointer describes a key of an object - which cannot be found. See example below. - - @throw out_of_range.404 if the JSON pointer @a ptr can not be resolved. - See example below. - - @exceptionsafety Strong guarantee: if an exception is thrown, there are no - changes in the JSON value. - - @complexity Constant. - - @since version 2.0.0 - - @liveexample{The behavior is shown in the example.,at_json_pointer_const} - */ - const_reference at(const json_pointer& ptr) const - { - return ptr.get_checked(this); - } - - /*! - @brief return flattened JSON value - - The function creates a JSON object whose keys are JSON pointers (see [RFC - 6901](https://tools.ietf.org/html/rfc6901)) and whose values are all - primitive. The original JSON value can be restored using the @ref - unflatten() function. - - @return an object that maps JSON pointers to primitive values - - @note Empty objects and arrays are flattened to `null` and will not be - reconstructed correctly by the @ref unflatten() function. - - @complexity Linear in the size the JSON value. - - @liveexample{The following code shows how a JSON object is flattened to an - object whose keys consist of JSON pointers.,flatten} - - @sa @ref unflatten() for the reverse function - - @since version 2.0.0 - */ - basic_json flatten() const - { - basic_json result(value_t::object); - json_pointer::flatten("", *this, result); - return result; - } - - /*! - @brief unflatten a previously flattened JSON value - - The function restores the arbitrary nesting of a JSON value that has been - flattened before using the @ref flatten() function. The JSON value must - meet certain constraints: - 1. The value must be an object. - 2. The keys must be JSON pointers (see - [RFC 6901](https://tools.ietf.org/html/rfc6901)) - 3. The mapped values must be primitive JSON types. - - @return the original JSON from a flattened version - - @note Empty objects and arrays are flattened by @ref flatten() to `null` - values and can not unflattened to their original type. Apart from - this example, for a JSON value `j`, the following is always true: - `j == j.flatten().unflatten()`. - - @complexity Linear in the size the JSON value. - - @throw type_error.314 if value is not an object - @throw type_error.315 if object values are not primitive - - @liveexample{The following code shows how a flattened JSON object is - unflattened into the original nested JSON object.,unflatten} - - @sa @ref flatten() for the reverse function - - @since version 2.0.0 - */ - basic_json unflatten() const - { - return json_pointer::unflatten(*this); - } - - /// @} - - ////////////////////////// - // JSON Patch functions // - ////////////////////////// - - /// @name JSON Patch functions - /// @{ - - /*! - @brief applies a JSON patch - - [JSON Patch](http://jsonpatch.com) defines a JSON document structure for - expressing a sequence of operations to apply to a JSON) document. With - this function, a JSON Patch is applied to the current JSON value by - executing all operations from the patch. - - @param[in] json_patch JSON patch document - @return patched document - - @note The application of a patch is atomic: Either all operations succeed - and the patched document is returned or an exception is thrown. In - any case, the original value is not changed: the patch is applied - to a copy of the value. - - @throw parse_error.104 if the JSON patch does not consist of an array of - objects - - @throw parse_error.105 if the JSON patch is malformed (e.g., mandatory - attributes are missing); example: `"operation add must have member path"` - - @throw out_of_range.401 if an array index is out of range. - - @throw out_of_range.403 if a JSON pointer inside the patch could not be - resolved successfully in the current JSON value; example: `"key baz not - found"` - - @throw out_of_range.405 if JSON pointer has no parent ("add", "remove", - "move") - - @throw other_error.501 if "test" operation was unsuccessful - - @complexity Linear in the size of the JSON value and the length of the - JSON patch. As usually only a fraction of the JSON value is affected by - the patch, the complexity can usually be neglected. - - @liveexample{The following code shows how a JSON patch is applied to a - value.,patch} - - @sa @ref diff -- create a JSON patch by comparing two JSON values - - @sa [RFC 6902 (JSON Patch)](https://tools.ietf.org/html/rfc6902) - @sa [RFC 6901 (JSON Pointer)](https://tools.ietf.org/html/rfc6901) - - @since version 2.0.0 - */ - basic_json patch(const basic_json& json_patch) const - { - // make a working copy to apply the patch to - basic_json result = *this; - - // the valid JSON Patch operations - enum class patch_operations {add, remove, replace, move, copy, test, invalid}; - - const auto get_op = [](const std::string & op) - { - if (op == "add") - { - return patch_operations::add; - } - if (op == "remove") - { - return patch_operations::remove; - } - if (op == "replace") - { - return patch_operations::replace; - } - if (op == "move") - { - return patch_operations::move; - } - if (op == "copy") - { - return patch_operations::copy; - } - if (op == "test") - { - return patch_operations::test; - } - - return patch_operations::invalid; - }; - - // wrapper for "add" operation; add value at ptr - const auto operation_add = [&result](json_pointer & ptr, basic_json val) - { - // adding to the root of the target document means replacing it - if (ptr.is_root()) - { - result = val; - } - else - { - // make sure the top element of the pointer exists - json_pointer top_pointer = ptr.top(); - if (top_pointer != ptr) - { - result.at(top_pointer); - } - - // get reference to parent of JSON pointer ptr - const auto last_path = ptr.pop_back(); - basic_json& parent = result[ptr]; - - switch (parent.m_type) - { - case value_t::null: - case value_t::object: - { - // use operator[] to add value - parent[last_path] = val; - break; - } - - case value_t::array: - { - if (last_path == "-") - { - // special case: append to back - parent.push_back(val); - } - else - { - const auto idx = json_pointer::array_index(last_path); - if (JSON_UNLIKELY(static_cast<size_type>(idx) > parent.size())) - { - // avoid undefined behavior - JSON_THROW(out_of_range::create(401, "array index " + std::to_string(idx) + " is out of range")); - } - else - { - // default case: insert add offset - parent.insert(parent.begin() + static_cast<difference_type>(idx), val); - } - } - break; - } - - default: - { - // if there exists a parent it cannot be primitive - assert(false); // LCOV_EXCL_LINE - } - } - } - }; - - // wrapper for "remove" operation; remove value at ptr - const auto operation_remove = [&result](json_pointer & ptr) - { - // get reference to parent of JSON pointer ptr - const auto last_path = ptr.pop_back(); - basic_json& parent = result.at(ptr); - - // remove child - if (parent.is_object()) - { - // perform range check - auto it = parent.find(last_path); - if (JSON_LIKELY(it != parent.end())) - { - parent.erase(it); - } - else - { - JSON_THROW(out_of_range::create(403, "key '" + last_path + "' not found")); - } - } - else if (parent.is_array()) - { - // note erase performs range check - parent.erase(static_cast<size_type>(json_pointer::array_index(last_path))); - } - }; - - // type check: top level value must be an array - if (JSON_UNLIKELY(not json_patch.is_array())) - { - JSON_THROW(parse_error::create(104, 0, "JSON patch must be an array of objects")); - } - - // iterate and apply the operations - for (const auto& val : json_patch) - { - // wrapper to get a value for an operation - const auto get_value = [&val](const std::string & op, - const std::string & member, - bool string_type) -> basic_json & - { - // find value - auto it = val.m_value.object->find(member); - - // context-sensitive error message - const auto error_msg = (op == "op") ? "operation" : "operation '" + op + "'"; - - // check if desired value is present - if (JSON_UNLIKELY(it == val.m_value.object->end())) - { - JSON_THROW(parse_error::create(105, 0, error_msg + " must have member '" + member + "'")); - } - - // check if result is of type string - if (JSON_UNLIKELY(string_type and not it->second.is_string())) - { - JSON_THROW(parse_error::create(105, 0, error_msg + " must have string member '" + member + "'")); - } - - // no error: return value - return it->second; - }; - - // type check: every element of the array must be an object - if (JSON_UNLIKELY(not val.is_object())) - { - JSON_THROW(parse_error::create(104, 0, "JSON patch must be an array of objects")); - } - - // collect mandatory members - const std::string op = get_value("op", "op", true); - const std::string path = get_value(op, "path", true); - json_pointer ptr(path); - - switch (get_op(op)) - { - case patch_operations::add: - { - operation_add(ptr, get_value("add", "value", false)); - break; - } - - case patch_operations::remove: - { - operation_remove(ptr); - break; - } - - case patch_operations::replace: - { - // the "path" location must exist - use at() - result.at(ptr) = get_value("replace", "value", false); - break; - } - - case patch_operations::move: - { - const std::string from_path = get_value("move", "from", true); - json_pointer from_ptr(from_path); - - // the "from" location must exist - use at() - basic_json v = result.at(from_ptr); - - // The move operation is functionally identical to a - // "remove" operation on the "from" location, followed - // immediately by an "add" operation at the target - // location with the value that was just removed. - operation_remove(from_ptr); - operation_add(ptr, v); - break; - } - - case patch_operations::copy: - { - const std::string from_path = get_value("copy", "from", true); - const json_pointer from_ptr(from_path); - - // the "from" location must exist - use at() - basic_json v = result.at(from_ptr); - - // The copy is functionally identical to an "add" - // operation at the target location using the value - // specified in the "from" member. - operation_add(ptr, v); - break; - } - - case patch_operations::test: - { - bool success = false; - JSON_TRY - { - // check if "value" matches the one at "path" - // the "path" location must exist - use at() - success = (result.at(ptr) == get_value("test", "value", false)); - } - JSON_CATCH (out_of_range&) - { - // ignore out of range errors: success remains false - } - - // throw an exception if test fails - if (JSON_UNLIKELY(not success)) - { - JSON_THROW(other_error::create(501, "unsuccessful: " + val.dump())); - } - - break; - } - - case patch_operations::invalid: - { - // op must be "add", "remove", "replace", "move", "copy", or - // "test" - JSON_THROW(parse_error::create(105, 0, "operation value '" + op + "' is invalid")); - } - } - } - - return result; - } - - /*! - @brief creates a diff as a JSON patch - - Creates a [JSON Patch](http://jsonpatch.com) so that value @a source can - be changed into the value @a target by calling @ref patch function. - - @invariant For two JSON values @a source and @a target, the following code - yields always `true`: - @code {.cpp} - source.patch(diff(source, target)) == target; - @endcode - - @note Currently, only `remove`, `add`, and `replace` operations are - generated. - - @param[in] source JSON value to compare from - @param[in] target JSON value to compare against - @param[in] path helper value to create JSON pointers - - @return a JSON patch to convert the @a source to @a target - - @complexity Linear in the lengths of @a source and @a target. - - @liveexample{The following code shows how a JSON patch is created as a - diff for two JSON values.,diff} - - @sa @ref patch -- apply a JSON patch - @sa @ref merge_patch -- apply a JSON Merge Patch - - @sa [RFC 6902 (JSON Patch)](https://tools.ietf.org/html/rfc6902) - - @since version 2.0.0 - */ - static basic_json diff(const basic_json& source, const basic_json& target, - const std::string& path = "") - { - // the patch - basic_json result(value_t::array); - - // if the values are the same, return empty patch - if (source == target) - { - return result; - } - - if (source.type() != target.type()) - { - // different types: replace value - result.push_back( - { - {"op", "replace"}, {"path", path}, {"value", target} - }); - } - else - { - switch (source.type()) - { - case value_t::array: - { - // first pass: traverse common elements - std::size_t i = 0; - while (i < source.size() and i < target.size()) - { - // recursive call to compare array values at index i - auto temp_diff = diff(source[i], target[i], path + "/" + std::to_string(i)); - result.insert(result.end(), temp_diff.begin(), temp_diff.end()); - ++i; - } - - // i now reached the end of at least one array - // in a second pass, traverse the remaining elements - - // remove my remaining elements - const auto end_index = static_cast<difference_type>(result.size()); - while (i < source.size()) - { - // add operations in reverse order to avoid invalid - // indices - result.insert(result.begin() + end_index, object( - { - {"op", "remove"}, - {"path", path + "/" + std::to_string(i)} - })); - ++i; - } - - // add other remaining elements - while (i < target.size()) - { - result.push_back( - { - {"op", "add"}, - {"path", path + "/" + std::to_string(i)}, - {"value", target[i]} - }); - ++i; - } - - break; - } - - case value_t::object: - { - // first pass: traverse this object's elements - for (auto it = source.cbegin(); it != source.cend(); ++it) - { - // escape the key name to be used in a JSON patch - const auto key = json_pointer::escape(it.key()); - - if (target.find(it.key()) != target.end()) - { - // recursive call to compare object values at key it - auto temp_diff = diff(it.value(), target[it.key()], path + "/" + key); - result.insert(result.end(), temp_diff.begin(), temp_diff.end()); - } - else - { - // found a key that is not in o -> remove it - result.push_back(object( - { - {"op", "remove"}, {"path", path + "/" + key} - })); - } - } - - // second pass: traverse other object's elements - for (auto it = target.cbegin(); it != target.cend(); ++it) - { - if (source.find(it.key()) == source.end()) - { - // found a key that is not in this -> add it - const auto key = json_pointer::escape(it.key()); - result.push_back( - { - {"op", "add"}, {"path", path + "/" + key}, - {"value", it.value()} - }); - } - } - - break; - } - - default: - { - // both primitive type: replace value - result.push_back( - { - {"op", "replace"}, {"path", path}, {"value", target} - }); - break; - } - } - } - - return result; - } - - /// @} - - //////////////////////////////// - // JSON Merge Patch functions // - //////////////////////////////// - - /// @name JSON Merge Patch functions - /// @{ - - /*! - @brief applies a JSON Merge Patch - - The merge patch format is primarily intended for use with the HTTP PATCH - method as a means of describing a set of modifications to a target - resource's content. This function applies a merge patch to the current - JSON value. - - The function implements the following algorithm from Section 2 of - [RFC 7396 (JSON Merge Patch)](https://tools.ietf.org/html/rfc7396): - - ``` - define MergePatch(Target, Patch): - if Patch is an Object: - if Target is not an Object: - Target = {} // Ignore the contents and set it to an empty Object - for each Name/Value pair in Patch: - if Value is null: - if Name exists in Target: - remove the Name/Value pair from Target - else: - Target[Name] = MergePatch(Target[Name], Value) - return Target - else: - return Patch - ``` - - Thereby, `Target` is the current object; that is, the patch is applied to - the current value. - - @param[in] patch the patch to apply - - @complexity Linear in the lengths of @a patch. - - @liveexample{The following code shows how a JSON Merge Patch is applied to - a JSON document.,merge_patch} - - @sa @ref patch -- apply a JSON patch - @sa [RFC 7396 (JSON Merge Patch)](https://tools.ietf.org/html/rfc7396) - - @since version 3.0.0 - */ - void merge_patch(const basic_json& patch) - { - if (patch.is_object()) - { - if (not is_object()) - { - *this = object(); - } - for (auto it = patch.begin(); it != patch.end(); ++it) - { - if (it.value().is_null()) - { - erase(it.key()); - } - else - { - operator[](it.key()).merge_patch(it.value()); - } - } - } - else - { - *this = patch; - } - } - - /// @} -}; -} // namespace nlohmann - -/////////////////////// -// nonmember support // -/////////////////////// - -// specialization of std::swap, and std::hash -namespace std -{ -/*! -@brief exchanges the values of two JSON objects - -@since version 1.0.0 -*/ -template<> -inline void swap(nlohmann::json& j1, - nlohmann::json& j2) noexcept( - is_nothrow_move_constructible<nlohmann::json>::value and - is_nothrow_move_assignable<nlohmann::json>::value - ) -{ - j1.swap(j2); -} - -/// hash value for JSON objects -template<> -struct hash<nlohmann::json> -{ - /*! - @brief return a hash value for a JSON object - - @since version 1.0.0 - */ - std::size_t operator()(const nlohmann::json& j) const - { - // a naive hashing via the string representation - const auto& h = hash<nlohmann::json::string_t>(); - return h(j.dump()); - } -}; - -/// specialization for std::less<value_t> -/// @note: do not remove the space after '<', -/// see https://github.com/nlohmann/json/pull/679 -template<> -struct less< ::nlohmann::detail::value_t> -{ - /*! - @brief compare two value_t enum values - @since version 3.0.0 - */ - bool operator()(nlohmann::detail::value_t lhs, - nlohmann::detail::value_t rhs) const noexcept - { - return nlohmann::detail::operator<(lhs, rhs); - } -}; - -} // namespace std - -/*! -@brief user-defined string literal for JSON values - -This operator implements a user-defined string literal for JSON objects. It -can be used by adding `"_json"` to a string literal and returns a JSON object -if no parse error occurred. - -@param[in] s a string representation of a JSON object -@param[in] n the length of string @a s -@return a JSON object - -@since version 1.0.0 -*/ -inline nlohmann::json operator "" _json(const char* s, std::size_t n) -{ - return nlohmann::json::parse(s, s + n); -} - -/*! -@brief user-defined string literal for JSON pointer - -This operator implements a user-defined string literal for JSON Pointers. It -can be used by adding `"_json_pointer"` to a string literal and returns a JSON pointer -object if no parse error occurred. - -@param[in] s a string representation of a JSON Pointer -@param[in] n the length of string @a s -@return a JSON pointer object - -@since version 2.0.0 -*/ -inline nlohmann::json::json_pointer operator "" _json_pointer(const char* s, std::size_t n) -{ - return nlohmann::json::json_pointer(std::string(s, n)); -} - -// #include <nlohmann/detail/macro_unscope.hpp> - - -// restore GCC/clang diagnostic settings -#if defined(__clang__) || defined(__GNUC__) || defined(__GNUG__) - #pragma GCC diagnostic pop -#endif -#if defined(__clang__) - #pragma GCC diagnostic pop -#endif - -// clean up -#undef JSON_CATCH -#undef JSON_THROW -#undef JSON_TRY -#undef JSON_LIKELY -#undef JSON_UNLIKELY -#undef JSON_DEPRECATED -#undef JSON_HAS_CPP_14 -#undef JSON_HAS_CPP_17 -#undef NLOHMANN_BASIC_JSON_TPL_DECLARATION -#undef NLOHMANN_BASIC_JSON_TPL -#undef NLOHMANN_JSON_HAS_HELPER - - -#endif diff --git a/externals/lz4 b/externals/lz4 deleted file mode 160000 -Subproject 4db65c1d99280a757823914efaa96d2e87f4184 diff --git a/externals/opus/CMakeLists.txt b/externals/opus/CMakeLists.txt deleted file mode 100644 index 94a86551f..000000000 --- a/externals/opus/CMakeLists.txt +++ /dev/null @@ -1,254 +0,0 @@ -cmake_minimum_required(VERSION 3.8) - -project(opus) - -option(OPUS_STACK_PROTECTOR "Use stack protection" OFF) -option(OPUS_USE_ALLOCA "Use alloca for stack arrays (on non-C99 compilers)" OFF) -option(OPUS_CUSTOM_MODES "Enable non-Opus modes, e.g. 44.1 kHz & 2^n frames" OFF) -option(OPUS_FIXED_POINT "Compile as fixed-point (for machines without a fast enough FPU)" OFF) -option(OPUS_ENABLE_FLOAT_API "Compile with the floating point API (for machines with float library" ON) - -include(opus/opus_functions.cmake) - -if(OPUS_STACK_PROTECTOR) - if(NOT MSVC) # GC on by default on MSVC - check_and_set_flag(STACK_PROTECTION_STRONG -fstack-protector-strong) - endif() -else() - if(MSVC) - check_and_set_flag(BUFFER_SECURITY_CHECK /GS-) - endif() -endif() - -add_library(opus STATIC - # CELT sources - opus/celt/bands.c - opus/celt/celt.c - opus/celt/celt_decoder.c - opus/celt/celt_encoder.c - opus/celt/celt_lpc.c - opus/celt/cwrs.c - opus/celt/entcode.c - opus/celt/entdec.c - opus/celt/entenc.c - opus/celt/kiss_fft.c - opus/celt/laplace.c - opus/celt/mathops.c - opus/celt/mdct.c - opus/celt/modes.c - opus/celt/pitch.c - opus/celt/quant_bands.c - opus/celt/rate.c - opus/celt/vq.c - - # SILK sources - opus/silk/A2NLSF.c - opus/silk/CNG.c - opus/silk/HP_variable_cutoff.c - opus/silk/LPC_analysis_filter.c - opus/silk/LPC_fit.c - opus/silk/LPC_inv_pred_gain.c - opus/silk/LP_variable_cutoff.c - opus/silk/NLSF2A.c - opus/silk/NLSF_VQ.c - opus/silk/NLSF_VQ_weights_laroia.c - opus/silk/NLSF_decode.c - opus/silk/NLSF_del_dec_quant.c - opus/silk/NLSF_encode.c - opus/silk/NLSF_stabilize.c - opus/silk/NLSF_unpack.c - opus/silk/NSQ.c - opus/silk/NSQ_del_dec.c - opus/silk/PLC.c - opus/silk/VAD.c - opus/silk/VQ_WMat_EC.c - opus/silk/ana_filt_bank_1.c - opus/silk/biquad_alt.c - opus/silk/bwexpander.c - opus/silk/bwexpander_32.c - opus/silk/check_control_input.c - opus/silk/code_signs.c - opus/silk/control_SNR.c - opus/silk/control_audio_bandwidth.c - opus/silk/control_codec.c - opus/silk/dec_API.c - opus/silk/decode_core.c - opus/silk/decode_frame.c - opus/silk/decode_indices.c - opus/silk/decode_parameters.c - opus/silk/decode_pitch.c - opus/silk/decode_pulses.c - opus/silk/decoder_set_fs.c - opus/silk/enc_API.c - opus/silk/encode_indices.c - opus/silk/encode_pulses.c - opus/silk/gain_quant.c - opus/silk/init_decoder.c - opus/silk/init_encoder.c - opus/silk/inner_prod_aligned.c - opus/silk/interpolate.c - opus/silk/lin2log.c - opus/silk/log2lin.c - opus/silk/pitch_est_tables.c - opus/silk/process_NLSFs.c - opus/silk/quant_LTP_gains.c - opus/silk/resampler.c - opus/silk/resampler_down2.c - opus/silk/resampler_down2_3.c - opus/silk/resampler_private_AR2.c - opus/silk/resampler_private_IIR_FIR.c - opus/silk/resampler_private_down_FIR.c - opus/silk/resampler_private_up2_HQ.c - opus/silk/resampler_rom.c - opus/silk/shell_coder.c - opus/silk/sigm_Q15.c - opus/silk/sort.c - opus/silk/stereo_LR_to_MS.c - opus/silk/stereo_MS_to_LR.c - opus/silk/stereo_decode_pred.c - opus/silk/stereo_encode_pred.c - opus/silk/stereo_find_predictor.c - opus/silk/stereo_quant_pred.c - opus/silk/sum_sqr_shift.c - opus/silk/table_LSF_cos.c - opus/silk/tables_LTP.c - opus/silk/tables_NLSF_CB_NB_MB.c - opus/silk/tables_NLSF_CB_WB.c - opus/silk/tables_gain.c - opus/silk/tables_other.c - opus/silk/tables_pitch_lag.c - opus/silk/tables_pulses_per_block.c - - # Opus sources - opus/src/analysis.c - opus/src/mapping_matrix.c - opus/src/mlp.c - opus/src/mlp_data.c - opus/src/opus.c - opus/src/opus_decoder.c - opus/src/opus_encoder.c - opus/src/opus_multistream.c - opus/src/opus_multistream_decoder.c - opus/src/opus_multistream_encoder.c - opus/src/opus_projection_decoder.c - opus/src/opus_projection_encoder.c - opus/src/repacketizer.c -) - -if (DEBUG) - target_sources(opus PRIVATE opus/silk/debug.c) -endif() - -if (OPUS_FIXED_POINT) - target_sources(opus PRIVATE - opus/silk/fixed/LTP_analysis_filter_FIX.c - opus/silk/fixed/LTP_scale_ctrl_FIX.c - opus/silk/fixed/apply_sine_window_FIX.c - opus/silk/fixed/autocorr_FIX.c - opus/silk/fixed/burg_modified_FIX.c - opus/silk/fixed/corrMatrix_FIX.c - opus/silk/fixed/encode_frame_FIX.c - opus/silk/fixed/find_LPC_FIX.c - opus/silk/fixed/find_LTP_FIX.c - opus/silk/fixed/find_pitch_lags_FIX.c - opus/silk/fixed/find_pred_coefs_FIX.c - opus/silk/fixed/k2a_FIX.c - opus/silk/fixed/k2a_Q16_FIX.c - opus/silk/fixed/noise_shape_analysis_FIX.c - opus/silk/fixed/pitch_analysis_core_FIX.c - opus/silk/fixed/prefilter_FIX.c - opus/silk/fixed/process_gains_FIX.c - opus/silk/fixed/regularize_correlations_FIX.c - opus/silk/fixed/residual_energy16_FIX.c - opus/silk/fixed/residual_energy_FIX.c - opus/silk/fixed/schur64_FIX.c - opus/silk/fixed/schur_FIX.c - opus/silk/fixed/solve_LS_FIX.c - opus/silk/fixed/vector_ops_FIX.c - opus/silk/fixed/warped_autocorrelation_FIX.c - ) -else() - target_sources(opus PRIVATE - opus/silk/float/LPC_analysis_filter_FLP.c - opus/silk/float/LPC_inv_pred_gain_FLP.c - opus/silk/float/LTP_analysis_filter_FLP.c - opus/silk/float/LTP_scale_ctrl_FLP.c - opus/silk/float/apply_sine_window_FLP.c - opus/silk/float/autocorrelation_FLP.c - opus/silk/float/burg_modified_FLP.c - opus/silk/float/bwexpander_FLP.c - opus/silk/float/corrMatrix_FLP.c - opus/silk/float/encode_frame_FLP.c - opus/silk/float/energy_FLP.c - opus/silk/float/find_LPC_FLP.c - opus/silk/float/find_LTP_FLP.c - opus/silk/float/find_pitch_lags_FLP.c - opus/silk/float/find_pred_coefs_FLP.c - opus/silk/float/inner_product_FLP.c - opus/silk/float/k2a_FLP.c - opus/silk/float/noise_shape_analysis_FLP.c - opus/silk/float/pitch_analysis_core_FLP.c - opus/silk/float/process_gains_FLP.c - opus/silk/float/regularize_correlations_FLP.c - opus/silk/float/residual_energy_FLP.c - opus/silk/float/scale_copy_vector_FLP.c - opus/silk/float/scale_vector_FLP.c - opus/silk/float/schur_FLP.c - opus/silk/float/sort_FLP.c - opus/silk/float/warped_autocorrelation_FLP.c - opus/silk/float/wrappers_FLP.c - ) -endif() - -target_compile_definitions(opus PRIVATE OPUS_BUILD ENABLE_HARDENING) - -if(NOT MSVC) - if(MINGW) - target_compile_definitions(opus PRIVATE _FORTIFY_SOURCE=0) - else() - target_compile_definitions(opus PRIVATE _FORTIFY_SOURCE=2) - endif() -endif() - -# It is strongly recommended to uncomment one of these VAR_ARRAYS: Use C99 -# variable-length arrays for stack allocation USE_ALLOCA: Use alloca() for stack -# allocation If none is defined, then the fallback is a non-threadsafe global -# array -if(OPUS_USE_ALLOCA OR MSVC) - target_compile_definitions(opus PRIVATE USE_ALLOCA) -else() - target_compile_definitions(opus PRIVATE VAR_ARRAYS) -endif() - -if(OPUS_CUSTOM_MODES) - target_compile_definitions(opus PRIVATE CUSTOM_MODES) -endif() - -if(NOT OPUS_ENABLE_FLOAT_API) - target_compile_definitions(opus PRIVATE DISABLE_FLOAT_API) -endif() - -target_compile_definitions(opus -PUBLIC - -DOPUS_VERSION="\\"1.3.1\\"" - -PRIVATE - # Use C99 intrinsics to speed up float-to-int conversion - HAVE_LRINTF -) - -if (FIXED_POINT) - target_compile_definitions(opus PRIVATE -DFIXED_POINT=1 -DDISABLE_FLOAT_API) -endif() - -target_include_directories(opus -PUBLIC - opus/include - -PRIVATE - opus/celt - opus/silk - opus/silk/fixed - opus/silk/float - opus/src -) diff --git a/externals/opus/opus b/externals/opus/opus deleted file mode 160000 -Subproject ad8fe90db79b7d2a135e3dfd2ed6631b0c5662a diff --git a/externals/sirit b/externals/sirit -Subproject 414fc4dbd28d8fe48f735a0c389db8a234f733c +Subproject eefca56afd49379bdebc97ded8b480839f93088 diff --git a/externals/xbyak b/externals/xbyak new file mode 160000 +Subproject 82b70e665918efc2ee348091742fd0237b3b68c diff --git a/externals/zlib/CMakeLists.txt b/externals/zlib/CMakeLists.txt deleted file mode 100644 index 0ca32aae4..000000000 --- a/externals/zlib/CMakeLists.txt +++ /dev/null @@ -1,81 +0,0 @@ -project(zlib C) - -include(CheckTypeSize) -include(CheckFunctionExists) -include(CheckIncludeFile) - -check_include_file(sys/types.h HAVE_SYS_TYPES_H) -check_include_file(stdint.h HAVE_STDINT_H) -check_include_file(stddef.h HAVE_STDDEF_H) - -# Check to see if we have large file support -set(CMAKE_REQUIRED_DEFINITIONS -D_LARGEFILE64_SOURCE=1) -# We add these other definitions here because CheckTypeSize.cmake -# in CMake 2.4.x does not automatically do so and we want -# compatibility with CMake 2.4.x. -if(HAVE_SYS_TYPES_H) - list(APPEND CMAKE_REQUIRED_DEFINITIONS -DHAVE_SYS_TYPES_H) -endif() -if(HAVE_STDINT_H) - list(APPEND CMAKE_REQUIRED_DEFINITIONS -DHAVE_STDINT_H) -endif() -if(HAVE_STDDEF_H) - list(APPEND CMAKE_REQUIRED_DEFINITIONS -DHAVE_STDDEF_H) -endif() -check_type_size(off64_t OFF64_T) -if(HAVE_OFF64_T) - add_definitions(-D_LARGEFILE64_SOURCE=1) -endif() -set(CMAKE_REQUIRED_DEFINITIONS) # clear variable - -# Check for fseeko -check_function_exists(fseeko HAVE_FSEEKO) -if(NOT HAVE_FSEEKO) - add_definitions(-DNO_FSEEKO) -endif() - -# Check for unistd.h -check_include_file(unistd.h HAVE_UNISTD_H) -if(HAVE_UNISTD_H) - add_definitions(-DHAVE_UNISTD_H) -endif() - -if(MSVC) - add_definitions(-D_CRT_SECURE_NO_DEPRECATE) - add_definitions(-D_CRT_NONSTDC_NO_DEPRECATE) -endif() - -add_library(z STATIC - zlib/adler32.c - zlib/compress.c - zlib/crc32.c - zlib/crc32.h - zlib/deflate.c - zlib/deflate.h - zlib/gzclose.c - zlib/gzguts.h - zlib/gzlib.c - zlib/gzread.c - zlib/gzwrite.c - zlib/inffast.h - zlib/inffixed.h - zlib/inflate.c - zlib/inflate.h - zlib/infback.c - zlib/inftrees.c - zlib/inftrees.h - zlib/inffast.c - zlib/trees.c - zlib/trees.h - zlib/uncompr.c - zlib/zconf.h - zlib/zlib.h - zlib/zutil.c - zlib/zutil.h -) -add_library(ZLIB::ZLIB ALIAS z) - -target_include_directories(z -PUBLIC - zlib/ -) diff --git a/externals/zlib/zlib b/externals/zlib/zlib deleted file mode 160000 -Subproject cacf7f1d4e3d44d871b605da3b647f07d718623 diff --git a/externals/zstd b/externals/zstd deleted file mode 160000 -Subproject 470344d33e1d52a2ada75d278466da8d4ee2faf diff --git a/src/CMakeLists.txt b/src/CMakeLists.txt index 3a57356ab..1e977e8a8 100644 --- a/src/CMakeLists.txt +++ b/src/CMakeLists.txt @@ -62,6 +62,10 @@ else() -Wno-unused-parameter ) + if (ARCHITECTURE_x86_64) + add_compile_options("-mcx16") + endif() + if (APPLE AND CMAKE_CXX_COMPILER_ID STREQUAL Clang) add_compile_options("-stdlib=libc++") endif() diff --git a/src/audio_core/audio_renderer.cpp b/src/audio_core/audio_renderer.cpp index d18ef6940..d64452617 100644 --- a/src/audio_core/audio_renderer.cpp +++ b/src/audio_core/audio_renderer.cpp @@ -17,7 +17,7 @@ namespace AudioCore { constexpr u32 STREAM_SAMPLE_RATE{48000}; constexpr u32 STREAM_NUM_CHANNELS{2}; - +using VoiceChannelHolder = std::array<VoiceResourceInformation*, 6>; class AudioRenderer::VoiceState { public: bool IsPlaying() const { @@ -37,9 +37,10 @@ public: } void SetWaveIndex(std::size_t index); - std::vector<s16> DequeueSamples(std::size_t sample_count, Core::Memory::Memory& memory); + std::vector<s16> DequeueSamples(std::size_t sample_count, Core::Memory::Memory& memory, + const VoiceChannelHolder& voice_resources); void UpdateState(); - void RefreshBuffer(Core::Memory::Memory& memory); + void RefreshBuffer(Core::Memory::Memory& memory, const VoiceChannelHolder& voice_resources); private: bool is_in_use{}; @@ -79,7 +80,7 @@ AudioRenderer::AudioRenderer(Core::Timing::CoreTiming& core_timing, Core::Memory std::shared_ptr<Kernel::WritableEvent> buffer_event, std::size_t instance_number) : worker_params{params}, buffer_event{buffer_event}, voices(params.voice_count), - effects(params.effect_count), memory{memory_} { + voice_resources(params.voice_count), effects(params.effect_count), memory{memory_} { behavior_info.SetUserRevision(params.revision); audio_out = std::make_unique<AudioCore::AudioOut>(); stream = audio_out->OpenStream(core_timing, STREAM_SAMPLE_RATE, STREAM_NUM_CHANNELS, @@ -127,6 +128,12 @@ ResultVal<std::vector<u8>> AudioRenderer::UpdateAudioRenderer(const std::vector< input_params.data() + sizeof(UpdateDataHeader) + config.behavior_size, memory_pool_count * sizeof(MemoryPoolInfo)); + // Copy voice resources + const std::size_t voice_resource_offset{sizeof(UpdateDataHeader) + config.behavior_size + + config.memory_pools_size}; + std::memcpy(voice_resources.data(), input_params.data() + voice_resource_offset, + sizeof(VoiceResourceInformation) * voice_resources.size()); + // Copy VoiceInfo structs std::size_t voice_offset{sizeof(UpdateDataHeader) + config.behavior_size + config.memory_pools_size + config.voice_resource_size}; @@ -173,11 +180,12 @@ ResultVal<std::vector<u8>> AudioRenderer::UpdateAudioRenderer(const std::vector< // Copy output header UpdateDataHeader response_data{worker_params}; - std::vector<u8> output_params(response_data.total_size); if (behavior_info.IsElapsedFrameCountSupported()) { - response_data.frame_count = 0x10; - response_data.total_size += 0x10; + response_data.render_info = sizeof(RendererInfo); + response_data.total_size += sizeof(RendererInfo); } + + std::vector<u8> output_params(response_data.total_size); std::memcpy(output_params.data(), &response_data, sizeof(UpdateDataHeader)); // Copy output memory pool entries @@ -212,6 +220,17 @@ ResultVal<std::vector<u8>> AudioRenderer::UpdateAudioRenderer(const std::vector< return Audren::ERR_INVALID_PARAMETERS; } + if (behavior_info.IsElapsedFrameCountSupported()) { + const std::size_t renderer_info_offset{ + sizeof(UpdateDataHeader) + response_data.memory_pools_size + response_data.voices_size + + response_data.effects_size + response_data.sinks_size + + response_data.performance_manager_size + response_data.behavior_size}; + RendererInfo renderer_info{}; + renderer_info.elasped_frame_count = elapsed_frame_count; + std::memcpy(output_params.data() + renderer_info_offset, &renderer_info, + sizeof(RendererInfo)); + } + return MakeResult(output_params); } @@ -220,14 +239,15 @@ void AudioRenderer::VoiceState::SetWaveIndex(std::size_t index) { is_refresh_pending = true; } -std::vector<s16> AudioRenderer::VoiceState::DequeueSamples(std::size_t sample_count, - Core::Memory::Memory& memory) { +std::vector<s16> AudioRenderer::VoiceState::DequeueSamples( + std::size_t sample_count, Core::Memory::Memory& memory, + const VoiceChannelHolder& voice_resources) { if (!IsPlaying()) { return {}; } if (is_refresh_pending) { - RefreshBuffer(memory); + RefreshBuffer(memory, voice_resources); } const std::size_t max_size{samples.size() - offset}; @@ -271,7 +291,8 @@ void AudioRenderer::VoiceState::UpdateState() { is_in_use = info.is_in_use; } -void AudioRenderer::VoiceState::RefreshBuffer(Core::Memory::Memory& memory) { +void AudioRenderer::VoiceState::RefreshBuffer(Core::Memory::Memory& memory, + const VoiceChannelHolder& voice_resources) { const auto wave_buffer_address = info.wave_buffer[wave_index].buffer_addr; const auto wave_buffer_size = info.wave_buffer[wave_index].buffer_sz; std::vector<s16> new_samples(wave_buffer_size / sizeof(s16)); @@ -296,17 +317,77 @@ void AudioRenderer::VoiceState::RefreshBuffer(Core::Memory::Memory& memory) { } switch (info.channel_count) { - case 1: + case 1: { // 1 channel is upsampled to 2 channel samples.resize(new_samples.size() * 2); + for (std::size_t index = 0; index < new_samples.size(); ++index) { - samples[index * 2] = new_samples[index]; - samples[index * 2 + 1] = new_samples[index]; + auto sample = static_cast<float>(new_samples[index]); + if (voice_resources[0]->in_use) { + sample *= voice_resources[0]->mix_volumes[0]; + } + + samples[index * 2] = static_cast<s16>(sample * info.volume); + samples[index * 2 + 1] = static_cast<s16>(sample * info.volume); } break; + } case 2: { // 2 channel is played as is samples = std::move(new_samples); + const std::size_t sample_count = (samples.size() / 2); + for (std::size_t index = 0; index < sample_count; ++index) { + const std::size_t index_l = index * 2; + const std::size_t index_r = index * 2 + 1; + + auto sample_l = static_cast<float>(samples[index_l]); + auto sample_r = static_cast<float>(samples[index_r]); + + if (voice_resources[0]->in_use) { + sample_l *= voice_resources[0]->mix_volumes[0]; + } + + if (voice_resources[1]->in_use) { + sample_r *= voice_resources[1]->mix_volumes[1]; + } + + samples[index_l] = static_cast<s16>(sample_l * info.volume); + samples[index_r] = static_cast<s16>(sample_r * info.volume); + } + break; + } + case 6: { + samples.resize((new_samples.size() / 6) * 2); + const std::size_t sample_count = samples.size() / 2; + + for (std::size_t index = 0; index < sample_count; ++index) { + auto FL = static_cast<float>(new_samples[index * 6]); + auto FR = static_cast<float>(new_samples[index * 6 + 1]); + auto FC = static_cast<float>(new_samples[index * 6 + 2]); + auto BL = static_cast<float>(new_samples[index * 6 + 4]); + auto BR = static_cast<float>(new_samples[index * 6 + 5]); + + if (voice_resources[0]->in_use) { + FL *= voice_resources[0]->mix_volumes[0]; + } + if (voice_resources[1]->in_use) { + FR *= voice_resources[1]->mix_volumes[1]; + } + if (voice_resources[2]->in_use) { + FC *= voice_resources[2]->mix_volumes[2]; + } + if (voice_resources[4]->in_use) { + BL *= voice_resources[4]->mix_volumes[4]; + } + if (voice_resources[5]->in_use) { + BR *= voice_resources[5]->mix_volumes[5]; + } + + samples[index * 2] = + static_cast<s16>((0.3694f * FL + 0.2612f * FC + 0.3694f * BL) * info.volume); + samples[index * 2 + 1] = + static_cast<s16>((0.3694f * FR + 0.2612f * FC + 0.3694f * BR) * info.volume); + } break; } default: @@ -352,11 +433,17 @@ void AudioRenderer::QueueMixedBuffer(Buffer::Tag tag) { if (!voice.IsPlaying()) { continue; } + VoiceChannelHolder resources{}; + for (u32 channel = 0; channel < voice.GetInfo().channel_count; channel++) { + const auto channel_resource_id = voice.GetInfo().voice_channel_resource_ids[channel]; + resources[channel] = &voice_resources[channel_resource_id]; + } std::size_t offset{}; s64 samples_remaining{BUFFER_SIZE}; while (samples_remaining > 0) { - const std::vector<s16> samples{voice.DequeueSamples(samples_remaining, memory)}; + const std::vector<s16> samples{ + voice.DequeueSamples(samples_remaining, memory, resources)}; if (samples.empty()) { break; @@ -372,6 +459,7 @@ void AudioRenderer::QueueMixedBuffer(Buffer::Tag tag) { } } audio_out->QueueBuffer(stream, tag, std::move(buffer)); + elapsed_frame_count++; } void AudioRenderer::ReleaseAndQueueBuffers() { diff --git a/src/audio_core/audio_renderer.h b/src/audio_core/audio_renderer.h index b42770fae..f0b691a86 100644 --- a/src/audio_core/audio_renderer.h +++ b/src/audio_core/audio_renderer.h @@ -9,6 +9,7 @@ #include <vector> #include "audio_core/behavior_info.h" +#include "audio_core/common.h" #include "audio_core/stream.h" #include "common/common_funcs.h" #include "common/common_types.h" @@ -116,6 +117,14 @@ struct WaveBuffer { }; static_assert(sizeof(WaveBuffer) == 0x38, "WaveBuffer has wrong size"); +struct VoiceResourceInformation { + s32_le id{}; + std::array<float_le, MAX_MIX_BUFFERS> mix_volumes{}; + bool in_use{}; + INSERT_PADDING_BYTES(11); +}; +static_assert(sizeof(VoiceResourceInformation) == 0x70, "VoiceResourceInformation has wrong size"); + struct VoiceInfo { u32_le id; u32_le node_id; @@ -187,6 +196,12 @@ struct EffectOutStatus { }; static_assert(sizeof(EffectOutStatus) == 0x10, "EffectOutStatus is an invalid size"); +struct RendererInfo { + u64_le elasped_frame_count{}; + INSERT_PADDING_WORDS(2); +}; +static_assert(sizeof(RendererInfo) == 0x10, "RendererInfo is an invalid size"); + struct UpdateDataHeader { UpdateDataHeader() {} @@ -200,7 +215,7 @@ struct UpdateDataHeader { mixes_size = 0x0; sinks_size = config.sink_count * 0x20; performance_manager_size = 0x10; - frame_count = 0; + render_info = 0; total_size = sizeof(UpdateDataHeader) + behavior_size + memory_pools_size + voices_size + effects_size + sinks_size + performance_manager_size; } @@ -214,8 +229,8 @@ struct UpdateDataHeader { u32_le mixes_size{}; u32_le sinks_size{}; u32_le performance_manager_size{}; - INSERT_PADDING_WORDS(1); - u32_le frame_count{}; + u32_le splitter_size{}; + u32_le render_info{}; INSERT_PADDING_WORDS(4); u32_le total_size{}; }; @@ -244,10 +259,12 @@ private: AudioRendererParameter worker_params; std::shared_ptr<Kernel::WritableEvent> buffer_event; std::vector<VoiceState> voices; + std::vector<VoiceResourceInformation> voice_resources; std::vector<EffectState> effects; std::unique_ptr<AudioOut> audio_out; StreamPtr stream; Core::Memory::Memory& memory; + std::size_t elapsed_frame_count{}; }; } // namespace AudioCore diff --git a/src/audio_core/common.h b/src/audio_core/common.h index 98478b66b..7bb145c53 100644 --- a/src/audio_core/common.h +++ b/src/audio_core/common.h @@ -14,6 +14,7 @@ constexpr ResultCode ERR_INVALID_PARAMETERS{ErrorModule::Audio, 41}; } constexpr u32_le CURRENT_PROCESS_REVISION = Common::MakeMagic('R', 'E', 'V', '8'); +constexpr std::size_t MAX_MIX_BUFFERS = 24; static constexpr u32 VersionFromRevision(u32_le rev) { // "REV7" -> 7 diff --git a/src/audio_core/stream.cpp b/src/audio_core/stream.cpp index 4ca98f8ea..dfc4805d9 100644 --- a/src/audio_core/stream.cpp +++ b/src/audio_core/stream.cpp @@ -59,15 +59,24 @@ Stream::State Stream::GetState() const { return state; } -s64 Stream::GetBufferReleaseCycles(const Buffer& buffer) const { +s64 Stream::GetBufferReleaseNS(const Buffer& buffer) const { const std::size_t num_samples{buffer.GetSamples().size() / GetNumChannels()}; - const auto us = - std::chrono::microseconds((static_cast<u64>(num_samples) * 1000000) / sample_rate); - return Core::Timing::usToCycles(us); + const auto ns = + std::chrono::nanoseconds((static_cast<u64>(num_samples) * 1000000000ULL) / sample_rate); + return ns.count(); +} + +s64 Stream::GetBufferReleaseNSHostTiming(const Buffer& buffer) const { + const std::size_t num_samples{buffer.GetSamples().size() / GetNumChannels()}; + /// DSP signals before playing the last sample, in HLE we emulate this in this way + s64 base_samples = std::max<s64>(static_cast<s64>(num_samples) - 1, 0); + const auto ns = + std::chrono::nanoseconds((static_cast<u64>(base_samples) * 1000000000ULL) / sample_rate); + return ns.count(); } static void VolumeAdjustSamples(std::vector<s16>& samples, float game_volume) { - const float volume{std::clamp(Settings::values.volume - (1.0f - game_volume), 0.0f, 1.0f)}; + const float volume{std::clamp(Settings::Volume() - (1.0f - game_volume), 0.0f, 1.0f)}; if (volume == 1.0f) { return; @@ -105,7 +114,11 @@ void Stream::PlayNextBuffer() { sink_stream.EnqueueSamples(GetNumChannels(), active_buffer->GetSamples()); - core_timing.ScheduleEvent(GetBufferReleaseCycles(*active_buffer), release_event, {}); + if (core_timing.IsHostTiming()) { + core_timing.ScheduleEvent(GetBufferReleaseNSHostTiming(*active_buffer), release_event, {}); + } else { + core_timing.ScheduleEvent(GetBufferReleaseNS(*active_buffer), release_event, {}); + } } void Stream::ReleaseActiveBuffer() { diff --git a/src/audio_core/stream.h b/src/audio_core/stream.h index 1708a4d98..e309d60fe 100644 --- a/src/audio_core/stream.h +++ b/src/audio_core/stream.h @@ -96,7 +96,10 @@ private: void ReleaseActiveBuffer(); /// Gets the number of core cycles when the specified buffer will be released - s64 GetBufferReleaseCycles(const Buffer& buffer) const; + s64 GetBufferReleaseNS(const Buffer& buffer) const; + + /// Gets the number of core cycles when the specified buffer will be released + s64 GetBufferReleaseNSHostTiming(const Buffer& buffer) const; u32 sample_rate; ///< Sample rate of the stream Format format; ///< Format of the stream diff --git a/src/common/CMakeLists.txt b/src/common/CMakeLists.txt index 6ffc612e7..d120c8d3d 100644 --- a/src/common/CMakeLists.txt +++ b/src/common/CMakeLists.txt @@ -32,6 +32,8 @@ add_custom_command(OUTPUT scm_rev.cpp DEPENDS # WARNING! It was too much work to try and make a common location for this list, # so if you need to change it, please update CMakeModules/GenerateSCMRev.cmake as well + "${VIDEO_CORE}/renderer_opengl/gl_arb_decompiler.cpp" + "${VIDEO_CORE}/renderer_opengl/gl_arb_decompiler.h" "${VIDEO_CORE}/renderer_opengl/gl_shader_cache.cpp" "${VIDEO_CORE}/renderer_opengl/gl_shader_cache.h" "${VIDEO_CORE}/renderer_opengl/gl_shader_decompiler.cpp" @@ -96,6 +98,8 @@ add_library(common STATIC algorithm.h alignment.h assert.h + atomic_ops.cpp + atomic_ops.h detached_tasks.cpp detached_tasks.h bit_field.h @@ -108,6 +112,8 @@ add_library(common STATIC common_types.h dynamic_library.cpp dynamic_library.h + fiber.cpp + fiber.h file_util.cpp file_util.h hash.h @@ -123,6 +129,8 @@ add_library(common STATIC lz4_compression.cpp lz4_compression.h math_util.h + memory_detect.cpp + memory_detect.h memory_hook.cpp memory_hook.h microprofile.cpp @@ -139,6 +147,8 @@ add_library(common STATIC scm_rev.cpp scm_rev.h scope_exit.h + spin_lock.cpp + spin_lock.h string_util.cpp string_util.h swap.h @@ -148,6 +158,8 @@ add_library(common STATIC thread.h thread_queue_list.h threadsafe_queue.h + time_zone.cpp + time_zone.h timer.cpp timer.h uint128.cpp @@ -157,6 +169,8 @@ add_library(common STATIC vector_math.h virtual_buffer.cpp virtual_buffer.h + wall_clock.cpp + wall_clock.h web_result.h zstd_compression.cpp zstd_compression.h @@ -167,10 +181,15 @@ if(ARCHITECTURE_x86_64) PRIVATE x64/cpu_detect.cpp x64/cpu_detect.h + x64/native_clock.cpp + x64/native_clock.h + x64/xbyak_abi.h + x64/xbyak_util.h ) endif() create_target_directory_groups(common) +find_package(Boost 1.71 COMPONENTS context headers REQUIRED) -target_link_libraries(common PUBLIC Boost::boost fmt microprofile) -target_link_libraries(common PRIVATE lz4_static libzstd_static) +target_link_libraries(common PUBLIC ${Boost_LIBRARIES} fmt::fmt microprofile) +target_link_libraries(common PRIVATE lz4::lz4 zstd::zstd xbyak) diff --git a/src/common/atomic_ops.cpp b/src/common/atomic_ops.cpp new file mode 100644 index 000000000..1098e21ff --- /dev/null +++ b/src/common/atomic_ops.cpp @@ -0,0 +1,70 @@ +// Copyright 2020 yuzu Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include <cstring> + +#include "common/atomic_ops.h" + +#if _MSC_VER +#include <intrin.h> +#endif + +namespace Common { + +#if _MSC_VER + +bool AtomicCompareAndSwap(u8 volatile* pointer, u8 value, u8 expected) { + u8 result = _InterlockedCompareExchange8((char*)pointer, value, expected); + return result == expected; +} + +bool AtomicCompareAndSwap(u16 volatile* pointer, u16 value, u16 expected) { + u16 result = _InterlockedCompareExchange16((short*)pointer, value, expected); + return result == expected; +} + +bool AtomicCompareAndSwap(u32 volatile* pointer, u32 value, u32 expected) { + u32 result = _InterlockedCompareExchange((long*)pointer, value, expected); + return result == expected; +} + +bool AtomicCompareAndSwap(u64 volatile* pointer, u64 value, u64 expected) { + u64 result = _InterlockedCompareExchange64((__int64*)pointer, value, expected); + return result == expected; +} + +bool AtomicCompareAndSwap(u64 volatile* pointer, u128 value, u128 expected) { + return _InterlockedCompareExchange128((__int64*)pointer, value[1], value[0], + (__int64*)expected.data()) != 0; +} + +#else + +bool AtomicCompareAndSwap(u8 volatile* pointer, u8 value, u8 expected) { + return __sync_bool_compare_and_swap(pointer, expected, value); +} + +bool AtomicCompareAndSwap(u16 volatile* pointer, u16 value, u16 expected) { + return __sync_bool_compare_and_swap(pointer, expected, value); +} + +bool AtomicCompareAndSwap(u32 volatile* pointer, u32 value, u32 expected) { + return __sync_bool_compare_and_swap(pointer, expected, value); +} + +bool AtomicCompareAndSwap(u64 volatile* pointer, u64 value, u64 expected) { + return __sync_bool_compare_and_swap(pointer, expected, value); +} + +bool AtomicCompareAndSwap(u64 volatile* pointer, u128 value, u128 expected) { + unsigned __int128 value_a; + unsigned __int128 expected_a; + std::memcpy(&value_a, value.data(), sizeof(u128)); + std::memcpy(&expected_a, expected.data(), sizeof(u128)); + return __sync_bool_compare_and_swap((unsigned __int128*)pointer, expected_a, value_a); +} + +#endif + +} // namespace Common diff --git a/src/common/atomic_ops.h b/src/common/atomic_ops.h new file mode 100644 index 000000000..e6181d521 --- /dev/null +++ b/src/common/atomic_ops.h @@ -0,0 +1,17 @@ +// Copyright 2020 yuzu Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include "common/common_types.h" + +namespace Common { + +bool AtomicCompareAndSwap(u8 volatile* pointer, u8 value, u8 expected); +bool AtomicCompareAndSwap(u16 volatile* pointer, u16 value, u16 expected); +bool AtomicCompareAndSwap(u32 volatile* pointer, u32 value, u32 expected); +bool AtomicCompareAndSwap(u64 volatile* pointer, u64 value, u64 expected); +bool AtomicCompareAndSwap(u64 volatile* pointer, u128 value, u128 expected); + +} // namespace Common diff --git a/src/common/fiber.cpp b/src/common/fiber.cpp new file mode 100644 index 000000000..1c1d09ccb --- /dev/null +++ b/src/common/fiber.cpp @@ -0,0 +1,222 @@ +// Copyright 2020 yuzu Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include "common/assert.h" +#include "common/fiber.h" +#if defined(_WIN32) || defined(WIN32) +#include <windows.h> +#else +#include <boost/context/detail/fcontext.hpp> +#endif + +namespace Common { + +constexpr std::size_t default_stack_size = 256 * 1024; // 256kb + +#if defined(_WIN32) || defined(WIN32) + +struct Fiber::FiberImpl { + LPVOID handle = nullptr; + LPVOID rewind_handle = nullptr; +}; + +void Fiber::Start() { + ASSERT(previous_fiber != nullptr); + previous_fiber->guard.unlock(); + previous_fiber.reset(); + entry_point(start_parameter); + UNREACHABLE(); +} + +void Fiber::OnRewind() { + ASSERT(impl->handle != nullptr); + DeleteFiber(impl->handle); + impl->handle = impl->rewind_handle; + impl->rewind_handle = nullptr; + rewind_point(rewind_parameter); + UNREACHABLE(); +} + +void Fiber::FiberStartFunc(void* fiber_parameter) { + auto fiber = static_cast<Fiber*>(fiber_parameter); + fiber->Start(); +} + +void Fiber::RewindStartFunc(void* fiber_parameter) { + auto fiber = static_cast<Fiber*>(fiber_parameter); + fiber->OnRewind(); +} + +Fiber::Fiber(std::function<void(void*)>&& entry_point_func, void* start_parameter) + : entry_point{std::move(entry_point_func)}, start_parameter{start_parameter} { + impl = std::make_unique<FiberImpl>(); + impl->handle = CreateFiber(default_stack_size, &FiberStartFunc, this); +} + +Fiber::Fiber() : impl{std::make_unique<FiberImpl>()} {} + +Fiber::~Fiber() { + if (released) { + return; + } + // Make sure the Fiber is not being used + const bool locked = guard.try_lock(); + ASSERT_MSG(locked, "Destroying a fiber that's still running"); + if (locked) { + guard.unlock(); + } + DeleteFiber(impl->handle); +} + +void Fiber::Exit() { + ASSERT_MSG(is_thread_fiber, "Exitting non main thread fiber"); + if (!is_thread_fiber) { + return; + } + ConvertFiberToThread(); + guard.unlock(); + released = true; +} + +void Fiber::SetRewindPoint(std::function<void(void*)>&& rewind_func, void* start_parameter) { + rewind_point = std::move(rewind_func); + rewind_parameter = start_parameter; +} + +void Fiber::Rewind() { + ASSERT(rewind_point); + ASSERT(impl->rewind_handle == nullptr); + impl->rewind_handle = CreateFiber(default_stack_size, &RewindStartFunc, this); + SwitchToFiber(impl->rewind_handle); +} + +void Fiber::YieldTo(std::shared_ptr<Fiber>& from, std::shared_ptr<Fiber>& to) { + ASSERT_MSG(from != nullptr, "Yielding fiber is null!"); + ASSERT_MSG(to != nullptr, "Next fiber is null!"); + to->guard.lock(); + to->previous_fiber = from; + SwitchToFiber(to->impl->handle); + ASSERT(from->previous_fiber != nullptr); + from->previous_fiber->guard.unlock(); + from->previous_fiber.reset(); +} + +std::shared_ptr<Fiber> Fiber::ThreadToFiber() { + std::shared_ptr<Fiber> fiber = std::shared_ptr<Fiber>{new Fiber()}; + fiber->guard.lock(); + fiber->impl->handle = ConvertThreadToFiber(nullptr); + fiber->is_thread_fiber = true; + return fiber; +} + +#else + +struct Fiber::FiberImpl { + alignas(64) std::array<u8, default_stack_size> stack; + alignas(64) std::array<u8, default_stack_size> rewind_stack; + u8* stack_limit; + u8* rewind_stack_limit; + boost::context::detail::fcontext_t context; + boost::context::detail::fcontext_t rewind_context; +}; + +void Fiber::Start(boost::context::detail::transfer_t& transfer) { + ASSERT(previous_fiber != nullptr); + previous_fiber->impl->context = transfer.fctx; + previous_fiber->guard.unlock(); + previous_fiber.reset(); + entry_point(start_parameter); + UNREACHABLE(); +} + +void Fiber::OnRewind([[maybe_unused]] boost::context::detail::transfer_t& transfer) { + ASSERT(impl->context != nullptr); + impl->context = impl->rewind_context; + impl->rewind_context = nullptr; + u8* tmp = impl->stack_limit; + impl->stack_limit = impl->rewind_stack_limit; + impl->rewind_stack_limit = tmp; + rewind_point(rewind_parameter); + UNREACHABLE(); +} + +void Fiber::FiberStartFunc(boost::context::detail::transfer_t transfer) { + auto fiber = static_cast<Fiber*>(transfer.data); + fiber->Start(transfer); +} + +void Fiber::RewindStartFunc(boost::context::detail::transfer_t transfer) { + auto fiber = static_cast<Fiber*>(transfer.data); + fiber->OnRewind(transfer); +} + +Fiber::Fiber(std::function<void(void*)>&& entry_point_func, void* start_parameter) + : entry_point{std::move(entry_point_func)}, start_parameter{start_parameter} { + impl = std::make_unique<FiberImpl>(); + impl->stack_limit = impl->stack.data(); + impl->rewind_stack_limit = impl->rewind_stack.data(); + u8* stack_base = impl->stack_limit + default_stack_size; + impl->context = + boost::context::detail::make_fcontext(stack_base, impl->stack.size(), FiberStartFunc); +} + +void Fiber::SetRewindPoint(std::function<void(void*)>&& rewind_func, void* start_parameter) { + rewind_point = std::move(rewind_func); + rewind_parameter = start_parameter; +} + +Fiber::Fiber() : impl{std::make_unique<FiberImpl>()} {} + +Fiber::~Fiber() { + if (released) { + return; + } + // Make sure the Fiber is not being used + const bool locked = guard.try_lock(); + ASSERT_MSG(locked, "Destroying a fiber that's still running"); + if (locked) { + guard.unlock(); + } +} + +void Fiber::Exit() { + + ASSERT_MSG(is_thread_fiber, "Exitting non main thread fiber"); + if (!is_thread_fiber) { + return; + } + guard.unlock(); + released = true; +} + +void Fiber::Rewind() { + ASSERT(rewind_point); + ASSERT(impl->rewind_context == nullptr); + u8* stack_base = impl->rewind_stack_limit + default_stack_size; + impl->rewind_context = + boost::context::detail::make_fcontext(stack_base, impl->stack.size(), RewindStartFunc); + boost::context::detail::jump_fcontext(impl->rewind_context, this); +} + +void Fiber::YieldTo(std::shared_ptr<Fiber>& from, std::shared_ptr<Fiber>& to) { + ASSERT_MSG(from != nullptr, "Yielding fiber is null!"); + ASSERT_MSG(to != nullptr, "Next fiber is null!"); + to->guard.lock(); + to->previous_fiber = from; + auto transfer = boost::context::detail::jump_fcontext(to->impl->context, to.get()); + ASSERT(from->previous_fiber != nullptr); + from->previous_fiber->impl->context = transfer.fctx; + from->previous_fiber->guard.unlock(); + from->previous_fiber.reset(); +} + +std::shared_ptr<Fiber> Fiber::ThreadToFiber() { + std::shared_ptr<Fiber> fiber = std::shared_ptr<Fiber>{new Fiber()}; + fiber->guard.lock(); + fiber->is_thread_fiber = true; + return fiber; +} + +#endif +} // namespace Common diff --git a/src/common/fiber.h b/src/common/fiber.h new file mode 100644 index 000000000..dafc1100e --- /dev/null +++ b/src/common/fiber.h @@ -0,0 +1,92 @@ +// Copyright 2020 yuzu Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include <functional> +#include <memory> + +#include "common/common_types.h" +#include "common/spin_lock.h" + +#if !defined(_WIN32) && !defined(WIN32) +namespace boost::context::detail { +struct transfer_t; +} +#endif + +namespace Common { + +/** + * Fiber class + * a fiber is a userspace thread with it's own context. They can be used to + * implement coroutines, emulated threading systems and certain asynchronous + * patterns. + * + * This class implements fibers at a low level, thus allowing greater freedom + * to implement such patterns. This fiber class is 'threadsafe' only one fiber + * can be running at a time and threads will be locked while trying to yield to + * a running fiber until it yields. WARNING exchanging two running fibers between + * threads will cause a deadlock. In order to prevent a deadlock, each thread should + * have an intermediary fiber, you switch to the intermediary fiber of the current + * thread and then from it switch to the expected fiber. This way you can exchange + * 2 fibers within 2 different threads. + */ +class Fiber { +public: + Fiber(std::function<void(void*)>&& entry_point_func, void* start_parameter); + ~Fiber(); + + Fiber(const Fiber&) = delete; + Fiber& operator=(const Fiber&) = delete; + + Fiber(Fiber&&) = default; + Fiber& operator=(Fiber&&) = default; + + /// Yields control from Fiber 'from' to Fiber 'to' + /// Fiber 'from' must be the currently running fiber. + static void YieldTo(std::shared_ptr<Fiber>& from, std::shared_ptr<Fiber>& to); + static std::shared_ptr<Fiber> ThreadToFiber(); + + void SetRewindPoint(std::function<void(void*)>&& rewind_func, void* start_parameter); + + void Rewind(); + + /// Only call from main thread's fiber + void Exit(); + + /// Changes the start parameter of the fiber. Has no effect if the fiber already started + void SetStartParameter(void* new_parameter) { + start_parameter = new_parameter; + } + +private: + Fiber(); + +#if defined(_WIN32) || defined(WIN32) + void OnRewind(); + void Start(); + static void FiberStartFunc(void* fiber_parameter); + static void RewindStartFunc(void* fiber_parameter); +#else + void OnRewind(boost::context::detail::transfer_t& transfer); + void Start(boost::context::detail::transfer_t& transfer); + static void FiberStartFunc(boost::context::detail::transfer_t transfer); + static void RewindStartFunc(boost::context::detail::transfer_t transfer); +#endif + + struct FiberImpl; + + SpinLock guard{}; + std::function<void(void*)> entry_point; + std::function<void(void*)> rewind_point; + void* rewind_parameter{}; + void* start_parameter{}; + std::shared_ptr<Fiber> previous_fiber; + std::unique_ptr<FiberImpl> impl; + bool is_thread_fiber{}; + bool released{}; +}; + +} // namespace Common diff --git a/src/common/memory_detect.cpp b/src/common/memory_detect.cpp new file mode 100644 index 000000000..3fdc309a2 --- /dev/null +++ b/src/common/memory_detect.cpp @@ -0,0 +1,60 @@ +// Copyright 2020 yuzu Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#ifdef _WIN32 +// clang-format off +#include <windows.h> +#include <sysinfoapi.h> +// clang-format on +#else +#include <sys/types.h> +#ifdef __APPLE__ +#include <sys/sysctl.h> +#else +#include <sys/sysinfo.h> +#endif +#endif + +#include "common/memory_detect.h" + +namespace Common { + +// Detects the RAM and Swapfile sizes +static MemoryInfo Detect() { + MemoryInfo mem_info{}; + +#ifdef _WIN32 + MEMORYSTATUSEX memorystatus; + memorystatus.dwLength = sizeof(memorystatus); + GlobalMemoryStatusEx(&memorystatus); + mem_info.TotalPhysicalMemory = memorystatus.ullTotalPhys; + mem_info.TotalSwapMemory = memorystatus.ullTotalPageFile - mem_info.TotalPhysicalMemory; +#elif defined(__APPLE__) + u64 ramsize; + struct xsw_usage vmusage; + std::size_t sizeof_ramsize = sizeof(ramsize); + std::size_t sizeof_vmusage = sizeof(vmusage); + // hw and vm are defined in sysctl.h + // https://github.com/apple/darwin-xnu/blob/master/bsd/sys/sysctl.h#L471 + // sysctlbyname(const char *, void *, size_t *, void *, size_t); + sysctlbyname("hw.memsize", &ramsize, &sizeof_ramsize, NULL, 0); + sysctlbyname("vm.swapusage", &vmusage, &sizeof_vmusage, NULL, 0); + mem_info.TotalPhysicalMemory = ramsize; + mem_info.TotalSwapMemory = vmusage.xsu_total; +#else + struct sysinfo meminfo; + sysinfo(&meminfo); + mem_info.TotalPhysicalMemory = meminfo.totalram; + mem_info.TotalSwapMemory = meminfo.totalswap; +#endif + + return mem_info; +} + +const MemoryInfo& GetMemInfo() { + static MemoryInfo mem_info = Detect(); + return mem_info; +} + +} // namespace Common
\ No newline at end of file diff --git a/src/common/memory_detect.h b/src/common/memory_detect.h new file mode 100644 index 000000000..a73c0f3f4 --- /dev/null +++ b/src/common/memory_detect.h @@ -0,0 +1,22 @@ +// Copyright 2020 yuzu Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include "common/common_types.h" + +namespace Common { + +struct MemoryInfo { + u64 TotalPhysicalMemory{}; + u64 TotalSwapMemory{}; +}; + +/** + * Gets the memory info of the host system + * @return Reference to a MemoryInfo struct with the physical and swap memory sizes in bytes + */ +const MemoryInfo& GetMemInfo(); + +} // namespace Common
\ No newline at end of file diff --git a/src/common/spin_lock.cpp b/src/common/spin_lock.cpp new file mode 100644 index 000000000..c1524220f --- /dev/null +++ b/src/common/spin_lock.cpp @@ -0,0 +1,54 @@ +// Copyright 2020 yuzu Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include "common/spin_lock.h" + +#if _MSC_VER +#include <intrin.h> +#if _M_AMD64 +#define __x86_64__ 1 +#endif +#if _M_ARM64 +#define __aarch64__ 1 +#endif +#else +#if __x86_64__ +#include <xmmintrin.h> +#endif +#endif + +namespace { + +void ThreadPause() { +#if __x86_64__ + _mm_pause(); +#elif __aarch64__ && _MSC_VER + __yield(); +#elif __aarch64__ + asm("yield"); +#endif +} + +} // Anonymous namespace + +namespace Common { + +void SpinLock::lock() { + while (lck.test_and_set(std::memory_order_acquire)) { + ThreadPause(); + } +} + +void SpinLock::unlock() { + lck.clear(std::memory_order_release); +} + +bool SpinLock::try_lock() { + if (lck.test_and_set(std::memory_order_acquire)) { + return false; + } + return true; +} + +} // namespace Common diff --git a/src/common/spin_lock.h b/src/common/spin_lock.h new file mode 100644 index 000000000..1df5528c4 --- /dev/null +++ b/src/common/spin_lock.h @@ -0,0 +1,26 @@ +// Copyright 2020 yuzu Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include <atomic> + +namespace Common { + +/** + * SpinLock class + * a lock similar to mutex that forces a thread to spin wait instead calling the + * supervisor. Should be used on short sequences of code. + */ +class SpinLock { +public: + void lock(); + void unlock(); + bool try_lock(); + +private: + std::atomic_flag lck = ATOMIC_FLAG_INIT; +}; + +} // namespace Common diff --git a/src/common/telemetry.cpp b/src/common/telemetry.cpp index 200c6489a..16d42facd 100644 --- a/src/common/telemetry.cpp +++ b/src/common/telemetry.cpp @@ -60,6 +60,7 @@ void AppendCPUInfo(FieldCollection& fc) { fc.AddField(FieldType::UserSystem, "CPU_Extension_x64_AES", Common::GetCPUCaps().aes); fc.AddField(FieldType::UserSystem, "CPU_Extension_x64_AVX", Common::GetCPUCaps().avx); fc.AddField(FieldType::UserSystem, "CPU_Extension_x64_AVX2", Common::GetCPUCaps().avx2); + fc.AddField(FieldType::UserSystem, "CPU_Extension_x64_AVX512", Common::GetCPUCaps().avx512); fc.AddField(FieldType::UserSystem, "CPU_Extension_x64_BMI1", Common::GetCPUCaps().bmi1); fc.AddField(FieldType::UserSystem, "CPU_Extension_x64_BMI2", Common::GetCPUCaps().bmi2); fc.AddField(FieldType::UserSystem, "CPU_Extension_x64_FMA", Common::GetCPUCaps().fma); diff --git a/src/common/thread.cpp b/src/common/thread.cpp index 0cd2d10bf..8e5935e6a 100644 --- a/src/common/thread.cpp +++ b/src/common/thread.cpp @@ -25,6 +25,52 @@ namespace Common { +#ifdef _WIN32 + +void SetCurrentThreadPriority(ThreadPriority new_priority) { + auto handle = GetCurrentThread(); + int windows_priority = 0; + switch (new_priority) { + case ThreadPriority::Low: + windows_priority = THREAD_PRIORITY_BELOW_NORMAL; + break; + case ThreadPriority::Normal: + windows_priority = THREAD_PRIORITY_NORMAL; + break; + case ThreadPriority::High: + windows_priority = THREAD_PRIORITY_ABOVE_NORMAL; + break; + case ThreadPriority::VeryHigh: + windows_priority = THREAD_PRIORITY_HIGHEST; + break; + default: + windows_priority = THREAD_PRIORITY_NORMAL; + break; + } + SetThreadPriority(handle, windows_priority); +} + +#else + +void SetCurrentThreadPriority(ThreadPriority new_priority) { + pthread_t this_thread = pthread_self(); + + s32 max_prio = sched_get_priority_max(SCHED_OTHER); + s32 min_prio = sched_get_priority_min(SCHED_OTHER); + u32 level = static_cast<u32>(new_priority) + 1; + + struct sched_param params; + if (max_prio > min_prio) { + params.sched_priority = min_prio + ((max_prio - min_prio) * level) / 4; + } else { + params.sched_priority = min_prio - ((min_prio - max_prio) * level) / 4; + } + + pthread_setschedparam(this_thread, SCHED_OTHER, ¶ms); +} + +#endif + #ifdef _MSC_VER // Sets the debugger-visible name of the current thread. @@ -70,6 +116,12 @@ void SetCurrentThreadName(const char* name) { } #endif +#if defined(_WIN32) +void SetCurrentThreadName(const char* name) { + // Do Nothing on MingW +} +#endif + #endif } // namespace Common diff --git a/src/common/thread.h b/src/common/thread.h index 2fc071685..52b359413 100644 --- a/src/common/thread.h +++ b/src/common/thread.h @@ -9,6 +9,7 @@ #include <cstddef> #include <mutex> #include <thread> +#include "common/common_types.h" namespace Common { @@ -28,8 +29,7 @@ public: is_set = false; } - template <class Duration> - bool WaitFor(const std::chrono::duration<Duration>& time) { + bool WaitFor(const std::chrono::nanoseconds& time) { std::unique_lock lk{mutex}; if (!condvar.wait_for(lk, time, [this] { return is_set; })) return false; @@ -86,6 +86,15 @@ private: std::size_t generation = 0; // Incremented once each time the barrier is used }; +enum class ThreadPriority : u32 { + Low = 0, + Normal = 1, + High = 2, + VeryHigh = 3, +}; + +void SetCurrentThreadPriority(ThreadPriority new_priority); + void SetCurrentThreadName(const char* name); } // namespace Common diff --git a/src/common/time_zone.cpp b/src/common/time_zone.cpp new file mode 100644 index 000000000..ce239eb63 --- /dev/null +++ b/src/common/time_zone.cpp @@ -0,0 +1,49 @@ +// Copyright 2020 yuzu Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include <chrono> +#include <iomanip> +#include <sstream> + +#include "common/logging/log.h" +#include "common/time_zone.h" + +namespace Common::TimeZone { + +std::string GetDefaultTimeZone() { + return "GMT"; +} + +static std::string GetOsTimeZoneOffset() { + const std::time_t t{std::time(nullptr)}; + const std::tm tm{*std::localtime(&t)}; + + std::stringstream ss; + ss << std::put_time(&tm, "%z"); // Get the current timezone offset, e.g. "-400", as a string + + return ss.str(); +} + +static int ConvertOsTimeZoneOffsetToInt(const std::string& timezone) { + try { + return std::stoi(timezone); + } catch (const std::invalid_argument&) { + LOG_CRITICAL(Common, "invalid_argument with {}!", timezone); + return 0; + } catch (const std::out_of_range&) { + LOG_CRITICAL(Common, "out_of_range with {}!", timezone); + return 0; + } +} + +std::chrono::seconds GetCurrentOffsetSeconds() { + const int offset{ConvertOsTimeZoneOffsetToInt(GetOsTimeZoneOffset())}; + + int seconds{(offset / 100) * 60 * 60}; // Convert hour component to seconds + seconds += (offset % 100) * 60; // Convert minute component to seconds + + return std::chrono::seconds{seconds}; +} + +} // namespace Common::TimeZone diff --git a/src/common/time_zone.h b/src/common/time_zone.h new file mode 100644 index 000000000..945daa09c --- /dev/null +++ b/src/common/time_zone.h @@ -0,0 +1,18 @@ +// Copyright 2020 yuzu Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include <chrono> +#include <string> + +namespace Common::TimeZone { + +/// Gets the default timezone, i.e. "GMT" +std::string GetDefaultTimeZone(); + +/// Gets the offset of the current timezone (from the default), in seconds +std::chrono::seconds GetCurrentOffsetSeconds(); + +} // namespace Common::TimeZone diff --git a/src/common/uint128.cpp b/src/common/uint128.cpp index 32bf56730..16bf7c828 100644 --- a/src/common/uint128.cpp +++ b/src/common/uint128.cpp @@ -6,12 +6,38 @@ #include <intrin.h> #pragma intrinsic(_umul128) +#pragma intrinsic(_udiv128) #endif #include <cstring> #include "common/uint128.h" namespace Common { +#ifdef _MSC_VER + +u64 MultiplyAndDivide64(u64 a, u64 b, u64 d) { + u128 r{}; + r[0] = _umul128(a, b, &r[1]); + u64 remainder; +#if _MSC_VER < 1923 + return udiv128(r[1], r[0], d, &remainder); +#else + return _udiv128(r[1], r[0], d, &remainder); +#endif +} + +#else + +u64 MultiplyAndDivide64(u64 a, u64 b, u64 d) { + const u64 diva = a / d; + const u64 moda = a % d; + const u64 divb = b / d; + const u64 modb = b % d; + return diva * b + moda * divb + moda * modb / d; +} + +#endif + u128 Multiply64Into128(u64 a, u64 b) { u128 result; #ifdef _MSC_VER diff --git a/src/common/uint128.h b/src/common/uint128.h index a3be2a2cb..503cd2d0c 100644 --- a/src/common/uint128.h +++ b/src/common/uint128.h @@ -9,6 +9,9 @@ namespace Common { +// This function multiplies 2 u64 values and divides it by a u64 value. +u64 MultiplyAndDivide64(u64 a, u64 b, u64 d); + // This function multiplies 2 u64 values and produces a u128 value; u128 Multiply64Into128(u64 a, u64 b); diff --git a/src/common/wall_clock.cpp b/src/common/wall_clock.cpp new file mode 100644 index 000000000..3afbdb898 --- /dev/null +++ b/src/common/wall_clock.cpp @@ -0,0 +1,91 @@ +// Copyright 2020 yuzu Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include "common/uint128.h" +#include "common/wall_clock.h" + +#ifdef ARCHITECTURE_x86_64 +#include "common/x64/cpu_detect.h" +#include "common/x64/native_clock.h" +#endif + +namespace Common { + +using base_timer = std::chrono::steady_clock; +using base_time_point = std::chrono::time_point<base_timer>; + +class StandardWallClock : public WallClock { +public: + StandardWallClock(u64 emulated_cpu_frequency, u64 emulated_clock_frequency) + : WallClock(emulated_cpu_frequency, emulated_clock_frequency, false) { + start_time = base_timer::now(); + } + + std::chrono::nanoseconds GetTimeNS() override { + base_time_point current = base_timer::now(); + auto elapsed = current - start_time; + return std::chrono::duration_cast<std::chrono::nanoseconds>(elapsed); + } + + std::chrono::microseconds GetTimeUS() override { + base_time_point current = base_timer::now(); + auto elapsed = current - start_time; + return std::chrono::duration_cast<std::chrono::microseconds>(elapsed); + } + + std::chrono::milliseconds GetTimeMS() override { + base_time_point current = base_timer::now(); + auto elapsed = current - start_time; + return std::chrono::duration_cast<std::chrono::milliseconds>(elapsed); + } + + u64 GetClockCycles() override { + std::chrono::nanoseconds time_now = GetTimeNS(); + const u128 temporary = + Common::Multiply64Into128(time_now.count(), emulated_clock_frequency); + return Common::Divide128On32(temporary, 1000000000).first; + } + + u64 GetCPUCycles() override { + std::chrono::nanoseconds time_now = GetTimeNS(); + const u128 temporary = Common::Multiply64Into128(time_now.count(), emulated_cpu_frequency); + return Common::Divide128On32(temporary, 1000000000).first; + } + + void Pause(bool is_paused) override { + // Do nothing in this clock type. + } + +private: + base_time_point start_time; +}; + +#ifdef ARCHITECTURE_x86_64 + +std::unique_ptr<WallClock> CreateBestMatchingClock(u32 emulated_cpu_frequency, + u32 emulated_clock_frequency) { + const auto& caps = GetCPUCaps(); + u64 rtsc_frequency = 0; + if (caps.invariant_tsc) { + rtsc_frequency = EstimateRDTSCFrequency(); + } + if (rtsc_frequency == 0) { + return std::make_unique<StandardWallClock>(emulated_cpu_frequency, + emulated_clock_frequency); + } else { + return std::make_unique<X64::NativeClock>(emulated_cpu_frequency, emulated_clock_frequency, + rtsc_frequency); + } +} + +#else + +std::unique_ptr<WallClock> CreateBestMatchingClock(u32 emulated_cpu_frequency, + u32 emulated_clock_frequency) { + return std::make_unique<StandardWallClock>(emulated_cpu_frequency, emulated_clock_frequency); +} + +#endif + +} // namespace Common diff --git a/src/common/wall_clock.h b/src/common/wall_clock.h new file mode 100644 index 000000000..367d72134 --- /dev/null +++ b/src/common/wall_clock.h @@ -0,0 +1,53 @@ +// Copyright 2020 yuzu Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include <chrono> +#include <memory> + +#include "common/common_types.h" + +namespace Common { + +class WallClock { +public: + /// Returns current wall time in nanoseconds + virtual std::chrono::nanoseconds GetTimeNS() = 0; + + /// Returns current wall time in microseconds + virtual std::chrono::microseconds GetTimeUS() = 0; + + /// Returns current wall time in milliseconds + virtual std::chrono::milliseconds GetTimeMS() = 0; + + /// Returns current wall time in emulated clock cycles + virtual u64 GetClockCycles() = 0; + + /// Returns current wall time in emulated cpu cycles + virtual u64 GetCPUCycles() = 0; + + virtual void Pause(bool is_paused) = 0; + + /// Tells if the wall clock, uses the host CPU's hardware clock + bool IsNative() const { + return is_native; + } + +protected: + WallClock(u64 emulated_cpu_frequency, u64 emulated_clock_frequency, bool is_native) + : emulated_cpu_frequency{emulated_cpu_frequency}, + emulated_clock_frequency{emulated_clock_frequency}, is_native{is_native} {} + + u64 emulated_cpu_frequency; + u64 emulated_clock_frequency; + +private: + bool is_native; +}; + +std::unique_ptr<WallClock> CreateBestMatchingClock(u32 emulated_cpu_frequency, + u32 emulated_clock_frequency); + +} // namespace Common diff --git a/src/common/x64/cpu_detect.cpp b/src/common/x64/cpu_detect.cpp index c9349a6b4..fccd2eee5 100644 --- a/src/common/x64/cpu_detect.cpp +++ b/src/common/x64/cpu_detect.cpp @@ -62,6 +62,17 @@ static CPUCaps Detect() { std::memcpy(&caps.brand_string[0], &cpu_id[1], sizeof(int)); std::memcpy(&caps.brand_string[4], &cpu_id[3], sizeof(int)); std::memcpy(&caps.brand_string[8], &cpu_id[2], sizeof(int)); + if (cpu_id[1] == 0x756e6547 && cpu_id[2] == 0x6c65746e && cpu_id[3] == 0x49656e69) + caps.manufacturer = Manufacturer::Intel; + else if (cpu_id[1] == 0x68747541 && cpu_id[2] == 0x444d4163 && cpu_id[3] == 0x69746e65) + caps.manufacturer = Manufacturer::AMD; + else if (cpu_id[1] == 0x6f677948 && cpu_id[2] == 0x656e6975 && cpu_id[3] == 0x6e65476e) + caps.manufacturer = Manufacturer::Hygon; + else + caps.manufacturer = Manufacturer::Unknown; + + u32 family = {}; + u32 model = {}; __cpuid(cpu_id, 0x80000000); @@ -73,6 +84,14 @@ static CPUCaps Detect() { // Detect family and other miscellaneous features if (max_std_fn >= 1) { __cpuid(cpu_id, 0x00000001); + family = (cpu_id[0] >> 8) & 0xf; + model = (cpu_id[0] >> 4) & 0xf; + if (family == 0xf) { + family += (cpu_id[0] >> 20) & 0xff; + } + if (family >= 6) { + model += ((cpu_id[0] >> 16) & 0xf) << 4; + } if ((cpu_id[3] >> 25) & 1) caps.sse = true; @@ -110,6 +129,11 @@ static CPUCaps Detect() { caps.bmi1 = true; if ((cpu_id[1] >> 8) & 1) caps.bmi2 = true; + // Checks for AVX512F, AVX512CD, AVX512VL, AVX512DQ, AVX512BW (Intel Skylake-X/SP) + if ((cpu_id[1] >> 16) & 1 && (cpu_id[1] >> 28) & 1 && (cpu_id[1] >> 31) & 1 && + (cpu_id[1] >> 17) & 1 && (cpu_id[1] >> 30) & 1) { + caps.avx512 = caps.avx2; + } } } @@ -130,6 +154,20 @@ static CPUCaps Detect() { caps.fma4 = true; } + if (max_ex_fn >= 0x80000007) { + __cpuid(cpu_id, 0x80000007); + if (cpu_id[3] & (1 << 8)) { + caps.invariant_tsc = true; + } + } + + if (max_std_fn >= 0x16) { + __cpuid(cpu_id, 0x16); + caps.base_frequency = cpu_id[0]; + caps.max_frequency = cpu_id[1]; + caps.bus_frequency = cpu_id[2]; + } + return caps; } diff --git a/src/common/x64/cpu_detect.h b/src/common/x64/cpu_detect.h index 20f2ba234..e3b63302e 100644 --- a/src/common/x64/cpu_detect.h +++ b/src/common/x64/cpu_detect.h @@ -6,8 +6,16 @@ namespace Common { +enum class Manufacturer : u32 { + Intel = 0, + AMD = 1, + Hygon = 2, + Unknown = 3, +}; + /// x86/x64 CPU capabilities that may be detected by this module struct CPUCaps { + Manufacturer manufacturer; char cpu_string[0x21]; char brand_string[0x41]; bool sse; @@ -19,11 +27,16 @@ struct CPUCaps { bool lzcnt; bool avx; bool avx2; + bool avx512; bool bmi1; bool bmi2; bool fma; bool fma4; bool aes; + bool invariant_tsc; + u32 base_frequency; + u32 max_frequency; + u32 bus_frequency; }; /** diff --git a/src/common/x64/native_clock.cpp b/src/common/x64/native_clock.cpp new file mode 100644 index 000000000..424b39b1f --- /dev/null +++ b/src/common/x64/native_clock.cpp @@ -0,0 +1,103 @@ +// Copyright 2020 yuzu Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include <chrono> +#include <mutex> +#include <thread> + +#ifdef _MSC_VER +#include <intrin.h> +#else +#include <x86intrin.h> +#endif + +#include "common/uint128.h" +#include "common/x64/native_clock.h" + +namespace Common { + +u64 EstimateRDTSCFrequency() { + const auto milli_10 = std::chrono::milliseconds{10}; + // get current time + _mm_mfence(); + const u64 tscStart = __rdtsc(); + const auto startTime = std::chrono::high_resolution_clock::now(); + // wait roughly 3 seconds + while (true) { + auto milli = std::chrono::duration_cast<std::chrono::milliseconds>( + std::chrono::high_resolution_clock::now() - startTime); + if (milli.count() >= 3000) + break; + std::this_thread::sleep_for(milli_10); + } + const auto endTime = std::chrono::high_resolution_clock::now(); + _mm_mfence(); + const u64 tscEnd = __rdtsc(); + // calculate difference + const u64 timer_diff = + std::chrono::duration_cast<std::chrono::nanoseconds>(endTime - startTime).count(); + const u64 tsc_diff = tscEnd - tscStart; + const u64 tsc_freq = MultiplyAndDivide64(tsc_diff, 1000000000ULL, timer_diff); + return tsc_freq; +} + +namespace X64 { +NativeClock::NativeClock(u64 emulated_cpu_frequency, u64 emulated_clock_frequency, + u64 rtsc_frequency) + : WallClock(emulated_cpu_frequency, emulated_clock_frequency, true), rtsc_frequency{ + rtsc_frequency} { + _mm_mfence(); + last_measure = __rdtsc(); + accumulated_ticks = 0U; +} + +u64 NativeClock::GetRTSC() { + std::scoped_lock scope{rtsc_serialize}; + _mm_mfence(); + const u64 current_measure = __rdtsc(); + u64 diff = current_measure - last_measure; + diff = diff & ~static_cast<u64>(static_cast<s64>(diff) >> 63); // max(diff, 0) + if (current_measure > last_measure) { + last_measure = current_measure; + } + accumulated_ticks += diff; + /// The clock cannot be more precise than the guest timer, remove the lower bits + return accumulated_ticks & inaccuracy_mask; +} + +void NativeClock::Pause(bool is_paused) { + if (!is_paused) { + _mm_mfence(); + last_measure = __rdtsc(); + } +} + +std::chrono::nanoseconds NativeClock::GetTimeNS() { + const u64 rtsc_value = GetRTSC(); + return std::chrono::nanoseconds{MultiplyAndDivide64(rtsc_value, 1000000000, rtsc_frequency)}; +} + +std::chrono::microseconds NativeClock::GetTimeUS() { + const u64 rtsc_value = GetRTSC(); + return std::chrono::microseconds{MultiplyAndDivide64(rtsc_value, 1000000, rtsc_frequency)}; +} + +std::chrono::milliseconds NativeClock::GetTimeMS() { + const u64 rtsc_value = GetRTSC(); + return std::chrono::milliseconds{MultiplyAndDivide64(rtsc_value, 1000, rtsc_frequency)}; +} + +u64 NativeClock::GetClockCycles() { + const u64 rtsc_value = GetRTSC(); + return MultiplyAndDivide64(rtsc_value, emulated_clock_frequency, rtsc_frequency); +} + +u64 NativeClock::GetCPUCycles() { + const u64 rtsc_value = GetRTSC(); + return MultiplyAndDivide64(rtsc_value, emulated_cpu_frequency, rtsc_frequency); +} + +} // namespace X64 + +} // namespace Common diff --git a/src/common/x64/native_clock.h b/src/common/x64/native_clock.h new file mode 100644 index 000000000..891a3bbfd --- /dev/null +++ b/src/common/x64/native_clock.h @@ -0,0 +1,48 @@ +// Copyright 2020 yuzu Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include <optional> + +#include "common/spin_lock.h" +#include "common/wall_clock.h" + +namespace Common { + +namespace X64 { +class NativeClock : public WallClock { +public: + NativeClock(u64 emulated_cpu_frequency, u64 emulated_clock_frequency, u64 rtsc_frequency); + + std::chrono::nanoseconds GetTimeNS() override; + + std::chrono::microseconds GetTimeUS() override; + + std::chrono::milliseconds GetTimeMS() override; + + u64 GetClockCycles() override; + + u64 GetCPUCycles() override; + + void Pause(bool is_paused) override; + +private: + u64 GetRTSC(); + + /// value used to reduce the native clocks accuracy as some apss rely on + /// undefined behavior where the level of accuracy in the clock shouldn't + /// be higher. + static constexpr u64 inaccuracy_mask = ~(0x400 - 1); + + SpinLock rtsc_serialize{}; + u64 last_measure{}; + u64 accumulated_ticks{}; + u64 rtsc_frequency; +}; +} // namespace X64 + +u64 EstimateRDTSCFrequency(); + +} // namespace Common diff --git a/src/common/x64/xbyak_abi.h b/src/common/x64/xbyak_abi.h new file mode 100644 index 000000000..a5f5d4fc1 --- /dev/null +++ b/src/common/x64/xbyak_abi.h @@ -0,0 +1,229 @@ +// Copyright 2016 Citra Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include <bitset> +#include <initializer_list> +#include <xbyak.h> +#include "common/assert.h" + +namespace Common::X64 { + +inline std::size_t RegToIndex(const Xbyak::Reg& reg) { + using Kind = Xbyak::Reg::Kind; + ASSERT_MSG((reg.getKind() & (Kind::REG | Kind::XMM)) != 0, + "RegSet only support GPRs and XMM registers."); + ASSERT_MSG(reg.getIdx() < 16, "RegSet only supports XXM0-15."); + return reg.getIdx() + (reg.getKind() == Kind::REG ? 0 : 16); +} + +inline Xbyak::Reg64 IndexToReg64(std::size_t reg_index) { + ASSERT(reg_index < 16); + return Xbyak::Reg64(static_cast<int>(reg_index)); +} + +inline Xbyak::Xmm IndexToXmm(std::size_t reg_index) { + ASSERT(reg_index >= 16 && reg_index < 32); + return Xbyak::Xmm(static_cast<int>(reg_index - 16)); +} + +inline Xbyak::Reg IndexToReg(std::size_t reg_index) { + if (reg_index < 16) { + return IndexToReg64(reg_index); + } else { + return IndexToXmm(reg_index); + } +} + +inline std::bitset<32> BuildRegSet(std::initializer_list<Xbyak::Reg> regs) { + std::bitset<32> bits; + for (const Xbyak::Reg& reg : regs) { + bits[RegToIndex(reg)] = true; + } + return bits; +} + +const std::bitset<32> ABI_ALL_GPRS(0x0000FFFF); +const std::bitset<32> ABI_ALL_XMMS(0xFFFF0000); + +#ifdef _WIN32 + +// Microsoft x64 ABI +const Xbyak::Reg ABI_RETURN = Xbyak::util::rax; +const Xbyak::Reg ABI_PARAM1 = Xbyak::util::rcx; +const Xbyak::Reg ABI_PARAM2 = Xbyak::util::rdx; +const Xbyak::Reg ABI_PARAM3 = Xbyak::util::r8; +const Xbyak::Reg ABI_PARAM4 = Xbyak::util::r9; + +const std::bitset<32> ABI_ALL_CALLER_SAVED = BuildRegSet({ + // GPRs + Xbyak::util::rcx, + Xbyak::util::rdx, + Xbyak::util::r8, + Xbyak::util::r9, + Xbyak::util::r10, + Xbyak::util::r11, + // XMMs + Xbyak::util::xmm0, + Xbyak::util::xmm1, + Xbyak::util::xmm2, + Xbyak::util::xmm3, + Xbyak::util::xmm4, + Xbyak::util::xmm5, +}); + +const std::bitset<32> ABI_ALL_CALLEE_SAVED = BuildRegSet({ + // GPRs + Xbyak::util::rbx, + Xbyak::util::rsi, + Xbyak::util::rdi, + Xbyak::util::rbp, + Xbyak::util::r12, + Xbyak::util::r13, + Xbyak::util::r14, + Xbyak::util::r15, + // XMMs + Xbyak::util::xmm6, + Xbyak::util::xmm7, + Xbyak::util::xmm8, + Xbyak::util::xmm9, + Xbyak::util::xmm10, + Xbyak::util::xmm11, + Xbyak::util::xmm12, + Xbyak::util::xmm13, + Xbyak::util::xmm14, + Xbyak::util::xmm15, +}); + +constexpr size_t ABI_SHADOW_SPACE = 0x20; + +#else + +// System V x86-64 ABI +const Xbyak::Reg ABI_RETURN = Xbyak::util::rax; +const Xbyak::Reg ABI_PARAM1 = Xbyak::util::rdi; +const Xbyak::Reg ABI_PARAM2 = Xbyak::util::rsi; +const Xbyak::Reg ABI_PARAM3 = Xbyak::util::rdx; +const Xbyak::Reg ABI_PARAM4 = Xbyak::util::rcx; + +const std::bitset<32> ABI_ALL_CALLER_SAVED = BuildRegSet({ + // GPRs + Xbyak::util::rcx, + Xbyak::util::rdx, + Xbyak::util::rdi, + Xbyak::util::rsi, + Xbyak::util::r8, + Xbyak::util::r9, + Xbyak::util::r10, + Xbyak::util::r11, + // XMMs + Xbyak::util::xmm0, + Xbyak::util::xmm1, + Xbyak::util::xmm2, + Xbyak::util::xmm3, + Xbyak::util::xmm4, + Xbyak::util::xmm5, + Xbyak::util::xmm6, + Xbyak::util::xmm7, + Xbyak::util::xmm8, + Xbyak::util::xmm9, + Xbyak::util::xmm10, + Xbyak::util::xmm11, + Xbyak::util::xmm12, + Xbyak::util::xmm13, + Xbyak::util::xmm14, + Xbyak::util::xmm15, +}); + +const std::bitset<32> ABI_ALL_CALLEE_SAVED = BuildRegSet({ + // GPRs + Xbyak::util::rbx, + Xbyak::util::rbp, + Xbyak::util::r12, + Xbyak::util::r13, + Xbyak::util::r14, + Xbyak::util::r15, +}); + +constexpr size_t ABI_SHADOW_SPACE = 0; + +#endif + +struct ABIFrameInfo { + s32 subtraction; + s32 xmm_offset; +}; + +inline ABIFrameInfo ABI_CalculateFrameSize(std::bitset<32> regs, size_t rsp_alignment, + size_t needed_frame_size) { + const auto count = (regs & ABI_ALL_GPRS).count(); + rsp_alignment -= count * 8; + size_t subtraction = 0; + const auto xmm_count = (regs & ABI_ALL_XMMS).count(); + if (xmm_count) { + // If we have any XMMs to save, we must align the stack here. + subtraction = rsp_alignment & 0xF; + } + subtraction += 0x10 * xmm_count; + size_t xmm_base_subtraction = subtraction; + subtraction += needed_frame_size; + subtraction += ABI_SHADOW_SPACE; + // Final alignment. + rsp_alignment -= subtraction; + subtraction += rsp_alignment & 0xF; + + return ABIFrameInfo{static_cast<s32>(subtraction), + static_cast<s32>(subtraction - xmm_base_subtraction)}; +} + +inline size_t ABI_PushRegistersAndAdjustStack(Xbyak::CodeGenerator& code, std::bitset<32> regs, + size_t rsp_alignment, size_t needed_frame_size = 0) { + auto frame_info = ABI_CalculateFrameSize(regs, rsp_alignment, needed_frame_size); + + for (std::size_t i = 0; i < regs.size(); ++i) { + if (regs[i] && ABI_ALL_GPRS[i]) { + code.push(IndexToReg64(i)); + } + } + + if (frame_info.subtraction != 0) { + code.sub(code.rsp, frame_info.subtraction); + } + + for (std::size_t i = 0; i < regs.size(); ++i) { + if (regs[i] && ABI_ALL_XMMS[i]) { + code.movaps(code.xword[code.rsp + frame_info.xmm_offset], IndexToXmm(i)); + frame_info.xmm_offset += 0x10; + } + } + + return ABI_SHADOW_SPACE; +} + +inline void ABI_PopRegistersAndAdjustStack(Xbyak::CodeGenerator& code, std::bitset<32> regs, + size_t rsp_alignment, size_t needed_frame_size = 0) { + auto frame_info = ABI_CalculateFrameSize(regs, rsp_alignment, needed_frame_size); + + for (std::size_t i = 0; i < regs.size(); ++i) { + if (regs[i] && ABI_ALL_XMMS[i]) { + code.movaps(IndexToXmm(i), code.xword[code.rsp + frame_info.xmm_offset]); + frame_info.xmm_offset += 0x10; + } + } + + if (frame_info.subtraction != 0) { + code.add(code.rsp, frame_info.subtraction); + } + + // GPRs need to be popped in reverse order + for (std::size_t j = 0; j < regs.size(); ++j) { + const std::size_t i = regs.size() - j - 1; + if (regs[i] && ABI_ALL_GPRS[i]) { + code.pop(IndexToReg64(i)); + } + } +} + +} // namespace Common::X64 diff --git a/src/common/x64/xbyak_util.h b/src/common/x64/xbyak_util.h new file mode 100644 index 000000000..df17f8cbe --- /dev/null +++ b/src/common/x64/xbyak_util.h @@ -0,0 +1,47 @@ +// Copyright 2016 Citra Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include <type_traits> +#include <xbyak.h> +#include "common/x64/xbyak_abi.h" + +namespace Common::X64 { + +// Constants for use with cmpps/cmpss +enum { + CMP_EQ = 0, + CMP_LT = 1, + CMP_LE = 2, + CMP_UNORD = 3, + CMP_NEQ = 4, + CMP_NLT = 5, + CMP_NLE = 6, + CMP_ORD = 7, +}; + +constexpr bool IsWithin2G(uintptr_t ref, uintptr_t target) { + const u64 distance = target - (ref + 5); + return !(distance >= 0x8000'0000ULL && distance <= ~0x8000'0000ULL); +} + +inline bool IsWithin2G(const Xbyak::CodeGenerator& code, uintptr_t target) { + return IsWithin2G(reinterpret_cast<uintptr_t>(code.getCurr()), target); +} + +template <typename T> +inline void CallFarFunction(Xbyak::CodeGenerator& code, const T f) { + static_assert(std::is_pointer_v<T>, "Argument must be a (function) pointer."); + size_t addr = reinterpret_cast<size_t>(f); + if (IsWithin2G(code, addr)) { + code.call(f); + } else { + // ABI_RETURN is a safe temp register to use before a call + code.mov(ABI_RETURN, addr); + code.call(ABI_RETURN); + } +} + +} // namespace Common::X64 diff --git a/src/core/CMakeLists.txt b/src/core/CMakeLists.txt index 8546d3602..f87d67db5 100644 --- a/src/core/CMakeLists.txt +++ b/src/core/CMakeLists.txt @@ -7,6 +7,16 @@ endif() add_library(core STATIC arm/arm_interface.h arm/arm_interface.cpp + arm/cpu_interrupt_handler.cpp + arm/cpu_interrupt_handler.h + arm/dynarmic/arm_dynarmic_32.cpp + arm/dynarmic/arm_dynarmic_32.h + arm/dynarmic/arm_dynarmic_64.cpp + arm/dynarmic/arm_dynarmic_64.h + arm/dynarmic/arm_dynarmic_cp15.cpp + arm/dynarmic/arm_dynarmic_cp15.h + arm/dynarmic/arm_exclusive_monitor.cpp + arm/dynarmic/arm_exclusive_monitor.h arm/exclusive_monitor.cpp arm/exclusive_monitor.h arm/unicorn/arm_unicorn.cpp @@ -15,8 +25,6 @@ add_library(core STATIC constants.h core.cpp core.h - core_manager.cpp - core_manager.h core_timing.cpp core_timing.h core_timing_util.cpp @@ -606,11 +614,11 @@ endif() create_target_directory_groups(core) target_link_libraries(core PUBLIC common PRIVATE audio_core video_core) -target_link_libraries(core PUBLIC Boost::boost PRIVATE fmt json-headers mbedtls opus unicorn) +target_link_libraries(core PUBLIC Boost::boost PRIVATE fmt::fmt nlohmann_json::nlohmann_json mbedtls Opus::Opus unicorn zip) if (YUZU_ENABLE_BOXCAT) target_compile_definitions(core PRIVATE -DYUZU_ENABLE_BOXCAT) - target_link_libraries(core PRIVATE httplib json-headers zip) + target_link_libraries(core PRIVATE httplib nlohmann_json::nlohmann_json) endif() if (ENABLE_WEB_SERVICE) diff --git a/src/core/arm/arm_interface.cpp b/src/core/arm/arm_interface.cpp index d079a1bc8..d2295ed90 100644 --- a/src/core/arm/arm_interface.cpp +++ b/src/core/arm/arm_interface.cpp @@ -139,6 +139,63 @@ std::optional<std::string> GetSymbolName(const Symbols& symbols, VAddr func_addr constexpr u64 SEGMENT_BASE = 0x7100000000ull; +std::vector<ARM_Interface::BacktraceEntry> ARM_Interface::GetBacktraceFromContext( + System& system, const ThreadContext64& ctx) { + std::vector<BacktraceEntry> out; + auto& memory = system.Memory(); + + auto fp = ctx.cpu_registers[29]; + auto lr = ctx.cpu_registers[30]; + while (true) { + out.push_back({"", 0, lr, 0}); + if (!fp) { + break; + } + lr = memory.Read64(fp + 8) - 4; + fp = memory.Read64(fp); + } + + std::map<VAddr, std::string> modules; + auto& loader{system.GetAppLoader()}; + if (loader.ReadNSOModules(modules) != Loader::ResultStatus::Success) { + return {}; + } + + std::map<std::string, Symbols> symbols; + for (const auto& module : modules) { + symbols.insert_or_assign(module.second, GetSymbols(module.first, memory)); + } + + for (auto& entry : out) { + VAddr base = 0; + for (auto iter = modules.rbegin(); iter != modules.rend(); ++iter) { + const auto& module{*iter}; + if (entry.original_address >= module.first) { + entry.module = module.second; + base = module.first; + break; + } + } + + entry.offset = entry.original_address - base; + entry.address = SEGMENT_BASE + entry.offset; + + if (entry.module.empty()) + entry.module = "unknown"; + + const auto symbol_set = symbols.find(entry.module); + if (symbol_set != symbols.end()) { + const auto symbol = GetSymbolName(symbol_set->second, entry.offset); + if (symbol.has_value()) { + // TODO(DarkLordZach): Add demangling of symbol names. + entry.name = *symbol; + } + } + } + + return out; +} + std::vector<ARM_Interface::BacktraceEntry> ARM_Interface::GetBacktrace() const { std::vector<BacktraceEntry> out; auto& memory = system.Memory(); diff --git a/src/core/arm/arm_interface.h b/src/core/arm/arm_interface.h index cb2e640e2..1f24051e4 100644 --- a/src/core/arm/arm_interface.h +++ b/src/core/arm/arm_interface.h @@ -7,6 +7,7 @@ #include <array> #include <vector> #include "common/common_types.h" +#include "core/hardware_properties.h" namespace Common { struct PageTable; @@ -18,25 +19,29 @@ enum class VMAPermission : u8; namespace Core { class System; +class CPUInterruptHandler; + +using CPUInterrupts = std::array<CPUInterruptHandler, Core::Hardware::NUM_CPU_CORES>; /// Generic ARMv8 CPU interface class ARM_Interface : NonCopyable { public: - explicit ARM_Interface(System& system_) : system{system_} {} + explicit ARM_Interface(System& system_, CPUInterrupts& interrupt_handlers, bool uses_wall_clock) + : system{system_}, interrupt_handlers{interrupt_handlers}, uses_wall_clock{ + uses_wall_clock} {} virtual ~ARM_Interface() = default; struct ThreadContext32 { std::array<u32, 16> cpu_registers{}; + std::array<u32, 64> extension_registers{}; u32 cpsr{}; - std::array<u8, 4> padding{}; - std::array<u64, 32> fprs{}; u32 fpscr{}; u32 fpexc{}; u32 tpidr{}; }; // Internally within the kernel, it expects the AArch32 version of the // thread context to be 344 bytes in size. - static_assert(sizeof(ThreadContext32) == 0x158); + static_assert(sizeof(ThreadContext32) == 0x150); struct ThreadContext64 { std::array<u64, 31> cpu_registers{}; @@ -143,6 +148,8 @@ public: */ virtual void SetTPIDR_EL0(u64 value) = 0; + virtual void ChangeProcessorID(std::size_t new_core_id) = 0; + virtual void SaveContext(ThreadContext32& ctx) = 0; virtual void SaveContext(ThreadContext64& ctx) = 0; virtual void LoadContext(const ThreadContext32& ctx) = 0; @@ -162,6 +169,9 @@ public: std::string name; }; + static std::vector<BacktraceEntry> GetBacktraceFromContext(System& system, + const ThreadContext64& ctx); + std::vector<BacktraceEntry> GetBacktrace() const; /// fp (= r29) points to the last frame record. @@ -175,6 +185,8 @@ public: protected: /// System context that this ARM interface is running under. System& system; + CPUInterrupts& interrupt_handlers; + bool uses_wall_clock; }; } // namespace Core diff --git a/src/core/arm/cpu_interrupt_handler.cpp b/src/core/arm/cpu_interrupt_handler.cpp new file mode 100644 index 000000000..2f1a1a269 --- /dev/null +++ b/src/core/arm/cpu_interrupt_handler.cpp @@ -0,0 +1,29 @@ +// Copyright 2020 yuzu emulator team +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include "common/thread.h" +#include "core/arm/cpu_interrupt_handler.h" + +namespace Core { + +CPUInterruptHandler::CPUInterruptHandler() : is_interrupted{} { + interrupt_event = std::make_unique<Common::Event>(); +} + +CPUInterruptHandler::~CPUInterruptHandler() = default; + +void CPUInterruptHandler::SetInterrupt(bool is_interrupted_) { + if (is_interrupted_) { + interrupt_event->Set(); + } + this->is_interrupted = is_interrupted_; +} + +void CPUInterruptHandler::AwaitInterrupt() { + interrupt_event->Wait(); +} + +} // namespace Core diff --git a/src/core/arm/cpu_interrupt_handler.h b/src/core/arm/cpu_interrupt_handler.h new file mode 100644 index 000000000..3d062d326 --- /dev/null +++ b/src/core/arm/cpu_interrupt_handler.h @@ -0,0 +1,39 @@ +// Copyright 2020 yuzu emulator team +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include <memory> + +namespace Common { +class Event; +} + +namespace Core { + +class CPUInterruptHandler { +public: + CPUInterruptHandler(); + ~CPUInterruptHandler(); + + CPUInterruptHandler(const CPUInterruptHandler&) = delete; + CPUInterruptHandler& operator=(const CPUInterruptHandler&) = delete; + + CPUInterruptHandler(CPUInterruptHandler&&) = default; + CPUInterruptHandler& operator=(CPUInterruptHandler&&) = default; + + bool IsInterrupted() const { + return is_interrupted; + } + + void SetInterrupt(bool is_interrupted); + + void AwaitInterrupt(); + +private: + bool is_interrupted{}; + std::unique_ptr<Common::Event> interrupt_event; +}; + +} // namespace Core diff --git a/src/core/arm/dynarmic/arm_dynarmic_32.cpp b/src/core/arm/dynarmic/arm_dynarmic_32.cpp index 9bc86e3b9..0d4ab95b7 100644 --- a/src/core/arm/dynarmic/arm_dynarmic_32.cpp +++ b/src/core/arm/dynarmic/arm_dynarmic_32.cpp @@ -7,15 +7,17 @@ #include <dynarmic/A32/a32.h> #include <dynarmic/A32/config.h> #include <dynarmic/A32/context.h> -#include "common/microprofile.h" +#include "common/logging/log.h" +#include "common/page_table.h" +#include "core/arm/cpu_interrupt_handler.h" #include "core/arm/dynarmic/arm_dynarmic_32.h" -#include "core/arm/dynarmic/arm_dynarmic_64.h" #include "core/arm/dynarmic/arm_dynarmic_cp15.h" +#include "core/arm/dynarmic/arm_exclusive_monitor.h" #include "core/core.h" -#include "core/core_manager.h" #include "core/core_timing.h" #include "core/hle/kernel/svc.h" #include "core/memory.h" +#include "core/settings.h" namespace Core { @@ -49,8 +51,22 @@ public: parent.system.Memory().Write64(vaddr, value); } + bool MemoryWriteExclusive8(u32 vaddr, u8 value, u8 expected) override { + return parent.system.Memory().WriteExclusive8(vaddr, value, expected); + } + bool MemoryWriteExclusive16(u32 vaddr, u16 value, u16 expected) override { + return parent.system.Memory().WriteExclusive16(vaddr, value, expected); + } + bool MemoryWriteExclusive32(u32 vaddr, u32 value, u32 expected) override { + return parent.system.Memory().WriteExclusive32(vaddr, value, expected); + } + bool MemoryWriteExclusive64(u32 vaddr, u64 value, u64 expected) override { + return parent.system.Memory().WriteExclusive64(vaddr, value, expected); + } + void InterpreterFallback(u32 pc, std::size_t num_instructions) override { - UNIMPLEMENTED(); + UNIMPLEMENTED_MSG("This should never happen, pc = {:08X}, code = {:08X}", pc, + MemoryReadCode(pc)); } void ExceptionRaised(u32 pc, Dynarmic::A32::Exception exception) override { @@ -61,7 +77,7 @@ public: case Dynarmic::A32::Exception::Breakpoint: break; } - LOG_CRITICAL(HW_GPU, "ExceptionRaised(exception = {}, pc = {:08X}, code = {:08X})", + LOG_CRITICAL(Core_ARM, "ExceptionRaised(exception = {}, pc = {:08X}, code = {:08X})", static_cast<std::size_t>(exception), pc, MemoryReadCode(pc)); UNIMPLEMENTED(); } @@ -71,26 +87,36 @@ public: } void AddTicks(u64 ticks) override { + if (parent.uses_wall_clock) { + return; + } // Divide the number of ticks by the amount of CPU cores. TODO(Subv): This yields only a // rough approximation of the amount of executed ticks in the system, it may be thrown off // if not all cores are doing a similar amount of work. Instead of doing this, we should // device a way so that timing is consistent across all cores without increasing the ticks 4 // times. - u64 amortized_ticks = (ticks - num_interpreted_instructions) / Core::NUM_CPU_CORES; + u64 amortized_ticks = + (ticks - num_interpreted_instructions) / Core::Hardware::NUM_CPU_CORES; // Always execute at least one tick. amortized_ticks = std::max<u64>(amortized_ticks, 1); parent.system.CoreTiming().AddTicks(amortized_ticks); num_interpreted_instructions = 0; } + u64 GetTicksRemaining() override { - return std::max(parent.system.CoreTiming().GetDowncount(), {}); + if (parent.uses_wall_clock) { + if (!parent.interrupt_handlers[parent.core_index].IsInterrupted()) { + return minimum_run_cycles; + } + return 0U; + } + return std::max<s64>(parent.system.CoreTiming().GetDowncount(), 0); } ARM_Dynarmic_32& parent; std::size_t num_interpreted_instructions{}; - u64 tpidrro_el0{}; - u64 tpidr_el0{}; + static constexpr u64 minimum_run_cycles = 1000U; }; std::shared_ptr<Dynarmic::A32::Jit> ARM_Dynarmic_32::MakeJit(Common::PageTable& page_table, @@ -99,26 +125,46 @@ std::shared_ptr<Dynarmic::A32::Jit> ARM_Dynarmic_32::MakeJit(Common::PageTable& config.callbacks = cb.get(); // TODO(bunnei): Implement page table for 32-bit // config.page_table = &page_table.pointers; - config.coprocessors[15] = std::make_shared<DynarmicCP15>((u32*)&CP15_regs[0]); + config.coprocessors[15] = cp15; config.define_unpredictable_behaviour = true; + static constexpr std::size_t PAGE_BITS = 12; + static constexpr std::size_t NUM_PAGE_TABLE_ENTRIES = 1 << (32 - PAGE_BITS); + config.page_table = reinterpret_cast<std::array<std::uint8_t*, NUM_PAGE_TABLE_ENTRIES>*>( + page_table.pointers.data()); + config.absolute_offset_page_table = true; + config.detect_misaligned_access_via_page_table = 16 | 32 | 64 | 128; + config.only_detect_misalignment_via_page_table_on_page_boundary = true; + + // Multi-process state + config.processor_id = core_index; + config.global_monitor = &exclusive_monitor.monitor; + + // Timing + config.wall_clock_cntpct = uses_wall_clock; + + // Optimizations + if (Settings::values.disable_cpu_opt) { + config.enable_optimizations = false; + config.enable_fast_dispatch = false; + } + return std::make_unique<Dynarmic::A32::Jit>(config); } -MICROPROFILE_DEFINE(ARM_Jit_Dynarmic_32, "ARM JIT", "Dynarmic", MP_RGB(255, 64, 64)); - void ARM_Dynarmic_32::Run() { - MICROPROFILE_SCOPE(ARM_Jit_Dynarmic_32); jit->Run(); } void ARM_Dynarmic_32::Step() { - cb->InterpreterFallback(jit->Regs()[15], 1); + jit->Step(); } -ARM_Dynarmic_32::ARM_Dynarmic_32(System& system, ExclusiveMonitor& exclusive_monitor, +ARM_Dynarmic_32::ARM_Dynarmic_32(System& system, CPUInterrupts& interrupt_handlers, + bool uses_wall_clock, ExclusiveMonitor& exclusive_monitor, std::size_t core_index) - : ARM_Interface{system}, - cb(std::make_unique<DynarmicCallbacks32>(*this)), core_index{core_index}, + : ARM_Interface{system, interrupt_handlers, uses_wall_clock}, + cb(std::make_unique<DynarmicCallbacks32>(*this)), + cp15(std::make_shared<DynarmicCP15>(*this)), core_index{core_index}, exclusive_monitor{dynamic_cast<DynarmicExclusiveMonitor&>(exclusive_monitor)} {} ARM_Dynarmic_32::~ARM_Dynarmic_32() = default; @@ -154,32 +200,40 @@ void ARM_Dynarmic_32::SetPSTATE(u32 cpsr) { } u64 ARM_Dynarmic_32::GetTlsAddress() const { - return CP15_regs[static_cast<std::size_t>(CP15Register::CP15_THREAD_URO)]; + return cp15->uro; } void ARM_Dynarmic_32::SetTlsAddress(VAddr address) { - CP15_regs[static_cast<std::size_t>(CP15Register::CP15_THREAD_URO)] = static_cast<u32>(address); + cp15->uro = static_cast<u32>(address); } u64 ARM_Dynarmic_32::GetTPIDR_EL0() const { - return cb->tpidr_el0; + return cp15->uprw; } void ARM_Dynarmic_32::SetTPIDR_EL0(u64 value) { - cb->tpidr_el0 = value; + cp15->uprw = static_cast<u32>(value); +} + +void ARM_Dynarmic_32::ChangeProcessorID(std::size_t new_core_id) { + jit->ChangeProcessorID(new_core_id); } void ARM_Dynarmic_32::SaveContext(ThreadContext32& ctx) { Dynarmic::A32::Context context; jit->SaveContext(context); ctx.cpu_registers = context.Regs(); + ctx.extension_registers = context.ExtRegs(); ctx.cpsr = context.Cpsr(); + ctx.fpscr = context.Fpscr(); } void ARM_Dynarmic_32::LoadContext(const ThreadContext32& ctx) { Dynarmic::A32::Context context; context.Regs() = ctx.cpu_registers; + context.ExtRegs() = ctx.extension_registers; context.SetCpsr(ctx.cpsr); + context.SetFpscr(ctx.fpscr); jit->LoadContext(context); } @@ -188,10 +242,15 @@ void ARM_Dynarmic_32::PrepareReschedule() { } void ARM_Dynarmic_32::ClearInstructionCache() { + if (!jit) { + return; + } jit->ClearCache(); } -void ARM_Dynarmic_32::ClearExclusiveState() {} +void ARM_Dynarmic_32::ClearExclusiveState() { + jit->ClearExclusiveState(); +} void ARM_Dynarmic_32::PageTableChanged(Common::PageTable& page_table, std::size_t new_address_space_size_in_bits) { diff --git a/src/core/arm/dynarmic/arm_dynarmic_32.h b/src/core/arm/dynarmic/arm_dynarmic_32.h index 8ba9cea8f..2bab31b92 100644 --- a/src/core/arm/dynarmic/arm_dynarmic_32.h +++ b/src/core/arm/dynarmic/arm_dynarmic_32.h @@ -9,7 +9,7 @@ #include <dynarmic/A32/a32.h> #include <dynarmic/A64/a64.h> -#include <dynarmic/A64/exclusive_monitor.h> +#include <dynarmic/exclusive_monitor.h> #include "common/common_types.h" #include "common/hash.h" #include "core/arm/arm_interface.h" @@ -21,13 +21,16 @@ class Memory; namespace Core { +class CPUInterruptHandler; class DynarmicCallbacks32; +class DynarmicCP15; class DynarmicExclusiveMonitor; class System; class ARM_Dynarmic_32 final : public ARM_Interface { public: - ARM_Dynarmic_32(System& system, ExclusiveMonitor& exclusive_monitor, std::size_t core_index); + ARM_Dynarmic_32(System& system, CPUInterrupts& interrupt_handlers, bool uses_wall_clock, + ExclusiveMonitor& exclusive_monitor, std::size_t core_index); ~ARM_Dynarmic_32() override; void SetPC(u64 pc) override; @@ -44,6 +47,7 @@ public: void SetTlsAddress(VAddr address) override; void SetTPIDR_EL0(u64 value) override; u64 GetTPIDR_EL0() const override; + void ChangeProcessorID(std::size_t new_core_id) override; void SaveContext(ThreadContext32& ctx) override; void SaveContext(ThreadContext64& ctx) override {} @@ -66,12 +70,14 @@ private: std::unordered_map<JitCacheKey, std::shared_ptr<Dynarmic::A32::Jit>, Common::PairHash>; friend class DynarmicCallbacks32; + friend class DynarmicCP15; + std::unique_ptr<DynarmicCallbacks32> cb; JitCacheType jit_cache; std::shared_ptr<Dynarmic::A32::Jit> jit; + std::shared_ptr<DynarmicCP15> cp15; std::size_t core_index; DynarmicExclusiveMonitor& exclusive_monitor; - std::array<u32, 84> CP15_regs{}; }; } // namespace Core diff --git a/src/core/arm/dynarmic/arm_dynarmic_64.cpp b/src/core/arm/dynarmic/arm_dynarmic_64.cpp index 337b97be9..790981034 100644 --- a/src/core/arm/dynarmic/arm_dynarmic_64.cpp +++ b/src/core/arm/dynarmic/arm_dynarmic_64.cpp @@ -7,11 +7,11 @@ #include <dynarmic/A64/a64.h> #include <dynarmic/A64/config.h> #include "common/logging/log.h" -#include "common/microprofile.h" #include "common/page_table.h" +#include "core/arm/cpu_interrupt_handler.h" #include "core/arm/dynarmic/arm_dynarmic_64.h" +#include "core/arm/dynarmic/arm_exclusive_monitor.h" #include "core/core.h" -#include "core/core_manager.h" #include "core/core_timing.h" #include "core/core_timing_util.h" #include "core/gdbstub/gdbstub.h" @@ -65,6 +65,22 @@ public: memory.Write64(vaddr + 8, value[1]); } + bool MemoryWriteExclusive8(u64 vaddr, std::uint8_t value, std::uint8_t expected) override { + return parent.system.Memory().WriteExclusive8(vaddr, value, expected); + } + bool MemoryWriteExclusive16(u64 vaddr, std::uint16_t value, std::uint16_t expected) override { + return parent.system.Memory().WriteExclusive16(vaddr, value, expected); + } + bool MemoryWriteExclusive32(u64 vaddr, std::uint32_t value, std::uint32_t expected) override { + return parent.system.Memory().WriteExclusive32(vaddr, value, expected); + } + bool MemoryWriteExclusive64(u64 vaddr, std::uint64_t value, std::uint64_t expected) override { + return parent.system.Memory().WriteExclusive64(vaddr, value, expected); + } + bool MemoryWriteExclusive128(u64 vaddr, Vector value, Vector expected) override { + return parent.system.Memory().WriteExclusive128(vaddr, value, expected); + } + void InterpreterFallback(u64 pc, std::size_t num_instructions) override { LOG_INFO(Core_ARM, "Unicorn fallback @ 0x{:X} for {} instructions (instr = {:08X})", pc, num_instructions, MemoryReadCode(pc)); @@ -98,8 +114,8 @@ public: } [[fallthrough]]; default: - ASSERT_MSG(false, "ExceptionRaised(exception = {}, pc = {:X})", - static_cast<std::size_t>(exception), pc); + ASSERT_MSG(false, "ExceptionRaised(exception = {}, pc = {:08X}, code = {:08X})", + static_cast<std::size_t>(exception), pc, MemoryReadCode(pc)); } } @@ -108,29 +124,42 @@ public: } void AddTicks(u64 ticks) override { + if (parent.uses_wall_clock) { + return; + } // Divide the number of ticks by the amount of CPU cores. TODO(Subv): This yields only a // rough approximation of the amount of executed ticks in the system, it may be thrown off // if not all cores are doing a similar amount of work. Instead of doing this, we should // device a way so that timing is consistent across all cores without increasing the ticks 4 // times. - u64 amortized_ticks = (ticks - num_interpreted_instructions) / Core::NUM_CPU_CORES; + u64 amortized_ticks = + (ticks - num_interpreted_instructions) / Core::Hardware::NUM_CPU_CORES; // Always execute at least one tick. amortized_ticks = std::max<u64>(amortized_ticks, 1); parent.system.CoreTiming().AddTicks(amortized_ticks); num_interpreted_instructions = 0; } + u64 GetTicksRemaining() override { - return std::max(parent.system.CoreTiming().GetDowncount(), s64{0}); + if (parent.uses_wall_clock) { + if (!parent.interrupt_handlers[parent.core_index].IsInterrupted()) { + return minimum_run_cycles; + } + return 0U; + } + return std::max<s64>(parent.system.CoreTiming().GetDowncount(), 0); } + u64 GetCNTPCT() override { - return Timing::CpuCyclesToClockCycles(parent.system.CoreTiming().GetTicks()); + return parent.system.CoreTiming().GetClockTicks(); } ARM_Dynarmic_64& parent; std::size_t num_interpreted_instructions = 0; u64 tpidrro_el0 = 0; u64 tpidr_el0 = 0; + static constexpr u64 minimum_run_cycles = 1000U; }; std::shared_ptr<Dynarmic::A64::Jit> ARM_Dynarmic_64::MakeJit(Common::PageTable& page_table, @@ -168,14 +197,13 @@ std::shared_ptr<Dynarmic::A64::Jit> ARM_Dynarmic_64::MakeJit(Common::PageTable& config.enable_fast_dispatch = false; } + // Timing + config.wall_clock_cntpct = uses_wall_clock; + return std::make_shared<Dynarmic::A64::Jit>(config); } -MICROPROFILE_DEFINE(ARM_Jit_Dynarmic_64, "ARM JIT", "Dynarmic", MP_RGB(255, 64, 64)); - void ARM_Dynarmic_64::Run() { - MICROPROFILE_SCOPE(ARM_Jit_Dynarmic_64); - jit->Run(); } @@ -183,11 +211,16 @@ void ARM_Dynarmic_64::Step() { cb->InterpreterFallback(jit->GetPC(), 1); } -ARM_Dynarmic_64::ARM_Dynarmic_64(System& system, ExclusiveMonitor& exclusive_monitor, +ARM_Dynarmic_64::ARM_Dynarmic_64(System& system, CPUInterrupts& interrupt_handlers, + bool uses_wall_clock, ExclusiveMonitor& exclusive_monitor, std::size_t core_index) - : ARM_Interface{system}, cb(std::make_unique<DynarmicCallbacks64>(*this)), - inner_unicorn{system, ARM_Unicorn::Arch::AArch64}, core_index{core_index}, - exclusive_monitor{dynamic_cast<DynarmicExclusiveMonitor&>(exclusive_monitor)} {} + : ARM_Interface{system, interrupt_handlers, uses_wall_clock}, + cb(std::make_unique<DynarmicCallbacks64>(*this)), inner_unicorn{system, interrupt_handlers, + uses_wall_clock, + ARM_Unicorn::Arch::AArch64, + core_index}, + core_index{core_index}, exclusive_monitor{ + dynamic_cast<DynarmicExclusiveMonitor&>(exclusive_monitor)} {} ARM_Dynarmic_64::~ARM_Dynarmic_64() = default; @@ -239,6 +272,10 @@ void ARM_Dynarmic_64::SetTPIDR_EL0(u64 value) { cb->tpidr_el0 = value; } +void ARM_Dynarmic_64::ChangeProcessorID(std::size_t new_core_id) { + jit->ChangeProcessorID(new_core_id); +} + void ARM_Dynarmic_64::SaveContext(ThreadContext64& ctx) { ctx.cpu_registers = jit->GetRegisters(); ctx.sp = jit->GetSP(); @@ -266,6 +303,9 @@ void ARM_Dynarmic_64::PrepareReschedule() { } void ARM_Dynarmic_64::ClearInstructionCache() { + if (!jit) { + return; + } jit->ClearCache(); } @@ -285,44 +325,4 @@ void ARM_Dynarmic_64::PageTableChanged(Common::PageTable& page_table, jit_cache.emplace(key, jit); } -DynarmicExclusiveMonitor::DynarmicExclusiveMonitor(Memory::Memory& memory, std::size_t core_count) - : monitor(core_count), memory{memory} {} - -DynarmicExclusiveMonitor::~DynarmicExclusiveMonitor() = default; - -void DynarmicExclusiveMonitor::SetExclusive(std::size_t core_index, VAddr addr) { - // Size doesn't actually matter. - monitor.Mark(core_index, addr, 16); -} - -void DynarmicExclusiveMonitor::ClearExclusive() { - monitor.Clear(); -} - -bool DynarmicExclusiveMonitor::ExclusiveWrite8(std::size_t core_index, VAddr vaddr, u8 value) { - return monitor.DoExclusiveOperation(core_index, vaddr, 1, [&] { memory.Write8(vaddr, value); }); -} - -bool DynarmicExclusiveMonitor::ExclusiveWrite16(std::size_t core_index, VAddr vaddr, u16 value) { - return monitor.DoExclusiveOperation(core_index, vaddr, 2, - [&] { memory.Write16(vaddr, value); }); -} - -bool DynarmicExclusiveMonitor::ExclusiveWrite32(std::size_t core_index, VAddr vaddr, u32 value) { - return monitor.DoExclusiveOperation(core_index, vaddr, 4, - [&] { memory.Write32(vaddr, value); }); -} - -bool DynarmicExclusiveMonitor::ExclusiveWrite64(std::size_t core_index, VAddr vaddr, u64 value) { - return monitor.DoExclusiveOperation(core_index, vaddr, 8, - [&] { memory.Write64(vaddr, value); }); -} - -bool DynarmicExclusiveMonitor::ExclusiveWrite128(std::size_t core_index, VAddr vaddr, u128 value) { - return monitor.DoExclusiveOperation(core_index, vaddr, 16, [&] { - memory.Write64(vaddr + 0, value[0]); - memory.Write64(vaddr + 8, value[1]); - }); -} - } // namespace Core diff --git a/src/core/arm/dynarmic/arm_dynarmic_64.h b/src/core/arm/dynarmic/arm_dynarmic_64.h index 647cecaf0..403c55961 100644 --- a/src/core/arm/dynarmic/arm_dynarmic_64.h +++ b/src/core/arm/dynarmic/arm_dynarmic_64.h @@ -8,7 +8,6 @@ #include <unordered_map> #include <dynarmic/A64/a64.h> -#include <dynarmic/A64/exclusive_monitor.h> #include "common/common_types.h" #include "common/hash.h" #include "core/arm/arm_interface.h" @@ -22,12 +21,14 @@ class Memory; namespace Core { class DynarmicCallbacks64; +class CPUInterruptHandler; class DynarmicExclusiveMonitor; class System; class ARM_Dynarmic_64 final : public ARM_Interface { public: - ARM_Dynarmic_64(System& system, ExclusiveMonitor& exclusive_monitor, std::size_t core_index); + ARM_Dynarmic_64(System& system, CPUInterrupts& interrupt_handlers, bool uses_wall_clock, + ExclusiveMonitor& exclusive_monitor, std::size_t core_index); ~ARM_Dynarmic_64() override; void SetPC(u64 pc) override; @@ -44,6 +45,7 @@ public: void SetTlsAddress(VAddr address) override; void SetTPIDR_EL0(u64 value) override; u64 GetTPIDR_EL0() const override; + void ChangeProcessorID(std::size_t new_core_id) override; void SaveContext(ThreadContext32& ctx) override {} void SaveContext(ThreadContext64& ctx) override; @@ -75,24 +77,4 @@ private: DynarmicExclusiveMonitor& exclusive_monitor; }; -class DynarmicExclusiveMonitor final : public ExclusiveMonitor { -public: - explicit DynarmicExclusiveMonitor(Memory::Memory& memory, std::size_t core_count); - ~DynarmicExclusiveMonitor() override; - - void SetExclusive(std::size_t core_index, VAddr addr) override; - void ClearExclusive() override; - - bool ExclusiveWrite8(std::size_t core_index, VAddr vaddr, u8 value) override; - bool ExclusiveWrite16(std::size_t core_index, VAddr vaddr, u16 value) override; - bool ExclusiveWrite32(std::size_t core_index, VAddr vaddr, u32 value) override; - bool ExclusiveWrite64(std::size_t core_index, VAddr vaddr, u64 value) override; - bool ExclusiveWrite128(std::size_t core_index, VAddr vaddr, u128 value) override; - -private: - friend class ARM_Dynarmic_64; - Dynarmic::A64::ExclusiveMonitor monitor; - Core::Memory::Memory& memory; -}; - } // namespace Core diff --git a/src/core/arm/dynarmic/arm_dynarmic_cp15.cpp b/src/core/arm/dynarmic/arm_dynarmic_cp15.cpp index 3fdcdebde..54556e0f9 100644 --- a/src/core/arm/dynarmic/arm_dynarmic_cp15.cpp +++ b/src/core/arm/dynarmic/arm_dynarmic_cp15.cpp @@ -2,79 +2,132 @@ // Licensed under GPLv2 or any later version // Refer to the license.txt file included. +#include <fmt/format.h> +#include "common/logging/log.h" +#include "core/arm/dynarmic/arm_dynarmic_32.h" #include "core/arm/dynarmic/arm_dynarmic_cp15.h" +#include "core/core.h" +#include "core/core_timing.h" +#include "core/core_timing_util.h" using Callback = Dynarmic::A32::Coprocessor::Callback; using CallbackOrAccessOneWord = Dynarmic::A32::Coprocessor::CallbackOrAccessOneWord; using CallbackOrAccessTwoWords = Dynarmic::A32::Coprocessor::CallbackOrAccessTwoWords; +template <> +struct fmt::formatter<Dynarmic::A32::CoprocReg> { + constexpr auto parse(format_parse_context& ctx) { + return ctx.begin(); + } + template <typename FormatContext> + auto format(const Dynarmic::A32::CoprocReg& reg, FormatContext& ctx) { + return format_to(ctx.out(), "cp{}", static_cast<size_t>(reg)); + } +}; + +namespace Core { + +static u32 dummy_value; + std::optional<Callback> DynarmicCP15::CompileInternalOperation(bool two, unsigned opc1, CoprocReg CRd, CoprocReg CRn, CoprocReg CRm, unsigned opc2) { + LOG_CRITICAL(Core_ARM, "CP15: cdp{} p15, {}, {}, {}, {}, {}", two ? "2" : "", opc1, CRd, CRn, + CRm, opc2); return {}; } CallbackOrAccessOneWord DynarmicCP15::CompileSendOneWord(bool two, unsigned opc1, CoprocReg CRn, CoprocReg CRm, unsigned opc2) { - // TODO(merry): Privileged CP15 registers - if (!two && CRn == CoprocReg::C7 && opc1 == 0 && CRm == CoprocReg::C5 && opc2 == 4) { + // CP15_FLUSH_PREFETCH_BUFFER // This is a dummy write, we ignore the value written here. - return &CP15[static_cast<std::size_t>(CP15Register::CP15_FLUSH_PREFETCH_BUFFER)]; + return &dummy_value; } if (!two && CRn == CoprocReg::C7 && opc1 == 0 && CRm == CoprocReg::C10) { switch (opc2) { case 4: + // CP15_DATA_SYNC_BARRIER // This is a dummy write, we ignore the value written here. - return &CP15[static_cast<std::size_t>(CP15Register::CP15_DATA_SYNC_BARRIER)]; + return &dummy_value; case 5: + // CP15_DATA_MEMORY_BARRIER // This is a dummy write, we ignore the value written here. - return &CP15[static_cast<std::size_t>(CP15Register::CP15_DATA_MEMORY_BARRIER)]; - default: - return {}; + return &dummy_value; } } if (!two && CRn == CoprocReg::C13 && opc1 == 0 && CRm == CoprocReg::C0 && opc2 == 2) { - return &CP15[static_cast<std::size_t>(CP15Register::CP15_THREAD_UPRW)]; + // CP15_THREAD_UPRW + return &uprw; } + LOG_CRITICAL(Core_ARM, "CP15: mcr{} p15, {}, <Rt>, {}, {}, {}", two ? "2" : "", opc1, CRn, CRm, + opc2); return {}; } CallbackOrAccessTwoWords DynarmicCP15::CompileSendTwoWords(bool two, unsigned opc, CoprocReg CRm) { + LOG_CRITICAL(Core_ARM, "CP15: mcrr{} p15, {}, <Rt>, <Rt2>, {}", two ? "2" : "", opc, CRm); return {}; } CallbackOrAccessOneWord DynarmicCP15::CompileGetOneWord(bool two, unsigned opc1, CoprocReg CRn, CoprocReg CRm, unsigned opc2) { - // TODO(merry): Privileged CP15 registers - if (!two && CRn == CoprocReg::C13 && opc1 == 0 && CRm == CoprocReg::C0) { switch (opc2) { case 2: - return &CP15[static_cast<std::size_t>(CP15Register::CP15_THREAD_UPRW)]; + // CP15_THREAD_UPRW + return &uprw; case 3: - return &CP15[static_cast<std::size_t>(CP15Register::CP15_THREAD_URO)]; - default: - return {}; + // CP15_THREAD_URO + return &uro; } } + LOG_CRITICAL(Core_ARM, "CP15: mrc{} p15, {}, <Rt>, {}, {}, {}", two ? "2" : "", opc1, CRn, CRm, + opc2); return {}; } CallbackOrAccessTwoWords DynarmicCP15::CompileGetTwoWords(bool two, unsigned opc, CoprocReg CRm) { + if (!two && opc == 0 && CRm == CoprocReg::C14) { + // CNTPCT + const auto callback = static_cast<u64 (*)(Dynarmic::A32::Jit*, void*, u32, u32)>( + [](Dynarmic::A32::Jit*, void* arg, u32, u32) -> u64 { + ARM_Dynarmic_32& parent = *(ARM_Dynarmic_32*)arg; + return parent.system.CoreTiming().GetClockTicks(); + }); + return Dynarmic::A32::Coprocessor::Callback{callback, (void*)&parent}; + } + + LOG_CRITICAL(Core_ARM, "CP15: mrrc{} p15, {}, <Rt>, <Rt2>, {}", two ? "2" : "", opc, CRm); return {}; } std::optional<Callback> DynarmicCP15::CompileLoadWords(bool two, bool long_transfer, CoprocReg CRd, std::optional<u8> option) { + if (option) { + LOG_CRITICAL(Core_ARM, "CP15: mrrc{}{} p15, {}, [...], {}", two ? "2" : "", + long_transfer ? "l" : "", CRd, *option); + } else { + LOG_CRITICAL(Core_ARM, "CP15: mrrc{}{} p15, {}, [...]", two ? "2" : "", + long_transfer ? "l" : "", CRd); + } return {}; } std::optional<Callback> DynarmicCP15::CompileStoreWords(bool two, bool long_transfer, CoprocReg CRd, std::optional<u8> option) { + if (option) { + LOG_CRITICAL(Core_ARM, "CP15: mrrc{}{} p15, {}, [...], {}", two ? "2" : "", + long_transfer ? "l" : "", CRd, *option); + } else { + LOG_CRITICAL(Core_ARM, "CP15: mrrc{}{} p15, {}, [...]", two ? "2" : "", + long_transfer ? "l" : "", CRd); + } return {}; } + +} // namespace Core diff --git a/src/core/arm/dynarmic/arm_dynarmic_cp15.h b/src/core/arm/dynarmic/arm_dynarmic_cp15.h index 07bcde5f9..7356d252e 100644 --- a/src/core/arm/dynarmic/arm_dynarmic_cp15.h +++ b/src/core/arm/dynarmic/arm_dynarmic_cp15.h @@ -10,128 +10,15 @@ #include <dynarmic/A32/coprocessor.h> #include "common/common_types.h" -enum class CP15Register { - // c0 - Information registers - CP15_MAIN_ID, - CP15_CACHE_TYPE, - CP15_TCM_STATUS, - CP15_TLB_TYPE, - CP15_CPU_ID, - CP15_PROCESSOR_FEATURE_0, - CP15_PROCESSOR_FEATURE_1, - CP15_DEBUG_FEATURE_0, - CP15_AUXILIARY_FEATURE_0, - CP15_MEMORY_MODEL_FEATURE_0, - CP15_MEMORY_MODEL_FEATURE_1, - CP15_MEMORY_MODEL_FEATURE_2, - CP15_MEMORY_MODEL_FEATURE_3, - CP15_ISA_FEATURE_0, - CP15_ISA_FEATURE_1, - CP15_ISA_FEATURE_2, - CP15_ISA_FEATURE_3, - CP15_ISA_FEATURE_4, +namespace Core { - // c1 - Control registers - CP15_CONTROL, - CP15_AUXILIARY_CONTROL, - CP15_COPROCESSOR_ACCESS_CONTROL, - - // c2 - Translation table registers - CP15_TRANSLATION_BASE_TABLE_0, - CP15_TRANSLATION_BASE_TABLE_1, - CP15_TRANSLATION_BASE_CONTROL, - CP15_DOMAIN_ACCESS_CONTROL, - CP15_RESERVED, - - // c5 - Fault status registers - CP15_FAULT_STATUS, - CP15_INSTR_FAULT_STATUS, - CP15_COMBINED_DATA_FSR = CP15_FAULT_STATUS, - CP15_INST_FSR, - - // c6 - Fault Address registers - CP15_FAULT_ADDRESS, - CP15_COMBINED_DATA_FAR = CP15_FAULT_ADDRESS, - CP15_WFAR, - CP15_IFAR, - - // c7 - Cache operation registers - CP15_WAIT_FOR_INTERRUPT, - CP15_PHYS_ADDRESS, - CP15_INVALIDATE_INSTR_CACHE, - CP15_INVALIDATE_INSTR_CACHE_USING_MVA, - CP15_INVALIDATE_INSTR_CACHE_USING_INDEX, - CP15_FLUSH_PREFETCH_BUFFER, - CP15_FLUSH_BRANCH_TARGET_CACHE, - CP15_FLUSH_BRANCH_TARGET_CACHE_ENTRY, - CP15_INVALIDATE_DATA_CACHE, - CP15_INVALIDATE_DATA_CACHE_LINE_USING_MVA, - CP15_INVALIDATE_DATA_CACHE_LINE_USING_INDEX, - CP15_INVALIDATE_DATA_AND_INSTR_CACHE, - CP15_CLEAN_DATA_CACHE, - CP15_CLEAN_DATA_CACHE_LINE_USING_MVA, - CP15_CLEAN_DATA_CACHE_LINE_USING_INDEX, - CP15_DATA_SYNC_BARRIER, - CP15_DATA_MEMORY_BARRIER, - CP15_CLEAN_AND_INVALIDATE_DATA_CACHE, - CP15_CLEAN_AND_INVALIDATE_DATA_CACHE_LINE_USING_MVA, - CP15_CLEAN_AND_INVALIDATE_DATA_CACHE_LINE_USING_INDEX, - - // c8 - TLB operations - CP15_INVALIDATE_ITLB, - CP15_INVALIDATE_ITLB_SINGLE_ENTRY, - CP15_INVALIDATE_ITLB_ENTRY_ON_ASID_MATCH, - CP15_INVALIDATE_ITLB_ENTRY_ON_MVA, - CP15_INVALIDATE_DTLB, - CP15_INVALIDATE_DTLB_SINGLE_ENTRY, - CP15_INVALIDATE_DTLB_ENTRY_ON_ASID_MATCH, - CP15_INVALIDATE_DTLB_ENTRY_ON_MVA, - CP15_INVALIDATE_UTLB, - CP15_INVALIDATE_UTLB_SINGLE_ENTRY, - CP15_INVALIDATE_UTLB_ENTRY_ON_ASID_MATCH, - CP15_INVALIDATE_UTLB_ENTRY_ON_MVA, - - // c9 - Data cache lockdown register - CP15_DATA_CACHE_LOCKDOWN, - - // c10 - TLB/Memory map registers - CP15_TLB_LOCKDOWN, - CP15_PRIMARY_REGION_REMAP, - CP15_NORMAL_REGION_REMAP, - - // c13 - Thread related registers - CP15_PID, - CP15_CONTEXT_ID, - CP15_THREAD_UPRW, // Thread ID register - User/Privileged Read/Write - CP15_THREAD_URO, // Thread ID register - User Read Only (Privileged R/W) - CP15_THREAD_PRW, // Thread ID register - Privileged R/W only. - - // c15 - Performance and TLB lockdown registers - CP15_PERFORMANCE_MONITOR_CONTROL, - CP15_CYCLE_COUNTER, - CP15_COUNT_0, - CP15_COUNT_1, - CP15_READ_MAIN_TLB_LOCKDOWN_ENTRY, - CP15_WRITE_MAIN_TLB_LOCKDOWN_ENTRY, - CP15_MAIN_TLB_LOCKDOWN_VIRT_ADDRESS, - CP15_MAIN_TLB_LOCKDOWN_PHYS_ADDRESS, - CP15_MAIN_TLB_LOCKDOWN_ATTRIBUTE, - CP15_TLB_DEBUG_CONTROL, - - // Skyeye defined - CP15_TLB_FAULT_ADDR, - CP15_TLB_FAULT_STATUS, - - // Not an actual register. - // All registers should be defined above this. - CP15_REGISTER_COUNT, -}; +class ARM_Dynarmic_32; class DynarmicCP15 final : public Dynarmic::A32::Coprocessor { public: using CoprocReg = Dynarmic::A32::CoprocReg; - explicit DynarmicCP15(u32* cp15) : CP15(cp15){}; + explicit DynarmicCP15(ARM_Dynarmic_32& parent) : parent(parent) {} std::optional<Callback> CompileInternalOperation(bool two, unsigned opc1, CoprocReg CRd, CoprocReg CRn, CoprocReg CRm, @@ -147,6 +34,9 @@ public: std::optional<Callback> CompileStoreWords(bool two, bool long_transfer, CoprocReg CRd, std::optional<u8> option) override; -private: - u32* CP15{}; + ARM_Dynarmic_32& parent; + u32 uprw; + u32 uro; }; + +} // namespace Core diff --git a/src/core/arm/dynarmic/arm_exclusive_monitor.cpp b/src/core/arm/dynarmic/arm_exclusive_monitor.cpp new file mode 100644 index 000000000..4e209f6a5 --- /dev/null +++ b/src/core/arm/dynarmic/arm_exclusive_monitor.cpp @@ -0,0 +1,76 @@ +// Copyright 2018 yuzu emulator team +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include <cinttypes> +#include <memory> +#include "core/arm/dynarmic/arm_exclusive_monitor.h" +#include "core/memory.h" + +namespace Core { + +DynarmicExclusiveMonitor::DynarmicExclusiveMonitor(Memory::Memory& memory, std::size_t core_count) + : monitor(core_count), memory{memory} {} + +DynarmicExclusiveMonitor::~DynarmicExclusiveMonitor() = default; + +u8 DynarmicExclusiveMonitor::ExclusiveRead8(std::size_t core_index, VAddr addr) { + return monitor.ReadAndMark<u8>(core_index, addr, [&]() -> u8 { return memory.Read8(addr); }); +} + +u16 DynarmicExclusiveMonitor::ExclusiveRead16(std::size_t core_index, VAddr addr) { + return monitor.ReadAndMark<u16>(core_index, addr, [&]() -> u16 { return memory.Read16(addr); }); +} + +u32 DynarmicExclusiveMonitor::ExclusiveRead32(std::size_t core_index, VAddr addr) { + return monitor.ReadAndMark<u32>(core_index, addr, [&]() -> u32 { return memory.Read32(addr); }); +} + +u64 DynarmicExclusiveMonitor::ExclusiveRead64(std::size_t core_index, VAddr addr) { + return monitor.ReadAndMark<u64>(core_index, addr, [&]() -> u64 { return memory.Read64(addr); }); +} + +u128 DynarmicExclusiveMonitor::ExclusiveRead128(std::size_t core_index, VAddr addr) { + return monitor.ReadAndMark<u128>(core_index, addr, [&]() -> u128 { + u128 result; + result[0] = memory.Read64(addr); + result[1] = memory.Read64(addr + 8); + return result; + }); +} + +void DynarmicExclusiveMonitor::ClearExclusive() { + monitor.Clear(); +} + +bool DynarmicExclusiveMonitor::ExclusiveWrite8(std::size_t core_index, VAddr vaddr, u8 value) { + return monitor.DoExclusiveOperation<u8>(core_index, vaddr, [&](u8 expected) -> bool { + return memory.WriteExclusive8(vaddr, value, expected); + }); +} + +bool DynarmicExclusiveMonitor::ExclusiveWrite16(std::size_t core_index, VAddr vaddr, u16 value) { + return monitor.DoExclusiveOperation<u16>(core_index, vaddr, [&](u16 expected) -> bool { + return memory.WriteExclusive16(vaddr, value, expected); + }); +} + +bool DynarmicExclusiveMonitor::ExclusiveWrite32(std::size_t core_index, VAddr vaddr, u32 value) { + return monitor.DoExclusiveOperation<u32>(core_index, vaddr, [&](u32 expected) -> bool { + return memory.WriteExclusive32(vaddr, value, expected); + }); +} + +bool DynarmicExclusiveMonitor::ExclusiveWrite64(std::size_t core_index, VAddr vaddr, u64 value) { + return monitor.DoExclusiveOperation<u64>(core_index, vaddr, [&](u64 expected) -> bool { + return memory.WriteExclusive64(vaddr, value, expected); + }); +} + +bool DynarmicExclusiveMonitor::ExclusiveWrite128(std::size_t core_index, VAddr vaddr, u128 value) { + return monitor.DoExclusiveOperation<u128>(core_index, vaddr, [&](u128 expected) -> bool { + return memory.WriteExclusive128(vaddr, value, expected); + }); +} + +} // namespace Core diff --git a/src/core/arm/dynarmic/arm_exclusive_monitor.h b/src/core/arm/dynarmic/arm_exclusive_monitor.h new file mode 100644 index 000000000..964f4a55d --- /dev/null +++ b/src/core/arm/dynarmic/arm_exclusive_monitor.h @@ -0,0 +1,48 @@ +// Copyright 2020 yuzu emulator team +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include <memory> +#include <unordered_map> + +#include <dynarmic/exclusive_monitor.h> + +#include "common/common_types.h" +#include "core/arm/dynarmic/arm_dynarmic_32.h" +#include "core/arm/dynarmic/arm_dynarmic_64.h" +#include "core/arm/exclusive_monitor.h" + +namespace Core::Memory { +class Memory; +} + +namespace Core { + +class DynarmicExclusiveMonitor final : public ExclusiveMonitor { +public: + explicit DynarmicExclusiveMonitor(Memory::Memory& memory, std::size_t core_count); + ~DynarmicExclusiveMonitor() override; + + u8 ExclusiveRead8(std::size_t core_index, VAddr addr) override; + u16 ExclusiveRead16(std::size_t core_index, VAddr addr) override; + u32 ExclusiveRead32(std::size_t core_index, VAddr addr) override; + u64 ExclusiveRead64(std::size_t core_index, VAddr addr) override; + u128 ExclusiveRead128(std::size_t core_index, VAddr addr) override; + void ClearExclusive() override; + + bool ExclusiveWrite8(std::size_t core_index, VAddr vaddr, u8 value) override; + bool ExclusiveWrite16(std::size_t core_index, VAddr vaddr, u16 value) override; + bool ExclusiveWrite32(std::size_t core_index, VAddr vaddr, u32 value) override; + bool ExclusiveWrite64(std::size_t core_index, VAddr vaddr, u64 value) override; + bool ExclusiveWrite128(std::size_t core_index, VAddr vaddr, u128 value) override; + +private: + friend class ARM_Dynarmic_32; + friend class ARM_Dynarmic_64; + Dynarmic::ExclusiveMonitor monitor; + Core::Memory::Memory& memory; +}; + +} // namespace Core diff --git a/src/core/arm/exclusive_monitor.cpp b/src/core/arm/exclusive_monitor.cpp index b32401e0b..d8cba369d 100644 --- a/src/core/arm/exclusive_monitor.cpp +++ b/src/core/arm/exclusive_monitor.cpp @@ -3,7 +3,7 @@ // Refer to the license.txt file included. #ifdef ARCHITECTURE_x86_64 -#include "core/arm/dynarmic/arm_dynarmic_64.h" +#include "core/arm/dynarmic/arm_exclusive_monitor.h" #endif #include "core/arm/exclusive_monitor.h" #include "core/memory.h" diff --git a/src/core/arm/exclusive_monitor.h b/src/core/arm/exclusive_monitor.h index ccd73b80f..62f6e6023 100644 --- a/src/core/arm/exclusive_monitor.h +++ b/src/core/arm/exclusive_monitor.h @@ -18,7 +18,11 @@ class ExclusiveMonitor { public: virtual ~ExclusiveMonitor(); - virtual void SetExclusive(std::size_t core_index, VAddr addr) = 0; + virtual u8 ExclusiveRead8(std::size_t core_index, VAddr addr) = 0; + virtual u16 ExclusiveRead16(std::size_t core_index, VAddr addr) = 0; + virtual u32 ExclusiveRead32(std::size_t core_index, VAddr addr) = 0; + virtual u64 ExclusiveRead64(std::size_t core_index, VAddr addr) = 0; + virtual u128 ExclusiveRead128(std::size_t core_index, VAddr addr) = 0; virtual void ClearExclusive() = 0; virtual bool ExclusiveWrite8(std::size_t core_index, VAddr vaddr, u8 value) = 0; diff --git a/src/core/arm/unicorn/arm_unicorn.cpp b/src/core/arm/unicorn/arm_unicorn.cpp index e40e9626a..1df3f3ed1 100644 --- a/src/core/arm/unicorn/arm_unicorn.cpp +++ b/src/core/arm/unicorn/arm_unicorn.cpp @@ -6,6 +6,7 @@ #include <unicorn/arm64.h> #include "common/assert.h" #include "common/microprofile.h" +#include "core/arm/cpu_interrupt_handler.h" #include "core/arm/unicorn/arm_unicorn.h" #include "core/core.h" #include "core/core_timing.h" @@ -62,7 +63,9 @@ static bool UnmappedMemoryHook(uc_engine* uc, uc_mem_type type, u64 addr, int si return false; } -ARM_Unicorn::ARM_Unicorn(System& system, Arch architecture) : ARM_Interface{system} { +ARM_Unicorn::ARM_Unicorn(System& system, CPUInterrupts& interrupt_handlers, bool uses_wall_clock, + Arch architecture, std::size_t core_index) + : ARM_Interface{system, interrupt_handlers, uses_wall_clock}, core_index{core_index} { const auto arch = architecture == Arch::AArch32 ? UC_ARCH_ARM : UC_ARCH_ARM64; CHECKED(uc_open(arch, UC_MODE_ARM, &uc)); @@ -156,12 +159,20 @@ void ARM_Unicorn::SetTPIDR_EL0(u64 value) { CHECKED(uc_reg_write(uc, UC_ARM64_REG_TPIDR_EL0, &value)); } +void ARM_Unicorn::ChangeProcessorID(std::size_t new_core_id) { + core_index = new_core_id; +} + void ARM_Unicorn::Run() { if (GDBStub::IsServerEnabled()) { ExecuteInstructions(std::max(4000000U, 0U)); } else { - ExecuteInstructions( - std::max(std::size_t(system.CoreTiming().GetDowncount()), std::size_t{0})); + while (true) { + if (interrupt_handlers[core_index].IsInterrupted()) { + return; + } + ExecuteInstructions(10); + } } } @@ -183,8 +194,6 @@ void ARM_Unicorn::ExecuteInstructions(std::size_t num_instructions) { UC_PROT_READ | UC_PROT_WRITE | UC_PROT_EXEC, page_buffer.data())); CHECKED(uc_emu_start(uc, GetPC(), 1ULL << 63, 0, num_instructions)); CHECKED(uc_mem_unmap(uc, map_addr, page_buffer.size())); - - system.CoreTiming().AddTicks(num_instructions); if (GDBStub::IsServerEnabled()) { if (last_bkpt_hit && last_bkpt.type == GDBStub::BreakpointType::Execute) { uc_reg_write(uc, UC_ARM64_REG_PC, &last_bkpt.address); diff --git a/src/core/arm/unicorn/arm_unicorn.h b/src/core/arm/unicorn/arm_unicorn.h index 725c65085..810aff311 100644 --- a/src/core/arm/unicorn/arm_unicorn.h +++ b/src/core/arm/unicorn/arm_unicorn.h @@ -20,7 +20,8 @@ public: AArch64, // 64-bit ARM }; - explicit ARM_Unicorn(System& system, Arch architecture); + explicit ARM_Unicorn(System& system, CPUInterrupts& interrupt_handlers, bool uses_wall_clock, + Arch architecture, std::size_t core_index); ~ARM_Unicorn() override; void SetPC(u64 pc) override; @@ -35,6 +36,7 @@ public: void SetTlsAddress(VAddr address) override; void SetTPIDR_EL0(u64 value) override; u64 GetTPIDR_EL0() const override; + void ChangeProcessorID(std::size_t new_core_id) override; void PrepareReschedule() override; void ClearExclusiveState() override; void ExecuteInstructions(std::size_t num_instructions); @@ -55,6 +57,7 @@ private: uc_engine* uc{}; GDBStub::BreakpointAddress last_bkpt{}; bool last_bkpt_hit = false; + std::size_t core_index; }; } // namespace Core diff --git a/src/core/core.cpp b/src/core/core.cpp index f9f8a3000..1a243c515 100644 --- a/src/core/core.cpp +++ b/src/core/core.cpp @@ -8,10 +8,10 @@ #include "common/file_util.h" #include "common/logging/log.h" +#include "common/microprofile.h" #include "common/string_util.h" #include "core/arm/exclusive_monitor.h" #include "core/core.h" -#include "core/core_manager.h" #include "core/core_timing.h" #include "core/cpu_manager.h" #include "core/device_memory.h" @@ -51,6 +51,11 @@ #include "video_core/renderer_base.h" #include "video_core/video_core.h" +MICROPROFILE_DEFINE(ARM_Jit_Dynarmic_CPU0, "ARM JIT", "Dynarmic CPU 0", MP_RGB(255, 64, 64)); +MICROPROFILE_DEFINE(ARM_Jit_Dynarmic_CPU1, "ARM JIT", "Dynarmic CPU 1", MP_RGB(255, 64, 64)); +MICROPROFILE_DEFINE(ARM_Jit_Dynarmic_CPU2, "ARM JIT", "Dynarmic CPU 2", MP_RGB(255, 64, 64)); +MICROPROFILE_DEFINE(ARM_Jit_Dynarmic_CPU3, "ARM JIT", "Dynarmic CPU 3", MP_RGB(255, 64, 64)); + namespace Core { namespace { @@ -117,23 +122,22 @@ struct System::Impl { : kernel{system}, fs_controller{system}, memory{system}, cpu_manager{system}, reporter{system}, applet_manager{system} {} - CoreManager& CurrentCoreManager() { - return cpu_manager.GetCurrentCoreManager(); - } + ResultStatus Run() { + status = ResultStatus::Success; - Kernel::PhysicalCore& CurrentPhysicalCore() { - const auto index = cpu_manager.GetActiveCoreIndex(); - return kernel.PhysicalCore(index); - } + kernel.Suspend(false); + core_timing.SyncPause(false); + cpu_manager.Pause(false); - Kernel::PhysicalCore& GetPhysicalCore(std::size_t index) { - return kernel.PhysicalCore(index); + return status; } - ResultStatus RunLoop(bool tight_loop) { + ResultStatus Pause() { status = ResultStatus::Success; - cpu_manager.RunLoop(tight_loop); + core_timing.SyncPause(true); + kernel.Suspend(true); + cpu_manager.Pause(true); return status; } @@ -143,7 +147,15 @@ struct System::Impl { device_memory = std::make_unique<Core::DeviceMemory>(system); - core_timing.Initialize(); + is_multicore = Settings::values.use_multi_core; + is_async_gpu = is_multicore || Settings::values.use_asynchronous_gpu_emulation; + + kernel.SetMulticore(is_multicore); + cpu_manager.SetMulticore(is_multicore); + cpu_manager.SetAsyncGpu(is_async_gpu); + core_timing.SetMulticore(is_multicore); + + core_timing.Initialize([&system]() { system.RegisterHostThread(); }); kernel.Initialize(); cpu_manager.Initialize(); @@ -180,6 +192,11 @@ struct System::Impl { is_powered_on = true; exit_lock = false; + microprofile_dynarmic[0] = MICROPROFILE_TOKEN(ARM_Jit_Dynarmic_CPU0); + microprofile_dynarmic[1] = MICROPROFILE_TOKEN(ARM_Jit_Dynarmic_CPU1); + microprofile_dynarmic[2] = MICROPROFILE_TOKEN(ARM_Jit_Dynarmic_CPU2); + microprofile_dynarmic[3] = MICROPROFILE_TOKEN(ARM_Jit_Dynarmic_CPU3); + LOG_DEBUG(Core, "Initialized OK"); return ResultStatus::Success; @@ -277,8 +294,6 @@ struct System::Impl { service_manager.reset(); cheat_engine.reset(); telemetry_session.reset(); - perf_stats.reset(); - gpu_core.reset(); device_memory.reset(); // Close all CPU/threading state @@ -290,6 +305,8 @@ struct System::Impl { // Close app loader app_loader.reset(); + gpu_core.reset(); + perf_stats.reset(); // Clear all applets applet_manager.ClearAll(); @@ -382,25 +399,35 @@ struct System::Impl { std::unique_ptr<Core::PerfStats> perf_stats; Core::FrameLimiter frame_limiter; + + bool is_multicore{}; + bool is_async_gpu{}; + + std::array<u64, Core::Hardware::NUM_CPU_CORES> dynarmic_ticks{}; + std::array<MicroProfileToken, Core::Hardware::NUM_CPU_CORES> microprofile_dynarmic{}; }; System::System() : impl{std::make_unique<Impl>(*this)} {} System::~System() = default; -CoreManager& System::CurrentCoreManager() { - return impl->CurrentCoreManager(); +CpuManager& System::GetCpuManager() { + return impl->cpu_manager; +} + +const CpuManager& System::GetCpuManager() const { + return impl->cpu_manager; } -const CoreManager& System::CurrentCoreManager() const { - return impl->CurrentCoreManager(); +System::ResultStatus System::Run() { + return impl->Run(); } -System::ResultStatus System::RunLoop(bool tight_loop) { - return impl->RunLoop(tight_loop); +System::ResultStatus System::Pause() { + return impl->Pause(); } System::ResultStatus System::SingleStep() { - return RunLoop(false); + return ResultStatus::Success; } void System::InvalidateCpuInstructionCaches() { @@ -416,7 +443,7 @@ bool System::IsPoweredOn() const { } void System::PrepareReschedule() { - impl->CurrentPhysicalCore().Stop(); + // Deprecated, does nothing, kept for backward compatibility. } void System::PrepareReschedule(const u32 core_index) { @@ -436,31 +463,41 @@ const TelemetrySession& System::TelemetrySession() const { } ARM_Interface& System::CurrentArmInterface() { - return impl->CurrentPhysicalCore().ArmInterface(); + return impl->kernel.CurrentScheduler().GetCurrentThread()->ArmInterface(); } const ARM_Interface& System::CurrentArmInterface() const { - return impl->CurrentPhysicalCore().ArmInterface(); + return impl->kernel.CurrentScheduler().GetCurrentThread()->ArmInterface(); } std::size_t System::CurrentCoreIndex() const { - return impl->cpu_manager.GetActiveCoreIndex(); + std::size_t core = impl->kernel.GetCurrentHostThreadID(); + ASSERT(core < Core::Hardware::NUM_CPU_CORES); + return core; } Kernel::Scheduler& System::CurrentScheduler() { - return impl->CurrentPhysicalCore().Scheduler(); + return impl->kernel.CurrentScheduler(); } const Kernel::Scheduler& System::CurrentScheduler() const { - return impl->CurrentPhysicalCore().Scheduler(); + return impl->kernel.CurrentScheduler(); +} + +Kernel::PhysicalCore& System::CurrentPhysicalCore() { + return impl->kernel.CurrentPhysicalCore(); +} + +const Kernel::PhysicalCore& System::CurrentPhysicalCore() const { + return impl->kernel.CurrentPhysicalCore(); } Kernel::Scheduler& System::Scheduler(std::size_t core_index) { - return impl->GetPhysicalCore(core_index).Scheduler(); + return impl->kernel.Scheduler(core_index); } const Kernel::Scheduler& System::Scheduler(std::size_t core_index) const { - return impl->GetPhysicalCore(core_index).Scheduler(); + return impl->kernel.Scheduler(core_index); } /// Gets the global scheduler @@ -490,20 +527,15 @@ const Kernel::Process* System::CurrentProcess() const { } ARM_Interface& System::ArmInterface(std::size_t core_index) { - return impl->GetPhysicalCore(core_index).ArmInterface(); + auto* thread = impl->kernel.Scheduler(core_index).GetCurrentThread(); + ASSERT(thread && !thread->IsHLEThread()); + return thread->ArmInterface(); } const ARM_Interface& System::ArmInterface(std::size_t core_index) const { - return impl->GetPhysicalCore(core_index).ArmInterface(); -} - -CoreManager& System::GetCoreManager(std::size_t core_index) { - return impl->cpu_manager.GetCoreManager(core_index); -} - -const CoreManager& System::GetCoreManager(std::size_t core_index) const { - ASSERT(core_index < NUM_CPU_CORES); - return impl->cpu_manager.GetCoreManager(core_index); + auto* thread = impl->kernel.Scheduler(core_index).GetCurrentThread(); + ASSERT(thread && !thread->IsHLEThread()); + return thread->ArmInterface(); } ExclusiveMonitor& System::Monitor() { @@ -722,4 +754,18 @@ void System::RegisterHostThread() { impl->kernel.RegisterHostThread(); } +void System::EnterDynarmicProfile() { + std::size_t core = impl->kernel.GetCurrentHostThreadID(); + impl->dynarmic_ticks[core] = MicroProfileEnter(impl->microprofile_dynarmic[core]); +} + +void System::ExitDynarmicProfile() { + std::size_t core = impl->kernel.GetCurrentHostThreadID(); + MicroProfileLeave(impl->microprofile_dynarmic[core], impl->dynarmic_ticks[core]); +} + +bool System::IsMulticore() const { + return impl->is_multicore; +} + } // namespace Core diff --git a/src/core/core.h b/src/core/core.h index acc53d6a1..5c6cfbffe 100644 --- a/src/core/core.h +++ b/src/core/core.h @@ -27,6 +27,7 @@ class VfsFilesystem; namespace Kernel { class GlobalScheduler; class KernelCore; +class PhysicalCore; class Process; class Scheduler; } // namespace Kernel @@ -90,7 +91,7 @@ class InterruptManager; namespace Core { class ARM_Interface; -class CoreManager; +class CpuManager; class DeviceMemory; class ExclusiveMonitor; class FrameLimiter; @@ -136,16 +137,16 @@ public: }; /** - * Run the core CPU loop - * This function runs the core for the specified number of CPU instructions before trying to - * update hardware. This is much faster than SingleStep (and should be equivalent), as the CPU - * is not required to do a full dispatch with each instruction. NOTE: the number of instructions - * requested is not guaranteed to run, as this will be interrupted preemptively if a hardware - * update is requested (e.g. on a thread switch). - * @param tight_loop If false, the CPU single-steps. - * @return Result status, indicating whether or not the operation succeeded. + * Run the OS and Application + * This function will start emulation and run the relevant devices + */ + ResultStatus Run(); + + /** + * Pause the OS and Application + * This function will pause emulation and stop the relevant devices */ - ResultStatus RunLoop(bool tight_loop = true); + ResultStatus Pause(); /** * Step the CPU one instruction @@ -209,17 +210,21 @@ public: /// Gets the scheduler for the CPU core that is currently running const Kernel::Scheduler& CurrentScheduler() const; + /// Gets the physical core for the CPU core that is currently running + Kernel::PhysicalCore& CurrentPhysicalCore(); + + /// Gets the physical core for the CPU core that is currently running + const Kernel::PhysicalCore& CurrentPhysicalCore() const; + /// Gets a reference to an ARM interface for the CPU core with the specified index ARM_Interface& ArmInterface(std::size_t core_index); /// Gets a const reference to an ARM interface from the CPU core with the specified index const ARM_Interface& ArmInterface(std::size_t core_index) const; - /// Gets a CPU interface to the CPU core with the specified index - CoreManager& GetCoreManager(std::size_t core_index); + CpuManager& GetCpuManager(); - /// Gets a CPU interface to the CPU core with the specified index - const CoreManager& GetCoreManager(std::size_t core_index) const; + const CpuManager& GetCpuManager() const; /// Gets a reference to the exclusive monitor ExclusiveMonitor& Monitor(); @@ -370,14 +375,17 @@ public: /// Register a host thread as an auxiliary thread. void RegisterHostThread(); -private: - System(); + /// Enter Dynarmic Microprofile + void EnterDynarmicProfile(); + + /// Exit Dynarmic Microprofile + void ExitDynarmicProfile(); - /// Returns the currently running CPU core - CoreManager& CurrentCoreManager(); + /// Tells if system is running on multicore. + bool IsMulticore() const; - /// Returns the currently running CPU core - const CoreManager& CurrentCoreManager() const; +private: + System(); /** * Initialize the emulated system. diff --git a/src/core/core_manager.cpp b/src/core/core_manager.cpp deleted file mode 100644 index b6b797c80..000000000 --- a/src/core/core_manager.cpp +++ /dev/null @@ -1,67 +0,0 @@ -// Copyright 2018 yuzu emulator team -// Licensed under GPLv2 or any later version -// Refer to the license.txt file included. - -#include <condition_variable> -#include <mutex> - -#include "common/logging/log.h" -#include "core/arm/exclusive_monitor.h" -#include "core/arm/unicorn/arm_unicorn.h" -#include "core/core.h" -#include "core/core_manager.h" -#include "core/core_timing.h" -#include "core/hle/kernel/kernel.h" -#include "core/hle/kernel/physical_core.h" -#include "core/hle/kernel/scheduler.h" -#include "core/hle/kernel/thread.h" -#include "core/hle/lock.h" -#include "core/settings.h" - -namespace Core { - -CoreManager::CoreManager(System& system, std::size_t core_index) - : global_scheduler{system.GlobalScheduler()}, physical_core{system.Kernel().PhysicalCore( - core_index)}, - core_timing{system.CoreTiming()}, core_index{core_index} {} - -CoreManager::~CoreManager() = default; - -void CoreManager::RunLoop(bool tight_loop) { - Reschedule(); - - // If we don't have a currently active thread then don't execute instructions, - // instead advance to the next event and try to yield to the next thread - if (Kernel::GetCurrentThread() == nullptr) { - LOG_TRACE(Core, "Core-{} idling", core_index); - core_timing.Idle(); - } else { - if (tight_loop) { - physical_core.Run(); - } else { - physical_core.Step(); - } - } - core_timing.Advance(); - - Reschedule(); -} - -void CoreManager::SingleStep() { - return RunLoop(false); -} - -void CoreManager::PrepareReschedule() { - physical_core.Stop(); -} - -void CoreManager::Reschedule() { - // Lock the global kernel mutex when we manipulate the HLE state - std::lock_guard lock(HLE::g_hle_lock); - - global_scheduler.SelectThread(core_index); - - physical_core.Scheduler().TryDoContextSwitch(); -} - -} // namespace Core diff --git a/src/core/core_manager.h b/src/core/core_manager.h deleted file mode 100644 index d525de00a..000000000 --- a/src/core/core_manager.h +++ /dev/null @@ -1,63 +0,0 @@ -// Copyright 2018 yuzu emulator team -// Licensed under GPLv2 or any later version -// Refer to the license.txt file included. - -#pragma once - -#include <atomic> -#include <cstddef> -#include <memory> -#include "common/common_types.h" - -namespace Kernel { -class GlobalScheduler; -class PhysicalCore; -} // namespace Kernel - -namespace Core { -class System; -} - -namespace Core::Timing { -class CoreTiming; -} - -namespace Core::Memory { -class Memory; -} - -namespace Core { - -constexpr unsigned NUM_CPU_CORES{4}; - -class CoreManager { -public: - CoreManager(System& system, std::size_t core_index); - ~CoreManager(); - - void RunLoop(bool tight_loop = true); - - void SingleStep(); - - void PrepareReschedule(); - - bool IsMainCore() const { - return core_index == 0; - } - - std::size_t CoreIndex() const { - return core_index; - } - -private: - void Reschedule(); - - Kernel::GlobalScheduler& global_scheduler; - Kernel::PhysicalCore& physical_core; - Timing::CoreTiming& core_timing; - - std::atomic<bool> reschedule_pending = false; - std::size_t core_index; -}; - -} // namespace Core diff --git a/src/core/core_timing.cpp b/src/core/core_timing.cpp index 46d4178c4..5c83c41a4 100644 --- a/src/core/core_timing.cpp +++ b/src/core/core_timing.cpp @@ -1,29 +1,27 @@ -// Copyright 2008 Dolphin Emulator Project / 2017 Citra Emulator Project -// Licensed under GPLv2+ +// Copyright 2020 yuzu Emulator Project +// Licensed under GPLv2 or any later version // Refer to the license.txt file included. -#include "core/core_timing.h" - #include <algorithm> #include <mutex> #include <string> #include <tuple> #include "common/assert.h" -#include "common/thread.h" +#include "common/microprofile.h" +#include "core/core_timing.h" #include "core/core_timing_util.h" -#include "core/hardware_properties.h" namespace Core::Timing { -constexpr int MAX_SLICE_LENGTH = 10000; +constexpr u64 MAX_SLICE_LENGTH = 4000; std::shared_ptr<EventType> CreateEvent(std::string name, TimedCallback&& callback) { return std::make_shared<EventType>(std::move(callback), std::move(name)); } struct CoreTiming::Event { - s64 time; + u64 time; u64 fifo_order; u64 userdata; std::weak_ptr<EventType> type; @@ -39,51 +37,90 @@ struct CoreTiming::Event { } }; -CoreTiming::CoreTiming() = default; -CoreTiming::~CoreTiming() = default; +CoreTiming::CoreTiming() { + clock = + Common::CreateBestMatchingClock(Core::Hardware::BASE_CLOCK_RATE, Core::Hardware::CNTFREQ); +} -void CoreTiming::Initialize() { - downcounts.fill(MAX_SLICE_LENGTH); - time_slice.fill(MAX_SLICE_LENGTH); - slice_length = MAX_SLICE_LENGTH; - global_timer = 0; - idled_cycles = 0; - current_context = 0; +CoreTiming::~CoreTiming() = default; - // The time between CoreTiming being initialized and the first call to Advance() is considered - // the slice boundary between slice -1 and slice 0. Dispatcher loops must call Advance() before - // executing the first cycle of each slice to prepare the slice length and downcount for - // that slice. - is_global_timer_sane = true; +void CoreTiming::ThreadEntry(CoreTiming& instance) { + constexpr char name[] = "yuzu:HostTiming"; + MicroProfileOnThreadCreate(name); + Common::SetCurrentThreadName(name); + Common::SetCurrentThreadPriority(Common::ThreadPriority::VeryHigh); + instance.on_thread_init(); + instance.ThreadLoop(); +} +void CoreTiming::Initialize(std::function<void(void)>&& on_thread_init_) { + on_thread_init = std::move(on_thread_init_); event_fifo_id = 0; - + shutting_down = false; + ticks = 0; const auto empty_timed_callback = [](u64, s64) {}; ev_lost = CreateEvent("_lost_event", empty_timed_callback); + if (is_multicore) { + timer_thread = std::make_unique<std::thread>(ThreadEntry, std::ref(*this)); + } } void CoreTiming::Shutdown() { + paused = true; + shutting_down = true; + pause_event.Set(); + event.Set(); + if (timer_thread) { + timer_thread->join(); + } ClearPendingEvents(); + timer_thread.reset(); + has_started = false; } -void CoreTiming::ScheduleEvent(s64 cycles_into_future, const std::shared_ptr<EventType>& event_type, - u64 userdata) { - std::lock_guard guard{inner_mutex}; - const s64 timeout = GetTicks() + cycles_into_future; +void CoreTiming::Pause(bool is_paused) { + paused = is_paused; + pause_event.Set(); +} - // If this event needs to be scheduled before the next advance(), force one early - if (!is_global_timer_sane) { - ForceExceptionCheck(cycles_into_future); +void CoreTiming::SyncPause(bool is_paused) { + if (is_paused == paused && paused_set == paused) { + return; + } + Pause(is_paused); + if (timer_thread) { + if (!is_paused) { + pause_event.Set(); + } + event.Set(); + while (paused_set != is_paused) + ; } +} - event_queue.emplace_back(Event{timeout, event_fifo_id++, userdata, event_type}); +bool CoreTiming::IsRunning() const { + return !paused_set; +} - std::push_heap(event_queue.begin(), event_queue.end(), std::greater<>()); +bool CoreTiming::HasPendingEvents() const { + return !(wait_set && event_queue.empty()); } -void CoreTiming::UnscheduleEvent(const std::shared_ptr<EventType>& event_type, u64 userdata) { - std::lock_guard guard{inner_mutex}; +void CoreTiming::ScheduleEvent(s64 ns_into_future, const std::shared_ptr<EventType>& event_type, + u64 userdata) { + { + std::scoped_lock scope{basic_lock}; + const u64 timeout = static_cast<u64>(GetGlobalTimeNs().count() + ns_into_future); + + event_queue.emplace_back(Event{timeout, event_fifo_id++, userdata, event_type}); + std::push_heap(event_queue.begin(), event_queue.end(), std::greater<>()); + } + event.Set(); +} + +void CoreTiming::UnscheduleEvent(const std::shared_ptr<EventType>& event_type, u64 userdata) { + std::scoped_lock scope{basic_lock}; const auto itr = std::remove_if(event_queue.begin(), event_queue.end(), [&](const Event& e) { return e.type.lock().get() == event_type.get() && e.userdata == userdata; }); @@ -95,21 +132,39 @@ void CoreTiming::UnscheduleEvent(const std::shared_ptr<EventType>& event_type, u } } -u64 CoreTiming::GetTicks() const { - u64 ticks = static_cast<u64>(global_timer); - if (!is_global_timer_sane) { - ticks += accumulated_ticks; +void CoreTiming::AddTicks(u64 ticks) { + this->ticks += ticks; + downcount -= ticks; +} + +void CoreTiming::Idle() { + if (!event_queue.empty()) { + const u64 next_event_time = event_queue.front().time; + const u64 next_ticks = nsToCycles(std::chrono::nanoseconds(next_event_time)) + 10U; + if (next_ticks > ticks) { + ticks = next_ticks; + } + return; } - return ticks; + ticks += 1000U; } -u64 CoreTiming::GetIdleTicks() const { - return static_cast<u64>(idled_cycles); +void CoreTiming::ResetTicks() { + downcount = MAX_SLICE_LENGTH; } -void CoreTiming::AddTicks(u64 ticks) { - accumulated_ticks += ticks; - downcounts[current_context] -= static_cast<s64>(ticks); +u64 CoreTiming::GetCPUTicks() const { + if (is_multicore) { + return clock->GetCPUCycles(); + } + return ticks; +} + +u64 CoreTiming::GetClockTicks() const { + if (is_multicore) { + return clock->GetClockCycles(); + } + return CpuCyclesToClockCycles(ticks); } void CoreTiming::ClearPendingEvents() { @@ -117,7 +172,7 @@ void CoreTiming::ClearPendingEvents() { } void CoreTiming::RemoveEvent(const std::shared_ptr<EventType>& event_type) { - std::lock_guard guard{inner_mutex}; + basic_lock.lock(); const auto itr = std::remove_if(event_queue.begin(), event_queue.end(), [&](const Event& e) { return e.type.lock().get() == event_type.get(); @@ -128,99 +183,72 @@ void CoreTiming::RemoveEvent(const std::shared_ptr<EventType>& event_type) { event_queue.erase(itr, event_queue.end()); std::make_heap(event_queue.begin(), event_queue.end(), std::greater<>()); } + basic_lock.unlock(); } -void CoreTiming::ForceExceptionCheck(s64 cycles) { - cycles = std::max<s64>(0, cycles); - if (downcounts[current_context] <= cycles) { - return; - } - - // downcount is always (much) smaller than MAX_INT so we can safely cast cycles to an int - // here. Account for cycles already executed by adjusting the g.slice_length - downcounts[current_context] = static_cast<int>(cycles); -} - -std::optional<u64> CoreTiming::NextAvailableCore(const s64 needed_ticks) const { - const u64 original_context = current_context; - u64 next_context = (original_context + 1) % num_cpu_cores; - while (next_context != original_context) { - if (time_slice[next_context] >= needed_ticks) { - return {next_context}; - } else if (time_slice[next_context] >= 0) { - return std::nullopt; - } - next_context = (next_context + 1) % num_cpu_cores; - } - return std::nullopt; -} - -void CoreTiming::Advance() { - std::unique_lock<std::mutex> guard(inner_mutex); - - const u64 cycles_executed = accumulated_ticks; - time_slice[current_context] = std::max<s64>(0, time_slice[current_context] - accumulated_ticks); - global_timer += cycles_executed; - - is_global_timer_sane = true; +std::optional<s64> CoreTiming::Advance() { + std::scoped_lock advance_scope{advance_lock}; + std::scoped_lock basic_scope{basic_lock}; + global_timer = GetGlobalTimeNs().count(); while (!event_queue.empty() && event_queue.front().time <= global_timer) { Event evt = std::move(event_queue.front()); std::pop_heap(event_queue.begin(), event_queue.end(), std::greater<>()); event_queue.pop_back(); - inner_mutex.unlock(); + basic_lock.unlock(); if (auto event_type{evt.type.lock()}) { event_type->callback(evt.userdata, global_timer - evt.time); } - inner_mutex.lock(); + basic_lock.lock(); + global_timer = GetGlobalTimeNs().count(); } - is_global_timer_sane = false; - - // Still events left (scheduled in the future) if (!event_queue.empty()) { - const s64 needed_ticks = - std::min<s64>(event_queue.front().time - global_timer, MAX_SLICE_LENGTH); - const auto next_core = NextAvailableCore(needed_ticks); - if (next_core) { - downcounts[*next_core] = needed_ticks; - } + const s64 next_time = event_queue.front().time - global_timer; + return next_time; + } else { + return std::nullopt; } - - accumulated_ticks = 0; - - downcounts[current_context] = time_slice[current_context]; } -void CoreTiming::ResetRun() { - downcounts.fill(MAX_SLICE_LENGTH); - time_slice.fill(MAX_SLICE_LENGTH); - current_context = 0; - // Still events left (scheduled in the future) - if (!event_queue.empty()) { - const s64 needed_ticks = - std::min<s64>(event_queue.front().time - global_timer, MAX_SLICE_LENGTH); - downcounts[current_context] = needed_ticks; +void CoreTiming::ThreadLoop() { + has_started = true; + while (!shutting_down) { + while (!paused) { + paused_set = false; + const auto next_time = Advance(); + if (next_time) { + if (*next_time > 0) { + std::chrono::nanoseconds next_time_ns = std::chrono::nanoseconds(*next_time); + event.WaitFor(next_time_ns); + } + } else { + wait_set = true; + event.Wait(); + } + wait_set = false; + } + paused_set = true; + clock->Pause(true); + pause_event.Wait(); + clock->Pause(false); } - - is_global_timer_sane = false; - accumulated_ticks = 0; } -void CoreTiming::Idle() { - accumulated_ticks += downcounts[current_context]; - idled_cycles += downcounts[current_context]; - downcounts[current_context] = 0; +std::chrono::nanoseconds CoreTiming::GetGlobalTimeNs() const { + if (is_multicore) { + return clock->GetTimeNS(); + } + return CyclesToNs(ticks); } std::chrono::microseconds CoreTiming::GetGlobalTimeUs() const { - return std::chrono::microseconds{GetTicks() * 1000000 / Hardware::BASE_CLOCK_RATE}; -} - -s64 CoreTiming::GetDowncount() const { - return downcounts[current_context]; + if (is_multicore) { + return clock->GetTimeUS(); + } + return CyclesToUs(ticks); } } // namespace Core::Timing diff --git a/src/core/core_timing.h b/src/core/core_timing.h index d50f4eb8a..72faaab64 100644 --- a/src/core/core_timing.h +++ b/src/core/core_timing.h @@ -1,19 +1,25 @@ -// Copyright 2008 Dolphin Emulator Project / 2017 Citra Emulator Project -// Licensed under GPLv2+ +// Copyright 2020 yuzu Emulator Project +// Licensed under GPLv2 or any later version // Refer to the license.txt file included. #pragma once +#include <atomic> #include <chrono> #include <functional> #include <memory> #include <mutex> #include <optional> #include <string> +#include <thread> #include <vector> #include "common/common_types.h" +#include "common/spin_lock.h" +#include "common/thread.h" #include "common/threadsafe_queue.h" +#include "common/wall_clock.h" +#include "core/hardware_properties.h" namespace Core::Timing { @@ -56,16 +62,40 @@ public: /// CoreTiming begins at the boundary of timing slice -1. An initial call to Advance() is /// required to end slice - 1 and start slice 0 before the first cycle of code is executed. - void Initialize(); + void Initialize(std::function<void(void)>&& on_thread_init_); /// Tears down all timing related functionality. void Shutdown(); - /// After the first Advance, the slice lengths and the downcount will be reduced whenever an - /// event is scheduled earlier than the current values. - /// - /// Scheduling from a callback will not update the downcount until the Advance() completes. - void ScheduleEvent(s64 cycles_into_future, const std::shared_ptr<EventType>& event_type, + /// Sets if emulation is multicore or single core, must be set before Initialize + void SetMulticore(bool is_multicore) { + this->is_multicore = is_multicore; + } + + /// Check if it's using host timing. + bool IsHostTiming() const { + return is_multicore; + } + + /// Pauses/Unpauses the execution of the timer thread. + void Pause(bool is_paused); + + /// Pauses/Unpauses the execution of the timer thread and waits until paused. + void SyncPause(bool is_paused); + + /// Checks if core timing is running. + bool IsRunning() const; + + /// Checks if the timer thread has started. + bool HasStarted() const { + return has_started; + } + + /// Checks if there are any pending time events. + bool HasPendingEvents() const; + + /// Schedules an event in core timing + void ScheduleEvent(s64 ns_into_future, const std::shared_ptr<EventType>& event_type, u64 userdata = 0); void UnscheduleEvent(const std::shared_ptr<EventType>& event_type, u64 userdata); @@ -73,41 +103,30 @@ public: /// We only permit one event of each type in the queue at a time. void RemoveEvent(const std::shared_ptr<EventType>& event_type); - void ForceExceptionCheck(s64 cycles); - - /// This should only be called from the emu thread, if you are calling it any other thread, - /// you are doing something evil - u64 GetTicks() const; - - u64 GetIdleTicks() const; - void AddTicks(u64 ticks); - /// Advance must be called at the beginning of dispatcher loops, not the end. Advance() ends - /// the previous timing slice and begins the next one, you must Advance from the previous - /// slice to the current one before executing any cycles. CoreTiming starts in slice -1 so an - /// Advance() is required to initialize the slice length before the first cycle of emulated - /// instructions is executed. - void Advance(); + void ResetTicks(); - /// Pretend that the main CPU has executed enough cycles to reach the next event. void Idle(); - std::chrono::microseconds GetGlobalTimeUs() const; + s64 GetDowncount() const { + return downcount; + } - void ResetRun(); + /// Returns current time in emulated CPU cycles + u64 GetCPUTicks() const; - s64 GetDowncount() const; + /// Returns current time in emulated in Clock cycles + u64 GetClockTicks() const; - void SwitchContext(u64 new_context) { - current_context = new_context; - } + /// Returns current time in microseconds. + std::chrono::microseconds GetGlobalTimeUs() const; - bool CanCurrentContextRun() const { - return time_slice[current_context] > 0; - } + /// Returns current time in nanoseconds. + std::chrono::nanoseconds GetGlobalTimeNs() const; - std::optional<u64> NextAvailableCore(const s64 needed_ticks) const; + /// Checks for events manually and returns time in nanoseconds for next event, threadsafe. + std::optional<s64> Advance(); private: struct Event; @@ -115,21 +134,14 @@ private: /// Clear all pending events. This should ONLY be done on exit. void ClearPendingEvents(); - static constexpr u64 num_cpu_cores = 4; + static void ThreadEntry(CoreTiming& instance); + void ThreadLoop(); - s64 global_timer = 0; - s64 idled_cycles = 0; - s64 slice_length = 0; - u64 accumulated_ticks = 0; - std::array<s64, num_cpu_cores> downcounts{}; - // Slice of time assigned to each core per run. - std::array<s64, num_cpu_cores> time_slice{}; - u64 current_context = 0; + std::unique_ptr<Common::WallClock> clock; - // Are we in a function that has been called from Advance() - // If events are scheduled from a function that gets called from Advance(), - // don't change slice_length and downcount. - bool is_global_timer_sane = false; + u64 global_timer = 0; + + std::chrono::nanoseconds start_point; // The queue is a min-heap using std::make_heap/push_heap/pop_heap. // We don't use std::priority_queue because we need to be able to serialize, unserialize and @@ -139,8 +151,23 @@ private: u64 event_fifo_id = 0; std::shared_ptr<EventType> ev_lost; - - std::mutex inner_mutex; + Common::Event event{}; + Common::Event pause_event{}; + Common::SpinLock basic_lock{}; + Common::SpinLock advance_lock{}; + std::unique_ptr<std::thread> timer_thread; + std::atomic<bool> paused{}; + std::atomic<bool> paused_set{}; + std::atomic<bool> wait_set{}; + std::atomic<bool> shutting_down{}; + std::atomic<bool> has_started{}; + std::function<void(void)> on_thread_init{}; + + bool is_multicore{}; + + /// Cycle timing + u64 ticks{}; + s64 downcount{}; }; /// Creates a core timing event with the given name and callback. diff --git a/src/core/core_timing_util.cpp b/src/core/core_timing_util.cpp index de50d3b14..aefc63663 100644 --- a/src/core/core_timing_util.cpp +++ b/src/core/core_timing_util.cpp @@ -38,15 +38,23 @@ s64 usToCycles(std::chrono::microseconds us) { } s64 nsToCycles(std::chrono::nanoseconds ns) { - if (static_cast<u64>(ns.count() / 1000000000) > MAX_VALUE_TO_MULTIPLY) { - LOG_ERROR(Core_Timing, "Integer overflow, use max value"); - return std::numeric_limits<s64>::max(); - } - if (static_cast<u64>(ns.count()) > MAX_VALUE_TO_MULTIPLY) { - LOG_DEBUG(Core_Timing, "Time very big, do rounding"); - return Hardware::BASE_CLOCK_RATE * (ns.count() / 1000000000); - } - return (Hardware::BASE_CLOCK_RATE * ns.count()) / 1000000000; + const u128 temporal = Common::Multiply64Into128(ns.count(), Hardware::BASE_CLOCK_RATE); + return Common::Divide128On32(temporal, static_cast<u32>(1000000000)).first; +} + +u64 msToClockCycles(std::chrono::milliseconds ns) { + const u128 temp = Common::Multiply64Into128(ns.count(), Hardware::CNTFREQ); + return Common::Divide128On32(temp, 1000).first; +} + +u64 usToClockCycles(std::chrono::microseconds ns) { + const u128 temp = Common::Multiply64Into128(ns.count(), Hardware::CNTFREQ); + return Common::Divide128On32(temp, 1000000).first; +} + +u64 nsToClockCycles(std::chrono::nanoseconds ns) { + const u128 temp = Common::Multiply64Into128(ns.count(), Hardware::CNTFREQ); + return Common::Divide128On32(temp, 1000000000).first; } u64 CpuCyclesToClockCycles(u64 ticks) { @@ -54,4 +62,22 @@ u64 CpuCyclesToClockCycles(u64 ticks) { return Common::Divide128On32(temporal, static_cast<u32>(Hardware::BASE_CLOCK_RATE)).first; } +std::chrono::milliseconds CyclesToMs(s64 cycles) { + const u128 temporal = Common::Multiply64Into128(cycles, 1000); + u64 ms = Common::Divide128On32(temporal, static_cast<u32>(Hardware::BASE_CLOCK_RATE)).first; + return std::chrono::milliseconds(ms); +} + +std::chrono::nanoseconds CyclesToNs(s64 cycles) { + const u128 temporal = Common::Multiply64Into128(cycles, 1000000000); + u64 ns = Common::Divide128On32(temporal, static_cast<u32>(Hardware::BASE_CLOCK_RATE)).first; + return std::chrono::nanoseconds(ns); +} + +std::chrono::microseconds CyclesToUs(s64 cycles) { + const u128 temporal = Common::Multiply64Into128(cycles, 1000000); + u64 us = Common::Divide128On32(temporal, static_cast<u32>(Hardware::BASE_CLOCK_RATE)).first; + return std::chrono::microseconds(us); +} + } // namespace Core::Timing diff --git a/src/core/core_timing_util.h b/src/core/core_timing_util.h index addc72b19..2ed979e14 100644 --- a/src/core/core_timing_util.h +++ b/src/core/core_timing_util.h @@ -13,18 +13,12 @@ namespace Core::Timing { s64 msToCycles(std::chrono::milliseconds ms); s64 usToCycles(std::chrono::microseconds us); s64 nsToCycles(std::chrono::nanoseconds ns); - -inline std::chrono::milliseconds CyclesToMs(s64 cycles) { - return std::chrono::milliseconds(cycles * 1000 / Hardware::BASE_CLOCK_RATE); -} - -inline std::chrono::nanoseconds CyclesToNs(s64 cycles) { - return std::chrono::nanoseconds(cycles * 1000000000 / Hardware::BASE_CLOCK_RATE); -} - -inline std::chrono::microseconds CyclesToUs(s64 cycles) { - return std::chrono::microseconds(cycles * 1000000 / Hardware::BASE_CLOCK_RATE); -} +u64 msToClockCycles(std::chrono::milliseconds ns); +u64 usToClockCycles(std::chrono::microseconds ns); +u64 nsToClockCycles(std::chrono::nanoseconds ns); +std::chrono::milliseconds CyclesToMs(s64 cycles); +std::chrono::nanoseconds CyclesToNs(s64 cycles); +std::chrono::microseconds CyclesToUs(s64 cycles); u64 CpuCyclesToClockCycles(u64 ticks); diff --git a/src/core/cpu_manager.cpp b/src/core/cpu_manager.cpp index 70ddbdcca..32afcf3ae 100644 --- a/src/core/cpu_manager.cpp +++ b/src/core/cpu_manager.cpp @@ -2,80 +2,372 @@ // Licensed under GPLv2 or any later version // Refer to the license.txt file included. +#include "common/fiber.h" +#include "common/microprofile.h" +#include "common/thread.h" #include "core/arm/exclusive_monitor.h" #include "core/core.h" -#include "core/core_manager.h" #include "core/core_timing.h" #include "core/cpu_manager.h" #include "core/gdbstub/gdbstub.h" +#include "core/hle/kernel/kernel.h" +#include "core/hle/kernel/physical_core.h" +#include "core/hle/kernel/scheduler.h" +#include "core/hle/kernel/thread.h" +#include "video_core/gpu.h" namespace Core { CpuManager::CpuManager(System& system) : system{system} {} CpuManager::~CpuManager() = default; +void CpuManager::ThreadStart(CpuManager& cpu_manager, std::size_t core) { + cpu_manager.RunThread(core); +} + void CpuManager::Initialize() { - for (std::size_t index = 0; index < core_managers.size(); ++index) { - core_managers[index] = std::make_unique<CoreManager>(system, index); + running_mode = true; + if (is_multicore) { + for (std::size_t core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) { + core_data[core].host_thread = + std::make_unique<std::thread>(ThreadStart, std::ref(*this), core); + } + } else { + core_data[0].host_thread = std::make_unique<std::thread>(ThreadStart, std::ref(*this), 0); } } void CpuManager::Shutdown() { - for (auto& cpu_core : core_managers) { - cpu_core.reset(); + running_mode = false; + Pause(false); + if (is_multicore) { + for (std::size_t core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) { + core_data[core].host_thread->join(); + core_data[core].host_thread.reset(); + } + } else { + core_data[0].host_thread->join(); + core_data[0].host_thread.reset(); } } -CoreManager& CpuManager::GetCoreManager(std::size_t index) { - return *core_managers.at(index); +std::function<void(void*)> CpuManager::GetGuestThreadStartFunc() { + return std::function<void(void*)>(GuestThreadFunction); } -const CoreManager& CpuManager::GetCoreManager(std::size_t index) const { - return *core_managers.at(index); +std::function<void(void*)> CpuManager::GetIdleThreadStartFunc() { + return std::function<void(void*)>(IdleThreadFunction); } -CoreManager& CpuManager::GetCurrentCoreManager() { - // Otherwise, use single-threaded mode active_core variable - return *core_managers[active_core]; +std::function<void(void*)> CpuManager::GetSuspendThreadStartFunc() { + return std::function<void(void*)>(SuspendThreadFunction); } -const CoreManager& CpuManager::GetCurrentCoreManager() const { - // Otherwise, use single-threaded mode active_core variable - return *core_managers[active_core]; +void CpuManager::GuestThreadFunction(void* cpu_manager_) { + CpuManager* cpu_manager = static_cast<CpuManager*>(cpu_manager_); + if (cpu_manager->is_multicore) { + cpu_manager->MultiCoreRunGuestThread(); + } else { + cpu_manager->SingleCoreRunGuestThread(); + } } -void CpuManager::RunLoop(bool tight_loop) { - if (GDBStub::IsServerEnabled()) { - GDBStub::HandlePacket(); +void CpuManager::GuestRewindFunction(void* cpu_manager_) { + CpuManager* cpu_manager = static_cast<CpuManager*>(cpu_manager_); + if (cpu_manager->is_multicore) { + cpu_manager->MultiCoreRunGuestLoop(); + } else { + cpu_manager->SingleCoreRunGuestLoop(); + } +} - // If the loop is halted and we want to step, use a tiny (1) number of instructions to - // execute. Otherwise, get out of the loop function. - if (GDBStub::GetCpuHaltFlag()) { - if (GDBStub::GetCpuStepFlag()) { - tight_loop = false; - } else { - return; +void CpuManager::IdleThreadFunction(void* cpu_manager_) { + CpuManager* cpu_manager = static_cast<CpuManager*>(cpu_manager_); + if (cpu_manager->is_multicore) { + cpu_manager->MultiCoreRunIdleThread(); + } else { + cpu_manager->SingleCoreRunIdleThread(); + } +} + +void CpuManager::SuspendThreadFunction(void* cpu_manager_) { + CpuManager* cpu_manager = static_cast<CpuManager*>(cpu_manager_); + if (cpu_manager->is_multicore) { + cpu_manager->MultiCoreRunSuspendThread(); + } else { + cpu_manager->SingleCoreRunSuspendThread(); + } +} + +void* CpuManager::GetStartFuncParamater() { + return static_cast<void*>(this); +} + +/////////////////////////////////////////////////////////////////////////////// +/// MultiCore /// +/////////////////////////////////////////////////////////////////////////////// + +void CpuManager::MultiCoreRunGuestThread() { + auto& kernel = system.Kernel(); + { + auto& sched = kernel.CurrentScheduler(); + sched.OnThreadStart(); + } + MultiCoreRunGuestLoop(); +} + +void CpuManager::MultiCoreRunGuestLoop() { + auto& kernel = system.Kernel(); + auto* thread = kernel.CurrentScheduler().GetCurrentThread(); + while (true) { + auto* physical_core = &kernel.CurrentPhysicalCore(); + auto& arm_interface = thread->ArmInterface(); + system.EnterDynarmicProfile(); + while (!physical_core->IsInterrupted()) { + arm_interface.Run(); + physical_core = &kernel.CurrentPhysicalCore(); + } + system.ExitDynarmicProfile(); + arm_interface.ClearExclusiveState(); + auto& scheduler = kernel.CurrentScheduler(); + scheduler.TryDoContextSwitch(); + } +} + +void CpuManager::MultiCoreRunIdleThread() { + auto& kernel = system.Kernel(); + while (true) { + auto& physical_core = kernel.CurrentPhysicalCore(); + physical_core.Idle(); + auto& scheduler = kernel.CurrentScheduler(); + scheduler.TryDoContextSwitch(); + } +} + +void CpuManager::MultiCoreRunSuspendThread() { + auto& kernel = system.Kernel(); + { + auto& sched = kernel.CurrentScheduler(); + sched.OnThreadStart(); + } + while (true) { + auto core = kernel.GetCurrentHostThreadID(); + auto& scheduler = kernel.CurrentScheduler(); + Kernel::Thread* current_thread = scheduler.GetCurrentThread(); + Common::Fiber::YieldTo(current_thread->GetHostContext(), core_data[core].host_context); + ASSERT(scheduler.ContextSwitchPending()); + ASSERT(core == kernel.GetCurrentHostThreadID()); + scheduler.TryDoContextSwitch(); + } +} + +void CpuManager::MultiCorePause(bool paused) { + if (!paused) { + bool all_not_barrier = false; + while (!all_not_barrier) { + all_not_barrier = true; + for (std::size_t core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) { + all_not_barrier &= + !core_data[core].is_running.load() && core_data[core].initialized.load(); + } + } + for (std::size_t core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) { + core_data[core].enter_barrier->Set(); + } + if (paused_state.load()) { + bool all_barrier = false; + while (!all_barrier) { + all_barrier = true; + for (std::size_t core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) { + all_barrier &= + core_data[core].is_paused.load() && core_data[core].initialized.load(); + } + } + for (std::size_t core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) { + core_data[core].exit_barrier->Set(); + } + } + } else { + /// Wait until all cores are paused. + bool all_barrier = false; + while (!all_barrier) { + all_barrier = true; + for (std::size_t core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) { + all_barrier &= + core_data[core].is_paused.load() && core_data[core].initialized.load(); } } + /// Don't release the barrier } + paused_state = paused; +} + +/////////////////////////////////////////////////////////////////////////////// +/// SingleCore /// +/////////////////////////////////////////////////////////////////////////////// - auto& core_timing = system.CoreTiming(); - core_timing.ResetRun(); - bool keep_running{}; - do { - keep_running = false; - for (active_core = 0; active_core < NUM_CPU_CORES; ++active_core) { - core_timing.SwitchContext(active_core); - if (core_timing.CanCurrentContextRun()) { - core_managers[active_core]->RunLoop(tight_loop); +void CpuManager::SingleCoreRunGuestThread() { + auto& kernel = system.Kernel(); + { + auto& sched = kernel.CurrentScheduler(); + sched.OnThreadStart(); + } + SingleCoreRunGuestLoop(); +} + +void CpuManager::SingleCoreRunGuestLoop() { + auto& kernel = system.Kernel(); + auto* thread = kernel.CurrentScheduler().GetCurrentThread(); + while (true) { + auto* physical_core = &kernel.CurrentPhysicalCore(); + auto& arm_interface = thread->ArmInterface(); + system.EnterDynarmicProfile(); + if (!physical_core->IsInterrupted()) { + arm_interface.Run(); + physical_core = &kernel.CurrentPhysicalCore(); + } + system.ExitDynarmicProfile(); + thread->SetPhantomMode(true); + system.CoreTiming().Advance(); + thread->SetPhantomMode(false); + arm_interface.ClearExclusiveState(); + PreemptSingleCore(); + auto& scheduler = kernel.Scheduler(current_core); + scheduler.TryDoContextSwitch(); + } +} + +void CpuManager::SingleCoreRunIdleThread() { + auto& kernel = system.Kernel(); + while (true) { + auto& physical_core = kernel.CurrentPhysicalCore(); + PreemptSingleCore(false); + system.CoreTiming().AddTicks(1000U); + idle_count++; + auto& scheduler = physical_core.Scheduler(); + scheduler.TryDoContextSwitch(); + } +} + +void CpuManager::SingleCoreRunSuspendThread() { + auto& kernel = system.Kernel(); + { + auto& sched = kernel.CurrentScheduler(); + sched.OnThreadStart(); + } + while (true) { + auto core = kernel.GetCurrentHostThreadID(); + auto& scheduler = kernel.CurrentScheduler(); + Kernel::Thread* current_thread = scheduler.GetCurrentThread(); + Common::Fiber::YieldTo(current_thread->GetHostContext(), core_data[0].host_context); + ASSERT(scheduler.ContextSwitchPending()); + ASSERT(core == kernel.GetCurrentHostThreadID()); + scheduler.TryDoContextSwitch(); + } +} + +void CpuManager::PreemptSingleCore(bool from_running_enviroment) { + std::size_t old_core = current_core; + auto& scheduler = system.Kernel().Scheduler(old_core); + Kernel::Thread* current_thread = scheduler.GetCurrentThread(); + if (idle_count >= 4 || from_running_enviroment) { + if (!from_running_enviroment) { + system.CoreTiming().Idle(); + idle_count = 0; + } + current_thread->SetPhantomMode(true); + system.CoreTiming().Advance(); + current_thread->SetPhantomMode(false); + } + current_core.store((current_core + 1) % Core::Hardware::NUM_CPU_CORES); + system.CoreTiming().ResetTicks(); + scheduler.Unload(); + auto& next_scheduler = system.Kernel().Scheduler(current_core); + Common::Fiber::YieldTo(current_thread->GetHostContext(), next_scheduler.ControlContext()); + /// May have changed scheduler + auto& current_scheduler = system.Kernel().Scheduler(current_core); + current_scheduler.Reload(); + auto* currrent_thread2 = current_scheduler.GetCurrentThread(); + if (!currrent_thread2->IsIdleThread()) { + idle_count = 0; + } +} + +void CpuManager::SingleCorePause(bool paused) { + if (!paused) { + bool all_not_barrier = false; + while (!all_not_barrier) { + all_not_barrier = !core_data[0].is_running.load() && core_data[0].initialized.load(); + } + core_data[0].enter_barrier->Set(); + if (paused_state.load()) { + bool all_barrier = false; + while (!all_barrier) { + all_barrier = core_data[0].is_paused.load() && core_data[0].initialized.load(); } - keep_running |= core_timing.CanCurrentContextRun(); + core_data[0].exit_barrier->Set(); } - } while (keep_running); + } else { + /// Wait until all cores are paused. + bool all_barrier = false; + while (!all_barrier) { + all_barrier = core_data[0].is_paused.load() && core_data[0].initialized.load(); + } + /// Don't release the barrier + } + paused_state = paused; +} + +void CpuManager::Pause(bool paused) { + if (is_multicore) { + MultiCorePause(paused); + } else { + SingleCorePause(paused); + } +} - if (GDBStub::IsServerEnabled()) { - GDBStub::SetCpuStepFlag(false); +void CpuManager::RunThread(std::size_t core) { + /// Initialization + system.RegisterCoreThread(core); + std::string name; + if (is_multicore) { + name = "yuzu:CoreCPUThread_" + std::to_string(core); + } else { + name = "yuzu:CPUThread"; + } + MicroProfileOnThreadCreate(name.c_str()); + Common::SetCurrentThreadName(name.c_str()); + Common::SetCurrentThreadPriority(Common::ThreadPriority::High); + auto& data = core_data[core]; + data.enter_barrier = std::make_unique<Common::Event>(); + data.exit_barrier = std::make_unique<Common::Event>(); + data.host_context = Common::Fiber::ThreadToFiber(); + data.is_running = false; + data.initialized = true; + const bool sc_sync = !is_async_gpu && !is_multicore; + bool sc_sync_first_use = sc_sync; + /// Running + while (running_mode) { + data.is_running = false; + data.enter_barrier->Wait(); + if (sc_sync_first_use) { + system.GPU().ObtainContext(); + sc_sync_first_use = false; + } + auto& scheduler = system.Kernel().CurrentScheduler(); + Kernel::Thread* current_thread = scheduler.GetCurrentThread(); + data.is_running = true; + Common::Fiber::YieldTo(data.host_context, current_thread->GetHostContext()); + data.is_running = false; + data.is_paused = true; + data.exit_barrier->Wait(); + data.is_paused = false; } + /// Time to cleanup + data.host_context->Exit(); + data.enter_barrier.reset(); + data.exit_barrier.reset(); + data.initialized = false; } } // namespace Core diff --git a/src/core/cpu_manager.h b/src/core/cpu_manager.h index 97554d1bb..35929ed94 100644 --- a/src/core/cpu_manager.h +++ b/src/core/cpu_manager.h @@ -5,12 +5,19 @@ #pragma once #include <array> +#include <atomic> +#include <functional> #include <memory> +#include <thread> #include "core/hardware_properties.h" +namespace Common { +class Event; +class Fiber; +} // namespace Common + namespace Core { -class CoreManager; class System; class CpuManager { @@ -24,24 +31,75 @@ public: CpuManager& operator=(const CpuManager&) = delete; CpuManager& operator=(CpuManager&&) = delete; + /// Sets if emulation is multicore or single core, must be set before Initialize + void SetMulticore(bool is_multicore) { + this->is_multicore = is_multicore; + } + + /// Sets if emulation is using an asynchronous GPU. + void SetAsyncGpu(bool is_async_gpu) { + this->is_async_gpu = is_async_gpu; + } + void Initialize(); void Shutdown(); - CoreManager& GetCoreManager(std::size_t index); - const CoreManager& GetCoreManager(std::size_t index) const; + void Pause(bool paused); - CoreManager& GetCurrentCoreManager(); - const CoreManager& GetCurrentCoreManager() const; + std::function<void(void*)> GetGuestThreadStartFunc(); + std::function<void(void*)> GetIdleThreadStartFunc(); + std::function<void(void*)> GetSuspendThreadStartFunc(); + void* GetStartFuncParamater(); - std::size_t GetActiveCoreIndex() const { - return active_core; - } + void PreemptSingleCore(bool from_running_enviroment = true); - void RunLoop(bool tight_loop); + std::size_t CurrentCore() const { + return current_core.load(); + } private: - std::array<std::unique_ptr<CoreManager>, Hardware::NUM_CPU_CORES> core_managers; - std::size_t active_core{}; ///< Active core, only used in single thread mode + static void GuestThreadFunction(void* cpu_manager); + static void GuestRewindFunction(void* cpu_manager); + static void IdleThreadFunction(void* cpu_manager); + static void SuspendThreadFunction(void* cpu_manager); + + void MultiCoreRunGuestThread(); + void MultiCoreRunGuestLoop(); + void MultiCoreRunIdleThread(); + void MultiCoreRunSuspendThread(); + void MultiCorePause(bool paused); + + void SingleCoreRunGuestThread(); + void SingleCoreRunGuestLoop(); + void SingleCoreRunIdleThread(); + void SingleCoreRunSuspendThread(); + void SingleCorePause(bool paused); + + static void ThreadStart(CpuManager& cpu_manager, std::size_t core); + + void RunThread(std::size_t core); + + struct CoreData { + std::shared_ptr<Common::Fiber> host_context; + std::unique_ptr<Common::Event> enter_barrier; + std::unique_ptr<Common::Event> exit_barrier; + std::atomic<bool> is_running; + std::atomic<bool> is_paused; + std::atomic<bool> initialized; + std::unique_ptr<std::thread> host_thread; + }; + + std::atomic<bool> running_mode{}; + std::atomic<bool> paused_state{}; + + std::array<CoreData, Core::Hardware::NUM_CPU_CORES> core_data{}; + + bool is_async_gpu{}; + bool is_multicore{}; + std::atomic<std::size_t> current_core{}; + std::size_t preemption_count{}; + std::size_t idle_count{}; + static constexpr std::size_t max_cycle_runs = 5; System& system; }; diff --git a/src/core/file_sys/control_metadata.cpp b/src/core/file_sys/control_metadata.cpp index f155a1341..63cd2eead 100644 --- a/src/core/file_sys/control_metadata.cpp +++ b/src/core/file_sys/control_metadata.cpp @@ -95,6 +95,10 @@ u32 NACP::GetSupportedLanguages() const { return raw.supported_languages; } +u64 NACP::GetDeviceSaveDataSize() const { + return raw.device_save_data_size; +} + std::vector<u8> NACP::GetRawBytes() const { std::vector<u8> out(sizeof(RawNACP)); std::memcpy(out.data(), &raw, sizeof(RawNACP)); diff --git a/src/core/file_sys/control_metadata.h b/src/core/file_sys/control_metadata.h index 2d8c251ac..e37b2fadf 100644 --- a/src/core/file_sys/control_metadata.h +++ b/src/core/file_sys/control_metadata.h @@ -113,6 +113,7 @@ public: u32 GetSupportedLanguages() const; std::vector<u8> GetRawBytes() const; bool GetUserAccountSwitchLock() const; + u64 GetDeviceSaveDataSize() const; private: RawNACP raw{}; diff --git a/src/core/file_sys/patch_manager.cpp b/src/core/file_sys/patch_manager.cpp index b93aa6935..c47ff863e 100644 --- a/src/core/file_sys/patch_manager.cpp +++ b/src/core/file_sys/patch_manager.cpp @@ -10,6 +10,7 @@ #include "common/file_util.h" #include "common/hex_util.h" #include "common/logging/log.h" +#include "common/string_util.h" #include "core/core.h" #include "core/file_sys/content_archive.h" #include "core/file_sys/control_metadata.h" @@ -48,6 +49,23 @@ std::string FormatTitleVersion(u32 version, TitleVersionFormat format) { return fmt::format("v{}.{}.{}", bytes[3], bytes[2], bytes[1]); } +std::shared_ptr<VfsDirectory> FindSubdirectoryCaseless(const std::shared_ptr<VfsDirectory> dir, + std::string_view name) { +#ifdef _WIN32 + return dir->GetSubdirectory(name); +#else + const auto subdirs = dir->GetSubdirectories(); + for (const auto& subdir : subdirs) { + std::string dir_name = Common::ToLower(subdir->GetName()); + if (dir_name == name) { + return subdir; + } + } + + return nullptr; +#endif +} + PatchManager::PatchManager(u64 title_id) : title_id(title_id) {} PatchManager::~PatchManager() = default; @@ -104,7 +122,7 @@ VirtualDir PatchManager::PatchExeFS(VirtualDir exefs) const { if (std::find(disabled.begin(), disabled.end(), subdir->GetName()) != disabled.end()) continue; - auto exefs_dir = subdir->GetSubdirectory("exefs"); + auto exefs_dir = FindSubdirectoryCaseless(subdir, "exefs"); if (exefs_dir != nullptr) layers.push_back(std::move(exefs_dir)); } @@ -130,7 +148,7 @@ std::vector<VirtualFile> PatchManager::CollectPatches(const std::vector<VirtualD if (std::find(disabled.cbegin(), disabled.cend(), subdir->GetName()) != disabled.cend()) continue; - auto exefs_dir = subdir->GetSubdirectory("exefs"); + auto exefs_dir = FindSubdirectoryCaseless(subdir, "exefs"); if (exefs_dir != nullptr) { for (const auto& file : exefs_dir->GetFiles()) { if (file->GetExtension() == "ips") { @@ -295,7 +313,7 @@ std::vector<Core::Memory::CheatEntry> PatchManager::CreateCheatList( continue; } - auto cheats_dir = subdir->GetSubdirectory("cheats"); + auto cheats_dir = FindSubdirectoryCaseless(subdir, "cheats"); if (cheats_dir != nullptr) { auto res = ReadCheatFileFromFolder(system, title_id, build_id_, cheats_dir, true); if (res.has_value()) { @@ -340,11 +358,11 @@ static void ApplyLayeredFS(VirtualFile& romfs, u64 title_id, ContentRecordType t continue; } - auto romfs_dir = subdir->GetSubdirectory("romfs"); + auto romfs_dir = FindSubdirectoryCaseless(subdir, "romfs"); if (romfs_dir != nullptr) layers.push_back(std::move(romfs_dir)); - auto ext_dir = subdir->GetSubdirectory("romfs_ext"); + auto ext_dir = FindSubdirectoryCaseless(subdir, "romfs_ext"); if (ext_dir != nullptr) layers_ext.push_back(std::move(ext_dir)); } @@ -470,7 +488,7 @@ std::map<std::string, std::string, std::less<>> PatchManager::GetPatchVersionNam for (const auto& mod : mod_dir->GetSubdirectories()) { std::string types; - const auto exefs_dir = mod->GetSubdirectory("exefs"); + const auto exefs_dir = FindSubdirectoryCaseless(mod, "exefs"); if (IsDirValidAndNonEmpty(exefs_dir)) { bool ips = false; bool ipswitch = false; @@ -494,9 +512,9 @@ std::map<std::string, std::string, std::less<>> PatchManager::GetPatchVersionNam if (layeredfs) AppendCommaIfNotEmpty(types, "LayeredExeFS"); } - if (IsDirValidAndNonEmpty(mod->GetSubdirectory("romfs"))) + if (IsDirValidAndNonEmpty(FindSubdirectoryCaseless(mod, "romfs"))) AppendCommaIfNotEmpty(types, "LayeredFS"); - if (IsDirValidAndNonEmpty(mod->GetSubdirectory("cheats"))) + if (IsDirValidAndNonEmpty(FindSubdirectoryCaseless(mod, "cheats"))) AppendCommaIfNotEmpty(types, "Cheats"); if (types.empty()) diff --git a/src/core/file_sys/patch_manager.h b/src/core/file_sys/patch_manager.h index ec6db524d..f4cb918dd 100644 --- a/src/core/file_sys/patch_manager.h +++ b/src/core/file_sys/patch_manager.h @@ -29,6 +29,11 @@ enum class TitleVersionFormat : u8 { std::string FormatTitleVersion(u32 version, TitleVersionFormat format = TitleVersionFormat::ThreeElements); +// Returns a directory with name matching name case-insensitive. Returns nullptr if directory +// doesn't have a directory with name. +std::shared_ptr<VfsDirectory> FindSubdirectoryCaseless(const std::shared_ptr<VfsDirectory> dir, + std::string_view name); + // A centralized class to manage patches to games. class PatchManager { public: diff --git a/src/core/file_sys/savedata_factory.cpp b/src/core/file_sys/savedata_factory.cpp index f3def93ab..adfd2c1a4 100644 --- a/src/core/file_sys/savedata_factory.cpp +++ b/src/core/file_sys/savedata_factory.cpp @@ -57,7 +57,8 @@ void PrintSaveDataDescriptorWarnings(SaveDataDescriptor meta) { bool ShouldSaveDataBeAutomaticallyCreated(SaveDataSpaceId space, const SaveDataDescriptor& desc) { return desc.type == SaveDataType::CacheStorage || desc.type == SaveDataType::TemporaryStorage || (space == SaveDataSpaceId::NandUser && ///< Normal Save Data -- Current Title & User - desc.type == SaveDataType::SaveData && desc.title_id == 0 && desc.save_id == 0); + (desc.type == SaveDataType::SaveData || desc.type == SaveDataType::DeviceSaveData) && + desc.title_id == 0 && desc.save_id == 0); } } // Anonymous namespace @@ -139,8 +140,10 @@ std::string SaveDataFactory::GetFullPath(SaveDataSpaceId space, SaveDataType typ u128 user_id, u64 save_id) { // According to switchbrew, if a save is of type SaveData and the title id field is 0, it should // be interpreted as the title id of the current process. - if (type == SaveDataType::SaveData && title_id == 0) { - title_id = Core::System::GetInstance().CurrentProcess()->GetTitleID(); + if (type == SaveDataType::SaveData || type == SaveDataType::DeviceSaveData) { + if (title_id == 0) { + title_id = Core::System::GetInstance().CurrentProcess()->GetTitleID(); + } } std::string out = GetSaveDataSpaceIdPath(space); diff --git a/src/core/file_sys/system_archive/mii_model.cpp b/src/core/file_sys/system_archive/mii_model.cpp index 6a9add87c..61bb67945 100644 --- a/src/core/file_sys/system_archive/mii_model.cpp +++ b/src/core/file_sys/system_archive/mii_model.cpp @@ -40,7 +40,7 @@ VirtualDir MiiModel() { out->AddFile(std::make_shared<ArrayVfsFile<MiiModelData::SHAPE_MID.size()>>( MiiModelData::SHAPE_MID, "ShapeMid.dat")); - return std::move(out); + return out; } } // namespace FileSys::SystemArchive diff --git a/src/core/file_sys/system_archive/shared_font.cpp b/src/core/file_sys/system_archive/shared_font.cpp index 2c05eb42e..c5cdf7d9b 100644 --- a/src/core/file_sys/system_archive/shared_font.cpp +++ b/src/core/file_sys/system_archive/shared_font.cpp @@ -23,7 +23,7 @@ VirtualFile PackBFTTF(const std::array<u8, Size>& data, const std::string& name) std::vector<u8> bfttf(Size + sizeof(u64)); - u64 offset = 0; + size_t offset = 0; Service::NS::EncryptSharedFont(vec, bfttf, offset); return std::make_shared<VectorVfsFile>(std::move(bfttf), name); } diff --git a/src/core/file_sys/system_archive/system_version.cpp b/src/core/file_sys/system_archive/system_version.cpp index 6e22f97b0..aa313de66 100644 --- a/src/core/file_sys/system_archive/system_version.cpp +++ b/src/core/file_sys/system_archive/system_version.cpp @@ -12,17 +12,17 @@ namespace SystemVersionData { // This section should reflect the best system version to describe yuzu's HLE api. // TODO(DarkLordZach): Update when HLE gets better. -constexpr u8 VERSION_MAJOR = 5; -constexpr u8 VERSION_MINOR = 1; -constexpr u8 VERSION_MICRO = 0; +constexpr u8 VERSION_MAJOR = 10; +constexpr u8 VERSION_MINOR = 0; +constexpr u8 VERSION_MICRO = 2; -constexpr u8 REVISION_MAJOR = 3; +constexpr u8 REVISION_MAJOR = 1; constexpr u8 REVISION_MINOR = 0; constexpr char PLATFORM_STRING[] = "NX"; -constexpr char VERSION_HASH[] = "23f9df53e25709d756e0c76effcb2473bd3447dd"; -constexpr char DISPLAY_VERSION[] = "5.1.0"; -constexpr char DISPLAY_TITLE[] = "NintendoSDK Firmware for NX 5.1.0-3.0"; +constexpr char VERSION_HASH[] = "f90143fa8bbc061d4f68c35f95f04f8080c0ecdc"; +constexpr char DISPLAY_VERSION[] = "10.0.2"; +constexpr char DISPLAY_TITLE[] = "NintendoSDK Firmware for NX 10.0.2-1.0"; } // namespace SystemVersionData diff --git a/src/core/frontend/emu_window.cpp b/src/core/frontend/emu_window.cpp index eda466a5d..9a081fbd4 100644 --- a/src/core/frontend/emu_window.cpp +++ b/src/core/frontend/emu_window.cpp @@ -46,7 +46,7 @@ private: EmuWindow::EmuWindow() { // TODO: Find a better place to set this. config.min_client_area_size = - std::make_pair(Layout::ScreenUndocked::Width, Layout::ScreenUndocked::Height); + std::make_pair(Layout::MinimumSize::Width, Layout::MinimumSize::Height); active_config = config; touch_state = std::make_shared<TouchState>(); Input::RegisterFactory<Input::TouchDevice>("emu_window", touch_state); diff --git a/src/core/frontend/framebuffer_layout.h b/src/core/frontend/framebuffer_layout.h index 15ecfb13d..91ecc30ab 100644 --- a/src/core/frontend/framebuffer_layout.h +++ b/src/core/frontend/framebuffer_layout.h @@ -8,6 +8,11 @@ namespace Layout { +namespace MinimumSize { +constexpr u32 Width = 640; +constexpr u32 Height = 360; +} // namespace MinimumSize + namespace ScreenUndocked { constexpr u32 Width = 1280; constexpr u32 Height = 720; diff --git a/src/core/gdbstub/gdbstub.cpp b/src/core/gdbstub/gdbstub.cpp index 70c0f8b80..79f22a403 100644 --- a/src/core/gdbstub/gdbstub.cpp +++ b/src/core/gdbstub/gdbstub.cpp @@ -35,7 +35,6 @@ #include "common/swap.h" #include "core/arm/arm_interface.h" #include "core/core.h" -#include "core/core_manager.h" #include "core/gdbstub/gdbstub.h" #include "core/hle/kernel/memory/page_table.h" #include "core/hle/kernel/process.h" diff --git a/src/core/hardware_properties.h b/src/core/hardware_properties.h index b04e046ed..456b41e1b 100644 --- a/src/core/hardware_properties.h +++ b/src/core/hardware_properties.h @@ -42,6 +42,10 @@ struct EmuThreadHandle { constexpr u32 invalid_handle = 0xFFFFFFFF; return {invalid_handle, invalid_handle}; } + + bool IsInvalid() const { + return (*this) == InvalidHandle(); + } }; } // namespace Core diff --git a/src/core/hle/kernel/address_arbiter.cpp b/src/core/hle/kernel/address_arbiter.cpp index 8475b698c..4d2a9b35d 100644 --- a/src/core/hle/kernel/address_arbiter.cpp +++ b/src/core/hle/kernel/address_arbiter.cpp @@ -7,11 +7,15 @@ #include "common/assert.h" #include "common/common_types.h" +#include "core/arm/exclusive_monitor.h" #include "core/core.h" #include "core/hle/kernel/address_arbiter.h" #include "core/hle/kernel/errors.h" +#include "core/hle/kernel/handle_table.h" +#include "core/hle/kernel/kernel.h" #include "core/hle/kernel/scheduler.h" #include "core/hle/kernel/thread.h" +#include "core/hle/kernel/time_manager.h" #include "core/hle/result.h" #include "core/memory.h" @@ -20,6 +24,7 @@ namespace Kernel { // Wake up num_to_wake (or all) threads in a vector. void AddressArbiter::WakeThreads(const std::vector<std::shared_ptr<Thread>>& waiting_threads, s32 num_to_wake) { + auto& time_manager = system.Kernel().TimeManager(); // Only process up to 'target' threads, unless 'target' is <= 0, in which case process // them all. std::size_t last = waiting_threads.size(); @@ -29,12 +34,10 @@ void AddressArbiter::WakeThreads(const std::vector<std::shared_ptr<Thread>>& wai // Signal the waiting threads. for (std::size_t i = 0; i < last; i++) { - ASSERT(waiting_threads[i]->GetStatus() == ThreadStatus::WaitArb); - waiting_threads[i]->SetWaitSynchronizationResult(RESULT_SUCCESS); + waiting_threads[i]->SetSynchronizationResults(nullptr, RESULT_SUCCESS); RemoveThread(waiting_threads[i]); - waiting_threads[i]->SetArbiterWaitAddress(0); + waiting_threads[i]->WaitForArbitration(false); waiting_threads[i]->ResumeFromWait(); - system.PrepareReschedule(waiting_threads[i]->GetProcessorID()); } } @@ -56,6 +59,7 @@ ResultCode AddressArbiter::SignalToAddress(VAddr address, SignalType type, s32 v } ResultCode AddressArbiter::SignalToAddressOnly(VAddr address, s32 num_to_wake) { + SchedulerLock lock(system.Kernel()); const std::vector<std::shared_ptr<Thread>> waiting_threads = GetThreadsWaitingOnAddress(address); WakeThreads(waiting_threads, num_to_wake); @@ -64,6 +68,7 @@ ResultCode AddressArbiter::SignalToAddressOnly(VAddr address, s32 num_to_wake) { ResultCode AddressArbiter::IncrementAndSignalToAddressIfEqual(VAddr address, s32 value, s32 num_to_wake) { + SchedulerLock lock(system.Kernel()); auto& memory = system.Memory(); // Ensure that we can write to the address. @@ -71,16 +76,24 @@ ResultCode AddressArbiter::IncrementAndSignalToAddressIfEqual(VAddr address, s32 return ERR_INVALID_ADDRESS_STATE; } - if (static_cast<s32>(memory.Read32(address)) != value) { - return ERR_INVALID_STATE; - } + const std::size_t current_core = system.CurrentCoreIndex(); + auto& monitor = system.Monitor(); + u32 current_value; + do { + current_value = monitor.ExclusiveRead32(current_core, address); + + if (current_value != value) { + return ERR_INVALID_STATE; + } + current_value++; + } while (!monitor.ExclusiveWrite32(current_core, address, current_value)); - memory.Write32(address, static_cast<u32>(value + 1)); return SignalToAddressOnly(address, num_to_wake); } ResultCode AddressArbiter::ModifyByWaitingCountAndSignalToAddressIfEqual(VAddr address, s32 value, s32 num_to_wake) { + SchedulerLock lock(system.Kernel()); auto& memory = system.Memory(); // Ensure that we can write to the address. @@ -92,29 +105,33 @@ ResultCode AddressArbiter::ModifyByWaitingCountAndSignalToAddressIfEqual(VAddr a const std::vector<std::shared_ptr<Thread>> waiting_threads = GetThreadsWaitingOnAddress(address); - // Determine the modified value depending on the waiting count. + const std::size_t current_core = system.CurrentCoreIndex(); + auto& monitor = system.Monitor(); s32 updated_value; - if (num_to_wake <= 0) { - if (waiting_threads.empty()) { - updated_value = value + 1; - } else { - updated_value = value - 1; + do { + updated_value = monitor.ExclusiveRead32(current_core, address); + + if (updated_value != value) { + return ERR_INVALID_STATE; } - } else { - if (waiting_threads.empty()) { - updated_value = value + 1; - } else if (waiting_threads.size() <= static_cast<u32>(num_to_wake)) { - updated_value = value - 1; + // Determine the modified value depending on the waiting count. + if (num_to_wake <= 0) { + if (waiting_threads.empty()) { + updated_value = value + 1; + } else { + updated_value = value - 1; + } } else { - updated_value = value; + if (waiting_threads.empty()) { + updated_value = value + 1; + } else if (waiting_threads.size() <= static_cast<u32>(num_to_wake)) { + updated_value = value - 1; + } else { + updated_value = value; + } } - } + } while (!monitor.ExclusiveWrite32(current_core, address, updated_value)); - if (static_cast<s32>(memory.Read32(address)) != value) { - return ERR_INVALID_STATE; - } - - memory.Write32(address, static_cast<u32>(updated_value)); WakeThreads(waiting_threads, num_to_wake); return RESULT_SUCCESS; } @@ -136,60 +153,127 @@ ResultCode AddressArbiter::WaitForAddress(VAddr address, ArbitrationType type, s ResultCode AddressArbiter::WaitForAddressIfLessThan(VAddr address, s32 value, s64 timeout, bool should_decrement) { auto& memory = system.Memory(); + auto& kernel = system.Kernel(); + Thread* current_thread = system.CurrentScheduler().GetCurrentThread(); - // Ensure that we can read the address. - if (!memory.IsValidVirtualAddress(address)) { - return ERR_INVALID_ADDRESS_STATE; - } + Handle event_handle = InvalidHandle; + { + SchedulerLockAndSleep lock(kernel, event_handle, current_thread, timeout); + + if (current_thread->IsPendingTermination()) { + lock.CancelSleep(); + return ERR_THREAD_TERMINATING; + } + + // Ensure that we can read the address. + if (!memory.IsValidVirtualAddress(address)) { + lock.CancelSleep(); + return ERR_INVALID_ADDRESS_STATE; + } + + s32 current_value = static_cast<s32>(memory.Read32(address)); + if (current_value >= value) { + lock.CancelSleep(); + return ERR_INVALID_STATE; + } + + current_thread->SetSynchronizationResults(nullptr, RESULT_TIMEOUT); + + s32 decrement_value; + + const std::size_t current_core = system.CurrentCoreIndex(); + auto& monitor = system.Monitor(); + do { + current_value = static_cast<s32>(monitor.ExclusiveRead32(current_core, address)); + if (should_decrement) { + decrement_value = current_value - 1; + } else { + decrement_value = current_value; + } + } while ( + !monitor.ExclusiveWrite32(current_core, address, static_cast<u32>(decrement_value))); + + // Short-circuit without rescheduling, if timeout is zero. + if (timeout == 0) { + lock.CancelSleep(); + return RESULT_TIMEOUT; + } - const s32 cur_value = static_cast<s32>(memory.Read32(address)); - if (cur_value >= value) { - return ERR_INVALID_STATE; + current_thread->SetArbiterWaitAddress(address); + InsertThread(SharedFrom(current_thread)); + current_thread->SetStatus(ThreadStatus::WaitArb); + current_thread->WaitForArbitration(true); } - if (should_decrement) { - memory.Write32(address, static_cast<u32>(cur_value - 1)); + if (event_handle != InvalidHandle) { + auto& time_manager = kernel.TimeManager(); + time_manager.UnscheduleTimeEvent(event_handle); } - // Short-circuit without rescheduling, if timeout is zero. - if (timeout == 0) { - return RESULT_TIMEOUT; + { + SchedulerLock lock(kernel); + if (current_thread->IsWaitingForArbitration()) { + RemoveThread(SharedFrom(current_thread)); + current_thread->WaitForArbitration(false); + } } - return WaitForAddressImpl(address, timeout); + return current_thread->GetSignalingResult(); } ResultCode AddressArbiter::WaitForAddressIfEqual(VAddr address, s32 value, s64 timeout) { auto& memory = system.Memory(); + auto& kernel = system.Kernel(); + Thread* current_thread = system.CurrentScheduler().GetCurrentThread(); - // Ensure that we can read the address. - if (!memory.IsValidVirtualAddress(address)) { - return ERR_INVALID_ADDRESS_STATE; - } + Handle event_handle = InvalidHandle; + { + SchedulerLockAndSleep lock(kernel, event_handle, current_thread, timeout); + + if (current_thread->IsPendingTermination()) { + lock.CancelSleep(); + return ERR_THREAD_TERMINATING; + } + + // Ensure that we can read the address. + if (!memory.IsValidVirtualAddress(address)) { + lock.CancelSleep(); + return ERR_INVALID_ADDRESS_STATE; + } - // Only wait for the address if equal. - if (static_cast<s32>(memory.Read32(address)) != value) { - return ERR_INVALID_STATE; + s32 current_value = static_cast<s32>(memory.Read32(address)); + if (current_value != value) { + lock.CancelSleep(); + return ERR_INVALID_STATE; + } + + // Short-circuit without rescheduling, if timeout is zero. + if (timeout == 0) { + lock.CancelSleep(); + return RESULT_TIMEOUT; + } + + current_thread->SetSynchronizationResults(nullptr, RESULT_TIMEOUT); + current_thread->SetArbiterWaitAddress(address); + InsertThread(SharedFrom(current_thread)); + current_thread->SetStatus(ThreadStatus::WaitArb); + current_thread->WaitForArbitration(true); } - // Short-circuit without rescheduling if timeout is zero. - if (timeout == 0) { - return RESULT_TIMEOUT; + if (event_handle != InvalidHandle) { + auto& time_manager = kernel.TimeManager(); + time_manager.UnscheduleTimeEvent(event_handle); } - return WaitForAddressImpl(address, timeout); -} + { + SchedulerLock lock(kernel); + if (current_thread->IsWaitingForArbitration()) { + RemoveThread(SharedFrom(current_thread)); + current_thread->WaitForArbitration(false); + } + } -ResultCode AddressArbiter::WaitForAddressImpl(VAddr address, s64 timeout) { - Thread* current_thread = system.CurrentScheduler().GetCurrentThread(); - current_thread->SetArbiterWaitAddress(address); - InsertThread(SharedFrom(current_thread)); - current_thread->SetStatus(ThreadStatus::WaitArb); - current_thread->InvalidateWakeupCallback(); - current_thread->WakeAfterDelay(timeout); - - system.PrepareReschedule(current_thread->GetProcessorID()); - return RESULT_TIMEOUT; + return current_thread->GetSignalingResult(); } void AddressArbiter::HandleWakeupThread(std::shared_ptr<Thread> thread) { @@ -221,9 +305,9 @@ void AddressArbiter::RemoveThread(std::shared_ptr<Thread> thread) { const auto iter = std::find_if(thread_list.cbegin(), thread_list.cend(), [&thread](const auto& entry) { return thread == entry; }); - ASSERT(iter != thread_list.cend()); - - thread_list.erase(iter); + if (iter != thread_list.cend()) { + thread_list.erase(iter); + } } std::vector<std::shared_ptr<Thread>> AddressArbiter::GetThreadsWaitingOnAddress( diff --git a/src/core/hle/kernel/address_arbiter.h b/src/core/hle/kernel/address_arbiter.h index f958eee5a..0b05d533c 100644 --- a/src/core/hle/kernel/address_arbiter.h +++ b/src/core/hle/kernel/address_arbiter.h @@ -73,9 +73,6 @@ private: /// Waits on an address if the value passed is equal to the argument value. ResultCode WaitForAddressIfEqual(VAddr address, s32 value, s64 timeout); - // Waits on the given address with a timeout in nanoseconds - ResultCode WaitForAddressImpl(VAddr address, s64 timeout); - /// Wake up num_to_wake (or all) threads in a vector. void WakeThreads(const std::vector<std::shared_ptr<Thread>>& waiting_threads, s32 num_to_wake); diff --git a/src/core/hle/kernel/client_port.cpp b/src/core/hle/kernel/client_port.cpp index 5498fd313..8aff2227a 100644 --- a/src/core/hle/kernel/client_port.cpp +++ b/src/core/hle/kernel/client_port.cpp @@ -34,7 +34,7 @@ ResultVal<std::shared_ptr<ClientSession>> ClientPort::Connect() { } // Wake the threads waiting on the ServerPort - server_port->WakeupAllWaitingThreads(); + server_port->Signal(); return MakeResult(std::move(client)); } diff --git a/src/core/hle/kernel/errors.h b/src/core/hle/kernel/errors.h index 29bfa3621..d4e5d88cf 100644 --- a/src/core/hle/kernel/errors.h +++ b/src/core/hle/kernel/errors.h @@ -12,6 +12,7 @@ namespace Kernel { constexpr ResultCode ERR_MAX_CONNECTIONS_REACHED{ErrorModule::Kernel, 7}; constexpr ResultCode ERR_INVALID_CAPABILITY_DESCRIPTOR{ErrorModule::Kernel, 14}; +constexpr ResultCode ERR_THREAD_TERMINATING{ErrorModule::Kernel, 59}; constexpr ResultCode ERR_INVALID_SIZE{ErrorModule::Kernel, 101}; constexpr ResultCode ERR_INVALID_ADDRESS{ErrorModule::Kernel, 102}; constexpr ResultCode ERR_OUT_OF_RESOURCES{ErrorModule::Kernel, 103}; diff --git a/src/core/hle/kernel/hle_ipc.cpp b/src/core/hle/kernel/hle_ipc.cpp index 91d94025c..9277b5d08 100644 --- a/src/core/hle/kernel/hle_ipc.cpp +++ b/src/core/hle/kernel/hle_ipc.cpp @@ -13,16 +13,18 @@ #include "common/common_funcs.h" #include "common/common_types.h" #include "common/logging/log.h" -#include "core/core.h" #include "core/hle/ipc_helpers.h" +#include "core/hle/kernel/errors.h" #include "core/hle/kernel/handle_table.h" #include "core/hle/kernel/hle_ipc.h" #include "core/hle/kernel/kernel.h" #include "core/hle/kernel/object.h" #include "core/hle/kernel/process.h" #include "core/hle/kernel/readable_event.h" +#include "core/hle/kernel/scheduler.h" #include "core/hle/kernel/server_session.h" #include "core/hle/kernel/thread.h" +#include "core/hle/kernel/time_manager.h" #include "core/hle/kernel/writable_event.h" #include "core/memory.h" @@ -47,31 +49,33 @@ std::shared_ptr<WritableEvent> HLERequestContext::SleepClientThread( const std::string& reason, u64 timeout, WakeupCallback&& callback, std::shared_ptr<WritableEvent> writable_event) { // Put the client thread to sleep until the wait event is signaled or the timeout expires. - thread->SetWakeupCallback( - [context = *this, callback](ThreadWakeupReason reason, std::shared_ptr<Thread> thread, - std::shared_ptr<SynchronizationObject> object, - std::size_t index) mutable -> bool { - ASSERT(thread->GetStatus() == ThreadStatus::WaitHLEEvent); - callback(thread, context, reason); - context.WriteToOutgoingCommandBuffer(*thread); - return true; - }); - - auto& kernel = Core::System::GetInstance().Kernel(); + if (!writable_event) { // Create event if not provided const auto pair = WritableEvent::CreateEventPair(kernel, "HLE Pause Event: " + reason); writable_event = pair.writable; } - const auto readable_event{writable_event->GetReadableEvent()}; - writable_event->Clear(); - thread->SetStatus(ThreadStatus::WaitHLEEvent); - thread->SetSynchronizationObjects({readable_event}); - readable_event->AddWaitingThread(thread); - - if (timeout > 0) { - thread->WakeAfterDelay(timeout); + { + Handle event_handle = InvalidHandle; + SchedulerLockAndSleep lock(kernel, event_handle, thread.get(), timeout); + thread->SetHLECallback( + [context = *this, callback](std::shared_ptr<Thread> thread) mutable -> bool { + ThreadWakeupReason reason = thread->GetSignalingResult() == RESULT_TIMEOUT + ? ThreadWakeupReason::Timeout + : ThreadWakeupReason::Signal; + callback(thread, context, reason); + context.WriteToOutgoingCommandBuffer(*thread); + return true; + }); + const auto readable_event{writable_event->GetReadableEvent()}; + writable_event->Clear(); + thread->SetHLESyncObject(readable_event.get()); + thread->SetStatus(ThreadStatus::WaitHLEEvent); + thread->SetSynchronizationResults(nullptr, RESULT_TIMEOUT); + readable_event->AddWaitingThread(thread); + lock.Release(); + thread->SetHLETimeEvent(event_handle); } is_thread_waiting = true; @@ -79,9 +83,11 @@ std::shared_ptr<WritableEvent> HLERequestContext::SleepClientThread( return writable_event; } -HLERequestContext::HLERequestContext(std::shared_ptr<Kernel::ServerSession> server_session, +HLERequestContext::HLERequestContext(KernelCore& kernel, Core::Memory::Memory& memory, + std::shared_ptr<ServerSession> server_session, std::shared_ptr<Thread> thread) - : server_session(std::move(server_session)), thread(std::move(thread)) { + : server_session(std::move(server_session)), + thread(std::move(thread)), kernel{kernel}, memory{memory} { cmd_buf[0] = 0; } @@ -216,7 +222,6 @@ ResultCode HLERequestContext::PopulateFromIncomingCommandBuffer(const HandleTabl ResultCode HLERequestContext::WriteToOutgoingCommandBuffer(Thread& thread) { auto& owner_process = *thread.GetOwnerProcess(); auto& handle_table = owner_process.GetHandleTable(); - auto& memory = Core::System::GetInstance().Memory(); std::array<u32, IPC::COMMAND_BUFFER_LENGTH> dst_cmdbuf; memory.ReadBlock(owner_process, thread.GetTLSAddress(), dst_cmdbuf.data(), @@ -283,19 +288,18 @@ ResultCode HLERequestContext::WriteToOutgoingCommandBuffer(Thread& thread) { } std::vector<u8> HLERequestContext::ReadBuffer(std::size_t buffer_index) const { - std::vector<u8> buffer; + std::vector<u8> buffer{}; const bool is_buffer_a{BufferDescriptorA().size() > buffer_index && BufferDescriptorA()[buffer_index].Size()}; - auto& memory = Core::System::GetInstance().Memory(); if (is_buffer_a) { - ASSERT_MSG(BufferDescriptorA().size() > buffer_index, - "BufferDescriptorA invalid buffer_index {}", buffer_index); + ASSERT_OR_EXECUTE_MSG(BufferDescriptorA().size() > buffer_index, { return buffer; }, + "BufferDescriptorA invalid buffer_index {}", buffer_index); buffer.resize(BufferDescriptorA()[buffer_index].Size()); memory.ReadBlock(BufferDescriptorA()[buffer_index].Address(), buffer.data(), buffer.size()); } else { - ASSERT_MSG(BufferDescriptorX().size() > buffer_index, - "BufferDescriptorX invalid buffer_index {}", buffer_index); + ASSERT_OR_EXECUTE_MSG(BufferDescriptorX().size() > buffer_index, { return buffer; }, + "BufferDescriptorX invalid buffer_index {}", buffer_index); buffer.resize(BufferDescriptorX()[buffer_index].Size()); memory.ReadBlock(BufferDescriptorX()[buffer_index].Address(), buffer.data(), buffer.size()); } @@ -319,18 +323,17 @@ std::size_t HLERequestContext::WriteBuffer(const void* buffer, std::size_t size, size = buffer_size; // TODO(bunnei): This needs to be HW tested } - auto& memory = Core::System::GetInstance().Memory(); if (is_buffer_b) { - ASSERT_MSG(BufferDescriptorB().size() > buffer_index, - "BufferDescriptorB invalid buffer_index {}", buffer_index); - ASSERT_MSG(BufferDescriptorB()[buffer_index].Size() >= size, - "BufferDescriptorB buffer_index {} is not large enough", buffer_index); + ASSERT_OR_EXECUTE_MSG(BufferDescriptorB().size() > buffer_index && + BufferDescriptorB()[buffer_index].Size() >= size, + { return 0; }, "BufferDescriptorB is invalid, index={}, size={}", + buffer_index, size); memory.WriteBlock(BufferDescriptorB()[buffer_index].Address(), buffer, size); } else { - ASSERT_MSG(BufferDescriptorC().size() > buffer_index, - "BufferDescriptorC invalid buffer_index {}", buffer_index); - ASSERT_MSG(BufferDescriptorC()[buffer_index].Size() >= size, - "BufferDescriptorC buffer_index {} is not large enough", buffer_index); + ASSERT_OR_EXECUTE_MSG(BufferDescriptorC().size() > buffer_index && + BufferDescriptorC()[buffer_index].Size() >= size, + { return 0; }, "BufferDescriptorC is invalid, index={}, size={}", + buffer_index, size); memory.WriteBlock(BufferDescriptorC()[buffer_index].Address(), buffer, size); } @@ -341,16 +344,12 @@ std::size_t HLERequestContext::GetReadBufferSize(std::size_t buffer_index) const const bool is_buffer_a{BufferDescriptorA().size() > buffer_index && BufferDescriptorA()[buffer_index].Size()}; if (is_buffer_a) { - ASSERT_MSG(BufferDescriptorA().size() > buffer_index, - "BufferDescriptorA invalid buffer_index {}", buffer_index); - ASSERT_MSG(BufferDescriptorA()[buffer_index].Size() > 0, - "BufferDescriptorA buffer_index {} is empty", buffer_index); + ASSERT_OR_EXECUTE_MSG(BufferDescriptorA().size() > buffer_index, { return 0; }, + "BufferDescriptorA invalid buffer_index {}", buffer_index); return BufferDescriptorA()[buffer_index].Size(); } else { - ASSERT_MSG(BufferDescriptorX().size() > buffer_index, - "BufferDescriptorX invalid buffer_index {}", buffer_index); - ASSERT_MSG(BufferDescriptorX()[buffer_index].Size() > 0, - "BufferDescriptorX buffer_index {} is empty", buffer_index); + ASSERT_OR_EXECUTE_MSG(BufferDescriptorX().size() > buffer_index, { return 0; }, + "BufferDescriptorX invalid buffer_index {}", buffer_index); return BufferDescriptorX()[buffer_index].Size(); } } @@ -359,14 +358,15 @@ std::size_t HLERequestContext::GetWriteBufferSize(std::size_t buffer_index) cons const bool is_buffer_b{BufferDescriptorB().size() > buffer_index && BufferDescriptorB()[buffer_index].Size()}; if (is_buffer_b) { - ASSERT_MSG(BufferDescriptorB().size() > buffer_index, - "BufferDescriptorB invalid buffer_index {}", buffer_index); + ASSERT_OR_EXECUTE_MSG(BufferDescriptorB().size() > buffer_index, { return 0; }, + "BufferDescriptorB invalid buffer_index {}", buffer_index); return BufferDescriptorB()[buffer_index].Size(); } else { - ASSERT_MSG(BufferDescriptorC().size() > buffer_index, - "BufferDescriptorC invalid buffer_index {}", buffer_index); + ASSERT_OR_EXECUTE_MSG(BufferDescriptorC().size() > buffer_index, { return 0; }, + "BufferDescriptorC invalid buffer_index {}", buffer_index); return BufferDescriptorC()[buffer_index].Size(); } + return 0; } std::string HLERequestContext::Description() const { diff --git a/src/core/hle/kernel/hle_ipc.h b/src/core/hle/kernel/hle_ipc.h index af3330297..b31673928 100644 --- a/src/core/hle/kernel/hle_ipc.h +++ b/src/core/hle/kernel/hle_ipc.h @@ -19,6 +19,10 @@ union ResultCode; +namespace Core::Memory { +class Memory; +} + namespace Service { class ServiceFrameworkBase; } @@ -28,6 +32,7 @@ namespace Kernel { class Domain; class HandleTable; class HLERequestContext; +class KernelCore; class Process; class ServerSession; class Thread; @@ -98,7 +103,8 @@ protected: */ class HLERequestContext { public: - explicit HLERequestContext(std::shared_ptr<ServerSession> session, + explicit HLERequestContext(KernelCore& kernel, Core::Memory::Memory& memory, + std::shared_ptr<ServerSession> session, std::shared_ptr<Thread> thread); ~HLERequestContext(); @@ -305,6 +311,9 @@ private: std::vector<std::shared_ptr<SessionRequestHandler>> domain_request_handlers; bool is_thread_waiting{}; + + KernelCore& kernel; + Core::Memory::Memory& memory; }; } // namespace Kernel diff --git a/src/core/hle/kernel/kernel.cpp b/src/core/hle/kernel/kernel.cpp index 7655382fa..1f2af7a1b 100644 --- a/src/core/hle/kernel/kernel.cpp +++ b/src/core/hle/kernel/kernel.cpp @@ -2,6 +2,7 @@ // Licensed under GPLv2 or any later version // Refer to the license.txt file included. +#include <array> #include <atomic> #include <bitset> #include <functional> @@ -13,11 +14,15 @@ #include "common/assert.h" #include "common/logging/log.h" +#include "common/microprofile.h" +#include "common/thread.h" #include "core/arm/arm_interface.h" +#include "core/arm/cpu_interrupt_handler.h" #include "core/arm/exclusive_monitor.h" #include "core/core.h" #include "core/core_timing.h" #include "core/core_timing_util.h" +#include "core/cpu_manager.h" #include "core/device_memory.h" #include "core/hardware_properties.h" #include "core/hle/kernel/client_port.h" @@ -39,85 +44,28 @@ #include "core/hle/result.h" #include "core/memory.h" -namespace Kernel { - -/** - * Callback that will wake up the thread it was scheduled for - * @param thread_handle The handle of the thread that's been awoken - * @param cycles_late The number of CPU cycles that have passed since the desired wakeup time - */ -static void ThreadWakeupCallback(u64 thread_handle, [[maybe_unused]] s64 cycles_late) { - const auto proper_handle = static_cast<Handle>(thread_handle); - const auto& system = Core::System::GetInstance(); - - // Lock the global kernel mutex when we enter the kernel HLE. - std::lock_guard lock{HLE::g_hle_lock}; - - std::shared_ptr<Thread> thread = - system.Kernel().RetrieveThreadFromGlobalHandleTable(proper_handle); - if (thread == nullptr) { - LOG_CRITICAL(Kernel, "Callback fired for invalid thread {:08X}", proper_handle); - return; - } - - bool resume = true; - - if (thread->GetStatus() == ThreadStatus::WaitSynch || - thread->GetStatus() == ThreadStatus::WaitHLEEvent) { - // Remove the thread from each of its waiting objects' waitlists - for (const auto& object : thread->GetSynchronizationObjects()) { - object->RemoveWaitingThread(thread); - } - thread->ClearSynchronizationObjects(); - - // Invoke the wakeup callback before clearing the wait objects - if (thread->HasWakeupCallback()) { - resume = thread->InvokeWakeupCallback(ThreadWakeupReason::Timeout, thread, nullptr, 0); - } - } else if (thread->GetStatus() == ThreadStatus::WaitMutex || - thread->GetStatus() == ThreadStatus::WaitCondVar) { - thread->SetMutexWaitAddress(0); - thread->SetWaitHandle(0); - if (thread->GetStatus() == ThreadStatus::WaitCondVar) { - thread->GetOwnerProcess()->RemoveConditionVariableThread(thread); - thread->SetCondVarWaitAddress(0); - } - - auto* const lock_owner = thread->GetLockOwner(); - // Threads waking up by timeout from WaitProcessWideKey do not perform priority inheritance - // and don't have a lock owner unless SignalProcessWideKey was called first and the thread - // wasn't awakened due to the mutex already being acquired. - if (lock_owner != nullptr) { - lock_owner->RemoveMutexWaiter(thread); - } - } +MICROPROFILE_DEFINE(Kernel_SVC, "Kernel", "SVC", MP_RGB(70, 200, 70)); - if (thread->GetStatus() == ThreadStatus::WaitArb) { - auto& address_arbiter = thread->GetOwnerProcess()->GetAddressArbiter(); - address_arbiter.HandleWakeupThread(thread); - } - - if (resume) { - if (thread->GetStatus() == ThreadStatus::WaitCondVar || - thread->GetStatus() == ThreadStatus::WaitArb) { - thread->SetWaitSynchronizationResult(RESULT_TIMEOUT); - } - thread->ResumeFromWait(); - } -} +namespace Kernel { struct KernelCore::Impl { explicit Impl(Core::System& system, KernelCore& kernel) : global_scheduler{kernel}, synchronization{system}, time_manager{system}, system{system} {} + void SetMulticore(bool is_multicore) { + this->is_multicore = is_multicore; + } + void Initialize(KernelCore& kernel) { Shutdown(); + RegisterHostThread(); InitializePhysicalCores(); InitializeSystemResourceLimit(kernel); InitializeMemoryLayout(); - InitializeThreads(); - InitializePreemption(); + InitializePreemption(kernel); + InitializeSchedulers(); + InitializeSuspendThreads(); } void Shutdown() { @@ -126,13 +74,26 @@ struct KernelCore::Impl { next_user_process_id = Process::ProcessIDMin; next_thread_id = 1; + for (std::size_t i = 0; i < Core::Hardware::NUM_CPU_CORES; i++) { + if (suspend_threads[i]) { + suspend_threads[i].reset(); + } + } + + for (std::size_t i = 0; i < cores.size(); i++) { + cores[i].Shutdown(); + schedulers[i].reset(); + } + cores.clear(); + + registered_core_threads.reset(); + process_list.clear(); current_process = nullptr; system_resource_limit = nullptr; global_handle_table.Clear(); - thread_wakeup_event_type = nullptr; preemption_event = nullptr; global_scheduler.Shutdown(); @@ -145,13 +106,21 @@ struct KernelCore::Impl { cores.clear(); exclusive_monitor.reset(); + host_thread_ids.clear(); } void InitializePhysicalCores() { exclusive_monitor = Core::MakeExclusiveMonitor(system.Memory(), Core::Hardware::NUM_CPU_CORES); for (std::size_t i = 0; i < Core::Hardware::NUM_CPU_CORES; i++) { - cores.emplace_back(system, i, *exclusive_monitor); + schedulers[i] = std::make_unique<Kernel::Scheduler>(system, i); + cores.emplace_back(system, i, *schedulers[i], interrupts[i]); + } + } + + void InitializeSchedulers() { + for (std::size_t i = 0; i < Core::Hardware::NUM_CPU_CORES; i++) { + cores[i].Scheduler().Initialize(); } } @@ -173,15 +142,13 @@ struct KernelCore::Impl { } } - void InitializeThreads() { - thread_wakeup_event_type = - Core::Timing::CreateEvent("ThreadWakeupCallback", ThreadWakeupCallback); - } - - void InitializePreemption() { - preemption_event = - Core::Timing::CreateEvent("PreemptionCallback", [this](u64 userdata, s64 cycles_late) { - global_scheduler.PreemptThreads(); + void InitializePreemption(KernelCore& kernel) { + preemption_event = Core::Timing::CreateEvent( + "PreemptionCallback", [this, &kernel](u64 userdata, s64 cycles_late) { + { + SchedulerLock lock(kernel); + global_scheduler.PreemptThreads(); + } s64 time_interval = Core::Timing::msToCycles(std::chrono::milliseconds(10)); system.CoreTiming().ScheduleEvent(time_interval, preemption_event); }); @@ -190,6 +157,20 @@ struct KernelCore::Impl { system.CoreTiming().ScheduleEvent(time_interval, preemption_event); } + void InitializeSuspendThreads() { + for (std::size_t i = 0; i < Core::Hardware::NUM_CPU_CORES; i++) { + std::string name = "Suspend Thread Id:" + std::to_string(i); + std::function<void(void*)> init_func = + system.GetCpuManager().GetSuspendThreadStartFunc(); + void* init_func_parameter = system.GetCpuManager().GetStartFuncParamater(); + ThreadType type = + static_cast<ThreadType>(THREADTYPE_KERNEL | THREADTYPE_HLE | THREADTYPE_SUSPEND); + auto thread_res = Thread::Create(system, type, name, 0, 0, 0, static_cast<u32>(i), 0, + nullptr, std::move(init_func), init_func_parameter); + suspend_threads[i] = std::move(thread_res).Unwrap(); + } + } + void MakeCurrentProcess(Process* process) { current_process = process; @@ -197,15 +178,17 @@ struct KernelCore::Impl { return; } - for (auto& core : cores) { - core.SetIs64Bit(process->Is64BitProcess()); + u32 core_id = GetCurrentHostThreadID(); + if (core_id < Core::Hardware::NUM_CPU_CORES) { + system.Memory().SetCurrentPageTable(*process, core_id); } - - system.Memory().SetCurrentPageTable(*process); } void RegisterCoreThread(std::size_t core_id) { std::unique_lock lock{register_thread_mutex}; + if (!is_multicore) { + single_core_thread_id = std::this_thread::get_id(); + } const std::thread::id this_id = std::this_thread::get_id(); const auto it = host_thread_ids.find(this_id); ASSERT(core_id < Core::Hardware::NUM_CPU_CORES); @@ -219,12 +202,19 @@ struct KernelCore::Impl { std::unique_lock lock{register_thread_mutex}; const std::thread::id this_id = std::this_thread::get_id(); const auto it = host_thread_ids.find(this_id); - ASSERT(it == host_thread_ids.end()); + if (it != host_thread_ids.end()) { + return; + } host_thread_ids[this_id] = registered_thread_ids++; } u32 GetCurrentHostThreadID() const { const std::thread::id this_id = std::this_thread::get_id(); + if (!is_multicore) { + if (single_core_thread_id == this_id) { + return static_cast<u32>(system.GetCpuManager().CurrentCore()); + } + } const auto it = host_thread_ids.find(this_id); if (it == host_thread_ids.end()) { return Core::INVALID_HOST_THREAD_ID; @@ -240,7 +230,7 @@ struct KernelCore::Impl { } const Kernel::Scheduler& sched = cores[result.host_handle].Scheduler(); const Kernel::Thread* current = sched.GetCurrentThread(); - if (current != nullptr) { + if (current != nullptr && !current->IsPhantomMode()) { result.guest_handle = current->GetGlobalHandle(); } else { result.guest_handle = InvalidHandle; @@ -313,7 +303,6 @@ struct KernelCore::Impl { std::shared_ptr<ResourceLimit> system_resource_limit; - std::shared_ptr<Core::Timing::EventType> thread_wakeup_event_type; std::shared_ptr<Core::Timing::EventType> preemption_event; // This is the kernel's handle table or supervisor handle table which @@ -343,6 +332,15 @@ struct KernelCore::Impl { std::shared_ptr<Kernel::SharedMemory> irs_shared_mem; std::shared_ptr<Kernel::SharedMemory> time_shared_mem; + std::array<std::shared_ptr<Thread>, Core::Hardware::NUM_CPU_CORES> suspend_threads{}; + std::array<Core::CPUInterruptHandler, Core::Hardware::NUM_CPU_CORES> interrupts{}; + std::array<std::unique_ptr<Kernel::Scheduler>, Core::Hardware::NUM_CPU_CORES> schedulers{}; + + bool is_multicore{}; + std::thread::id single_core_thread_id{}; + + std::array<u64, Core::Hardware::NUM_CPU_CORES> svc_ticks{}; + // System context Core::System& system; }; @@ -352,6 +350,10 @@ KernelCore::~KernelCore() { Shutdown(); } +void KernelCore::SetMulticore(bool is_multicore) { + impl->SetMulticore(is_multicore); +} + void KernelCore::Initialize() { impl->Initialize(*this); } @@ -397,11 +399,11 @@ const Kernel::GlobalScheduler& KernelCore::GlobalScheduler() const { } Kernel::Scheduler& KernelCore::Scheduler(std::size_t id) { - return impl->cores[id].Scheduler(); + return *impl->schedulers[id]; } const Kernel::Scheduler& KernelCore::Scheduler(std::size_t id) const { - return impl->cores[id].Scheduler(); + return *impl->schedulers[id]; } Kernel::PhysicalCore& KernelCore::PhysicalCore(std::size_t id) { @@ -412,6 +414,39 @@ const Kernel::PhysicalCore& KernelCore::PhysicalCore(std::size_t id) const { return impl->cores[id]; } +Kernel::PhysicalCore& KernelCore::CurrentPhysicalCore() { + u32 core_id = impl->GetCurrentHostThreadID(); + ASSERT(core_id < Core::Hardware::NUM_CPU_CORES); + return impl->cores[core_id]; +} + +const Kernel::PhysicalCore& KernelCore::CurrentPhysicalCore() const { + u32 core_id = impl->GetCurrentHostThreadID(); + ASSERT(core_id < Core::Hardware::NUM_CPU_CORES); + return impl->cores[core_id]; +} + +Kernel::Scheduler& KernelCore::CurrentScheduler() { + u32 core_id = impl->GetCurrentHostThreadID(); + ASSERT(core_id < Core::Hardware::NUM_CPU_CORES); + return *impl->schedulers[core_id]; +} + +const Kernel::Scheduler& KernelCore::CurrentScheduler() const { + u32 core_id = impl->GetCurrentHostThreadID(); + ASSERT(core_id < Core::Hardware::NUM_CPU_CORES); + return *impl->schedulers[core_id]; +} + +std::array<Core::CPUInterruptHandler, Core::Hardware::NUM_CPU_CORES>& KernelCore::Interrupts() { + return impl->interrupts; +} + +const std::array<Core::CPUInterruptHandler, Core::Hardware::NUM_CPU_CORES>& KernelCore::Interrupts() + const { + return impl->interrupts; +} + Kernel::Synchronization& KernelCore::Synchronization() { return impl->synchronization; } @@ -437,15 +472,17 @@ const Core::ExclusiveMonitor& KernelCore::GetExclusiveMonitor() const { } void KernelCore::InvalidateAllInstructionCaches() { - for (std::size_t i = 0; i < impl->global_scheduler.CpuCoresCount(); i++) { - PhysicalCore(i).ArmInterface().ClearInstructionCache(); + auto& threads = GlobalScheduler().GetThreadList(); + for (auto& thread : threads) { + if (!thread->IsHLEThread()) { + auto& arm_interface = thread->ArmInterface(); + arm_interface.ClearInstructionCache(); + } } } void KernelCore::PrepareReschedule(std::size_t id) { - if (id < impl->global_scheduler.CpuCoresCount()) { - impl->cores[id].Stop(); - } + // TODO: Reimplement, this } void KernelCore::AddNamedPort(std::string name, std::shared_ptr<ClientPort> port) { @@ -481,10 +518,6 @@ u64 KernelCore::CreateNewUserProcessID() { return impl->next_user_process_id++; } -const std::shared_ptr<Core::Timing::EventType>& KernelCore::ThreadWakeupCallbackEventType() const { - return impl->thread_wakeup_event_type; -} - Kernel::HandleTable& KernelCore::GlobalHandleTable() { return impl->global_handle_table; } @@ -557,4 +590,34 @@ const Kernel::SharedMemory& KernelCore::GetTimeSharedMem() const { return *impl->time_shared_mem; } +void KernelCore::Suspend(bool in_suspention) { + const bool should_suspend = exception_exited || in_suspention; + { + SchedulerLock lock(*this); + ThreadStatus status = should_suspend ? ThreadStatus::Ready : ThreadStatus::WaitSleep; + for (std::size_t i = 0; i < Core::Hardware::NUM_CPU_CORES; i++) { + impl->suspend_threads[i]->SetStatus(status); + } + } +} + +bool KernelCore::IsMulticore() const { + return impl->is_multicore; +} + +void KernelCore::ExceptionalExit() { + exception_exited = true; + Suspend(true); +} + +void KernelCore::EnterSVCProfile() { + std::size_t core = impl->GetCurrentHostThreadID(); + impl->svc_ticks[core] = MicroProfileEnter(MICROPROFILE_TOKEN(Kernel_SVC)); +} + +void KernelCore::ExitSVCProfile() { + std::size_t core = impl->GetCurrentHostThreadID(); + MicroProfileLeave(MICROPROFILE_TOKEN(Kernel_SVC), impl->svc_ticks[core]); +} + } // namespace Kernel diff --git a/src/core/hle/kernel/kernel.h b/src/core/hle/kernel/kernel.h index 83de1f542..49bd47e89 100644 --- a/src/core/hle/kernel/kernel.h +++ b/src/core/hle/kernel/kernel.h @@ -4,15 +4,17 @@ #pragma once +#include <array> #include <memory> #include <string> #include <unordered_map> #include <vector> +#include "core/hardware_properties.h" #include "core/hle/kernel/memory/memory_types.h" #include "core/hle/kernel/object.h" namespace Core { -struct EmuThreadHandle; +class CPUInterruptHandler; class ExclusiveMonitor; class System; } // namespace Core @@ -65,6 +67,9 @@ public: KernelCore(KernelCore&&) = delete; KernelCore& operator=(KernelCore&&) = delete; + /// Sets if emulation is multicore or single core, must be set before Initialize + void SetMulticore(bool is_multicore); + /// Resets the kernel to a clean slate for use. void Initialize(); @@ -110,6 +115,18 @@ public: /// Gets the an instance of the respective physical CPU core. const Kernel::PhysicalCore& PhysicalCore(std::size_t id) const; + /// Gets the sole instance of the Scheduler at the current running core. + Kernel::Scheduler& CurrentScheduler(); + + /// Gets the sole instance of the Scheduler at the current running core. + const Kernel::Scheduler& CurrentScheduler() const; + + /// Gets the an instance of the current physical CPU core. + Kernel::PhysicalCore& CurrentPhysicalCore(); + + /// Gets the an instance of the current physical CPU core. + const Kernel::PhysicalCore& CurrentPhysicalCore() const; + /// Gets the an instance of the Synchronization Interface. Kernel::Synchronization& Synchronization(); @@ -129,6 +146,10 @@ public: const Core::ExclusiveMonitor& GetExclusiveMonitor() const; + std::array<Core::CPUInterruptHandler, Core::Hardware::NUM_CPU_CORES>& Interrupts(); + + const std::array<Core::CPUInterruptHandler, Core::Hardware::NUM_CPU_CORES>& Interrupts() const; + void InvalidateAllInstructionCaches(); /// Adds a port to the named port table @@ -191,6 +212,18 @@ public: /// Gets the shared memory object for Time services. const Kernel::SharedMemory& GetTimeSharedMem() const; + /// Suspend/unsuspend the OS. + void Suspend(bool in_suspention); + + /// Exceptional exit the OS. + void ExceptionalExit(); + + bool IsMulticore() const; + + void EnterSVCProfile(); + + void ExitSVCProfile(); + private: friend class Object; friend class Process; @@ -208,9 +241,6 @@ private: /// Creates a new thread ID, incrementing the internal thread ID counter. u64 CreateNewThreadID(); - /// Retrieves the event type used for thread wakeup callbacks. - const std::shared_ptr<Core::Timing::EventType>& ThreadWakeupCallbackEventType() const; - /// Provides a reference to the global handle table. Kernel::HandleTable& GlobalHandleTable(); @@ -219,6 +249,7 @@ private: struct Impl; std::unique_ptr<Impl> impl; + bool exception_exited{}; }; } // namespace Kernel diff --git a/src/core/hle/kernel/memory/memory_manager.cpp b/src/core/hle/kernel/memory/memory_manager.cpp index 6b432e1b2..acf13585c 100644 --- a/src/core/hle/kernel/memory/memory_manager.cpp +++ b/src/core/hle/kernel/memory/memory_manager.cpp @@ -104,7 +104,7 @@ ResultCode MemoryManager::Allocate(PageLinkedList& page_list, std::size_t num_pa // Ensure that we don't leave anything un-freed auto group_guard = detail::ScopeExit([&] { for (const auto& it : page_list.Nodes()) { - const auto min_num_pages{std::min( + const auto min_num_pages{std::min<size_t>( it.GetNumPages(), (chosen_manager.GetEndAddress() - it.GetAddress()) / PageSize)}; chosen_manager.Free(it.GetAddress(), min_num_pages); } @@ -139,7 +139,6 @@ ResultCode MemoryManager::Allocate(PageLinkedList& page_list, std::size_t num_pa } // Only succeed if we allocated as many pages as we wanted - ASSERT(num_pages >= 0); if (num_pages) { return ERR_OUT_OF_MEMORY; } @@ -165,7 +164,7 @@ ResultCode MemoryManager::Free(PageLinkedList& page_list, std::size_t num_pages, // Free all of the pages for (const auto& it : page_list.Nodes()) { - const auto min_num_pages{std::min( + const auto min_num_pages{std::min<size_t>( it.GetNumPages(), (chosen_manager.GetEndAddress() - it.GetAddress()) / PageSize)}; chosen_manager.Free(it.GetAddress(), min_num_pages); } diff --git a/src/core/hle/kernel/mutex.cpp b/src/core/hle/kernel/mutex.cpp index 7869eb32b..8f6c944d1 100644 --- a/src/core/hle/kernel/mutex.cpp +++ b/src/core/hle/kernel/mutex.cpp @@ -34,8 +34,6 @@ static std::pair<std::shared_ptr<Thread>, u32> GetHighestPriorityMutexWaitingThr if (thread->GetMutexWaitAddress() != mutex_addr) continue; - ASSERT(thread->GetStatus() == ThreadStatus::WaitMutex); - ++num_waiters; if (highest_priority_thread == nullptr || thread->GetPriority() < highest_priority_thread->GetPriority()) { @@ -49,6 +47,7 @@ static std::pair<std::shared_ptr<Thread>, u32> GetHighestPriorityMutexWaitingThr /// Update the mutex owner field of all threads waiting on the mutex to point to the new owner. static void TransferMutexOwnership(VAddr mutex_addr, std::shared_ptr<Thread> current_thread, std::shared_ptr<Thread> new_owner) { + current_thread->RemoveMutexWaiter(new_owner); const auto threads = current_thread->GetMutexWaitingThreads(); for (const auto& thread : threads) { if (thread->GetMutexWaitAddress() != mutex_addr) @@ -72,85 +71,100 @@ ResultCode Mutex::TryAcquire(VAddr address, Handle holding_thread_handle, return ERR_INVALID_ADDRESS; } - const auto& handle_table = system.Kernel().CurrentProcess()->GetHandleTable(); + auto& kernel = system.Kernel(); std::shared_ptr<Thread> current_thread = - SharedFrom(system.CurrentScheduler().GetCurrentThread()); - std::shared_ptr<Thread> holding_thread = handle_table.Get<Thread>(holding_thread_handle); - std::shared_ptr<Thread> requesting_thread = handle_table.Get<Thread>(requesting_thread_handle); + SharedFrom(kernel.CurrentScheduler().GetCurrentThread()); + { + SchedulerLock lock(kernel); + // The mutex address must be 4-byte aligned + if ((address % sizeof(u32)) != 0) { + return ERR_INVALID_ADDRESS; + } - // TODO(Subv): It is currently unknown if it is possible to lock a mutex in behalf of another - // thread. - ASSERT(requesting_thread == current_thread); + const auto& handle_table = kernel.CurrentProcess()->GetHandleTable(); + std::shared_ptr<Thread> holding_thread = handle_table.Get<Thread>(holding_thread_handle); + std::shared_ptr<Thread> requesting_thread = + handle_table.Get<Thread>(requesting_thread_handle); - const u32 addr_value = system.Memory().Read32(address); + // TODO(Subv): It is currently unknown if it is possible to lock a mutex in behalf of + // another thread. + ASSERT(requesting_thread == current_thread); - // If the mutex isn't being held, just return success. - if (addr_value != (holding_thread_handle | Mutex::MutexHasWaitersFlag)) { - return RESULT_SUCCESS; - } + current_thread->SetSynchronizationResults(nullptr, RESULT_SUCCESS); - if (holding_thread == nullptr) { - LOG_ERROR(Kernel, "Holding thread does not exist! thread_handle={:08X}", - holding_thread_handle); - return ERR_INVALID_HANDLE; - } + const u32 addr_value = system.Memory().Read32(address); + + // If the mutex isn't being held, just return success. + if (addr_value != (holding_thread_handle | Mutex::MutexHasWaitersFlag)) { + return RESULT_SUCCESS; + } - // Wait until the mutex is released - current_thread->SetMutexWaitAddress(address); - current_thread->SetWaitHandle(requesting_thread_handle); + if (holding_thread == nullptr) { + return ERR_INVALID_HANDLE; + } - current_thread->SetStatus(ThreadStatus::WaitMutex); - current_thread->InvalidateWakeupCallback(); + // Wait until the mutex is released + current_thread->SetMutexWaitAddress(address); + current_thread->SetWaitHandle(requesting_thread_handle); - // Update the lock holder thread's priority to prevent priority inversion. - holding_thread->AddMutexWaiter(current_thread); + current_thread->SetStatus(ThreadStatus::WaitMutex); - system.PrepareReschedule(); + // Update the lock holder thread's priority to prevent priority inversion. + holding_thread->AddMutexWaiter(current_thread); + } - return RESULT_SUCCESS; + { + SchedulerLock lock(kernel); + auto* owner = current_thread->GetLockOwner(); + if (owner != nullptr) { + owner->RemoveMutexWaiter(current_thread); + } + } + return current_thread->GetSignalingResult(); } -ResultCode Mutex::Release(VAddr address) { +std::pair<ResultCode, std::shared_ptr<Thread>> Mutex::Unlock(std::shared_ptr<Thread> owner, + VAddr address) { // The mutex address must be 4-byte aligned if ((address % sizeof(u32)) != 0) { LOG_ERROR(Kernel, "Address is not 4-byte aligned! address={:016X}", address); - return ERR_INVALID_ADDRESS; + return {ERR_INVALID_ADDRESS, nullptr}; } - std::shared_ptr<Thread> current_thread = - SharedFrom(system.CurrentScheduler().GetCurrentThread()); - auto [thread, num_waiters] = GetHighestPriorityMutexWaitingThread(current_thread, address); - - // There are no more threads waiting for the mutex, release it completely. - if (thread == nullptr) { + auto [new_owner, num_waiters] = GetHighestPriorityMutexWaitingThread(owner, address); + if (new_owner == nullptr) { system.Memory().Write32(address, 0); - return RESULT_SUCCESS; + return {RESULT_SUCCESS, nullptr}; } - // Transfer the ownership of the mutex from the previous owner to the new one. - TransferMutexOwnership(address, current_thread, thread); - - u32 mutex_value = thread->GetWaitHandle(); - + TransferMutexOwnership(address, owner, new_owner); + u32 mutex_value = new_owner->GetWaitHandle(); if (num_waiters >= 2) { // Notify the guest that there are still some threads waiting for the mutex mutex_value |= Mutex::MutexHasWaitersFlag; } + new_owner->SetSynchronizationResults(nullptr, RESULT_SUCCESS); + new_owner->SetLockOwner(nullptr); + new_owner->ResumeFromWait(); - // Grant the mutex to the next waiting thread and resume it. system.Memory().Write32(address, mutex_value); + return {RESULT_SUCCESS, new_owner}; +} - ASSERT(thread->GetStatus() == ThreadStatus::WaitMutex); - thread->ResumeFromWait(); +ResultCode Mutex::Release(VAddr address) { + auto& kernel = system.Kernel(); + SchedulerLock lock(kernel); - thread->SetLockOwner(nullptr); - thread->SetCondVarWaitAddress(0); - thread->SetMutexWaitAddress(0); - thread->SetWaitHandle(0); - thread->SetWaitSynchronizationResult(RESULT_SUCCESS); + std::shared_ptr<Thread> current_thread = + SharedFrom(kernel.CurrentScheduler().GetCurrentThread()); - system.PrepareReschedule(); + auto [result, new_owner] = Unlock(current_thread, address); - return RESULT_SUCCESS; + if (result != RESULT_SUCCESS && new_owner != nullptr) { + new_owner->SetSynchronizationResults(nullptr, result); + } + + return result; } + } // namespace Kernel diff --git a/src/core/hle/kernel/mutex.h b/src/core/hle/kernel/mutex.h index b904de2e8..3b81dc3df 100644 --- a/src/core/hle/kernel/mutex.h +++ b/src/core/hle/kernel/mutex.h @@ -28,6 +28,10 @@ public: ResultCode TryAcquire(VAddr address, Handle holding_thread_handle, Handle requesting_thread_handle); + /// Unlocks a mutex for owner at address + std::pair<ResultCode, std::shared_ptr<Thread>> Unlock(std::shared_ptr<Thread> owner, + VAddr address); + /// Releases the mutex at the specified address. ResultCode Release(VAddr address); diff --git a/src/core/hle/kernel/physical_core.cpp b/src/core/hle/kernel/physical_core.cpp index a15011076..c6bbdb080 100644 --- a/src/core/hle/kernel/physical_core.cpp +++ b/src/core/hle/kernel/physical_core.cpp @@ -2,12 +2,15 @@ // Licensed under GPLv2 or any later version // Refer to the license.txt file included. +#include "common/assert.h" #include "common/logging/log.h" +#include "common/spin_lock.h" #include "core/arm/arm_interface.h" #ifdef ARCHITECTURE_x86_64 #include "core/arm/dynarmic/arm_dynarmic_32.h" #include "core/arm/dynarmic/arm_dynarmic_64.h" #endif +#include "core/arm/cpu_interrupt_handler.h" #include "core/arm/exclusive_monitor.h" #include "core/arm/unicorn/arm_unicorn.h" #include "core/core.h" @@ -17,50 +20,37 @@ namespace Kernel { -PhysicalCore::PhysicalCore(Core::System& system, std::size_t id, - Core::ExclusiveMonitor& exclusive_monitor) - : core_index{id} { -#ifdef ARCHITECTURE_x86_64 - arm_interface_32 = - std::make_unique<Core::ARM_Dynarmic_32>(system, exclusive_monitor, core_index); - arm_interface_64 = - std::make_unique<Core::ARM_Dynarmic_64>(system, exclusive_monitor, core_index); - -#else - using Core::ARM_Unicorn; - arm_interface_32 = std::make_unique<ARM_Unicorn>(system, ARM_Unicorn::Arch::AArch32); - arm_interface_64 = std::make_unique<ARM_Unicorn>(system, ARM_Unicorn::Arch::AArch64); - LOG_WARNING(Core, "CPU JIT requested, but Dynarmic not available"); -#endif +PhysicalCore::PhysicalCore(Core::System& system, std::size_t id, Kernel::Scheduler& scheduler, + Core::CPUInterruptHandler& interrupt_handler) + : interrupt_handler{interrupt_handler}, core_index{id}, scheduler{scheduler} { - scheduler = std::make_unique<Kernel::Scheduler>(system, core_index); + guard = std::make_unique<Common::SpinLock>(); } PhysicalCore::~PhysicalCore() = default; -void PhysicalCore::Run() { - arm_interface->Run(); - arm_interface->ClearExclusiveState(); +void PhysicalCore::Idle() { + interrupt_handler.AwaitInterrupt(); } -void PhysicalCore::Step() { - arm_interface->Step(); +void PhysicalCore::Shutdown() { + scheduler.Shutdown(); } -void PhysicalCore::Stop() { - arm_interface->PrepareReschedule(); +bool PhysicalCore::IsInterrupted() const { + return interrupt_handler.IsInterrupted(); } -void PhysicalCore::Shutdown() { - scheduler->Shutdown(); +void PhysicalCore::Interrupt() { + guard->lock(); + interrupt_handler.SetInterrupt(true); + guard->unlock(); } -void PhysicalCore::SetIs64Bit(bool is_64_bit) { - if (is_64_bit) { - arm_interface = arm_interface_64.get(); - } else { - arm_interface = arm_interface_32.get(); - } +void PhysicalCore::ClearInterrupt() { + guard->lock(); + interrupt_handler.SetInterrupt(false); + guard->unlock(); } } // namespace Kernel diff --git a/src/core/hle/kernel/physical_core.h b/src/core/hle/kernel/physical_core.h index 3269166be..d7a7a951c 100644 --- a/src/core/hle/kernel/physical_core.h +++ b/src/core/hle/kernel/physical_core.h @@ -7,12 +7,17 @@ #include <cstddef> #include <memory> +namespace Common { +class SpinLock; +} + namespace Kernel { class Scheduler; } // namespace Kernel namespace Core { class ARM_Interface; +class CPUInterruptHandler; class ExclusiveMonitor; class System; } // namespace Core @@ -21,7 +26,8 @@ namespace Kernel { class PhysicalCore { public: - PhysicalCore(Core::System& system, std::size_t id, Core::ExclusiveMonitor& exclusive_monitor); + PhysicalCore(Core::System& system, std::size_t id, Kernel::Scheduler& scheduler, + Core::CPUInterruptHandler& interrupt_handler); ~PhysicalCore(); PhysicalCore(const PhysicalCore&) = delete; @@ -30,23 +36,18 @@ public: PhysicalCore(PhysicalCore&&) = default; PhysicalCore& operator=(PhysicalCore&&) = default; - /// Execute current jit state - void Run(); - /// Execute a single instruction in current jit. - void Step(); - /// Stop JIT execution/exit - void Stop(); + void Idle(); + /// Interrupt this physical core. + void Interrupt(); - // Shutdown this physical core. - void Shutdown(); + /// Clear this core's interrupt + void ClearInterrupt(); - Core::ARM_Interface& ArmInterface() { - return *arm_interface; - } + /// Check if this core is interrupted + bool IsInterrupted() const; - const Core::ARM_Interface& ArmInterface() const { - return *arm_interface; - } + // Shutdown this physical core. + void Shutdown(); bool IsMainCore() const { return core_index == 0; @@ -61,21 +62,18 @@ public: } Kernel::Scheduler& Scheduler() { - return *scheduler; + return scheduler; } const Kernel::Scheduler& Scheduler() const { - return *scheduler; + return scheduler; } - void SetIs64Bit(bool is_64_bit); - private: + Core::CPUInterruptHandler& interrupt_handler; std::size_t core_index; - std::unique_ptr<Core::ARM_Interface> arm_interface_32; - std::unique_ptr<Core::ARM_Interface> arm_interface_64; - std::unique_ptr<Kernel::Scheduler> scheduler; - Core::ARM_Interface* arm_interface{}; + Kernel::Scheduler& scheduler; + std::unique_ptr<Common::SpinLock> guard; }; } // namespace Kernel diff --git a/src/core/hle/kernel/process.cpp b/src/core/hle/kernel/process.cpp index 36724569f..f9d7c024d 100644 --- a/src/core/hle/kernel/process.cpp +++ b/src/core/hle/kernel/process.cpp @@ -22,6 +22,7 @@ #include "core/hle/kernel/resource_limit.h" #include "core/hle/kernel/scheduler.h" #include "core/hle/kernel/thread.h" +#include "core/hle/lock.h" #include "core/memory.h" #include "core/settings.h" @@ -30,14 +31,15 @@ namespace { /** * Sets up the primary application thread * + * @param system The system instance to create the main thread under. * @param owner_process The parent process for the main thread - * @param kernel The kernel instance to create the main thread under. * @param priority The priority to give the main thread */ -void SetupMainThread(Process& owner_process, KernelCore& kernel, u32 priority, VAddr stack_top) { +void SetupMainThread(Core::System& system, Process& owner_process, u32 priority, VAddr stack_top) { const VAddr entry_point = owner_process.PageTable().GetCodeRegionStart(); - auto thread_res = Thread::Create(kernel, "main", entry_point, priority, 0, - owner_process.GetIdealCore(), stack_top, owner_process); + ThreadType type = THREADTYPE_USER; + auto thread_res = Thread::Create(system, type, "main", entry_point, priority, 0, + owner_process.GetIdealCore(), stack_top, &owner_process); std::shared_ptr<Thread> thread = std::move(thread_res).Unwrap(); @@ -48,8 +50,12 @@ void SetupMainThread(Process& owner_process, KernelCore& kernel, u32 priority, V thread->GetContext32().cpu_registers[1] = thread_handle; thread->GetContext64().cpu_registers[1] = thread_handle; + auto& kernel = system.Kernel(); // Threads by default are dormant, wake up the main thread so it runs when the scheduler fires - thread->ResumeFromWait(); + { + SchedulerLock lock{kernel}; + thread->SetStatus(ThreadStatus::Ready); + } } } // Anonymous namespace @@ -132,7 +138,8 @@ std::shared_ptr<ResourceLimit> Process::GetResourceLimit() const { u64 Process::GetTotalPhysicalMemoryAvailable() const { const u64 capacity{resource_limit->GetCurrentResourceValue(ResourceType::PhysicalMemory) + - page_table->GetTotalHeapSize() + image_size + main_thread_stack_size}; + page_table->GetTotalHeapSize() + GetSystemResourceSize() + image_size + + main_thread_stack_size}; if (capacity < memory_usage_capacity) { return capacity; @@ -146,7 +153,8 @@ u64 Process::GetTotalPhysicalMemoryAvailableWithoutSystemResource() const { } u64 Process::GetTotalPhysicalMemoryUsed() const { - return image_size + main_thread_stack_size + page_table->GetTotalHeapSize(); + return image_size + main_thread_stack_size + page_table->GetTotalHeapSize() + + GetSystemResourceSize(); } u64 Process::GetTotalPhysicalMemoryUsedWithoutSystemResource() const { @@ -180,7 +188,6 @@ void Process::RemoveConditionVariableThread(std::shared_ptr<Thread> thread) { } ++it; } - UNREACHABLE(); } std::vector<std::shared_ptr<Thread>> Process::GetConditionVariableThreads( @@ -205,6 +212,7 @@ void Process::UnregisterThread(const Thread* thread) { } ResultCode Process::ClearSignalState() { + SchedulerLock lock(system.Kernel()); if (status == ProcessStatus::Exited) { LOG_ERROR(Kernel, "called on a terminated process instance."); return ERR_INVALID_STATE; @@ -292,7 +300,7 @@ void Process::Run(s32 main_thread_priority, u64 stack_size) { ChangeStatus(ProcessStatus::Running); - SetupMainThread(*this, kernel, main_thread_priority, main_thread_stack_top); + SetupMainThread(system, *this, main_thread_priority, main_thread_stack_top); resource_limit->Reserve(ResourceType::Threads, 1); resource_limit->Reserve(ResourceType::PhysicalMemory, main_thread_stack_size); } @@ -338,6 +346,7 @@ static auto FindTLSPageWithAvailableSlots(std::vector<TLSPage>& tls_pages) { } VAddr Process::CreateTLSRegion() { + SchedulerLock lock(system.Kernel()); if (auto tls_page_iter{FindTLSPageWithAvailableSlots(tls_pages)}; tls_page_iter != tls_pages.cend()) { return *tls_page_iter->ReserveSlot(); @@ -368,6 +377,7 @@ VAddr Process::CreateTLSRegion() { } void Process::FreeTLSRegion(VAddr tls_address) { + SchedulerLock lock(system.Kernel()); const VAddr aligned_address = Common::AlignDown(tls_address, Core::Memory::PAGE_SIZE); auto iter = std::find_if(tls_pages.begin(), tls_pages.end(), [aligned_address](const auto& page) { @@ -382,6 +392,7 @@ void Process::FreeTLSRegion(VAddr tls_address) { } void Process::LoadModule(CodeSet code_set, VAddr base_addr) { + std::lock_guard lock{HLE::g_hle_lock}; const auto ReprotectSegment = [&](const CodeSet::Segment& segment, Memory::MemoryPermission permission) { page_table->SetCodeMemoryPermission(segment.addr + base_addr, segment.size, permission); diff --git a/src/core/hle/kernel/readable_event.cpp b/src/core/hle/kernel/readable_event.cpp index 00860fcbd..6e286419e 100644 --- a/src/core/hle/kernel/readable_event.cpp +++ b/src/core/hle/kernel/readable_event.cpp @@ -6,8 +6,10 @@ #include "common/assert.h" #include "common/logging/log.h" #include "core/hle/kernel/errors.h" +#include "core/hle/kernel/kernel.h" #include "core/hle/kernel/object.h" #include "core/hle/kernel/readable_event.h" +#include "core/hle/kernel/scheduler.h" #include "core/hle/kernel/thread.h" namespace Kernel { @@ -37,8 +39,9 @@ void ReadableEvent::Clear() { } ResultCode ReadableEvent::Reset() { + SchedulerLock lock(kernel); if (!is_signaled) { - LOG_ERROR(Kernel, "Handle is not signaled! object_id={}, object_type={}, object_name={}", + LOG_TRACE(Kernel, "Handle is not signaled! object_id={}, object_type={}, object_name={}", GetObjectId(), GetTypeName(), GetName()); return ERR_INVALID_STATE; } diff --git a/src/core/hle/kernel/resource_limit.cpp b/src/core/hle/kernel/resource_limit.cpp index d9beaa3a4..212e442f4 100644 --- a/src/core/hle/kernel/resource_limit.cpp +++ b/src/core/hle/kernel/resource_limit.cpp @@ -24,13 +24,9 @@ bool ResourceLimit::Reserve(ResourceType resource, s64 amount, u64 timeout) { const std::size_t index{ResourceTypeToIndex(resource)}; s64 new_value = current[index] + amount; - while (new_value > limit[index] && available[index] + amount <= limit[index]) { + if (new_value > limit[index] && available[index] + amount <= limit[index]) { // TODO(bunnei): This is wrong for multicore, we should wait the calling thread for timeout new_value = current[index] + amount; - - if (timeout >= 0) { - break; - } } if (new_value <= limit[index]) { diff --git a/src/core/hle/kernel/scheduler.cpp b/src/core/hle/kernel/scheduler.cpp index 1140c72a3..2b12c0dbf 100644 --- a/src/core/hle/kernel/scheduler.cpp +++ b/src/core/hle/kernel/scheduler.cpp @@ -11,11 +11,15 @@ #include <utility> #include "common/assert.h" +#include "common/bit_util.h" +#include "common/fiber.h" #include "common/logging/log.h" #include "core/arm/arm_interface.h" #include "core/core.h" #include "core/core_timing.h" +#include "core/cpu_manager.h" #include "core/hle/kernel/kernel.h" +#include "core/hle/kernel/physical_core.h" #include "core/hle/kernel/process.h" #include "core/hle/kernel/scheduler.h" #include "core/hle/kernel/time_manager.h" @@ -27,103 +31,151 @@ GlobalScheduler::GlobalScheduler(KernelCore& kernel) : kernel{kernel} {} GlobalScheduler::~GlobalScheduler() = default; void GlobalScheduler::AddThread(std::shared_ptr<Thread> thread) { + global_list_guard.lock(); thread_list.push_back(std::move(thread)); + global_list_guard.unlock(); } void GlobalScheduler::RemoveThread(std::shared_ptr<Thread> thread) { + global_list_guard.lock(); thread_list.erase(std::remove(thread_list.begin(), thread_list.end(), thread), thread_list.end()); + global_list_guard.unlock(); } -void GlobalScheduler::UnloadThread(std::size_t core) { - Scheduler& sched = kernel.Scheduler(core); - sched.UnloadThread(); -} - -void GlobalScheduler::SelectThread(std::size_t core) { +u32 GlobalScheduler::SelectThreads() { + ASSERT(is_locked); const auto update_thread = [](Thread* thread, Scheduler& sched) { - if (thread != sched.selected_thread.get()) { + sched.guard.lock(); + if (thread != sched.selected_thread_set.get()) { if (thread == nullptr) { ++sched.idle_selection_count; } - sched.selected_thread = SharedFrom(thread); + sched.selected_thread_set = SharedFrom(thread); } - sched.is_context_switch_pending = sched.selected_thread != sched.current_thread; + const bool reschedule_pending = + sched.is_context_switch_pending || (sched.selected_thread_set != sched.current_thread); + sched.is_context_switch_pending = reschedule_pending; std::atomic_thread_fence(std::memory_order_seq_cst); + sched.guard.unlock(); + return reschedule_pending; }; - Scheduler& sched = kernel.Scheduler(core); - Thread* current_thread = nullptr; - // Step 1: Get top thread in schedule queue. - current_thread = scheduled_queue[core].empty() ? nullptr : scheduled_queue[core].front(); - if (current_thread) { - update_thread(current_thread, sched); - return; + if (!is_reselection_pending.load()) { + return 0; } - // Step 2: Try selecting a suggested thread. - Thread* winner = nullptr; - std::set<s32> sug_cores; - for (auto thread : suggested_queue[core]) { - s32 this_core = thread->GetProcessorID(); - Thread* thread_on_core = nullptr; - if (this_core >= 0) { - thread_on_core = scheduled_queue[this_core].front(); - } - if (this_core < 0 || thread != thread_on_core) { - winner = thread; - break; + std::array<Thread*, Core::Hardware::NUM_CPU_CORES> top_threads{}; + + u32 idle_cores{}; + + // Step 1: Get top thread in schedule queue. + for (u32 core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) { + Thread* top_thread = + scheduled_queue[core].empty() ? nullptr : scheduled_queue[core].front(); + if (top_thread != nullptr) { + // TODO(Blinkhawk): Implement Thread Pinning + } else { + idle_cores |= (1ul << core); } - sug_cores.insert(this_core); + top_threads[core] = top_thread; } - // if we got a suggested thread, select it, else do a second pass. - if (winner && winner->GetPriority() > 2) { - if (winner->IsRunning()) { - UnloadThread(static_cast<u32>(winner->GetProcessorID())); + + while (idle_cores != 0) { + u32 core_id = Common::CountTrailingZeroes32(idle_cores); + + if (!suggested_queue[core_id].empty()) { + std::array<s32, Core::Hardware::NUM_CPU_CORES> migration_candidates{}; + std::size_t num_candidates = 0; + auto iter = suggested_queue[core_id].begin(); + Thread* suggested = nullptr; + // Step 2: Try selecting a suggested thread. + while (iter != suggested_queue[core_id].end()) { + suggested = *iter; + iter++; + s32 suggested_core_id = suggested->GetProcessorID(); + Thread* top_thread = + suggested_core_id >= 0 ? top_threads[suggested_core_id] : nullptr; + if (top_thread != suggested) { + if (top_thread != nullptr && + top_thread->GetPriority() < THREADPRIO_MAX_CORE_MIGRATION) { + suggested = nullptr; + break; + // There's a too high thread to do core migration, cancel + } + TransferToCore(suggested->GetPriority(), static_cast<s32>(core_id), suggested); + break; + } + suggested = nullptr; + migration_candidates[num_candidates++] = suggested_core_id; + } + // Step 3: Select a suggested thread from another core + if (suggested == nullptr) { + for (std::size_t i = 0; i < num_candidates; i++) { + s32 candidate_core = migration_candidates[i]; + suggested = top_threads[candidate_core]; + auto it = scheduled_queue[candidate_core].begin(); + it++; + Thread* next = it != scheduled_queue[candidate_core].end() ? *it : nullptr; + if (next != nullptr) { + TransferToCore(suggested->GetPriority(), static_cast<s32>(core_id), + suggested); + top_threads[candidate_core] = next; + break; + } else { + suggested = nullptr; + } + } + } + top_threads[core_id] = suggested; } - TransferToCore(winner->GetPriority(), static_cast<s32>(core), winner); - update_thread(winner, sched); - return; + + idle_cores &= ~(1ul << core_id); } - // Step 3: Select a suggested thread from another core - for (auto& src_core : sug_cores) { - auto it = scheduled_queue[src_core].begin(); - it++; - if (it != scheduled_queue[src_core].end()) { - Thread* thread_on_core = scheduled_queue[src_core].front(); - Thread* to_change = *it; - if (thread_on_core->IsRunning() || to_change->IsRunning()) { - UnloadThread(static_cast<u32>(src_core)); - } - TransferToCore(thread_on_core->GetPriority(), static_cast<s32>(core), thread_on_core); - current_thread = thread_on_core; - break; + u32 cores_needing_context_switch{}; + for (u32 core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) { + Scheduler& sched = kernel.Scheduler(core); + ASSERT(top_threads[core] == nullptr || top_threads[core]->GetProcessorID() == core); + if (update_thread(top_threads[core], sched)) { + cores_needing_context_switch |= (1ul << core); } } - update_thread(current_thread, sched); + return cores_needing_context_switch; } bool GlobalScheduler::YieldThread(Thread* yielding_thread) { + ASSERT(is_locked); // Note: caller should use critical section, etc. + if (!yielding_thread->IsRunnable()) { + // Normally this case shouldn't happen except for SetThreadActivity. + is_reselection_pending.store(true, std::memory_order_release); + return false; + } const u32 core_id = static_cast<u32>(yielding_thread->GetProcessorID()); const u32 priority = yielding_thread->GetPriority(); // Yield the thread - const Thread* const winner = scheduled_queue[core_id].front(priority); - ASSERT_MSG(yielding_thread == winner, "Thread yielding without being in front"); - scheduled_queue[core_id].yield(priority); + Reschedule(priority, core_id, yielding_thread); + const Thread* const winner = scheduled_queue[core_id].front(); + if (kernel.GetCurrentHostThreadID() != core_id) { + is_reselection_pending.store(true, std::memory_order_release); + } return AskForReselectionOrMarkRedundant(yielding_thread, winner); } bool GlobalScheduler::YieldThreadAndBalanceLoad(Thread* yielding_thread) { + ASSERT(is_locked); // Note: caller should check if !thread.IsSchedulerOperationRedundant and use critical section, // etc. + if (!yielding_thread->IsRunnable()) { + // Normally this case shouldn't happen except for SetThreadActivity. + is_reselection_pending.store(true, std::memory_order_release); + return false; + } const u32 core_id = static_cast<u32>(yielding_thread->GetProcessorID()); const u32 priority = yielding_thread->GetPriority(); // Yield the thread - ASSERT_MSG(yielding_thread == scheduled_queue[core_id].front(priority), - "Thread yielding without being in front"); - scheduled_queue[core_id].yield(priority); + Reschedule(priority, core_id, yielding_thread); std::array<Thread*, Core::Hardware::NUM_CPU_CORES> current_threads; for (std::size_t i = 0; i < current_threads.size(); i++) { @@ -153,21 +205,28 @@ bool GlobalScheduler::YieldThreadAndBalanceLoad(Thread* yielding_thread) { if (winner != nullptr) { if (winner != yielding_thread) { - if (winner->IsRunning()) { - UnloadThread(static_cast<u32>(winner->GetProcessorID())); - } TransferToCore(winner->GetPriority(), s32(core_id), winner); } } else { winner = next_thread; } + if (kernel.GetCurrentHostThreadID() != core_id) { + is_reselection_pending.store(true, std::memory_order_release); + } + return AskForReselectionOrMarkRedundant(yielding_thread, winner); } bool GlobalScheduler::YieldThreadAndWaitForLoadBalancing(Thread* yielding_thread) { + ASSERT(is_locked); // Note: caller should check if !thread.IsSchedulerOperationRedundant and use critical section, // etc. + if (!yielding_thread->IsRunnable()) { + // Normally this case shouldn't happen except for SetThreadActivity. + is_reselection_pending.store(true, std::memory_order_release); + return false; + } Thread* winner = nullptr; const u32 core_id = static_cast<u32>(yielding_thread->GetProcessorID()); @@ -195,25 +254,31 @@ bool GlobalScheduler::YieldThreadAndWaitForLoadBalancing(Thread* yielding_thread } if (winner != nullptr) { if (winner != yielding_thread) { - if (winner->IsRunning()) { - UnloadThread(static_cast<u32>(winner->GetProcessorID())); - } TransferToCore(winner->GetPriority(), static_cast<s32>(core_id), winner); } } else { winner = yielding_thread; } + } else { + winner = scheduled_queue[core_id].front(); + } + + if (kernel.GetCurrentHostThreadID() != core_id) { + is_reselection_pending.store(true, std::memory_order_release); } return AskForReselectionOrMarkRedundant(yielding_thread, winner); } void GlobalScheduler::PreemptThreads() { + ASSERT(is_locked); for (std::size_t core_id = 0; core_id < Core::Hardware::NUM_CPU_CORES; core_id++) { const u32 priority = preemption_priorities[core_id]; if (scheduled_queue[core_id].size(priority) > 0) { - scheduled_queue[core_id].front(priority)->IncrementYieldCount(); + if (scheduled_queue[core_id].size(priority) > 1) { + scheduled_queue[core_id].front(priority)->IncrementYieldCount(); + } scheduled_queue[core_id].yield(priority); if (scheduled_queue[core_id].size(priority) > 1) { scheduled_queue[core_id].front(priority)->IncrementYieldCount(); @@ -247,9 +312,6 @@ void GlobalScheduler::PreemptThreads() { } if (winner != nullptr) { - if (winner->IsRunning()) { - UnloadThread(static_cast<u32>(winner->GetProcessorID())); - } TransferToCore(winner->GetPriority(), s32(core_id), winner); current_thread = winner->GetPriority() <= current_thread->GetPriority() ? winner : current_thread; @@ -280,9 +342,6 @@ void GlobalScheduler::PreemptThreads() { } if (winner != nullptr) { - if (winner->IsRunning()) { - UnloadThread(static_cast<u32>(winner->GetProcessorID())); - } TransferToCore(winner->GetPriority(), s32(core_id), winner); current_thread = winner; } @@ -292,34 +351,65 @@ void GlobalScheduler::PreemptThreads() { } } +void GlobalScheduler::EnableInterruptAndSchedule(u32 cores_pending_reschedule, + Core::EmuThreadHandle global_thread) { + u32 current_core = global_thread.host_handle; + bool must_context_switch = global_thread.guest_handle != InvalidHandle && + (current_core < Core::Hardware::NUM_CPU_CORES); + while (cores_pending_reschedule != 0) { + u32 core = Common::CountTrailingZeroes32(cores_pending_reschedule); + ASSERT(core < Core::Hardware::NUM_CPU_CORES); + if (!must_context_switch || core != current_core) { + auto& phys_core = kernel.PhysicalCore(core); + phys_core.Interrupt(); + } else { + must_context_switch = true; + } + cores_pending_reschedule &= ~(1ul << core); + } + if (must_context_switch) { + auto& core_scheduler = kernel.CurrentScheduler(); + kernel.ExitSVCProfile(); + core_scheduler.TryDoContextSwitch(); + kernel.EnterSVCProfile(); + } +} + void GlobalScheduler::Suggest(u32 priority, std::size_t core, Thread* thread) { + ASSERT(is_locked); suggested_queue[core].add(thread, priority); } void GlobalScheduler::Unsuggest(u32 priority, std::size_t core, Thread* thread) { + ASSERT(is_locked); suggested_queue[core].remove(thread, priority); } void GlobalScheduler::Schedule(u32 priority, std::size_t core, Thread* thread) { + ASSERT(is_locked); ASSERT_MSG(thread->GetProcessorID() == s32(core), "Thread must be assigned to this core."); scheduled_queue[core].add(thread, priority); } void GlobalScheduler::SchedulePrepend(u32 priority, std::size_t core, Thread* thread) { + ASSERT(is_locked); ASSERT_MSG(thread->GetProcessorID() == s32(core), "Thread must be assigned to this core."); scheduled_queue[core].add(thread, priority, false); } void GlobalScheduler::Reschedule(u32 priority, std::size_t core, Thread* thread) { + ASSERT(is_locked); scheduled_queue[core].remove(thread, priority); scheduled_queue[core].add(thread, priority); } void GlobalScheduler::Unschedule(u32 priority, std::size_t core, Thread* thread) { + ASSERT(is_locked); scheduled_queue[core].remove(thread, priority); } void GlobalScheduler::TransferToCore(u32 priority, s32 destination_core, Thread* thread) { + ASSERT(is_locked); const bool schedulable = thread->GetPriority() < THREADPRIO_COUNT; const s32 source_core = thread->GetProcessorID(); if (source_core == destination_core || !schedulable) { @@ -349,6 +439,108 @@ bool GlobalScheduler::AskForReselectionOrMarkRedundant(Thread* current_thread, } } +void GlobalScheduler::AdjustSchedulingOnStatus(Thread* thread, u32 old_flags) { + if (old_flags == thread->scheduling_state) { + return; + } + ASSERT(is_locked); + + if (old_flags == static_cast<u32>(ThreadSchedStatus::Runnable)) { + // In this case the thread was running, now it's pausing/exitting + if (thread->processor_id >= 0) { + Unschedule(thread->current_priority, static_cast<u32>(thread->processor_id), thread); + } + + for (u32 core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) { + if (core != static_cast<u32>(thread->processor_id) && + ((thread->affinity_mask >> core) & 1) != 0) { + Unsuggest(thread->current_priority, core, thread); + } + } + } else if (thread->scheduling_state == static_cast<u32>(ThreadSchedStatus::Runnable)) { + // The thread is now set to running from being stopped + if (thread->processor_id >= 0) { + Schedule(thread->current_priority, static_cast<u32>(thread->processor_id), thread); + } + + for (u32 core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) { + if (core != static_cast<u32>(thread->processor_id) && + ((thread->affinity_mask >> core) & 1) != 0) { + Suggest(thread->current_priority, core, thread); + } + } + } + + SetReselectionPending(); +} + +void GlobalScheduler::AdjustSchedulingOnPriority(Thread* thread, u32 old_priority) { + if (thread->scheduling_state != static_cast<u32>(ThreadSchedStatus::Runnable)) { + return; + } + ASSERT(is_locked); + if (thread->processor_id >= 0) { + Unschedule(old_priority, static_cast<u32>(thread->processor_id), thread); + } + + for (u32 core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) { + if (core != static_cast<u32>(thread->processor_id) && + ((thread->affinity_mask >> core) & 1) != 0) { + Unsuggest(old_priority, core, thread); + } + } + + if (thread->processor_id >= 0) { + if (thread == kernel.CurrentScheduler().GetCurrentThread()) { + SchedulePrepend(thread->current_priority, static_cast<u32>(thread->processor_id), + thread); + } else { + Schedule(thread->current_priority, static_cast<u32>(thread->processor_id), thread); + } + } + + for (u32 core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) { + if (core != static_cast<u32>(thread->processor_id) && + ((thread->affinity_mask >> core) & 1) != 0) { + Suggest(thread->current_priority, core, thread); + } + } + thread->IncrementYieldCount(); + SetReselectionPending(); +} + +void GlobalScheduler::AdjustSchedulingOnAffinity(Thread* thread, u64 old_affinity_mask, + s32 old_core) { + if (thread->scheduling_state != static_cast<u32>(ThreadSchedStatus::Runnable) || + thread->current_priority >= THREADPRIO_COUNT) { + return; + } + ASSERT(is_locked); + + for (u32 core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) { + if (((old_affinity_mask >> core) & 1) != 0) { + if (core == static_cast<u32>(old_core)) { + Unschedule(thread->current_priority, core, thread); + } else { + Unsuggest(thread->current_priority, core, thread); + } + } + } + + for (u32 core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) { + if (((thread->affinity_mask >> core) & 1) != 0) { + if (core == static_cast<u32>(thread->processor_id)) { + Schedule(thread->current_priority, core, thread); + } else { + Suggest(thread->current_priority, core, thread); + } + } + } + + thread->IncrementYieldCount(); + SetReselectionPending(); +} + void GlobalScheduler::Shutdown() { for (std::size_t core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) { scheduled_queue[core].clear(); @@ -359,10 +551,12 @@ void GlobalScheduler::Shutdown() { void GlobalScheduler::Lock() { Core::EmuThreadHandle current_thread = kernel.GetCurrentEmuThreadID(); + ASSERT(!current_thread.IsInvalid()); if (current_thread == current_owner) { ++scope_lock; } else { inner_lock.lock(); + is_locked = true; current_owner = current_thread; ASSERT(current_owner != Core::EmuThreadHandle::InvalidHandle()); scope_lock = 1; @@ -374,17 +568,18 @@ void GlobalScheduler::Unlock() { ASSERT(scope_lock > 0); return; } - for (std::size_t i = 0; i < Core::Hardware::NUM_CPU_CORES; i++) { - SelectThread(i); - } + u32 cores_pending_reschedule = SelectThreads(); + Core::EmuThreadHandle leaving_thread = current_owner; current_owner = Core::EmuThreadHandle::InvalidHandle(); scope_lock = 1; + is_locked = false; inner_lock.unlock(); - // TODO(Blinkhawk): Setup the interrupts and change context on current core. + EnableInterruptAndSchedule(cores_pending_reschedule, leaving_thread); } -Scheduler::Scheduler(Core::System& system, std::size_t core_id) - : system{system}, core_id{core_id} {} +Scheduler::Scheduler(Core::System& system, std::size_t core_id) : system(system), core_id(core_id) { + switch_fiber = std::make_shared<Common::Fiber>(std::function<void(void*)>(OnSwitch), this); +} Scheduler::~Scheduler() = default; @@ -393,56 +588,128 @@ bool Scheduler::HaveReadyThreads() const { } Thread* Scheduler::GetCurrentThread() const { - return current_thread.get(); + if (current_thread) { + return current_thread.get(); + } + return idle_thread.get(); } Thread* Scheduler::GetSelectedThread() const { return selected_thread.get(); } -void Scheduler::SelectThreads() { - system.GlobalScheduler().SelectThread(core_id); -} - u64 Scheduler::GetLastContextSwitchTicks() const { return last_context_switch_time; } void Scheduler::TryDoContextSwitch() { + auto& phys_core = system.Kernel().CurrentPhysicalCore(); + if (phys_core.IsInterrupted()) { + phys_core.ClearInterrupt(); + } + guard.lock(); if (is_context_switch_pending) { SwitchContext(); + } else { + guard.unlock(); } } -void Scheduler::UnloadThread() { - Thread* const previous_thread = GetCurrentThread(); - Process* const previous_process = system.Kernel().CurrentProcess(); +void Scheduler::OnThreadStart() { + SwitchContextStep2(); +} - UpdateLastContextSwitchTime(previous_thread, previous_process); +void Scheduler::Unload() { + Thread* thread = current_thread.get(); + if (thread) { + thread->SetContinuousOnSVC(false); + thread->last_running_ticks = system.CoreTiming().GetCPUTicks(); + thread->SetIsRunning(false); + if (!thread->IsHLEThread() && !thread->HasExited()) { + Core::ARM_Interface& cpu_core = thread->ArmInterface(); + cpu_core.SaveContext(thread->GetContext32()); + cpu_core.SaveContext(thread->GetContext64()); + // Save the TPIDR_EL0 system register in case it was modified. + thread->SetTPIDR_EL0(cpu_core.GetTPIDR_EL0()); + cpu_core.ClearExclusiveState(); + } + thread->context_guard.unlock(); + } +} - // Save context for previous thread - if (previous_thread) { - system.ArmInterface(core_id).SaveContext(previous_thread->GetContext32()); - system.ArmInterface(core_id).SaveContext(previous_thread->GetContext64()); - // Save the TPIDR_EL0 system register in case it was modified. - previous_thread->SetTPIDR_EL0(system.ArmInterface(core_id).GetTPIDR_EL0()); +void Scheduler::Reload() { + Thread* thread = current_thread.get(); + if (thread) { + ASSERT_MSG(thread->GetSchedulingStatus() == ThreadSchedStatus::Runnable, + "Thread must be runnable."); - if (previous_thread->GetStatus() == ThreadStatus::Running) { - // This is only the case when a reschedule is triggered without the current thread - // yielding execution (i.e. an event triggered, system core time-sliced, etc) - previous_thread->SetStatus(ThreadStatus::Ready); + // Cancel any outstanding wakeup events for this thread + thread->SetIsRunning(true); + thread->SetWasRunning(false); + thread->last_running_ticks = system.CoreTiming().GetCPUTicks(); + + auto* const thread_owner_process = thread->GetOwnerProcess(); + if (thread_owner_process != nullptr) { + system.Kernel().MakeCurrentProcess(thread_owner_process); + } + if (!thread->IsHLEThread()) { + Core::ARM_Interface& cpu_core = thread->ArmInterface(); + cpu_core.LoadContext(thread->GetContext32()); + cpu_core.LoadContext(thread->GetContext64()); + cpu_core.SetTlsAddress(thread->GetTLSAddress()); + cpu_core.SetTPIDR_EL0(thread->GetTPIDR_EL0()); + cpu_core.ChangeProcessorID(this->core_id); + cpu_core.ClearExclusiveState(); } - previous_thread->SetIsRunning(false); } - current_thread = nullptr; +} + +void Scheduler::SwitchContextStep2() { + Thread* previous_thread = current_thread_prev.get(); + Thread* new_thread = selected_thread.get(); + + // Load context of new thread + Process* const previous_process = + previous_thread != nullptr ? previous_thread->GetOwnerProcess() : nullptr; + + if (new_thread) { + ASSERT_MSG(new_thread->GetSchedulingStatus() == ThreadSchedStatus::Runnable, + "Thread must be runnable."); + + // Cancel any outstanding wakeup events for this thread + new_thread->SetIsRunning(true); + new_thread->last_running_ticks = system.CoreTiming().GetCPUTicks(); + new_thread->SetWasRunning(false); + + auto* const thread_owner_process = current_thread->GetOwnerProcess(); + if (thread_owner_process != nullptr) { + system.Kernel().MakeCurrentProcess(thread_owner_process); + } + if (!new_thread->IsHLEThread()) { + Core::ARM_Interface& cpu_core = new_thread->ArmInterface(); + cpu_core.LoadContext(new_thread->GetContext32()); + cpu_core.LoadContext(new_thread->GetContext64()); + cpu_core.SetTlsAddress(new_thread->GetTLSAddress()); + cpu_core.SetTPIDR_EL0(new_thread->GetTPIDR_EL0()); + cpu_core.ChangeProcessorID(this->core_id); + cpu_core.ClearExclusiveState(); + } + } + + TryDoContextSwitch(); } void Scheduler::SwitchContext() { - Thread* const previous_thread = GetCurrentThread(); - Thread* const new_thread = GetSelectedThread(); + current_thread_prev = current_thread; + selected_thread = selected_thread_set; + Thread* previous_thread = current_thread_prev.get(); + Thread* new_thread = selected_thread.get(); + current_thread = selected_thread; is_context_switch_pending = false; + if (new_thread == previous_thread) { + guard.unlock(); return; } @@ -452,51 +719,75 @@ void Scheduler::SwitchContext() { // Save context for previous thread if (previous_thread) { - system.ArmInterface(core_id).SaveContext(previous_thread->GetContext32()); - system.ArmInterface(core_id).SaveContext(previous_thread->GetContext64()); - // Save the TPIDR_EL0 system register in case it was modified. - previous_thread->SetTPIDR_EL0(system.ArmInterface(core_id).GetTPIDR_EL0()); - - if (previous_thread->GetStatus() == ThreadStatus::Running) { - // This is only the case when a reschedule is triggered without the current thread - // yielding execution (i.e. an event triggered, system core time-sliced, etc) - previous_thread->SetStatus(ThreadStatus::Ready); + if (new_thread != nullptr && new_thread->IsSuspendThread()) { + previous_thread->SetWasRunning(true); } + previous_thread->SetContinuousOnSVC(false); + previous_thread->last_running_ticks = system.CoreTiming().GetCPUTicks(); previous_thread->SetIsRunning(false); - } - - // Load context of new thread - if (new_thread) { - ASSERT_MSG(new_thread->GetProcessorID() == s32(this->core_id), - "Thread must be assigned to this core."); - ASSERT_MSG(new_thread->GetStatus() == ThreadStatus::Ready, - "Thread must be ready to become running."); - - // Cancel any outstanding wakeup events for this thread - new_thread->CancelWakeupTimer(); - current_thread = SharedFrom(new_thread); - new_thread->SetStatus(ThreadStatus::Running); - new_thread->SetIsRunning(true); - - auto* const thread_owner_process = current_thread->GetOwnerProcess(); - if (previous_process != thread_owner_process) { - system.Kernel().MakeCurrentProcess(thread_owner_process); + if (!previous_thread->IsHLEThread() && !previous_thread->HasExited()) { + Core::ARM_Interface& cpu_core = previous_thread->ArmInterface(); + cpu_core.SaveContext(previous_thread->GetContext32()); + cpu_core.SaveContext(previous_thread->GetContext64()); + // Save the TPIDR_EL0 system register in case it was modified. + previous_thread->SetTPIDR_EL0(cpu_core.GetTPIDR_EL0()); + cpu_core.ClearExclusiveState(); } + previous_thread->context_guard.unlock(); + } - system.ArmInterface(core_id).LoadContext(new_thread->GetContext32()); - system.ArmInterface(core_id).LoadContext(new_thread->GetContext64()); - system.ArmInterface(core_id).SetTlsAddress(new_thread->GetTLSAddress()); - system.ArmInterface(core_id).SetTPIDR_EL0(new_thread->GetTPIDR_EL0()); + std::shared_ptr<Common::Fiber>* old_context; + if (previous_thread != nullptr) { + old_context = &previous_thread->GetHostContext(); } else { - current_thread = nullptr; - // Note: We do not reset the current process and current page table when idling because - // technically we haven't changed processes, our threads are just paused. + old_context = &idle_thread->GetHostContext(); + } + guard.unlock(); + + Common::Fiber::YieldTo(*old_context, switch_fiber); + /// When a thread wakes up, the scheduler may have changed to other in another core. + auto& next_scheduler = system.Kernel().CurrentScheduler(); + next_scheduler.SwitchContextStep2(); +} + +void Scheduler::OnSwitch(void* this_scheduler) { + Scheduler* sched = static_cast<Scheduler*>(this_scheduler); + sched->SwitchToCurrent(); +} + +void Scheduler::SwitchToCurrent() { + while (true) { + guard.lock(); + selected_thread = selected_thread_set; + current_thread = selected_thread; + is_context_switch_pending = false; + guard.unlock(); + while (!is_context_switch_pending) { + if (current_thread != nullptr && !current_thread->IsHLEThread()) { + current_thread->context_guard.lock(); + if (!current_thread->IsRunnable()) { + current_thread->context_guard.unlock(); + break; + } + if (current_thread->GetProcessorID() != core_id) { + current_thread->context_guard.unlock(); + break; + } + } + std::shared_ptr<Common::Fiber>* next_context; + if (current_thread != nullptr) { + next_context = ¤t_thread->GetHostContext(); + } else { + next_context = &idle_thread->GetHostContext(); + } + Common::Fiber::YieldTo(switch_fiber, *next_context); + } } } void Scheduler::UpdateLastContextSwitchTime(Thread* thread, Process* process) { const u64 prev_switch_ticks = last_context_switch_time; - const u64 most_recent_switch_ticks = system.CoreTiming().GetTicks(); + const u64 most_recent_switch_ticks = system.CoreTiming().GetCPUTicks(); const u64 update_ticks = most_recent_switch_ticks - prev_switch_ticks; if (thread != nullptr) { @@ -510,6 +801,16 @@ void Scheduler::UpdateLastContextSwitchTime(Thread* thread, Process* process) { last_context_switch_time = most_recent_switch_ticks; } +void Scheduler::Initialize() { + std::string name = "Idle Thread Id:" + std::to_string(core_id); + std::function<void(void*)> init_func = system.GetCpuManager().GetIdleThreadStartFunc(); + void* init_func_parameter = system.GetCpuManager().GetStartFuncParamater(); + ThreadType type = static_cast<ThreadType>(THREADTYPE_KERNEL | THREADTYPE_HLE | THREADTYPE_IDLE); + auto thread_res = Thread::Create(system, type, name, 0, 64, 0, static_cast<u32>(core_id), 0, + nullptr, std::move(init_func), init_func_parameter); + idle_thread = std::move(thread_res).Unwrap(); +} + void Scheduler::Shutdown() { current_thread = nullptr; selected_thread = nullptr; @@ -538,4 +839,13 @@ SchedulerLockAndSleep::~SchedulerLockAndSleep() { time_manager.ScheduleTimeEvent(event_handle, time_task, nanoseconds); } +void SchedulerLockAndSleep::Release() { + if (sleep_cancelled) { + return; + } + auto& time_manager = kernel.TimeManager(); + time_manager.ScheduleTimeEvent(event_handle, time_task, nanoseconds); + sleep_cancelled = true; +} + } // namespace Kernel diff --git a/src/core/hle/kernel/scheduler.h b/src/core/hle/kernel/scheduler.h index 07df33f9c..b3b4b5169 100644 --- a/src/core/hle/kernel/scheduler.h +++ b/src/core/hle/kernel/scheduler.h @@ -11,9 +11,14 @@ #include "common/common_types.h" #include "common/multi_level_queue.h" +#include "common/spin_lock.h" #include "core/hardware_properties.h" #include "core/hle/kernel/thread.h" +namespace Common { +class Fiber; +} + namespace Core { class ARM_Interface; class System; @@ -41,41 +46,17 @@ public: return thread_list; } - /** - * Add a thread to the suggested queue of a cpu core. Suggested threads may be - * picked if no thread is scheduled to run on the core. - */ - void Suggest(u32 priority, std::size_t core, Thread* thread); - - /** - * Remove a thread to the suggested queue of a cpu core. Suggested threads may be - * picked if no thread is scheduled to run on the core. - */ - void Unsuggest(u32 priority, std::size_t core, Thread* thread); - - /** - * Add a thread to the scheduling queue of a cpu core. The thread is added at the - * back the queue in its priority level. - */ - void Schedule(u32 priority, std::size_t core, Thread* thread); - - /** - * Add a thread to the scheduling queue of a cpu core. The thread is added at the - * front the queue in its priority level. - */ - void SchedulePrepend(u32 priority, std::size_t core, Thread* thread); + /// Notify the scheduler a thread's status has changed. + void AdjustSchedulingOnStatus(Thread* thread, u32 old_flags); - /// Reschedule an already scheduled thread based on a new priority - void Reschedule(u32 priority, std::size_t core, Thread* thread); - - /// Unschedules a thread. - void Unschedule(u32 priority, std::size_t core, Thread* thread); + /// Notify the scheduler a thread's priority has changed. + void AdjustSchedulingOnPriority(Thread* thread, u32 old_priority); - /// Selects a core and forces it to unload its current thread's context - void UnloadThread(std::size_t core); + /// Notify the scheduler a thread's core and/or affinity mask has changed. + void AdjustSchedulingOnAffinity(Thread* thread, u64 old_affinity_mask, s32 old_core); /** - * Takes care of selecting the new scheduled thread in three steps: + * Takes care of selecting the new scheduled threads in three steps: * * 1. First a thread is selected from the top of the priority queue. If no thread * is obtained then we move to step two, else we are done. @@ -85,8 +66,10 @@ public: * * 3. Third is no suggested thread is found, we do a second pass and pick a running * thread in another core and swap it with its current thread. + * + * returns the cores needing scheduling. */ - void SelectThread(std::size_t core); + u32 SelectThreads(); bool HaveReadyThreads(std::size_t core_id) const { return !scheduled_queue[core_id].empty(); @@ -149,6 +132,40 @@ private: /// Unlocks the scheduler, reselects threads, interrupts cores for rescheduling /// and reschedules current core if needed. void Unlock(); + + void EnableInterruptAndSchedule(u32 cores_pending_reschedule, + Core::EmuThreadHandle global_thread); + + /** + * Add a thread to the suggested queue of a cpu core. Suggested threads may be + * picked if no thread is scheduled to run on the core. + */ + void Suggest(u32 priority, std::size_t core, Thread* thread); + + /** + * Remove a thread to the suggested queue of a cpu core. Suggested threads may be + * picked if no thread is scheduled to run on the core. + */ + void Unsuggest(u32 priority, std::size_t core, Thread* thread); + + /** + * Add a thread to the scheduling queue of a cpu core. The thread is added at the + * back the queue in its priority level. + */ + void Schedule(u32 priority, std::size_t core, Thread* thread); + + /** + * Add a thread to the scheduling queue of a cpu core. The thread is added at the + * front the queue in its priority level. + */ + void SchedulePrepend(u32 priority, std::size_t core, Thread* thread); + + /// Reschedule an already scheduled thread based on a new priority + void Reschedule(u32 priority, std::size_t core, Thread* thread); + + /// Unschedules a thread. + void Unschedule(u32 priority, std::size_t core, Thread* thread); + /** * Transfers a thread into an specific core. If the destination_core is -1 * it will be unscheduled from its source code and added into its suggested @@ -170,10 +187,13 @@ private: std::array<u32, Core::Hardware::NUM_CPU_CORES> preemption_priorities = {59, 59, 59, 62}; /// Scheduler lock mechanisms. - std::mutex inner_lock{}; // TODO(Blinkhawk): Replace for a SpinLock + bool is_locked{}; + Common::SpinLock inner_lock{}; std::atomic<s64> scope_lock{}; Core::EmuThreadHandle current_owner{Core::EmuThreadHandle::InvalidHandle()}; + Common::SpinLock global_list_guard{}; + /// Lists all thread ids that aren't deleted/etc. std::vector<std::shared_ptr<Thread>> thread_list; KernelCore& kernel; @@ -190,11 +210,11 @@ public: /// Reschedules to the next available thread (call after current thread is suspended) void TryDoContextSwitch(); - /// Unloads currently running thread - void UnloadThread(); - - /// Select the threads in top of the scheduling multilist. - void SelectThreads(); + /// The next two are for SingleCore Only. + /// Unload current thread before preempting core. + void Unload(); + /// Reload current thread after core preemption. + void Reload(); /// Gets the current running thread Thread* GetCurrentThread() const; @@ -209,15 +229,30 @@ public: return is_context_switch_pending; } + void Initialize(); + /// Shutdowns the scheduler. void Shutdown(); + void OnThreadStart(); + + std::shared_ptr<Common::Fiber>& ControlContext() { + return switch_fiber; + } + + const std::shared_ptr<Common::Fiber>& ControlContext() const { + return switch_fiber; + } + private: friend class GlobalScheduler; /// Switches the CPU's active thread context to that of the specified thread void SwitchContext(); + /// When a thread wakes up, it must run this through it's new scheduler + void SwitchContextStep2(); + /** * Called on every context switch to update the internal timestamp * This also updates the running time ticks for the given thread and @@ -231,14 +266,24 @@ private: */ void UpdateLastContextSwitchTime(Thread* thread, Process* process); + static void OnSwitch(void* this_scheduler); + void SwitchToCurrent(); + std::shared_ptr<Thread> current_thread = nullptr; std::shared_ptr<Thread> selected_thread = nullptr; + std::shared_ptr<Thread> current_thread_prev = nullptr; + std::shared_ptr<Thread> selected_thread_set = nullptr; + std::shared_ptr<Thread> idle_thread = nullptr; + + std::shared_ptr<Common::Fiber> switch_fiber = nullptr; Core::System& system; u64 last_context_switch_time = 0; u64 idle_selection_count = 0; const std::size_t core_id; + Common::SpinLock guard{}; + bool is_context_switch_pending = false; }; @@ -261,6 +306,8 @@ public: sleep_cancelled = true; } + void Release(); + private: Handle& event_handle; Thread* time_task; diff --git a/src/core/hle/kernel/server_session.cpp b/src/core/hle/kernel/server_session.cpp index 0f102ca44..7b23a6889 100644 --- a/src/core/hle/kernel/server_session.cpp +++ b/src/core/hle/kernel/server_session.cpp @@ -17,6 +17,7 @@ #include "core/hle/kernel/hle_ipc.h" #include "core/hle/kernel/kernel.h" #include "core/hle/kernel/process.h" +#include "core/hle/kernel/scheduler.h" #include "core/hle/kernel/server_session.h" #include "core/hle/kernel/session.h" #include "core/hle/kernel/thread.h" @@ -137,8 +138,8 @@ ResultCode ServerSession::HandleDomainSyncRequest(Kernel::HLERequestContext& con ResultCode ServerSession::QueueSyncRequest(std::shared_ptr<Thread> thread, Core::Memory::Memory& memory) { u32* cmd_buf{reinterpret_cast<u32*>(memory.GetPointer(thread->GetTLSAddress()))}; - std::shared_ptr<Kernel::HLERequestContext> context{ - std::make_shared<Kernel::HLERequestContext>(SharedFrom(this), std::move(thread))}; + auto context = + std::make_shared<HLERequestContext>(kernel, memory, SharedFrom(this), std::move(thread)); context->PopulateFromIncomingCommandBuffer(kernel.CurrentProcess()->GetHandleTable(), cmd_buf); request_queue.Push(std::move(context)); @@ -168,9 +169,12 @@ ResultCode ServerSession::CompleteSyncRequest() { } // Some service requests require the thread to block - if (!context.IsThreadWaiting()) { - context.GetThread().ResumeFromWait(); - context.GetThread().SetWaitSynchronizationResult(result); + { + SchedulerLock lock(kernel); + if (!context.IsThreadWaiting()) { + context.GetThread().ResumeFromWait(); + context.GetThread().SetSynchronizationResults(nullptr, result); + } } request_queue.Pop(); @@ -180,8 +184,10 @@ ResultCode ServerSession::CompleteSyncRequest() { ResultCode ServerSession::HandleSyncRequest(std::shared_ptr<Thread> thread, Core::Memory::Memory& memory) { - Core::System::GetInstance().CoreTiming().ScheduleEvent(20000, request_event, {}); - return QueueSyncRequest(std::move(thread), memory); + ResultCode result = QueueSyncRequest(std::move(thread), memory); + const u64 delay = kernel.IsMulticore() ? 0U : 20000U; + Core::System::GetInstance().CoreTiming().ScheduleEvent(delay, request_event, {}); + return result; } } // namespace Kernel diff --git a/src/core/hle/kernel/svc.cpp b/src/core/hle/kernel/svc.cpp index 4ae4529f5..5db19dcf3 100644 --- a/src/core/hle/kernel/svc.cpp +++ b/src/core/hle/kernel/svc.cpp @@ -10,14 +10,15 @@ #include "common/alignment.h" #include "common/assert.h" +#include "common/fiber.h" #include "common/logging/log.h" #include "common/microprofile.h" #include "common/string_util.h" #include "core/arm/exclusive_monitor.h" #include "core/core.h" -#include "core/core_manager.h" #include "core/core_timing.h" #include "core/core_timing_util.h" +#include "core/cpu_manager.h" #include "core/hle/kernel/address_arbiter.h" #include "core/hle/kernel/client_port.h" #include "core/hle/kernel/client_session.h" @@ -27,6 +28,7 @@ #include "core/hle/kernel/memory/memory_block.h" #include "core/hle/kernel/memory/page_table.h" #include "core/hle/kernel/mutex.h" +#include "core/hle/kernel/physical_core.h" #include "core/hle/kernel/process.h" #include "core/hle/kernel/readable_event.h" #include "core/hle/kernel/resource_limit.h" @@ -37,6 +39,7 @@ #include "core/hle/kernel/svc_wrap.h" #include "core/hle/kernel/synchronization.h" #include "core/hle/kernel/thread.h" +#include "core/hle/kernel/time_manager.h" #include "core/hle/kernel/transfer_memory.h" #include "core/hle/kernel/writable_event.h" #include "core/hle/lock.h" @@ -133,6 +136,7 @@ enum class ResourceLimitValueType { ResultVal<s64> RetrieveResourceLimitValue(Core::System& system, Handle resource_limit, u32 resource_type, ResourceLimitValueType value_type) { + std::lock_guard lock{HLE::g_hle_lock}; const auto type = static_cast<ResourceType>(resource_type); if (!IsValidResourceType(type)) { LOG_ERROR(Kernel_SVC, "Invalid resource limit type: '{}'", resource_type); @@ -160,6 +164,7 @@ ResultVal<s64> RetrieveResourceLimitValue(Core::System& system, Handle resource_ /// Set the process heap to a given Size. It can both extend and shrink the heap. static ResultCode SetHeapSize(Core::System& system, VAddr* heap_addr, u64 heap_size) { + std::lock_guard lock{HLE::g_hle_lock}; LOG_TRACE(Kernel_SVC, "called, heap_size=0x{:X}", heap_size); // Size must be a multiple of 0x200000 (2MB) and be equal to or less than 8GB. @@ -190,6 +195,7 @@ static ResultCode SetHeapSize32(Core::System& system, u32* heap_addr, u32 heap_s static ResultCode SetMemoryAttribute(Core::System& system, VAddr address, u64 size, u32 mask, u32 attribute) { + std::lock_guard lock{HLE::g_hle_lock}; LOG_DEBUG(Kernel_SVC, "called, address=0x{:016X}, size=0x{:X}, mask=0x{:08X}, attribute=0x{:08X}", address, size, mask, attribute); @@ -226,8 +232,15 @@ static ResultCode SetMemoryAttribute(Core::System& system, VAddr address, u64 si static_cast<Memory::MemoryAttribute>(attribute)); } +static ResultCode SetMemoryAttribute32(Core::System& system, u32 address, u32 size, u32 mask, + u32 attribute) { + return SetMemoryAttribute(system, static_cast<VAddr>(address), static_cast<std::size_t>(size), + mask, attribute); +} + /// Maps a memory range into a different range. static ResultCode MapMemory(Core::System& system, VAddr dst_addr, VAddr src_addr, u64 size) { + std::lock_guard lock{HLE::g_hle_lock}; LOG_TRACE(Kernel_SVC, "called, dst_addr=0x{:X}, src_addr=0x{:X}, size=0x{:X}", dst_addr, src_addr, size); @@ -241,8 +254,14 @@ static ResultCode MapMemory(Core::System& system, VAddr dst_addr, VAddr src_addr return page_table.Map(dst_addr, src_addr, size); } +static ResultCode MapMemory32(Core::System& system, u32 dst_addr, u32 src_addr, u32 size) { + return MapMemory(system, static_cast<VAddr>(dst_addr), static_cast<VAddr>(src_addr), + static_cast<std::size_t>(size)); +} + /// Unmaps a region that was previously mapped with svcMapMemory static ResultCode UnmapMemory(Core::System& system, VAddr dst_addr, VAddr src_addr, u64 size) { + std::lock_guard lock{HLE::g_hle_lock}; LOG_TRACE(Kernel_SVC, "called, dst_addr=0x{:X}, src_addr=0x{:X}, size=0x{:X}", dst_addr, src_addr, size); @@ -256,9 +275,15 @@ static ResultCode UnmapMemory(Core::System& system, VAddr dst_addr, VAddr src_ad return page_table.Unmap(dst_addr, src_addr, size); } +static ResultCode UnmapMemory32(Core::System& system, u32 dst_addr, u32 src_addr, u32 size) { + return UnmapMemory(system, static_cast<VAddr>(dst_addr), static_cast<VAddr>(src_addr), + static_cast<std::size_t>(size)); +} + /// Connect to an OS service given the port name, returns the handle to the port to out static ResultCode ConnectToNamedPort(Core::System& system, Handle* out_handle, VAddr port_name_address) { + std::lock_guard lock{HLE::g_hle_lock}; auto& memory = system.Memory(); if (!memory.IsValidVirtualAddress(port_name_address)) { @@ -317,11 +342,30 @@ static ResultCode SendSyncRequest(Core::System& system, Handle handle) { LOG_TRACE(Kernel_SVC, "called handle=0x{:08X}({})", handle, session->GetName()); auto thread = system.CurrentScheduler().GetCurrentThread(); - thread->InvalidateWakeupCallback(); - thread->SetStatus(ThreadStatus::WaitIPC); - system.PrepareReschedule(thread->GetProcessorID()); + { + SchedulerLock lock(system.Kernel()); + thread->InvalidateHLECallback(); + thread->SetStatus(ThreadStatus::WaitIPC); + session->SendSyncRequest(SharedFrom(thread), system.Memory()); + } + + if (thread->HasHLECallback()) { + Handle event_handle = thread->GetHLETimeEvent(); + if (event_handle != InvalidHandle) { + auto& time_manager = system.Kernel().TimeManager(); + time_manager.UnscheduleTimeEvent(event_handle); + } + + { + SchedulerLock lock(system.Kernel()); + auto* sync_object = thread->GetHLESyncObject(); + sync_object->RemoveWaitingThread(SharedFrom(thread)); + } + + thread->InvokeHLECallback(SharedFrom(thread)); + } - return session->SendSyncRequest(SharedFrom(thread), system.Memory()); + return thread->GetSignalingResult(); } static ResultCode SendSyncRequest32(Core::System& system, Handle handle) { @@ -383,6 +427,15 @@ static ResultCode GetProcessId(Core::System& system, u64* process_id, Handle han return ERR_INVALID_HANDLE; } +static ResultCode GetProcessId32(Core::System& system, u32* process_id_low, u32* process_id_high, + Handle handle) { + u64 process_id{}; + const auto result = GetProcessId(system, &process_id, handle); + *process_id_low = static_cast<u32>(process_id); + *process_id_high = static_cast<u32>(process_id >> 32); + return result; +} + /// Wait for the given handles to synchronize, timeout after the specified nanoseconds static ResultCode WaitSynchronization(Core::System& system, Handle* index, VAddr handles_address, u64 handle_count, s64 nano_seconds) { @@ -447,10 +500,13 @@ static ResultCode CancelSynchronization(Core::System& system, Handle thread_hand } thread->CancelWait(); - system.PrepareReschedule(thread->GetProcessorID()); return RESULT_SUCCESS; } +static ResultCode CancelSynchronization32(Core::System& system, Handle thread_handle) { + return CancelSynchronization(system, thread_handle); +} + /// Attempts to locks a mutex, creating it if it does not already exist static ResultCode ArbitrateLock(Core::System& system, Handle holding_thread_handle, VAddr mutex_addr, Handle requesting_thread_handle) { @@ -475,6 +531,12 @@ static ResultCode ArbitrateLock(Core::System& system, Handle holding_thread_hand requesting_thread_handle); } +static ResultCode ArbitrateLock32(Core::System& system, Handle holding_thread_handle, + u32 mutex_addr, Handle requesting_thread_handle) { + return ArbitrateLock(system, holding_thread_handle, static_cast<VAddr>(mutex_addr), + requesting_thread_handle); +} + /// Unlock a mutex static ResultCode ArbitrateUnlock(Core::System& system, VAddr mutex_addr) { LOG_TRACE(Kernel_SVC, "called mutex_addr=0x{:X}", mutex_addr); @@ -494,6 +556,10 @@ static ResultCode ArbitrateUnlock(Core::System& system, VAddr mutex_addr) { return current_process->GetMutex().Release(mutex_addr); } +static ResultCode ArbitrateUnlock32(Core::System& system, u32 mutex_addr) { + return ArbitrateUnlock(system, static_cast<VAddr>(mutex_addr)); +} + enum class BreakType : u32 { Panic = 0, AssertionFailed = 1, @@ -594,6 +660,7 @@ static void Break(Core::System& system, u32 reason, u64 info1, u64 info2) { info2, has_dumped_buffer ? std::make_optional(debug_buffer) : std::nullopt); if (!break_reason.signal_debugger) { + SchedulerLock lock(system.Kernel()); LOG_CRITICAL( Debug_Emulated, "Emulated program broke execution! reason=0x{:016X}, info1=0x{:016X}, info2=0x{:016X}", @@ -605,14 +672,16 @@ static void Break(Core::System& system, u32 reason, u64 info1, u64 info2) { const auto thread_processor_id = current_thread->GetProcessorID(); system.ArmInterface(static_cast<std::size_t>(thread_processor_id)).LogBacktrace(); - system.Kernel().CurrentProcess()->PrepareForTermination(); - // Kill the current thread + system.Kernel().ExceptionalExit(); current_thread->Stop(); - system.PrepareReschedule(); } } +static void Break32(Core::System& system, u32 reason, u32 info1, u32 info2) { + Break(system, reason, static_cast<u64>(info1), static_cast<u64>(info2)); +} + /// Used to output a message on a debug hardware unit - does nothing on a retail unit static void OutputDebugString([[maybe_unused]] Core::System& system, VAddr address, u64 len) { if (len == 0) { @@ -627,6 +696,7 @@ static void OutputDebugString([[maybe_unused]] Core::System& system, VAddr addre /// Gets system/memory information for the current process static ResultCode GetInfo(Core::System& system, u64* result, u64 info_id, u64 handle, u64 info_sub_id) { + std::lock_guard lock{HLE::g_hle_lock}; LOG_TRACE(Kernel_SVC, "called info_id=0x{:X}, info_sub_id=0x{:X}, handle=0x{:08X}", info_id, info_sub_id, handle); @@ -863,9 +933,9 @@ static ResultCode GetInfo(Core::System& system, u64* result, u64 info_id, u64 ha if (same_thread && info_sub_id == 0xFFFFFFFFFFFFFFFF) { const u64 thread_ticks = current_thread->GetTotalCPUTimeTicks(); - out_ticks = thread_ticks + (core_timing.GetTicks() - prev_ctx_ticks); + out_ticks = thread_ticks + (core_timing.GetCPUTicks() - prev_ctx_ticks); } else if (same_thread && info_sub_id == system.CurrentCoreIndex()) { - out_ticks = core_timing.GetTicks() - prev_ctx_ticks; + out_ticks = core_timing.GetCPUTicks() - prev_ctx_ticks; } *result = out_ticks; @@ -892,6 +962,7 @@ static ResultCode GetInfo32(Core::System& system, u32* result_low, u32* result_h /// Maps memory at a desired address static ResultCode MapPhysicalMemory(Core::System& system, VAddr addr, u64 size) { + std::lock_guard lock{HLE::g_hle_lock}; LOG_DEBUG(Kernel_SVC, "called, addr=0x{:016X}, size=0x{:X}", addr, size); if (!Common::Is4KBAligned(addr)) { @@ -939,8 +1010,13 @@ static ResultCode MapPhysicalMemory(Core::System& system, VAddr addr, u64 size) return page_table.MapPhysicalMemory(addr, size); } +static ResultCode MapPhysicalMemory32(Core::System& system, u32 addr, u32 size) { + return MapPhysicalMemory(system, static_cast<VAddr>(addr), static_cast<std::size_t>(size)); +} + /// Unmaps memory previously mapped via MapPhysicalMemory static ResultCode UnmapPhysicalMemory(Core::System& system, VAddr addr, u64 size) { + std::lock_guard lock{HLE::g_hle_lock}; LOG_DEBUG(Kernel_SVC, "called, addr=0x{:016X}, size=0x{:X}", addr, size); if (!Common::Is4KBAligned(addr)) { @@ -988,6 +1064,10 @@ static ResultCode UnmapPhysicalMemory(Core::System& system, VAddr addr, u64 size return page_table.UnmapPhysicalMemory(addr, size); } +static ResultCode UnmapPhysicalMemory32(Core::System& system, u32 addr, u32 size) { + return UnmapPhysicalMemory(system, static_cast<VAddr>(addr), static_cast<std::size_t>(size)); +} + /// Sets the thread activity static ResultCode SetThreadActivity(Core::System& system, Handle handle, u32 activity) { LOG_DEBUG(Kernel_SVC, "called, handle=0x{:08X}, activity=0x{:08X}", handle, activity); @@ -1017,10 +1097,11 @@ static ResultCode SetThreadActivity(Core::System& system, Handle handle, u32 act return ERR_BUSY; } - thread->SetActivity(static_cast<ThreadActivity>(activity)); + return thread->SetActivity(static_cast<ThreadActivity>(activity)); +} - system.PrepareReschedule(thread->GetProcessorID()); - return RESULT_SUCCESS; +static ResultCode SetThreadActivity32(Core::System& system, Handle handle, u32 activity) { + return SetThreadActivity(system, handle, activity); } /// Gets the thread context @@ -1064,6 +1145,10 @@ static ResultCode GetThreadContext(Core::System& system, VAddr thread_context, H return RESULT_SUCCESS; } +static ResultCode GetThreadContext32(Core::System& system, u32 thread_context, Handle handle) { + return GetThreadContext(system, static_cast<VAddr>(thread_context), handle); +} + /// Gets the priority for the specified thread static ResultCode GetThreadPriority(Core::System& system, u32* priority, Handle handle) { LOG_TRACE(Kernel_SVC, "called"); @@ -1071,6 +1156,7 @@ static ResultCode GetThreadPriority(Core::System& system, u32* priority, Handle const auto& handle_table = system.Kernel().CurrentProcess()->GetHandleTable(); const std::shared_ptr<Thread> thread = handle_table.Get<Thread>(handle); if (!thread) { + *priority = 0; LOG_ERROR(Kernel_SVC, "Thread handle does not exist, handle=0x{:08X}", handle); return ERR_INVALID_HANDLE; } @@ -1105,18 +1191,26 @@ static ResultCode SetThreadPriority(Core::System& system, Handle handle, u32 pri thread->SetPriority(priority); - system.PrepareReschedule(thread->GetProcessorID()); return RESULT_SUCCESS; } +static ResultCode SetThreadPriority32(Core::System& system, Handle handle, u32 priority) { + return SetThreadPriority(system, handle, priority); +} + /// Get which CPU core is executing the current thread static u32 GetCurrentProcessorNumber(Core::System& system) { LOG_TRACE(Kernel_SVC, "called"); - return system.CurrentScheduler().GetCurrentThread()->GetProcessorID(); + return static_cast<u32>(system.CurrentPhysicalCore().CoreIndex()); +} + +static u32 GetCurrentProcessorNumber32(Core::System& system) { + return GetCurrentProcessorNumber(system); } static ResultCode MapSharedMemory(Core::System& system, Handle shared_memory_handle, VAddr addr, u64 size, u32 permissions) { + std::lock_guard lock{HLE::g_hle_lock}; LOG_TRACE(Kernel_SVC, "called, shared_memory_handle=0x{:X}, addr=0x{:X}, size=0x{:X}, permissions=0x{:08X}", shared_memory_handle, addr, size, permissions); @@ -1187,9 +1281,16 @@ static ResultCode MapSharedMemory(Core::System& system, Handle shared_memory_han return shared_memory->Map(*current_process, addr, size, permission_type); } +static ResultCode MapSharedMemory32(Core::System& system, Handle shared_memory_handle, u32 addr, + u32 size, u32 permissions) { + return MapSharedMemory(system, shared_memory_handle, static_cast<VAddr>(addr), + static_cast<std::size_t>(size), permissions); +} + static ResultCode QueryProcessMemory(Core::System& system, VAddr memory_info_address, VAddr page_info_address, Handle process_handle, VAddr address) { + std::lock_guard lock{HLE::g_hle_lock}; LOG_TRACE(Kernel_SVC, "called process=0x{:08X} address={:X}", process_handle, address); const auto& handle_table = system.Kernel().CurrentProcess()->GetHandleTable(); std::shared_ptr<Process> process = handle_table.Get<Process>(process_handle); @@ -1372,6 +1473,7 @@ static ResultCode UnmapProcessCodeMemory(Core::System& system, Handle process_ha /// Exits the current process static void ExitProcess(Core::System& system) { auto* current_process = system.Kernel().CurrentProcess(); + UNIMPLEMENTED(); LOG_INFO(Kernel_SVC, "Process {} exiting", current_process->GetProcessID()); ASSERT_MSG(current_process->GetStatus() == ProcessStatus::Running, @@ -1381,8 +1483,10 @@ static void ExitProcess(Core::System& system) { // Kill the current thread system.CurrentScheduler().GetCurrentThread()->Stop(); +} - system.PrepareReschedule(); +static void ExitProcess32(Core::System& system) { + ExitProcess(system); } /// Creates a new thread @@ -1428,9 +1532,10 @@ static ResultCode CreateThread(Core::System& system, Handle* out_handle, VAddr e ASSERT(kernel.CurrentProcess()->GetResourceLimit()->Reserve(ResourceType::Threads, 1)); + ThreadType type = THREADTYPE_USER; CASCADE_RESULT(std::shared_ptr<Thread> thread, - Thread::Create(kernel, "", entry_point, priority, arg, processor_id, stack_top, - *current_process)); + Thread::Create(system, type, "", entry_point, priority, arg, processor_id, + stack_top, current_process)); const auto new_thread_handle = current_process->GetHandleTable().Create(thread); if (new_thread_handle.Failed()) { @@ -1444,11 +1549,15 @@ static ResultCode CreateThread(Core::System& system, Handle* out_handle, VAddr e thread->SetName( fmt::format("thread[entry_point={:X}, handle={:X}]", entry_point, *new_thread_handle)); - system.PrepareReschedule(thread->GetProcessorID()); - return RESULT_SUCCESS; } +static ResultCode CreateThread32(Core::System& system, Handle* out_handle, u32 priority, + u32 entry_point, u32 arg, u32 stack_top, s32 processor_id) { + return CreateThread(system, out_handle, static_cast<VAddr>(entry_point), static_cast<u64>(arg), + static_cast<VAddr>(stack_top), priority, processor_id); +} + /// Starts the thread for the provided handle static ResultCode StartThread(Core::System& system, Handle thread_handle) { LOG_DEBUG(Kernel_SVC, "called thread=0x{:08X}", thread_handle); @@ -1463,13 +1572,11 @@ static ResultCode StartThread(Core::System& system, Handle thread_handle) { ASSERT(thread->GetStatus() == ThreadStatus::Dormant); - thread->ResumeFromWait(); - - if (thread->GetStatus() == ThreadStatus::Ready) { - system.PrepareReschedule(thread->GetProcessorID()); - } + return thread->Start(); +} - return RESULT_SUCCESS; +static ResultCode StartThread32(Core::System& system, Handle thread_handle) { + return StartThread(system, thread_handle); } /// Called when a thread exits @@ -1477,9 +1584,12 @@ static void ExitThread(Core::System& system) { LOG_DEBUG(Kernel_SVC, "called, pc=0x{:08X}", system.CurrentArmInterface().GetPC()); auto* const current_thread = system.CurrentScheduler().GetCurrentThread(); - current_thread->Stop(); system.GlobalScheduler().RemoveThread(SharedFrom(current_thread)); - system.PrepareReschedule(); + current_thread->Stop(); +} + +static void ExitThread32(Core::System& system) { + ExitThread(system); } /// Sleep the current thread @@ -1498,15 +1608,21 @@ static void SleepThread(Core::System& system, s64 nanoseconds) { if (nanoseconds <= 0) { switch (static_cast<SleepType>(nanoseconds)) { - case SleepType::YieldWithoutLoadBalancing: - is_redundant = current_thread->YieldSimple(); + case SleepType::YieldWithoutLoadBalancing: { + auto pair = current_thread->YieldSimple(); + is_redundant = pair.second; break; - case SleepType::YieldWithLoadBalancing: - is_redundant = current_thread->YieldAndBalanceLoad(); + } + case SleepType::YieldWithLoadBalancing: { + auto pair = current_thread->YieldAndBalanceLoad(); + is_redundant = pair.second; break; - case SleepType::YieldAndWaitForLoadBalancing: - is_redundant = current_thread->YieldAndWaitForLoadBalancing(); + } + case SleepType::YieldAndWaitForLoadBalancing: { + auto pair = current_thread->YieldAndWaitForLoadBalancing(); + is_redundant = pair.second; break; + } default: UNREACHABLE_MSG("Unimplemented sleep yield type '{:016X}'!", nanoseconds); } @@ -1514,13 +1630,18 @@ static void SleepThread(Core::System& system, s64 nanoseconds) { current_thread->Sleep(nanoseconds); } - if (is_redundant) { - // If it's redundant, the core is pretty much idle. Some games keep idling - // a core while it's doing nothing, we advance timing to avoid costly continuous - // calls. - system.CoreTiming().AddTicks(2000); + if (is_redundant && !system.Kernel().IsMulticore()) { + system.Kernel().ExitSVCProfile(); + system.CoreTiming().AddTicks(1000U); + system.GetCpuManager().PreemptSingleCore(); + system.Kernel().EnterSVCProfile(); } - system.PrepareReschedule(current_thread->GetProcessorID()); +} + +static void SleepThread32(Core::System& system, u32 nanoseconds_low, u32 nanoseconds_high) { + const s64 nanoseconds = static_cast<s64>(static_cast<u64>(nanoseconds_low) | + (static_cast<u64>(nanoseconds_high) << 32)); + SleepThread(system, nanoseconds); } /// Wait process wide key atomic @@ -1547,31 +1668,69 @@ static ResultCode WaitProcessWideKeyAtomic(Core::System& system, VAddr mutex_add } ASSERT(condition_variable_addr == Common::AlignDown(condition_variable_addr, 4)); - + auto& kernel = system.Kernel(); + Handle event_handle; + Thread* current_thread = system.CurrentScheduler().GetCurrentThread(); auto* const current_process = system.Kernel().CurrentProcess(); - const auto& handle_table = current_process->GetHandleTable(); - std::shared_ptr<Thread> thread = handle_table.Get<Thread>(thread_handle); - ASSERT(thread); + { + SchedulerLockAndSleep lock(kernel, event_handle, current_thread, nano_seconds); + const auto& handle_table = current_process->GetHandleTable(); + std::shared_ptr<Thread> thread = handle_table.Get<Thread>(thread_handle); + ASSERT(thread); + + current_thread->SetSynchronizationResults(nullptr, RESULT_TIMEOUT); + + if (thread->IsPendingTermination()) { + lock.CancelSleep(); + return ERR_THREAD_TERMINATING; + } + + const auto release_result = current_process->GetMutex().Release(mutex_addr); + if (release_result.IsError()) { + lock.CancelSleep(); + return release_result; + } + + if (nano_seconds == 0) { + lock.CancelSleep(); + return RESULT_TIMEOUT; + } - const auto release_result = current_process->GetMutex().Release(mutex_addr); - if (release_result.IsError()) { - return release_result; + current_thread->SetCondVarWaitAddress(condition_variable_addr); + current_thread->SetMutexWaitAddress(mutex_addr); + current_thread->SetWaitHandle(thread_handle); + current_thread->SetStatus(ThreadStatus::WaitCondVar); + current_process->InsertConditionVariableThread(SharedFrom(current_thread)); } - Thread* current_thread = system.CurrentScheduler().GetCurrentThread(); - current_thread->SetCondVarWaitAddress(condition_variable_addr); - current_thread->SetMutexWaitAddress(mutex_addr); - current_thread->SetWaitHandle(thread_handle); - current_thread->SetStatus(ThreadStatus::WaitCondVar); - current_thread->InvalidateWakeupCallback(); - current_process->InsertConditionVariableThread(SharedFrom(current_thread)); + if (event_handle != InvalidHandle) { + auto& time_manager = kernel.TimeManager(); + time_manager.UnscheduleTimeEvent(event_handle); + } + + { + SchedulerLock lock(kernel); - current_thread->WakeAfterDelay(nano_seconds); + auto* owner = current_thread->GetLockOwner(); + if (owner != nullptr) { + owner->RemoveMutexWaiter(SharedFrom(current_thread)); + } + current_process->RemoveConditionVariableThread(SharedFrom(current_thread)); + } // Note: Deliberately don't attempt to inherit the lock owner's priority. - system.PrepareReschedule(current_thread->GetProcessorID()); - return RESULT_SUCCESS; + return current_thread->GetSignalingResult(); +} + +static ResultCode WaitProcessWideKeyAtomic32(Core::System& system, u32 mutex_addr, + u32 condition_variable_addr, Handle thread_handle, + u32 nanoseconds_low, u32 nanoseconds_high) { + const s64 nanoseconds = + static_cast<s64>(nanoseconds_low | (static_cast<u64>(nanoseconds_high) << 32)); + return WaitProcessWideKeyAtomic(system, static_cast<VAddr>(mutex_addr), + static_cast<VAddr>(condition_variable_addr), thread_handle, + nanoseconds); } /// Signal process wide key @@ -1582,7 +1741,9 @@ static void SignalProcessWideKey(Core::System& system, VAddr condition_variable_ ASSERT(condition_variable_addr == Common::AlignDown(condition_variable_addr, 4)); // Retrieve a list of all threads that are waiting for this condition variable. - auto* const current_process = system.Kernel().CurrentProcess(); + auto& kernel = system.Kernel(); + SchedulerLock lock(kernel); + auto* const current_process = kernel.CurrentProcess(); std::vector<std::shared_ptr<Thread>> waiting_threads = current_process->GetConditionVariableThreads(condition_variable_addr); @@ -1591,7 +1752,7 @@ static void SignalProcessWideKey(Core::System& system, VAddr condition_variable_ std::size_t last = waiting_threads.size(); if (target > 0) last = std::min(waiting_threads.size(), static_cast<std::size_t>(target)); - + auto& time_manager = kernel.TimeManager(); for (std::size_t index = 0; index < last; ++index) { auto& thread = waiting_threads[index]; @@ -1599,7 +1760,6 @@ static void SignalProcessWideKey(Core::System& system, VAddr condition_variable_ // liberate Cond Var Thread. current_process->RemoveConditionVariableThread(thread); - thread->SetCondVarWaitAddress(0); const std::size_t current_core = system.CurrentCoreIndex(); auto& monitor = system.Monitor(); @@ -1610,10 +1770,8 @@ static void SignalProcessWideKey(Core::System& system, VAddr condition_variable_ u32 update_val = 0; const VAddr mutex_address = thread->GetMutexWaitAddress(); do { - monitor.SetExclusive(current_core, mutex_address); - // If the mutex is not yet acquired, acquire it. - mutex_val = memory.Read32(mutex_address); + mutex_val = monitor.ExclusiveRead32(current_core, mutex_address); if (mutex_val != 0) { update_val = mutex_val | Mutex::MutexHasWaitersFlag; @@ -1621,33 +1779,28 @@ static void SignalProcessWideKey(Core::System& system, VAddr condition_variable_ update_val = thread->GetWaitHandle(); } } while (!monitor.ExclusiveWrite32(current_core, mutex_address, update_val)); + monitor.ClearExclusive(); if (mutex_val == 0) { // We were able to acquire the mutex, resume this thread. - ASSERT(thread->GetStatus() == ThreadStatus::WaitCondVar); - thread->ResumeFromWait(); - auto* const lock_owner = thread->GetLockOwner(); if (lock_owner != nullptr) { lock_owner->RemoveMutexWaiter(thread); } thread->SetLockOwner(nullptr); - thread->SetMutexWaitAddress(0); - thread->SetWaitHandle(0); - thread->SetWaitSynchronizationResult(RESULT_SUCCESS); - system.PrepareReschedule(thread->GetProcessorID()); + thread->SetSynchronizationResults(nullptr, RESULT_SUCCESS); + thread->ResumeFromWait(); } else { // The mutex is already owned by some other thread, make this thread wait on it. const Handle owner_handle = static_cast<Handle>(mutex_val & Mutex::MutexOwnerMask); const auto& handle_table = system.Kernel().CurrentProcess()->GetHandleTable(); auto owner = handle_table.Get<Thread>(owner_handle); ASSERT(owner); - ASSERT(thread->GetStatus() == ThreadStatus::WaitCondVar); - thread->InvalidateWakeupCallback(); - thread->SetStatus(ThreadStatus::WaitMutex); + if (thread->GetStatus() == ThreadStatus::WaitCondVar) { + thread->SetStatus(ThreadStatus::WaitMutex); + } owner->AddMutexWaiter(thread); - system.PrepareReschedule(thread->GetProcessorID()); } } } @@ -1678,12 +1831,15 @@ static ResultCode WaitForAddress(Core::System& system, VAddr address, u32 type, auto& address_arbiter = system.Kernel().CurrentProcess()->GetAddressArbiter(); const ResultCode result = address_arbiter.WaitForAddress(address, arbitration_type, value, timeout); - if (result == RESULT_SUCCESS) { - system.PrepareReschedule(); - } return result; } +static ResultCode WaitForAddress32(Core::System& system, u32 address, u32 type, s32 value, + u32 timeout_low, u32 timeout_high) { + s64 timeout = static_cast<s64>(timeout_low | (static_cast<u64>(timeout_high) << 32)); + return WaitForAddress(system, static_cast<VAddr>(address), type, value, timeout); +} + // Signals to an address (via Address Arbiter) static ResultCode SignalToAddress(Core::System& system, VAddr address, u32 type, s32 value, s32 num_to_wake) { @@ -1707,6 +1863,11 @@ static ResultCode SignalToAddress(Core::System& system, VAddr address, u32 type, return address_arbiter.SignalToAddress(address, signal_type, value, num_to_wake); } +static ResultCode SignalToAddress32(Core::System& system, u32 address, u32 type, s32 value, + s32 num_to_wake) { + return SignalToAddress(system, static_cast<VAddr>(address), type, value, num_to_wake); +} + static void KernelDebug([[maybe_unused]] Core::System& system, [[maybe_unused]] u32 kernel_debug_type, [[maybe_unused]] u64 param1, [[maybe_unused]] u64 param2, [[maybe_unused]] u64 param3) { @@ -1725,14 +1886,21 @@ static u64 GetSystemTick(Core::System& system) { auto& core_timing = system.CoreTiming(); // Returns the value of cntpct_el0 (https://switchbrew.org/wiki/SVC#svcGetSystemTick) - const u64 result{Core::Timing::CpuCyclesToClockCycles(system.CoreTiming().GetTicks())}; + const u64 result{system.CoreTiming().GetClockTicks()}; - // Advance time to defeat dumb games that busy-wait for the frame to end. - core_timing.AddTicks(400); + if (!system.Kernel().IsMulticore()) { + core_timing.AddTicks(400U); + } return result; } +static void GetSystemTick32(Core::System& system, u32* time_low, u32* time_high) { + u64 time = GetSystemTick(system); + *time_low = static_cast<u32>(time); + *time_high = static_cast<u32>(time >> 32); +} + /// Close a handle static ResultCode CloseHandle(Core::System& system, Handle handle) { LOG_TRACE(Kernel_SVC, "Closing handle 0x{:08X}", handle); @@ -1765,9 +1933,14 @@ static ResultCode ResetSignal(Core::System& system, Handle handle) { return ERR_INVALID_HANDLE; } +static ResultCode ResetSignal32(Core::System& system, Handle handle) { + return ResetSignal(system, handle); +} + /// Creates a TransferMemory object static ResultCode CreateTransferMemory(Core::System& system, Handle* handle, VAddr addr, u64 size, u32 permissions) { + std::lock_guard lock{HLE::g_hle_lock}; LOG_DEBUG(Kernel_SVC, "called addr=0x{:X}, size=0x{:X}, perms=0x{:08X}", addr, size, permissions); @@ -1812,6 +1985,12 @@ static ResultCode CreateTransferMemory(Core::System& system, Handle* handle, VAd return RESULT_SUCCESS; } +static ResultCode CreateTransferMemory32(Core::System& system, Handle* handle, u32 addr, u32 size, + u32 permissions) { + return CreateTransferMemory(system, handle, static_cast<VAddr>(addr), + static_cast<std::size_t>(size), permissions); +} + static ResultCode GetThreadCoreMask(Core::System& system, Handle thread_handle, u32* core, u64* mask) { LOG_TRACE(Kernel_SVC, "called, handle=0x{:08X}", thread_handle); @@ -1821,6 +2000,8 @@ static ResultCode GetThreadCoreMask(Core::System& system, Handle thread_handle, if (!thread) { LOG_ERROR(Kernel_SVC, "Thread handle does not exist, thread_handle=0x{:08X}", thread_handle); + *core = 0; + *mask = 0; return ERR_INVALID_HANDLE; } @@ -1830,6 +2011,15 @@ static ResultCode GetThreadCoreMask(Core::System& system, Handle thread_handle, return RESULT_SUCCESS; } +static ResultCode GetThreadCoreMask32(Core::System& system, Handle thread_handle, u32* core, + u32* mask_low, u32* mask_high) { + u64 mask{}; + const auto result = GetThreadCoreMask(system, thread_handle, core, &mask); + *mask_high = static_cast<u32>(mask >> 32); + *mask_low = static_cast<u32>(mask); + return result; +} + static ResultCode SetThreadCoreMask(Core::System& system, Handle thread_handle, u32 core, u64 affinity_mask) { LOG_DEBUG(Kernel_SVC, "called, handle=0x{:08X}, core=0x{:X}, affinity_mask=0x{:016X}", @@ -1861,7 +2051,7 @@ static ResultCode SetThreadCoreMask(Core::System& system, Handle thread_handle, return ERR_INVALID_COMBINATION; } - if (core < Core::NUM_CPU_CORES) { + if (core < Core::Hardware::NUM_CPU_CORES) { if ((affinity_mask & (1ULL << core)) == 0) { LOG_ERROR(Kernel_SVC, "Core is not enabled for the current mask, core={}, mask={:016X}", core, @@ -1883,11 +2073,14 @@ static ResultCode SetThreadCoreMask(Core::System& system, Handle thread_handle, return ERR_INVALID_HANDLE; } - system.PrepareReschedule(thread->GetProcessorID()); - thread->ChangeCore(core, affinity_mask); - system.PrepareReschedule(thread->GetProcessorID()); + return thread->SetCoreAndAffinityMask(core, affinity_mask); +} - return RESULT_SUCCESS; +static ResultCode SetThreadCoreMask32(Core::System& system, Handle thread_handle, u32 core, + u32 affinity_mask_low, u32 affinity_mask_high) { + const u64 affinity_mask = + static_cast<u64>(affinity_mask_low) | (static_cast<u64>(affinity_mask_high) << 32); + return SetThreadCoreMask(system, thread_handle, core, affinity_mask); } static ResultCode CreateEvent(Core::System& system, Handle* write_handle, Handle* read_handle) { @@ -1918,6 +2111,10 @@ static ResultCode CreateEvent(Core::System& system, Handle* write_handle, Handle return RESULT_SUCCESS; } +static ResultCode CreateEvent32(Core::System& system, Handle* write_handle, Handle* read_handle) { + return CreateEvent(system, write_handle, read_handle); +} + static ResultCode ClearEvent(Core::System& system, Handle handle) { LOG_TRACE(Kernel_SVC, "called, event=0x{:08X}", handle); @@ -1939,6 +2136,10 @@ static ResultCode ClearEvent(Core::System& system, Handle handle) { return ERR_INVALID_HANDLE; } +static ResultCode ClearEvent32(Core::System& system, Handle handle) { + return ClearEvent(system, handle); +} + static ResultCode SignalEvent(Core::System& system, Handle handle) { LOG_DEBUG(Kernel_SVC, "called. Handle=0x{:08X}", handle); @@ -1951,10 +2152,13 @@ static ResultCode SignalEvent(Core::System& system, Handle handle) { } writable_event->Signal(); - system.PrepareReschedule(); return RESULT_SUCCESS; } +static ResultCode SignalEvent32(Core::System& system, Handle handle) { + return SignalEvent(system, handle); +} + static ResultCode GetProcessInfo(Core::System& system, u64* out, Handle process_handle, u32 type) { LOG_DEBUG(Kernel_SVC, "called, handle=0x{:08X}, type=0x{:X}", process_handle, type); @@ -1982,6 +2186,7 @@ static ResultCode GetProcessInfo(Core::System& system, u64* out, Handle process_ } static ResultCode CreateResourceLimit(Core::System& system, Handle* out_handle) { + std::lock_guard lock{HLE::g_hle_lock}; LOG_DEBUG(Kernel_SVC, "called"); auto& kernel = system.Kernel(); @@ -2139,6 +2344,15 @@ static ResultCode GetThreadList(Core::System& system, u32* out_num_threads, VAdd return RESULT_SUCCESS; } +static ResultCode FlushProcessDataCache32(Core::System& system, Handle handle, u32 address, + u32 size) { + // Note(Blinkhawk): For emulation purposes of the data cache this is mostly a nope + // as all emulation is done in the same cache level in host architecture, thus data cache + // does not need flushing. + LOG_DEBUG(Kernel_SVC, "called"); + return RESULT_SUCCESS; +} + namespace { struct FunctionDef { using Func = void(Core::System&); @@ -2153,57 +2367,57 @@ static const FunctionDef SVC_Table_32[] = { {0x00, nullptr, "Unknown"}, {0x01, SvcWrap32<SetHeapSize32>, "SetHeapSize32"}, {0x02, nullptr, "Unknown"}, - {0x03, nullptr, "SetMemoryAttribute32"}, - {0x04, nullptr, "MapMemory32"}, - {0x05, nullptr, "UnmapMemory32"}, + {0x03, SvcWrap32<SetMemoryAttribute32>, "SetMemoryAttribute32"}, + {0x04, SvcWrap32<MapMemory32>, "MapMemory32"}, + {0x05, SvcWrap32<UnmapMemory32>, "UnmapMemory32"}, {0x06, SvcWrap32<QueryMemory32>, "QueryMemory32"}, - {0x07, nullptr, "ExitProcess32"}, - {0x08, nullptr, "CreateThread32"}, - {0x09, nullptr, "StartThread32"}, - {0x0a, nullptr, "ExitThread32"}, - {0x0b, nullptr, "SleepThread32"}, + {0x07, SvcWrap32<ExitProcess32>, "ExitProcess32"}, + {0x08, SvcWrap32<CreateThread32>, "CreateThread32"}, + {0x09, SvcWrap32<StartThread32>, "StartThread32"}, + {0x0a, SvcWrap32<ExitThread32>, "ExitThread32"}, + {0x0b, SvcWrap32<SleepThread32>, "SleepThread32"}, {0x0c, SvcWrap32<GetThreadPriority32>, "GetThreadPriority32"}, - {0x0d, nullptr, "SetThreadPriority32"}, - {0x0e, nullptr, "GetThreadCoreMask32"}, - {0x0f, nullptr, "SetThreadCoreMask32"}, - {0x10, nullptr, "GetCurrentProcessorNumber32"}, - {0x11, nullptr, "SignalEvent32"}, - {0x12, nullptr, "ClearEvent32"}, - {0x13, nullptr, "MapSharedMemory32"}, + {0x0d, SvcWrap32<SetThreadPriority32>, "SetThreadPriority32"}, + {0x0e, SvcWrap32<GetThreadCoreMask32>, "GetThreadCoreMask32"}, + {0x0f, SvcWrap32<SetThreadCoreMask32>, "SetThreadCoreMask32"}, + {0x10, SvcWrap32<GetCurrentProcessorNumber32>, "GetCurrentProcessorNumber32"}, + {0x11, SvcWrap32<SignalEvent32>, "SignalEvent32"}, + {0x12, SvcWrap32<ClearEvent32>, "ClearEvent32"}, + {0x13, SvcWrap32<MapSharedMemory32>, "MapSharedMemory32"}, {0x14, nullptr, "UnmapSharedMemory32"}, - {0x15, nullptr, "CreateTransferMemory32"}, + {0x15, SvcWrap32<CreateTransferMemory32>, "CreateTransferMemory32"}, {0x16, SvcWrap32<CloseHandle32>, "CloseHandle32"}, - {0x17, nullptr, "ResetSignal32"}, + {0x17, SvcWrap32<ResetSignal32>, "ResetSignal32"}, {0x18, SvcWrap32<WaitSynchronization32>, "WaitSynchronization32"}, - {0x19, nullptr, "CancelSynchronization32"}, - {0x1a, nullptr, "ArbitrateLock32"}, - {0x1b, nullptr, "ArbitrateUnlock32"}, - {0x1c, nullptr, "WaitProcessWideKeyAtomic32"}, + {0x19, SvcWrap32<CancelSynchronization32>, "CancelSynchronization32"}, + {0x1a, SvcWrap32<ArbitrateLock32>, "ArbitrateLock32"}, + {0x1b, SvcWrap32<ArbitrateUnlock32>, "ArbitrateUnlock32"}, + {0x1c, SvcWrap32<WaitProcessWideKeyAtomic32>, "WaitProcessWideKeyAtomic32"}, {0x1d, SvcWrap32<SignalProcessWideKey32>, "SignalProcessWideKey32"}, - {0x1e, nullptr, "GetSystemTick32"}, + {0x1e, SvcWrap32<GetSystemTick32>, "GetSystemTick32"}, {0x1f, SvcWrap32<ConnectToNamedPort32>, "ConnectToNamedPort32"}, {0x20, nullptr, "Unknown"}, {0x21, SvcWrap32<SendSyncRequest32>, "SendSyncRequest32"}, {0x22, nullptr, "SendSyncRequestWithUserBuffer32"}, {0x23, nullptr, "Unknown"}, - {0x24, nullptr, "GetProcessId32"}, + {0x24, SvcWrap32<GetProcessId32>, "GetProcessId32"}, {0x25, SvcWrap32<GetThreadId32>, "GetThreadId32"}, - {0x26, nullptr, "Break32"}, + {0x26, SvcWrap32<Break32>, "Break32"}, {0x27, nullptr, "OutputDebugString32"}, {0x28, nullptr, "Unknown"}, {0x29, SvcWrap32<GetInfo32>, "GetInfo32"}, {0x2a, nullptr, "Unknown"}, {0x2b, nullptr, "Unknown"}, - {0x2c, nullptr, "MapPhysicalMemory32"}, - {0x2d, nullptr, "UnmapPhysicalMemory32"}, + {0x2c, SvcWrap32<MapPhysicalMemory32>, "MapPhysicalMemory32"}, + {0x2d, SvcWrap32<UnmapPhysicalMemory32>, "UnmapPhysicalMemory32"}, {0x2e, nullptr, "Unknown"}, {0x2f, nullptr, "Unknown"}, {0x30, nullptr, "Unknown"}, {0x31, nullptr, "Unknown"}, - {0x32, nullptr, "SetThreadActivity32"}, - {0x33, nullptr, "GetThreadContext32"}, - {0x34, nullptr, "WaitForAddress32"}, - {0x35, nullptr, "SignalToAddress32"}, + {0x32, SvcWrap32<SetThreadActivity32>, "SetThreadActivity32"}, + {0x33, SvcWrap32<GetThreadContext32>, "GetThreadContext32"}, + {0x34, SvcWrap32<WaitForAddress32>, "WaitForAddress32"}, + {0x35, SvcWrap32<SignalToAddress32>, "SignalToAddress32"}, {0x36, nullptr, "Unknown"}, {0x37, nullptr, "Unknown"}, {0x38, nullptr, "Unknown"}, @@ -2219,7 +2433,7 @@ static const FunctionDef SVC_Table_32[] = { {0x42, nullptr, "Unknown"}, {0x43, nullptr, "ReplyAndReceive32"}, {0x44, nullptr, "Unknown"}, - {0x45, nullptr, "CreateEvent32"}, + {0x45, SvcWrap32<CreateEvent32>, "CreateEvent32"}, {0x46, nullptr, "Unknown"}, {0x47, nullptr, "Unknown"}, {0x48, nullptr, "Unknown"}, @@ -2245,7 +2459,7 @@ static const FunctionDef SVC_Table_32[] = { {0x5c, nullptr, "Unknown"}, {0x5d, nullptr, "Unknown"}, {0x5e, nullptr, "Unknown"}, - {0x5F, nullptr, "FlushProcessDataCache32"}, + {0x5F, SvcWrap32<FlushProcessDataCache32>, "FlushProcessDataCache32"}, {0x60, nullptr, "Unknown"}, {0x61, nullptr, "Unknown"}, {0x62, nullptr, "Unknown"}, @@ -2423,13 +2637,10 @@ static const FunctionDef* GetSVCInfo64(u32 func_num) { return &SVC_Table_64[func_num]; } -MICROPROFILE_DEFINE(Kernel_SVC, "Kernel", "SVC", MP_RGB(70, 200, 70)); - void Call(Core::System& system, u32 immediate) { - MICROPROFILE_SCOPE(Kernel_SVC); - - // Lock the global kernel mutex when we enter the kernel HLE. - std::lock_guard lock{HLE::g_hle_lock}; + system.ExitDynarmicProfile(); + auto& kernel = system.Kernel(); + kernel.EnterSVCProfile(); const FunctionDef* info = system.CurrentProcess()->Is64BitProcess() ? GetSVCInfo64(immediate) : GetSVCInfo32(immediate); @@ -2442,6 +2653,9 @@ void Call(Core::System& system, u32 immediate) { } else { LOG_CRITICAL(Kernel_SVC, "Unknown SVC function 0x{:X}", immediate); } + + kernel.ExitSVCProfile(); + system.EnterDynarmicProfile(); } } // namespace Kernel::Svc diff --git a/src/core/hle/kernel/svc_wrap.h b/src/core/hle/kernel/svc_wrap.h index 7d735e3fa..0b6dd9df0 100644 --- a/src/core/hle/kernel/svc_wrap.h +++ b/src/core/hle/kernel/svc_wrap.h @@ -350,13 +350,50 @@ void SvcWrap64(Core::System& system) { func(system, static_cast<u32>(Param(system, 0)), Param(system, 1), Param(system, 2)); } -// Used by QueryMemory32 +// Used by QueryMemory32, ArbitrateLock32 template <ResultCode func(Core::System&, u32, u32, u32)> void SvcWrap32(Core::System& system) { FuncReturn32(system, func(system, Param32(system, 0), Param32(system, 1), Param32(system, 2)).raw); } +// Used by Break32 +template <void func(Core::System&, u32, u32, u32)> +void SvcWrap32(Core::System& system) { + func(system, Param32(system, 0), Param32(system, 1), Param32(system, 2)); +} + +// Used by ExitProcess32, ExitThread32 +template <void func(Core::System&)> +void SvcWrap32(Core::System& system) { + func(system); +} + +// Used by GetCurrentProcessorNumber32 +template <u32 func(Core::System&)> +void SvcWrap32(Core::System& system) { + FuncReturn32(system, func(system)); +} + +// Used by SleepThread32 +template <void func(Core::System&, u32, u32)> +void SvcWrap32(Core::System& system) { + func(system, Param32(system, 0), Param32(system, 1)); +} + +// Used by CreateThread32 +template <ResultCode func(Core::System&, Handle*, u32, u32, u32, u32, s32)> +void SvcWrap32(Core::System& system) { + Handle param_1 = 0; + + const u32 retval = func(system, ¶m_1, Param32(system, 0), Param32(system, 1), + Param32(system, 2), Param32(system, 3), Param32(system, 4)) + .raw; + + system.CurrentArmInterface().SetReg(1, param_1); + FuncReturn(system, retval); +} + // Used by GetInfo32 template <ResultCode func(Core::System&, u32*, u32*, u32, u32, u32, u32)> void SvcWrap32(Core::System& system) { @@ -393,18 +430,114 @@ void SvcWrap32(Core::System& system) { FuncReturn(system, retval); } +// Used by GetSystemTick32 +template <void func(Core::System&, u32*, u32*)> +void SvcWrap32(Core::System& system) { + u32 param_1 = 0; + u32 param_2 = 0; + + func(system, ¶m_1, ¶m_2); + system.CurrentArmInterface().SetReg(0, param_1); + system.CurrentArmInterface().SetReg(1, param_2); +} + +// Used by CreateEvent32 +template <ResultCode func(Core::System&, Handle*, Handle*)> +void SvcWrap32(Core::System& system) { + Handle param_1 = 0; + Handle param_2 = 0; + + const u32 retval = func(system, ¶m_1, ¶m_2).raw; + system.CurrentArmInterface().SetReg(1, param_1); + system.CurrentArmInterface().SetReg(2, param_2); + FuncReturn(system, retval); +} + +// Used by GetThreadId32 +template <ResultCode func(Core::System&, Handle, u32*, u32*, u32*)> +void SvcWrap32(Core::System& system) { + u32 param_1 = 0; + u32 param_2 = 0; + u32 param_3 = 0; + + const u32 retval = func(system, Param32(system, 2), ¶m_1, ¶m_2, ¶m_3).raw; + system.CurrentArmInterface().SetReg(1, param_1); + system.CurrentArmInterface().SetReg(2, param_2); + system.CurrentArmInterface().SetReg(3, param_3); + FuncReturn(system, retval); +} + // Used by SignalProcessWideKey32 template <void func(Core::System&, u32, s32)> void SvcWrap32(Core::System& system) { func(system, static_cast<u32>(Param(system, 0)), static_cast<s32>(Param(system, 1))); } -// Used by SendSyncRequest32 +// Used by SetThreadPriority32 +template <ResultCode func(Core::System&, Handle, u32)> +void SvcWrap32(Core::System& system) { + const u32 retval = + func(system, static_cast<Handle>(Param(system, 0)), static_cast<u32>(Param(system, 1))).raw; + FuncReturn(system, retval); +} + +// Used by SetThreadCoreMask32 +template <ResultCode func(Core::System&, Handle, u32, u32, u32)> +void SvcWrap32(Core::System& system) { + const u32 retval = + func(system, static_cast<Handle>(Param(system, 0)), static_cast<u32>(Param(system, 1)), + static_cast<u32>(Param(system, 2)), static_cast<u32>(Param(system, 3))) + .raw; + FuncReturn(system, retval); +} + +// Used by WaitProcessWideKeyAtomic32 +template <ResultCode func(Core::System&, u32, u32, Handle, u32, u32)> +void SvcWrap32(Core::System& system) { + const u32 retval = + func(system, static_cast<u32>(Param(system, 0)), static_cast<u32>(Param(system, 1)), + static_cast<Handle>(Param(system, 2)), static_cast<u32>(Param(system, 3)), + static_cast<u32>(Param(system, 4))) + .raw; + FuncReturn(system, retval); +} + +// Used by WaitForAddress32 +template <ResultCode func(Core::System&, u32, u32, s32, u32, u32)> +void SvcWrap32(Core::System& system) { + const u32 retval = func(system, static_cast<u32>(Param(system, 0)), + static_cast<u32>(Param(system, 1)), static_cast<s32>(Param(system, 2)), + static_cast<u32>(Param(system, 3)), static_cast<u32>(Param(system, 4))) + .raw; + FuncReturn(system, retval); +} + +// Used by SignalToAddress32 +template <ResultCode func(Core::System&, u32, u32, s32, s32)> +void SvcWrap32(Core::System& system) { + const u32 retval = + func(system, static_cast<u32>(Param(system, 0)), static_cast<u32>(Param(system, 1)), + static_cast<s32>(Param(system, 2)), static_cast<s32>(Param(system, 3))) + .raw; + FuncReturn(system, retval); +} + +// Used by SendSyncRequest32, ArbitrateUnlock32 template <ResultCode func(Core::System&, u32)> void SvcWrap32(Core::System& system) { FuncReturn(system, func(system, static_cast<u32>(Param(system, 0))).raw); } +// Used by CreateTransferMemory32 +template <ResultCode func(Core::System&, Handle*, u32, u32, u32)> +void SvcWrap32(Core::System& system) { + Handle handle = 0; + const u32 retval = + func(system, &handle, Param32(system, 1), Param32(system, 2), Param32(system, 3)).raw; + system.CurrentArmInterface().SetReg(1, handle); + FuncReturn(system, retval); +} + // Used by WaitSynchronization32 template <ResultCode func(Core::System&, u32, u32, s32, u32, Handle*)> void SvcWrap32(Core::System& system) { diff --git a/src/core/hle/kernel/synchronization.cpp b/src/core/hle/kernel/synchronization.cpp index dc37fad1a..851b702a5 100644 --- a/src/core/hle/kernel/synchronization.cpp +++ b/src/core/hle/kernel/synchronization.cpp @@ -10,78 +10,107 @@ #include "core/hle/kernel/synchronization.h" #include "core/hle/kernel/synchronization_object.h" #include "core/hle/kernel/thread.h" +#include "core/hle/kernel/time_manager.h" namespace Kernel { -/// Default thread wakeup callback for WaitSynchronization -static bool DefaultThreadWakeupCallback(ThreadWakeupReason reason, std::shared_ptr<Thread> thread, - std::shared_ptr<SynchronizationObject> object, - std::size_t index) { - ASSERT(thread->GetStatus() == ThreadStatus::WaitSynch); - - if (reason == ThreadWakeupReason::Timeout) { - thread->SetWaitSynchronizationResult(RESULT_TIMEOUT); - return true; - } - - ASSERT(reason == ThreadWakeupReason::Signal); - thread->SetWaitSynchronizationResult(RESULT_SUCCESS); - thread->SetWaitSynchronizationOutput(static_cast<u32>(index)); - return true; -} - Synchronization::Synchronization(Core::System& system) : system{system} {} void Synchronization::SignalObject(SynchronizationObject& obj) const { + auto& kernel = system.Kernel(); + SchedulerLock lock(kernel); + auto& time_manager = kernel.TimeManager(); if (obj.IsSignaled()) { - obj.WakeupAllWaitingThreads(); + for (auto thread : obj.GetWaitingThreads()) { + if (thread->GetSchedulingStatus() == ThreadSchedStatus::Paused) { + if (thread->GetStatus() != ThreadStatus::WaitHLEEvent) { + ASSERT(thread->GetStatus() == ThreadStatus::WaitSynch); + ASSERT(thread->IsWaitingSync()); + } + thread->SetSynchronizationResults(&obj, RESULT_SUCCESS); + thread->ResumeFromWait(); + } + } + obj.ClearWaitingThreads(); } } std::pair<ResultCode, Handle> Synchronization::WaitFor( std::vector<std::shared_ptr<SynchronizationObject>>& sync_objects, s64 nano_seconds) { + auto& kernel = system.Kernel(); auto* const thread = system.CurrentScheduler().GetCurrentThread(); - // Find the first object that is acquirable in the provided list of objects - const auto itr = std::find_if(sync_objects.begin(), sync_objects.end(), - [thread](const std::shared_ptr<SynchronizationObject>& object) { - return object->IsSignaled(); - }); - - if (itr != sync_objects.end()) { - // We found a ready object, acquire it and set the result value - SynchronizationObject* object = itr->get(); - object->Acquire(thread); - const u32 index = static_cast<s32>(std::distance(sync_objects.begin(), itr)); - return {RESULT_SUCCESS, index}; + Handle event_handle = InvalidHandle; + { + SchedulerLockAndSleep lock(kernel, event_handle, thread, nano_seconds); + const auto itr = + std::find_if(sync_objects.begin(), sync_objects.end(), + [thread](const std::shared_ptr<SynchronizationObject>& object) { + return object->IsSignaled(); + }); + + if (itr != sync_objects.end()) { + // We found a ready object, acquire it and set the result value + SynchronizationObject* object = itr->get(); + object->Acquire(thread); + const u32 index = static_cast<s32>(std::distance(sync_objects.begin(), itr)); + lock.CancelSleep(); + return {RESULT_SUCCESS, index}; + } + + if (nano_seconds == 0) { + lock.CancelSleep(); + return {RESULT_TIMEOUT, InvalidHandle}; + } + + if (thread->IsPendingTermination()) { + lock.CancelSleep(); + return {ERR_THREAD_TERMINATING, InvalidHandle}; + } + + if (thread->IsSyncCancelled()) { + thread->SetSyncCancelled(false); + lock.CancelSleep(); + return {ERR_SYNCHRONIZATION_CANCELED, InvalidHandle}; + } + + for (auto& object : sync_objects) { + object->AddWaitingThread(SharedFrom(thread)); + } + + thread->SetSynchronizationObjects(&sync_objects); + thread->SetSynchronizationResults(nullptr, RESULT_TIMEOUT); + thread->SetStatus(ThreadStatus::WaitSynch); + thread->SetWaitingSync(true); } + thread->SetWaitingSync(false); - // No objects were ready to be acquired, prepare to suspend the thread. - - // If a timeout value of 0 was provided, just return the Timeout error code instead of - // suspending the thread. - if (nano_seconds == 0) { - return {RESULT_TIMEOUT, InvalidHandle}; + if (event_handle != InvalidHandle) { + auto& time_manager = kernel.TimeManager(); + time_manager.UnscheduleTimeEvent(event_handle); } - if (thread->IsSyncCancelled()) { - thread->SetSyncCancelled(false); - return {ERR_SYNCHRONIZATION_CANCELED, InvalidHandle}; + { + SchedulerLock lock(kernel); + ResultCode signaling_result = thread->GetSignalingResult(); + SynchronizationObject* signaling_object = thread->GetSignalingObject(); + thread->SetSynchronizationObjects(nullptr); + auto shared_thread = SharedFrom(thread); + for (auto& obj : sync_objects) { + obj->RemoveWaitingThread(shared_thread); + } + if (signaling_object != nullptr) { + const auto itr = std::find_if( + sync_objects.begin(), sync_objects.end(), + [signaling_object](const std::shared_ptr<SynchronizationObject>& object) { + return object.get() == signaling_object; + }); + ASSERT(itr != sync_objects.end()); + signaling_object->Acquire(thread); + const u32 index = static_cast<s32>(std::distance(sync_objects.begin(), itr)); + return {signaling_result, index}; + } + return {signaling_result, -1}; } - - for (auto& object : sync_objects) { - object->AddWaitingThread(SharedFrom(thread)); - } - - thread->SetSynchronizationObjects(std::move(sync_objects)); - thread->SetStatus(ThreadStatus::WaitSynch); - - // Create an event to wake the thread up after the specified nanosecond delay has passed - thread->WakeAfterDelay(nano_seconds); - thread->SetWakeupCallback(DefaultThreadWakeupCallback); - - system.PrepareReschedule(thread->GetProcessorID()); - - return {RESULT_TIMEOUT, InvalidHandle}; } } // namespace Kernel diff --git a/src/core/hle/kernel/synchronization_object.cpp b/src/core/hle/kernel/synchronization_object.cpp index 43f3eef18..ba4d39157 100644 --- a/src/core/hle/kernel/synchronization_object.cpp +++ b/src/core/hle/kernel/synchronization_object.cpp @@ -38,68 +38,8 @@ void SynchronizationObject::RemoveWaitingThread(std::shared_ptr<Thread> thread) waiting_threads.erase(itr); } -std::shared_ptr<Thread> SynchronizationObject::GetHighestPriorityReadyThread() const { - Thread* candidate = nullptr; - u32 candidate_priority = THREADPRIO_LOWEST + 1; - - for (const auto& thread : waiting_threads) { - const ThreadStatus thread_status = thread->GetStatus(); - - // The list of waiting threads must not contain threads that are not waiting to be awakened. - ASSERT_MSG(thread_status == ThreadStatus::WaitSynch || - thread_status == ThreadStatus::WaitHLEEvent, - "Inconsistent thread statuses in waiting_threads"); - - if (thread->GetPriority() >= candidate_priority) - continue; - - if (ShouldWait(thread.get())) - continue; - - candidate = thread.get(); - candidate_priority = thread->GetPriority(); - } - - return SharedFrom(candidate); -} - -void SynchronizationObject::WakeupWaitingThread(std::shared_ptr<Thread> thread) { - ASSERT(!ShouldWait(thread.get())); - - if (!thread) { - return; - } - - if (thread->IsSleepingOnWait()) { - for (const auto& object : thread->GetSynchronizationObjects()) { - ASSERT(!object->ShouldWait(thread.get())); - object->Acquire(thread.get()); - } - } else { - Acquire(thread.get()); - } - - const std::size_t index = thread->GetSynchronizationObjectIndex(SharedFrom(this)); - - thread->ClearSynchronizationObjects(); - - thread->CancelWakeupTimer(); - - bool resume = true; - if (thread->HasWakeupCallback()) { - resume = thread->InvokeWakeupCallback(ThreadWakeupReason::Signal, thread, SharedFrom(this), - index); - } - if (resume) { - thread->ResumeFromWait(); - kernel.PrepareReschedule(thread->GetProcessorID()); - } -} - -void SynchronizationObject::WakeupAllWaitingThreads() { - while (auto thread = GetHighestPriorityReadyThread()) { - WakeupWaitingThread(thread); - } +void SynchronizationObject::ClearWaitingThreads() { + waiting_threads.clear(); } const std::vector<std::shared_ptr<Thread>>& SynchronizationObject::GetWaitingThreads() const { diff --git a/src/core/hle/kernel/synchronization_object.h b/src/core/hle/kernel/synchronization_object.h index 741c31faf..f89b24204 100644 --- a/src/core/hle/kernel/synchronization_object.h +++ b/src/core/hle/kernel/synchronization_object.h @@ -12,6 +12,7 @@ namespace Kernel { class KernelCore; +class Synchronization; class Thread; /// Class that represents a Kernel object that a thread can be waiting on @@ -49,24 +50,11 @@ public: */ void RemoveWaitingThread(std::shared_ptr<Thread> thread); - /** - * Wake up all threads waiting on this object that can be awoken, in priority order, - * and set the synchronization result and output of the thread. - */ - void WakeupAllWaitingThreads(); - - /** - * Wakes up a single thread waiting on this object. - * @param thread Thread that is waiting on this object to wakeup. - */ - void WakeupWaitingThread(std::shared_ptr<Thread> thread); - - /// Obtains the highest priority thread that is ready to run from this object's waiting list. - std::shared_ptr<Thread> GetHighestPriorityReadyThread() const; - /// Get a const reference to the waiting threads list for debug use const std::vector<std::shared_ptr<Thread>>& GetWaitingThreads() const; + void ClearWaitingThreads(); + protected: bool is_signaled{}; // Tells if this sync object is signalled; diff --git a/src/core/hle/kernel/thread.cpp b/src/core/hle/kernel/thread.cpp index db7f379ac..2b1092697 100644 --- a/src/core/hle/kernel/thread.cpp +++ b/src/core/hle/kernel/thread.cpp @@ -9,12 +9,21 @@ #include "common/assert.h" #include "common/common_types.h" +#include "common/fiber.h" #include "common/logging/log.h" #include "common/thread_queue_list.h" #include "core/arm/arm_interface.h" +#ifdef ARCHITECTURE_x86_64 +#include "core/arm/dynarmic/arm_dynarmic_32.h" +#include "core/arm/dynarmic/arm_dynarmic_64.h" +#endif +#include "core/arm/cpu_interrupt_handler.h" +#include "core/arm/exclusive_monitor.h" +#include "core/arm/unicorn/arm_unicorn.h" #include "core/core.h" #include "core/core_timing.h" #include "core/core_timing_util.h" +#include "core/cpu_manager.h" #include "core/hardware_properties.h" #include "core/hle/kernel/errors.h" #include "core/hle/kernel/handle_table.h" @@ -23,6 +32,7 @@ #include "core/hle/kernel/process.h" #include "core/hle/kernel/scheduler.h" #include "core/hle/kernel/thread.h" +#include "core/hle/kernel/time_manager.h" #include "core/hle/result.h" #include "core/memory.h" @@ -44,46 +54,26 @@ Thread::Thread(KernelCore& kernel) : SynchronizationObject{kernel} {} Thread::~Thread() = default; void Thread::Stop() { - // Cancel any outstanding wakeup events for this thread - Core::System::GetInstance().CoreTiming().UnscheduleEvent(kernel.ThreadWakeupCallbackEventType(), - global_handle); - kernel.GlobalHandleTable().Close(global_handle); - global_handle = 0; - SetStatus(ThreadStatus::Dead); - Signal(); - - // Clean up any dangling references in objects that this thread was waiting for - for (auto& wait_object : wait_objects) { - wait_object->RemoveWaitingThread(SharedFrom(this)); - } - wait_objects.clear(); - - owner_process->UnregisterThread(this); - - // Mark the TLS slot in the thread's page as free. - owner_process->FreeTLSRegion(tls_address); -} - -void Thread::WakeAfterDelay(s64 nanoseconds) { - // Don't schedule a wakeup if the thread wants to wait forever - if (nanoseconds == -1) - return; + { + SchedulerLock lock(kernel); + SetStatus(ThreadStatus::Dead); + Signal(); + kernel.GlobalHandleTable().Close(global_handle); - // This function might be called from any thread so we have to be cautious and use the - // thread-safe version of ScheduleEvent. - const s64 cycles = Core::Timing::nsToCycles(std::chrono::nanoseconds{nanoseconds}); - Core::System::GetInstance().CoreTiming().ScheduleEvent( - cycles, kernel.ThreadWakeupCallbackEventType(), global_handle); -} + if (owner_process) { + owner_process->UnregisterThread(this); -void Thread::CancelWakeupTimer() { - Core::System::GetInstance().CoreTiming().UnscheduleEvent(kernel.ThreadWakeupCallbackEventType(), - global_handle); + // Mark the TLS slot in the thread's page as free. + owner_process->FreeTLSRegion(tls_address); + } + arm_interface.reset(); + has_exited = true; + } + global_handle = 0; } void Thread::ResumeFromWait() { - ASSERT_MSG(wait_objects.empty(), "Thread is waking up while waiting for objects"); - + SchedulerLock lock(kernel); switch (status) { case ThreadStatus::Paused: case ThreadStatus::WaitSynch: @@ -99,7 +89,7 @@ void Thread::ResumeFromWait() { case ThreadStatus::Ready: // The thread's wakeup callback must have already been cleared when the thread was first // awoken. - ASSERT(wakeup_callback == nullptr); + ASSERT(hle_callback == nullptr); // If the thread is waiting on multiple wait objects, it might be awoken more than once // before actually resuming. We can ignore subsequent wakeups if the thread status has // already been set to ThreadStatus::Ready. @@ -115,24 +105,31 @@ void Thread::ResumeFromWait() { return; } - wakeup_callback = nullptr; + SetStatus(ThreadStatus::Ready); +} + +void Thread::OnWakeUp() { + SchedulerLock lock(kernel); - if (activity == ThreadActivity::Paused) { - SetStatus(ThreadStatus::Paused); - return; - } + SetStatus(ThreadStatus::Ready); +} +ResultCode Thread::Start() { + SchedulerLock lock(kernel); SetStatus(ThreadStatus::Ready); + return RESULT_SUCCESS; } void Thread::CancelWait() { - if (GetSchedulingStatus() != ThreadSchedStatus::Paused) { + SchedulerLock lock(kernel); + if (GetSchedulingStatus() != ThreadSchedStatus::Paused || !is_waiting_on_sync) { is_sync_cancelled = true; return; } + // TODO(Blinkhawk): Implement cancel of server session is_sync_cancelled = false; - SetWaitSynchronizationResult(ERR_SYNCHRONIZATION_CANCELED); - ResumeFromWait(); + SetSynchronizationResults(nullptr, ERR_SYNCHRONIZATION_CANCELED); + SetStatus(ThreadStatus::Ready); } static void ResetThreadContext32(Core::ARM_Interface::ThreadContext32& context, u32 stack_top, @@ -153,12 +150,29 @@ static void ResetThreadContext64(Core::ARM_Interface::ThreadContext64& context, context.fpcr = 0; } -ResultVal<std::shared_ptr<Thread>> Thread::Create(KernelCore& kernel, std::string name, - VAddr entry_point, u32 priority, u64 arg, - s32 processor_id, VAddr stack_top, - Process& owner_process) { +std::shared_ptr<Common::Fiber>& Thread::GetHostContext() { + return host_context; +} + +ResultVal<std::shared_ptr<Thread>> Thread::Create(Core::System& system, ThreadType type_flags, + std::string name, VAddr entry_point, u32 priority, + u64 arg, s32 processor_id, VAddr stack_top, + Process* owner_process) { + std::function<void(void*)> init_func = system.GetCpuManager().GetGuestThreadStartFunc(); + void* init_func_parameter = system.GetCpuManager().GetStartFuncParamater(); + return Create(system, type_flags, name, entry_point, priority, arg, processor_id, stack_top, + owner_process, std::move(init_func), init_func_parameter); +} + +ResultVal<std::shared_ptr<Thread>> Thread::Create(Core::System& system, ThreadType type_flags, + std::string name, VAddr entry_point, u32 priority, + u64 arg, s32 processor_id, VAddr stack_top, + Process* owner_process, + std::function<void(void*)>&& thread_start_func, + void* thread_start_parameter) { + auto& kernel = system.Kernel(); // Check if priority is in ranged. Lowest priority -> highest priority id. - if (priority > THREADPRIO_LOWEST) { + if (priority > THREADPRIO_LOWEST && ((type_flags & THREADTYPE_IDLE) == 0)) { LOG_ERROR(Kernel_SVC, "Invalid thread priority: {}", priority); return ERR_INVALID_THREAD_PRIORITY; } @@ -168,11 +182,12 @@ ResultVal<std::shared_ptr<Thread>> Thread::Create(KernelCore& kernel, std::strin return ERR_INVALID_PROCESSOR_ID; } - auto& system = Core::System::GetInstance(); - if (!system.Memory().IsValidVirtualAddress(owner_process, entry_point)) { - LOG_ERROR(Kernel_SVC, "(name={}): invalid entry {:016X}", name, entry_point); - // TODO (bunnei): Find the correct error code to use here - return RESULT_UNKNOWN; + if (owner_process) { + if (!system.Memory().IsValidVirtualAddress(*owner_process, entry_point)) { + LOG_ERROR(Kernel_SVC, "(name={}): invalid entry {:016X}", name, entry_point); + // TODO (bunnei): Find the correct error code to use here + return RESULT_UNKNOWN; + } } std::shared_ptr<Thread> thread = std::make_shared<Thread>(kernel); @@ -183,51 +198,82 @@ ResultVal<std::shared_ptr<Thread>> Thread::Create(KernelCore& kernel, std::strin thread->stack_top = stack_top; thread->tpidr_el0 = 0; thread->nominal_priority = thread->current_priority = priority; - thread->last_running_ticks = system.CoreTiming().GetTicks(); + thread->last_running_ticks = 0; thread->processor_id = processor_id; thread->ideal_core = processor_id; thread->affinity_mask = 1ULL << processor_id; - thread->wait_objects.clear(); + thread->wait_objects = nullptr; thread->mutex_wait_address = 0; thread->condvar_wait_address = 0; thread->wait_handle = 0; thread->name = std::move(name); thread->global_handle = kernel.GlobalHandleTable().Create(thread).Unwrap(); - thread->owner_process = &owner_process; - auto& scheduler = kernel.GlobalScheduler(); - scheduler.AddThread(thread); - thread->tls_address = thread->owner_process->CreateTLSRegion(); - - thread->owner_process->RegisterThread(thread.get()); + thread->owner_process = owner_process; + thread->type = type_flags; + if ((type_flags & THREADTYPE_IDLE) == 0) { + auto& scheduler = kernel.GlobalScheduler(); + scheduler.AddThread(thread); + } + if (owner_process) { + thread->tls_address = thread->owner_process->CreateTLSRegion(); + thread->owner_process->RegisterThread(thread.get()); + } else { + thread->tls_address = 0; + } + // TODO(peachum): move to ScheduleThread() when scheduler is added so selected core is used + // to initialize the context + thread->arm_interface.reset(); + if ((type_flags & THREADTYPE_HLE) == 0) { +#ifdef ARCHITECTURE_x86_64 + if (owner_process && !owner_process->Is64BitProcess()) { + thread->arm_interface = std::make_unique<Core::ARM_Dynarmic_32>( + system, kernel.Interrupts(), kernel.IsMulticore(), kernel.GetExclusiveMonitor(), + processor_id); + } else { + thread->arm_interface = std::make_unique<Core::ARM_Dynarmic_64>( + system, kernel.Interrupts(), kernel.IsMulticore(), kernel.GetExclusiveMonitor(), + processor_id); + } - ResetThreadContext32(thread->context_32, static_cast<u32>(stack_top), - static_cast<u32>(entry_point), static_cast<u32>(arg)); - ResetThreadContext64(thread->context_64, stack_top, entry_point, arg); +#else + if (owner_process && !owner_process->Is64BitProcess()) { + thread->arm_interface = std::make_shared<Core::ARM_Unicorn>( + system, kernel.Interrupts(), kernel.IsMulticore(), ARM_Unicorn::Arch::AArch32, + processor_id); + } else { + thread->arm_interface = std::make_shared<Core::ARM_Unicorn>( + system, kernel.Interrupts(), kernel.IsMulticore(), ARM_Unicorn::Arch::AArch64, + processor_id); + } + LOG_WARNING(Core, "CPU JIT requested, but Dynarmic not available"); +#endif + ResetThreadContext32(thread->context_32, static_cast<u32>(stack_top), + static_cast<u32>(entry_point), static_cast<u32>(arg)); + ResetThreadContext64(thread->context_64, stack_top, entry_point, arg); + } + thread->host_context = + std::make_shared<Common::Fiber>(std::move(thread_start_func), thread_start_parameter); return MakeResult<std::shared_ptr<Thread>>(std::move(thread)); } void Thread::SetPriority(u32 priority) { + SchedulerLock lock(kernel); ASSERT_MSG(priority <= THREADPRIO_LOWEST && priority >= THREADPRIO_HIGHEST, "Invalid priority value."); nominal_priority = priority; UpdatePriority(); } -void Thread::SetWaitSynchronizationResult(ResultCode result) { - context_32.cpu_registers[0] = result.raw; - context_64.cpu_registers[0] = result.raw; -} - -void Thread::SetWaitSynchronizationOutput(s32 output) { - context_32.cpu_registers[1] = output; - context_64.cpu_registers[1] = output; +void Thread::SetSynchronizationResults(SynchronizationObject* object, ResultCode result) { + signaling_object = object; + signaling_result = result; } s32 Thread::GetSynchronizationObjectIndex(std::shared_ptr<SynchronizationObject> object) const { - ASSERT_MSG(!wait_objects.empty(), "Thread is not waiting for anything"); - const auto match = std::find(wait_objects.rbegin(), wait_objects.rend(), object); - return static_cast<s32>(std::distance(match, wait_objects.rend()) - 1); + ASSERT_MSG(!wait_objects->empty(), "Thread is not waiting for anything"); + const auto match = std::find(wait_objects->rbegin(), wait_objects->rend(), object); + return static_cast<s32>(std::distance(match, wait_objects->rend()) - 1); } VAddr Thread::GetCommandBufferAddress() const { @@ -236,6 +282,14 @@ VAddr Thread::GetCommandBufferAddress() const { return GetTLSAddress() + command_header_offset; } +Core::ARM_Interface& Thread::ArmInterface() { + return *arm_interface; +} + +const Core::ARM_Interface& Thread::ArmInterface() const { + return *arm_interface; +} + void Thread::SetStatus(ThreadStatus new_status) { if (new_status == status) { return; @@ -257,10 +311,6 @@ void Thread::SetStatus(ThreadStatus new_status) { break; } - if (status == ThreadStatus::Running) { - last_running_ticks = Core::System::GetInstance().CoreTiming().GetTicks(); - } - status = new_status; } @@ -341,75 +391,116 @@ void Thread::UpdatePriority() { lock_owner->UpdatePriority(); } -void Thread::ChangeCore(u32 core, u64 mask) { - SetCoreAndAffinityMask(core, mask); -} - bool Thread::AllSynchronizationObjectsReady() const { - return std::none_of(wait_objects.begin(), wait_objects.end(), + return std::none_of(wait_objects->begin(), wait_objects->end(), [this](const std::shared_ptr<SynchronizationObject>& object) { return object->ShouldWait(this); }); } -bool Thread::InvokeWakeupCallback(ThreadWakeupReason reason, std::shared_ptr<Thread> thread, - std::shared_ptr<SynchronizationObject> object, - std::size_t index) { - ASSERT(wakeup_callback); - return wakeup_callback(reason, std::move(thread), std::move(object), index); +bool Thread::InvokeHLECallback(std::shared_ptr<Thread> thread) { + ASSERT(hle_callback); + return hle_callback(std::move(thread)); } -void Thread::SetActivity(ThreadActivity value) { - activity = value; +ResultCode Thread::SetActivity(ThreadActivity value) { + SchedulerLock lock(kernel); + + auto sched_status = GetSchedulingStatus(); + + if (sched_status != ThreadSchedStatus::Runnable && sched_status != ThreadSchedStatus::Paused) { + return ERR_INVALID_STATE; + } + + if (IsPendingTermination()) { + return RESULT_SUCCESS; + } if (value == ThreadActivity::Paused) { - // Set status if not waiting - if (status == ThreadStatus::Ready || status == ThreadStatus::Running) { - SetStatus(ThreadStatus::Paused); - kernel.PrepareReschedule(processor_id); + if ((pausing_state & static_cast<u32>(ThreadSchedFlags::ThreadPauseFlag)) != 0) { + return ERR_INVALID_STATE; + } + AddSchedulingFlag(ThreadSchedFlags::ThreadPauseFlag); + } else { + if ((pausing_state & static_cast<u32>(ThreadSchedFlags::ThreadPauseFlag)) == 0) { + return ERR_INVALID_STATE; } - } else if (status == ThreadStatus::Paused) { - // Ready to reschedule - ResumeFromWait(); + RemoveSchedulingFlag(ThreadSchedFlags::ThreadPauseFlag); } + return RESULT_SUCCESS; } -void Thread::Sleep(s64 nanoseconds) { - // Sleep current thread and check for next thread to schedule - SetStatus(ThreadStatus::WaitSleep); +ResultCode Thread::Sleep(s64 nanoseconds) { + Handle event_handle{}; + { + SchedulerLockAndSleep lock(kernel, event_handle, this, nanoseconds); + SetStatus(ThreadStatus::WaitSleep); + } - // Create an event to wake the thread up after the specified nanosecond delay has passed - WakeAfterDelay(nanoseconds); + if (event_handle != InvalidHandle) { + auto& time_manager = kernel.TimeManager(); + time_manager.UnscheduleTimeEvent(event_handle); + } + return RESULT_SUCCESS; +} + +std::pair<ResultCode, bool> Thread::YieldSimple() { + bool is_redundant = false; + { + SchedulerLock lock(kernel); + is_redundant = kernel.GlobalScheduler().YieldThread(this); + } + return {RESULT_SUCCESS, is_redundant}; +} + +std::pair<ResultCode, bool> Thread::YieldAndBalanceLoad() { + bool is_redundant = false; + { + SchedulerLock lock(kernel); + is_redundant = kernel.GlobalScheduler().YieldThreadAndBalanceLoad(this); + } + return {RESULT_SUCCESS, is_redundant}; } -bool Thread::YieldSimple() { - auto& scheduler = kernel.GlobalScheduler(); - return scheduler.YieldThread(this); +std::pair<ResultCode, bool> Thread::YieldAndWaitForLoadBalancing() { + bool is_redundant = false; + { + SchedulerLock lock(kernel); + is_redundant = kernel.GlobalScheduler().YieldThreadAndWaitForLoadBalancing(this); + } + return {RESULT_SUCCESS, is_redundant}; } -bool Thread::YieldAndBalanceLoad() { - auto& scheduler = kernel.GlobalScheduler(); - return scheduler.YieldThreadAndBalanceLoad(this); +void Thread::AddSchedulingFlag(ThreadSchedFlags flag) { + const u32 old_state = scheduling_state; + pausing_state |= static_cast<u32>(flag); + const u32 base_scheduling = static_cast<u32>(GetSchedulingStatus()); + scheduling_state = base_scheduling | pausing_state; + kernel.GlobalScheduler().AdjustSchedulingOnStatus(this, old_state); } -bool Thread::YieldAndWaitForLoadBalancing() { - auto& scheduler = kernel.GlobalScheduler(); - return scheduler.YieldThreadAndWaitForLoadBalancing(this); +void Thread::RemoveSchedulingFlag(ThreadSchedFlags flag) { + const u32 old_state = scheduling_state; + pausing_state &= ~static_cast<u32>(flag); + const u32 base_scheduling = static_cast<u32>(GetSchedulingStatus()); + scheduling_state = base_scheduling | pausing_state; + kernel.GlobalScheduler().AdjustSchedulingOnStatus(this, old_state); } void Thread::SetSchedulingStatus(ThreadSchedStatus new_status) { - const u32 old_flags = scheduling_state; + const u32 old_state = scheduling_state; scheduling_state = (scheduling_state & static_cast<u32>(ThreadSchedMasks::HighMask)) | static_cast<u32>(new_status); - AdjustSchedulingOnStatus(old_flags); + kernel.GlobalScheduler().AdjustSchedulingOnStatus(this, old_state); } void Thread::SetCurrentPriority(u32 new_priority) { const u32 old_priority = std::exchange(current_priority, new_priority); - AdjustSchedulingOnPriority(old_priority); + kernel.GlobalScheduler().AdjustSchedulingOnPriority(this, old_priority); } ResultCode Thread::SetCoreAndAffinityMask(s32 new_core, u64 new_affinity_mask) { + SchedulerLock lock(kernel); const auto HighestSetCore = [](u64 mask, u32 max_cores) { for (s32 core = static_cast<s32>(max_cores - 1); core >= 0; core--) { if (((mask >> core) & 1) != 0) { @@ -443,111 +534,12 @@ ResultCode Thread::SetCoreAndAffinityMask(s32 new_core, u64 new_affinity_mask) { processor_id = ideal_core; } } - AdjustSchedulingOnAffinity(old_affinity_mask, old_core); + kernel.GlobalScheduler().AdjustSchedulingOnAffinity(this, old_affinity_mask, old_core); } } return RESULT_SUCCESS; } -void Thread::AdjustSchedulingOnStatus(u32 old_flags) { - if (old_flags == scheduling_state) { - return; - } - - auto& scheduler = kernel.GlobalScheduler(); - if (static_cast<ThreadSchedStatus>(old_flags & static_cast<u32>(ThreadSchedMasks::LowMask)) == - ThreadSchedStatus::Runnable) { - // In this case the thread was running, now it's pausing/exitting - if (processor_id >= 0) { - scheduler.Unschedule(current_priority, static_cast<u32>(processor_id), this); - } - - for (u32 core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) { - if (core != static_cast<u32>(processor_id) && ((affinity_mask >> core) & 1) != 0) { - scheduler.Unsuggest(current_priority, core, this); - } - } - } else if (GetSchedulingStatus() == ThreadSchedStatus::Runnable) { - // The thread is now set to running from being stopped - if (processor_id >= 0) { - scheduler.Schedule(current_priority, static_cast<u32>(processor_id), this); - } - - for (u32 core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) { - if (core != static_cast<u32>(processor_id) && ((affinity_mask >> core) & 1) != 0) { - scheduler.Suggest(current_priority, core, this); - } - } - } - - scheduler.SetReselectionPending(); -} - -void Thread::AdjustSchedulingOnPriority(u32 old_priority) { - if (GetSchedulingStatus() != ThreadSchedStatus::Runnable) { - return; - } - auto& scheduler = kernel.GlobalScheduler(); - if (processor_id >= 0) { - scheduler.Unschedule(old_priority, static_cast<u32>(processor_id), this); - } - - for (u32 core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) { - if (core != static_cast<u32>(processor_id) && ((affinity_mask >> core) & 1) != 0) { - scheduler.Unsuggest(old_priority, core, this); - } - } - - // Add thread to the new priority queues. - Thread* current_thread = GetCurrentThread(); - - if (processor_id >= 0) { - if (current_thread == this) { - scheduler.SchedulePrepend(current_priority, static_cast<u32>(processor_id), this); - } else { - scheduler.Schedule(current_priority, static_cast<u32>(processor_id), this); - } - } - - for (u32 core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) { - if (core != static_cast<u32>(processor_id) && ((affinity_mask >> core) & 1) != 0) { - scheduler.Suggest(current_priority, core, this); - } - } - - scheduler.SetReselectionPending(); -} - -void Thread::AdjustSchedulingOnAffinity(u64 old_affinity_mask, s32 old_core) { - auto& scheduler = kernel.GlobalScheduler(); - if (GetSchedulingStatus() != ThreadSchedStatus::Runnable || - current_priority >= THREADPRIO_COUNT) { - return; - } - - for (u32 core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) { - if (((old_affinity_mask >> core) & 1) != 0) { - if (core == static_cast<u32>(old_core)) { - scheduler.Unschedule(current_priority, core, this); - } else { - scheduler.Unsuggest(current_priority, core, this); - } - } - } - - for (u32 core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) { - if (((affinity_mask >> core) & 1) != 0) { - if (core == static_cast<u32>(processor_id)) { - scheduler.Schedule(current_priority, core, this); - } else { - scheduler.Suggest(current_priority, core, this); - } - } - } - - scheduler.SetReselectionPending(); -} - //////////////////////////////////////////////////////////////////////////////////////////////////// /** diff --git a/src/core/hle/kernel/thread.h b/src/core/hle/kernel/thread.h index 23fdef8a4..c0342c462 100644 --- a/src/core/hle/kernel/thread.h +++ b/src/core/hle/kernel/thread.h @@ -6,26 +6,47 @@ #include <functional> #include <string> +#include <utility> #include <vector> #include "common/common_types.h" +#include "common/spin_lock.h" #include "core/arm/arm_interface.h" #include "core/hle/kernel/object.h" #include "core/hle/kernel/synchronization_object.h" #include "core/hle/result.h" +namespace Common { +class Fiber; +} + +namespace Core { +class ARM_Interface; +class System; +} // namespace Core + namespace Kernel { +class GlobalScheduler; class KernelCore; class Process; class Scheduler; enum ThreadPriority : u32 { - THREADPRIO_HIGHEST = 0, ///< Highest thread priority - THREADPRIO_USERLAND_MAX = 24, ///< Highest thread priority for userland apps - THREADPRIO_DEFAULT = 44, ///< Default thread priority for userland apps - THREADPRIO_LOWEST = 63, ///< Lowest thread priority - THREADPRIO_COUNT = 64, ///< Total number of possible thread priorities. + THREADPRIO_HIGHEST = 0, ///< Highest thread priority + THREADPRIO_MAX_CORE_MIGRATION = 2, ///< Highest priority for a core migration + THREADPRIO_USERLAND_MAX = 24, ///< Highest thread priority for userland apps + THREADPRIO_DEFAULT = 44, ///< Default thread priority for userland apps + THREADPRIO_LOWEST = 63, ///< Lowest thread priority + THREADPRIO_COUNT = 64, ///< Total number of possible thread priorities. +}; + +enum ThreadType : u32 { + THREADTYPE_USER = 0x1, + THREADTYPE_KERNEL = 0x2, + THREADTYPE_HLE = 0x4, + THREADTYPE_IDLE = 0x8, + THREADTYPE_SUSPEND = 0x10, }; enum ThreadProcessorId : s32 { @@ -107,26 +128,45 @@ public: using ThreadSynchronizationObjects = std::vector<std::shared_ptr<SynchronizationObject>>; - using WakeupCallback = - std::function<bool(ThreadWakeupReason reason, std::shared_ptr<Thread> thread, - std::shared_ptr<SynchronizationObject> object, std::size_t index)>; + using HLECallback = std::function<bool(std::shared_ptr<Thread> thread)>; + + /** + * Creates and returns a new thread. The new thread is immediately scheduled + * @param system The instance of the whole system + * @param name The friendly name desired for the thread + * @param entry_point The address at which the thread should start execution + * @param priority The thread's priority + * @param arg User data to pass to the thread + * @param processor_id The ID(s) of the processors on which the thread is desired to be run + * @param stack_top The address of the thread's stack top + * @param owner_process The parent process for the thread, if null, it's a kernel thread + * @return A shared pointer to the newly created thread + */ + static ResultVal<std::shared_ptr<Thread>> Create(Core::System& system, ThreadType type_flags, + std::string name, VAddr entry_point, + u32 priority, u64 arg, s32 processor_id, + VAddr stack_top, Process* owner_process); /** * Creates and returns a new thread. The new thread is immediately scheduled - * @param kernel The kernel instance this thread will be created under. + * @param system The instance of the whole system * @param name The friendly name desired for the thread * @param entry_point The address at which the thread should start execution * @param priority The thread's priority * @param arg User data to pass to the thread * @param processor_id The ID(s) of the processors on which the thread is desired to be run * @param stack_top The address of the thread's stack top - * @param owner_process The parent process for the thread + * @param owner_process The parent process for the thread, if null, it's a kernel thread + * @param thread_start_func The function where the host context will start. + * @param thread_start_parameter The parameter which will passed to host context on init * @return A shared pointer to the newly created thread */ - static ResultVal<std::shared_ptr<Thread>> Create(KernelCore& kernel, std::string name, - VAddr entry_point, u32 priority, u64 arg, - s32 processor_id, VAddr stack_top, - Process& owner_process); + static ResultVal<std::shared_ptr<Thread>> Create(Core::System& system, ThreadType type_flags, + std::string name, VAddr entry_point, + u32 priority, u64 arg, s32 processor_id, + VAddr stack_top, Process* owner_process, + std::function<void(void*)>&& thread_start_func, + void* thread_start_parameter); std::string GetName() const override { return name; @@ -181,7 +221,7 @@ public: void UpdatePriority(); /// Changes the core that the thread is running or scheduled to run on. - void ChangeCore(u32 core, u64 mask); + ResultCode SetCoreAndAffinityMask(s32 new_core, u64 new_affinity_mask); /** * Gets the thread's thread ID @@ -194,6 +234,10 @@ public: /// Resumes a thread from waiting void ResumeFromWait(); + void OnWakeUp(); + + ResultCode Start(); + /// Cancels a waiting operation that this thread may or may not be within. /// /// When the thread is within a waiting state, this will set the thread's @@ -202,26 +246,19 @@ public: /// void CancelWait(); - /** - * Schedules an event to wake up the specified thread after the specified delay - * @param nanoseconds The time this thread will be allowed to sleep for - */ - void WakeAfterDelay(s64 nanoseconds); + void SetSynchronizationResults(SynchronizationObject* object, ResultCode result); - /// Cancel any outstanding wakeup events for this thread - void CancelWakeupTimer(); + Core::ARM_Interface& ArmInterface(); - /** - * Sets the result after the thread awakens (from svcWaitSynchronization) - * @param result Value to set to the returned result - */ - void SetWaitSynchronizationResult(ResultCode result); + const Core::ARM_Interface& ArmInterface() const; - /** - * Sets the output parameter value after the thread awakens (from svcWaitSynchronization) - * @param output Value to set to the output parameter - */ - void SetWaitSynchronizationOutput(s32 output); + SynchronizationObject* GetSignalingObject() const { + return signaling_object; + } + + ResultCode GetSignalingResult() const { + return signaling_result; + } /** * Retrieves the index that this particular object occupies in the list of objects @@ -269,11 +306,6 @@ public: */ VAddr GetCommandBufferAddress() const; - /// Returns whether this thread is waiting on objects from a WaitSynchronization call. - bool IsSleepingOnWait() const { - return status == ThreadStatus::WaitSynch; - } - ThreadContext32& GetContext32() { return context_32; } @@ -290,6 +322,28 @@ public: return context_64; } + bool IsHLEThread() const { + return (type & THREADTYPE_HLE) != 0; + } + + bool IsSuspendThread() const { + return (type & THREADTYPE_SUSPEND) != 0; + } + + bool IsIdleThread() const { + return (type & THREADTYPE_IDLE) != 0; + } + + bool WasRunning() const { + return was_running; + } + + void SetWasRunning(bool value) { + was_running = value; + } + + std::shared_ptr<Common::Fiber>& GetHostContext(); + ThreadStatus GetStatus() const { return status; } @@ -325,18 +379,18 @@ public: } const ThreadSynchronizationObjects& GetSynchronizationObjects() const { - return wait_objects; + return *wait_objects; } - void SetSynchronizationObjects(ThreadSynchronizationObjects objects) { - wait_objects = std::move(objects); + void SetSynchronizationObjects(ThreadSynchronizationObjects* objects) { + wait_objects = objects; } void ClearSynchronizationObjects() { - for (const auto& waiting_object : wait_objects) { + for (const auto& waiting_object : *wait_objects) { waiting_object->RemoveWaitingThread(SharedFrom(this)); } - wait_objects.clear(); + wait_objects->clear(); } /// Determines whether all the objects this thread is waiting on are ready. @@ -386,26 +440,35 @@ public: arb_wait_address = address; } - bool HasWakeupCallback() const { - return wakeup_callback != nullptr; + bool HasHLECallback() const { + return hle_callback != nullptr; } - void SetWakeupCallback(WakeupCallback callback) { - wakeup_callback = std::move(callback); + void SetHLECallback(HLECallback callback) { + hle_callback = std::move(callback); } - void InvalidateWakeupCallback() { - SetWakeupCallback(nullptr); + void SetHLETimeEvent(Handle time_event) { + hle_time_event = time_event; } - /** - * Invokes the thread's wakeup callback. - * - * @pre A valid wakeup callback has been set. Violating this precondition - * will cause an assertion to trigger. - */ - bool InvokeWakeupCallback(ThreadWakeupReason reason, std::shared_ptr<Thread> thread, - std::shared_ptr<SynchronizationObject> object, std::size_t index); + void SetHLESyncObject(SynchronizationObject* object) { + hle_object = object; + } + + Handle GetHLETimeEvent() const { + return hle_time_event; + } + + SynchronizationObject* GetHLESyncObject() const { + return hle_object; + } + + void InvalidateHLECallback() { + SetHLECallback(nullptr); + } + + bool InvokeHLECallback(std::shared_ptr<Thread> thread); u32 GetIdealCore() const { return ideal_core; @@ -415,23 +478,19 @@ public: return affinity_mask; } - ThreadActivity GetActivity() const { - return activity; - } - - void SetActivity(ThreadActivity value); + ResultCode SetActivity(ThreadActivity value); /// Sleeps this thread for the given amount of nanoseconds. - void Sleep(s64 nanoseconds); + ResultCode Sleep(s64 nanoseconds); /// Yields this thread without rebalancing loads. - bool YieldSimple(); + std::pair<ResultCode, bool> YieldSimple(); /// Yields this thread and does a load rebalancing. - bool YieldAndBalanceLoad(); + std::pair<ResultCode, bool> YieldAndBalanceLoad(); /// Yields this thread and if the core is left idle, loads are rebalanced - bool YieldAndWaitForLoadBalancing(); + std::pair<ResultCode, bool> YieldAndWaitForLoadBalancing(); void IncrementYieldCount() { yield_count++; @@ -446,6 +505,10 @@ public: static_cast<u32>(ThreadSchedMasks::LowMask)); } + bool IsRunnable() const { + return scheduling_state == static_cast<u32>(ThreadSchedStatus::Runnable); + } + bool IsRunning() const { return is_running; } @@ -466,17 +529,67 @@ public: return global_handle; } + bool IsWaitingForArbitration() const { + return waiting_for_arbitration; + } + + void WaitForArbitration(bool set) { + waiting_for_arbitration = set; + } + + bool IsWaitingSync() const { + return is_waiting_on_sync; + } + + void SetWaitingSync(bool is_waiting) { + is_waiting_on_sync = is_waiting; + } + + bool IsPendingTermination() const { + return will_be_terminated || GetSchedulingStatus() == ThreadSchedStatus::Exited; + } + + bool IsPaused() const { + return pausing_state != 0; + } + + bool IsContinuousOnSVC() const { + return is_continuous_on_svc; + } + + void SetContinuousOnSVC(bool is_continuous) { + is_continuous_on_svc = is_continuous; + } + + bool IsPhantomMode() const { + return is_phantom_mode; + } + + void SetPhantomMode(bool phantom) { + is_phantom_mode = phantom; + } + + bool HasExited() const { + return has_exited; + } + private: + friend class GlobalScheduler; + friend class Scheduler; + void SetSchedulingStatus(ThreadSchedStatus new_status); + void AddSchedulingFlag(ThreadSchedFlags flag); + void RemoveSchedulingFlag(ThreadSchedFlags flag); + void SetCurrentPriority(u32 new_priority); - ResultCode SetCoreAndAffinityMask(s32 new_core, u64 new_affinity_mask); - void AdjustSchedulingOnStatus(u32 old_flags); - void AdjustSchedulingOnPriority(u32 old_priority); void AdjustSchedulingOnAffinity(u64 old_affinity_mask, s32 old_core); + Common::SpinLock context_guard{}; ThreadContext32 context_32{}; ThreadContext64 context_64{}; + std::unique_ptr<Core::ARM_Interface> arm_interface{}; + std::shared_ptr<Common::Fiber> host_context{}; u64 thread_id = 0; @@ -485,6 +598,8 @@ private: VAddr entry_point = 0; VAddr stack_top = 0; + ThreadType type; + /// Nominal thread priority, as set by the emulated application. /// The nominal priority is the thread priority without priority /// inheritance taken into account. @@ -509,7 +624,10 @@ private: /// Objects that the thread is waiting on, in the same order as they were /// passed to WaitSynchronization. - ThreadSynchronizationObjects wait_objects; + ThreadSynchronizationObjects* wait_objects; + + SynchronizationObject* signaling_object; + ResultCode signaling_result{RESULT_SUCCESS}; /// List of threads that are waiting for a mutex that is held by this thread. MutexWaitingThreads wait_mutex_threads; @@ -526,30 +644,39 @@ private: /// If waiting for an AddressArbiter, this is the address being waited on. VAddr arb_wait_address{0}; + bool waiting_for_arbitration{}; /// Handle used as userdata to reference this object when inserting into the CoreTiming queue. Handle global_handle = 0; - /// Callback that will be invoked when the thread is resumed from a waiting state. If the thread - /// was waiting via WaitSynchronization then the object will be the last object that became - /// available. In case of a timeout, the object will be nullptr. - WakeupCallback wakeup_callback; + /// Callback for HLE Events + HLECallback hle_callback; + Handle hle_time_event; + SynchronizationObject* hle_object; Scheduler* scheduler = nullptr; u32 ideal_core{0xFFFFFFFF}; u64 affinity_mask{0x1}; - ThreadActivity activity = ThreadActivity::Normal; - s32 ideal_core_override = -1; u64 affinity_mask_override = 0x1; u32 affinity_override_count = 0; u32 scheduling_state = 0; + u32 pausing_state = 0; bool is_running = false; + bool is_waiting_on_sync = false; bool is_sync_cancelled = false; + bool is_continuous_on_svc = false; + + bool will_be_terminated = false; + bool is_phantom_mode = false; + bool has_exited = false; + + bool was_running = false; + std::string name; }; diff --git a/src/core/hle/kernel/time_manager.cpp b/src/core/hle/kernel/time_manager.cpp index 21b290468..941305e8e 100644 --- a/src/core/hle/kernel/time_manager.cpp +++ b/src/core/hle/kernel/time_manager.cpp @@ -8,30 +8,37 @@ #include "core/core_timing_util.h" #include "core/hle/kernel/handle_table.h" #include "core/hle/kernel/kernel.h" +#include "core/hle/kernel/scheduler.h" #include "core/hle/kernel/thread.h" #include "core/hle/kernel/time_manager.h" namespace Kernel { -TimeManager::TimeManager(Core::System& system) : system{system} { +TimeManager::TimeManager(Core::System& system_) : system{system_} { time_manager_event_type = Core::Timing::CreateEvent( "Kernel::TimeManagerCallback", [this](u64 thread_handle, [[maybe_unused]] s64 cycles_late) { + SchedulerLock lock(system.Kernel()); Handle proper_handle = static_cast<Handle>(thread_handle); + if (cancelled_events[proper_handle]) { + return; + } std::shared_ptr<Thread> thread = this->system.Kernel().RetrieveThreadFromGlobalHandleTable(proper_handle); - thread->ResumeFromWait(); + thread->OnWakeUp(); }); } void TimeManager::ScheduleTimeEvent(Handle& event_handle, Thread* timetask, s64 nanoseconds) { + event_handle = timetask->GetGlobalHandle(); if (nanoseconds > 0) { ASSERT(timetask); - event_handle = timetask->GetGlobalHandle(); - const s64 cycles = Core::Timing::nsToCycles(std::chrono::nanoseconds{nanoseconds}); - system.CoreTiming().ScheduleEvent(cycles, time_manager_event_type, event_handle); + ASSERT(timetask->GetStatus() != ThreadStatus::Ready); + ASSERT(timetask->GetStatus() != ThreadStatus::WaitMutex); + system.CoreTiming().ScheduleEvent(nanoseconds, time_manager_event_type, event_handle); } else { event_handle = InvalidHandle; } + cancelled_events[event_handle] = false; } void TimeManager::UnscheduleTimeEvent(Handle event_handle) { @@ -39,6 +46,12 @@ void TimeManager::UnscheduleTimeEvent(Handle event_handle) { return; } system.CoreTiming().UnscheduleEvent(time_manager_event_type, event_handle); + cancelled_events[event_handle] = true; +} + +void TimeManager::CancelTimeEvent(Thread* time_task) { + Handle event_handle = time_task->GetGlobalHandle(); + UnscheduleTimeEvent(event_handle); } } // namespace Kernel diff --git a/src/core/hle/kernel/time_manager.h b/src/core/hle/kernel/time_manager.h index eaec486d1..307a18765 100644 --- a/src/core/hle/kernel/time_manager.h +++ b/src/core/hle/kernel/time_manager.h @@ -5,6 +5,7 @@ #pragma once #include <memory> +#include <unordered_map> #include "core/hle/kernel/object.h" @@ -35,9 +36,12 @@ public: /// Unschedule an existing time event void UnscheduleTimeEvent(Handle event_handle); + void CancelTimeEvent(Thread* time_task); + private: Core::System& system; std::shared_ptr<Core::Timing::EventType> time_manager_event_type; + std::unordered_map<Handle, bool> cancelled_events; }; } // namespace Kernel diff --git a/src/core/hle/service/acc/acc.cpp b/src/core/hle/service/acc/acc.cpp index 630a8b048..94d8c1fc6 100644 --- a/src/core/hle/service/acc/acc.cpp +++ b/src/core/hle/service/acc/acc.cpp @@ -44,6 +44,218 @@ static constexpr u32 SanitizeJPEGSize(std::size_t size) { return static_cast<u32>(std::min(size, max_jpeg_image_size)); } +class IManagerForSystemService final : public ServiceFramework<IManagerForSystemService> { +public: + explicit IManagerForSystemService(Common::UUID user_id) + : ServiceFramework("IManagerForSystemService") { + // clang-format off + static const FunctionInfo functions[] = { + {0, nullptr, "CheckAvailability"}, + {1, nullptr, "GetAccountId"}, + {2, nullptr, "EnsureIdTokenCacheAsync"}, + {3, nullptr, "LoadIdTokenCache"}, + {100, nullptr, "SetSystemProgramIdentification"}, + {101, nullptr, "RefreshNotificationTokenAsync"}, // 7.0.0+ + {110, nullptr, "GetServiceEntryRequirementCache"}, // 4.0.0+ + {111, nullptr, "InvalidateServiceEntryRequirementCache"}, // 4.0.0+ + {112, nullptr, "InvalidateTokenCache"}, // 4.0.0 - 6.2.0 + {113, nullptr, "GetServiceEntryRequirementCacheForOnlinePlay"}, // 6.1.0+ + {120, nullptr, "GetNintendoAccountId"}, + {121, nullptr, "CalculateNintendoAccountAuthenticationFingerprint"}, // 9.0.0+ + {130, nullptr, "GetNintendoAccountUserResourceCache"}, + {131, nullptr, "RefreshNintendoAccountUserResourceCacheAsync"}, + {132, nullptr, "RefreshNintendoAccountUserResourceCacheAsyncIfSecondsElapsed"}, + {133, nullptr, "GetNintendoAccountVerificationUrlCache"}, // 9.0.0+ + {134, nullptr, "RefreshNintendoAccountVerificationUrlCache"}, // 9.0.0+ + {135, nullptr, "RefreshNintendoAccountVerificationUrlCacheAsyncIfSecondsElapsed"}, // 9.0.0+ + {140, nullptr, "GetNetworkServiceLicenseCache"}, // 5.0.0+ + {141, nullptr, "RefreshNetworkServiceLicenseCacheAsync"}, // 5.0.0+ + {142, nullptr, "RefreshNetworkServiceLicenseCacheAsyncIfSecondsElapsed"}, // 5.0.0+ + {150, nullptr, "CreateAuthorizationRequest"}, + }; + // clang-format on + + RegisterHandlers(functions); + } +}; + +// 3.0.0+ +class IFloatingRegistrationRequest final : public ServiceFramework<IFloatingRegistrationRequest> { +public: + explicit IFloatingRegistrationRequest(Common::UUID user_id) + : ServiceFramework("IFloatingRegistrationRequest") { + // clang-format off + static const FunctionInfo functions[] = { + {0, nullptr, "GetSessionId"}, + {12, nullptr, "GetAccountId"}, + {13, nullptr, "GetLinkedNintendoAccountId"}, + {14, nullptr, "GetNickname"}, + {15, nullptr, "GetProfileImage"}, + {21, nullptr, "LoadIdTokenCache"}, + {100, nullptr, "RegisterUser"}, // [1.0.0-3.0.2] RegisterAsync + {101, nullptr, "RegisterUserWithUid"}, // [1.0.0-3.0.2] RegisterWithUidAsync + {102, nullptr, "RegisterNetworkServiceAccountAsync"}, // 4.0.0+ + {103, nullptr, "RegisterNetworkServiceAccountWithUidAsync"}, // 4.0.0+ + {110, nullptr, "SetSystemProgramIdentification"}, + {111, nullptr, "EnsureIdTokenCacheAsync"}, + }; + // clang-format on + + RegisterHandlers(functions); + } +}; + +class IAdministrator final : public ServiceFramework<IAdministrator> { +public: + explicit IAdministrator(Common::UUID user_id) : ServiceFramework("IAdministrator") { + // clang-format off + static const FunctionInfo functions[] = { + {0, nullptr, "CheckAvailability"}, + {1, nullptr, "GetAccountId"}, + {2, nullptr, "EnsureIdTokenCacheAsync"}, + {3, nullptr, "LoadIdTokenCache"}, + {100, nullptr, "SetSystemProgramIdentification"}, + {101, nullptr, "RefreshNotificationTokenAsync"}, // 7.0.0+ + {110, nullptr, "GetServiceEntryRequirementCache"}, // 4.0.0+ + {111, nullptr, "InvalidateServiceEntryRequirementCache"}, // 4.0.0+ + {112, nullptr, "InvalidateTokenCache"}, // 4.0.0 - 6.2.0 + {113, nullptr, "GetServiceEntryRequirementCacheForOnlinePlay"}, // 6.1.0+ + {120, nullptr, "GetNintendoAccountId"}, + {121, nullptr, "CalculateNintendoAccountAuthenticationFingerprint"}, // 9.0.0+ + {130, nullptr, "GetNintendoAccountUserResourceCache"}, + {131, nullptr, "RefreshNintendoAccountUserResourceCacheAsync"}, + {132, nullptr, "RefreshNintendoAccountUserResourceCacheAsyncIfSecondsElapsed"}, + {133, nullptr, "GetNintendoAccountVerificationUrlCache"}, // 9.0.0+ + {134, nullptr, "RefreshNintendoAccountVerificationUrlCacheAsync"}, // 9.0.0+ + {135, nullptr, "RefreshNintendoAccountVerificationUrlCacheAsyncIfSecondsElapsed"}, // 9.0.0+ + {140, nullptr, "GetNetworkServiceLicenseCache"}, // 5.0.0+ + {141, nullptr, "RefreshNetworkServiceLicenseCacheAsync"}, // 5.0.0+ + {142, nullptr, "RefreshNetworkServiceLicenseCacheAsyncIfSecondsElapsed"}, // 5.0.0+ + {150, nullptr, "CreateAuthorizationRequest"}, + {200, nullptr, "IsRegistered"}, + {201, nullptr, "RegisterAsync"}, + {202, nullptr, "UnregisterAsync"}, + {203, nullptr, "DeleteRegistrationInfoLocally"}, + {220, nullptr, "SynchronizeProfileAsync"}, + {221, nullptr, "UploadProfileAsync"}, + {222, nullptr, "SynchronizaProfileAsyncIfSecondsElapsed"}, + {250, nullptr, "IsLinkedWithNintendoAccount"}, + {251, nullptr, "CreateProcedureToLinkWithNintendoAccount"}, + {252, nullptr, "ResumeProcedureToLinkWithNintendoAccount"}, + {255, nullptr, "CreateProcedureToUpdateLinkageStateOfNintendoAccount"}, + {256, nullptr, "ResumeProcedureToUpdateLinkageStateOfNintendoAccount"}, + {260, nullptr, "CreateProcedureToLinkNnidWithNintendoAccount"}, // 3.0.0+ + {261, nullptr, "ResumeProcedureToLinkNnidWithNintendoAccount"}, // 3.0.0+ + {280, nullptr, "ProxyProcedureToAcquireApplicationAuthorizationForNintendoAccount"}, + {290, nullptr, "GetRequestForNintendoAccountUserResourceView"}, // 8.0.0+ + {300, nullptr, "TryRecoverNintendoAccountUserStateAsync"}, // 6.0.0+ + {400, nullptr, "IsServiceEntryRequirementCacheRefreshRequiredForOnlinePlay"}, // 6.1.0+ + {401, nullptr, "RefreshServiceEntryRequirementCacheForOnlinePlayAsync"}, // 6.1.0+ + {900, nullptr, "GetAuthenticationInfoForWin"}, // 9.0.0+ + {901, nullptr, "ImportAsyncForWin"}, // 9.0.0+ + {997, nullptr, "DebugUnlinkNintendoAccountAsync"}, + {998, nullptr, "DebugSetAvailabilityErrorDetail"}, + }; + // clang-format on + + RegisterHandlers(functions); + } +}; + +class IAuthorizationRequest final : public ServiceFramework<IAuthorizationRequest> { +public: + explicit IAuthorizationRequest(Common::UUID user_id) + : ServiceFramework("IAuthorizationRequest") { + // clang-format off + static const FunctionInfo functions[] = { + {0, nullptr, "GetSessionId"}, + {10, nullptr, "InvokeWithoutInteractionAsync"}, + {19, nullptr, "IsAuthorized"}, + {20, nullptr, "GetAuthorizationCode"}, + {21, nullptr, "GetIdToken"}, + {22, nullptr, "GetState"}, + }; + // clang-format on + + RegisterHandlers(functions); + } +}; + +class IOAuthProcedure final : public ServiceFramework<IOAuthProcedure> { +public: + explicit IOAuthProcedure(Common::UUID user_id) : ServiceFramework("IOAuthProcedure") { + // clang-format off + static const FunctionInfo functions[] = { + {0, nullptr, "PrepareAsync"}, + {1, nullptr, "GetRequest"}, + {2, nullptr, "ApplyResponse"}, + {3, nullptr, "ApplyResponseAsync"}, + {10, nullptr, "Suspend"}, + }; + // clang-format on + + RegisterHandlers(functions); + } +}; + +// 3.0.0+ +class IOAuthProcedureForExternalNsa final : public ServiceFramework<IOAuthProcedureForExternalNsa> { +public: + explicit IOAuthProcedureForExternalNsa(Common::UUID user_id) + : ServiceFramework("IOAuthProcedureForExternalNsa") { + // clang-format off + static const FunctionInfo functions[] = { + {0, nullptr, "PrepareAsync"}, + {1, nullptr, "GetRequest"}, + {2, nullptr, "ApplyResponse"}, + {3, nullptr, "ApplyResponseAsync"}, + {10, nullptr, "Suspend"}, + {100, nullptr, "GetAccountId"}, + {101, nullptr, "GetLinkedNintendoAccountId"}, + {102, nullptr, "GetNickname"}, + {103, nullptr, "GetProfileImage"}, + }; + // clang-format on + + RegisterHandlers(functions); + } +}; + +class IOAuthProcedureForNintendoAccountLinkage final + : public ServiceFramework<IOAuthProcedureForNintendoAccountLinkage> { +public: + explicit IOAuthProcedureForNintendoAccountLinkage(Common::UUID user_id) + : ServiceFramework("IOAuthProcedureForNintendoAccountLinkage") { + // clang-format off + static const FunctionInfo functions[] = { + {0, nullptr, "PrepareAsync"}, + {1, nullptr, "GetRequest"}, + {2, nullptr, "ApplyResponse"}, + {3, nullptr, "ApplyResponseAsync"}, + {10, nullptr, "Suspend"}, + {100, nullptr, "GetRequestWithTheme"}, + {101, nullptr, "IsNetworkServiceAccountReplaced"}, + {199, nullptr, "GetUrlForIntroductionOfExtraMembership"}, // 2.0.0 - 5.1.0 + }; + // clang-format on + + RegisterHandlers(functions); + } +}; + +class INotifier final : public ServiceFramework<INotifier> { +public: + explicit INotifier(Common::UUID user_id) : ServiceFramework("INotifier") { + // clang-format off + static const FunctionInfo functions[] = { + {0, nullptr, "GetSystemEvent"}, + }; + // clang-format on + + RegisterHandlers(functions); + } +}; + class IProfileCommon : public ServiceFramework<IProfileCommon> { public: explicit IProfileCommon(const char* name, bool editor_commands, Common::UUID user_id, @@ -226,6 +438,54 @@ public: : IProfileCommon("IProfileEditor", true, user_id, profile_manager) {} }; +class IAsyncContext final : public ServiceFramework<IAsyncContext> { +public: + explicit IAsyncContext(Common::UUID user_id) : ServiceFramework("IAsyncContext") { + // clang-format off + static const FunctionInfo functions[] = { + {0, nullptr, "GetSystemEvent"}, + {1, nullptr, "Cancel"}, + {2, nullptr, "HasDone"}, + {3, nullptr, "GetResult"}, + }; + // clang-format on + + RegisterHandlers(functions); + } +}; + +class ISessionObject final : public ServiceFramework<ISessionObject> { +public: + explicit ISessionObject(Common::UUID user_id) : ServiceFramework("ISessionObject") { + // clang-format off + static const FunctionInfo functions[] = { + {999, nullptr, "Dummy"}, + }; + // clang-format on + + RegisterHandlers(functions); + } +}; + +class IGuestLoginRequest final : public ServiceFramework<IGuestLoginRequest> { +public: + explicit IGuestLoginRequest(Common::UUID) : ServiceFramework("IGuestLoginRequest") { + // clang-format off + static const FunctionInfo functions[] = { + {0, nullptr, "GetSessionId"}, + {11, nullptr, "Unknown"}, // 1.0.0 - 2.3.0 (the name is blank on Switchbrew) + {12, nullptr, "GetAccountId"}, + {13, nullptr, "GetLinkedNintendoAccountId"}, + {14, nullptr, "GetNickname"}, + {15, nullptr, "GetProfileImage"}, + {21, nullptr, "LoadIdTokenCache"}, // 3.0.0+ + }; + // clang-format on + + RegisterHandlers(functions); + } +}; + class IManagerForApplication final : public ServiceFramework<IManagerForApplication> { public: explicit IManagerForApplication(Common::UUID user_id) @@ -265,6 +525,87 @@ private: Common::UUID user_id; }; +// 6.0.0+ +class IAsyncNetworkServiceLicenseKindContext final + : public ServiceFramework<IAsyncNetworkServiceLicenseKindContext> { +public: + explicit IAsyncNetworkServiceLicenseKindContext(Common::UUID user_id) + : ServiceFramework("IAsyncNetworkServiceLicenseKindContext") { + // clang-format off + static const FunctionInfo functions[] = { + {0, nullptr, "GetSystemEvent"}, + {1, nullptr, "Cancel"}, + {2, nullptr, "HasDone"}, + {3, nullptr, "GetResult"}, + {4, nullptr, "GetNetworkServiceLicenseKind"}, + }; + // clang-format on + + RegisterHandlers(functions); + } +}; + +// 8.0.0+ +class IOAuthProcedureForUserRegistration final + : public ServiceFramework<IOAuthProcedureForUserRegistration> { +public: + explicit IOAuthProcedureForUserRegistration(Common::UUID user_id) + : ServiceFramework("IOAuthProcedureForUserRegistration") { + // clang-format off + static const FunctionInfo functions[] = { + {0, nullptr, "PrepareAsync"}, + {1, nullptr, "GetRequest"}, + {2, nullptr, "ApplyResponse"}, + {3, nullptr, "ApplyResponseAsync"}, + {10, nullptr, "Suspend"}, + {100, nullptr, "GetAccountId"}, + {101, nullptr, "GetLinkedNintendoAccountId"}, + {102, nullptr, "GetNickname"}, + {103, nullptr, "GetProfileImage"}, + {110, nullptr, "RegisterUserAsync"}, + {111, nullptr, "GetUid"}, + }; + // clang-format on + + RegisterHandlers(functions); + } +}; + +class DAUTH_O final : public ServiceFramework<DAUTH_O> { +public: + explicit DAUTH_O(Common::UUID) : ServiceFramework("dauth:o") { + // clang-format off + static const FunctionInfo functions[] = { + {0, nullptr, "EnsureAuthenticationTokenCacheAsync"}, // [5.0.0-5.1.0] GeneratePostData + {1, nullptr, "LoadAuthenticationTokenCache"}, // 6.0.0+ + {2, nullptr, "InvalidateAuthenticationTokenCache"}, // 6.0.0+ + {10, nullptr, "EnsureEdgeTokenCacheAsync"}, // 6.0.0+ + {11, nullptr, "LoadEdgeTokenCache"}, // 6.0.0+ + {12, nullptr, "InvalidateEdgeTokenCache"}, // 6.0.0+ + }; + // clang-format on + + RegisterHandlers(functions); + } +}; + +// 6.0.0+ +class IAsyncResult final : public ServiceFramework<IAsyncResult> { +public: + explicit IAsyncResult(Common::UUID user_id) : ServiceFramework("IAsyncResult") { + // clang-format off + static const FunctionInfo functions[] = { + {0, nullptr, "GetResult"}, + {1, nullptr, "Cancel"}, + {2, nullptr, "IsAvailable"}, + {3, nullptr, "GetSystemEvent"}, + }; + // clang-format on + + RegisterHandlers(functions); + } +}; + void Module::Interface::GetUserCount(Kernel::HLERequestContext& ctx) { LOG_DEBUG(Service_ACC, "called"); IPC::ResponseBuilder rb{ctx, 3}; diff --git a/src/core/hle/service/acc/acc_aa.cpp b/src/core/hle/service/acc/acc_aa.cpp index 3bac6bcd1..51f119b12 100644 --- a/src/core/hle/service/acc/acc_aa.cpp +++ b/src/core/hle/service/acc/acc_aa.cpp @@ -13,8 +13,8 @@ ACC_AA::ACC_AA(std::shared_ptr<Module> module, std::shared_ptr<ProfileManager> p {0, nullptr, "EnsureCacheAsync"}, {1, nullptr, "LoadCache"}, {2, nullptr, "GetDeviceAccountId"}, - {50, nullptr, "RegisterNotificationTokenAsync"}, - {51, nullptr, "UnregisterNotificationTokenAsync"}, + {50, nullptr, "RegisterNotificationTokenAsync"}, // 1.0.0 - 6.2.0 + {51, nullptr, "UnregisterNotificationTokenAsync"}, // 1.0.0 - 6.2.0 }; RegisterHandlers(functions); } diff --git a/src/core/hle/service/acc/acc_su.cpp b/src/core/hle/service/acc/acc_su.cpp index 2eefc6df5..85620bde3 100644 --- a/src/core/hle/service/acc/acc_su.cpp +++ b/src/core/hle/service/acc/acc_su.cpp @@ -17,28 +17,28 @@ ACC_SU::ACC_SU(std::shared_ptr<Module> module, std::shared_ptr<ProfileManager> p {3, &ACC_SU::ListOpenUsers, "ListOpenUsers"}, {4, &ACC_SU::GetLastOpenedUser, "GetLastOpenedUser"}, {5, &ACC_SU::GetProfile, "GetProfile"}, - {6, nullptr, "GetProfileDigest"}, + {6, nullptr, "GetProfileDigest"}, // 3.0.0+ {50, &ACC_SU::IsUserRegistrationRequestPermitted, "IsUserRegistrationRequestPermitted"}, {51, &ACC_SU::TrySelectUserWithoutInteraction, "TrySelectUserWithoutInteraction"}, - {60, nullptr, "ListOpenContextStoredUsers"}, - {99, nullptr, "DebugActivateOpenContextRetention"}, + {60, nullptr, "ListOpenContextStoredUsers"}, // 5.0.0 - 5.1.0 + {99, nullptr, "DebugActivateOpenContextRetention"}, // 6.0.0+ {100, nullptr, "GetUserRegistrationNotifier"}, {101, nullptr, "GetUserStateChangeNotifier"}, {102, nullptr, "GetBaasAccountManagerForSystemService"}, {103, nullptr, "GetBaasUserAvailabilityChangeNotifier"}, {104, nullptr, "GetProfileUpdateNotifier"}, - {105, nullptr, "CheckNetworkServiceAvailabilityAsync"}, - {106, nullptr, "GetProfileSyncNotifier"}, + {105, nullptr, "CheckNetworkServiceAvailabilityAsync"}, // 4.0.0+ + {106, nullptr, "GetProfileSyncNotifier"}, // 9.0.0+ {110, nullptr, "StoreSaveDataThumbnail"}, {111, nullptr, "ClearSaveDataThumbnail"}, {112, nullptr, "LoadSaveDataThumbnail"}, - {113, nullptr, "GetSaveDataThumbnailExistence"}, - {120, nullptr, "ListOpenUsersInApplication"}, - {130, nullptr, "ActivateOpenContextRetention"}, - {140, &ACC_SU::ListQualifiedUsers, "ListQualifiedUsers"}, - {150, nullptr, "AuthenticateApplicationAsync"}, - {190, nullptr, "GetUserLastOpenedApplication"}, - {191, nullptr, "ActivateOpenContextHolder"}, + {113, nullptr, "GetSaveDataThumbnailExistence"}, // 5.0.0+ + {120, nullptr, "ListOpenUsersInApplication"}, // 10.0.0+ + {130, nullptr, "ActivateOpenContextRetention"}, // 6.0.0+ + {140, &ACC_SU::ListQualifiedUsers, "ListQualifiedUsers"}, // 6.0.0+ + {150, nullptr, "AuthenticateApplicationAsync"}, // 10.0.0+ + {190, nullptr, "GetUserLastOpenedApplication"}, // 1.0.0 - 9.2.0 + {191, nullptr, "ActivateOpenContextHolder"}, // 7.0.0+ {200, nullptr, "BeginUserRegistration"}, {201, nullptr, "CompleteUserRegistration"}, {202, nullptr, "CancelUserRegistration"}, @@ -46,15 +46,15 @@ ACC_SU::ACC_SU(std::shared_ptr<Module> module, std::shared_ptr<ProfileManager> p {204, nullptr, "SetUserPosition"}, {205, &ACC_SU::GetProfileEditor, "GetProfileEditor"}, {206, nullptr, "CompleteUserRegistrationForcibly"}, - {210, nullptr, "CreateFloatingRegistrationRequest"}, - {211, nullptr, "CreateProcedureToRegisterUserWithNintendoAccount"}, - {212, nullptr, "ResumeProcedureToRegisterUserWithNintendoAccount"}, + {210, nullptr, "CreateFloatingRegistrationRequest"}, // 3.0.0+ + {211, nullptr, "CreateProcedureToRegisterUserWithNintendoAccount"}, // 8.0.0+ + {212, nullptr, "ResumeProcedureToRegisterUserWithNintendoAccount"}, // 8.0.0+ {230, nullptr, "AuthenticateServiceAsync"}, {250, nullptr, "GetBaasAccountAdministrator"}, {290, nullptr, "ProxyProcedureForGuestLoginWithNintendoAccount"}, - {291, nullptr, "ProxyProcedureForFloatingRegistrationWithNintendoAccount"}, + {291, nullptr, "ProxyProcedureForFloatingRegistrationWithNintendoAccount"}, // 3.0.0+ {299, nullptr, "SuspendBackgroundDaemon"}, - {997, nullptr, "DebugInvalidateTokenCacheForUser"}, + {997, nullptr, "DebugInvalidateTokenCacheForUser"}, // 3.0.0+ {998, nullptr, "DebugSetUserStateClose"}, {999, nullptr, "DebugSetUserStateOpen"}, }; diff --git a/src/core/hle/service/acc/acc_u0.cpp b/src/core/hle/service/acc/acc_u0.cpp index fb4e7e772..49f6e20f1 100644 --- a/src/core/hle/service/acc/acc_u0.cpp +++ b/src/core/hle/service/acc/acc_u0.cpp @@ -17,23 +17,23 @@ ACC_U0::ACC_U0(std::shared_ptr<Module> module, std::shared_ptr<ProfileManager> p {3, &ACC_U0::ListOpenUsers, "ListOpenUsers"}, {4, &ACC_U0::GetLastOpenedUser, "GetLastOpenedUser"}, {5, &ACC_U0::GetProfile, "GetProfile"}, - {6, nullptr, "GetProfileDigest"}, + {6, nullptr, "GetProfileDigest"}, // 3.0.0+ {50, &ACC_U0::IsUserRegistrationRequestPermitted, "IsUserRegistrationRequestPermitted"}, {51, &ACC_U0::TrySelectUserWithoutInteraction, "TrySelectUserWithoutInteraction"}, - {60, nullptr, "ListOpenContextStoredUsers"}, - {99, nullptr, "DebugActivateOpenContextRetention"}, + {60, nullptr, "ListOpenContextStoredUsers"}, // 5.0.0 - 5.1.0 + {99, nullptr, "DebugActivateOpenContextRetention"}, // 6.0.0+ {100, &ACC_U0::InitializeApplicationInfo, "InitializeApplicationInfo"}, {101, &ACC_U0::GetBaasAccountManagerForApplication, "GetBaasAccountManagerForApplication"}, {102, nullptr, "AuthenticateApplicationAsync"}, - {103, nullptr, "CheckNetworkServiceAvailabilityAsync"}, + {103, nullptr, "CheckNetworkServiceAvailabilityAsync"}, // 4.0.0+ {110, nullptr, "StoreSaveDataThumbnail"}, {111, nullptr, "ClearSaveDataThumbnail"}, {120, nullptr, "CreateGuestLoginRequest"}, - {130, nullptr, "LoadOpenContext"}, - {131, nullptr, "ListOpenContextStoredUsers"}, - {140, &ACC_U0::InitializeApplicationInfoRestricted, "InitializeApplicationInfoRestricted"}, - {141, &ACC_U0::ListQualifiedUsers, "ListQualifiedUsers"}, - {150, &ACC_U0::IsUserAccountSwitchLocked, "IsUserAccountSwitchLocked"}, + {130, nullptr, "LoadOpenContext"}, // 5.0.0+ + {131, nullptr, "ListOpenContextStoredUsers"}, // 6.0.0+ + {140, &ACC_U0::InitializeApplicationInfoRestricted, "InitializeApplicationInfoRestricted"}, // 6.0.0+ + {141, &ACC_U0::ListQualifiedUsers, "ListQualifiedUsers"}, // 6.0.0+ + {150, &ACC_U0::IsUserAccountSwitchLocked, "IsUserAccountSwitchLocked"}, // 6.0.0+ }; // clang-format on diff --git a/src/core/hle/service/acc/acc_u1.cpp b/src/core/hle/service/acc/acc_u1.cpp index 9f29cdc82..f47004f84 100644 --- a/src/core/hle/service/acc/acc_u1.cpp +++ b/src/core/hle/service/acc/acc_u1.cpp @@ -17,28 +17,29 @@ ACC_U1::ACC_U1(std::shared_ptr<Module> module, std::shared_ptr<ProfileManager> p {3, &ACC_U1::ListOpenUsers, "ListOpenUsers"}, {4, &ACC_U1::GetLastOpenedUser, "GetLastOpenedUser"}, {5, &ACC_U1::GetProfile, "GetProfile"}, - {6, nullptr, "GetProfileDigest"}, + {6, nullptr, "GetProfileDigest"}, // 3.0.0+ {50, &ACC_U1::IsUserRegistrationRequestPermitted, "IsUserRegistrationRequestPermitted"}, {51, &ACC_U1::TrySelectUserWithoutInteraction, "TrySelectUserWithoutInteraction"}, - {60, nullptr, "ListOpenContextStoredUsers"}, - {99, nullptr, "DebugActivateOpenContextRetention"}, + {60, nullptr, "ListOpenContextStoredUsers"}, // 5.0.0 - 5.1.0 + {99, nullptr, "DebugActivateOpenContextRetention"}, // 6.0.0+ {100, nullptr, "GetUserRegistrationNotifier"}, {101, nullptr, "GetUserStateChangeNotifier"}, {102, nullptr, "GetBaasAccountManagerForSystemService"}, - {103, nullptr, "GetProfileUpdateNotifier"}, - {104, nullptr, "CheckNetworkServiceAvailabilityAsync"}, - {105, nullptr, "GetBaasUserAvailabilityChangeNotifier"}, - {106, nullptr, "GetProfileSyncNotifier"}, + {103, nullptr, "GetBaasUserAvailabilityChangeNotifier"}, + {104, nullptr, "GetProfileUpdateNotifier"}, + {105, nullptr, "CheckNetworkServiceAvailabilityAsync"}, // 4.0.0+ + {106, nullptr, "GetProfileSyncNotifier"}, // 9.0.0+ {110, nullptr, "StoreSaveDataThumbnail"}, {111, nullptr, "ClearSaveDataThumbnail"}, {112, nullptr, "LoadSaveDataThumbnail"}, - {113, nullptr, "GetSaveDataThumbnailExistence"}, - {130, nullptr, "ActivateOpenContextRetention"}, - {140, &ACC_U1::ListQualifiedUsers, "ListQualifiedUsers"}, - {150, nullptr, "AuthenticateApplicationAsync"}, - {190, nullptr, "GetUserLastOpenedApplication"}, - {191, nullptr, "ActivateOpenContextHolder"}, - {997, nullptr, "DebugInvalidateTokenCacheForUser"}, + {113, nullptr, "GetSaveDataThumbnailExistence"}, // 5.0.0+ + {120, nullptr, "ListOpenUsersInApplication"}, // 10.0.0+ + {130, nullptr, "ActivateOpenContextRetention"}, // 6.0.0+ + {140, &ACC_U1::ListQualifiedUsers, "ListQualifiedUsers"}, // 6.0.0+ + {150, nullptr, "AuthenticateApplicationAsync"}, // 10.0.0+ + {190, nullptr, "GetUserLastOpenedApplication"}, // 1.0.0 - 9.2.0 + {191, nullptr, "ActivateOpenContextHolder"}, // 7.0.0+ + {997, nullptr, "DebugInvalidateTokenCacheForUser"}, // 3.0.0+ {998, nullptr, "DebugSetUserStateClose"}, {999, nullptr, "DebugSetUserStateOpen"}, }; diff --git a/src/core/hle/service/am/am.cpp b/src/core/hle/service/am/am.cpp index 4df74c4f9..20f366635 100644 --- a/src/core/hle/service/am/am.cpp +++ b/src/core/hle/service/am/am.cpp @@ -68,6 +68,7 @@ IWindowController::IWindowController(Core::System& system_) static const FunctionInfo functions[] = { {0, nullptr, "CreateWindow"}, {1, &IWindowController::GetAppletResourceUserId, "GetAppletResourceUserId"}, + {2, nullptr, "GetAppletResourceUserIdOfCallerApplet"}, {10, &IWindowController::AcquireForegroundRights, "AcquireForegroundRights"}, {11, nullptr, "ReleaseForegroundRights"}, {12, nullptr, "RejectToChangeIntoBackground"}, @@ -189,8 +190,8 @@ IDisplayController::IDisplayController() : ServiceFramework("IDisplayController" {5, nullptr, "GetLastForegroundCaptureImageEx"}, {6, nullptr, "GetLastApplicationCaptureImageEx"}, {7, nullptr, "GetCallerAppletCaptureImageEx"}, - {8, nullptr, "TakeScreenShotOfOwnLayer"}, // 2.0.0+ - {9, nullptr, "CopyBetweenCaptureBuffers"}, // 5.0.0+ + {8, nullptr, "TakeScreenShotOfOwnLayer"}, + {9, nullptr, "CopyBetweenCaptureBuffers"}, {10, nullptr, "AcquireLastApplicationCaptureBuffer"}, {11, nullptr, "ReleaseLastApplicationCaptureBuffer"}, {12, nullptr, "AcquireLastForegroundCaptureBuffer"}, @@ -200,17 +201,14 @@ IDisplayController::IDisplayController() : ServiceFramework("IDisplayController" {16, nullptr, "AcquireLastApplicationCaptureBufferEx"}, {17, nullptr, "AcquireLastForegroundCaptureBufferEx"}, {18, nullptr, "AcquireCallerAppletCaptureBufferEx"}, - // 2.0.0+ {20, nullptr, "ClearCaptureBuffer"}, {21, nullptr, "ClearAppletTransitionBuffer"}, - // 4.0.0+ {22, nullptr, "AcquireLastApplicationCaptureSharedBuffer"}, {23, nullptr, "ReleaseLastApplicationCaptureSharedBuffer"}, {24, nullptr, "AcquireLastForegroundCaptureSharedBuffer"}, {25, nullptr, "ReleaseLastForegroundCaptureSharedBuffer"}, {26, nullptr, "AcquireCallerAppletCaptureSharedBuffer"}, {27, nullptr, "ReleaseCallerAppletCaptureSharedBuffer"}, - // 6.0.0+ {28, nullptr, "TakeScreenShotOfOwnLayerEx"}, }; // clang-format on @@ -225,7 +223,7 @@ IDebugFunctions::IDebugFunctions() : ServiceFramework{"IDebugFunctions"} { static const FunctionInfo functions[] = { {0, nullptr, "NotifyMessageToHomeMenuForDebug"}, {1, nullptr, "OpenMainApplication"}, - {10, nullptr, "EmulateButtonEvent"}, + {10, nullptr, "PerformSystemButtonPressing"}, {20, nullptr, "InvalidateTransitionLayer"}, {30, nullptr, "RequestLaunchApplicationWithUserAndArgumentForDebug"}, {40, nullptr, "GetAppletResourceUsageInfo"}, @@ -267,7 +265,7 @@ ISelfController::ISelfController(Core::System& system, {16, &ISelfController::SetOutOfFocusSuspendingEnabled, "SetOutOfFocusSuspendingEnabled"}, {17, nullptr, "SetControllerFirmwareUpdateSection"}, {18, nullptr, "SetRequiresCaptureButtonShortPressedMessage"}, - {19, &ISelfController::SetScreenShotImageOrientation, "SetScreenShotImageOrientation"}, + {19, &ISelfController::SetAlbumImageOrientation, "SetAlbumImageOrientation"}, {20, nullptr, "SetDesirableKeyboardLayout"}, {40, &ISelfController::CreateManagedDisplayLayer, "CreateManagedDisplayLayer"}, {41, nullptr, "IsSystemBufferSharingEnabled"}, @@ -443,7 +441,7 @@ void ISelfController::SetOutOfFocusSuspendingEnabled(Kernel::HLERequestContext& rb.Push(RESULT_SUCCESS); } -void ISelfController::SetScreenShotImageOrientation(Kernel::HLERequestContext& ctx) { +void ISelfController::SetAlbumImageOrientation(Kernel::HLERequestContext& ctx) { LOG_WARNING(Service_AM, "(STUBBED) called"); IPC::ResponseBuilder rb{ctx, 2}; @@ -607,6 +605,7 @@ ICommonStateGetter::ICommonStateGetter(Core::System& system, {20, nullptr, "PushToGeneralChannel"}, {30, nullptr, "GetHomeButtonReaderLockAccessor"}, {31, nullptr, "GetReaderLockAccessorEx"}, + {32, nullptr, "GetWriterLockAccessorEx"}, {40, nullptr, "GetCradleFwVersion"}, {50, &ICommonStateGetter::IsVrModeEnabled, "IsVrModeEnabled"}, {51, &ICommonStateGetter::SetVrModeEnabled, "SetVrModeEnabled"}, @@ -1132,6 +1131,7 @@ IApplicationFunctions::IApplicationFunctions(Core::System& system_) {24, nullptr, "GetLaunchStorageInfoForDebug"}, {25, &IApplicationFunctions::ExtendSaveData, "ExtendSaveData"}, {26, &IApplicationFunctions::GetSaveDataSize, "GetSaveDataSize"}, + {27, nullptr, "CreateCacheStorage"}, {30, &IApplicationFunctions::BeginBlockingHomeButtonShortAndLongPressed, "BeginBlockingHomeButtonShortAndLongPressed"}, {31, &IApplicationFunctions::EndBlockingHomeButtonShortAndLongPressed, "EndBlockingHomeButtonShortAndLongPressed"}, {32, &IApplicationFunctions::BeginBlockingHomeButton, "BeginBlockingHomeButton"}, @@ -1157,6 +1157,8 @@ IApplicationFunctions::IApplicationFunctions(Core::System& system_) {120, nullptr, "ExecuteProgram"}, {121, nullptr, "ClearUserChannel"}, {122, nullptr, "UnpopToUserChannel"}, + {123, nullptr, "GetPreviousProgramIndex"}, + {124, nullptr, "EnableApplicationAllThreadDumpOnCrash"}, {130, &IApplicationFunctions::GetGpuErrorDetectedSystemEvent, "GetGpuErrorDetectedSystemEvent"}, {140, &IApplicationFunctions::GetFriendInvitationStorageChannelEvent, "GetFriendInvitationStorageChannelEvent"}, {141, nullptr, "TryPopFromFriendInvitationStorageChannel"}, diff --git a/src/core/hle/service/am/am.h b/src/core/hle/service/am/am.h index 469f7f814..2f69466ec 100644 --- a/src/core/hle/service/am/am.h +++ b/src/core/hle/service/am/am.h @@ -138,7 +138,7 @@ private: void SetFocusHandlingMode(Kernel::HLERequestContext& ctx); void SetRestartMessageEnabled(Kernel::HLERequestContext& ctx); void SetOutOfFocusSuspendingEnabled(Kernel::HLERequestContext& ctx); - void SetScreenShotImageOrientation(Kernel::HLERequestContext& ctx); + void SetAlbumImageOrientation(Kernel::HLERequestContext& ctx); void CreateManagedDisplayLayer(Kernel::HLERequestContext& ctx); void SetHandlesRequestToDisplay(Kernel::HLERequestContext& ctx); void SetIdleTimeDetectionExtension(Kernel::HLERequestContext& ctx); diff --git a/src/core/hle/service/am/applets/software_keyboard.cpp b/src/core/hle/service/am/applets/software_keyboard.cpp index 54e63c138..fbe3686ae 100644 --- a/src/core/hle/service/am/applets/software_keyboard.cpp +++ b/src/core/hle/service/am/applets/software_keyboard.cpp @@ -30,7 +30,7 @@ static Core::Frontend::SoftwareKeyboardParameters ConvertToFrontendParameters( config.sub_text.size()); params.guide_text = Common::UTF16StringFromFixedZeroTerminatedBuffer(config.guide_text.data(), config.guide_text.size()); - params.initial_text = initial_text; + params.initial_text = std::move(initial_text); params.max_length = config.length_limit == 0 ? DEFAULT_MAX_LENGTH : config.length_limit; params.password = static_cast<bool>(config.is_password); params.cursor_at_beginning = static_cast<bool>(config.initial_cursor_position); @@ -60,7 +60,7 @@ void SoftwareKeyboard::Initialize() { std::memcpy(&config, keyboard_config.data(), sizeof(KeyboardConfig)); const auto work_buffer_storage = broker.PopNormalDataToApplet(); - ASSERT(work_buffer_storage != nullptr); + ASSERT_OR_EXECUTE(work_buffer_storage != nullptr, { return; }); const auto& work_buffer = work_buffer_storage->GetData(); if (config.initial_string_size == 0) @@ -109,7 +109,7 @@ void SoftwareKeyboard::Execute() { const auto parameters = ConvertToFrontendParameters(config, initial_text); - frontend.RequestText([this](std::optional<std::u16string> text) { WriteText(text); }, + frontend.RequestText([this](std::optional<std::u16string> text) { WriteText(std::move(text)); }, parameters); } diff --git a/src/core/hle/service/am/spsm.cpp b/src/core/hle/service/am/spsm.cpp index 003ee8667..f27729ce7 100644 --- a/src/core/hle/service/am/spsm.cpp +++ b/src/core/hle/service/am/spsm.cpp @@ -10,17 +10,17 @@ SPSM::SPSM() : ServiceFramework{"spsm"} { // clang-format off static const FunctionInfo functions[] = { {0, nullptr, "GetState"}, - {1, nullptr, "SleepSystemAndWaitAwake"}, - {2, nullptr, "Unknown1"}, - {3, nullptr, "Unknown2"}, + {1, nullptr, "EnterSleep"}, + {2, nullptr, "GetLastWakeReason"}, + {3, nullptr, "Shutdown"}, {4, nullptr, "GetNotificationMessageEventHandle"}, - {5, nullptr, "Unknown3"}, - {6, nullptr, "Unknown4"}, - {7, nullptr, "Unknown5"}, + {5, nullptr, "ReceiveNotificationMessage"}, + {6, nullptr, "AnalyzeLogForLastSleepWakeSequence"}, + {7, nullptr, "ResetEventLog"}, {8, nullptr, "AnalyzePerformanceLogForLastSleepWakeSequence"}, {9, nullptr, "ChangeHomeButtonLongPressingTime"}, - {10, nullptr, "Unknown6"}, - {11, nullptr, "Unknown7"}, + {10, nullptr, "PutErrorState"}, + {11, nullptr, "InvalidateCurrentHomeButtonPressing"}, }; // clang-format on diff --git a/src/core/hle/service/aoc/aoc_u.cpp b/src/core/hle/service/aoc/aoc_u.cpp index 4227a4adf..8e79f707b 100644 --- a/src/core/hle/service/aoc/aoc_u.cpp +++ b/src/core/hle/service/aoc/aoc_u.cpp @@ -60,6 +60,7 @@ AOC_U::AOC_U(Core::System& system) {6, nullptr, "PrepareAddOnContentByApplicationId"}, {7, &AOC_U::PrepareAddOnContent, "PrepareAddOnContent"}, {8, &AOC_U::GetAddOnContentListChangedEvent, "GetAddOnContentListChangedEvent"}, + {9, nullptr, "GetAddOnContentLostErrorCode"}, {100, nullptr, "CreateEcPurchasedEventManager"}, {101, nullptr, "CreatePermanentEcPurchasedEventManager"}, }; diff --git a/src/core/hle/service/bcat/backend/boxcat.cpp b/src/core/hle/service/bcat/backend/boxcat.cpp index 5febe8fc1..d29e78d7e 100644 --- a/src/core/hle/service/bcat/backend/boxcat.cpp +++ b/src/core/hle/service/bcat/backend/boxcat.cpp @@ -4,8 +4,8 @@ #include <fmt/ostream.h> #include <httplib.h> -#include <json.hpp> #include <mbedtls/sha256.h> +#include <nlohmann/json.hpp> #include "common/hex_util.h" #include "common/logging/backend.h" #include "common/logging/log.h" diff --git a/src/core/hle/service/bcat/bcat.cpp b/src/core/hle/service/bcat/bcat.cpp index 8bb2528c9..b31766212 100644 --- a/src/core/hle/service/bcat/bcat.cpp +++ b/src/core/hle/service/bcat/bcat.cpp @@ -14,6 +14,8 @@ BCAT::BCAT(Core::System& system, std::shared_ptr<Module> module, {0, &BCAT::CreateBcatService, "CreateBcatService"}, {1, &BCAT::CreateDeliveryCacheStorageService, "CreateDeliveryCacheStorageService"}, {2, &BCAT::CreateDeliveryCacheStorageServiceWithApplicationId, "CreateDeliveryCacheStorageServiceWithApplicationId"}, + {3, nullptr, "CreateDeliveryCacheProgressService"}, + {4, nullptr, "CreateDeliveryCacheProgressServiceWithApplicationId"}, }; // clang-format on RegisterHandlers(functions); diff --git a/src/core/hle/service/bcat/module.cpp b/src/core/hle/service/bcat/module.cpp index 34aba7a27..603b64d4f 100644 --- a/src/core/hle/service/bcat/module.cpp +++ b/src/core/hle/service/bcat/module.cpp @@ -143,10 +143,13 @@ public: {20401, nullptr, "UnregisterSystemApplicationDeliveryTask"}, {20410, nullptr, "SetSystemApplicationDeliveryTaskTimer"}, {30100, &IBcatService::SetPassphrase, "SetPassphrase"}, + {30101, nullptr, "Unknown"}, + {30102, nullptr, "Unknown2"}, {30200, nullptr, "RegisterBackgroundDeliveryTask"}, {30201, nullptr, "UnregisterBackgroundDeliveryTask"}, {30202, nullptr, "BlockDeliveryTask"}, {30203, nullptr, "UnblockDeliveryTask"}, + {30210, nullptr, "SetDeliveryTaskTimer"}, {30300, nullptr, "RegisterSystemApplicationDeliveryTasks"}, {90100, nullptr, "EnumerateBackgroundDeliveryTask"}, {90200, nullptr, "GetDeliveryList"}, diff --git a/src/core/hle/service/bpc/bpc.cpp b/src/core/hle/service/bpc/bpc.cpp index 1c1ecdb60..fac6b2f9c 100644 --- a/src/core/hle/service/bpc/bpc.cpp +++ b/src/core/hle/service/bpc/bpc.cpp @@ -23,9 +23,14 @@ public: {5, nullptr, "GetBoardPowerControlEvent"}, {6, nullptr, "GetSleepButtonState"}, {7, nullptr, "GetPowerEvent"}, - {8, nullptr, "Unknown1"}, - {9, nullptr, "Unknown2"}, - {10, nullptr, "Unknown3"}, + {8, nullptr, "CreateWakeupTimer"}, + {9, nullptr, "CancelWakeupTimer"}, + {10, nullptr, "EnableWakeupTimerOnDevice"}, + {11, nullptr, "CreateWakeupTimerEx"}, + {12, nullptr, "GetLastEnabledWakeupTimerType"}, + {13, nullptr, "CleanAllWakeupTimers"}, + {14, nullptr, "Unknown"}, + {15, nullptr, "Unknown2"}, }; // clang-format on @@ -38,10 +43,11 @@ public: explicit BPC_R() : ServiceFramework{"bpc:r"} { // clang-format off static const FunctionInfo functions[] = { - {0, nullptr, "GetExternalRtcValue"}, - {1, nullptr, "SetExternalRtcValue"}, - {2, nullptr, "ReadExternalRtcResetFlag"}, - {3, nullptr, "ClearExternalRtcResetFlag"}, + {0, nullptr, "GetRtcTime"}, + {1, nullptr, "SetRtcTime"}, + {2, nullptr, "GetRtcResetDetected"}, + {3, nullptr, "ClearRtcResetDetected"}, + {4, nullptr, "SetUpRtcResetOnShutdown"}, }; // clang-format on diff --git a/src/core/hle/service/btdrv/btdrv.cpp b/src/core/hle/service/btdrv/btdrv.cpp index 40a06c9fd..f311afa2f 100644 --- a/src/core/hle/service/btdrv/btdrv.cpp +++ b/src/core/hle/service/btdrv/btdrv.cpp @@ -58,102 +58,103 @@ public: {1, nullptr, "InitializeBluetooth"}, {2, nullptr, "EnableBluetooth"}, {3, nullptr, "DisableBluetooth"}, - {4, nullptr, "CleanupBluetooth"}, + {4, nullptr, "FinalizeBluetooth"}, {5, nullptr, "GetAdapterProperties"}, {6, nullptr, "GetAdapterProperty"}, {7, nullptr, "SetAdapterProperty"}, - {8, nullptr, "StartDiscovery"}, - {9, nullptr, "CancelDiscovery"}, + {8, nullptr, "StartInquiry"}, + {9, nullptr, "StopInquiry"}, {10, nullptr, "CreateBond"}, {11, nullptr, "RemoveBond"}, {12, nullptr, "CancelBond"}, - {13, nullptr, "PinReply"}, - {14, nullptr, "SspReply"}, + {13, nullptr, "RespondToPinRequest"}, + {14, nullptr, "RespondToSspRequest"}, {15, nullptr, "GetEventInfo"}, {16, nullptr, "InitializeHid"}, - {17, nullptr, "HidConnect"}, - {18, nullptr, "HidDisconnect"}, - {19, nullptr, "HidSendData"}, - {20, nullptr, "HidSendData2"}, - {21, nullptr, "HidSetReport"}, - {22, nullptr, "HidGetReport"}, - {23, nullptr, "HidWakeController"}, - {24, nullptr, "HidAddPairedDevice"}, - {25, nullptr, "HidGetPairedDevice"}, - {26, nullptr, "CleanupHid"}, - {27, nullptr, "HidGetEventInfo"}, - {28, nullptr, "ExtSetTsi"}, - {29, nullptr, "ExtSetBurstMode"}, - {30, nullptr, "ExtSetZeroRetran"}, - {31, nullptr, "ExtSetMcMode"}, - {32, nullptr, "ExtStartLlrMode"}, - {33, nullptr, "ExtExitLlrMode"}, - {34, nullptr, "ExtSetRadio"}, - {35, nullptr, "ExtSetVisibility"}, - {36, nullptr, "ExtSetTbfcScan"}, + {17, nullptr, "OpenHidConnection"}, + {18, nullptr, "CloseHidConnection"}, + {19, nullptr, "WriteHidData"}, + {20, nullptr, "WriteHidData2"}, + {21, nullptr, "SetHidReport"}, + {22, nullptr, "GetHidReport"}, + {23, nullptr, "TriggerConnection"}, + {24, nullptr, "AddPairedDeviceInfo"}, + {25, nullptr, "GetPairedDeviceInfo"}, + {26, nullptr, "FinalizeHid"}, + {27, nullptr, "GetHidEventInfo"}, + {28, nullptr, "SetTsi"}, + {29, nullptr, "EnableBurstMode"}, + {30, nullptr, "SetZeroRetransmission"}, + {31, nullptr, "EnableMcMode"}, + {32, nullptr, "EnableLlrScan"}, + {33, nullptr, "DisableLlrScan"}, + {34, nullptr, "EnableRadio"}, + {35, nullptr, "SetVisibility"}, + {36, nullptr, "EnableTbfcScan"}, {37, nullptr, "RegisterHidReportEvent"}, - {38, nullptr, "HidGetReportEventInfo"}, + {38, nullptr, "GetHidReportEventInfo"}, {39, nullptr, "GetLatestPlr"}, - {40, nullptr, "ExtGetPendingConnections"}, + {40, nullptr, "GetPendingConnections"}, {41, nullptr, "GetChannelMap"}, - {42, nullptr, "EnableBluetoothBoostSetting"}, - {43, nullptr, "IsBluetoothBoostSettingEnabled"}, - {44, nullptr, "EnableBluetoothAfhSetting"}, - {45, nullptr, "IsBluetoothAfhSettingEnabled"}, - {46, nullptr, "InitializeBluetoothLe"}, - {47, nullptr, "EnableBluetoothLe"}, - {48, nullptr, "DisableBluetoothLe"}, - {49, nullptr, "CleanupBluetoothLe"}, - {50, nullptr, "SetLeVisibility"}, - {51, nullptr, "SetLeConnectionParameter"}, - {52, nullptr, "SetLeDefaultConnectionParameter"}, - {53, nullptr, "SetLeAdvertiseData"}, - {54, nullptr, "SetLeAdvertiseParameter"}, - {55, nullptr, "StartLeScan"}, - {56, nullptr, "StopLeScan"}, - {57, nullptr, "AddLeScanFilterCondition"}, - {58, nullptr, "DeleteLeScanFilterCondition"}, - {59, nullptr, "DeleteLeScanFilter"}, - {60, nullptr, "ClearLeScanFilters"}, - {61, nullptr, "EnableLeScanFilter"}, - {62, nullptr, "RegisterLeClient"}, - {63, nullptr, "UnregisterLeClient"}, - {64, nullptr, "UnregisterLeClientAll"}, - {65, nullptr, "LeClientConnect"}, - {66, nullptr, "LeClientCancelConnection"}, - {67, nullptr, "LeClientDisconnect"}, - {68, nullptr, "LeClientGetAttributes"}, - {69, nullptr, "LeClientDiscoverService"}, - {70, nullptr, "LeClientConfigureMtu"}, - {71, nullptr, "RegisterLeServer"}, - {72, nullptr, "UnregisterLeServer"}, - {73, nullptr, "LeServerConnect"}, - {74, nullptr, "LeServerDisconnect"}, - {75, nullptr, "CreateLeService"}, - {76, nullptr, "StartLeService"}, - {77, nullptr, "AddLeCharacteristic"}, - {78, nullptr, "AddLeDescriptor"}, - {79, nullptr, "GetLeCoreEventInfo"}, - {80, nullptr, "LeGetFirstCharacteristic"}, - {81, nullptr, "LeGetNextCharacteristic"}, - {82, nullptr, "LeGetFirstDescriptor"}, - {83, nullptr, "LeGetNextDescriptor"}, - {84, nullptr, "RegisterLeCoreDataPath"}, - {85, nullptr, "UnregisterLeCoreDataPath"}, - {86, nullptr, "RegisterLeHidDataPath"}, - {87, nullptr, "UnregisterLeHidDataPath"}, - {88, nullptr, "RegisterLeDataPath"}, - {89, nullptr, "UnregisterLeDataPath"}, - {90, nullptr, "LeClientReadCharacteristic"}, - {91, nullptr, "LeClientReadDescriptor"}, - {92, nullptr, "LeClientWriteCharacteristic"}, - {93, nullptr, "LeClientWriteDescriptor"}, - {94, nullptr, "LeClientRegisterNotification"}, - {95, nullptr, "LeClientDeregisterNotification"}, + {42, nullptr, "EnableTxPowerBoostSetting"}, + {43, nullptr, "IsTxPowerBoostSettingEnabled"}, + {44, nullptr, "EnableAfhSetting"}, + {45, nullptr, "IsAfhSettingEnabled"}, + {46, nullptr, "InitializeBle"}, + {47, nullptr, "EnableBle"}, + {48, nullptr, "DisableBle"}, + {49, nullptr, "FinalizeBle"}, + {50, nullptr, "SetBleVisibility"}, + {51, nullptr, "SetBleConnectionParameter"}, + {52, nullptr, "SetBleDefaultConnectionParameter"}, + {53, nullptr, "SetBleAdvertiseData"}, + {54, nullptr, "SetBleAdvertiseParameter"}, + {55, nullptr, "StartBleScan"}, + {56, nullptr, "StopBleScan"}, + {57, nullptr, "AddBleScanFilterCondition"}, + {58, nullptr, "DeleteBleScanFilterCondition"}, + {59, nullptr, "DeleteBleScanFilter"}, + {60, nullptr, "ClearBleScanFilters"}, + {61, nullptr, "EnableBleScanFilter"}, + {62, nullptr, "RegisterGattClient"}, + {63, nullptr, "UnregisterGattClient"}, + {64, nullptr, "UnregisterAllGattClients"}, + {65, nullptr, "ConnectGattServer"}, + {66, nullptr, "CancelConnectGattServer"}, + {67, nullptr, "DisconnectGattServer"}, + {68, nullptr, "GetGattAttribute"}, + {69, nullptr, "GetGattService"}, + {70, nullptr, "ConfigureAttMtu"}, + {71, nullptr, "RegisterGattServer"}, + {72, nullptr, "UnregisterGattServer"}, + {73, nullptr, "ConnectGattClient"}, + {74, nullptr, "DisconnectGattClient"}, + {75, nullptr, "AddGattService"}, + {76, nullptr, "EnableGattService"}, + {77, nullptr, "AddGattCharacteristic"}, + {78, nullptr, "AddGattDescriptor"}, + {79, nullptr, "GetBleManagedEventInfo"}, + {80, nullptr, "GetGattFirstCharacteristic"}, + {81, nullptr, "GetGattNextCharacteristic"}, + {82, nullptr, "GetGattFirstDescriptor"}, + {83, nullptr, "GetGattNextDescriptor"}, + {84, nullptr, "RegisterGattManagedDataPath"}, + {85, nullptr, "UnregisterGattManagedDataPath"}, + {86, nullptr, "RegisterGattHidDataPath"}, + {87, nullptr, "UnregisterGattHidDataPath"}, + {88, nullptr, "RegisterGattDataPath"}, + {89, nullptr, "UnregisterGattDataPath"}, + {90, nullptr, "ReadGattCharacteristic"}, + {91, nullptr, "ReadGattDescriptor"}, + {92, nullptr, "WriteGattCharacteristic"}, + {93, nullptr, "WriteGattDescriptor"}, + {94, nullptr, "RegisterGattNotification"}, + {95, nullptr, "UnregisterGattNotification"}, {96, nullptr, "GetLeHidEventInfo"}, {97, nullptr, "RegisterBleHidEvent"}, - {98, nullptr, "SetLeScanParameter"}, - {256, nullptr, "GetIsManufacturingMode"}, + {98, nullptr, "SetBleScanParameter"}, + {99, nullptr, "MoveToSecondaryPiconet"}, + {256, nullptr, "IsManufacturingMode"}, {257, nullptr, "EmulateBluetoothCrash"}, {258, nullptr, "GetBleChannelMap"}, }; diff --git a/src/core/hle/service/btm/btm.cpp b/src/core/hle/service/btm/btm.cpp index 251b3c9df..0d251c6d0 100644 --- a/src/core/hle/service/btm/btm.cpp +++ b/src/core/hle/service/btm/btm.cpp @@ -132,66 +132,71 @@ public: explicit BTM() : ServiceFramework{"btm"} { // clang-format off static const FunctionInfo functions[] = { - {0, nullptr, "Unknown1"}, - {1, nullptr, "Unknown2"}, - {2, nullptr, "RegisterSystemEventForConnectedDeviceCondition"}, - {3, nullptr, "Unknown3"}, - {4, nullptr, "Unknown4"}, - {5, nullptr, "Unknown5"}, - {6, nullptr, "Unknown6"}, - {7, nullptr, "Unknown7"}, - {8, nullptr, "RegisterSystemEventForRegisteredDeviceInfo"}, - {9, nullptr, "Unknown8"}, - {10, nullptr, "Unknown9"}, - {11, nullptr, "Unknown10"}, - {12, nullptr, "Unknown11"}, - {13, nullptr, "Unknown12"}, + {0, nullptr, "GetState"}, + {1, nullptr, "GetHostDeviceProperty"}, + {2, nullptr, "AcquireDeviceConditionEvent"}, + {3, nullptr, "GetDeviceCondition"}, + {4, nullptr, "SetBurstMode"}, + {5, nullptr, "SetSlotMode"}, + {6, nullptr, "SetBluetoothMode"}, + {7, nullptr, "SetWlanMode"}, + {8, nullptr, "AcquireDeviceInfoEvent"}, + {9, nullptr, "GetDeviceInfo"}, + {10, nullptr, "AddDeviceInfo"}, + {11, nullptr, "RemoveDeviceInfo"}, + {12, nullptr, "IncreaseDeviceInfoOrder"}, + {13, nullptr, "LlrNotify"}, {14, nullptr, "EnableRadio"}, {15, nullptr, "DisableRadio"}, - {16, nullptr, "Unknown13"}, - {17, nullptr, "Unknown14"}, - {18, nullptr, "Unknown15"}, - {19, nullptr, "Unknown16"}, - {20, nullptr, "Unknown17"}, - {21, nullptr, "Unknown18"}, - {22, nullptr, "Unknown19"}, - {23, nullptr, "Unknown20"}, - {24, nullptr, "Unknown21"}, - {25, nullptr, "Unknown22"}, - {26, nullptr, "Unknown23"}, - {27, nullptr, "Unknown24"}, - {28, nullptr, "Unknown25"}, - {29, nullptr, "Unknown26"}, - {30, nullptr, "Unknown27"}, - {31, nullptr, "Unknown28"}, - {32, nullptr, "Unknown29"}, - {33, nullptr, "Unknown30"}, - {34, nullptr, "Unknown31"}, - {35, nullptr, "Unknown32"}, - {36, nullptr, "Unknown33"}, - {37, nullptr, "Unknown34"}, - {38, nullptr, "Unknown35"}, - {39, nullptr, "Unknown36"}, - {40, nullptr, "Unknown37"}, - {41, nullptr, "Unknown38"}, - {42, nullptr, "Unknown39"}, - {43, nullptr, "Unknown40"}, - {44, nullptr, "Unknown41"}, - {45, nullptr, "Unknown42"}, - {46, nullptr, "Unknown43"}, - {47, nullptr, "Unknown44"}, - {48, nullptr, "Unknown45"}, - {49, nullptr, "Unknown46"}, - {50, nullptr, "Unknown47"}, - {51, nullptr, "Unknown48"}, - {52, nullptr, "Unknown49"}, - {53, nullptr, "Unknown50"}, - {54, nullptr, "Unknown51"}, - {55, nullptr, "Unknown52"}, - {56, nullptr, "Unknown53"}, - {57, nullptr, "Unknown54"}, - {58, nullptr, "Unknown55"}, - {59, nullptr, "Unknown56"}, + {16, nullptr, "HidDisconnect"}, + {17, nullptr, "HidSetRetransmissionMode"}, + {18, nullptr, "AcquireAwakeReqEvent"}, + {19, nullptr, "AcquireLlrStateEvent"}, + {20, nullptr, "IsLlrStarted"}, + {21, nullptr, "EnableSlotSaving"}, + {22, nullptr, "ProtectDeviceInfo"}, + {23, nullptr, "AcquireBleScanEvent"}, + {24, nullptr, "GetBleScanParameterGeneral"}, + {25, nullptr, "GetBleScanParameterSmartDevice"}, + {26, nullptr, "StartBleScanForGeneral"}, + {27, nullptr, "StopBleScanForGeneral"}, + {28, nullptr, "GetBleScanResultsForGeneral"}, + {29, nullptr, "StartBleScanForPairedDevice"}, + {30, nullptr, "StopBleScanForPairedDevice"}, + {31, nullptr, "StartBleScanForSmartDevice"}, + {32, nullptr, "StopBleScanForSmartDevice"}, + {33, nullptr, "GetBleScanResultsForSmartDevice"}, + {34, nullptr, "AcquireBleConnectionEvent"}, + {35, nullptr, "BleConnect"}, + {36, nullptr, "BleOverrideConnection"}, + {37, nullptr, "BleDisconnect"}, + {38, nullptr, "BleGetConnectionState"}, + {39, nullptr, "BleGetGattClientConditionList"}, + {40, nullptr, "AcquireBlePairingEvent"}, + {41, nullptr, "BlePairDevice"}, + {42, nullptr, "BleUnpairDeviceOnBoth"}, + {43, nullptr, "BleUnpairDevice"}, + {44, nullptr, "BleGetPairedAddresses"}, + {45, nullptr, "AcquireBleServiceDiscoveryEvent"}, + {46, nullptr, "GetGattServices"}, + {47, nullptr, "GetGattService"}, + {48, nullptr, "GetGattIncludedServices"}, + {49, nullptr, "GetBelongingService"}, + {50, nullptr, "GetGattCharacteristics"}, + {51, nullptr, "GetGattDescriptors"}, + {52, nullptr, "AcquireBleMtuConfigEvent"}, + {53, nullptr, "ConfigureBleMtu"}, + {54, nullptr, "GetBleMtu"}, + {55, nullptr, "RegisterBleGattDataPath"}, + {56, nullptr, "UnregisterBleGattDataPath"}, + {57, nullptr, "RegisterAppletResourceUserId"}, + {58, nullptr, "UnregisterAppletResourceUserId"}, + {59, nullptr, "SetAppletResourceUserId"}, + {60, nullptr, "Unknown60"}, + {61, nullptr, "Unknown61"}, + {62, nullptr, "Unknown62"}, + {63, nullptr, "Unknown63"}, + {64, nullptr, "Unknown64"}, }; // clang-format on @@ -204,19 +209,19 @@ public: explicit BTM_DBG() : ServiceFramework{"btm:dbg"} { // clang-format off static const FunctionInfo functions[] = { - {0, nullptr, "RegisterSystemEventForDiscovery"}, - {1, nullptr, "Unknown1"}, - {2, nullptr, "Unknown2"}, - {3, nullptr, "Unknown3"}, - {4, nullptr, "Unknown4"}, - {5, nullptr, "Unknown5"}, - {6, nullptr, "Unknown6"}, - {7, nullptr, "Unknown7"}, - {8, nullptr, "Unknown8"}, - {9, nullptr, "Unknown9"}, - {10, nullptr, "Unknown10"}, - {11, nullptr, "Unknown11"}, - {12, nullptr, "Unknown11"}, + {0, nullptr, "AcquireDiscoveryEvent"}, + {1, nullptr, "StartDiscovery"}, + {2, nullptr, "CancelDiscovery"}, + {3, nullptr, "GetDeviceProperty"}, + {4, nullptr, "CreateBond"}, + {5, nullptr, "CancelBond"}, + {6, nullptr, "SetTsiMode"}, + {7, nullptr, "GeneralTest"}, + {8, nullptr, "HidConnect"}, + {9, nullptr, "GeneralGet"}, + {10, nullptr, "GetGattClientDisconnectionReason"}, + {11, nullptr, "GetBleConnectionParameter"}, + {12, nullptr, "GetBleConnectionParameterRequest"}, }; // clang-format on diff --git a/src/core/hle/service/caps/caps.cpp b/src/core/hle/service/caps/caps.cpp index 26c8a7081..ba5749b84 100644 --- a/src/core/hle/service/caps/caps.cpp +++ b/src/core/hle/service/caps/caps.cpp @@ -1,4 +1,4 @@ -// Copyright 2018 yuzu emulator team +// Copyright 2018 yuzu Emulator Project // Licensed under GPLv2 or any later version // Refer to the license.txt file included. diff --git a/src/core/hle/service/caps/caps.h b/src/core/hle/service/caps/caps.h index fc70a4c27..b8c67b6e2 100644 --- a/src/core/hle/service/caps/caps.h +++ b/src/core/hle/service/caps/caps.h @@ -1,4 +1,4 @@ -// Copyright 2018 yuzu emulator team +// Copyright 2018 yuzu Emulator Project // Licensed under GPLv2 or any later version // Refer to the license.txt file included. @@ -12,73 +12,79 @@ class ServiceManager; namespace Service::Capture { -enum AlbumImageOrientation { +enum class AlbumImageOrientation { Orientation0 = 0, Orientation1 = 1, Orientation2 = 2, Orientation3 = 3, }; -enum AlbumReportOption { +enum class AlbumReportOption { Disable = 0, Enable = 1, }; -enum ContentType : u8 { +enum class ContentType : u8 { Screenshot = 0, Movie = 1, ExtraMovie = 3, }; -enum AlbumStorage : u8 { +enum class AlbumStorage : u8 { NAND = 0, SD = 1, }; struct AlbumFileDateTime { - u16 year; - u8 month; - u8 day; - u8 hour; - u8 minute; - u8 second; - u8 uid; + s16 year{}; + s8 month{}; + s8 day{}; + s8 hour{}; + s8 minute{}; + s8 second{}; + s8 uid{}; }; +static_assert(sizeof(AlbumFileDateTime) == 0x8, "AlbumFileDateTime has incorrect size."); struct AlbumEntry { - u64 size; - u64 application_id; - AlbumFileDateTime datetime; - AlbumStorage storage; - ContentType content; - u8 padding[6]; + u64 size{}; + u64 application_id{}; + AlbumFileDateTime datetime{}; + AlbumStorage storage{}; + ContentType content{}; + INSERT_PADDING_BYTES(6); }; +static_assert(sizeof(AlbumEntry) == 0x20, "AlbumEntry has incorrect size."); struct AlbumFileEntry { - u64 size; - u64 hash; - AlbumFileDateTime datetime; - AlbumStorage storage; - ContentType content; - u8 padding[5]; - u8 unknown; + u64 size{}; // Size of the entry + u64 hash{}; // AES256 with hardcoded key over AlbumEntry + AlbumFileDateTime datetime{}; + AlbumStorage storage{}; + ContentType content{}; + INSERT_PADDING_BYTES(5); + u8 unknown{1}; // Set to 1 on official SW }; +static_assert(sizeof(AlbumFileEntry) == 0x20, "AlbumFileEntry has incorrect size."); struct ApplicationAlbumEntry { - u64 size; - u64 hash; - AlbumFileDateTime datetime; - AlbumStorage storage; - ContentType content; - u8 padding[5]; - u8 unknown; + u64 size{}; // Size of the entry + u64 hash{}; // AES256 with hardcoded key over AlbumEntry + AlbumFileDateTime datetime{}; + AlbumStorage storage{}; + ContentType content{}; + INSERT_PADDING_BYTES(5); + u8 unknown{1}; // Set to 1 on official SW }; +static_assert(sizeof(ApplicationAlbumEntry) == 0x20, "ApplicationAlbumEntry has incorrect size."); struct ApplicationAlbumFileEntry { - ApplicationAlbumEntry entry; - AlbumFileDateTime datetime; - u64 unknown; + ApplicationAlbumEntry entry{}; + AlbumFileDateTime datetime{}; + u64 unknown{}; }; +static_assert(sizeof(ApplicationAlbumFileEntry) == 0x30, + "ApplicationAlbumFileEntry has incorrect size."); /// Registers all Capture services with the specified service manager. void InstallInterfaces(SM::ServiceManager& sm); diff --git a/src/core/hle/service/caps/caps_a.cpp b/src/core/hle/service/caps/caps_a.cpp index 88a3fdc05..a0a3b2ae3 100644 --- a/src/core/hle/service/caps/caps_a.cpp +++ b/src/core/hle/service/caps/caps_a.cpp @@ -1,4 +1,4 @@ -// Copyright 2020 yuzu emulator team +// Copyright 2020 yuzu Emulator Project // Licensed under GPLv2 or any later version // Refer to the license.txt file included. diff --git a/src/core/hle/service/caps/caps_a.h b/src/core/hle/service/caps/caps_a.h index 8de832491..cb93aad5b 100644 --- a/src/core/hle/service/caps/caps_a.h +++ b/src/core/hle/service/caps/caps_a.h @@ -1,4 +1,4 @@ -// Copyright 2020 yuzu emulator team +// Copyright 2020 yuzu Emulator Project // Licensed under GPLv2 or any later version // Refer to the license.txt file included. diff --git a/src/core/hle/service/caps/caps_c.cpp b/src/core/hle/service/caps/caps_c.cpp index ea6452ffa..ab17a187e 100644 --- a/src/core/hle/service/caps/caps_c.cpp +++ b/src/core/hle/service/caps/caps_c.cpp @@ -1,4 +1,4 @@ -// Copyright 2020 yuzu emulator team +// Copyright 2020 yuzu Emulator Project // Licensed under GPLv2 or any later version // Refer to the license.txt file included. diff --git a/src/core/hle/service/caps/caps_c.h b/src/core/hle/service/caps/caps_c.h index d07cdb441..a9d028689 100644 --- a/src/core/hle/service/caps/caps_c.h +++ b/src/core/hle/service/caps/caps_c.h @@ -1,4 +1,4 @@ -// Copyright 2020 yuzu emulator team +// Copyright 2020 yuzu Emulator Project // Licensed under GPLv2 or any later version // Refer to the license.txt file included. diff --git a/src/core/hle/service/caps/caps_sc.cpp b/src/core/hle/service/caps/caps_sc.cpp index d01a8a58e..822ee96c8 100644 --- a/src/core/hle/service/caps/caps_sc.cpp +++ b/src/core/hle/service/caps/caps_sc.cpp @@ -1,4 +1,4 @@ -// Copyright 2020 yuzu emulator team +// Copyright 2020 yuzu Emulator Project // Licensed under GPLv2 or any later version // Refer to the license.txt file included. diff --git a/src/core/hle/service/caps/caps_sc.h b/src/core/hle/service/caps/caps_sc.h index 9ba372f7a..ac3e929ca 100644 --- a/src/core/hle/service/caps/caps_sc.h +++ b/src/core/hle/service/caps/caps_sc.h @@ -1,4 +1,4 @@ -// Copyright 2020 yuzu emulator team +// Copyright 2020 yuzu Emulator Project // Licensed under GPLv2 or any later version // Refer to the license.txt file included. diff --git a/src/core/hle/service/caps/caps_ss.cpp b/src/core/hle/service/caps/caps_ss.cpp index eaa3a7494..24dc716e7 100644 --- a/src/core/hle/service/caps/caps_ss.cpp +++ b/src/core/hle/service/caps/caps_ss.cpp @@ -1,4 +1,4 @@ -// Copyright 2020 yuzu emulator team +// Copyright 2020 yuzu Emulator Project // Licensed under GPLv2 or any later version // Refer to the license.txt file included. diff --git a/src/core/hle/service/caps/caps_ss.h b/src/core/hle/service/caps/caps_ss.h index e258a6925..450686e4f 100644 --- a/src/core/hle/service/caps/caps_ss.h +++ b/src/core/hle/service/caps/caps_ss.h @@ -1,4 +1,4 @@ -// Copyright 2020 yuzu emulator team +// Copyright 2020 yuzu Emulator Project // Licensed under GPLv2 or any later version // Refer to the license.txt file included. diff --git a/src/core/hle/service/caps/caps_su.cpp b/src/core/hle/service/caps/caps_su.cpp index e8b0698e8..fffb2ecf9 100644 --- a/src/core/hle/service/caps/caps_su.cpp +++ b/src/core/hle/service/caps/caps_su.cpp @@ -1,4 +1,4 @@ -// Copyright 2020 yuzu emulator team +// Copyright 2020 yuzu Emulator Project // Licensed under GPLv2 or any later version // Refer to the license.txt file included. diff --git a/src/core/hle/service/caps/caps_su.h b/src/core/hle/service/caps/caps_su.h index c494d7c84..62c9603a9 100644 --- a/src/core/hle/service/caps/caps_su.h +++ b/src/core/hle/service/caps/caps_su.h @@ -1,4 +1,4 @@ -// Copyright 2020 yuzu emulator team +// Copyright 2020 yuzu Emulator Project // Licensed under GPLv2 or any later version // Refer to the license.txt file included. diff --git a/src/core/hle/service/caps/caps_u.cpp b/src/core/hle/service/caps/caps_u.cpp index 78bab6ed8..f36d8de2d 100644 --- a/src/core/hle/service/caps/caps_u.cpp +++ b/src/core/hle/service/caps/caps_u.cpp @@ -1,4 +1,4 @@ -// Copyright 2020 yuzu emulator team +// Copyright 2020 yuzu Emulator Project // Licensed under GPLv2 or any later version // Refer to the license.txt file included. @@ -58,19 +58,25 @@ void CAPS_U::GetAlbumContentsFileListForApplication(Kernel::HLERequestContext& c // u8 ContentType, two s64s, and an u64 AppletResourceUserId. Returns an output u64 for total // output entries (which is copied to a s32 by official SW). IPC::RequestParser rp{ctx}; - [[maybe_unused]] const auto application_album_file_entries = rp.PopRaw<std::array<u8, 0x30>>(); - const auto pid = rp.Pop<s32>(); - const auto content_type = rp.PopRaw<ContentType>(); - [[maybe_unused]] const auto start_datetime = rp.PopRaw<AlbumFileDateTime>(); - [[maybe_unused]] const auto end_datetime = rp.PopRaw<AlbumFileDateTime>(); - const auto applet_resource_user_id = rp.Pop<u64>(); + const auto pid{rp.Pop<s32>()}; + const auto content_type{rp.PopEnum<ContentType>()}; + const auto start_posix_time{rp.Pop<s64>()}; + const auto end_posix_time{rp.Pop<s64>()}; + const auto applet_resource_user_id{rp.Pop<u64>()}; + + // TODO: Update this when we implement the album. + // Currently we do not have a method of accessing album entries, set this to 0 for now. + constexpr s32 total_entries{0}; + LOG_WARNING(Service_Capture, - "(STUBBED) called. pid={}, content_type={}, applet_resource_user_id={}", pid, - content_type, applet_resource_user_id); + "(STUBBED) called. pid={}, content_type={}, start_posix_time={}, " + "end_posix_time={}, applet_resource_user_id={}, total_entries={}", + pid, content_type, start_posix_time, end_posix_time, applet_resource_user_id, + total_entries); IPC::ResponseBuilder rb{ctx, 3}; rb.Push(RESULT_SUCCESS); - rb.Push<s32>(0); + rb.Push(total_entries); } } // namespace Service::Capture diff --git a/src/core/hle/service/caps/caps_u.h b/src/core/hle/service/caps/caps_u.h index e6e0716ff..689364de4 100644 --- a/src/core/hle/service/caps/caps_u.h +++ b/src/core/hle/service/caps/caps_u.h @@ -1,4 +1,4 @@ -// Copyright 2020 yuzu emulator team +// Copyright 2020 yuzu Emulator Project // Licensed under GPLv2 or any later version // Refer to the license.txt file included. diff --git a/src/core/hle/service/es/es.cpp b/src/core/hle/service/es/es.cpp index ad6841a64..da6b74a22 100644 --- a/src/core/hle/service/es/es.cpp +++ b/src/core/hle/service/es/es.cpp @@ -27,8 +27,8 @@ public: {8, &ETicket::GetTitleKey, "GetTitleKey"}, {9, &ETicket::CountCommonTicket, "CountCommonTicket"}, {10, &ETicket::CountPersonalizedTicket, "CountPersonalizedTicket"}, - {11, &ETicket::ListCommonTicket, "ListCommonTicket"}, - {12, &ETicket::ListPersonalizedTicket, "ListPersonalizedTicket"}, + {11, &ETicket::ListCommonTicketRightsIds, "ListCommonTicketRightsIds"}, + {12, &ETicket::ListPersonalizedTicketRightsIds, "ListPersonalizedTicketRightsIds"}, {13, nullptr, "ListMissingPersonalizedTicket"}, {14, &ETicket::GetCommonTicketSize, "GetCommonTicketSize"}, {15, &ETicket::GetPersonalizedTicketSize, "GetPersonalizedTicketSize"}, @@ -55,7 +55,46 @@ public: {36, nullptr, "DeleteAllInactiveELicenseRequiredPersonalizedTicket"}, {37, nullptr, "OwnTicket2"}, {38, nullptr, "OwnTicket3"}, + {501, nullptr, "Unknown501"}, + {502, nullptr, "Unknown502"}, {503, nullptr, "GetTitleKey"}, + {504, nullptr, "Unknown504"}, + {508, nullptr, "Unknown508"}, + {509, nullptr, "Unknown509"}, + {510, nullptr, "Unknown510"}, + {511, nullptr, "Unknown511"}, + {1001, nullptr, "Unknown1001"}, + {1002, nullptr, "Unknown1001"}, + {1003, nullptr, "Unknown1003"}, + {1004, nullptr, "Unknown1004"}, + {1005, nullptr, "Unknown1005"}, + {1006, nullptr, "Unknown1006"}, + {1007, nullptr, "Unknown1007"}, + {1009, nullptr, "Unknown1009"}, + {1010, nullptr, "Unknown1010"}, + {1011, nullptr, "Unknown1011"}, + {1012, nullptr, "Unknown1012"}, + {1013, nullptr, "Unknown1013"}, + {1014, nullptr, "Unknown1014"}, + {1015, nullptr, "Unknown1015"}, + {1016, nullptr, "Unknown1016"}, + {1017, nullptr, "Unknown1017"}, + {1018, nullptr, "Unknown1018"}, + {1019, nullptr, "Unknown1019"}, + {1020, nullptr, "Unknown1020"}, + {1021, nullptr, "Unknown1021"}, + {1501, nullptr, "Unknown1501"}, + {1502, nullptr, "Unknown1502"}, + {1503, nullptr, "Unknown1503"}, + {1504, nullptr, "Unknown1504"}, + {1505, nullptr, "Unknown1505"}, + {2000, nullptr, "Unknown2000"}, + {2001, nullptr, "Unknown2001"}, + {2100, nullptr, "Unknown2100"}, + {2501, nullptr, "Unknown2501"}, + {2502, nullptr, "Unknown2502"}, + {3001, nullptr, "Unknown3001"}, + {3002, nullptr, "Unknown3002"}, }; // clang-format on RegisterHandlers(functions); @@ -147,7 +186,7 @@ private: rb.Push<u32>(count); } - void ListCommonTicket(Kernel::HLERequestContext& ctx) { + void ListCommonTicketRightsIds(Kernel::HLERequestContext& ctx) { u32 out_entries; if (keys.GetCommonTickets().empty()) out_entries = 0; @@ -170,7 +209,7 @@ private: rb.Push<u32>(out_entries); } - void ListPersonalizedTicket(Kernel::HLERequestContext& ctx) { + void ListPersonalizedTicketRightsIds(Kernel::HLERequestContext& ctx) { u32 out_entries; if (keys.GetPersonalizedTickets().empty()) out_entries = 0; diff --git a/src/core/hle/service/eupld/eupld.cpp b/src/core/hle/service/eupld/eupld.cpp index 2df30acee..0d6d244f4 100644 --- a/src/core/hle/service/eupld/eupld.cpp +++ b/src/core/hle/service/eupld/eupld.cpp @@ -19,6 +19,7 @@ public: {1, nullptr, "ImportCrt"}, {2, nullptr, "ImportPki"}, {3, nullptr, "SetAutoUpload"}, + {4, nullptr, "GetAutoUpload"}, }; // clang-format on diff --git a/src/core/hle/service/filesystem/fsp_srv.cpp b/src/core/hle/service/filesystem/fsp_srv.cpp index f6503fe2f..20c331b77 100644 --- a/src/core/hle/service/filesystem/fsp_srv.cpp +++ b/src/core/hle/service/filesystem/fsp_srv.cpp @@ -767,7 +767,7 @@ FSP_SRV::FSP_SRV(FileSystemController& fsc, const Core::Reporter& reporter) {1014, nullptr, "OutputMultiProgramTagAccessLog"}, {1100, nullptr, "OverrideSaveDataTransferTokenSignVerificationKey"}, {1110, nullptr, "CorruptSaveDataFileSystemBySaveDataSpaceId2"}, - {1200, nullptr, "OpenMultiCommitManager"}, + {1200, &FSP_SRV::OpenMultiCommitManager, "OpenMultiCommitManager"}, {1300, nullptr, "OpenBisWiper"}, }; // clang-format on @@ -988,4 +988,40 @@ void FSP_SRV::GetAccessLogVersionInfo(Kernel::HLERequestContext& ctx) { rb.Push(access_log_program_index); } +class IMultiCommitManager final : public ServiceFramework<IMultiCommitManager> { +public: + explicit IMultiCommitManager() : ServiceFramework("IMultiCommitManager") { + static const FunctionInfo functions[] = { + {1, &IMultiCommitManager::Add, "Add"}, + {2, &IMultiCommitManager::Commit, "Commit"}, + }; + RegisterHandlers(functions); + } + +private: + FileSys::VirtualFile backend; + + void Add(Kernel::HLERequestContext& ctx) { + LOG_WARNING(Service_FS, "(STUBBED) called"); + + IPC::ResponseBuilder rb{ctx, 2}; + rb.Push(RESULT_SUCCESS); + } + + void Commit(Kernel::HLERequestContext& ctx) { + LOG_WARNING(Service_FS, "(STUBBED) called"); + + IPC::ResponseBuilder rb{ctx, 2}; + rb.Push(RESULT_SUCCESS); + } +}; + +void FSP_SRV::OpenMultiCommitManager(Kernel::HLERequestContext& ctx) { + LOG_DEBUG(Service_FS, "called"); + + IPC::ResponseBuilder rb{ctx, 2, 0, 1}; + rb.Push(RESULT_SUCCESS); + rb.PushIpcInterface<IMultiCommitManager>(std::make_shared<IMultiCommitManager>()); +} + } // namespace Service::FileSystem diff --git a/src/core/hle/service/filesystem/fsp_srv.h b/src/core/hle/service/filesystem/fsp_srv.h index d52b55999..dfb3e395b 100644 --- a/src/core/hle/service/filesystem/fsp_srv.h +++ b/src/core/hle/service/filesystem/fsp_srv.h @@ -50,6 +50,7 @@ private: void OpenPatchDataStorageByCurrentProcess(Kernel::HLERequestContext& ctx); void OutputAccessLogToSdCard(Kernel::HLERequestContext& ctx); void GetAccessLogVersionInfo(Kernel::HLERequestContext& ctx); + void OpenMultiCommitManager(Kernel::HLERequestContext& ctx); FileSystemController& fsc; diff --git a/src/core/hle/service/friend/friend.cpp b/src/core/hle/service/friend/friend.cpp index 68f259b70..b7adaffc7 100644 --- a/src/core/hle/service/friend/friend.cpp +++ b/src/core/hle/service/friend/friend.cpp @@ -25,9 +25,13 @@ public: {10101, &IFriendService::GetFriendList, "GetFriendList"}, {10102, nullptr, "UpdateFriendInfo"}, {10110, nullptr, "GetFriendProfileImage"}, + {10120, nullptr, "Unknown10120"}, + {10121, nullptr, "Unknown10121"}, {10200, nullptr, "SendFriendRequestForApplication"}, {10211, nullptr, "AddFacedFriendRequestForApplication"}, {10400, &IFriendService::GetBlockedUserListIds, "GetBlockedUserListIds"}, + {10420, nullptr, "Unknown10420"}, + {10421, nullptr, "Unknown10421"}, {10500, nullptr, "GetProfileList"}, {10600, nullptr, "DeclareOpenOnlinePlaySession"}, {10601, &IFriendService::DeclareCloseOnlinePlaySession, "DeclareCloseOnlinePlaySession"}, @@ -97,6 +101,8 @@ public: {30900, nullptr, "SendFriendInvitation"}, {30910, nullptr, "ReadFriendInvitation"}, {30911, nullptr, "ReadAllFriendInvitations"}, + {40100, nullptr, "Unknown40100"}, + {40400, nullptr, "Unknown40400"}, {49900, nullptr, "DeleteNetworkServiceAccountCache"}, }; // clang-format on diff --git a/src/core/hle/service/grc/grc.cpp b/src/core/hle/service/grc/grc.cpp index 24910ac6c..401e0b208 100644 --- a/src/core/hle/service/grc/grc.cpp +++ b/src/core/hle/service/grc/grc.cpp @@ -17,6 +17,9 @@ public: static const FunctionInfo functions[] = { {1, nullptr, "OpenContinuousRecorder"}, {2, nullptr, "OpenGameMovieTrimmer"}, + {3, nullptr, "OpenOffscreenRecorder"}, + {101, nullptr, "CreateMovieMaker"}, + {9903, nullptr, "SetOffscreenRecordingMarker"} }; // clang-format on diff --git a/src/core/hle/service/hid/controllers/debug_pad.cpp b/src/core/hle/service/hid/controllers/debug_pad.cpp index 1f2131ec8..cb35919e9 100644 --- a/src/core/hle/service/hid/controllers/debug_pad.cpp +++ b/src/core/hle/service/hid/controllers/debug_pad.cpp @@ -23,7 +23,7 @@ void Controller_DebugPad::OnRelease() {} void Controller_DebugPad::OnUpdate(const Core::Timing::CoreTiming& core_timing, u8* data, std::size_t size) { - shared_memory.header.timestamp = core_timing.GetTicks(); + shared_memory.header.timestamp = core_timing.GetCPUTicks(); shared_memory.header.total_entry_count = 17; if (!IsControllerActivated()) { diff --git a/src/core/hle/service/hid/controllers/gesture.cpp b/src/core/hle/service/hid/controllers/gesture.cpp index 6e990dd00..b7b7bfeae 100644 --- a/src/core/hle/service/hid/controllers/gesture.cpp +++ b/src/core/hle/service/hid/controllers/gesture.cpp @@ -19,7 +19,7 @@ void Controller_Gesture::OnRelease() {} void Controller_Gesture::OnUpdate(const Core::Timing::CoreTiming& core_timing, u8* data, std::size_t size) { - shared_memory.header.timestamp = core_timing.GetTicks(); + shared_memory.header.timestamp = core_timing.GetCPUTicks(); shared_memory.header.total_entry_count = 17; if (!IsControllerActivated()) { diff --git a/src/core/hle/service/hid/controllers/keyboard.cpp b/src/core/hle/service/hid/controllers/keyboard.cpp index 358cb9329..feae89525 100644 --- a/src/core/hle/service/hid/controllers/keyboard.cpp +++ b/src/core/hle/service/hid/controllers/keyboard.cpp @@ -21,7 +21,7 @@ void Controller_Keyboard::OnRelease() {} void Controller_Keyboard::OnUpdate(const Core::Timing::CoreTiming& core_timing, u8* data, std::size_t size) { - shared_memory.header.timestamp = core_timing.GetTicks(); + shared_memory.header.timestamp = core_timing.GetCPUTicks(); shared_memory.header.total_entry_count = 17; if (!IsControllerActivated()) { @@ -38,10 +38,11 @@ void Controller_Keyboard::OnUpdate(const Core::Timing::CoreTiming& core_timing, cur_entry.sampling_number = last_entry.sampling_number + 1; cur_entry.sampling_number2 = cur_entry.sampling_number; + cur_entry.key.fill(0); + cur_entry.modifier = 0; + for (std::size_t i = 0; i < keyboard_keys.size(); ++i) { - for (std::size_t k = 0; k < KEYS_PER_BYTE; ++k) { - cur_entry.key[i / KEYS_PER_BYTE] |= (keyboard_keys[i]->GetStatus() << k); - } + cur_entry.key[i / KEYS_PER_BYTE] |= (keyboard_keys[i]->GetStatus() << (i % KEYS_PER_BYTE)); } for (std::size_t i = 0; i < keyboard_mods.size(); ++i) { diff --git a/src/core/hle/service/hid/controllers/mouse.cpp b/src/core/hle/service/hid/controllers/mouse.cpp index 93d88ea50..ac40989c5 100644 --- a/src/core/hle/service/hid/controllers/mouse.cpp +++ b/src/core/hle/service/hid/controllers/mouse.cpp @@ -19,7 +19,7 @@ void Controller_Mouse::OnRelease() {} void Controller_Mouse::OnUpdate(const Core::Timing::CoreTiming& core_timing, u8* data, std::size_t size) { - shared_memory.header.timestamp = core_timing.GetTicks(); + shared_memory.header.timestamp = core_timing.GetCPUTicks(); shared_memory.header.total_entry_count = 17; if (!IsControllerActivated()) { diff --git a/src/core/hle/service/hid/controllers/npad.cpp b/src/core/hle/service/hid/controllers/npad.cpp index c55d900e2..ef67ad690 100644 --- a/src/core/hle/service/hid/controllers/npad.cpp +++ b/src/core/hle/service/hid/controllers/npad.cpp @@ -328,7 +328,7 @@ void Controller_NPad::OnUpdate(const Core::Timing::CoreTiming& core_timing, u8* const auto& last_entry = main_controller->npad[main_controller->common.last_entry_index]; - main_controller->common.timestamp = core_timing.GetTicks(); + main_controller->common.timestamp = core_timing.GetCPUTicks(); main_controller->common.last_entry_index = (main_controller->common.last_entry_index + 1) % 17; @@ -566,6 +566,14 @@ void Controller_NPad::DisconnectNPad(u32 npad_id) { connected_controllers[NPadIdToIndex(npad_id)].is_connected = false; } +void Controller_NPad::SetGyroscopeZeroDriftMode(GyroscopeZeroDriftMode drift_mode) { + gyroscope_zero_drift_mode = drift_mode; +} + +Controller_NPad::GyroscopeZeroDriftMode Controller_NPad::GetGyroscopeZeroDriftMode() const { + return gyroscope_zero_drift_mode; +} + void Controller_NPad::StartLRAssignmentMode() { // Nothing internally is used for lr assignment mode. Since we have the ability to set the // controller types from boot, it doesn't really matter about showing a selection screen diff --git a/src/core/hle/service/hid/controllers/npad.h b/src/core/hle/service/hid/controllers/npad.h index 931f03430..5d4c58a43 100644 --- a/src/core/hle/service/hid/controllers/npad.h +++ b/src/core/hle/service/hid/controllers/npad.h @@ -58,6 +58,12 @@ public: }; static_assert(sizeof(Vibration) == 0x10, "Vibration is an invalid size"); + enum class GyroscopeZeroDriftMode : u32 { + Loose = 0, + Standard = 1, + Tight = 2, + }; + enum class NpadHoldType : u64 { Vertical = 0, Horizontal = 1, @@ -117,6 +123,8 @@ public: void ConnectNPad(u32 npad_id); void DisconnectNPad(u32 npad_id); + void SetGyroscopeZeroDriftMode(GyroscopeZeroDriftMode drift_mode); + GyroscopeZeroDriftMode GetGyroscopeZeroDriftMode() const; LedPattern GetLedPattern(u32 npad_id); void SetVibrationEnabled(bool can_vibrate); bool IsVibrationEnabled() const; @@ -324,8 +332,8 @@ private: std::array<Kernel::EventPair, 10> styleset_changed_events; Vibration last_processed_vibration{}; std::array<ControllerHolder, 10> connected_controllers{}; + GyroscopeZeroDriftMode gyroscope_zero_drift_mode{GyroscopeZeroDriftMode::Standard}; bool can_controllers_vibrate{true}; - std::array<ControllerPad, 10> npad_pad_states{}; bool is_in_lr_assignment_mode{false}; Core::System& system; diff --git a/src/core/hle/service/hid/controllers/stubbed.cpp b/src/core/hle/service/hid/controllers/stubbed.cpp index 9e527d176..e7483bfa2 100644 --- a/src/core/hle/service/hid/controllers/stubbed.cpp +++ b/src/core/hle/service/hid/controllers/stubbed.cpp @@ -23,7 +23,7 @@ void Controller_Stubbed::OnUpdate(const Core::Timing::CoreTiming& core_timing, u } CommonHeader header{}; - header.timestamp = core_timing.GetTicks(); + header.timestamp = core_timing.GetCPUTicks(); header.total_entry_count = 17; header.entry_count = 0; header.last_entry_index = 0; diff --git a/src/core/hle/service/hid/controllers/touchscreen.cpp b/src/core/hle/service/hid/controllers/touchscreen.cpp index 1c6e55566..e326f8f5c 100644 --- a/src/core/hle/service/hid/controllers/touchscreen.cpp +++ b/src/core/hle/service/hid/controllers/touchscreen.cpp @@ -22,7 +22,7 @@ void Controller_Touchscreen::OnRelease() {} void Controller_Touchscreen::OnUpdate(const Core::Timing::CoreTiming& core_timing, u8* data, std::size_t size) { - shared_memory.header.timestamp = core_timing.GetTicks(); + shared_memory.header.timestamp = core_timing.GetCPUTicks(); shared_memory.header.total_entry_count = 17; if (!IsControllerActivated()) { @@ -49,7 +49,7 @@ void Controller_Touchscreen::OnUpdate(const Core::Timing::CoreTiming& core_timin touch_entry.diameter_x = Settings::values.touchscreen.diameter_x; touch_entry.diameter_y = Settings::values.touchscreen.diameter_y; touch_entry.rotation_angle = Settings::values.touchscreen.rotation_angle; - const u64 tick = core_timing.GetTicks(); + const u64 tick = core_timing.GetCPUTicks(); touch_entry.delta_time = tick - last_touch; last_touch = tick; touch_entry.finger = Settings::values.touchscreen.finger; diff --git a/src/core/hle/service/hid/controllers/xpad.cpp b/src/core/hle/service/hid/controllers/xpad.cpp index 27511b27b..2503ef241 100644 --- a/src/core/hle/service/hid/controllers/xpad.cpp +++ b/src/core/hle/service/hid/controllers/xpad.cpp @@ -20,7 +20,7 @@ void Controller_XPad::OnRelease() {} void Controller_XPad::OnUpdate(const Core::Timing::CoreTiming& core_timing, u8* data, std::size_t size) { for (auto& xpad_entry : shared_memory.shared_memory_entries) { - xpad_entry.header.timestamp = core_timing.GetTicks(); + xpad_entry.header.timestamp = core_timing.GetCPUTicks(); xpad_entry.header.total_entry_count = 17; if (!IsControllerActivated()) { diff --git a/src/core/hle/service/hid/hid.cpp b/src/core/hle/service/hid/hid.cpp index 5559587e3..e9020e0dc 100644 --- a/src/core/hle/service/hid/hid.cpp +++ b/src/core/hle/service/hid/hid.cpp @@ -39,11 +39,9 @@ namespace Service::HID { // Updating period for each HID device. // TODO(ogniK): Find actual polling rate of hid -constexpr s64 pad_update_ticks = static_cast<s64>(Core::Hardware::BASE_CLOCK_RATE / 66); -[[maybe_unused]] constexpr s64 accelerometer_update_ticks = - static_cast<s64>(Core::Hardware::BASE_CLOCK_RATE / 100); -[[maybe_unused]] constexpr s64 gyroscope_update_ticks = - static_cast<s64>(Core::Hardware::BASE_CLOCK_RATE / 100); +constexpr s64 pad_update_ticks = static_cast<s64>(1000000000 / 66); +[[maybe_unused]] constexpr s64 accelerometer_update_ticks = static_cast<s64>(1000000000 / 100); +[[maybe_unused]] constexpr s64 gyroscope_update_ticks = static_cast<s64>(1000000000 / 100); constexpr std::size_t SHARED_MEMORY_SIZE = 0x40000; IAppletResource::IAppletResource(Core::System& system) @@ -78,8 +76,8 @@ IAppletResource::IAppletResource(Core::System& system) // Register update callbacks pad_update_event = - Core::Timing::CreateEvent("HID::UpdatePadCallback", [this](u64 userdata, s64 cycles_late) { - UpdateControllers(userdata, cycles_late); + Core::Timing::CreateEvent("HID::UpdatePadCallback", [this](u64 userdata, s64 ns_late) { + UpdateControllers(userdata, ns_late); }); // TODO(shinyquagsire23): Other update callbacks? (accel, gyro?) @@ -109,7 +107,7 @@ void IAppletResource::GetSharedMemoryHandle(Kernel::HLERequestContext& ctx) { rb.PushCopyObjects(shared_mem); } -void IAppletResource::UpdateControllers(u64 userdata, s64 cycles_late) { +void IAppletResource::UpdateControllers(u64 userdata, s64 ns_late) { auto& core_timing = system.CoreTiming(); const bool should_reload = Settings::values.is_device_reload_pending.exchange(false); @@ -120,7 +118,7 @@ void IAppletResource::UpdateControllers(u64 userdata, s64 cycles_late) { controller->OnUpdate(core_timing, shared_mem->GetPointer(), SHARED_MEMORY_SIZE); } - core_timing.ScheduleEvent(pad_update_ticks - cycles_late, pad_update_event); + core_timing.ScheduleEvent(pad_update_ticks - ns_late, pad_update_event); } class IActiveVibrationDeviceList final : public ServiceFramework<IActiveVibrationDeviceList> { @@ -157,11 +155,11 @@ Hid::Hid(Core::System& system) : ServiceFramework("hid"), system(system) { {11, &Hid::ActivateTouchScreen, "ActivateTouchScreen"}, {21, &Hid::ActivateMouse, "ActivateMouse"}, {31, &Hid::ActivateKeyboard, "ActivateKeyboard"}, - {32, nullptr, "SendKeyboardLockKeyEvent"}, + {32, &Hid::SendKeyboardLockKeyEvent, "SendKeyboardLockKeyEvent"}, {40, nullptr, "AcquireXpadIdEventHandle"}, {41, nullptr, "ReleaseXpadIdEventHandle"}, {51, &Hid::ActivateXpad, "ActivateXpad"}, - {55, nullptr, "GetXpadIds"}, + {55, &Hid::GetXpadIDs, "GetXpadIds"}, {56, nullptr, "ActivateJoyXpad"}, {58, nullptr, "GetJoyXpadLifoHandle"}, {59, nullptr, "GetJoyXpadIds"}, @@ -185,8 +183,8 @@ Hid::Hid(Core::System& system) : ServiceFramework("hid"), system(system) { {77, nullptr, "GetAccelerometerPlayMode"}, {78, nullptr, "ResetAccelerometerPlayMode"}, {79, &Hid::SetGyroscopeZeroDriftMode, "SetGyroscopeZeroDriftMode"}, - {80, nullptr, "GetGyroscopeZeroDriftMode"}, - {81, nullptr, "ResetGyroscopeZeroDriftMode"}, + {80, &Hid::GetGyroscopeZeroDriftMode, "GetGyroscopeZeroDriftMode"}, + {81, &Hid::ResetGyroscopeZeroDriftMode, "ResetGyroscopeZeroDriftMode"}, {82, &Hid::IsSixAxisSensorAtRest, "IsSixAxisSensorAtRest"}, {83, nullptr, "IsFirmwareUpdateAvailableForSixAxisSensor"}, {91, &Hid::ActivateGesture, "ActivateGesture"}, @@ -230,15 +228,15 @@ Hid::Hid(Core::System& system) : ServiceFramework("hid"), system(system) { {211, nullptr, "IsVibrationDeviceMounted"}, {300, &Hid::ActivateConsoleSixAxisSensor, "ActivateConsoleSixAxisSensor"}, {301, &Hid::StartConsoleSixAxisSensor, "StartConsoleSixAxisSensor"}, - {302, nullptr, "StopConsoleSixAxisSensor"}, - {303, nullptr, "ActivateSevenSixAxisSensor"}, - {304, nullptr, "StartSevenSixAxisSensor"}, + {302, &Hid::StopConsoleSixAxisSensor, "StopConsoleSixAxisSensor"}, + {303, &Hid::ActivateSevenSixAxisSensor, "ActivateSevenSixAxisSensor"}, + {304, &Hid::StartSevenSixAxisSensor, "StartSevenSixAxisSensor"}, {305, &Hid::StopSevenSixAxisSensor, "StopSevenSixAxisSensor"}, {306, &Hid::InitializeSevenSixAxisSensor, "InitializeSevenSixAxisSensor"}, - {307, nullptr, "FinalizeSevenSixAxisSensor"}, + {307, &Hid::FinalizeSevenSixAxisSensor, "FinalizeSevenSixAxisSensor"}, {308, nullptr, "SetSevenSixAxisSensorFusionStrength"}, {309, nullptr, "GetSevenSixAxisSensorFusionStrength"}, - {310, nullptr, "ResetSevenSixAxisSensorTimestamp"}, + {310, &Hid::ResetSevenSixAxisSensorTimestamp, "ResetSevenSixAxisSensorTimestamp"}, {400, nullptr, "IsUsbFullKeyControllerEnabled"}, {401, nullptr, "EnableUsbFullKeyController"}, {402, nullptr, "IsUsbFullKeyControllerConnected"}, @@ -319,6 +317,17 @@ void Hid::ActivateXpad(Kernel::HLERequestContext& ctx) { rb.Push(RESULT_SUCCESS); } +void Hid::GetXpadIDs(Kernel::HLERequestContext& ctx) { + IPC::RequestParser rp{ctx}; + const auto applet_resource_user_id{rp.Pop<u64>()}; + + LOG_DEBUG(Service_HID, "(STUBBED) called, applet_resource_user_id={}", applet_resource_user_id); + + IPC::ResponseBuilder rb{ctx, 3}; + rb.Push(RESULT_SUCCESS); + rb.Push(0); +} + void Hid::ActivateDebugPad(Kernel::HLERequestContext& ctx) { IPC::RequestParser rp{ctx}; const auto applet_resource_user_id{rp.Pop<u64>()}; @@ -363,6 +372,15 @@ void Hid::ActivateKeyboard(Kernel::HLERequestContext& ctx) { rb.Push(RESULT_SUCCESS); } +void Hid::SendKeyboardLockKeyEvent(Kernel::HLERequestContext& ctx) { + IPC::RequestParser rp{ctx}; + const auto flags{rp.Pop<u32>()}; + LOG_WARNING(Service_HID, "(STUBBED) called. flags={}", flags); + + IPC::ResponseBuilder rb{ctx, 2}; + rb.Push(RESULT_SUCCESS); +} + void Hid::ActivateGesture(Kernel::HLERequestContext& ctx) { IPC::RequestParser rp{ctx}; const auto unknown{rp.Pop<u32>()}; @@ -402,15 +420,59 @@ void Hid::StartSixAxisSensor(Kernel::HLERequestContext& ctx) { rb.Push(RESULT_SUCCESS); } +void Hid::StopSixAxisSensor(Kernel::HLERequestContext& ctx) { + IPC::RequestParser rp{ctx}; + const auto handle{rp.Pop<u32>()}; + const auto applet_resource_user_id{rp.Pop<u64>()}; + + LOG_WARNING(Service_HID, "(STUBBED) called, handle={}, applet_resource_user_id={}", handle, + applet_resource_user_id); + + IPC::ResponseBuilder rb{ctx, 2}; + rb.Push(RESULT_SUCCESS); +} + void Hid::SetGyroscopeZeroDriftMode(Kernel::HLERequestContext& ctx) { IPC::RequestParser rp{ctx}; const auto handle{rp.Pop<u32>()}; const auto drift_mode{rp.Pop<u32>()}; const auto applet_resource_user_id{rp.Pop<u64>()}; - LOG_WARNING(Service_HID, - "(STUBBED) called, handle={}, drift_mode={}, applet_resource_user_id={}", handle, - drift_mode, applet_resource_user_id); + applet_resource->GetController<Controller_NPad>(HidController::NPad) + .SetGyroscopeZeroDriftMode(Controller_NPad::GyroscopeZeroDriftMode{drift_mode}); + + LOG_DEBUG(Service_HID, "called, handle={}, drift_mode={}, applet_resource_user_id={}", handle, + drift_mode, applet_resource_user_id); + + IPC::ResponseBuilder rb{ctx, 2}; + rb.Push(RESULT_SUCCESS); +} + +void Hid::GetGyroscopeZeroDriftMode(Kernel::HLERequestContext& ctx) { + IPC::RequestParser rp{ctx}; + const auto handle{rp.Pop<u32>()}; + const auto applet_resource_user_id{rp.Pop<u64>()}; + + LOG_DEBUG(Service_HID, "called, handle={}, applet_resource_user_id={}", handle, + applet_resource_user_id); + + IPC::ResponseBuilder rb{ctx, 3}; + rb.Push(RESULT_SUCCESS); + rb.Push<u32>( + static_cast<u32>(applet_resource->GetController<Controller_NPad>(HidController::NPad) + .GetGyroscopeZeroDriftMode())); +} + +void Hid::ResetGyroscopeZeroDriftMode(Kernel::HLERequestContext& ctx) { + IPC::RequestParser rp{ctx}; + const auto handle{rp.Pop<u32>()}; + const auto applet_resource_user_id{rp.Pop<u64>()}; + + applet_resource->GetController<Controller_NPad>(HidController::NPad) + .SetGyroscopeZeroDriftMode(Controller_NPad::GyroscopeZeroDriftMode::Standard); + + LOG_DEBUG(Service_HID, "called, handle={}, applet_resource_user_id={}", handle, + applet_resource_user_id); IPC::ResponseBuilder rb{ctx, 2}; rb.Push(RESULT_SUCCESS); @@ -821,33 +883,35 @@ void Hid::StartConsoleSixAxisSensor(Kernel::HLERequestContext& ctx) { rb.Push(RESULT_SUCCESS); } -void Hid::StopSixAxisSensor(Kernel::HLERequestContext& ctx) { +void Hid::StopConsoleSixAxisSensor(Kernel::HLERequestContext& ctx) { IPC::RequestParser rp{ctx}; const auto handle{rp.Pop<u32>()}; + const auto applet_resource_user_id{rp.Pop<u64>()}; - LOG_WARNING(Service_HID, "(STUBBED) called, handle={}", handle); + LOG_WARNING(Service_HID, "(STUBBED) called, handle={}, applet_resource_user_id={}", handle, + applet_resource_user_id); IPC::ResponseBuilder rb{ctx, 2}; rb.Push(RESULT_SUCCESS); } -void Hid::SetIsPalmaAllConnectable(Kernel::HLERequestContext& ctx) { +void Hid::ActivateSevenSixAxisSensor(Kernel::HLERequestContext& ctx) { IPC::RequestParser rp{ctx}; const auto applet_resource_user_id{rp.Pop<u64>()}; - const auto unknown{rp.Pop<u32>()}; - LOG_WARNING(Service_HID, "(STUBBED) called, applet_resource_user_id={}, unknown={}", - applet_resource_user_id, unknown); + LOG_WARNING(Service_HID, "(STUBBED) called, applet_resource_user_id={}", + applet_resource_user_id); IPC::ResponseBuilder rb{ctx, 2}; rb.Push(RESULT_SUCCESS); } -void Hid::SetPalmaBoostMode(Kernel::HLERequestContext& ctx) { +void Hid::StartSevenSixAxisSensor(Kernel::HLERequestContext& ctx) { IPC::RequestParser rp{ctx}; - const auto unknown{rp.Pop<u32>()}; + const auto applet_resource_user_id{rp.Pop<u64>()}; - LOG_WARNING(Service_HID, "(STUBBED) called, unknown={}", unknown); + LOG_WARNING(Service_HID, "(STUBBED) called, applet_resource_user_id={}", + applet_resource_user_id); IPC::ResponseBuilder rb{ctx, 2}; rb.Push(RESULT_SUCCESS); @@ -871,6 +935,51 @@ void Hid::InitializeSevenSixAxisSensor(Kernel::HLERequestContext& ctx) { rb.Push(RESULT_SUCCESS); } +void Hid::FinalizeSevenSixAxisSensor(Kernel::HLERequestContext& ctx) { + IPC::RequestParser rp{ctx}; + const auto applet_resource_user_id{rp.Pop<u64>()}; + + LOG_WARNING(Service_HID, "(STUBBED) called, applet_resource_user_id={}", + applet_resource_user_id); + + IPC::ResponseBuilder rb{ctx, 2}; + rb.Push(RESULT_SUCCESS); +} + +void Hid::ResetSevenSixAxisSensorTimestamp(Kernel::HLERequestContext& ctx) { + IPC::RequestParser rp{ctx}; + const auto applet_resource_user_id{rp.Pop<u64>()}; + + LOG_WARNING(Service_HID, "(STUBBED) called, applet_resource_user_id={}", + applet_resource_user_id); + + IPC::ResponseBuilder rb{ctx, 2}; + rb.Push(RESULT_SUCCESS); +} + +void Hid::SetIsPalmaAllConnectable(Kernel::HLERequestContext& ctx) { + IPC::RequestParser rp{ctx}; + const auto applet_resource_user_id{rp.Pop<u64>()}; + const auto is_palma_all_connectable{rp.Pop<bool>()}; + + LOG_WARNING(Service_HID, + "(STUBBED) called, applet_resource_user_id={}, is_palma_all_connectable={}", + applet_resource_user_id, is_palma_all_connectable); + + IPC::ResponseBuilder rb{ctx, 2}; + rb.Push(RESULT_SUCCESS); +} + +void Hid::SetPalmaBoostMode(Kernel::HLERequestContext& ctx) { + IPC::RequestParser rp{ctx}; + const auto palma_boost_mode{rp.Pop<bool>()}; + + LOG_WARNING(Service_HID, "(STUBBED) called, palma_boost_mode={}", palma_boost_mode); + + IPC::ResponseBuilder rb{ctx, 2}; + rb.Push(RESULT_SUCCESS); +} + class HidDbg final : public ServiceFramework<HidDbg> { public: explicit HidDbg() : ServiceFramework{"hid:dbg"} { diff --git a/src/core/hle/service/hid/hid.h b/src/core/hle/service/hid/hid.h index 23552efb1..6fb048360 100644 --- a/src/core/hle/service/hid/hid.h +++ b/src/core/hle/service/hid/hid.h @@ -86,14 +86,19 @@ public: private: void CreateAppletResource(Kernel::HLERequestContext& ctx); void ActivateXpad(Kernel::HLERequestContext& ctx); + void GetXpadIDs(Kernel::HLERequestContext& ctx); void ActivateDebugPad(Kernel::HLERequestContext& ctx); void ActivateTouchScreen(Kernel::HLERequestContext& ctx); void ActivateMouse(Kernel::HLERequestContext& ctx); void ActivateKeyboard(Kernel::HLERequestContext& ctx); + void SendKeyboardLockKeyEvent(Kernel::HLERequestContext& ctx); void ActivateGesture(Kernel::HLERequestContext& ctx); void ActivateNpadWithRevision(Kernel::HLERequestContext& ctx); void StartSixAxisSensor(Kernel::HLERequestContext& ctx); + void StopSixAxisSensor(Kernel::HLERequestContext& ctx); void SetGyroscopeZeroDriftMode(Kernel::HLERequestContext& ctx); + void GetGyroscopeZeroDriftMode(Kernel::HLERequestContext& ctx); + void ResetGyroscopeZeroDriftMode(Kernel::HLERequestContext& ctx); void IsSixAxisSensorAtRest(Kernel::HLERequestContext& ctx); void SetSupportedNpadStyleSet(Kernel::HLERequestContext& ctx); void GetSupportedNpadStyleSet(Kernel::HLERequestContext& ctx); @@ -125,11 +130,15 @@ private: void IsVibrationPermitted(Kernel::HLERequestContext& ctx); void ActivateConsoleSixAxisSensor(Kernel::HLERequestContext& ctx); void StartConsoleSixAxisSensor(Kernel::HLERequestContext& ctx); - void StopSixAxisSensor(Kernel::HLERequestContext& ctx); - void SetIsPalmaAllConnectable(Kernel::HLERequestContext& ctx); - void SetPalmaBoostMode(Kernel::HLERequestContext& ctx); + void StopConsoleSixAxisSensor(Kernel::HLERequestContext& ctx); + void ActivateSevenSixAxisSensor(Kernel::HLERequestContext& ctx); + void StartSevenSixAxisSensor(Kernel::HLERequestContext& ctx); void StopSevenSixAxisSensor(Kernel::HLERequestContext& ctx); void InitializeSevenSixAxisSensor(Kernel::HLERequestContext& ctx); + void FinalizeSevenSixAxisSensor(Kernel::HLERequestContext& ctx); + void ResetSevenSixAxisSensorTimestamp(Kernel::HLERequestContext& ctx); + void SetIsPalmaAllConnectable(Kernel::HLERequestContext& ctx); + void SetPalmaBoostMode(Kernel::HLERequestContext& ctx); std::shared_ptr<IAppletResource> applet_resource; Core::System& system; diff --git a/src/core/hle/service/hid/irs.cpp b/src/core/hle/service/hid/irs.cpp index 36ed6f7da..e82fd031b 100644 --- a/src/core/hle/service/hid/irs.cpp +++ b/src/core/hle/service/hid/irs.cpp @@ -98,7 +98,7 @@ void IRS::GetImageTransferProcessorState(Kernel::HLERequestContext& ctx) { IPC::ResponseBuilder rb{ctx, 5}; rb.Push(RESULT_SUCCESS); - rb.PushRaw<u64>(system.CoreTiming().GetTicks()); + rb.PushRaw<u64>(system.CoreTiming().GetCPUTicks()); rb.PushRaw<u32>(0); } diff --git a/src/core/hle/service/lbl/lbl.cpp b/src/core/hle/service/lbl/lbl.cpp index e8f9f2d29..17350b403 100644 --- a/src/core/hle/service/lbl/lbl.cpp +++ b/src/core/hle/service/lbl/lbl.cpp @@ -47,6 +47,7 @@ public: {26, &LBL::EnableVrMode, "EnableVrMode"}, {27, &LBL::DisableVrMode, "DisableVrMode"}, {28, &LBL::IsVrModeEnabled, "IsVrModeEnabled"}, + {29, nullptr, "IsAutoBrightnessControlSupported"}, }; // clang-format on diff --git a/src/core/hle/service/ldn/ldn.cpp b/src/core/hle/service/ldn/ldn.cpp index 92adde6d4..49972cd69 100644 --- a/src/core/hle/service/ldn/ldn.cpp +++ b/src/core/hle/service/ldn/ldn.cpp @@ -69,6 +69,7 @@ public: {101, nullptr, "GetNetworkInfoLatestUpdate"}, {102, nullptr, "Scan"}, {103, nullptr, "ScanPrivate"}, + {104, nullptr, "SetWirelessControllerRestriction"}, {200, nullptr, "OpenAccessPoint"}, {201, nullptr, "CloseAccessPoint"}, {202, nullptr, "CreateNetwork"}, diff --git a/src/core/hle/service/ldr/ldr.cpp b/src/core/hle/service/ldr/ldr.cpp index 6ad3be1b3..64a526b9e 100644 --- a/src/core/hle/service/ldr/ldr.cpp +++ b/src/core/hle/service/ldr/ldr.cpp @@ -39,42 +39,61 @@ constexpr ResultCode ERROR_NOT_INITIALIZED{ErrorModule::Loader, 87}; constexpr std::size_t MAXIMUM_LOADED_RO{0x40}; constexpr std::size_t MAXIMUM_MAP_RETRIES{0x200}; +constexpr std::size_t TEXT_INDEX{0}; +constexpr std::size_t RO_INDEX{1}; +constexpr std::size_t DATA_INDEX{2}; + +struct NRRCertification { + u64_le application_id_mask; + u64_le application_id_pattern; + INSERT_PADDING_BYTES(0x10); + std::array<u8, 0x100> public_key; // Also known as modulus + std::array<u8, 0x100> signature; +}; +static_assert(sizeof(NRRCertification) == 0x220, "NRRCertification has invalid size."); + struct NRRHeader { u32_le magic; - INSERT_PADDING_BYTES(12); - u64_le title_id_mask; - u64_le title_id_pattern; - INSERT_PADDING_BYTES(16); - std::array<u8, 0x100> modulus; - std::array<u8, 0x100> signature_1; - std::array<u8, 0x100> signature_2; - u64_le title_id; + u32_le certification_signature_key_generation; // 9.0.0+ + INSERT_PADDING_WORDS(2); + NRRCertification certification; + std::array<u8, 0x100> signature; + u64_le application_id; u32_le size; - INSERT_PADDING_BYTES(4); + u8 nrr_kind; // 7.0.0+ + INSERT_PADDING_BYTES(3); u32_le hash_offset; u32_le hash_count; - INSERT_PADDING_BYTES(8); + INSERT_PADDING_WORDS(2); +}; +static_assert(sizeof(NRRHeader) == 0x350, "NRRHeader has invalid size."); + +struct SegmentHeader { + u32_le memory_offset; + u32_le memory_size; }; -static_assert(sizeof(NRRHeader) == 0x350, "NRRHeader has incorrect size."); +static_assert(sizeof(SegmentHeader) == 0x8, "SegmentHeader has invalid size."); struct NROHeader { + // Switchbrew calls this "Start" (0x10) INSERT_PADDING_WORDS(1); u32_le mod_offset; INSERT_PADDING_WORDS(2); + + // Switchbrew calls this "Header" (0x70) u32_le magic; u32_le version; u32_le nro_size; u32_le flags; - u32_le text_offset; - u32_le text_size; - u32_le ro_offset; - u32_le ro_size; - u32_le rw_offset; - u32_le rw_size; + // .text, .ro, .data + std::array<SegmentHeader, 3> segment_headers; u32_le bss_size; INSERT_PADDING_WORDS(1); std::array<u8, 0x20> build_id; - INSERT_PADDING_BYTES(0x20); + u32_le dso_handle_offset; + INSERT_PADDING_WORDS(1); + // .apiInfo, .dynstr, .dynsym + std::array<SegmentHeader, 3> segment_headers_2; }; static_assert(sizeof(NROHeader) == 0x80, "NROHeader has invalid size."); @@ -91,6 +110,7 @@ struct NROInfo { std::size_t data_size{}; VAddr src_addr{}; }; +static_assert(sizeof(NROInfo) == 0x60, "NROInfo has invalid size."); class DebugMonitor final : public ServiceFramework<DebugMonitor> { public: @@ -226,11 +246,11 @@ public: return; } - if (system.CurrentProcess()->GetTitleID() != header.title_id) { + if (system.CurrentProcess()->GetTitleID() != header.application_id) { LOG_ERROR(Service_LDR, "Attempting to load NRR with title ID other than current process. (actual " "{:016X})!", - header.title_id); + header.application_id); IPC::ResponseBuilder rb{ctx, 2}; rb.Push(ERROR_INVALID_NRR); return; @@ -348,10 +368,10 @@ public: ResultCode LoadNro(Kernel::Process* process, const NROHeader& nro_header, VAddr nro_addr, VAddr start) const { - const VAddr text_start{start + nro_header.text_offset}; - const VAddr ro_start{start + nro_header.ro_offset}; - const VAddr data_start{start + nro_header.rw_offset}; - const VAddr bss_start{data_start + nro_header.rw_size}; + const VAddr text_start{start + nro_header.segment_headers[TEXT_INDEX].memory_offset}; + const VAddr ro_start{start + nro_header.segment_headers[RO_INDEX].memory_offset}; + const VAddr data_start{start + nro_header.segment_headers[DATA_INDEX].memory_offset}; + const VAddr bss_start{data_start + nro_header.segment_headers[DATA_INDEX].memory_size}; const VAddr bss_end_addr{ Common::AlignUp(bss_start + nro_header.bss_size, Kernel::Memory::PageSize)}; @@ -360,9 +380,12 @@ public: system.Memory().ReadBlock(src_addr, source_data.data(), source_data.size()); system.Memory().WriteBlock(dst_addr, source_data.data(), source_data.size()); }}; - CopyCode(nro_addr + nro_header.text_offset, text_start, nro_header.text_size); - CopyCode(nro_addr + nro_header.ro_offset, ro_start, nro_header.ro_size); - CopyCode(nro_addr + nro_header.rw_offset, data_start, nro_header.rw_size); + CopyCode(nro_addr + nro_header.segment_headers[TEXT_INDEX].memory_offset, text_start, + nro_header.segment_headers[TEXT_INDEX].memory_size); + CopyCode(nro_addr + nro_header.segment_headers[RO_INDEX].memory_offset, ro_start, + nro_header.segment_headers[RO_INDEX].memory_size); + CopyCode(nro_addr + nro_header.segment_headers[DATA_INDEX].memory_offset, data_start, + nro_header.segment_headers[DATA_INDEX].memory_size); CASCADE_CODE(process->PageTable().SetCodeMemoryPermission( text_start, ro_start - text_start, Kernel::Memory::MemoryPermission::ReadAndExecute)); @@ -484,9 +507,11 @@ public: } // Track the loaded NRO - nro.insert_or_assign(*map_result, NROInfo{hash, *map_result, nro_size, bss_address, - bss_size, header.text_size, header.ro_size, - header.rw_size, nro_address}); + nro.insert_or_assign(*map_result, + NROInfo{hash, *map_result, nro_size, bss_address, bss_size, + header.segment_headers[TEXT_INDEX].memory_size, + header.segment_headers[RO_INDEX].memory_size, + header.segment_headers[DATA_INDEX].memory_size, nro_address}); // Invalidate JIT caches for the newly mapped process code system.InvalidateCpuInstructionCaches(); @@ -584,11 +609,21 @@ private: static bool IsValidNRO(const NROHeader& header, u64 nro_size, u64 bss_size) { return header.magic == Common::MakeMagic('N', 'R', 'O', '0') && header.nro_size == nro_size && header.bss_size == bss_size && - header.ro_offset == header.text_offset + header.text_size && - header.rw_offset == header.ro_offset + header.ro_size && - nro_size == header.rw_offset + header.rw_size && - Common::Is4KBAligned(header.text_size) && Common::Is4KBAligned(header.ro_size) && - Common::Is4KBAligned(header.rw_size); + + header.segment_headers[RO_INDEX].memory_offset == + header.segment_headers[TEXT_INDEX].memory_offset + + header.segment_headers[TEXT_INDEX].memory_size && + + header.segment_headers[DATA_INDEX].memory_offset == + header.segment_headers[RO_INDEX].memory_offset + + header.segment_headers[RO_INDEX].memory_size && + + nro_size == header.segment_headers[DATA_INDEX].memory_offset + + header.segment_headers[DATA_INDEX].memory_size && + + Common::Is4KBAligned(header.segment_headers[TEXT_INDEX].memory_size) && + Common::Is4KBAligned(header.segment_headers[RO_INDEX].memory_size) && + Common::Is4KBAligned(header.segment_headers[DATA_INDEX].memory_size); } Core::System& system; }; diff --git a/src/core/hle/service/lm/manager.cpp b/src/core/hle/service/lm/manager.cpp index b67081b86..3ee2374e7 100644 --- a/src/core/hle/service/lm/manager.cpp +++ b/src/core/hle/service/lm/manager.cpp @@ -86,7 +86,8 @@ std::string FormatField(Field type, const std::vector<u8>& data) { return Common::StringFromFixedZeroTerminatedBuffer( reinterpret_cast<const char*>(data.data()), data.size()); default: - UNIMPLEMENTED(); + UNIMPLEMENTED_MSG("Unimplemented field type={}", type); + return ""; } } diff --git a/src/core/hle/service/mig/mig.cpp b/src/core/hle/service/mig/mig.cpp index d16367f2c..113a4665c 100644 --- a/src/core/hle/service/mig/mig.cpp +++ b/src/core/hle/service/mig/mig.cpp @@ -20,6 +20,12 @@ public: {101, nullptr, "ResumeServer"}, {200, nullptr, "CreateClient"}, {201, nullptr, "ResumeClient"}, + {1001, nullptr, "Unknown1001"}, + {1010, nullptr, "Unknown1010"}, + {1100, nullptr, "Unknown1100"}, + {1101, nullptr, "Unknown1101"}, + {1200, nullptr, "Unknown1200"}, + {1201, nullptr, "Unknown1201"} }; // clang-format on diff --git a/src/core/hle/service/mm/mm_u.cpp b/src/core/hle/service/mm/mm_u.cpp index def63dc8a..25c24e537 100644 --- a/src/core/hle/service/mm/mm_u.cpp +++ b/src/core/hle/service/mm/mm_u.cpp @@ -14,14 +14,14 @@ public: explicit MM_U() : ServiceFramework{"mm:u"} { // clang-format off static const FunctionInfo functions[] = { - {0, &MM_U::Initialize, "Initialize"}, - {1, &MM_U::Finalize, "Finalize"}, - {2, &MM_U::SetAndWait, "SetAndWait"}, - {3, &MM_U::Get, "Get"}, - {4, &MM_U::InitializeWithId, "InitializeWithId"}, - {5, &MM_U::FinalizeWithId, "FinalizeWithId"}, - {6, &MM_U::SetAndWaitWithId, "SetAndWaitWithId"}, - {7, &MM_U::GetWithId, "GetWithId"}, + {0, &MM_U::InitializeOld, "InitializeOld"}, + {1, &MM_U::FinalizeOld, "FinalizeOld"}, + {2, &MM_U::SetAndWaitOld, "SetAndWaitOld"}, + {3, &MM_U::GetOld, "GetOld"}, + {4, &MM_U::Initialize, "Initialize"}, + {5, &MM_U::Finalize, "Finalize"}, + {6, &MM_U::SetAndWait, "SetAndWait"}, + {7, &MM_U::Get, "Get"}, }; // clang-format on @@ -29,21 +29,21 @@ public: } private: - void Initialize(Kernel::HLERequestContext& ctx) { + void InitializeOld(Kernel::HLERequestContext& ctx) { LOG_WARNING(Service_MM, "(STUBBED) called"); IPC::ResponseBuilder rb{ctx, 2}; rb.Push(RESULT_SUCCESS); } - void Finalize(Kernel::HLERequestContext& ctx) { + void FinalizeOld(Kernel::HLERequestContext& ctx) { LOG_WARNING(Service_MM, "(STUBBED) called"); IPC::ResponseBuilder rb{ctx, 2}; rb.Push(RESULT_SUCCESS); } - void SetAndWait(Kernel::HLERequestContext& ctx) { + void SetAndWaitOld(Kernel::HLERequestContext& ctx) { IPC::RequestParser rp{ctx}; min = rp.Pop<u32>(); max = rp.Pop<u32>(); @@ -54,7 +54,7 @@ private: rb.Push(RESULT_SUCCESS); } - void Get(Kernel::HLERequestContext& ctx) { + void GetOld(Kernel::HLERequestContext& ctx) { LOG_WARNING(Service_MM, "(STUBBED) called"); IPC::ResponseBuilder rb{ctx, 3}; @@ -62,7 +62,7 @@ private: rb.Push(current); } - void InitializeWithId(Kernel::HLERequestContext& ctx) { + void Initialize(Kernel::HLERequestContext& ctx) { LOG_WARNING(Service_MM, "(STUBBED) called"); IPC::ResponseBuilder rb{ctx, 3}; @@ -70,14 +70,14 @@ private: rb.Push<u32>(id); // Any non zero value } - void FinalizeWithId(Kernel::HLERequestContext& ctx) { + void Finalize(Kernel::HLERequestContext& ctx) { LOG_WARNING(Service_MM, "(STUBBED) called"); IPC::ResponseBuilder rb{ctx, 2}; rb.Push(RESULT_SUCCESS); } - void SetAndWaitWithId(Kernel::HLERequestContext& ctx) { + void SetAndWait(Kernel::HLERequestContext& ctx) { IPC::RequestParser rp{ctx}; u32 input_id = rp.Pop<u32>(); min = rp.Pop<u32>(); @@ -90,7 +90,7 @@ private: rb.Push(RESULT_SUCCESS); } - void GetWithId(Kernel::HLERequestContext& ctx) { + void Get(Kernel::HLERequestContext& ctx) { LOG_WARNING(Service_MM, "(STUBBED) called"); IPC::ResponseBuilder rb{ctx, 3}; diff --git a/src/core/hle/service/ncm/ncm.cpp b/src/core/hle/service/ncm/ncm.cpp index ec9aae04a..e38dea1f4 100644 --- a/src/core/hle/service/ncm/ncm.cpp +++ b/src/core/hle/service/ncm/ncm.cpp @@ -28,16 +28,16 @@ public: {7, nullptr, "ResolveApplicationLegalInformationPath"}, {8, nullptr, "RedirectApplicationLegalInformationPath"}, {9, nullptr, "Refresh"}, - {10, nullptr, "RedirectProgramPath2"}, - {11, nullptr, "Refresh2"}, - {12, nullptr, "DeleteProgramPath"}, - {13, nullptr, "DeleteApplicationControlPath"}, - {14, nullptr, "DeleteApplicationHtmlDocumentPath"}, - {15, nullptr, "DeleteApplicationLegalInformationPath"}, - {16, nullptr, ""}, - {17, nullptr, ""}, - {18, nullptr, ""}, - {19, nullptr, ""}, + {10, nullptr, "RedirectApplicationProgramPath"}, + {11, nullptr, "ClearApplicationRedirection"}, + {12, nullptr, "EraseProgramRedirection"}, + {13, nullptr, "EraseApplicationControlRedirection"}, + {14, nullptr, "EraseApplicationHtmlDocumentRedirection"}, + {15, nullptr, "EraseApplicationLegalInformationRedirection"}, + {16, nullptr, "ResolveProgramPathForDebug"}, + {17, nullptr, "RedirectProgramPathForDebug"}, + {18, nullptr, "RedirectApplicationProgramPathForDebug"}, + {19, nullptr, "EraseProgramRedirectionForDebug"}, }; // clang-format on diff --git a/src/core/hle/service/nfc/nfc.cpp b/src/core/hle/service/nfc/nfc.cpp index b7b34ce7e..780ea30fe 100644 --- a/src/core/hle/service/nfc/nfc.cpp +++ b/src/core/hle/service/nfc/nfc.cpp @@ -198,9 +198,9 @@ public: static const FunctionInfo functions[] = { {0, nullptr, "Initialize"}, {1, nullptr, "Finalize"}, - {2, nullptr, "GetState"}, - {3, nullptr, "IsNfcEnabled"}, - {100, nullptr, "SetNfcEnabled"}, + {2, nullptr, "GetStateOld"}, + {3, nullptr, "IsNfcEnabledOld"}, + {100, nullptr, "SetNfcEnabledOld"}, {400, nullptr, "InitializeSystem"}, {401, nullptr, "FinalizeSystem"}, {402, nullptr, "GetState"}, diff --git a/src/core/hle/service/nifm/nifm.cpp b/src/core/hle/service/nifm/nifm.cpp index 767158444..01ddcdbd6 100644 --- a/src/core/hle/service/nifm/nifm.cpp +++ b/src/core/hle/service/nifm/nifm.cpp @@ -177,7 +177,8 @@ private: void CreateTemporaryNetworkProfile(Kernel::HLERequestContext& ctx) { LOG_DEBUG(Service_NIFM, "called"); - ASSERT_MSG(ctx.GetReadBufferSize() == 0x17c, "NetworkProfileData is not the correct size"); + ASSERT_MSG(ctx.GetReadBufferSize() == 0x17c, + "SfNetworkProfileData is not the correct size"); u128 uuid{}; auto buffer = ctx.ReadBuffer(); std::memcpy(&uuid, buffer.data() + 8, sizeof(u128)); diff --git a/src/core/hle/service/nvdrv/devices/nvhost_ctrl_gpu.cpp b/src/core/hle/service/nvdrv/devices/nvhost_ctrl_gpu.cpp index cc2192e5c..fba89e7a6 100644 --- a/src/core/hle/service/nvdrv/devices/nvhost_ctrl_gpu.cpp +++ b/src/core/hle/service/nvdrv/devices/nvhost_ctrl_gpu.cpp @@ -25,7 +25,7 @@ u32 nvhost_ctrl_gpu::ioctl(Ioctl command, const std::vector<u8>& input, case IoctlCommand::IocGetCharacteristicsCommand: return GetCharacteristics(input, output, output2, version); case IoctlCommand::IocGetTPCMasksCommand: - return GetTPCMasks(input, output); + return GetTPCMasks(input, output, output2, version); case IoctlCommand::IocGetActiveSlotMaskCommand: return GetActiveSlotMask(input, output); case IoctlCommand::IocZcullGetCtxSizeCommand: @@ -98,17 +98,22 @@ u32 nvhost_ctrl_gpu::GetCharacteristics(const std::vector<u8>& input, std::vecto return 0; } -u32 nvhost_ctrl_gpu::GetTPCMasks(const std::vector<u8>& input, std::vector<u8>& output) { +u32 nvhost_ctrl_gpu::GetTPCMasks(const std::vector<u8>& input, std::vector<u8>& output, + std::vector<u8>& output2, IoctlVersion version) { IoctlGpuGetTpcMasksArgs params{}; std::memcpy(¶ms, input.data(), input.size()); - LOG_INFO(Service_NVDRV, "called, mask=0x{:X}, mask_buf_addr=0x{:X}", params.mask_buf_size, - params.mask_buf_addr); - // TODO(ogniK): Confirm value on hardware - if (params.mask_buf_size) - params.tpc_mask_size = 4 * 1; // 4 * num_gpc - else - params.tpc_mask_size = 0; - std::memcpy(output.data(), ¶ms, sizeof(params)); + LOG_DEBUG(Service_NVDRV, "called, mask_buffer_size=0x{:X}", params.mask_buffer_size); + if (params.mask_buffer_size != 0) { + params.tcp_mask = 3; + } + + if (version == IoctlVersion::Version3) { + std::memcpy(output.data(), input.data(), output.size()); + std::memcpy(output2.data(), ¶ms.tcp_mask, output2.size()); + } else { + std::memcpy(output.data(), ¶ms, output.size()); + } + return 0; } @@ -195,8 +200,7 @@ u32 nvhost_ctrl_gpu::GetGpuTime(const std::vector<u8>& input, std::vector<u8>& o IoctlGetGpuTime params{}; std::memcpy(¶ms, input.data(), input.size()); - const auto ns = Core::Timing::CyclesToNs(system.CoreTiming().GetTicks()); - params.gpu_time = static_cast<u64_le>(ns.count()); + params.gpu_time = static_cast<u64_le>(system.CoreTiming().GetGlobalTimeNs().count()); std::memcpy(output.data(), ¶ms, output.size()); return 0; } diff --git a/src/core/hle/service/nvdrv/devices/nvhost_ctrl_gpu.h b/src/core/hle/service/nvdrv/devices/nvhost_ctrl_gpu.h index 07b644ec5..ef60f72ce 100644 --- a/src/core/hle/service/nvdrv/devices/nvhost_ctrl_gpu.h +++ b/src/core/hle/service/nvdrv/devices/nvhost_ctrl_gpu.h @@ -92,16 +92,11 @@ private: "IoctlCharacteristics is incorrect size"); struct IoctlGpuGetTpcMasksArgs { - /// [in] TPC mask buffer size reserved by userspace. Should be at least - /// sizeof(__u32) * fls(gpc_mask) to receive TPC mask for each GPC. - /// [out] full kernel buffer size - u32_le mask_buf_size; - u32_le reserved; - - /// [in] pointer to TPC mask buffer. It will receive one 32-bit TPC mask per GPC or 0 if - /// GPC is not enabled or not present. This parameter is ignored if mask_buf_size is 0. - u64_le mask_buf_addr; - u64_le tpc_mask_size; // Nintendo add this? + u32_le mask_buffer_size{}; + INSERT_PADDING_WORDS(1); + u64_le mask_buffer_address{}; + u32_le tcp_mask{}; + INSERT_PADDING_WORDS(1); }; static_assert(sizeof(IoctlGpuGetTpcMasksArgs) == 24, "IoctlGpuGetTpcMasksArgs is incorrect size"); @@ -166,7 +161,8 @@ private: u32 GetCharacteristics(const std::vector<u8>& input, std::vector<u8>& output, std::vector<u8>& output2, IoctlVersion version); - u32 GetTPCMasks(const std::vector<u8>& input, std::vector<u8>& output); + u32 GetTPCMasks(const std::vector<u8>& input, std::vector<u8>& output, std::vector<u8>& output2, + IoctlVersion version); u32 GetActiveSlotMask(const std::vector<u8>& input, std::vector<u8>& output); u32 ZCullGetCtxSize(const std::vector<u8>& input, std::vector<u8>& output); u32 ZCullGetInfo(const std::vector<u8>& input, std::vector<u8>& output); diff --git a/src/core/hle/service/nvflinger/buffer_queue.cpp b/src/core/hle/service/nvflinger/buffer_queue.cpp index f1e3d832a..caca80dde 100644 --- a/src/core/hle/service/nvflinger/buffer_queue.cpp +++ b/src/core/hle/service/nvflinger/buffer_queue.cpp @@ -138,9 +138,7 @@ u32 BufferQueue::Query(QueryType type) { switch (type) { case QueryType::NativeWindowFormat: - // TODO(Subv): Use an enum for this - static constexpr u32 FormatABGR8 = 1; - return FormatABGR8; + return static_cast<u32>(PixelFormat::RGBA8888); } UNIMPLEMENTED(); diff --git a/src/core/hle/service/nvflinger/buffer_queue.h b/src/core/hle/service/nvflinger/buffer_queue.h index d5f31e567..8a837e5aa 100644 --- a/src/core/hle/service/nvflinger/buffer_queue.h +++ b/src/core/hle/service/nvflinger/buffer_queue.h @@ -66,6 +66,16 @@ public: Rotate270 = 0x07, }; + enum class PixelFormat : u32 { + RGBA8888 = 1, + RGBX8888 = 2, + RGB888 = 3, + RGB565 = 4, + BGRA8888 = 5, + RGBA5551 = 6, + RRGBA4444 = 7, + }; + struct Buffer { enum class Status { Free = 0, Queued = 1, Dequeued = 2, Acquired = 3 }; diff --git a/src/core/hle/service/nvflinger/nvflinger.cpp b/src/core/hle/service/nvflinger/nvflinger.cpp index 437bc5dee..2f44d3779 100644 --- a/src/core/hle/service/nvflinger/nvflinger.cpp +++ b/src/core/hle/service/nvflinger/nvflinger.cpp @@ -9,6 +9,7 @@ #include "common/logging/log.h" #include "common/microprofile.h" #include "common/scope_exit.h" +#include "common/thread.h" #include "core/core.h" #include "core/core_timing.h" #include "core/core_timing_util.h" @@ -27,8 +28,35 @@ namespace Service::NVFlinger { -constexpr s64 frame_ticks = static_cast<s64>(Core::Hardware::BASE_CLOCK_RATE / 60); -constexpr s64 frame_ticks_30fps = static_cast<s64>(Core::Hardware::BASE_CLOCK_RATE / 30); +constexpr s64 frame_ticks = static_cast<s64>(1000000000 / 60); +constexpr s64 frame_ticks_30fps = static_cast<s64>(1000000000 / 30); + +void NVFlinger::VSyncThread(NVFlinger& nv_flinger) { + nv_flinger.SplitVSync(); +} + +void NVFlinger::SplitVSync() { + system.RegisterHostThread(); + std::string name = "yuzu:VSyncThread"; + MicroProfileOnThreadCreate(name.c_str()); + Common::SetCurrentThreadName(name.c_str()); + Common::SetCurrentThreadPriority(Common::ThreadPriority::High); + s64 delay = 0; + while (is_running) { + guard->lock(); + const s64 time_start = system.CoreTiming().GetGlobalTimeNs().count(); + Compose(); + const auto ticks = GetNextTicks(); + const s64 time_end = system.CoreTiming().GetGlobalTimeNs().count(); + const s64 time_passed = time_end - time_start; + const s64 next_time = std::max<s64>(0, ticks - time_passed - delay); + guard->unlock(); + if (next_time > 0) { + wait_event->WaitFor(std::chrono::nanoseconds{next_time}); + } + delay = (system.CoreTiming().GetGlobalTimeNs().count() - time_end) - next_time; + } +} NVFlinger::NVFlinger(Core::System& system) : system(system) { displays.emplace_back(0, "Default", system); @@ -36,22 +64,36 @@ NVFlinger::NVFlinger(Core::System& system) : system(system) { displays.emplace_back(2, "Edid", system); displays.emplace_back(3, "Internal", system); displays.emplace_back(4, "Null", system); + guard = std::make_shared<std::mutex>(); // Schedule the screen composition events composition_event = - Core::Timing::CreateEvent("ScreenComposition", [this](u64 userdata, s64 cycles_late) { + Core::Timing::CreateEvent("ScreenComposition", [this](u64 userdata, s64 ns_late) { + Lock(); Compose(); - const auto ticks = - Settings::values.force_30fps_mode ? frame_ticks_30fps : GetNextTicks(); - this->system.CoreTiming().ScheduleEvent(std::max<s64>(0LL, ticks - cycles_late), + const auto ticks = GetNextTicks(); + this->system.CoreTiming().ScheduleEvent(std::max<s64>(0LL, ticks - ns_late), composition_event); }); - - system.CoreTiming().ScheduleEvent(frame_ticks, composition_event); + if (system.IsMulticore()) { + is_running = true; + wait_event = std::make_unique<Common::Event>(); + vsync_thread = std::make_unique<std::thread>(VSyncThread, std::ref(*this)); + } else { + system.CoreTiming().ScheduleEvent(frame_ticks, composition_event); + } } NVFlinger::~NVFlinger() { - system.CoreTiming().UnscheduleEvent(composition_event, 0); + if (system.IsMulticore()) { + is_running = false; + wait_event->Set(); + vsync_thread->join(); + vsync_thread.reset(); + wait_event.reset(); + } else { + system.CoreTiming().UnscheduleEvent(composition_event, 0); + } } void NVFlinger::SetNVDrvInstance(std::shared_ptr<Nvidia::Module> instance) { @@ -199,10 +241,12 @@ void NVFlinger::Compose() { auto& gpu = system.GPU(); const auto& multi_fence = buffer->get().multi_fence; + guard->unlock(); for (u32 fence_id = 0; fence_id < multi_fence.num_fences; fence_id++) { const auto& fence = multi_fence.fences[fence_id]; gpu.WaitFence(fence.id, fence.value); } + guard->lock(); MicroProfileFlip(); @@ -223,7 +267,7 @@ void NVFlinger::Compose() { s64 NVFlinger::GetNextTicks() const { constexpr s64 max_hertz = 120LL; - return (Core::Hardware::BASE_CLOCK_RATE * (1LL << swap_interval)) / max_hertz; + return (1000000000 * (1LL << swap_interval)) / max_hertz; } } // namespace Service::NVFlinger diff --git a/src/core/hle/service/nvflinger/nvflinger.h b/src/core/hle/service/nvflinger/nvflinger.h index 57a21f33b..e4959a9af 100644 --- a/src/core/hle/service/nvflinger/nvflinger.h +++ b/src/core/hle/service/nvflinger/nvflinger.h @@ -4,15 +4,22 @@ #pragma once +#include <atomic> #include <memory> +#include <mutex> #include <optional> #include <string> #include <string_view> +#include <thread> #include <vector> #include "common/common_types.h" #include "core/hle/kernel/object.h" +namespace Common { +class Event; +} // namespace Common + namespace Core::Timing { class CoreTiming; struct EventType; @@ -79,6 +86,10 @@ public: s64 GetNextTicks() const; + std::unique_lock<std::mutex> Lock() { + return std::unique_lock{*guard}; + } + private: /// Finds the display identified by the specified ID. VI::Display* FindDisplay(u64 display_id); @@ -92,6 +103,10 @@ private: /// Finds the layer identified by the specified ID in the desired display. const VI::Layer* FindLayer(u64 display_id, u64 layer_id) const; + static void VSyncThread(NVFlinger& nv_flinger); + + void SplitVSync(); + std::shared_ptr<Nvidia::Module> nvdrv; std::vector<VI::Display> displays; @@ -108,7 +123,13 @@ private: /// Event that handles screen composition. std::shared_ptr<Core::Timing::EventType> composition_event; + std::shared_ptr<std::mutex> guard; + Core::System& system; + + std::unique_ptr<std::thread> vsync_thread; + std::unique_ptr<Common::Event> wait_event; + std::atomic<bool> is_running{}; }; } // namespace Service::NVFlinger diff --git a/src/core/hle/service/prepo/prepo.cpp b/src/core/hle/service/prepo/prepo.cpp index 14309c679..67833d9af 100644 --- a/src/core/hle/service/prepo/prepo.cpp +++ b/src/core/hle/service/prepo/prepo.cpp @@ -75,8 +75,13 @@ private: const auto user_id = rp.PopRaw<u128>(); const auto process_id = rp.PopRaw<u64>(); std::vector<std::vector<u8>> data{ctx.ReadBuffer(0)}; + if constexpr (Type == Core::Reporter::PlayReportType::Old2) { - data.emplace_back(ctx.ReadBuffer(1)); + const auto read_buffer_count = + ctx.BufferDescriptorX().size() + ctx.BufferDescriptorA().size(); + if (read_buffer_count > 1) { + data.emplace_back(ctx.ReadBuffer(1)); + } } LOG_DEBUG( diff --git a/src/core/hle/service/sm/sm.cpp b/src/core/hle/service/sm/sm.cpp index 6ada13be4..d872de16c 100644 --- a/src/core/hle/service/sm/sm.cpp +++ b/src/core/hle/service/sm/sm.cpp @@ -142,7 +142,7 @@ void SM::GetService(Kernel::HLERequestContext& ctx) { } // Wake the threads waiting on the ServerPort - server_port->WakeupAllWaitingThreads(); + server_port->Signal(); LOG_DEBUG(Service_SM, "called service={} -> session={}", name, client->GetObjectId()); IPC::ResponseBuilder rb{ctx, 2, 0, 1, IPC::ResponseBuilder::Flags::AlwaysMoveHandles}; diff --git a/src/core/hle/service/time/standard_steady_clock_core.cpp b/src/core/hle/service/time/standard_steady_clock_core.cpp index 1575f0b49..59a272f4a 100644 --- a/src/core/hle/service/time/standard_steady_clock_core.cpp +++ b/src/core/hle/service/time/standard_steady_clock_core.cpp @@ -11,9 +11,8 @@ namespace Service::Time::Clock { TimeSpanType StandardSteadyClockCore::GetCurrentRawTimePoint(Core::System& system) { - const TimeSpanType ticks_time_span{TimeSpanType::FromTicks( - Core::Timing::CpuCyclesToClockCycles(system.CoreTiming().GetTicks()), - Core::Hardware::CNTFREQ)}; + const TimeSpanType ticks_time_span{ + TimeSpanType::FromTicks(system.CoreTiming().GetClockTicks(), Core::Hardware::CNTFREQ)}; TimeSpanType raw_time_point{setup_value.nanoseconds + ticks_time_span.nanoseconds}; if (raw_time_point.nanoseconds < cached_raw_time_point.nanoseconds) { diff --git a/src/core/hle/service/time/tick_based_steady_clock_core.cpp b/src/core/hle/service/time/tick_based_steady_clock_core.cpp index 44d5bc651..8baaa2a6a 100644 --- a/src/core/hle/service/time/tick_based_steady_clock_core.cpp +++ b/src/core/hle/service/time/tick_based_steady_clock_core.cpp @@ -11,9 +11,8 @@ namespace Service::Time::Clock { SteadyClockTimePoint TickBasedSteadyClockCore::GetTimePoint(Core::System& system) { - const TimeSpanType ticks_time_span{TimeSpanType::FromTicks( - Core::Timing::CpuCyclesToClockCycles(system.CoreTiming().GetTicks()), - Core::Hardware::CNTFREQ)}; + const TimeSpanType ticks_time_span{ + TimeSpanType::FromTicks(system.CoreTiming().GetClockTicks(), Core::Hardware::CNTFREQ)}; return {ticks_time_span.ToSeconds(), GetClockSourceId()}; } diff --git a/src/core/hle/service/time/time.cpp b/src/core/hle/service/time/time.cpp index 67f1bbcf3..4cf58a61a 100644 --- a/src/core/hle/service/time/time.cpp +++ b/src/core/hle/service/time/time.cpp @@ -234,9 +234,8 @@ void Module::Interface::CalculateMonotonicSystemClockBaseTimePoint(Kernel::HLERe const auto current_time_point{steady_clock_core.GetCurrentTimePoint(system)}; if (current_time_point.clock_source_id == context.steady_time_point.clock_source_id) { - const auto ticks{Clock::TimeSpanType::FromTicks( - Core::Timing::CpuCyclesToClockCycles(system.CoreTiming().GetTicks()), - Core::Hardware::CNTFREQ)}; + const auto ticks{Clock::TimeSpanType::FromTicks(system.CoreTiming().GetClockTicks(), + Core::Hardware::CNTFREQ)}; const s64 base_time_point{context.offset + current_time_point.time_point - ticks.ToSeconds()}; IPC::ResponseBuilder rb{ctx, (sizeof(s64) / 4) + 2}; diff --git a/src/core/hle/service/time/time_manager.cpp b/src/core/hle/service/time/time_manager.cpp index 9d6c55865..b4dfe45e5 100644 --- a/src/core/hle/service/time/time_manager.cpp +++ b/src/core/hle/service/time/time_manager.cpp @@ -5,6 +5,7 @@ #include <chrono> #include <ctime> +#include "common/time_zone.h" #include "core/hle/service/time/ephemeral_network_system_clock_context_writer.h" #include "core/hle/service/time/local_system_clock_context_writer.h" #include "core/hle/service/time/network_system_clock_context_writer.h" @@ -21,8 +22,16 @@ static std::chrono::seconds GetSecondsSinceEpoch() { Settings::values.custom_rtc_differential; } +static s64 GetExternalTimeZoneOffset() { + // With "auto" timezone setting, we use the external system's timezone offset + if (Settings::GetTimeZoneString() == "auto") { + return Common::TimeZone::GetCurrentOffsetSeconds().count(); + } + return 0; +} + static s64 GetExternalRtcValue() { - return GetSecondsSinceEpoch().count(); + return GetSecondsSinceEpoch().count() + GetExternalTimeZoneOffset(); } TimeManager::TimeManager(Core::System& system) diff --git a/src/core/hle/service/time/time_sharedmemory.cpp b/src/core/hle/service/time/time_sharedmemory.cpp index 999ec1e51..e0ae9f874 100644 --- a/src/core/hle/service/time/time_sharedmemory.cpp +++ b/src/core/hle/service/time/time_sharedmemory.cpp @@ -30,8 +30,7 @@ void SharedMemory::SetupStandardSteadyClock(Core::System& system, const Common::UUID& clock_source_id, Clock::TimeSpanType current_time_point) { const Clock::TimeSpanType ticks_time_span{Clock::TimeSpanType::FromTicks( - Core::Timing::CpuCyclesToClockCycles(system.CoreTiming().GetTicks()), - Core::Hardware::CNTFREQ)}; + system.CoreTiming().GetClockTicks(), Core::Hardware::CNTFREQ)}; const Clock::SteadyClockContext context{ static_cast<u64>(current_time_point.nanoseconds - ticks_time_span.nanoseconds), clock_source_id}; diff --git a/src/core/hle/service/time/time_zone_content_manager.cpp b/src/core/hle/service/time/time_zone_content_manager.cpp index 78d4acd95..c070d6e97 100644 --- a/src/core/hle/service/time/time_zone_content_manager.cpp +++ b/src/core/hle/service/time/time_zone_content_manager.cpp @@ -5,6 +5,7 @@ #include <sstream> #include "common/logging/log.h" +#include "common/time_zone.h" #include "core/core.h" #include "core/file_sys/content_archive.h" #include "core/file_sys/nca_metadata.h" @@ -14,6 +15,7 @@ #include "core/hle/service/filesystem/filesystem.h" #include "core/hle/service/time/time_manager.h" #include "core/hle/service/time/time_zone_content_manager.h" +#include "core/settings.h" namespace Service::Time::TimeZone { @@ -68,10 +70,22 @@ static std::vector<std::string> BuildLocationNameCache(Core::System& system) { TimeZoneContentManager::TimeZoneContentManager(TimeManager& time_manager, Core::System& system) : system{system}, location_name_cache{BuildLocationNameCache(system)} { - if (FileSys::VirtualFile vfs_file; GetTimeZoneInfoFile("GMT", vfs_file) == RESULT_SUCCESS) { + + std::string location_name; + const auto timezone_setting = Settings::GetTimeZoneString(); + if (timezone_setting == "auto") { + location_name = Common::TimeZone::GetDefaultTimeZone(); + } else if (timezone_setting == "default") { + location_name = location_name; + } else { + location_name = timezone_setting; + } + + if (FileSys::VirtualFile vfs_file; + GetTimeZoneInfoFile(location_name, vfs_file) == RESULT_SUCCESS) { const auto time_point{ time_manager.GetStandardSteadyClockCore().GetCurrentTimePoint(system)}; - time_manager.SetupTimeZoneManager("GMT", time_point, location_name_cache.size(), {}, + time_manager.SetupTimeZoneManager(location_name, time_point, location_name_cache.size(), {}, vfs_file); } else { time_zone_manager.MarkAsInitialized(); @@ -114,6 +128,12 @@ ResultCode TimeZoneContentManager::GetTimeZoneInfoFile(const std::string& locati vfs_file = zoneinfo_dir->GetFile(location_name); if (!vfs_file) { + LOG_ERROR(Service_Time, "{:016X} has no file \"{}\"! Using default timezone.", + time_zone_binary_titleid, location_name); + vfs_file = zoneinfo_dir->GetFile(Common::TimeZone::GetDefaultTimeZone()); + } + + if (!vfs_file) { LOG_ERROR(Service_Time, "{:016X} has no file \"{}\"!", time_zone_binary_titleid, location_name); return ERROR_TIME_NOT_FOUND; diff --git a/src/core/hle/service/vi/vi.cpp b/src/core/hle/service/vi/vi.cpp index 46e14c2a3..157092074 100644 --- a/src/core/hle/service/vi/vi.cpp +++ b/src/core/hle/service/vi/vi.cpp @@ -511,6 +511,7 @@ private: LOG_DEBUG(Service_VI, "called. id=0x{:08X} transaction={:X}, flags=0x{:08X}", id, static_cast<u32>(transaction), flags); + nv_flinger->Lock(); auto& buffer_queue = nv_flinger->FindBufferQueue(id); switch (transaction) { @@ -550,6 +551,7 @@ private: [=](std::shared_ptr<Kernel::Thread> thread, Kernel::HLERequestContext& ctx, Kernel::ThreadWakeupReason reason) { // Repeat TransactParcel DequeueBuffer when a buffer is available + nv_flinger->Lock(); auto& buffer_queue = nv_flinger->FindBufferQueue(id); auto result = buffer_queue.DequeueBuffer(width, height); ASSERT_MSG(result != std::nullopt, "Could not dequeue buffer."); diff --git a/src/core/memory.cpp b/src/core/memory.cpp index 9d87045a0..7def00768 100644 --- a/src/core/memory.cpp +++ b/src/core/memory.cpp @@ -8,6 +8,7 @@ #include <utility> #include "common/assert.h" +#include "common/atomic_ops.h" #include "common/common_types.h" #include "common/logging/log.h" #include "common/page_table.h" @@ -29,15 +30,12 @@ namespace Core::Memory { struct Memory::Impl { explicit Impl(Core::System& system_) : system{system_} {} - void SetCurrentPageTable(Kernel::Process& process) { + void SetCurrentPageTable(Kernel::Process& process, u32 core_id) { current_page_table = &process.PageTable().PageTableImpl(); const std::size_t address_space_width = process.PageTable().GetAddressSpaceWidth(); - system.ArmInterface(0).PageTableChanged(*current_page_table, address_space_width); - system.ArmInterface(1).PageTableChanged(*current_page_table, address_space_width); - system.ArmInterface(2).PageTableChanged(*current_page_table, address_space_width); - system.ArmInterface(3).PageTableChanged(*current_page_table, address_space_width); + system.ArmInterface(core_id).PageTableChanged(*current_page_table, address_space_width); } void MapMemoryRegion(Common::PageTable& page_table, VAddr base, u64 size, PAddr target) { @@ -179,6 +177,22 @@ struct Memory::Impl { } } + bool WriteExclusive8(const VAddr addr, const u8 data, const u8 expected) { + return WriteExclusive<u8>(addr, data, expected); + } + + bool WriteExclusive16(const VAddr addr, const u16 data, const u16 expected) { + return WriteExclusive<u16_le>(addr, data, expected); + } + + bool WriteExclusive32(const VAddr addr, const u32 data, const u32 expected) { + return WriteExclusive<u32_le>(addr, data, expected); + } + + bool WriteExclusive64(const VAddr addr, const u64 data, const u64 expected) { + return WriteExclusive<u64_le>(addr, data, expected); + } + std::string ReadCString(VAddr vaddr, std::size_t max_length) { std::string string; string.reserve(max_length); @@ -682,6 +696,67 @@ struct Memory::Impl { } } + template <typename T> + bool WriteExclusive(const VAddr vaddr, const T data, const T expected) { + u8* page_pointer = current_page_table->pointers[vaddr >> PAGE_BITS]; + if (page_pointer != nullptr) { + // NOTE: Avoid adding any extra logic to this fast-path block + T volatile* pointer = reinterpret_cast<T volatile*>(&page_pointer[vaddr]); + return Common::AtomicCompareAndSwap(pointer, data, expected); + } + + const Common::PageType type = current_page_table->attributes[vaddr >> PAGE_BITS]; + switch (type) { + case Common::PageType::Unmapped: + LOG_ERROR(HW_Memory, "Unmapped Write{} 0x{:08X} @ 0x{:016X}", sizeof(data) * 8, + static_cast<u32>(data), vaddr); + return true; + case Common::PageType::Memory: + ASSERT_MSG(false, "Mapped memory page without a pointer @ {:016X}", vaddr); + break; + case Common::PageType::RasterizerCachedMemory: { + u8* host_ptr{GetPointerFromRasterizerCachedMemory(vaddr)}; + system.GPU().InvalidateRegion(vaddr, sizeof(T)); + T volatile* pointer = reinterpret_cast<T volatile*>(&host_ptr); + return Common::AtomicCompareAndSwap(pointer, data, expected); + break; + } + default: + UNREACHABLE(); + } + return true; + } + + bool WriteExclusive128(const VAddr vaddr, const u128 data, const u128 expected) { + u8* const page_pointer = current_page_table->pointers[vaddr >> PAGE_BITS]; + if (page_pointer != nullptr) { + // NOTE: Avoid adding any extra logic to this fast-path block + u64 volatile* pointer = reinterpret_cast<u64 volatile*>(&page_pointer[vaddr]); + return Common::AtomicCompareAndSwap(pointer, data, expected); + } + + const Common::PageType type = current_page_table->attributes[vaddr >> PAGE_BITS]; + switch (type) { + case Common::PageType::Unmapped: + LOG_ERROR(HW_Memory, "Unmapped Write{} 0x{:08X} @ 0x{:016X}{:016X}", sizeof(data) * 8, + static_cast<u64>(data[1]), static_cast<u64>(data[0]), vaddr); + return true; + case Common::PageType::Memory: + ASSERT_MSG(false, "Mapped memory page without a pointer @ {:016X}", vaddr); + break; + case Common::PageType::RasterizerCachedMemory: { + u8* host_ptr{GetPointerFromRasterizerCachedMemory(vaddr)}; + system.GPU().InvalidateRegion(vaddr, sizeof(u128)); + u64 volatile* pointer = reinterpret_cast<u64 volatile*>(&host_ptr); + return Common::AtomicCompareAndSwap(pointer, data, expected); + break; + } + default: + UNREACHABLE(); + } + return true; + } + Common::PageTable* current_page_table = nullptr; Core::System& system; }; @@ -689,8 +764,8 @@ struct Memory::Impl { Memory::Memory(Core::System& system) : impl{std::make_unique<Impl>(system)} {} Memory::~Memory() = default; -void Memory::SetCurrentPageTable(Kernel::Process& process) { - impl->SetCurrentPageTable(process); +void Memory::SetCurrentPageTable(Kernel::Process& process, u32 core_id) { + impl->SetCurrentPageTable(process, core_id); } void Memory::MapMemoryRegion(Common::PageTable& page_table, VAddr base, u64 size, PAddr target) { @@ -764,6 +839,26 @@ void Memory::Write64(VAddr addr, u64 data) { impl->Write64(addr, data); } +bool Memory::WriteExclusive8(VAddr addr, u8 data, u8 expected) { + return impl->WriteExclusive8(addr, data, expected); +} + +bool Memory::WriteExclusive16(VAddr addr, u16 data, u16 expected) { + return impl->WriteExclusive16(addr, data, expected); +} + +bool Memory::WriteExclusive32(VAddr addr, u32 data, u32 expected) { + return impl->WriteExclusive32(addr, data, expected); +} + +bool Memory::WriteExclusive64(VAddr addr, u64 data, u64 expected) { + return impl->WriteExclusive64(addr, data, expected); +} + +bool Memory::WriteExclusive128(VAddr addr, u128 data, u128 expected) { + return impl->WriteExclusive128(addr, data, expected); +} + std::string Memory::ReadCString(VAddr vaddr, std::size_t max_length) { return impl->ReadCString(vaddr, max_length); } diff --git a/src/core/memory.h b/src/core/memory.h index 9292f3b0a..4a1cc63f4 100644 --- a/src/core/memory.h +++ b/src/core/memory.h @@ -64,7 +64,7 @@ public: * * @param process The process to use the page table of. */ - void SetCurrentPageTable(Kernel::Process& process); + void SetCurrentPageTable(Kernel::Process& process, u32 core_id); /** * Maps an allocated buffer onto a region of the emulated process address space. @@ -245,6 +245,71 @@ public: void Write64(VAddr addr, u64 data); /** + * Writes a 8-bit unsigned integer to the given virtual address in + * the current process' address space if and only if the address contains + * the expected value. This operation is atomic. + * + * @param addr The virtual address to write the 8-bit unsigned integer to. + * @param data The 8-bit unsigned integer to write to the given virtual address. + * @param expected The 8-bit unsigned integer to check against the given virtual address. + * + * @post The memory range [addr, sizeof(data)) contains the given data value. + */ + bool WriteExclusive8(VAddr addr, u8 data, u8 expected); + + /** + * Writes a 16-bit unsigned integer to the given virtual address in + * the current process' address space if and only if the address contains + * the expected value. This operation is atomic. + * + * @param addr The virtual address to write the 16-bit unsigned integer to. + * @param data The 16-bit unsigned integer to write to the given virtual address. + * @param expected The 16-bit unsigned integer to check against the given virtual address. + * + * @post The memory range [addr, sizeof(data)) contains the given data value. + */ + bool WriteExclusive16(VAddr addr, u16 data, u16 expected); + + /** + * Writes a 32-bit unsigned integer to the given virtual address in + * the current process' address space if and only if the address contains + * the expected value. This operation is atomic. + * + * @param addr The virtual address to write the 32-bit unsigned integer to. + * @param data The 32-bit unsigned integer to write to the given virtual address. + * @param expected The 32-bit unsigned integer to check against the given virtual address. + * + * @post The memory range [addr, sizeof(data)) contains the given data value. + */ + bool WriteExclusive32(VAddr addr, u32 data, u32 expected); + + /** + * Writes a 64-bit unsigned integer to the given virtual address in + * the current process' address space if and only if the address contains + * the expected value. This operation is atomic. + * + * @param addr The virtual address to write the 64-bit unsigned integer to. + * @param data The 64-bit unsigned integer to write to the given virtual address. + * @param expected The 64-bit unsigned integer to check against the given virtual address. + * + * @post The memory range [addr, sizeof(data)) contains the given data value. + */ + bool WriteExclusive64(VAddr addr, u64 data, u64 expected); + + /** + * Writes a 128-bit unsigned integer to the given virtual address in + * the current process' address space if and only if the address contains + * the expected value. This operation is atomic. + * + * @param addr The virtual address to write the 128-bit unsigned integer to. + * @param data The 128-bit unsigned integer to write to the given virtual address. + * @param expected The 128-bit unsigned integer to check against the given virtual address. + * + * @post The memory range [addr, sizeof(data)) contains the given data value. + */ + bool WriteExclusive128(VAddr addr, u128 data, u128 expected); + + /** * Reads a null-terminated string from the given virtual address. * This function will continually read characters until either: * diff --git a/src/core/memory/cheat_engine.cpp b/src/core/memory/cheat_engine.cpp index b139e8465..53d27859b 100644 --- a/src/core/memory/cheat_engine.cpp +++ b/src/core/memory/cheat_engine.cpp @@ -20,7 +20,7 @@ namespace Core::Memory { -constexpr s64 CHEAT_ENGINE_TICKS = static_cast<s64>(Core::Hardware::BASE_CLOCK_RATE / 12); +constexpr s64 CHEAT_ENGINE_TICKS = static_cast<s64>(1000000000 / 12); constexpr u32 KEYPAD_BITMASK = 0x3FFFFFF; StandardVmCallbacks::StandardVmCallbacks(Core::System& system, const CheatProcessMetadata& metadata) @@ -190,7 +190,7 @@ CheatEngine::~CheatEngine() { void CheatEngine::Initialize() { event = Core::Timing::CreateEvent( "CheatEngine::FrameCallback::" + Common::HexToString(metadata.main_nso_build_id), - [this](u64 userdata, s64 cycles_late) { FrameCallback(userdata, cycles_late); }); + [this](u64 userdata, s64 ns_late) { FrameCallback(userdata, ns_late); }); core_timing.ScheduleEvent(CHEAT_ENGINE_TICKS, event); metadata.process_id = system.CurrentProcess()->GetProcessID(); @@ -217,7 +217,7 @@ void CheatEngine::Reload(std::vector<CheatEntry> cheats) { MICROPROFILE_DEFINE(Cheat_Engine, "Add-Ons", "Cheat Engine", MP_RGB(70, 200, 70)); -void CheatEngine::FrameCallback(u64 userdata, s64 cycles_late) { +void CheatEngine::FrameCallback(u64 userdata, s64 ns_late) { if (is_pending_reload.exchange(false)) { vm.LoadProgram(cheats); } @@ -230,7 +230,7 @@ void CheatEngine::FrameCallback(u64 userdata, s64 cycles_late) { vm.Execute(metadata); - core_timing.ScheduleEvent(CHEAT_ENGINE_TICKS - cycles_late, event); + core_timing.ScheduleEvent(CHEAT_ENGINE_TICKS - ns_late, event); } } // namespace Core::Memory diff --git a/src/core/perf_stats.cpp b/src/core/perf_stats.cpp index f1ae9d4df..9f3a6b811 100644 --- a/src/core/perf_stats.cpp +++ b/src/core/perf_stats.cpp @@ -119,7 +119,7 @@ double PerfStats::GetLastFrameTimeScale() { } void FrameLimiter::DoFrameLimiting(microseconds current_system_time_us) { - if (!Settings::values.use_frame_limit) { + if (!Settings::values.use_frame_limit || Settings::values.use_multi_core) { return; } diff --git a/src/core/reporter.cpp b/src/core/reporter.cpp index 558cbe6d7..76cfa5a17 100644 --- a/src/core/reporter.cpp +++ b/src/core/reporter.cpp @@ -4,11 +4,12 @@ #include <ctime> #include <fstream> +#include <iomanip> #include <fmt/chrono.h> #include <fmt/format.h> #include <fmt/ostream.h> -#include <json.hpp> +#include <nlohmann/json.hpp> #include "common/file_util.h" #include "common/hex_util.h" diff --git a/src/core/settings.cpp b/src/core/settings.cpp index 2b0bdc4d3..56df5e925 100644 --- a/src/core/settings.cpp +++ b/src/core/settings.cpp @@ -63,6 +63,21 @@ const std::array<const char*, NumMouseButtons> mapping = {{ Values values = {}; +std::string GetTimeZoneString() { + static constexpr std::array<const char*, 46> timezones{{ + "auto", "default", "CET", "CST6CDT", "Cuba", "EET", "Egypt", "Eire", + "EST", "EST5EDT", "GB", "GB-Eire", "GMT", "GMT+0", "GMT-0", "GMT0", + "Greenwich", "Hongkong", "HST", "Iceland", "Iran", "Israel", "Jamaica", "Japan", + "Kwajalein", "Libya", "MET", "MST", "MST7MDT", "Navajo", "NZ", "NZ-CHAT", + "Poland", "Portugal", "PRC", "PST8PDT", "ROC", "ROK", "Singapore", "Turkey", + "UCT", "Universal", "UTC", "W-SU", "WET", "Zulu", + }}; + + ASSERT(Settings::values.time_zone_index < timezones.size()); + + return timezones[Settings::values.time_zone_index]; +} + void Apply() { GDBStub::SetServerPort(values.gdbstub_port); GDBStub::ToggleServer(values.use_gdbstub); @@ -87,6 +102,7 @@ void LogSettings() { LogSetting("System_CurrentUser", Settings::values.current_user); LogSetting("System_LanguageIndex", Settings::values.language_index); LogSetting("System_RegionIndex", Settings::values.region_index); + LogSetting("System_TimeZoneIndex", Settings::values.time_zone_index); LogSetting("Core_UseMultiCore", Settings::values.use_multi_core); LogSetting("Renderer_UseResolutionFactor", Settings::values.resolution_factor); LogSetting("Renderer_UseFrameLimit", Settings::values.use_frame_limit); @@ -96,6 +112,7 @@ void LogSettings() { LogSetting("Renderer_UseAsynchronousGpuEmulation", Settings::values.use_asynchronous_gpu_emulation); LogSetting("Renderer_UseVsync", Settings::values.use_vsync); + LogSetting("Renderer_UseAssemblyShaders", Settings::values.use_assembly_shaders); LogSetting("Renderer_AnisotropicFilteringLevel", Settings::values.max_anisotropy); LogSetting("Audio_OutputEngine", Settings::values.sink_id); LogSetting("Audio_EnableAudioStretching", Settings::values.enable_audio_stretching); @@ -110,6 +127,13 @@ void LogSettings() { LogSetting("Services_BCATBoxcatLocal", Settings::values.bcat_boxcat_local); } +float Volume() { + if (values.audio_muted) { + return 0.0f; + } + return values.volume; +} + bool IsGPULevelExtreme() { return values.gpu_accuracy == GPUAccuracy::Extreme; } diff --git a/src/core/settings.h b/src/core/settings.h index 163900f0b..a598ccbc1 100644 --- a/src/core/settings.h +++ b/src/core/settings.h @@ -394,6 +394,7 @@ struct Values { s32 current_user; s32 language_index; s32 region_index; + s32 time_zone_index; s32 sound_index; // Controls @@ -436,7 +437,7 @@ struct Values { bool renderer_debug; int vulkan_device; - float resolution_factor; + u16 resolution_factor{1}; int aspect_ratio; int max_anisotropy; bool use_frame_limit; @@ -445,6 +446,7 @@ struct Values { GPUAccuracy gpu_accuracy; bool use_asynchronous_gpu_emulation; bool use_vsync; + bool use_assembly_shaders; bool force_30fps_mode; bool use_fast_gpu_time; @@ -457,6 +459,7 @@ struct Values { bool use_dev_keys; // Audio + bool audio_muted; std::string sink_id; bool enable_audio_stretching; std::string audio_device_id; @@ -472,6 +475,7 @@ struct Values { bool reporting_services; bool quest_flag; bool disable_cpu_opt; + bool disable_macro_jit; // BCAT std::string bcat_backend; @@ -487,9 +491,14 @@ struct Values { std::map<u64, std::vector<std::string>> disabled_addons; } extern values; +float Volume(); + bool IsGPULevelExtreme(); bool IsGPULevelHigh(); +std::string GetTimeZoneString(); + void Apply(); void LogSettings(); + } // namespace Settings diff --git a/src/core/telemetry_session.cpp b/src/core/telemetry_session.cpp index 1c3b03a1c..c781b3cfc 100644 --- a/src/core/telemetry_session.cpp +++ b/src/core/telemetry_session.cpp @@ -201,6 +201,7 @@ void TelemetrySession::AddInitialInfo(Loader::AppLoader& app_loader) { AddField(field_type, "Renderer_UseAsynchronousGpuEmulation", Settings::values.use_asynchronous_gpu_emulation); AddField(field_type, "Renderer_UseVsync", Settings::values.use_vsync); + AddField(field_type, "Renderer_UseAssemblyShaders", Settings::values.use_assembly_shaders); AddField(field_type, "System_UseDockedMode", Settings::values.use_docked_mode); } diff --git a/src/core/tools/freezer.cpp b/src/core/tools/freezer.cpp index b2c6c537e..8b0c50d11 100644 --- a/src/core/tools/freezer.cpp +++ b/src/core/tools/freezer.cpp @@ -14,7 +14,7 @@ namespace Tools { namespace { -constexpr s64 MEMORY_FREEZER_TICKS = static_cast<s64>(Core::Hardware::BASE_CLOCK_RATE / 60); +constexpr s64 MEMORY_FREEZER_TICKS = static_cast<s64>(1000000000 / 60); u64 MemoryReadWidth(Core::Memory::Memory& memory, u32 width, VAddr addr) { switch (width) { @@ -57,7 +57,7 @@ Freezer::Freezer(Core::Timing::CoreTiming& core_timing_, Core::Memory::Memory& m : core_timing{core_timing_}, memory{memory_} { event = Core::Timing::CreateEvent( "MemoryFreezer::FrameCallback", - [this](u64 userdata, s64 cycles_late) { FrameCallback(userdata, cycles_late); }); + [this](u64 userdata, s64 ns_late) { FrameCallback(userdata, ns_late); }); core_timing.ScheduleEvent(MEMORY_FREEZER_TICKS, event); } @@ -158,7 +158,7 @@ std::vector<Freezer::Entry> Freezer::GetEntries() const { return entries; } -void Freezer::FrameCallback(u64 userdata, s64 cycles_late) { +void Freezer::FrameCallback(u64 userdata, s64 ns_late) { if (!IsActive()) { LOG_DEBUG(Common_Memory, "Memory freezer has been deactivated, ending callback events."); return; @@ -173,7 +173,7 @@ void Freezer::FrameCallback(u64 userdata, s64 cycles_late) { MemoryWriteWidth(memory, entry.width, entry.address, entry.value); } - core_timing.ScheduleEvent(MEMORY_FREEZER_TICKS - cycles_late, event); + core_timing.ScheduleEvent(MEMORY_FREEZER_TICKS - ns_late, event); } void Freezer::FillEntryReads() { diff --git a/src/input_common/keyboard.cpp b/src/input_common/keyboard.cpp index 078374be5..afb8e6612 100644 --- a/src/input_common/keyboard.cpp +++ b/src/input_common/keyboard.cpp @@ -76,7 +76,7 @@ std::unique_ptr<Input::ButtonDevice> Keyboard::Create(const Common::ParamPackage int key_code = params.Get("code", 0); std::unique_ptr<KeyButton> button = std::make_unique<KeyButton>(key_button_list); key_button_list->AddKeyButton(key_code, button.get()); - return std::move(button); + return button; } void Keyboard::PressKey(int key_code) { diff --git a/src/input_common/motion_emu.cpp b/src/input_common/motion_emu.cpp index 868251628..d4cdf76a3 100644 --- a/src/input_common/motion_emu.cpp +++ b/src/input_common/motion_emu.cpp @@ -145,7 +145,7 @@ std::unique_ptr<Input::MotionDevice> MotionEmu::Create(const Common::ParamPackag // Previously created device is disconnected here. Having two motion devices for 3DS is not // expected. current_device = device_wrapper->device; - return std::move(device_wrapper); + return device_wrapper; } void MotionEmu::BeginTilt(int x, int y) { diff --git a/src/tests/CMakeLists.txt b/src/tests/CMakeLists.txt index c7038b217..47ef30aa9 100644 --- a/src/tests/CMakeLists.txt +++ b/src/tests/CMakeLists.txt @@ -1,6 +1,7 @@ add_executable(tests common/bit_field.cpp common/bit_utils.cpp + common/fibers.cpp common/multi_level_queue.cpp common/param_package.cpp common/ring_buffer.cpp diff --git a/src/tests/common/fibers.cpp b/src/tests/common/fibers.cpp new file mode 100644 index 000000000..4fd92428f --- /dev/null +++ b/src/tests/common/fibers.cpp @@ -0,0 +1,358 @@ +// Copyright 2020 yuzu Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include <atomic> +#include <cstdlib> +#include <functional> +#include <memory> +#include <thread> +#include <unordered_map> +#include <vector> + +#include <catch2/catch.hpp> +#include <math.h> +#include "common/common_types.h" +#include "common/fiber.h" +#include "common/spin_lock.h" + +namespace Common { + +class TestControl1 { +public: + TestControl1() = default; + + void DoWork(); + + void ExecuteThread(u32 id); + + std::unordered_map<std::thread::id, u32> ids; + std::vector<std::shared_ptr<Common::Fiber>> thread_fibers; + std::vector<std::shared_ptr<Common::Fiber>> work_fibers; + std::vector<u32> items; + std::vector<u32> results; +}; + +static void WorkControl1(void* control) { + auto* test_control = static_cast<TestControl1*>(control); + test_control->DoWork(); +} + +void TestControl1::DoWork() { + std::thread::id this_id = std::this_thread::get_id(); + u32 id = ids[this_id]; + u32 value = items[id]; + for (u32 i = 0; i < id; i++) { + value++; + } + results[id] = value; + Fiber::YieldTo(work_fibers[id], thread_fibers[id]); +} + +void TestControl1::ExecuteThread(u32 id) { + std::thread::id this_id = std::this_thread::get_id(); + ids[this_id] = id; + auto thread_fiber = Fiber::ThreadToFiber(); + thread_fibers[id] = thread_fiber; + work_fibers[id] = std::make_shared<Fiber>(std::function<void(void*)>{WorkControl1}, this); + items[id] = rand() % 256; + Fiber::YieldTo(thread_fibers[id], work_fibers[id]); + thread_fibers[id]->Exit(); +} + +static void ThreadStart1(u32 id, TestControl1& test_control) { + test_control.ExecuteThread(id); +} + +/** This test checks for fiber setup configuration and validates that fibers are + * doing all the work required. + */ +TEST_CASE("Fibers::Setup", "[common]") { + constexpr std::size_t num_threads = 7; + TestControl1 test_control{}; + test_control.thread_fibers.resize(num_threads); + test_control.work_fibers.resize(num_threads); + test_control.items.resize(num_threads, 0); + test_control.results.resize(num_threads, 0); + std::vector<std::thread> threads; + for (u32 i = 0; i < num_threads; i++) { + threads.emplace_back(ThreadStart1, i, std::ref(test_control)); + } + for (u32 i = 0; i < num_threads; i++) { + threads[i].join(); + } + for (u32 i = 0; i < num_threads; i++) { + REQUIRE(test_control.items[i] + i == test_control.results[i]); + } +} + +class TestControl2 { +public: + TestControl2() = default; + + void DoWork1() { + trap2 = false; + while (trap.load()) + ; + for (u32 i = 0; i < 12000; i++) { + value1 += i; + } + Fiber::YieldTo(fiber1, fiber3); + std::thread::id this_id = std::this_thread::get_id(); + u32 id = ids[this_id]; + assert1 = id == 1; + value2 += 5000; + Fiber::YieldTo(fiber1, thread_fibers[id]); + } + + void DoWork2() { + while (trap2.load()) + ; + value2 = 2000; + trap = false; + Fiber::YieldTo(fiber2, fiber1); + assert3 = false; + } + + void DoWork3() { + std::thread::id this_id = std::this_thread::get_id(); + u32 id = ids[this_id]; + assert2 = id == 0; + value1 += 1000; + Fiber::YieldTo(fiber3, thread_fibers[id]); + } + + void ExecuteThread(u32 id); + + void CallFiber1() { + std::thread::id this_id = std::this_thread::get_id(); + u32 id = ids[this_id]; + Fiber::YieldTo(thread_fibers[id], fiber1); + } + + void CallFiber2() { + std::thread::id this_id = std::this_thread::get_id(); + u32 id = ids[this_id]; + Fiber::YieldTo(thread_fibers[id], fiber2); + } + + void Exit(); + + bool assert1{}; + bool assert2{}; + bool assert3{true}; + u32 value1{}; + u32 value2{}; + std::atomic<bool> trap{true}; + std::atomic<bool> trap2{true}; + std::unordered_map<std::thread::id, u32> ids; + std::vector<std::shared_ptr<Common::Fiber>> thread_fibers; + std::shared_ptr<Common::Fiber> fiber1; + std::shared_ptr<Common::Fiber> fiber2; + std::shared_ptr<Common::Fiber> fiber3; +}; + +static void WorkControl2_1(void* control) { + auto* test_control = static_cast<TestControl2*>(control); + test_control->DoWork1(); +} + +static void WorkControl2_2(void* control) { + auto* test_control = static_cast<TestControl2*>(control); + test_control->DoWork2(); +} + +static void WorkControl2_3(void* control) { + auto* test_control = static_cast<TestControl2*>(control); + test_control->DoWork3(); +} + +void TestControl2::ExecuteThread(u32 id) { + std::thread::id this_id = std::this_thread::get_id(); + ids[this_id] = id; + auto thread_fiber = Fiber::ThreadToFiber(); + thread_fibers[id] = thread_fiber; +} + +void TestControl2::Exit() { + std::thread::id this_id = std::this_thread::get_id(); + u32 id = ids[this_id]; + thread_fibers[id]->Exit(); +} + +static void ThreadStart2_1(u32 id, TestControl2& test_control) { + test_control.ExecuteThread(id); + test_control.CallFiber1(); + test_control.Exit(); +} + +static void ThreadStart2_2(u32 id, TestControl2& test_control) { + test_control.ExecuteThread(id); + test_control.CallFiber2(); + test_control.Exit(); +} + +/** This test checks for fiber thread exchange configuration and validates that fibers are + * that a fiber has been succesfully transfered from one thread to another and that the TLS + * region of the thread is kept while changing fibers. + */ +TEST_CASE("Fibers::InterExchange", "[common]") { + TestControl2 test_control{}; + test_control.thread_fibers.resize(2); + test_control.fiber1 = + std::make_shared<Fiber>(std::function<void(void*)>{WorkControl2_1}, &test_control); + test_control.fiber2 = + std::make_shared<Fiber>(std::function<void(void*)>{WorkControl2_2}, &test_control); + test_control.fiber3 = + std::make_shared<Fiber>(std::function<void(void*)>{WorkControl2_3}, &test_control); + std::thread thread1(ThreadStart2_1, 0, std::ref(test_control)); + std::thread thread2(ThreadStart2_2, 1, std::ref(test_control)); + thread1.join(); + thread2.join(); + REQUIRE(test_control.assert1); + REQUIRE(test_control.assert2); + REQUIRE(test_control.assert3); + REQUIRE(test_control.value2 == 7000); + u32 cal_value = 0; + for (u32 i = 0; i < 12000; i++) { + cal_value += i; + } + cal_value += 1000; + REQUIRE(test_control.value1 == cal_value); +} + +class TestControl3 { +public: + TestControl3() = default; + + void DoWork1() { + value1 += 1; + Fiber::YieldTo(fiber1, fiber2); + std::thread::id this_id = std::this_thread::get_id(); + u32 id = ids[this_id]; + value3 += 1; + Fiber::YieldTo(fiber1, thread_fibers[id]); + } + + void DoWork2() { + value2 += 1; + std::thread::id this_id = std::this_thread::get_id(); + u32 id = ids[this_id]; + Fiber::YieldTo(fiber2, thread_fibers[id]); + } + + void ExecuteThread(u32 id); + + void CallFiber1() { + std::thread::id this_id = std::this_thread::get_id(); + u32 id = ids[this_id]; + Fiber::YieldTo(thread_fibers[id], fiber1); + } + + void Exit(); + + u32 value1{}; + u32 value2{}; + u32 value3{}; + std::unordered_map<std::thread::id, u32> ids; + std::vector<std::shared_ptr<Common::Fiber>> thread_fibers; + std::shared_ptr<Common::Fiber> fiber1; + std::shared_ptr<Common::Fiber> fiber2; +}; + +static void WorkControl3_1(void* control) { + auto* test_control = static_cast<TestControl3*>(control); + test_control->DoWork1(); +} + +static void WorkControl3_2(void* control) { + auto* test_control = static_cast<TestControl3*>(control); + test_control->DoWork2(); +} + +void TestControl3::ExecuteThread(u32 id) { + std::thread::id this_id = std::this_thread::get_id(); + ids[this_id] = id; + auto thread_fiber = Fiber::ThreadToFiber(); + thread_fibers[id] = thread_fiber; +} + +void TestControl3::Exit() { + std::thread::id this_id = std::this_thread::get_id(); + u32 id = ids[this_id]; + thread_fibers[id]->Exit(); +} + +static void ThreadStart3(u32 id, TestControl3& test_control) { + test_control.ExecuteThread(id); + test_control.CallFiber1(); + test_control.Exit(); +} + +/** This test checks for one two threads racing for starting the same fiber. + * It checks execution occured in an ordered manner and by no time there were + * two contexts at the same time. + */ +TEST_CASE("Fibers::StartRace", "[common]") { + TestControl3 test_control{}; + test_control.thread_fibers.resize(2); + test_control.fiber1 = + std::make_shared<Fiber>(std::function<void(void*)>{WorkControl3_1}, &test_control); + test_control.fiber2 = + std::make_shared<Fiber>(std::function<void(void*)>{WorkControl3_2}, &test_control); + std::thread thread1(ThreadStart3, 0, std::ref(test_control)); + std::thread thread2(ThreadStart3, 1, std::ref(test_control)); + thread1.join(); + thread2.join(); + REQUIRE(test_control.value1 == 1); + REQUIRE(test_control.value2 == 1); + REQUIRE(test_control.value3 == 1); +} + +class TestControl4; + +static void WorkControl4(void* control); + +class TestControl4 { +public: + TestControl4() { + fiber1 = std::make_shared<Fiber>(std::function<void(void*)>{WorkControl4}, this); + goal_reached = false; + rewinded = false; + } + + void Execute() { + thread_fiber = Fiber::ThreadToFiber(); + Fiber::YieldTo(thread_fiber, fiber1); + thread_fiber->Exit(); + } + + void DoWork() { + fiber1->SetRewindPoint(std::function<void(void*)>{WorkControl4}, this); + if (rewinded) { + goal_reached = true; + Fiber::YieldTo(fiber1, thread_fiber); + } + rewinded = true; + fiber1->Rewind(); + } + + std::shared_ptr<Common::Fiber> fiber1; + std::shared_ptr<Common::Fiber> thread_fiber; + bool goal_reached; + bool rewinded; +}; + +static void WorkControl4(void* control) { + auto* test_control = static_cast<TestControl4*>(control); + test_control->DoWork(); +} + +TEST_CASE("Fibers::Rewind", "[common]") { + TestControl4 test_control{}; + test_control.Execute(); + REQUIRE(test_control.goal_reached); + REQUIRE(test_control.rewinded); +} + +} // namespace Common diff --git a/src/tests/core/core_timing.cpp b/src/tests/core/core_timing.cpp index ff2d11cc8..e66db1940 100644 --- a/src/tests/core/core_timing.cpp +++ b/src/tests/core/core_timing.cpp @@ -18,29 +18,26 @@ namespace { // Numbers are chosen randomly to make sure the correct one is given. constexpr std::array<u64, 5> CB_IDS{{42, 144, 93, 1026, UINT64_C(0xFFFF7FFFF7FFFF)}}; constexpr int MAX_SLICE_LENGTH = 10000; // Copied from CoreTiming internals +constexpr std::array<u64, 5> calls_order{{2, 0, 1, 4, 3}}; +std::array<s64, 5> delays{}; std::bitset<CB_IDS.size()> callbacks_ran_flags; u64 expected_callback = 0; -s64 lateness = 0; template <unsigned int IDX> -void CallbackTemplate(u64 userdata, s64 cycles_late) { +void HostCallbackTemplate(u64 userdata, s64 nanoseconds_late) { static_assert(IDX < CB_IDS.size(), "IDX out of range"); callbacks_ran_flags.set(IDX); REQUIRE(CB_IDS[IDX] == userdata); - REQUIRE(CB_IDS[IDX] == expected_callback); - REQUIRE(lateness == cycles_late); -} - -u64 callbacks_done = 0; - -void EmptyCallback(u64 userdata, s64 cycles_late) { - ++callbacks_done; + REQUIRE(CB_IDS[IDX] == CB_IDS[calls_order[expected_callback]]); + delays[IDX] = nanoseconds_late; + ++expected_callback; } struct ScopeInit final { ScopeInit() { - core_timing.Initialize(); + core_timing.SetMulticore(true); + core_timing.Initialize([]() {}); } ~ScopeInit() { core_timing.Shutdown(); @@ -49,110 +46,101 @@ struct ScopeInit final { Core::Timing::CoreTiming core_timing; }; -void AdvanceAndCheck(Core::Timing::CoreTiming& core_timing, u32 idx, u32 context = 0, - int expected_lateness = 0, int cpu_downcount = 0) { - callbacks_ran_flags = 0; - expected_callback = CB_IDS[idx]; - lateness = expected_lateness; - - // Pretend we executed X cycles of instructions. - core_timing.SwitchContext(context); - core_timing.AddTicks(core_timing.GetDowncount() - cpu_downcount); - core_timing.Advance(); - core_timing.SwitchContext((context + 1) % 4); +#pragma optimize("", off) - REQUIRE(decltype(callbacks_ran_flags)().set(idx) == callbacks_ran_flags); +u64 TestTimerSpeed(Core::Timing::CoreTiming& core_timing) { + u64 start = core_timing.GetGlobalTimeNs().count(); + u64 placebo = 0; + for (std::size_t i = 0; i < 1000; i++) { + placebo += core_timing.GetGlobalTimeNs().count(); + } + u64 end = core_timing.GetGlobalTimeNs().count(); + return (end - start); } + +#pragma optimize("", on) + } // Anonymous namespace TEST_CASE("CoreTiming[BasicOrder]", "[core]") { ScopeInit guard; auto& core_timing = guard.core_timing; + std::vector<std::shared_ptr<Core::Timing::EventType>> events{ + Core::Timing::CreateEvent("callbackA", HostCallbackTemplate<0>), + Core::Timing::CreateEvent("callbackB", HostCallbackTemplate<1>), + Core::Timing::CreateEvent("callbackC", HostCallbackTemplate<2>), + Core::Timing::CreateEvent("callbackD", HostCallbackTemplate<3>), + Core::Timing::CreateEvent("callbackE", HostCallbackTemplate<4>), + }; + + expected_callback = 0; + + core_timing.SyncPause(true); + + u64 one_micro = 1000U; + for (std::size_t i = 0; i < events.size(); i++) { + u64 order = calls_order[i]; + core_timing.ScheduleEvent(i * one_micro + 100U, events[order], CB_IDS[order]); + } + /// test pause + REQUIRE(callbacks_ran_flags.none()); - std::shared_ptr<Core::Timing::EventType> cb_a = - Core::Timing::CreateEvent("callbackA", CallbackTemplate<0>); - std::shared_ptr<Core::Timing::EventType> cb_b = - Core::Timing::CreateEvent("callbackB", CallbackTemplate<1>); - std::shared_ptr<Core::Timing::EventType> cb_c = - Core::Timing::CreateEvent("callbackC", CallbackTemplate<2>); - std::shared_ptr<Core::Timing::EventType> cb_d = - Core::Timing::CreateEvent("callbackD", CallbackTemplate<3>); - std::shared_ptr<Core::Timing::EventType> cb_e = - Core::Timing::CreateEvent("callbackE", CallbackTemplate<4>); - - // Enter slice 0 - core_timing.ResetRun(); - - // D -> B -> C -> A -> E - core_timing.SwitchContext(0); - core_timing.ScheduleEvent(1000, cb_a, CB_IDS[0]); - REQUIRE(1000 == core_timing.GetDowncount()); - core_timing.ScheduleEvent(500, cb_b, CB_IDS[1]); - REQUIRE(500 == core_timing.GetDowncount()); - core_timing.ScheduleEvent(800, cb_c, CB_IDS[2]); - REQUIRE(500 == core_timing.GetDowncount()); - core_timing.ScheduleEvent(100, cb_d, CB_IDS[3]); - REQUIRE(100 == core_timing.GetDowncount()); - core_timing.ScheduleEvent(1200, cb_e, CB_IDS[4]); - REQUIRE(100 == core_timing.GetDowncount()); - - AdvanceAndCheck(core_timing, 3, 0); - AdvanceAndCheck(core_timing, 1, 1); - AdvanceAndCheck(core_timing, 2, 2); - AdvanceAndCheck(core_timing, 0, 3); - AdvanceAndCheck(core_timing, 4, 0); -} - -TEST_CASE("CoreTiming[FairSharing]", "[core]") { + core_timing.Pause(false); // No need to sync - ScopeInit guard; - auto& core_timing = guard.core_timing; + while (core_timing.HasPendingEvents()) + ; - std::shared_ptr<Core::Timing::EventType> empty_callback = - Core::Timing::CreateEvent("empty_callback", EmptyCallback); + REQUIRE(callbacks_ran_flags.all()); - callbacks_done = 0; - u64 MAX_CALLBACKS = 10; - for (std::size_t i = 0; i < 10; i++) { - core_timing.ScheduleEvent(i * 3333U, empty_callback, 0); + for (std::size_t i = 0; i < delays.size(); i++) { + const double delay = static_cast<double>(delays[i]); + const double micro = delay / 1000.0f; + const double mili = micro / 1000.0f; + printf("HostTimer Pausing Delay[%zu]: %.3f %.6f\n", i, micro, mili); } - - const s64 advances = MAX_SLICE_LENGTH / 10; - core_timing.ResetRun(); - u64 current_time = core_timing.GetTicks(); - bool keep_running{}; - do { - keep_running = false; - for (u32 active_core = 0; active_core < 4; ++active_core) { - core_timing.SwitchContext(active_core); - if (core_timing.CanCurrentContextRun()) { - core_timing.AddTicks(std::min<s64>(advances, core_timing.GetDowncount())); - core_timing.Advance(); - } - keep_running |= core_timing.CanCurrentContextRun(); - } - } while (keep_running); - u64 current_time_2 = core_timing.GetTicks(); - - REQUIRE(MAX_CALLBACKS == callbacks_done); - REQUIRE(current_time_2 == current_time + MAX_SLICE_LENGTH * 4); } -TEST_CASE("Core::Timing[PredictableLateness]", "[core]") { +TEST_CASE("CoreTiming[BasicOrderNoPausing]", "[core]") { ScopeInit guard; auto& core_timing = guard.core_timing; + std::vector<std::shared_ptr<Core::Timing::EventType>> events{ + Core::Timing::CreateEvent("callbackA", HostCallbackTemplate<0>), + Core::Timing::CreateEvent("callbackB", HostCallbackTemplate<1>), + Core::Timing::CreateEvent("callbackC", HostCallbackTemplate<2>), + Core::Timing::CreateEvent("callbackD", HostCallbackTemplate<3>), + Core::Timing::CreateEvent("callbackE", HostCallbackTemplate<4>), + }; + + core_timing.SyncPause(true); + core_timing.SyncPause(false); + + expected_callback = 0; + + u64 start = core_timing.GetGlobalTimeNs().count(); + u64 one_micro = 1000U; + for (std::size_t i = 0; i < events.size(); i++) { + u64 order = calls_order[i]; + core_timing.ScheduleEvent(i * one_micro + 100U, events[order], CB_IDS[order]); + } + u64 end = core_timing.GetGlobalTimeNs().count(); + const double scheduling_time = static_cast<double>(end - start); + const double timer_time = static_cast<double>(TestTimerSpeed(core_timing)); - std::shared_ptr<Core::Timing::EventType> cb_a = - Core::Timing::CreateEvent("callbackA", CallbackTemplate<0>); - std::shared_ptr<Core::Timing::EventType> cb_b = - Core::Timing::CreateEvent("callbackB", CallbackTemplate<1>); + while (core_timing.HasPendingEvents()) + ; - // Enter slice 0 - core_timing.ResetRun(); + REQUIRE(callbacks_ran_flags.all()); - core_timing.ScheduleEvent(100, cb_a, CB_IDS[0]); - core_timing.ScheduleEvent(200, cb_b, CB_IDS[1]); + for (std::size_t i = 0; i < delays.size(); i++) { + const double delay = static_cast<double>(delays[i]); + const double micro = delay / 1000.0f; + const double mili = micro / 1000.0f; + printf("HostTimer No Pausing Delay[%zu]: %.3f %.6f\n", i, micro, mili); + } - AdvanceAndCheck(core_timing, 0, 0, 10, -10); // (100 - 10) - AdvanceAndCheck(core_timing, 1, 1, 50, -50); + const double micro = scheduling_time / 1000.0f; + const double mili = micro / 1000.0f; + printf("HostTimer No Pausing Scheduling Time: %.3f %.6f\n", micro, mili); + printf("HostTimer No Pausing Timer Time: %.3f %.6f\n", timer_time / 1000.f, + timer_time / 1000000.f); } diff --git a/src/video_core/CMakeLists.txt b/src/video_core/CMakeLists.txt index d23c53843..21c46a567 100644 --- a/src/video_core/CMakeLists.txt +++ b/src/video_core/CMakeLists.txt @@ -1,7 +1,10 @@ add_library(video_core STATIC buffer_cache/buffer_block.h buffer_cache/buffer_cache.h + buffer_cache/map_interval.cpp buffer_cache/map_interval.h + compatible_formats.cpp + compatible_formats.h dirty_flags.cpp dirty_flags.h dma_pusher.cpp @@ -24,6 +27,14 @@ add_library(video_core STATIC engines/shader_bytecode.h engines/shader_header.h engines/shader_type.h + macro/macro.cpp + macro/macro.h + macro/macro_hle.cpp + macro/macro_hle.h + macro/macro_interpreter.cpp + macro/macro_interpreter.h + macro/macro_jit_x64.cpp + macro/macro_jit_x64.h fence_manager.h gpu.cpp gpu.h @@ -35,8 +46,6 @@ add_library(video_core STATIC gpu_thread.h guest_driver.cpp guest_driver.h - macro_interpreter.cpp - macro_interpreter.h memory_manager.cpp memory_manager.h morton.cpp @@ -44,11 +53,11 @@ add_library(video_core STATIC query_cache.h rasterizer_accelerated.cpp rasterizer_accelerated.h - rasterizer_cache.cpp - rasterizer_cache.h rasterizer_interface.h renderer_base.cpp renderer_base.h + renderer_opengl/gl_arb_decompiler.cpp + renderer_opengl/gl_arb_decompiler.h renderer_opengl/gl_buffer_cache.cpp renderer_opengl/gl_buffer_cache.h renderer_opengl/gl_device.cpp @@ -88,6 +97,7 @@ add_library(video_core STATIC renderer_opengl/utils.h sampler_cache.cpp sampler_cache.h + shader_cache.h shader/decode/arithmetic.cpp shader/decode/arithmetic_immediate.cpp shader/decode/bfe.cpp @@ -228,7 +238,7 @@ endif() create_target_directory_groups(video_core) target_link_libraries(video_core PUBLIC common core) -target_link_libraries(video_core PRIVATE glad) +target_link_libraries(video_core PRIVATE glad xbyak) if (ENABLE_VULKAN) target_include_directories(video_core PRIVATE sirit ../../externals/Vulkan-Headers/include) diff --git a/src/video_core/buffer_cache/buffer_block.h b/src/video_core/buffer_cache/buffer_block.h index e35ee0b67..e64170e66 100644 --- a/src/video_core/buffer_cache/buffer_block.h +++ b/src/video_core/buffer_cache/buffer_block.h @@ -15,48 +15,47 @@ namespace VideoCommon { class BufferBlock { public: - bool Overlaps(const VAddr start, const VAddr end) const { + bool Overlaps(VAddr start, VAddr end) const { return (cpu_addr < end) && (cpu_addr_end > start); } - bool IsInside(const VAddr other_start, const VAddr other_end) const { + bool IsInside(VAddr other_start, VAddr other_end) const { return cpu_addr <= other_start && other_end <= cpu_addr_end; } - std::size_t GetOffset(const VAddr in_addr) { + std::size_t Offset(VAddr in_addr) const { return static_cast<std::size_t>(in_addr - cpu_addr); } - VAddr GetCpuAddr() const { + VAddr CpuAddr() const { return cpu_addr; } - VAddr GetCpuAddrEnd() const { + VAddr CpuAddrEnd() const { return cpu_addr_end; } - void SetCpuAddr(const VAddr new_addr) { + void SetCpuAddr(VAddr new_addr) { cpu_addr = new_addr; cpu_addr_end = new_addr + size; } - std::size_t GetSize() const { + std::size_t Size() const { return size; } - void SetEpoch(u64 new_epoch) { - epoch = new_epoch; + u64 Epoch() const { + return epoch; } - u64 GetEpoch() { - return epoch; + void SetEpoch(u64 new_epoch) { + epoch = new_epoch; } protected: - explicit BufferBlock(VAddr cpu_addr, const std::size_t size) : size{size} { - SetCpuAddr(cpu_addr); + explicit BufferBlock(VAddr cpu_addr_, std::size_t size_) : size{size_} { + SetCpuAddr(cpu_addr_); } - ~BufferBlock() = default; private: VAddr cpu_addr{}; diff --git a/src/video_core/buffer_cache/buffer_cache.h b/src/video_core/buffer_cache/buffer_cache.h index 56e570994..cf8bdd021 100644 --- a/src/video_core/buffer_cache/buffer_cache.h +++ b/src/video_core/buffer_cache/buffer_cache.h @@ -12,11 +12,12 @@ #include <utility> #include <vector> -#include <boost/icl/interval_map.hpp> +#include <boost/container/small_vector.hpp> #include <boost/icl/interval_set.hpp> -#include <boost/range/iterator_range.hpp> +#include <boost/intrusive/set.hpp> #include "common/alignment.h" +#include "common/assert.h" #include "common/common_types.h" #include "common/logging/log.h" #include "core/core.h" @@ -29,102 +30,124 @@ namespace VideoCommon { -using MapInterval = std::shared_ptr<MapIntervalBase>; - -template <typename OwnerBuffer, typename BufferType, typename StreamBuffer> +template <typename Buffer, typename BufferType, typename StreamBuffer> class BufferCache { + using IntervalSet = boost::icl::interval_set<VAddr>; + using IntervalType = typename IntervalSet::interval_type; + using VectorMapInterval = boost::container::small_vector<MapInterval*, 1>; + + static constexpr u64 WRITE_PAGE_BIT = 11; + static constexpr u64 BLOCK_PAGE_BITS = 21; + static constexpr u64 BLOCK_PAGE_SIZE = 1ULL << BLOCK_PAGE_BITS; + public: - using BufferInfo = std::pair<BufferType, u64>; + struct BufferInfo { + BufferType handle; + u64 offset; + u64 address; + }; BufferInfo UploadMemory(GPUVAddr gpu_addr, std::size_t size, std::size_t alignment = 4, bool is_written = false, bool use_fast_cbuf = false) { std::lock_guard lock{mutex}; - const std::optional<VAddr> cpu_addr_opt = - system.GPU().MemoryManager().GpuToCpuAddress(gpu_addr); - + auto& memory_manager = system.GPU().MemoryManager(); + const std::optional<VAddr> cpu_addr_opt = memory_manager.GpuToCpuAddress(gpu_addr); if (!cpu_addr_opt) { - return {GetEmptyBuffer(size), 0}; + return GetEmptyBuffer(size); } - - VAddr cpu_addr = *cpu_addr_opt; + const VAddr cpu_addr = *cpu_addr_opt; // Cache management is a big overhead, so only cache entries with a given size. // TODO: Figure out which size is the best for given games. constexpr std::size_t max_stream_size = 0x800; if (use_fast_cbuf || size < max_stream_size) { if (!is_written && !IsRegionWritten(cpu_addr, cpu_addr + size - 1)) { - auto& memory_manager = system.GPU().MemoryManager(); + const bool is_granular = memory_manager.IsGranularRange(gpu_addr, size); if (use_fast_cbuf) { - if (memory_manager.IsGranularRange(gpu_addr, size)) { - const auto host_ptr = memory_manager.GetPointer(gpu_addr); - return ConstBufferUpload(host_ptr, size); + u8* dest; + if (is_granular) { + dest = memory_manager.GetPointer(gpu_addr); } else { staging_buffer.resize(size); - memory_manager.ReadBlockUnsafe(gpu_addr, staging_buffer.data(), size); - return ConstBufferUpload(staging_buffer.data(), size); + dest = staging_buffer.data(); + memory_manager.ReadBlockUnsafe(gpu_addr, dest, size); } + return ConstBufferUpload(dest, size); + } + if (is_granular) { + u8* const host_ptr = memory_manager.GetPointer(gpu_addr); + return StreamBufferUpload(size, alignment, [host_ptr, size](u8* dest) { + std::memcpy(dest, host_ptr, size); + }); } else { - if (memory_manager.IsGranularRange(gpu_addr, size)) { - const auto host_ptr = memory_manager.GetPointer(gpu_addr); - return StreamBufferUpload(host_ptr, size, alignment); - } else { - staging_buffer.resize(size); - memory_manager.ReadBlockUnsafe(gpu_addr, staging_buffer.data(), size); - return StreamBufferUpload(staging_buffer.data(), size, alignment); - } + return StreamBufferUpload( + size, alignment, [&memory_manager, gpu_addr, size](u8* dest) { + memory_manager.ReadBlockUnsafe(gpu_addr, dest, size); + }); } } } - auto block = GetBlock(cpu_addr, size); - auto map = MapAddress(block, gpu_addr, cpu_addr, size); + Buffer* const block = GetBlock(cpu_addr, size); + MapInterval* const map = MapAddress(block, gpu_addr, cpu_addr, size); + if (!map) { + return GetEmptyBuffer(size); + } if (is_written) { map->MarkAsModified(true, GetModifiedTicks()); if (Settings::IsGPULevelHigh() && Settings::values.use_asynchronous_gpu_emulation) { MarkForAsyncFlush(map); } - if (!map->IsWritten()) { - map->MarkAsWritten(true); - MarkRegionAsWritten(map->GetStart(), map->GetEnd() - 1); + if (!map->is_written) { + map->is_written = true; + MarkRegionAsWritten(map->start, map->end - 1); } } - return {ToHandle(block), static_cast<u64>(block->GetOffset(cpu_addr))}; + return BufferInfo{block->Handle(), block->Offset(cpu_addr), block->Address()}; } /// Uploads from a host memory. Returns the OpenGL buffer where it's located and its offset. BufferInfo UploadHostMemory(const void* raw_pointer, std::size_t size, std::size_t alignment = 4) { std::lock_guard lock{mutex}; - return StreamBufferUpload(raw_pointer, size, alignment); + return StreamBufferUpload(size, alignment, [raw_pointer, size](u8* dest) { + std::memcpy(dest, raw_pointer, size); + }); } - void Map(std::size_t max_size) { + /// Prepares the buffer cache for data uploading + /// @param max_size Maximum number of bytes that will be uploaded + /// @return True when a stream buffer invalidation was required, false otherwise + bool Map(std::size_t max_size) { std::lock_guard lock{mutex}; + bool invalidated; std::tie(buffer_ptr, buffer_offset_base, invalidated) = stream_buffer->Map(max_size, 4); buffer_offset = buffer_offset_base; + + return invalidated; } - /// Finishes the upload stream, returns true on bindings invalidation. - bool Unmap() { + /// Finishes the upload stream + void Unmap() { std::lock_guard lock{mutex}; - stream_buffer->Unmap(buffer_offset - buffer_offset_base); - return std::exchange(invalidated, false); } + /// Function called at the end of each frame, inteded for deferred operations void TickFrame() { ++epoch; + while (!pending_destruction.empty()) { // Delay at least 4 frames before destruction. // This is due to triple buffering happening on some drivers. static constexpr u64 epochs_to_destroy = 5; - if (pending_destruction.front()->GetEpoch() + epochs_to_destroy > epoch) { + if (pending_destruction.front()->Epoch() + epochs_to_destroy > epoch) { break; } - pending_destruction.pop_front(); + pending_destruction.pop(); } } @@ -132,12 +155,11 @@ public: void FlushRegion(VAddr addr, std::size_t size) { std::lock_guard lock{mutex}; - std::vector<MapInterval> objects = GetMapsInRange(addr, size); - std::sort(objects.begin(), objects.end(), [](const MapInterval& a, const MapInterval& b) { - return a->GetModificationTick() < b->GetModificationTick(); - }); - for (auto& object : objects) { - if (object->IsModified() && object->IsRegistered()) { + VectorMapInterval objects = GetMapsInRange(addr, size); + std::sort(objects.begin(), objects.end(), + [](MapInterval* lhs, MapInterval* rhs) { return lhs->ticks < rhs->ticks; }); + for (MapInterval* object : objects) { + if (object->is_modified && object->is_registered) { mutex.unlock(); FlushMap(object); mutex.lock(); @@ -148,9 +170,9 @@ public: bool MustFlushRegion(VAddr addr, std::size_t size) { std::lock_guard lock{mutex}; - const std::vector<MapInterval> objects = GetMapsInRange(addr, size); - return std::any_of(objects.cbegin(), objects.cend(), [](const MapInterval& map) { - return map->IsModified() && map->IsRegistered(); + const VectorMapInterval objects = GetMapsInRange(addr, size); + return std::any_of(objects.cbegin(), objects.cend(), [](const MapInterval* map) { + return map->is_modified && map->is_registered; }); } @@ -158,9 +180,8 @@ public: void InvalidateRegion(VAddr addr, u64 size) { std::lock_guard lock{mutex}; - std::vector<MapInterval> objects = GetMapsInRange(addr, size); - for (auto& object : objects) { - if (object->IsRegistered()) { + for (auto& object : GetMapsInRange(addr, size)) { + if (object->is_registered) { Unregister(object); } } @@ -169,10 +190,10 @@ public: void OnCPUWrite(VAddr addr, std::size_t size) { std::lock_guard lock{mutex}; - for (const auto& object : GetMapsInRange(addr, size)) { - if (object->IsMemoryMarked() && object->IsRegistered()) { + for (MapInterval* object : GetMapsInRange(addr, size)) { + if (object->is_memory_marked && object->is_registered) { UnmarkMemory(object); - object->SetSyncPending(true); + object->is_sync_pending = true; marked_for_unregister.emplace_back(object); } } @@ -181,9 +202,9 @@ public: void SyncGuestHost() { std::lock_guard lock{mutex}; - for (const auto& object : marked_for_unregister) { - if (object->IsRegistered()) { - object->SetSyncPending(false); + for (auto& object : marked_for_unregister) { + if (object->is_registered) { + object->is_sync_pending = false; Unregister(object); } } @@ -192,9 +213,9 @@ public: void CommitAsyncFlushes() { if (uncommitted_flushes) { - auto commit_list = std::make_shared<std::list<MapInterval>>(); - for (auto& map : *uncommitted_flushes) { - if (map->IsRegistered() && map->IsModified()) { + auto commit_list = std::make_shared<std::list<MapInterval*>>(); + for (MapInterval* map : *uncommitted_flushes) { + if (map->is_registered && map->is_modified) { // TODO(Blinkhawk): Implement backend asynchronous flushing // AsyncFlushMap(map) commit_list->push_back(map); @@ -228,8 +249,8 @@ public: committed_flushes.pop_front(); return; } - for (MapInterval& map : *flush_list) { - if (map->IsRegistered()) { + for (MapInterval* map : *flush_list) { + if (map->is_registered) { // TODO(Blinkhawk): Replace this for reading the asynchronous flush FlushMap(map); } @@ -237,104 +258,90 @@ public: committed_flushes.pop_front(); } - virtual BufferType GetEmptyBuffer(std::size_t size) = 0; + virtual BufferInfo GetEmptyBuffer(std::size_t size) = 0; protected: explicit BufferCache(VideoCore::RasterizerInterface& rasterizer, Core::System& system, std::unique_ptr<StreamBuffer> stream_buffer) - : rasterizer{rasterizer}, system{system}, stream_buffer{std::move(stream_buffer)}, - stream_buffer_handle{this->stream_buffer->GetHandle()} {} + : rasterizer{rasterizer}, system{system}, stream_buffer{std::move(stream_buffer)} {} ~BufferCache() = default; - virtual BufferType ToHandle(const OwnerBuffer& storage) = 0; - - virtual OwnerBuffer CreateBlock(VAddr cpu_addr, std::size_t size) = 0; - - virtual void UploadBlockData(const OwnerBuffer& buffer, std::size_t offset, std::size_t size, - const u8* data) = 0; - - virtual void DownloadBlockData(const OwnerBuffer& buffer, std::size_t offset, std::size_t size, - u8* data) = 0; - - virtual void CopyBlock(const OwnerBuffer& src, const OwnerBuffer& dst, std::size_t src_offset, - std::size_t dst_offset, std::size_t size) = 0; + virtual std::shared_ptr<Buffer> CreateBlock(VAddr cpu_addr, std::size_t size) = 0; virtual BufferInfo ConstBufferUpload(const void* raw_pointer, std::size_t size) { return {}; } /// Register an object into the cache - void Register(const MapInterval& new_map, bool inherit_written = false) { - const VAddr cpu_addr = new_map->GetStart(); + MapInterval* Register(MapInterval new_map, bool inherit_written = false) { + const VAddr cpu_addr = new_map.start; if (!cpu_addr) { LOG_CRITICAL(HW_GPU, "Failed to register buffer with unmapped gpu_address 0x{:016x}", - new_map->GetGpuAddress()); - return; + new_map.gpu_addr); + return nullptr; } - const std::size_t size = new_map->GetEnd() - new_map->GetStart(); - new_map->MarkAsRegistered(true); - const IntervalType interval{new_map->GetStart(), new_map->GetEnd()}; - mapped_addresses.insert({interval, new_map}); + const std::size_t size = new_map.end - new_map.start; + new_map.is_registered = true; rasterizer.UpdatePagesCachedCount(cpu_addr, size, 1); - new_map->SetMemoryMarked(true); + new_map.is_memory_marked = true; if (inherit_written) { - MarkRegionAsWritten(new_map->GetStart(), new_map->GetEnd() - 1); - new_map->MarkAsWritten(true); + MarkRegionAsWritten(new_map.start, new_map.end - 1); + new_map.is_written = true; } + MapInterval* const storage = mapped_addresses_allocator.Allocate(); + *storage = new_map; + mapped_addresses.insert(*storage); + return storage; } - void UnmarkMemory(const MapInterval& map) { - if (!map->IsMemoryMarked()) { + void UnmarkMemory(MapInterval* map) { + if (!map->is_memory_marked) { return; } - const std::size_t size = map->GetEnd() - map->GetStart(); - rasterizer.UpdatePagesCachedCount(map->GetStart(), size, -1); - map->SetMemoryMarked(false); + const std::size_t size = map->end - map->start; + rasterizer.UpdatePagesCachedCount(map->start, size, -1); + map->is_memory_marked = false; } /// Unregisters an object from the cache - void Unregister(const MapInterval& map) { + void Unregister(MapInterval* map) { UnmarkMemory(map); - map->MarkAsRegistered(false); - if (map->IsSyncPending()) { + map->is_registered = false; + if (map->is_sync_pending) { + map->is_sync_pending = false; marked_for_unregister.remove(map); - map->SetSyncPending(false); } - if (map->IsWritten()) { - UnmarkRegionAsWritten(map->GetStart(), map->GetEnd() - 1); + if (map->is_written) { + UnmarkRegionAsWritten(map->start, map->end - 1); } - const IntervalType delete_interval{map->GetStart(), map->GetEnd()}; - mapped_addresses.erase(delete_interval); + const auto it = mapped_addresses.find(*map); + ASSERT(it != mapped_addresses.end()); + mapped_addresses.erase(it); + mapped_addresses_allocator.Release(map); } private: - MapInterval CreateMap(const VAddr start, const VAddr end, const GPUVAddr gpu_addr) { - return std::make_shared<MapIntervalBase>(start, end, gpu_addr); - } - - MapInterval MapAddress(const OwnerBuffer& block, const GPUVAddr gpu_addr, const VAddr cpu_addr, - const std::size_t size) { - std::vector<MapInterval> overlaps = GetMapsInRange(cpu_addr, size); + MapInterval* MapAddress(const Buffer* block, GPUVAddr gpu_addr, VAddr cpu_addr, + std::size_t size) { + const VectorMapInterval overlaps = GetMapsInRange(cpu_addr, size); if (overlaps.empty()) { auto& memory_manager = system.GPU().MemoryManager(); const VAddr cpu_addr_end = cpu_addr + size; - MapInterval new_map = CreateMap(cpu_addr, cpu_addr_end, gpu_addr); if (memory_manager.IsGranularRange(gpu_addr, size)) { u8* host_ptr = memory_manager.GetPointer(gpu_addr); - UploadBlockData(block, block->GetOffset(cpu_addr), size, host_ptr); + block->Upload(block->Offset(cpu_addr), size, host_ptr); } else { staging_buffer.resize(size); memory_manager.ReadBlockUnsafe(gpu_addr, staging_buffer.data(), size); - UploadBlockData(block, block->GetOffset(cpu_addr), size, staging_buffer.data()); + block->Upload(block->Offset(cpu_addr), size, staging_buffer.data()); } - Register(new_map); - return new_map; + return Register(MapInterval(cpu_addr, cpu_addr_end, gpu_addr)); } const VAddr cpu_addr_end = cpu_addr + size; if (overlaps.size() == 1) { - MapInterval& current_map = overlaps[0]; + MapInterval* const current_map = overlaps[0]; if (current_map->IsInside(cpu_addr, cpu_addr_end)) { return current_map; } @@ -344,60 +351,70 @@ private: bool write_inheritance = false; bool modified_inheritance = false; // Calculate new buffer parameters - for (auto& overlap : overlaps) { - new_start = std::min(overlap->GetStart(), new_start); - new_end = std::max(overlap->GetEnd(), new_end); - write_inheritance |= overlap->IsWritten(); - modified_inheritance |= overlap->IsModified(); + for (MapInterval* overlap : overlaps) { + new_start = std::min(overlap->start, new_start); + new_end = std::max(overlap->end, new_end); + write_inheritance |= overlap->is_written; + modified_inheritance |= overlap->is_modified; } GPUVAddr new_gpu_addr = gpu_addr + new_start - cpu_addr; for (auto& overlap : overlaps) { Unregister(overlap); } UpdateBlock(block, new_start, new_end, overlaps); - MapInterval new_map = CreateMap(new_start, new_end, new_gpu_addr); + + const MapInterval new_map{new_start, new_end, new_gpu_addr}; + MapInterval* const map = Register(new_map, write_inheritance); + if (!map) { + return nullptr; + } if (modified_inheritance) { - new_map->MarkAsModified(true, GetModifiedTicks()); + map->MarkAsModified(true, GetModifiedTicks()); if (Settings::IsGPULevelHigh() && Settings::values.use_asynchronous_gpu_emulation) { - MarkForAsyncFlush(new_map); + MarkForAsyncFlush(map); } } - Register(new_map, write_inheritance); - return new_map; + return map; } - void UpdateBlock(const OwnerBuffer& block, VAddr start, VAddr end, - std::vector<MapInterval>& overlaps) { + void UpdateBlock(const Buffer* block, VAddr start, VAddr end, + const VectorMapInterval& overlaps) { const IntervalType base_interval{start, end}; IntervalSet interval_set{}; interval_set.add(base_interval); for (auto& overlap : overlaps) { - const IntervalType subtract{overlap->GetStart(), overlap->GetEnd()}; + const IntervalType subtract{overlap->start, overlap->end}; interval_set.subtract(subtract); } for (auto& interval : interval_set) { - std::size_t size = interval.upper() - interval.lower(); - if (size > 0) { - staging_buffer.resize(size); - system.Memory().ReadBlockUnsafe(interval.lower(), staging_buffer.data(), size); - UploadBlockData(block, block->GetOffset(interval.lower()), size, - staging_buffer.data()); + const std::size_t size = interval.upper() - interval.lower(); + if (size == 0) { + continue; } + staging_buffer.resize(size); + system.Memory().ReadBlockUnsafe(interval.lower(), staging_buffer.data(), size); + block->Upload(block->Offset(interval.lower()), size, staging_buffer.data()); } } - std::vector<MapInterval> GetMapsInRange(VAddr addr, std::size_t size) { + VectorMapInterval GetMapsInRange(VAddr addr, std::size_t size) { + VectorMapInterval result; if (size == 0) { - return {}; + return result; } - std::vector<MapInterval> objects{}; - const IntervalType interval{addr, addr + size}; - for (auto& pair : boost::make_iterator_range(mapped_addresses.equal_range(interval))) { - objects.push_back(pair.second); + const VAddr addr_end = addr + size; + auto it = mapped_addresses.lower_bound(addr); + if (it != mapped_addresses.begin()) { + --it; } - - return objects; + while (it != mapped_addresses.end() && it->start < addr_end) { + if (it->Overlaps(addr, addr_end)) { + result.push_back(&*it); + } + ++it; + } + return result; } /// Returns a ticks counter used for tracking when cached objects were last modified @@ -405,24 +422,28 @@ private: return ++modified_ticks; } - void FlushMap(MapInterval map) { - std::size_t size = map->GetEnd() - map->GetStart(); - OwnerBuffer block = blocks[map->GetStart() >> block_page_bits]; + void FlushMap(MapInterval* map) { + const auto it = blocks.find(map->start >> BLOCK_PAGE_BITS); + ASSERT_OR_EXECUTE(it != blocks.end(), return;); + + std::shared_ptr<Buffer> block = it->second; + + const std::size_t size = map->end - map->start; staging_buffer.resize(size); - DownloadBlockData(block, block->GetOffset(map->GetStart()), size, staging_buffer.data()); - system.Memory().WriteBlockUnsafe(map->GetStart(), staging_buffer.data(), size); + block->Download(block->Offset(map->start), size, staging_buffer.data()); + system.Memory().WriteBlockUnsafe(map->start, staging_buffer.data(), size); map->MarkAsModified(false, 0); } - BufferInfo StreamBufferUpload(const void* raw_pointer, std::size_t size, - std::size_t alignment) { + template <typename Callable> + BufferInfo StreamBufferUpload(std::size_t size, std::size_t alignment, Callable&& callable) { AlignBuffer(alignment); const std::size_t uploaded_offset = buffer_offset; - std::memcpy(buffer_ptr, raw_pointer, size); + callable(buffer_ptr); buffer_ptr += size; buffer_offset += size; - return {stream_buffer_handle, uploaded_offset}; + return BufferInfo{stream_buffer->Handle(), uploaded_offset, stream_buffer->Address()}; } void AlignBuffer(std::size_t alignment) { @@ -432,97 +453,89 @@ private: buffer_offset = offset_aligned; } - OwnerBuffer EnlargeBlock(OwnerBuffer buffer) { - const std::size_t old_size = buffer->GetSize(); - const std::size_t new_size = old_size + block_page_size; - const VAddr cpu_addr = buffer->GetCpuAddr(); - OwnerBuffer new_buffer = CreateBlock(cpu_addr, new_size); - CopyBlock(buffer, new_buffer, 0, 0, old_size); - buffer->SetEpoch(epoch); - pending_destruction.push_back(buffer); + std::shared_ptr<Buffer> EnlargeBlock(std::shared_ptr<Buffer> buffer) { + const std::size_t old_size = buffer->Size(); + const std::size_t new_size = old_size + BLOCK_PAGE_SIZE; + const VAddr cpu_addr = buffer->CpuAddr(); + std::shared_ptr<Buffer> new_buffer = CreateBlock(cpu_addr, new_size); + new_buffer->CopyFrom(*buffer, 0, 0, old_size); + QueueDestruction(std::move(buffer)); + const VAddr cpu_addr_end = cpu_addr + new_size - 1; - u64 page_start = cpu_addr >> block_page_bits; - const u64 page_end = cpu_addr_end >> block_page_bits; - while (page_start <= page_end) { - blocks[page_start] = new_buffer; - ++page_start; + const u64 page_end = cpu_addr_end >> BLOCK_PAGE_BITS; + for (u64 page_start = cpu_addr >> BLOCK_PAGE_BITS; page_start <= page_end; ++page_start) { + blocks.insert_or_assign(page_start, new_buffer); } + return new_buffer; } - OwnerBuffer MergeBlocks(OwnerBuffer first, OwnerBuffer second) { - const std::size_t size_1 = first->GetSize(); - const std::size_t size_2 = second->GetSize(); - const VAddr first_addr = first->GetCpuAddr(); - const VAddr second_addr = second->GetCpuAddr(); + std::shared_ptr<Buffer> MergeBlocks(std::shared_ptr<Buffer> first, + std::shared_ptr<Buffer> second) { + const std::size_t size_1 = first->Size(); + const std::size_t size_2 = second->Size(); + const VAddr first_addr = first->CpuAddr(); + const VAddr second_addr = second->CpuAddr(); const VAddr new_addr = std::min(first_addr, second_addr); const std::size_t new_size = size_1 + size_2; - OwnerBuffer new_buffer = CreateBlock(new_addr, new_size); - CopyBlock(first, new_buffer, 0, new_buffer->GetOffset(first_addr), size_1); - CopyBlock(second, new_buffer, 0, new_buffer->GetOffset(second_addr), size_2); - first->SetEpoch(epoch); - second->SetEpoch(epoch); - pending_destruction.push_back(first); - pending_destruction.push_back(second); + + std::shared_ptr<Buffer> new_buffer = CreateBlock(new_addr, new_size); + new_buffer->CopyFrom(*first, 0, new_buffer->Offset(first_addr), size_1); + new_buffer->CopyFrom(*second, 0, new_buffer->Offset(second_addr), size_2); + QueueDestruction(std::move(first)); + QueueDestruction(std::move(second)); + const VAddr cpu_addr_end = new_addr + new_size - 1; - u64 page_start = new_addr >> block_page_bits; - const u64 page_end = cpu_addr_end >> block_page_bits; - while (page_start <= page_end) { - blocks[page_start] = new_buffer; - ++page_start; + const u64 page_end = cpu_addr_end >> BLOCK_PAGE_BITS; + for (u64 page_start = new_addr >> BLOCK_PAGE_BITS; page_start <= page_end; ++page_start) { + blocks.insert_or_assign(page_start, new_buffer); } return new_buffer; } - OwnerBuffer GetBlock(const VAddr cpu_addr, const std::size_t size) { - OwnerBuffer found; + Buffer* GetBlock(VAddr cpu_addr, std::size_t size) { + std::shared_ptr<Buffer> found; + const VAddr cpu_addr_end = cpu_addr + size - 1; - u64 page_start = cpu_addr >> block_page_bits; - const u64 page_end = cpu_addr_end >> block_page_bits; - while (page_start <= page_end) { + const u64 page_end = cpu_addr_end >> BLOCK_PAGE_BITS; + for (u64 page_start = cpu_addr >> BLOCK_PAGE_BITS; page_start <= page_end; ++page_start) { auto it = blocks.find(page_start); if (it == blocks.end()) { if (found) { found = EnlargeBlock(found); - } else { - const VAddr start_addr = (page_start << block_page_bits); - found = CreateBlock(start_addr, block_page_size); - blocks[page_start] = found; - } - } else { - if (found) { - if (found == it->second) { - ++page_start; - continue; - } - found = MergeBlocks(found, it->second); - } else { - found = it->second; + continue; } + const VAddr start_addr = page_start << BLOCK_PAGE_BITS; + found = CreateBlock(start_addr, BLOCK_PAGE_SIZE); + blocks.insert_or_assign(page_start, found); + continue; + } + if (!found) { + found = it->second; + continue; + } + if (found != it->second) { + found = MergeBlocks(std::move(found), it->second); } - ++page_start; } - return found; + return found.get(); } - void MarkRegionAsWritten(const VAddr start, const VAddr end) { - u64 page_start = start >> write_page_bit; - const u64 page_end = end >> write_page_bit; - while (page_start <= page_end) { + void MarkRegionAsWritten(VAddr start, VAddr end) { + const u64 page_end = end >> WRITE_PAGE_BIT; + for (u64 page_start = start >> WRITE_PAGE_BIT; page_start <= page_end; ++page_start) { auto it = written_pages.find(page_start); if (it != written_pages.end()) { it->second = it->second + 1; } else { - written_pages[page_start] = 1; + written_pages.insert_or_assign(page_start, 1); } - page_start++; } } - void UnmarkRegionAsWritten(const VAddr start, const VAddr end) { - u64 page_start = start >> write_page_bit; - const u64 page_end = end >> write_page_bit; - while (page_start <= page_end) { + void UnmarkRegionAsWritten(VAddr start, VAddr end) { + const u64 page_end = end >> WRITE_PAGE_BIT; + for (u64 page_start = start >> WRITE_PAGE_BIT; page_start <= page_end; ++page_start) { auto it = written_pages.find(page_start); if (it != written_pages.end()) { if (it->second > 1) { @@ -531,25 +544,27 @@ private: written_pages.erase(it); } } - page_start++; } } - bool IsRegionWritten(const VAddr start, const VAddr end) const { - u64 page_start = start >> write_page_bit; - const u64 page_end = end >> write_page_bit; - while (page_start <= page_end) { + bool IsRegionWritten(VAddr start, VAddr end) const { + const u64 page_end = end >> WRITE_PAGE_BIT; + for (u64 page_start = start >> WRITE_PAGE_BIT; page_start <= page_end; ++page_start) { if (written_pages.count(page_start) > 0) { return true; } - page_start++; } return false; } - void MarkForAsyncFlush(MapInterval& map) { + void QueueDestruction(std::shared_ptr<Buffer> buffer) { + buffer->SetEpoch(epoch); + pending_destruction.push(std::move(buffer)); + } + + void MarkForAsyncFlush(MapInterval* map) { if (!uncommitted_flushes) { - uncommitted_flushes = std::make_shared<std::unordered_set<MapInterval>>(); + uncommitted_flushes = std::make_shared<std::unordered_set<MapInterval*>>(); } uncommitted_flushes->insert(map); } @@ -558,35 +573,29 @@ private: Core::System& system; std::unique_ptr<StreamBuffer> stream_buffer; - BufferType stream_buffer_handle{}; - - bool invalidated = false; + BufferType stream_buffer_handle; u8* buffer_ptr = nullptr; u64 buffer_offset = 0; u64 buffer_offset_base = 0; - using IntervalSet = boost::icl::interval_set<VAddr>; - using IntervalCache = boost::icl::interval_map<VAddr, MapInterval>; - using IntervalType = typename IntervalCache::interval_type; - IntervalCache mapped_addresses; + MapIntervalAllocator mapped_addresses_allocator; + boost::intrusive::set<MapInterval, boost::intrusive::compare<MapIntervalCompare>> + mapped_addresses; - static constexpr u64 write_page_bit = 11; std::unordered_map<u64, u32> written_pages; + std::unordered_map<u64, std::shared_ptr<Buffer>> blocks; - static constexpr u64 block_page_bits = 21; - static constexpr u64 block_page_size = 1ULL << block_page_bits; - std::unordered_map<u64, OwnerBuffer> blocks; - - std::list<OwnerBuffer> pending_destruction; + std::queue<std::shared_ptr<Buffer>> pending_destruction; u64 epoch = 0; u64 modified_ticks = 0; std::vector<u8> staging_buffer; - std::list<MapInterval> marked_for_unregister; - std::shared_ptr<std::unordered_set<MapInterval>> uncommitted_flushes{}; - std::list<std::shared_ptr<std::list<MapInterval>>> committed_flushes; + std::list<MapInterval*> marked_for_unregister; + + std::shared_ptr<std::unordered_set<MapInterval*>> uncommitted_flushes; + std::list<std::shared_ptr<std::list<MapInterval*>>> committed_flushes; std::recursive_mutex mutex; }; diff --git a/src/video_core/buffer_cache/map_interval.cpp b/src/video_core/buffer_cache/map_interval.cpp new file mode 100644 index 000000000..62587e18a --- /dev/null +++ b/src/video_core/buffer_cache/map_interval.cpp @@ -0,0 +1,33 @@ +// Copyright 2020 yuzu Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include <algorithm> +#include <array> +#include <cstddef> +#include <memory> + +#include "video_core/buffer_cache/map_interval.h" + +namespace VideoCommon { + +MapIntervalAllocator::MapIntervalAllocator() { + FillFreeList(first_chunk); +} + +MapIntervalAllocator::~MapIntervalAllocator() = default; + +void MapIntervalAllocator::AllocateNewChunk() { + *new_chunk = std::make_unique<Chunk>(); + FillFreeList(**new_chunk); + new_chunk = &(*new_chunk)->next; +} + +void MapIntervalAllocator::FillFreeList(Chunk& chunk) { + const std::size_t old_size = free_list.size(); + free_list.resize(old_size + chunk.data.size()); + std::transform(chunk.data.rbegin(), chunk.data.rend(), free_list.begin() + old_size, + [](MapInterval& interval) { return &interval; }); +} + +} // namespace VideoCommon diff --git a/src/video_core/buffer_cache/map_interval.h b/src/video_core/buffer_cache/map_interval.h index 29d8b26f3..fe0bcd1d8 100644 --- a/src/video_core/buffer_cache/map_interval.h +++ b/src/video_core/buffer_cache/map_interval.h @@ -4,104 +4,89 @@ #pragma once +#include <array> +#include <cstddef> +#include <memory> +#include <vector> + +#include <boost/intrusive/set_hook.hpp> + #include "common/common_types.h" #include "video_core/gpu.h" namespace VideoCommon { -class MapIntervalBase { -public: - MapIntervalBase(const VAddr start, const VAddr end, const GPUVAddr gpu_addr) - : start{start}, end{end}, gpu_addr{gpu_addr} {} - - void SetCpuAddress(VAddr new_cpu_addr) { - cpu_addr = new_cpu_addr; - } - - VAddr GetCpuAddress() const { - return cpu_addr; - } - - GPUVAddr GetGpuAddress() const { - return gpu_addr; - } - - bool IsInside(const VAddr other_start, const VAddr other_end) const { - return (start <= other_start && other_end <= end); - } - - bool operator==(const MapIntervalBase& rhs) const { - return std::tie(start, end) == std::tie(rhs.start, rhs.end); - } - - bool operator!=(const MapIntervalBase& rhs) const { - return !operator==(rhs); - } +struct MapInterval : public boost::intrusive::set_base_hook<boost::intrusive::optimize_size<true>> { + MapInterval() = default; - void MarkAsRegistered(const bool registered) { - is_registered = registered; - } + /*implicit*/ MapInterval(VAddr start_) noexcept : start{start_} {} - bool IsRegistered() const { - return is_registered; - } + explicit MapInterval(VAddr start_, VAddr end_, GPUVAddr gpu_addr_) noexcept + : start{start_}, end{end_}, gpu_addr{gpu_addr_} {} - void SetMemoryMarked(bool is_memory_marked_) { - is_memory_marked = is_memory_marked_; + bool IsInside(VAddr other_start, VAddr other_end) const noexcept { + return start <= other_start && other_end <= end; } - bool IsMemoryMarked() const { - return is_memory_marked; + bool Overlaps(VAddr other_start, VAddr other_end) const noexcept { + return start < other_end && other_start < end; } - void SetSyncPending(bool is_sync_pending_) { - is_sync_pending = is_sync_pending_; - } + void MarkAsModified(bool is_modified_, u64 ticks_) noexcept { + is_modified = is_modified_; + ticks = ticks_; + } + + boost::intrusive::set_member_hook<> member_hook_; + VAddr start = 0; + VAddr end = 0; + GPUVAddr gpu_addr = 0; + u64 ticks = 0; + bool is_written = false; + bool is_modified = false; + bool is_registered = false; + bool is_memory_marked = false; + bool is_sync_pending = false; +}; - bool IsSyncPending() const { - return is_sync_pending; +struct MapIntervalCompare { + constexpr bool operator()(const MapInterval& lhs, const MapInterval& rhs) const noexcept { + return lhs.start < rhs.start; } +}; - VAddr GetStart() const { - return start; - } +class MapIntervalAllocator { +public: + MapIntervalAllocator(); + ~MapIntervalAllocator(); - VAddr GetEnd() const { - return end; + MapInterval* Allocate() { + if (free_list.empty()) { + AllocateNewChunk(); + } + MapInterval* const interval = free_list.back(); + free_list.pop_back(); + return interval; } - void MarkAsModified(const bool is_modified_, const u64 tick) { - is_modified = is_modified_; - ticks = tick; + void Release(MapInterval* interval) { + free_list.push_back(interval); } - bool IsModified() const { - return is_modified; - } +private: + struct Chunk { + std::unique_ptr<Chunk> next; + std::array<MapInterval, 0x8000> data; + }; - u64 GetModificationTick() const { - return ticks; - } + void AllocateNewChunk(); - void MarkAsWritten(const bool is_written_) { - is_written = is_written_; - } + void FillFreeList(Chunk& chunk); - bool IsWritten() const { - return is_written; - } + std::vector<MapInterval*> free_list; + std::unique_ptr<Chunk>* new_chunk = &first_chunk.next; -private: - VAddr start; - VAddr end; - GPUVAddr gpu_addr; - VAddr cpu_addr{}; - bool is_written{}; - bool is_modified{}; - bool is_registered{}; - bool is_memory_marked{}; - bool is_sync_pending{}; - u64 ticks{}; + Chunk first_chunk; }; } // namespace VideoCommon diff --git a/src/video_core/compatible_formats.cpp b/src/video_core/compatible_formats.cpp new file mode 100644 index 000000000..6c426b035 --- /dev/null +++ b/src/video_core/compatible_formats.cpp @@ -0,0 +1,162 @@ +// Copyright 2020 yuzu Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include <array> +#include <bitset> +#include <cstddef> + +#include "video_core/compatible_formats.h" +#include "video_core/surface.h" + +namespace VideoCore::Surface { + +namespace { + +// Compatibility table taken from Table 3.X.2 in: +// https://www.khronos.org/registry/OpenGL/extensions/ARB/ARB_texture_view.txt + +constexpr std::array VIEW_CLASS_128_BITS = { + PixelFormat::RGBA32F, + PixelFormat::RGBA32UI, +}; +// Missing formats: +// PixelFormat::RGBA32I + +constexpr std::array VIEW_CLASS_96_BITS = { + PixelFormat::RGB32F, +}; +// Missing formats: +// PixelFormat::RGB32UI, +// PixelFormat::RGB32I, + +constexpr std::array VIEW_CLASS_64_BITS = { + PixelFormat::RGBA16F, PixelFormat::RG32F, PixelFormat::RGBA16UI, PixelFormat::RG32UI, + PixelFormat::RGBA16U, PixelFormat::RGBA16F, PixelFormat::RGBA16S, +}; +// Missing formats: +// PixelFormat::RGBA16I +// PixelFormat::RG32I + +// TODO: How should we handle 48 bits? + +constexpr std::array VIEW_CLASS_32_BITS = { + PixelFormat::RG16F, PixelFormat::R11FG11FB10F, PixelFormat::R32F, + PixelFormat::A2B10G10R10U, PixelFormat::RG16UI, PixelFormat::R32UI, + PixelFormat::RG16I, PixelFormat::R32I, PixelFormat::ABGR8U, + PixelFormat::RG16, PixelFormat::ABGR8S, PixelFormat::RG16S, + PixelFormat::RGBA8_SRGB, PixelFormat::E5B9G9R9F, PixelFormat::BGRA8, + PixelFormat::BGRA8_SRGB, +}; +// Missing formats: +// PixelFormat::RGBA8UI +// PixelFormat::RGBA8I +// PixelFormat::RGB10_A2_UI + +// TODO: How should we handle 24 bits? + +constexpr std::array VIEW_CLASS_16_BITS = { + PixelFormat::R16F, PixelFormat::RG8UI, PixelFormat::R16UI, PixelFormat::R16I, + PixelFormat::RG8U, PixelFormat::R16U, PixelFormat::RG8S, PixelFormat::R16S, +}; +// Missing formats: +// PixelFormat::RG8I + +constexpr std::array VIEW_CLASS_8_BITS = { + PixelFormat::R8UI, + PixelFormat::R8U, +}; +// Missing formats: +// PixelFormat::R8I +// PixelFormat::R8S + +constexpr std::array VIEW_CLASS_RGTC1_RED = { + PixelFormat::DXN1, +}; +// Missing formats: +// COMPRESSED_SIGNED_RED_RGTC1 + +constexpr std::array VIEW_CLASS_RGTC2_RG = { + PixelFormat::DXN2UNORM, + PixelFormat::DXN2SNORM, +}; + +constexpr std::array VIEW_CLASS_BPTC_UNORM = { + PixelFormat::BC7U, + PixelFormat::BC7U_SRGB, +}; + +constexpr std::array VIEW_CLASS_BPTC_FLOAT = { + PixelFormat::BC6H_SF16, + PixelFormat::BC6H_UF16, +}; + +// Compatibility table taken from Table 4.X.1 in: +// https://www.khronos.org/registry/OpenGL/extensions/ARB/ARB_copy_image.txt + +constexpr std::array COPY_CLASS_128_BITS = { + PixelFormat::RGBA32UI, PixelFormat::RGBA32F, PixelFormat::DXT23, + PixelFormat::DXT23_SRGB, PixelFormat::DXT45, PixelFormat::DXT45_SRGB, + PixelFormat::DXN2SNORM, PixelFormat::BC7U, PixelFormat::BC7U_SRGB, + PixelFormat::BC6H_SF16, PixelFormat::BC6H_UF16, +}; +// Missing formats: +// PixelFormat::RGBA32I +// COMPRESSED_RG_RGTC2 + +constexpr std::array COPY_CLASS_64_BITS = { + PixelFormat::RGBA16F, PixelFormat::RG32F, PixelFormat::RGBA16UI, PixelFormat::RG32UI, + PixelFormat::RGBA16U, PixelFormat::RGBA16S, PixelFormat::DXT1_SRGB, PixelFormat::DXT1, + +}; +// Missing formats: +// PixelFormat::RGBA16I +// PixelFormat::RG32I, +// COMPRESSED_RGB_S3TC_DXT1_EXT +// COMPRESSED_SRGB_S3TC_DXT1_EXT +// COMPRESSED_RGBA_S3TC_DXT1_EXT +// COMPRESSED_SIGNED_RED_RGTC1 + +void Enable(FormatCompatibility::Table& compatiblity, size_t format_a, size_t format_b) { + compatiblity[format_a][format_b] = true; + compatiblity[format_b][format_a] = true; +} + +void Enable(FormatCompatibility::Table& compatibility, PixelFormat format_a, PixelFormat format_b) { + Enable(compatibility, static_cast<size_t>(format_a), static_cast<size_t>(format_b)); +} + +template <typename Range> +void EnableRange(FormatCompatibility::Table& compatibility, const Range& range) { + for (auto it_a = range.begin(); it_a != range.end(); ++it_a) { + for (auto it_b = it_a; it_b != range.end(); ++it_b) { + Enable(compatibility, *it_a, *it_b); + } + } +} + +} // Anonymous namespace + +FormatCompatibility::FormatCompatibility() { + for (size_t i = 0; i < MaxPixelFormat; ++i) { + // Identity is allowed + Enable(view, i, i); + } + + EnableRange(view, VIEW_CLASS_128_BITS); + EnableRange(view, VIEW_CLASS_96_BITS); + EnableRange(view, VIEW_CLASS_64_BITS); + EnableRange(view, VIEW_CLASS_32_BITS); + EnableRange(view, VIEW_CLASS_16_BITS); + EnableRange(view, VIEW_CLASS_8_BITS); + EnableRange(view, VIEW_CLASS_RGTC1_RED); + EnableRange(view, VIEW_CLASS_RGTC2_RG); + EnableRange(view, VIEW_CLASS_BPTC_UNORM); + EnableRange(view, VIEW_CLASS_BPTC_FLOAT); + + copy = view; + EnableRange(copy, COPY_CLASS_128_BITS); + EnableRange(copy, COPY_CLASS_64_BITS); +} + +} // namespace VideoCore::Surface diff --git a/src/video_core/compatible_formats.h b/src/video_core/compatible_formats.h new file mode 100644 index 000000000..d1082566d --- /dev/null +++ b/src/video_core/compatible_formats.h @@ -0,0 +1,32 @@ +// Copyright 2020 yuzu Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include <array> +#include <bitset> +#include <cstddef> + +#include "video_core/surface.h" + +namespace VideoCore::Surface { + +class FormatCompatibility { +public: + using Table = std::array<std::bitset<MaxPixelFormat>, MaxPixelFormat>; + + explicit FormatCompatibility(); + + bool TestView(PixelFormat format_a, PixelFormat format_b) const noexcept { + return view[static_cast<size_t>(format_a)][static_cast<size_t>(format_b)]; + } + + bool TestCopy(PixelFormat format_a, PixelFormat format_b) const noexcept { + return copy[static_cast<size_t>(format_a)][static_cast<size_t>(format_b)]; + } + +private: + Table view; + Table copy; +}; + +} // namespace VideoCore::Surface diff --git a/src/video_core/dma_pusher.cpp b/src/video_core/dma_pusher.cpp index bdc023d54..f2f96ac33 100644 --- a/src/video_core/dma_pusher.cpp +++ b/src/video_core/dma_pusher.cpp @@ -54,9 +54,7 @@ bool DmaPusher::Step() { return true; }); const CommandListHeader command_list_header{command_list[dma_pushbuffer_subindex++]}; - GPUVAddr dma_get = command_list_header.addr; - GPUVAddr dma_put = dma_get + command_list_header.size * sizeof(u32); - bool non_main = command_list_header.is_non_main; + const GPUVAddr dma_get = command_list_header.addr; if (dma_pushbuffer_subindex >= command_list.size()) { // We've gone through the current list, remove it from the queue @@ -133,11 +131,6 @@ bool DmaPusher::Step() { index++; } - if (!non_main) { - // TODO (degasus): This is dead code, as dma_mget is never read. - dma_mget = dma_put; - } - return true; } diff --git a/src/video_core/dma_pusher.h b/src/video_core/dma_pusher.h index e8b714e94..efa90d170 100644 --- a/src/video_core/dma_pusher.h +++ b/src/video_core/dma_pusher.h @@ -102,7 +102,6 @@ private: DmaState dma_state{}; bool dma_increment_once{}; - GPUVAddr dma_mget{}; ///< main pushbuffer last read address bool ib_enable{true}; ///< IB mode enabled std::array<Tegra::Engines::EngineInterface*, max_subchannels> subchannels{}; diff --git a/src/video_core/engines/const_buffer_engine_interface.h b/src/video_core/engines/const_buffer_engine_interface.h index ebe139504..f46e81bb7 100644 --- a/src/video_core/engines/const_buffer_engine_interface.h +++ b/src/video_core/engines/const_buffer_engine_interface.h @@ -93,6 +93,7 @@ public: virtual SamplerDescriptor AccessBoundSampler(ShaderType stage, u64 offset) const = 0; virtual SamplerDescriptor AccessBindlessSampler(ShaderType stage, u64 const_buffer, u64 offset) const = 0; + virtual SamplerDescriptor AccessSampler(u32 handle) const = 0; virtual u32 GetBoundBuffer() const = 0; virtual VideoCore::GuestDriverProfile& AccessGuestDriverProfile() = 0; diff --git a/src/video_core/engines/kepler_compute.cpp b/src/video_core/engines/kepler_compute.cpp index f6237fc6a..a82b06a38 100644 --- a/src/video_core/engines/kepler_compute.cpp +++ b/src/video_core/engines/kepler_compute.cpp @@ -92,8 +92,11 @@ SamplerDescriptor KeplerCompute::AccessBindlessSampler(ShaderType stage, u64 con ASSERT(stage == ShaderType::Compute); const auto& tex_info_buffer = launch_description.const_buffer_config[const_buffer]; const GPUVAddr tex_info_address = tex_info_buffer.Address() + offset; + return AccessSampler(memory_manager.Read<u32>(tex_info_address)); +} - const Texture::TextureHandle tex_handle{memory_manager.Read<u32>(tex_info_address)}; +SamplerDescriptor KeplerCompute::AccessSampler(u32 handle) const { + const Texture::TextureHandle tex_handle{handle}; const Texture::FullTextureInfo tex_info = GetTextureInfo(tex_handle); SamplerDescriptor result = SamplerDescriptor::FromTIC(tex_info.tic); result.is_shadow.Assign(tex_info.tsc.depth_compare_enabled.Value()); diff --git a/src/video_core/engines/kepler_compute.h b/src/video_core/engines/kepler_compute.h index 18ceedfaf..b7f668d88 100644 --- a/src/video_core/engines/kepler_compute.h +++ b/src/video_core/engines/kepler_compute.h @@ -219,6 +219,8 @@ public: SamplerDescriptor AccessBindlessSampler(ShaderType stage, u64 const_buffer, u64 offset) const override; + SamplerDescriptor AccessSampler(u32 handle) const override; + u32 GetBoundBuffer() const override { return regs.tex_cb_index; } diff --git a/src/video_core/engines/maxwell_3d.cpp b/src/video_core/engines/maxwell_3d.cpp index 33936e209..c01436295 100644 --- a/src/video_core/engines/maxwell_3d.cpp +++ b/src/video_core/engines/maxwell_3d.cpp @@ -25,9 +25,8 @@ constexpr u32 MacroRegistersStart = 0xE00; Maxwell3D::Maxwell3D(Core::System& system, VideoCore::RasterizerInterface& rasterizer, MemoryManager& memory_manager) : system{system}, rasterizer{rasterizer}, memory_manager{memory_manager}, - macro_interpreter{*this}, upload_state{memory_manager, regs.upload} { + macro_engine{GetMacroEngine(*this)}, upload_state{memory_manager, regs.upload} { dirty.flags.flip(); - InitializeRegisterDefaults(); } @@ -44,6 +43,12 @@ void Maxwell3D::InitializeRegisterDefaults() { viewport.depth_range_near = 0.0f; viewport.depth_range_far = 1.0f; } + for (auto& viewport : regs.viewport_transform) { + viewport.swizzle.x.Assign(Regs::ViewportSwizzle::PositiveX); + viewport.swizzle.y.Assign(Regs::ViewportSwizzle::PositiveY); + viewport.swizzle.z.Assign(Regs::ViewportSwizzle::PositiveZ); + viewport.swizzle.w.Assign(Regs::ViewportSwizzle::PositiveW); + } // Doom and Bomberman seems to use the uninitialized registers and just enable blend // so initialize blend registers with sane values @@ -100,7 +105,11 @@ void Maxwell3D::InitializeRegisterDefaults() { regs.rasterize_enable = 1; regs.rt_separate_frag_data = 1; regs.framebuffer_srgb = 1; + regs.line_width_aliased = 1.0f; + regs.line_width_smooth = 1.0f; regs.front_face = Maxwell3D::Regs::FrontFace::ClockWise; + regs.polygon_mode_back = Maxwell3D::Regs::PolygonMode::Fill; + regs.polygon_mode_front = Maxwell3D::Regs::PolygonMode::Fill; shadow_state = regs; @@ -110,7 +119,7 @@ void Maxwell3D::InitializeRegisterDefaults() { mme_inline[MAXWELL3D_REG_INDEX(index_array.count)] = true; } -void Maxwell3D::CallMacroMethod(u32 method, std::size_t num_parameters, const u32* parameters) { +void Maxwell3D::CallMacroMethod(u32 method, const std::vector<u32>& parameters) { // Reset the current macro. executing_macro = 0; @@ -119,7 +128,7 @@ void Maxwell3D::CallMacroMethod(u32 method, std::size_t num_parameters, const u3 ((method - MacroRegistersStart) >> 1) % static_cast<u32>(macro_positions.size()); // Execute the current macro. - macro_interpreter.Execute(macro_positions[entry], num_parameters, parameters); + macro_engine->Execute(*this, macro_positions[entry], parameters); if (mme_draw.current_mode != MMEDrawMode::Undefined) { FlushMMEInlineDraw(); } @@ -155,7 +164,7 @@ void Maxwell3D::CallMethod(u32 method, u32 method_argument, bool is_last_call) { // Call the macro when there are no more parameters in the command buffer if (is_last_call) { - CallMacroMethod(executing_macro, macro_params.size(), macro_params.data()); + CallMacroMethod(executing_macro, macro_params); macro_params.clear(); } return; @@ -191,7 +200,7 @@ void Maxwell3D::CallMethod(u32 method, u32 method_argument, bool is_last_call) { break; } case MAXWELL3D_REG_INDEX(macros.data): { - ProcessMacroUpload(arg); + macro_engine->AddCode(regs.macros.upload_address, arg); break; } case MAXWELL3D_REG_INDEX(macros.bind): { @@ -300,7 +309,7 @@ void Maxwell3D::CallMultiMethod(u32 method, const u32* base_start, u32 amount, // Call the macro when there are no more parameters in the command buffer if (amount == methods_pending) { - CallMacroMethod(executing_macro, macro_params.size(), macro_params.data()); + CallMacroMethod(executing_macro, macro_params); macro_params.clear(); } return; @@ -414,9 +423,7 @@ void Maxwell3D::FlushMMEInlineDraw() { } void Maxwell3D::ProcessMacroUpload(u32 data) { - ASSERT_MSG(regs.macros.upload_address < macro_memory.size(), - "upload_address exceeded macro_memory size!"); - macro_memory[regs.macros.upload_address++] = data; + macro_engine->AddCode(regs.macros.upload_address++, data); } void Maxwell3D::ProcessMacroBind(u32 data) { @@ -451,8 +458,9 @@ void Maxwell3D::StampQueryResult(u64 payload, bool long_query) { void Maxwell3D::ProcessQueryGet() { // TODO(Subv): Support the other query units. - ASSERT_MSG(regs.query.query_get.unit == Regs::QueryUnit::Crop, - "Units other than CROP are unimplemented"); + if (regs.query.query_get.unit != Regs::QueryUnit::Crop) { + LOG_DEBUG(HW_GPU, "Units other than CROP are unimplemented"); + } switch (regs.query.query_get.operation) { case Regs::QueryOperation::Release: @@ -528,8 +536,8 @@ void Maxwell3D::ProcessCounterReset() { rasterizer.ResetCounter(QueryType::SamplesPassed); break; default: - LOG_WARNING(Render_OpenGL, "Unimplemented counter reset={}", - static_cast<int>(regs.counter_reset)); + LOG_DEBUG(Render_OpenGL, "Unimplemented counter reset={}", + static_cast<int>(regs.counter_reset)); break; } } @@ -586,8 +594,8 @@ std::optional<u64> Maxwell3D::GetQueryResult() { system.GPU().GetTicks()); return {}; default: - UNIMPLEMENTED_MSG("Unimplemented query select type {}", - static_cast<u32>(regs.query.query_get.select.Value())); + LOG_DEBUG(HW_GPU, "Unimplemented query select type {}", + static_cast<u32>(regs.query.query_get.select.Value())); return 1; } } @@ -732,8 +740,11 @@ SamplerDescriptor Maxwell3D::AccessBindlessSampler(ShaderType stage, u64 const_b const auto& shader = state.shader_stages[static_cast<std::size_t>(stage)]; const auto& tex_info_buffer = shader.const_buffers[const_buffer]; const GPUVAddr tex_info_address = tex_info_buffer.address + offset; + return AccessSampler(memory_manager.Read<u32>(tex_info_address)); +} - const Texture::TextureHandle tex_handle{memory_manager.Read<u32>(tex_info_address)}; +SamplerDescriptor Maxwell3D::AccessSampler(u32 handle) const { + const Texture::TextureHandle tex_handle{handle}; const Texture::FullTextureInfo tex_info = GetTextureInfo(tex_handle); SamplerDescriptor result = SamplerDescriptor::FromTIC(tex_info.tic); result.is_shadow.Assign(tex_info.tsc.depth_compare_enabled.Value()); diff --git a/src/video_core/engines/maxwell_3d.h b/src/video_core/engines/maxwell_3d.h index 1a5df05ce..ef1618990 100644 --- a/src/video_core/engines/maxwell_3d.h +++ b/src/video_core/engines/maxwell_3d.h @@ -23,7 +23,7 @@ #include "video_core/engines/engine_upload.h" #include "video_core/engines/shader_type.h" #include "video_core/gpu.h" -#include "video_core/macro_interpreter.h" +#include "video_core/macro/macro.h" #include "video_core/textures/texture.h" namespace Core { @@ -576,6 +576,17 @@ public: Replay = 3, }; + enum class ViewportSwizzle : u32 { + PositiveX = 0, + NegativeX = 1, + PositiveY = 2, + NegativeY = 3, + PositiveZ = 4, + NegativeZ = 5, + PositiveW = 6, + NegativeW = 7, + }; + struct RenderTargetConfig { u32 address_high; u32 address_low; @@ -587,6 +598,7 @@ public: BitField<4, 3, u32> block_height; BitField<8, 3, u32> block_depth; BitField<12, 1, InvMemoryLayout> type; + BitField<16, 1, u32> is_3d; } memory_layout; union { BitField<0, 16, u32> layers; @@ -619,7 +631,14 @@ public: f32 translate_x; f32 translate_y; f32 translate_z; - INSERT_UNION_PADDING_WORDS(2); + union { + u32 raw; + BitField<0, 3, ViewportSwizzle> x; + BitField<4, 3, ViewportSwizzle> y; + BitField<8, 3, ViewportSwizzle> z; + BitField<12, 3, ViewportSwizzle> w; + } swizzle; + INSERT_UNION_PADDING_WORDS(1); Common::Rectangle<f32> GetRect() const { return { @@ -1385,6 +1404,8 @@ public: SamplerDescriptor AccessBindlessSampler(ShaderType stage, u64 const_buffer, u64 offset) const override; + SamplerDescriptor AccessSampler(u32 handle) const override; + u32 GetBoundBuffer() const override { return regs.tex_cb_index; } @@ -1393,17 +1414,16 @@ public: const VideoCore::GuestDriverProfile& AccessGuestDriverProfile() const override; - /// Memory for macro code - it's undetermined how big this is, however 1MB is much larger than - /// we've seen used. - using MacroMemory = std::array<u32, 0x40000>; + bool ShouldExecute() const { + return execute_on; + } - /// Gets a reference to macro memory. - const MacroMemory& GetMacroMemory() const { - return macro_memory; + VideoCore::RasterizerInterface& GetRasterizer() { + return rasterizer; } - bool ShouldExecute() const { - return execute_on; + const VideoCore::RasterizerInterface& GetRasterizer() const { + return rasterizer; } /// Notify a memory write has happened. @@ -1450,16 +1470,13 @@ private: std::array<bool, Regs::NUM_REGS> mme_inline{}; - /// Memory for macro code - MacroMemory macro_memory; - /// Macro method that is currently being executed / being fed parameters. u32 executing_macro = 0; /// Parameters that have been submitted to the macro call so far. std::vector<u32> macro_params; /// Interpreter for the macro codes uploaded to the GPU. - MacroInterpreter macro_interpreter; + std::unique_ptr<MacroEngine> macro_engine; static constexpr u32 null_cb_data = 0xFFFFFFFF; struct { @@ -1488,7 +1505,7 @@ private: * @param num_parameters Number of arguments * @param parameters Arguments to the method call */ - void CallMacroMethod(u32 method, std::size_t num_parameters, const u32* parameters); + void CallMacroMethod(u32 method, const std::vector<u32>& parameters); /// Handles writes to the macro uploading register. void ProcessMacroUpload(u32 data); diff --git a/src/video_core/engines/shader_bytecode.h b/src/video_core/engines/shader_bytecode.h index 8dae754d4..d374b73cf 100644 --- a/src/video_core/engines/shader_bytecode.h +++ b/src/video_core/engines/shader_bytecode.h @@ -168,18 +168,22 @@ enum class Pred : u64 { }; enum class PredCondition : u64 { - LessThan = 1, - Equal = 2, - LessEqual = 3, - GreaterThan = 4, - NotEqual = 5, - GreaterEqual = 6, - LessThanWithNan = 9, - LessEqualWithNan = 11, - GreaterThanWithNan = 12, - NotEqualWithNan = 13, - GreaterEqualWithNan = 14, - // TODO(Subv): Other condition types + F = 0, // Always false + LT = 1, // Ordered less than + EQ = 2, // Ordered equal + LE = 3, // Ordered less than or equal + GT = 4, // Ordered greater than + NE = 5, // Ordered not equal + GE = 6, // Ordered greater than or equal + NUM = 7, // Ordered + NAN_ = 8, // Unordered + LTU = 9, // Unordered less than + EQU = 10, // Unordered equal + LEU = 11, // Unordered less than or equal + GTU = 12, // Unordered greater than + NEU = 13, // Unordered not equal + GEU = 14, // Unordered greater than or equal + T = 15, // Always true }; enum class PredOperation : u64 { @@ -657,6 +661,10 @@ union Instruction { constexpr Instruction(u64 value) : value{value} {} constexpr Instruction(const Instruction& instr) : value(instr.value) {} + constexpr bool Bit(u64 offset) const { + return ((value >> offset) & 1) != 0; + } + BitField<0, 8, Register> gpr0; BitField<8, 8, Register> gpr8; union { @@ -1870,7 +1878,9 @@ public: HSETP2_C, HSETP2_R, HSETP2_IMM, + HSET2_C, HSET2_R, + HSET2_IMM, POPC_C, POPC_R, POPC_IMM, @@ -2190,7 +2200,9 @@ private: INST("0111111-1-------", Id::HSETP2_C, Type::HalfSetPredicate, "HSETP2_C"), INST("0101110100100---", Id::HSETP2_R, Type::HalfSetPredicate, "HSETP2_R"), INST("0111111-0-------", Id::HSETP2_IMM, Type::HalfSetPredicate, "HSETP2_IMM"), + INST("0111110-1-------", Id::HSET2_C, Type::HalfSet, "HSET2_C"), INST("0101110100011---", Id::HSET2_R, Type::HalfSet, "HSET2_R"), + INST("0111110-0-------", Id::HSET2_IMM, Type::HalfSet, "HSET2_IMM"), INST("010110111010----", Id::FCMP_RR, Type::Arithmetic, "FCMP_RR"), INST("010010111010----", Id::FCMP_RC, Type::Arithmetic, "FCMP_RC"), INST("0101000010000---", Id::MUFU, Type::Arithmetic, "MUFU"), diff --git a/src/video_core/gpu.cpp b/src/video_core/gpu.cpp index 8eb017f65..482e49711 100644 --- a/src/video_core/gpu.cpp +++ b/src/video_core/gpu.cpp @@ -2,6 +2,8 @@ // Licensed under GPLv2 or any later version // Refer to the license.txt file included. +#include <chrono> + #include "common/assert.h" #include "common/microprofile.h" #include "core/core.h" @@ -154,8 +156,7 @@ u64 GPU::GetTicks() const { constexpr u64 gpu_ticks_num = 384; constexpr u64 gpu_ticks_den = 625; - const u64 cpu_ticks = system.CoreTiming().GetTicks(); - u64 nanoseconds = Core::Timing::CyclesToNs(cpu_ticks).count(); + u64 nanoseconds = system.CoreTiming().GetGlobalTimeNs().count(); if (Settings::values.use_fast_gpu_time) { nanoseconds /= 256; } diff --git a/src/video_core/gpu.h b/src/video_core/gpu.h index dd51c95b7..2c42483bd 100644 --- a/src/video_core/gpu.h +++ b/src/video_core/gpu.h @@ -68,6 +68,7 @@ enum class RenderTargetFormat : u32 { BGR5A1_UNORM = 0xE9, RG8_UNORM = 0xEA, RG8_SNORM = 0xEB, + RG8_UINT = 0xED, R16_UNORM = 0xEE, R16_SNORM = 0xEF, R16_SINT = 0xF0, @@ -283,6 +284,12 @@ public: /// core timing events. virtual void Start() = 0; + /// Obtain the CPU Context + virtual void ObtainContext() = 0; + + /// Release the CPU Context + virtual void ReleaseContext() = 0; + /// Push GPU command entries to be processed virtual void PushGPUEntries(Tegra::CommandList&& entries) = 0; diff --git a/src/video_core/gpu_asynch.cpp b/src/video_core/gpu_asynch.cpp index 53305ab43..7b855f63e 100644 --- a/src/video_core/gpu_asynch.cpp +++ b/src/video_core/gpu_asynch.cpp @@ -19,10 +19,17 @@ GPUAsynch::GPUAsynch(Core::System& system, std::unique_ptr<VideoCore::RendererBa GPUAsynch::~GPUAsynch() = default; void GPUAsynch::Start() { - cpu_context->MakeCurrent(); gpu_thread.StartThread(*renderer, *gpu_context, *dma_pusher); } +void GPUAsynch::ObtainContext() { + cpu_context->MakeCurrent(); +} + +void GPUAsynch::ReleaseContext() { + cpu_context->DoneCurrent(); +} + void GPUAsynch::PushGPUEntries(Tegra::CommandList&& entries) { gpu_thread.SubmitList(std::move(entries)); } diff --git a/src/video_core/gpu_asynch.h b/src/video_core/gpu_asynch.h index 517658612..15e9f1d38 100644 --- a/src/video_core/gpu_asynch.h +++ b/src/video_core/gpu_asynch.h @@ -25,6 +25,8 @@ public: ~GPUAsynch() override; void Start() override; + void ObtainContext() override; + void ReleaseContext() override; void PushGPUEntries(Tegra::CommandList&& entries) override; void SwapBuffers(const Tegra::FramebufferConfig* framebuffer) override; void FlushRegion(VAddr addr, u64 size) override; diff --git a/src/video_core/gpu_synch.cpp b/src/video_core/gpu_synch.cpp index 6f38a672a..aaeb9811d 100644 --- a/src/video_core/gpu_synch.cpp +++ b/src/video_core/gpu_synch.cpp @@ -13,10 +13,16 @@ GPUSynch::GPUSynch(Core::System& system, std::unique_ptr<VideoCore::RendererBase GPUSynch::~GPUSynch() = default; -void GPUSynch::Start() { +void GPUSynch::Start() {} + +void GPUSynch::ObtainContext() { context->MakeCurrent(); } +void GPUSynch::ReleaseContext() { + context->DoneCurrent(); +} + void GPUSynch::PushGPUEntries(Tegra::CommandList&& entries) { dma_pusher->Push(std::move(entries)); dma_pusher->DispatchCalls(); diff --git a/src/video_core/gpu_synch.h b/src/video_core/gpu_synch.h index 4a6e9a01d..762c20aa5 100644 --- a/src/video_core/gpu_synch.h +++ b/src/video_core/gpu_synch.h @@ -24,6 +24,8 @@ public: ~GPUSynch() override; void Start() override; + void ObtainContext() override; + void ReleaseContext() override; void PushGPUEntries(Tegra::CommandList&& entries) override; void SwapBuffers(const Tegra::FramebufferConfig* framebuffer) override; void FlushRegion(VAddr addr, u64 size) override; diff --git a/src/video_core/gpu_thread.cpp b/src/video_core/gpu_thread.cpp index c3bb4fe06..738c6f0c1 100644 --- a/src/video_core/gpu_thread.cpp +++ b/src/video_core/gpu_thread.cpp @@ -4,6 +4,7 @@ #include "common/assert.h" #include "common/microprofile.h" +#include "common/thread.h" #include "core/core.h" #include "core/frontend/emu_window.h" #include "core/settings.h" @@ -18,7 +19,11 @@ namespace VideoCommon::GPUThread { static void RunThread(Core::System& system, VideoCore::RendererBase& renderer, Core::Frontend::GraphicsContext& context, Tegra::DmaPusher& dma_pusher, SynchState& state) { - MicroProfileOnThreadCreate("GpuThread"); + std::string name = "yuzu:GPU"; + MicroProfileOnThreadCreate(name.c_str()); + Common::SetCurrentThreadName(name.c_str()); + Common::SetCurrentThreadPriority(Common::ThreadPriority::High); + system.RegisterHostThread(); // Wait for first GPU command before acquiring the window context while (state.queue.Empty()) diff --git a/src/video_core/macro/macro.cpp b/src/video_core/macro/macro.cpp new file mode 100644 index 000000000..a50e7b4e0 --- /dev/null +++ b/src/video_core/macro/macro.cpp @@ -0,0 +1,91 @@ +// Copyright 2020 yuzu Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include <optional> +#include <boost/container_hash/hash.hpp> +#include "common/assert.h" +#include "common/logging/log.h" +#include "core/settings.h" +#include "video_core/engines/maxwell_3d.h" +#include "video_core/macro/macro.h" +#include "video_core/macro/macro_hle.h" +#include "video_core/macro/macro_interpreter.h" +#include "video_core/macro/macro_jit_x64.h" + +namespace Tegra { + +MacroEngine::MacroEngine(Engines::Maxwell3D& maxwell3d) + : hle_macros{std::make_unique<Tegra::HLEMacro>(maxwell3d)} {} + +MacroEngine::~MacroEngine() = default; + +void MacroEngine::AddCode(u32 method, u32 data) { + uploaded_macro_code[method].push_back(data); +} + +void MacroEngine::Execute(Engines::Maxwell3D& maxwell3d, u32 method, + const std::vector<u32>& parameters) { + auto compiled_macro = macro_cache.find(method); + if (compiled_macro != macro_cache.end()) { + const auto& cache_info = compiled_macro->second; + if (cache_info.has_hle_program) { + cache_info.hle_program->Execute(parameters, method); + } else { + cache_info.lle_program->Execute(parameters, method); + } + } else { + // Macro not compiled, check if it's uploaded and if so, compile it + std::optional<u32> mid_method = std::nullopt; + const auto macro_code = uploaded_macro_code.find(method); + if (macro_code == uploaded_macro_code.end()) { + for (const auto& [method_base, code] : uploaded_macro_code) { + if (method >= method_base && (method - method_base) < code.size()) { + mid_method = method_base; + break; + } + } + if (!mid_method.has_value()) { + UNREACHABLE_MSG("Macro 0x{0:x} was not uploaded", method); + return; + } + } + auto& cache_info = macro_cache[method]; + + if (!mid_method.has_value()) { + cache_info.lle_program = Compile(macro_code->second); + cache_info.hash = boost::hash_value(macro_code->second); + } else { + const auto& macro_cached = uploaded_macro_code[mid_method.value()]; + const auto rebased_method = method - mid_method.value(); + auto& code = uploaded_macro_code[method]; + code.resize(macro_cached.size() - rebased_method); + std::memcpy(code.data(), macro_cached.data() + rebased_method, + code.size() * sizeof(u32)); + cache_info.hash = boost::hash_value(code); + cache_info.lle_program = Compile(code); + } + + auto hle_program = hle_macros->GetHLEProgram(cache_info.hash); + if (hle_program.has_value()) { + cache_info.has_hle_program = true; + cache_info.hle_program = std::move(hle_program.value()); + cache_info.hle_program->Execute(parameters, method); + } else { + cache_info.lle_program->Execute(parameters, method); + } + } +} + +std::unique_ptr<MacroEngine> GetMacroEngine(Engines::Maxwell3D& maxwell3d) { + if (Settings::values.disable_macro_jit) { + return std::make_unique<MacroInterpreter>(maxwell3d); + } +#ifdef ARCHITECTURE_x86_64 + return std::make_unique<MacroJITx64>(maxwell3d); +#else + return std::make_unique<MacroInterpreter>(maxwell3d); +#endif +} + +} // namespace Tegra diff --git a/src/video_core/macro/macro.h b/src/video_core/macro/macro.h new file mode 100644 index 000000000..4d00b84b0 --- /dev/null +++ b/src/video_core/macro/macro.h @@ -0,0 +1,141 @@ +// Copyright 2020 yuzu Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include <memory> +#include <unordered_map> +#include <vector> +#include "common/bit_field.h" +#include "common/common_types.h" + +namespace Tegra { + +namespace Engines { +class Maxwell3D; +} + +namespace Macro { +constexpr std::size_t NUM_MACRO_REGISTERS = 8; +enum class Operation : u32 { + ALU = 0, + AddImmediate = 1, + ExtractInsert = 2, + ExtractShiftLeftImmediate = 3, + ExtractShiftLeftRegister = 4, + Read = 5, + Unused = 6, // This operation doesn't seem to be a valid encoding. + Branch = 7, +}; + +enum class ALUOperation : u32 { + Add = 0, + AddWithCarry = 1, + Subtract = 2, + SubtractWithBorrow = 3, + // Operations 4-7 don't seem to be valid encodings. + Xor = 8, + Or = 9, + And = 10, + AndNot = 11, + Nand = 12 +}; + +enum class ResultOperation : u32 { + IgnoreAndFetch = 0, + Move = 1, + MoveAndSetMethod = 2, + FetchAndSend = 3, + MoveAndSend = 4, + FetchAndSetMethod = 5, + MoveAndSetMethodFetchAndSend = 6, + MoveAndSetMethodSend = 7 +}; + +enum class BranchCondition : u32 { + Zero = 0, + NotZero = 1, +}; + +union Opcode { + u32 raw; + BitField<0, 3, Operation> operation; + BitField<4, 3, ResultOperation> result_operation; + BitField<4, 1, BranchCondition> branch_condition; + // If set on a branch, then the branch doesn't have a delay slot. + BitField<5, 1, u32> branch_annul; + BitField<7, 1, u32> is_exit; + BitField<8, 3, u32> dst; + BitField<11, 3, u32> src_a; + BitField<14, 3, u32> src_b; + // The signed immediate overlaps the second source operand and the alu operation. + BitField<14, 18, s32> immediate; + + BitField<17, 5, ALUOperation> alu_operation; + + // Bitfield instructions data + BitField<17, 5, u32> bf_src_bit; + BitField<22, 5, u32> bf_size; + BitField<27, 5, u32> bf_dst_bit; + + u32 GetBitfieldMask() const { + return (1 << bf_size) - 1; + } + + s32 GetBranchTarget() const { + return static_cast<s32>(immediate * sizeof(u32)); + } +}; + +union MethodAddress { + u32 raw; + BitField<0, 12, u32> address; + BitField<12, 6, u32> increment; +}; + +} // namespace Macro + +class HLEMacro; + +class CachedMacro { +public: + virtual ~CachedMacro() = default; + /** + * Executes the macro code with the specified input parameters. + * @param code The macro byte code to execute + * @param parameters The parameters of the macro + */ + virtual void Execute(const std::vector<u32>& parameters, u32 method) = 0; +}; + +class MacroEngine { +public: + explicit MacroEngine(Engines::Maxwell3D& maxwell3d); + virtual ~MacroEngine(); + + // Store the uploaded macro code to compile them when they're called. + void AddCode(u32 method, u32 data); + + // Compiles the macro if its not in the cache, and executes the compiled macro + void Execute(Engines::Maxwell3D& maxwell3d, u32 method, const std::vector<u32>& parameters); + +protected: + virtual std::unique_ptr<CachedMacro> Compile(const std::vector<u32>& code) = 0; + +private: + struct CacheInfo { + std::unique_ptr<CachedMacro> lle_program{}; + std::unique_ptr<CachedMacro> hle_program{}; + u64 hash{}; + bool has_hle_program{}; + }; + + std::unordered_map<u32, CacheInfo> macro_cache; + std::unordered_map<u32, std::vector<u32>> uploaded_macro_code; + std::unique_ptr<HLEMacro> hle_macros; +}; + +std::unique_ptr<MacroEngine> GetMacroEngine(Engines::Maxwell3D& maxwell3d); + +} // namespace Tegra diff --git a/src/video_core/macro/macro_hle.cpp b/src/video_core/macro/macro_hle.cpp new file mode 100644 index 000000000..410f99018 --- /dev/null +++ b/src/video_core/macro/macro_hle.cpp @@ -0,0 +1,113 @@ +// Copyright 2020 yuzu Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include <array> +#include <vector> +#include "video_core/engines/maxwell_3d.h" +#include "video_core/macro/macro_hle.h" +#include "video_core/rasterizer_interface.h" + +namespace Tegra { + +namespace { +// HLE'd functions +static void HLE_771BB18C62444DA0(Engines::Maxwell3D& maxwell3d, + const std::vector<u32>& parameters) { + const u32 instance_count = parameters[2] & maxwell3d.GetRegisterValue(0xD1B); + + maxwell3d.regs.draw.topology.Assign( + static_cast<Tegra::Engines::Maxwell3D::Regs::PrimitiveTopology>(parameters[0] & + ~(0x3ffffff << 26))); + maxwell3d.regs.vb_base_instance = parameters[5]; + maxwell3d.mme_draw.instance_count = instance_count; + maxwell3d.regs.vb_element_base = parameters[3]; + maxwell3d.regs.index_array.count = parameters[1]; + maxwell3d.regs.index_array.first = parameters[4]; + + if (maxwell3d.ShouldExecute()) { + maxwell3d.GetRasterizer().Draw(true, true); + } + maxwell3d.regs.index_array.count = 0; + maxwell3d.mme_draw.instance_count = 0; + maxwell3d.mme_draw.current_mode = Engines::Maxwell3D::MMEDrawMode::Undefined; +} + +static void HLE_0D61FC9FAAC9FCAD(Engines::Maxwell3D& maxwell3d, + const std::vector<u32>& parameters) { + const u32 count = (maxwell3d.GetRegisterValue(0xD1B) & parameters[2]); + + maxwell3d.regs.vertex_buffer.first = parameters[3]; + maxwell3d.regs.vertex_buffer.count = parameters[1]; + maxwell3d.regs.vb_base_instance = parameters[4]; + maxwell3d.regs.draw.topology.Assign( + static_cast<Tegra::Engines::Maxwell3D::Regs::PrimitiveTopology>(parameters[0])); + maxwell3d.mme_draw.instance_count = count; + + if (maxwell3d.ShouldExecute()) { + maxwell3d.GetRasterizer().Draw(false, true); + } + maxwell3d.regs.vertex_buffer.count = 0; + maxwell3d.mme_draw.instance_count = 0; + maxwell3d.mme_draw.current_mode = Engines::Maxwell3D::MMEDrawMode::Undefined; +} + +static void HLE_0217920100488FF7(Engines::Maxwell3D& maxwell3d, + const std::vector<u32>& parameters) { + const u32 instance_count = (maxwell3d.GetRegisterValue(0xD1B) & parameters[2]); + const u32 element_base = parameters[4]; + const u32 base_instance = parameters[5]; + maxwell3d.regs.index_array.first = parameters[3]; + maxwell3d.regs.reg_array[0x446] = element_base; // vertex id base? + maxwell3d.regs.index_array.count = parameters[1]; + maxwell3d.regs.vb_element_base = element_base; + maxwell3d.regs.vb_base_instance = base_instance; + maxwell3d.mme_draw.instance_count = instance_count; + maxwell3d.CallMethodFromMME(0x8e3, 0x640); + maxwell3d.CallMethodFromMME(0x8e4, element_base); + maxwell3d.CallMethodFromMME(0x8e5, base_instance); + maxwell3d.regs.draw.topology.Assign( + static_cast<Tegra::Engines::Maxwell3D::Regs::PrimitiveTopology>(parameters[0])); + if (maxwell3d.ShouldExecute()) { + maxwell3d.GetRasterizer().Draw(true, true); + } + maxwell3d.regs.reg_array[0x446] = 0x0; // vertex id base? + maxwell3d.regs.index_array.count = 0; + maxwell3d.regs.vb_element_base = 0x0; + maxwell3d.regs.vb_base_instance = 0x0; + maxwell3d.mme_draw.instance_count = 0; + maxwell3d.CallMethodFromMME(0x8e3, 0x640); + maxwell3d.CallMethodFromMME(0x8e4, 0x0); + maxwell3d.CallMethodFromMME(0x8e5, 0x0); + maxwell3d.mme_draw.current_mode = Engines::Maxwell3D::MMEDrawMode::Undefined; +} +} // namespace + +constexpr std::array<std::pair<u64, HLEFunction>, 3> hle_funcs{{ + std::make_pair<u64, HLEFunction>(0x771BB18C62444DA0, &HLE_771BB18C62444DA0), + std::make_pair<u64, HLEFunction>(0x0D61FC9FAAC9FCAD, &HLE_0D61FC9FAAC9FCAD), + std::make_pair<u64, HLEFunction>(0x0217920100488FF7, &HLE_0217920100488FF7), +}}; + +HLEMacro::HLEMacro(Engines::Maxwell3D& maxwell3d) : maxwell3d(maxwell3d) {} +HLEMacro::~HLEMacro() = default; + +std::optional<std::unique_ptr<CachedMacro>> HLEMacro::GetHLEProgram(u64 hash) const { + const auto it = std::find_if(hle_funcs.cbegin(), hle_funcs.cend(), + [hash](const auto& pair) { return pair.first == hash; }); + if (it == hle_funcs.end()) { + return std::nullopt; + } + return std::make_unique<HLEMacroImpl>(maxwell3d, it->second); +} + +HLEMacroImpl::~HLEMacroImpl() = default; + +HLEMacroImpl::HLEMacroImpl(Engines::Maxwell3D& maxwell3d, HLEFunction func) + : maxwell3d(maxwell3d), func(func) {} + +void HLEMacroImpl::Execute(const std::vector<u32>& parameters, u32 method) { + func(maxwell3d, parameters); +} + +} // namespace Tegra diff --git a/src/video_core/macro/macro_hle.h b/src/video_core/macro/macro_hle.h new file mode 100644 index 000000000..37af875a0 --- /dev/null +++ b/src/video_core/macro/macro_hle.h @@ -0,0 +1,44 @@ +// Copyright 2020 yuzu Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include <memory> +#include <optional> +#include <vector> +#include "common/common_types.h" +#include "video_core/macro/macro.h" + +namespace Tegra { + +namespace Engines { +class Maxwell3D; +} + +using HLEFunction = void (*)(Engines::Maxwell3D& maxwell3d, const std::vector<u32>& parameters); + +class HLEMacro { +public: + explicit HLEMacro(Engines::Maxwell3D& maxwell3d); + ~HLEMacro(); + + std::optional<std::unique_ptr<CachedMacro>> GetHLEProgram(u64 hash) const; + +private: + Engines::Maxwell3D& maxwell3d; +}; + +class HLEMacroImpl : public CachedMacro { +public: + explicit HLEMacroImpl(Engines::Maxwell3D& maxwell3d, HLEFunction func); + ~HLEMacroImpl(); + + void Execute(const std::vector<u32>& parameters, u32 method) override; + +private: + Engines::Maxwell3D& maxwell3d; + HLEFunction func; +}; + +} // namespace Tegra diff --git a/src/video_core/macro_interpreter.cpp b/src/video_core/macro/macro_interpreter.cpp index 947364928..aa5256419 100644 --- a/src/video_core/macro_interpreter.cpp +++ b/src/video_core/macro/macro_interpreter.cpp @@ -1,4 +1,4 @@ -// Copyright 2018 yuzu Emulator Project +// Copyright 2020 yuzu Emulator Project // Licensed under GPLv2 or any later version // Refer to the license.txt file included. @@ -6,109 +6,47 @@ #include "common/logging/log.h" #include "common/microprofile.h" #include "video_core/engines/maxwell_3d.h" -#include "video_core/macro_interpreter.h" +#include "video_core/macro/macro_interpreter.h" MICROPROFILE_DEFINE(MacroInterp, "GPU", "Execute macro interpreter", MP_RGB(128, 128, 192)); namespace Tegra { -namespace { -enum class Operation : u32 { - ALU = 0, - AddImmediate = 1, - ExtractInsert = 2, - ExtractShiftLeftImmediate = 3, - ExtractShiftLeftRegister = 4, - Read = 5, - Unused = 6, // This operation doesn't seem to be a valid encoding. - Branch = 7, -}; -} // Anonymous namespace - -enum class MacroInterpreter::ALUOperation : u32 { - Add = 0, - AddWithCarry = 1, - Subtract = 2, - SubtractWithBorrow = 3, - // Operations 4-7 don't seem to be valid encodings. - Xor = 8, - Or = 9, - And = 10, - AndNot = 11, - Nand = 12 -}; - -enum class MacroInterpreter::ResultOperation : u32 { - IgnoreAndFetch = 0, - Move = 1, - MoveAndSetMethod = 2, - FetchAndSend = 3, - MoveAndSend = 4, - FetchAndSetMethod = 5, - MoveAndSetMethodFetchAndSend = 6, - MoveAndSetMethodSend = 7 -}; - -enum class MacroInterpreter::BranchCondition : u32 { - Zero = 0, - NotZero = 1, -}; - -union MacroInterpreter::Opcode { - u32 raw; - BitField<0, 3, Operation> operation; - BitField<4, 3, ResultOperation> result_operation; - BitField<4, 1, BranchCondition> branch_condition; - // If set on a branch, then the branch doesn't have a delay slot. - BitField<5, 1, u32> branch_annul; - BitField<7, 1, u32> is_exit; - BitField<8, 3, u32> dst; - BitField<11, 3, u32> src_a; - BitField<14, 3, u32> src_b; - // The signed immediate overlaps the second source operand and the alu operation. - BitField<14, 18, s32> immediate; - - BitField<17, 5, ALUOperation> alu_operation; - - // Bitfield instructions data - BitField<17, 5, u32> bf_src_bit; - BitField<22, 5, u32> bf_size; - BitField<27, 5, u32> bf_dst_bit; - - u32 GetBitfieldMask() const { - return (1 << bf_size) - 1; - } +MacroInterpreter::MacroInterpreter(Engines::Maxwell3D& maxwell3d) + : MacroEngine::MacroEngine(maxwell3d), maxwell3d(maxwell3d) {} - s32 GetBranchTarget() const { - return static_cast<s32>(immediate * sizeof(u32)); - } -}; +std::unique_ptr<CachedMacro> MacroInterpreter::Compile(const std::vector<u32>& code) { + return std::make_unique<MacroInterpreterImpl>(maxwell3d, code); +} -MacroInterpreter::MacroInterpreter(Engines::Maxwell3D& maxwell3d) : maxwell3d(maxwell3d) {} +MacroInterpreterImpl::MacroInterpreterImpl(Engines::Maxwell3D& maxwell3d, + const std::vector<u32>& code) + : maxwell3d(maxwell3d), code(code) {} -void MacroInterpreter::Execute(u32 offset, std::size_t num_parameters, const u32* parameters) { +void MacroInterpreterImpl::Execute(const std::vector<u32>& parameters, u32 method) { MICROPROFILE_SCOPE(MacroInterp); Reset(); registers[1] = parameters[0]; + num_parameters = parameters.size(); if (num_parameters > parameters_capacity) { parameters_capacity = num_parameters; this->parameters = std::make_unique<u32[]>(num_parameters); } - std::memcpy(this->parameters.get(), parameters, num_parameters * sizeof(u32)); + std::memcpy(this->parameters.get(), parameters.data(), num_parameters * sizeof(u32)); this->num_parameters = num_parameters; // Execute the code until we hit an exit condition. bool keep_executing = true; while (keep_executing) { - keep_executing = Step(offset, false); + keep_executing = Step(false); } // Assert the the macro used all the input parameters ASSERT(next_parameter_index == num_parameters); } -void MacroInterpreter::Reset() { +void MacroInterpreterImpl::Reset() { registers = {}; pc = 0; delayed_pc = {}; @@ -120,10 +58,10 @@ void MacroInterpreter::Reset() { carry_flag = false; } -bool MacroInterpreter::Step(u32 offset, bool is_delay_slot) { +bool MacroInterpreterImpl::Step(bool is_delay_slot) { u32 base_address = pc; - Opcode opcode = GetOpcode(offset); + Macro::Opcode opcode = GetOpcode(); pc += 4; // Update the program counter if we were delayed @@ -134,18 +72,18 @@ bool MacroInterpreter::Step(u32 offset, bool is_delay_slot) { } switch (opcode.operation) { - case Operation::ALU: { + case Macro::Operation::ALU: { u32 result = GetALUResult(opcode.alu_operation, GetRegister(opcode.src_a), GetRegister(opcode.src_b)); ProcessResult(opcode.result_operation, opcode.dst, result); break; } - case Operation::AddImmediate: { + case Macro::Operation::AddImmediate: { ProcessResult(opcode.result_operation, opcode.dst, GetRegister(opcode.src_a) + opcode.immediate); break; } - case Operation::ExtractInsert: { + case Macro::Operation::ExtractInsert: { u32 dst = GetRegister(opcode.src_a); u32 src = GetRegister(opcode.src_b); @@ -155,7 +93,7 @@ bool MacroInterpreter::Step(u32 offset, bool is_delay_slot) { ProcessResult(opcode.result_operation, opcode.dst, dst); break; } - case Operation::ExtractShiftLeftImmediate: { + case Macro::Operation::ExtractShiftLeftImmediate: { u32 dst = GetRegister(opcode.src_a); u32 src = GetRegister(opcode.src_b); @@ -164,7 +102,7 @@ bool MacroInterpreter::Step(u32 offset, bool is_delay_slot) { ProcessResult(opcode.result_operation, opcode.dst, result); break; } - case Operation::ExtractShiftLeftRegister: { + case Macro::Operation::ExtractShiftLeftRegister: { u32 dst = GetRegister(opcode.src_a); u32 src = GetRegister(opcode.src_b); @@ -173,12 +111,12 @@ bool MacroInterpreter::Step(u32 offset, bool is_delay_slot) { ProcessResult(opcode.result_operation, opcode.dst, result); break; } - case Operation::Read: { + case Macro::Operation::Read: { u32 result = Read(GetRegister(opcode.src_a) + opcode.immediate); ProcessResult(opcode.result_operation, opcode.dst, result); break; } - case Operation::Branch: { + case Macro::Operation::Branch: { ASSERT_MSG(!is_delay_slot, "Executing a branch in a delay slot is not valid"); u32 value = GetRegister(opcode.src_a); bool taken = EvaluateBranchCondition(opcode.branch_condition, value); @@ -191,7 +129,7 @@ bool MacroInterpreter::Step(u32 offset, bool is_delay_slot) { delayed_pc = base_address + opcode.GetBranchTarget(); // Execute one more instruction due to the delay slot. - return Step(offset, true); + return Step(true); } break; } @@ -204,51 +142,44 @@ bool MacroInterpreter::Step(u32 offset, bool is_delay_slot) { // cause an exit if it's executed inside a delay slot. if (opcode.is_exit && !is_delay_slot) { // Exit has a delay slot, execute the next instruction - Step(offset, true); + Step(true); return false; } return true; } -MacroInterpreter::Opcode MacroInterpreter::GetOpcode(u32 offset) const { - const auto& macro_memory{maxwell3d.GetMacroMemory()}; - ASSERT((pc % sizeof(u32)) == 0); - ASSERT((pc + offset) < macro_memory.size() * sizeof(u32)); - return {macro_memory[offset + pc / sizeof(u32)]}; -} - -u32 MacroInterpreter::GetALUResult(ALUOperation operation, u32 src_a, u32 src_b) { +u32 MacroInterpreterImpl::GetALUResult(Macro::ALUOperation operation, u32 src_a, u32 src_b) { switch (operation) { - case ALUOperation::Add: { + case Macro::ALUOperation::Add: { const u64 result{static_cast<u64>(src_a) + src_b}; carry_flag = result > 0xffffffff; return static_cast<u32>(result); } - case ALUOperation::AddWithCarry: { + case Macro::ALUOperation::AddWithCarry: { const u64 result{static_cast<u64>(src_a) + src_b + (carry_flag ? 1ULL : 0ULL)}; carry_flag = result > 0xffffffff; return static_cast<u32>(result); } - case ALUOperation::Subtract: { + case Macro::ALUOperation::Subtract: { const u64 result{static_cast<u64>(src_a) - src_b}; carry_flag = result < 0x100000000; return static_cast<u32>(result); } - case ALUOperation::SubtractWithBorrow: { + case Macro::ALUOperation::SubtractWithBorrow: { const u64 result{static_cast<u64>(src_a) - src_b - (carry_flag ? 0ULL : 1ULL)}; carry_flag = result < 0x100000000; return static_cast<u32>(result); } - case ALUOperation::Xor: + case Macro::ALUOperation::Xor: return src_a ^ src_b; - case ALUOperation::Or: + case Macro::ALUOperation::Or: return src_a | src_b; - case ALUOperation::And: + case Macro::ALUOperation::And: return src_a & src_b; - case ALUOperation::AndNot: + case Macro::ALUOperation::AndNot: return src_a & ~src_b; - case ALUOperation::Nand: + case Macro::ALUOperation::Nand: return ~(src_a & src_b); default: @@ -257,43 +188,43 @@ u32 MacroInterpreter::GetALUResult(ALUOperation operation, u32 src_a, u32 src_b) } } -void MacroInterpreter::ProcessResult(ResultOperation operation, u32 reg, u32 result) { +void MacroInterpreterImpl::ProcessResult(Macro::ResultOperation operation, u32 reg, u32 result) { switch (operation) { - case ResultOperation::IgnoreAndFetch: + case Macro::ResultOperation::IgnoreAndFetch: // Fetch parameter and ignore result. SetRegister(reg, FetchParameter()); break; - case ResultOperation::Move: + case Macro::ResultOperation::Move: // Move result. SetRegister(reg, result); break; - case ResultOperation::MoveAndSetMethod: + case Macro::ResultOperation::MoveAndSetMethod: // Move result and use as Method Address. SetRegister(reg, result); SetMethodAddress(result); break; - case ResultOperation::FetchAndSend: + case Macro::ResultOperation::FetchAndSend: // Fetch parameter and send result. SetRegister(reg, FetchParameter()); Send(result); break; - case ResultOperation::MoveAndSend: + case Macro::ResultOperation::MoveAndSend: // Move and send result. SetRegister(reg, result); Send(result); break; - case ResultOperation::FetchAndSetMethod: + case Macro::ResultOperation::FetchAndSetMethod: // Fetch parameter and use result as Method Address. SetRegister(reg, FetchParameter()); SetMethodAddress(result); break; - case ResultOperation::MoveAndSetMethodFetchAndSend: + case Macro::ResultOperation::MoveAndSetMethodFetchAndSend: // Move result and use as Method Address, then fetch and send parameter. SetRegister(reg, result); SetMethodAddress(result); Send(FetchParameter()); break; - case ResultOperation::MoveAndSetMethodSend: + case Macro::ResultOperation::MoveAndSetMethodSend: // Move result and use as Method Address, then send bits 12:17 of result. SetRegister(reg, result); SetMethodAddress(result); @@ -304,16 +235,28 @@ void MacroInterpreter::ProcessResult(ResultOperation operation, u32 reg, u32 res } } -u32 MacroInterpreter::FetchParameter() { - ASSERT(next_parameter_index < num_parameters); - return parameters[next_parameter_index++]; +bool MacroInterpreterImpl::EvaluateBranchCondition(Macro::BranchCondition cond, u32 value) const { + switch (cond) { + case Macro::BranchCondition::Zero: + return value == 0; + case Macro::BranchCondition::NotZero: + return value != 0; + } + UNREACHABLE(); + return true; +} + +Macro::Opcode MacroInterpreterImpl::GetOpcode() const { + ASSERT((pc % sizeof(u32)) == 0); + ASSERT(pc < code.size() * sizeof(u32)); + return {code[pc / sizeof(u32)]}; } -u32 MacroInterpreter::GetRegister(u32 register_id) const { +u32 MacroInterpreterImpl::GetRegister(u32 register_id) const { return registers.at(register_id); } -void MacroInterpreter::SetRegister(u32 register_id, u32 value) { +void MacroInterpreterImpl::SetRegister(u32 register_id, u32 value) { // Register 0 is hardwired as the zero register. // Ensure no writes to it actually occur. if (register_id == 0) { @@ -323,30 +266,24 @@ void MacroInterpreter::SetRegister(u32 register_id, u32 value) { registers.at(register_id) = value; } -void MacroInterpreter::SetMethodAddress(u32 address) { +void MacroInterpreterImpl::SetMethodAddress(u32 address) { method_address.raw = address; } -void MacroInterpreter::Send(u32 value) { +void MacroInterpreterImpl::Send(u32 value) { maxwell3d.CallMethodFromMME(method_address.address, value); // Increment the method address by the method increment. method_address.address.Assign(method_address.address.Value() + method_address.increment.Value()); } -u32 MacroInterpreter::Read(u32 method) const { +u32 MacroInterpreterImpl::Read(u32 method) const { return maxwell3d.GetRegisterValue(method); } -bool MacroInterpreter::EvaluateBranchCondition(BranchCondition cond, u32 value) const { - switch (cond) { - case BranchCondition::Zero: - return value == 0; - case BranchCondition::NotZero: - return value != 0; - } - UNREACHABLE(); - return true; +u32 MacroInterpreterImpl::FetchParameter() { + ASSERT(next_parameter_index < num_parameters); + return parameters[next_parameter_index++]; } } // namespace Tegra diff --git a/src/video_core/macro_interpreter.h b/src/video_core/macro/macro_interpreter.h index 631146d89..90217fc89 100644 --- a/src/video_core/macro_interpreter.h +++ b/src/video_core/macro/macro_interpreter.h @@ -1,44 +1,37 @@ -// Copyright 2018 yuzu Emulator Project +// Copyright 2020 yuzu Emulator Project // Licensed under GPLv2 or any later version // Refer to the license.txt file included. #pragma once - #include <array> #include <optional> - +#include <vector> #include "common/bit_field.h" #include "common/common_types.h" +#include "video_core/macro/macro.h" namespace Tegra { namespace Engines { class Maxwell3D; } -class MacroInterpreter final { +class MacroInterpreter final : public MacroEngine { public: explicit MacroInterpreter(Engines::Maxwell3D& maxwell3d); - /** - * Executes the macro code with the specified input parameters. - * @param offset Offset to start execution at. - * @param parameters The parameters of the macro. - */ - void Execute(u32 offset, std::size_t num_parameters, const u32* parameters); +protected: + std::unique_ptr<CachedMacro> Compile(const std::vector<u32>& code) override; private: - enum class ALUOperation : u32; - enum class BranchCondition : u32; - enum class ResultOperation : u32; - - union Opcode; + Engines::Maxwell3D& maxwell3d; +}; - union MethodAddress { - u32 raw; - BitField<0, 12, u32> address; - BitField<12, 6, u32> increment; - }; +class MacroInterpreterImpl : public CachedMacro { +public: + MacroInterpreterImpl(Engines::Maxwell3D& maxwell3d, const std::vector<u32>& code); + void Execute(const std::vector<u32>& parameters, u32 method) override; +private: /// Resets the execution engine state, zeroing registers, etc. void Reset(); @@ -49,20 +42,20 @@ private: * @param is_delay_slot Whether the current step is being executed due to a delay slot in a * previous instruction. */ - bool Step(u32 offset, bool is_delay_slot); + bool Step(bool is_delay_slot); /// Calculates the result of an ALU operation. src_a OP src_b; - u32 GetALUResult(ALUOperation operation, u32 src_a, u32 src_b); + u32 GetALUResult(Macro::ALUOperation operation, u32 src_a, u32 src_b); /// Performs the result operation on the input result and stores it in the specified register /// (if necessary). - void ProcessResult(ResultOperation operation, u32 reg, u32 result); + void ProcessResult(Macro::ResultOperation operation, u32 reg, u32 result); /// Evaluates the branch condition and returns whether the branch should be taken or not. - bool EvaluateBranchCondition(BranchCondition cond, u32 value) const; + bool EvaluateBranchCondition(Macro::BranchCondition cond, u32 value) const; /// Reads an opcode at the current program counter location. - Opcode GetOpcode(u32 offset) const; + Macro::Opcode GetOpcode() const; /// Returns the specified register's value. Register 0 is hardcoded to always return 0. u32 GetRegister(u32 register_id) const; @@ -89,13 +82,11 @@ private: /// Program counter to execute at after the delay slot is executed. std::optional<u32> delayed_pc; - static constexpr std::size_t NumMacroRegisters = 8; - /// General purpose macro registers. - std::array<u32, NumMacroRegisters> registers = {}; + std::array<u32, Macro::NUM_MACRO_REGISTERS> registers = {}; /// Method address to use for the next Send instruction. - MethodAddress method_address = {}; + Macro::MethodAddress method_address = {}; /// Input parameters of the current macro. std::unique_ptr<u32[]> parameters; @@ -105,5 +96,7 @@ private: u32 next_parameter_index = 0; bool carry_flag = false; + const std::vector<u32>& code; }; + } // namespace Tegra diff --git a/src/video_core/macro/macro_jit_x64.cpp b/src/video_core/macro/macro_jit_x64.cpp new file mode 100644 index 000000000..07292702f --- /dev/null +++ b/src/video_core/macro/macro_jit_x64.cpp @@ -0,0 +1,621 @@ +// Copyright 2020 yuzu Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include "common/assert.h" +#include "common/logging/log.h" +#include "common/microprofile.h" +#include "common/x64/xbyak_util.h" +#include "video_core/engines/maxwell_3d.h" +#include "video_core/macro/macro_interpreter.h" +#include "video_core/macro/macro_jit_x64.h" + +MICROPROFILE_DEFINE(MacroJitCompile, "GPU", "Compile macro JIT", MP_RGB(173, 255, 47)); +MICROPROFILE_DEFINE(MacroJitExecute, "GPU", "Execute macro JIT", MP_RGB(255, 255, 0)); + +namespace Tegra { +static const Xbyak::Reg64 STATE = Xbyak::util::rbx; +static const Xbyak::Reg32 RESULT = Xbyak::util::ebp; +static const Xbyak::Reg64 PARAMETERS = Xbyak::util::r12; +static const Xbyak::Reg32 METHOD_ADDRESS = Xbyak::util::r14d; +static const Xbyak::Reg64 BRANCH_HOLDER = Xbyak::util::r15; + +static const std::bitset<32> PERSISTENT_REGISTERS = Common::X64::BuildRegSet({ + STATE, + RESULT, + PARAMETERS, + METHOD_ADDRESS, + BRANCH_HOLDER, +}); + +MacroJITx64::MacroJITx64(Engines::Maxwell3D& maxwell3d) + : MacroEngine::MacroEngine(maxwell3d), maxwell3d(maxwell3d) {} + +std::unique_ptr<CachedMacro> MacroJITx64::Compile(const std::vector<u32>& code) { + return std::make_unique<MacroJITx64Impl>(maxwell3d, code); +} + +MacroJITx64Impl::MacroJITx64Impl(Engines::Maxwell3D& maxwell3d, const std::vector<u32>& code) + : Xbyak::CodeGenerator(MAX_CODE_SIZE), code(code), maxwell3d(maxwell3d) { + Compile(); +} + +MacroJITx64Impl::~MacroJITx64Impl() = default; + +void MacroJITx64Impl::Execute(const std::vector<u32>& parameters, u32 method) { + MICROPROFILE_SCOPE(MacroJitExecute); + ASSERT_OR_EXECUTE(program != nullptr, { return; }); + JITState state{}; + state.maxwell3d = &maxwell3d; + state.registers = {}; + program(&state, parameters.data()); +} + +void MacroJITx64Impl::Compile_ALU(Macro::Opcode opcode) { + const bool is_a_zero = opcode.src_a == 0; + const bool is_b_zero = opcode.src_b == 0; + const bool valid_operation = !is_a_zero && !is_b_zero; + [[maybe_unused]] const bool is_move_operation = !is_a_zero && is_b_zero; + const bool has_zero_register = is_a_zero || is_b_zero; + const bool no_zero_reg_skip = opcode.alu_operation == Macro::ALUOperation::AddWithCarry || + opcode.alu_operation == Macro::ALUOperation::SubtractWithBorrow; + + Xbyak::Reg32 src_a; + Xbyak::Reg32 src_b; + + if (!optimizer.zero_reg_skip || no_zero_reg_skip) { + src_a = Compile_GetRegister(opcode.src_a, RESULT); + src_b = Compile_GetRegister(opcode.src_b, eax); + } else { + if (!is_a_zero) { + src_a = Compile_GetRegister(opcode.src_a, RESULT); + } + if (!is_b_zero) { + src_b = Compile_GetRegister(opcode.src_b, eax); + } + } + + bool has_emitted = false; + + switch (opcode.alu_operation) { + case Macro::ALUOperation::Add: + if (optimizer.zero_reg_skip) { + if (valid_operation) { + add(src_a, src_b); + } + } else { + add(src_a, src_b); + } + + if (!optimizer.can_skip_carry) { + setc(byte[STATE + offsetof(JITState, carry_flag)]); + } + break; + case Macro::ALUOperation::AddWithCarry: + bt(dword[STATE + offsetof(JITState, carry_flag)], 0); + adc(src_a, src_b); + setc(byte[STATE + offsetof(JITState, carry_flag)]); + break; + case Macro::ALUOperation::Subtract: + if (optimizer.zero_reg_skip) { + if (valid_operation) { + sub(src_a, src_b); + has_emitted = true; + } + } else { + sub(src_a, src_b); + has_emitted = true; + } + if (!optimizer.can_skip_carry && has_emitted) { + setc(byte[STATE + offsetof(JITState, carry_flag)]); + } + break; + case Macro::ALUOperation::SubtractWithBorrow: + bt(dword[STATE + offsetof(JITState, carry_flag)], 0); + sbb(src_a, src_b); + setc(byte[STATE + offsetof(JITState, carry_flag)]); + break; + case Macro::ALUOperation::Xor: + if (optimizer.zero_reg_skip) { + if (valid_operation) { + xor_(src_a, src_b); + } + } else { + xor_(src_a, src_b); + } + break; + case Macro::ALUOperation::Or: + if (optimizer.zero_reg_skip) { + if (valid_operation) { + or_(src_a, src_b); + } + } else { + or_(src_a, src_b); + } + break; + case Macro::ALUOperation::And: + if (optimizer.zero_reg_skip) { + if (!has_zero_register) { + and_(src_a, src_b); + } + } else { + and_(src_a, src_b); + } + break; + case Macro::ALUOperation::AndNot: + if (optimizer.zero_reg_skip) { + if (!is_a_zero) { + not_(src_b); + and_(src_a, src_b); + } + } else { + not_(src_b); + and_(src_a, src_b); + } + break; + case Macro::ALUOperation::Nand: + if (optimizer.zero_reg_skip) { + if (!is_a_zero) { + and_(src_a, src_b); + not_(src_a); + } + } else { + and_(src_a, src_b); + not_(src_a); + } + break; + default: + UNIMPLEMENTED_MSG("Unimplemented ALU operation {}", + static_cast<std::size_t>(opcode.alu_operation.Value())); + break; + } + Compile_ProcessResult(opcode.result_operation, opcode.dst); +} + +void MacroJITx64Impl::Compile_AddImmediate(Macro::Opcode opcode) { + if (optimizer.skip_dummy_addimmediate) { + // Games tend to use this as an exit instruction placeholder. It's to encode an instruction + // without doing anything. In our case we can just not emit anything. + if (opcode.result_operation == Macro::ResultOperation::Move && opcode.dst == 0) { + return; + } + } + // Check for redundant moves + if (optimizer.optimize_for_method_move && + opcode.result_operation == Macro::ResultOperation::MoveAndSetMethod) { + if (next_opcode.has_value()) { + const auto next = *next_opcode; + if (next.result_operation == Macro::ResultOperation::MoveAndSetMethod && + opcode.dst == next.dst) { + return; + } + } + } + if (optimizer.zero_reg_skip && opcode.src_a == 0) { + if (opcode.immediate == 0) { + xor_(RESULT, RESULT); + } else { + mov(RESULT, opcode.immediate); + } + } else { + auto result = Compile_GetRegister(opcode.src_a, RESULT); + if (opcode.immediate > 2) { + add(result, opcode.immediate); + } else if (opcode.immediate == 1) { + inc(result); + } else if (opcode.immediate < 0) { + sub(result, opcode.immediate * -1); + } + } + Compile_ProcessResult(opcode.result_operation, opcode.dst); +} + +void MacroJITx64Impl::Compile_ExtractInsert(Macro::Opcode opcode) { + auto dst = Compile_GetRegister(opcode.src_a, RESULT); + auto src = Compile_GetRegister(opcode.src_b, eax); + + if (opcode.bf_src_bit != 0 && opcode.bf_src_bit != 31) { + shr(src, opcode.bf_src_bit); + } else if (opcode.bf_src_bit == 31) { + xor_(src, src); + } + // Don't bother masking the whole register since we're using a 32 bit register + if (opcode.bf_size != 31 && opcode.bf_size != 0) { + and_(src, opcode.GetBitfieldMask()); + } else if (opcode.bf_size == 0) { + xor_(src, src); + } + if (opcode.bf_dst_bit != 31 && opcode.bf_dst_bit != 0) { + shl(src, opcode.bf_dst_bit); + } else if (opcode.bf_dst_bit == 31) { + xor_(src, src); + } + + const u32 mask = ~(opcode.GetBitfieldMask() << opcode.bf_dst_bit); + if (mask != 0xffffffff) { + and_(dst, mask); + } + or_(dst, src); + Compile_ProcessResult(opcode.result_operation, opcode.dst); +} + +void MacroJITx64Impl::Compile_ExtractShiftLeftImmediate(Macro::Opcode opcode) { + const auto dst = Compile_GetRegister(opcode.src_a, ecx); + const auto src = Compile_GetRegister(opcode.src_b, RESULT); + + shr(src, dst.cvt8()); + if (opcode.bf_size != 0 && opcode.bf_size != 31) { + and_(src, opcode.GetBitfieldMask()); + } else if (opcode.bf_size == 0) { + xor_(src, src); + } + + if (opcode.bf_dst_bit != 0 && opcode.bf_dst_bit != 31) { + shl(src, opcode.bf_dst_bit); + } else if (opcode.bf_dst_bit == 31) { + xor_(src, src); + } + Compile_ProcessResult(opcode.result_operation, opcode.dst); +} + +void MacroJITx64Impl::Compile_ExtractShiftLeftRegister(Macro::Opcode opcode) { + const auto dst = Compile_GetRegister(opcode.src_a, ecx); + const auto src = Compile_GetRegister(opcode.src_b, RESULT); + + if (opcode.bf_src_bit != 0) { + shr(src, opcode.bf_src_bit); + } + + if (opcode.bf_size != 31) { + and_(src, opcode.GetBitfieldMask()); + } + shl(src, dst.cvt8()); + + Compile_ProcessResult(opcode.result_operation, opcode.dst); +} + +void MacroJITx64Impl::Compile_Read(Macro::Opcode opcode) { + if (optimizer.zero_reg_skip && opcode.src_a == 0) { + if (opcode.immediate == 0) { + xor_(RESULT, RESULT); + } else { + mov(RESULT, opcode.immediate); + } + } else { + auto result = Compile_GetRegister(opcode.src_a, RESULT); + if (opcode.immediate > 2) { + add(result, opcode.immediate); + } else if (opcode.immediate == 1) { + inc(result); + } else if (opcode.immediate < 0) { + sub(result, opcode.immediate * -1); + } + } + + // Equivalent to Engines::Maxwell3D::GetRegisterValue: + if (optimizer.enable_asserts) { + Xbyak::Label pass_range_check; + cmp(RESULT, static_cast<u32>(Engines::Maxwell3D::Regs::NUM_REGS)); + jb(pass_range_check); + int3(); + L(pass_range_check); + } + mov(rax, qword[STATE]); + mov(RESULT, + dword[rax + offsetof(Engines::Maxwell3D, regs) + + offsetof(Engines::Maxwell3D::Regs, reg_array) + RESULT.cvt64() * sizeof(u32)]); + + Compile_ProcessResult(opcode.result_operation, opcode.dst); +} + +static void Send(Engines::Maxwell3D* maxwell3d, Macro::MethodAddress method_address, u32 value) { + maxwell3d->CallMethodFromMME(method_address.address, value); +} + +void Tegra::MacroJITx64Impl::Compile_Send(Xbyak::Reg32 value) { + Common::X64::ABI_PushRegistersAndAdjustStack(*this, PersistentCallerSavedRegs(), 0); + mov(Common::X64::ABI_PARAM1, qword[STATE]); + mov(Common::X64::ABI_PARAM2, METHOD_ADDRESS); + mov(Common::X64::ABI_PARAM3, value); + Common::X64::CallFarFunction(*this, &Send); + Common::X64::ABI_PopRegistersAndAdjustStack(*this, PersistentCallerSavedRegs(), 0); + + Xbyak::Label dont_process{}; + // Get increment + test(METHOD_ADDRESS, 0x3f000); + // If zero, method address doesn't update + je(dont_process); + + mov(ecx, METHOD_ADDRESS); + and_(METHOD_ADDRESS, 0xfff); + shr(ecx, 12); + and_(ecx, 0x3f); + lea(eax, ptr[rcx + METHOD_ADDRESS.cvt64()]); + sal(ecx, 12); + or_(eax, ecx); + + mov(METHOD_ADDRESS, eax); + + L(dont_process); +} + +void Tegra::MacroJITx64Impl::Compile_Branch(Macro::Opcode opcode) { + ASSERT_MSG(!is_delay_slot, "Executing a branch in a delay slot is not valid"); + const s32 jump_address = + static_cast<s32>(pc) + static_cast<s32>(opcode.GetBranchTarget() / sizeof(s32)); + + Xbyak::Label end; + auto value = Compile_GetRegister(opcode.src_a, eax); + test(value, value); + if (optimizer.has_delayed_pc) { + switch (opcode.branch_condition) { + case Macro::BranchCondition::Zero: + jne(end, T_NEAR); + break; + case Macro::BranchCondition::NotZero: + je(end, T_NEAR); + break; + } + + if (opcode.branch_annul) { + xor_(BRANCH_HOLDER, BRANCH_HOLDER); + jmp(labels[jump_address], T_NEAR); + } else { + Xbyak::Label handle_post_exit{}; + Xbyak::Label skip{}; + jmp(skip, T_NEAR); + if (opcode.is_exit) { + L(handle_post_exit); + // Execute 1 instruction + mov(BRANCH_HOLDER, end_of_code); + // Jump to next instruction to skip delay slot check + jmp(labels[jump_address], T_NEAR); + } else { + L(handle_post_exit); + xor_(BRANCH_HOLDER, BRANCH_HOLDER); + jmp(labels[jump_address], T_NEAR); + } + L(skip); + mov(BRANCH_HOLDER, handle_post_exit); + jmp(delay_skip[pc], T_NEAR); + } + } else { + switch (opcode.branch_condition) { + case Macro::BranchCondition::Zero: + je(labels[jump_address], T_NEAR); + break; + case Macro::BranchCondition::NotZero: + jne(labels[jump_address], T_NEAR); + break; + } + } + + L(end); +} + +void Tegra::MacroJITx64Impl::Optimizer_ScanFlags() { + optimizer.can_skip_carry = true; + optimizer.has_delayed_pc = false; + for (auto raw_op : code) { + Macro::Opcode op{}; + op.raw = raw_op; + + if (op.operation == Macro::Operation::ALU) { + // Scan for any ALU operations which actually use the carry flag, if they don't exist in + // our current code we can skip emitting the carry flag handling operations + if (op.alu_operation == Macro::ALUOperation::AddWithCarry || + op.alu_operation == Macro::ALUOperation::SubtractWithBorrow) { + optimizer.can_skip_carry = false; + } + } + + if (op.operation == Macro::Operation::Branch) { + if (!op.branch_annul) { + optimizer.has_delayed_pc = true; + } + } + } +} + +void MacroJITx64Impl::Compile() { + MICROPROFILE_SCOPE(MacroJitCompile); + bool keep_executing = true; + labels.fill(Xbyak::Label()); + + Common::X64::ABI_PushRegistersAndAdjustStack(*this, Common::X64::ABI_ALL_CALLEE_SAVED, 8); + // JIT state + mov(STATE, Common::X64::ABI_PARAM1); + mov(PARAMETERS, Common::X64::ABI_PARAM2); + xor_(RESULT, RESULT); + xor_(METHOD_ADDRESS, METHOD_ADDRESS); + xor_(BRANCH_HOLDER, BRANCH_HOLDER); + + mov(dword[STATE + offsetof(JITState, registers) + 4], Compile_FetchParameter()); + + // Track get register for zero registers and mark it as no-op + optimizer.zero_reg_skip = true; + + // AddImmediate tends to be used as a NOP instruction, if we detect this we can + // completely skip the entire code path and no emit anything + optimizer.skip_dummy_addimmediate = true; + + // SMO tends to emit a lot of unnecessary method moves, we can mitigate this by only emitting + // one if our register isn't "dirty" + optimizer.optimize_for_method_move = true; + + // Enable run-time assertions in JITted code + optimizer.enable_asserts = false; + + // Check to see if we can skip emitting certain instructions + Optimizer_ScanFlags(); + + const u32 op_count = static_cast<u32>(code.size()); + for (u32 i = 0; i < op_count; i++) { + if (i < op_count - 1) { + pc = i + 1; + next_opcode = GetOpCode(); + } else { + next_opcode = {}; + } + pc = i; + Compile_NextInstruction(); + } + + L(end_of_code); + + Common::X64::ABI_PopRegistersAndAdjustStack(*this, Common::X64::ABI_ALL_CALLEE_SAVED, 8); + ret(); + ready(); + program = getCode<ProgramType>(); +} + +bool MacroJITx64Impl::Compile_NextInstruction() { + const auto opcode = GetOpCode(); + if (labels[pc].getAddress()) { + return false; + } + + L(labels[pc]); + + switch (opcode.operation) { + case Macro::Operation::ALU: + Compile_ALU(opcode); + break; + case Macro::Operation::AddImmediate: + Compile_AddImmediate(opcode); + break; + case Macro::Operation::ExtractInsert: + Compile_ExtractInsert(opcode); + break; + case Macro::Operation::ExtractShiftLeftImmediate: + Compile_ExtractShiftLeftImmediate(opcode); + break; + case Macro::Operation::ExtractShiftLeftRegister: + Compile_ExtractShiftLeftRegister(opcode); + break; + case Macro::Operation::Read: + Compile_Read(opcode); + break; + case Macro::Operation::Branch: + Compile_Branch(opcode); + break; + default: + UNIMPLEMENTED_MSG("Unimplemented opcode {}", opcode.operation.Value()); + break; + } + + if (optimizer.has_delayed_pc) { + if (opcode.is_exit) { + mov(rax, end_of_code); + test(BRANCH_HOLDER, BRANCH_HOLDER); + cmove(BRANCH_HOLDER, rax); + // Jump to next instruction to skip delay slot check + je(labels[pc + 1], T_NEAR); + } else { + // TODO(ogniK): Optimize delay slot branching + Xbyak::Label no_delay_slot{}; + test(BRANCH_HOLDER, BRANCH_HOLDER); + je(no_delay_slot, T_NEAR); + mov(rax, BRANCH_HOLDER); + xor_(BRANCH_HOLDER, BRANCH_HOLDER); + jmp(rax); + L(no_delay_slot); + } + L(delay_skip[pc]); + if (opcode.is_exit) { + return false; + } + } else { + test(BRANCH_HOLDER, BRANCH_HOLDER); + jne(end_of_code, T_NEAR); + if (opcode.is_exit) { + inc(BRANCH_HOLDER); + return false; + } + } + return true; +} + +Xbyak::Reg32 Tegra::MacroJITx64Impl::Compile_FetchParameter() { + mov(eax, dword[PARAMETERS]); + add(PARAMETERS, sizeof(u32)); + return eax; +} + +Xbyak::Reg32 MacroJITx64Impl::Compile_GetRegister(u32 index, Xbyak::Reg32 dst) { + if (index == 0) { + // Register 0 is always zero + xor_(dst, dst); + } else { + mov(dst, dword[STATE + offsetof(JITState, registers) + index * sizeof(u32)]); + } + + return dst; +} + +void MacroJITx64Impl::Compile_ProcessResult(Macro::ResultOperation operation, u32 reg) { + const auto SetRegister = [this](u32 reg, const Xbyak::Reg32& result) { + // Register 0 is supposed to always return 0. NOP is implemented as a store to the zero + // register. + if (reg == 0) { + return; + } + mov(dword[STATE + offsetof(JITState, registers) + reg * sizeof(u32)], result); + }; + const auto SetMethodAddress = [this](const Xbyak::Reg32& reg) { mov(METHOD_ADDRESS, reg); }; + + switch (operation) { + case Macro::ResultOperation::IgnoreAndFetch: + SetRegister(reg, Compile_FetchParameter()); + break; + case Macro::ResultOperation::Move: + SetRegister(reg, RESULT); + break; + case Macro::ResultOperation::MoveAndSetMethod: + SetRegister(reg, RESULT); + SetMethodAddress(RESULT); + break; + case Macro::ResultOperation::FetchAndSend: + // Fetch parameter and send result. + SetRegister(reg, Compile_FetchParameter()); + Compile_Send(RESULT); + break; + case Macro::ResultOperation::MoveAndSend: + // Move and send result. + SetRegister(reg, RESULT); + Compile_Send(RESULT); + break; + case Macro::ResultOperation::FetchAndSetMethod: + // Fetch parameter and use result as Method Address. + SetRegister(reg, Compile_FetchParameter()); + SetMethodAddress(RESULT); + break; + case Macro::ResultOperation::MoveAndSetMethodFetchAndSend: + // Move result and use as Method Address, then fetch and send parameter. + SetRegister(reg, RESULT); + SetMethodAddress(RESULT); + Compile_Send(Compile_FetchParameter()); + break; + case Macro::ResultOperation::MoveAndSetMethodSend: + // Move result and use as Method Address, then send bits 12:17 of result. + SetRegister(reg, RESULT); + SetMethodAddress(RESULT); + shr(RESULT, 12); + and_(RESULT, 0b111111); + Compile_Send(RESULT); + break; + default: + UNIMPLEMENTED_MSG("Unimplemented macro operation {}", static_cast<std::size_t>(operation)); + } +} + +Macro::Opcode MacroJITx64Impl::GetOpCode() const { + ASSERT(pc < code.size()); + return {code[pc]}; +} + +std::bitset<32> MacroJITx64Impl::PersistentCallerSavedRegs() const { + return PERSISTENT_REGISTERS & Common::X64::ABI_ALL_CALLER_SAVED; +} + +} // namespace Tegra diff --git a/src/video_core/macro/macro_jit_x64.h b/src/video_core/macro/macro_jit_x64.h new file mode 100644 index 000000000..a180e7428 --- /dev/null +++ b/src/video_core/macro/macro_jit_x64.h @@ -0,0 +1,98 @@ +// Copyright 2020 yuzu Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include <array> +#include <bitset> +#include <xbyak.h> +#include "common/bit_field.h" +#include "common/common_types.h" +#include "common/x64/xbyak_abi.h" +#include "video_core/macro/macro.h" + +namespace Tegra { + +namespace Engines { +class Maxwell3D; +} + +/// MAX_CODE_SIZE is arbitrarily chosen based on current booting games +constexpr size_t MAX_CODE_SIZE = 0x10000; + +class MacroJITx64 final : public MacroEngine { +public: + explicit MacroJITx64(Engines::Maxwell3D& maxwell3d); + +protected: + std::unique_ptr<CachedMacro> Compile(const std::vector<u32>& code) override; + +private: + Engines::Maxwell3D& maxwell3d; +}; + +class MacroJITx64Impl : public Xbyak::CodeGenerator, public CachedMacro { +public: + MacroJITx64Impl(Engines::Maxwell3D& maxwell3d, const std::vector<u32>& code); + ~MacroJITx64Impl(); + + void Execute(const std::vector<u32>& parameters, u32 method) override; + + void Compile_ALU(Macro::Opcode opcode); + void Compile_AddImmediate(Macro::Opcode opcode); + void Compile_ExtractInsert(Macro::Opcode opcode); + void Compile_ExtractShiftLeftImmediate(Macro::Opcode opcode); + void Compile_ExtractShiftLeftRegister(Macro::Opcode opcode); + void Compile_Read(Macro::Opcode opcode); + void Compile_Branch(Macro::Opcode opcode); + +private: + void Optimizer_ScanFlags(); + + void Compile(); + bool Compile_NextInstruction(); + + Xbyak::Reg32 Compile_FetchParameter(); + Xbyak::Reg32 Compile_GetRegister(u32 index, Xbyak::Reg32 dst); + + void Compile_ProcessResult(Macro::ResultOperation operation, u32 reg); + void Compile_Send(Xbyak::Reg32 value); + + Macro::Opcode GetOpCode() const; + std::bitset<32> PersistentCallerSavedRegs() const; + + struct JITState { + Engines::Maxwell3D* maxwell3d{}; + std::array<u32, Macro::NUM_MACRO_REGISTERS> registers{}; + u32 carry_flag{}; + }; + static_assert(offsetof(JITState, maxwell3d) == 0, "Maxwell3D is not at 0x0"); + using ProgramType = void (*)(JITState*, const u32*); + + struct OptimizerState { + bool can_skip_carry{}; + bool has_delayed_pc{}; + bool zero_reg_skip{}; + bool skip_dummy_addimmediate{}; + bool optimize_for_method_move{}; + bool enable_asserts{}; + }; + OptimizerState optimizer{}; + + std::optional<Macro::Opcode> next_opcode{}; + ProgramType program{nullptr}; + + std::array<Xbyak::Label, MAX_CODE_SIZE> labels; + std::array<Xbyak::Label, MAX_CODE_SIZE> delay_skip; + Xbyak::Label end_of_code{}; + + bool is_delay_slot{}; + u32 pc{}; + std::optional<u32> delayed_pc; + + const std::vector<u32>& code; + Engines::Maxwell3D& maxwell3d; +}; + +} // namespace Tegra diff --git a/src/video_core/memory_manager.cpp b/src/video_core/memory_manager.cpp index dbee9f634..ff5505d12 100644 --- a/src/video_core/memory_manager.cpp +++ b/src/video_core/memory_manager.cpp @@ -210,10 +210,11 @@ bool MemoryManager::IsBlockContinuous(const GPUVAddr start, const std::size_t si return range == inner_size; } -void MemoryManager::ReadBlock(GPUVAddr src_addr, void* dest_buffer, const std::size_t size) const { +void MemoryManager::ReadBlock(GPUVAddr gpu_src_addr, void* dest_buffer, + const std::size_t size) const { std::size_t remaining_size{size}; - std::size_t page_index{src_addr >> page_bits}; - std::size_t page_offset{src_addr & page_mask}; + std::size_t page_index{gpu_src_addr >> page_bits}; + std::size_t page_offset{gpu_src_addr & page_mask}; auto& memory = system.Memory(); @@ -234,11 +235,11 @@ void MemoryManager::ReadBlock(GPUVAddr src_addr, void* dest_buffer, const std::s } } -void MemoryManager::ReadBlockUnsafe(GPUVAddr src_addr, void* dest_buffer, +void MemoryManager::ReadBlockUnsafe(GPUVAddr gpu_src_addr, void* dest_buffer, const std::size_t size) const { std::size_t remaining_size{size}; - std::size_t page_index{src_addr >> page_bits}; - std::size_t page_offset{src_addr & page_mask}; + std::size_t page_index{gpu_src_addr >> page_bits}; + std::size_t page_offset{gpu_src_addr & page_mask}; auto& memory = system.Memory(); @@ -259,10 +260,11 @@ void MemoryManager::ReadBlockUnsafe(GPUVAddr src_addr, void* dest_buffer, } } -void MemoryManager::WriteBlock(GPUVAddr dest_addr, const void* src_buffer, const std::size_t size) { +void MemoryManager::WriteBlock(GPUVAddr gpu_dest_addr, const void* src_buffer, + const std::size_t size) { std::size_t remaining_size{size}; - std::size_t page_index{dest_addr >> page_bits}; - std::size_t page_offset{dest_addr & page_mask}; + std::size_t page_index{gpu_dest_addr >> page_bits}; + std::size_t page_offset{gpu_dest_addr & page_mask}; auto& memory = system.Memory(); @@ -283,11 +285,11 @@ void MemoryManager::WriteBlock(GPUVAddr dest_addr, const void* src_buffer, const } } -void MemoryManager::WriteBlockUnsafe(GPUVAddr dest_addr, const void* src_buffer, +void MemoryManager::WriteBlockUnsafe(GPUVAddr gpu_dest_addr, const void* src_buffer, const std::size_t size) { std::size_t remaining_size{size}; - std::size_t page_index{dest_addr >> page_bits}; - std::size_t page_offset{dest_addr & page_mask}; + std::size_t page_index{gpu_dest_addr >> page_bits}; + std::size_t page_offset{gpu_dest_addr & page_mask}; auto& memory = system.Memory(); @@ -306,16 +308,18 @@ void MemoryManager::WriteBlockUnsafe(GPUVAddr dest_addr, const void* src_buffer, } } -void MemoryManager::CopyBlock(GPUVAddr dest_addr, GPUVAddr src_addr, const std::size_t size) { +void MemoryManager::CopyBlock(GPUVAddr gpu_dest_addr, GPUVAddr gpu_src_addr, + const std::size_t size) { std::vector<u8> tmp_buffer(size); - ReadBlock(src_addr, tmp_buffer.data(), size); - WriteBlock(dest_addr, tmp_buffer.data(), size); + ReadBlock(gpu_src_addr, tmp_buffer.data(), size); + WriteBlock(gpu_dest_addr, tmp_buffer.data(), size); } -void MemoryManager::CopyBlockUnsafe(GPUVAddr dest_addr, GPUVAddr src_addr, const std::size_t size) { +void MemoryManager::CopyBlockUnsafe(GPUVAddr gpu_dest_addr, GPUVAddr gpu_src_addr, + const std::size_t size) { std::vector<u8> tmp_buffer(size); - ReadBlockUnsafe(src_addr, tmp_buffer.data(), size); - WriteBlockUnsafe(dest_addr, tmp_buffer.data(), size); + ReadBlockUnsafe(gpu_src_addr, tmp_buffer.data(), size); + WriteBlockUnsafe(gpu_dest_addr, tmp_buffer.data(), size); } bool MemoryManager::IsGranularRange(GPUVAddr gpu_addr, std::size_t size) { diff --git a/src/video_core/memory_manager.h b/src/video_core/memory_manager.h index 0ddd52d5a..87658e87a 100644 --- a/src/video_core/memory_manager.h +++ b/src/video_core/memory_manager.h @@ -79,9 +79,9 @@ public: * in the Host Memory counterpart. Note: This functions cause Host GPU Memory * Flushes and Invalidations, respectively to each operation. */ - void ReadBlock(GPUVAddr src_addr, void* dest_buffer, std::size_t size) const; - void WriteBlock(GPUVAddr dest_addr, const void* src_buffer, std::size_t size); - void CopyBlock(GPUVAddr dest_addr, GPUVAddr src_addr, std::size_t size); + void ReadBlock(GPUVAddr gpu_src_addr, void* dest_buffer, std::size_t size) const; + void WriteBlock(GPUVAddr gpu_dest_addr, const void* src_buffer, std::size_t size); + void CopyBlock(GPUVAddr gpu_dest_addr, GPUVAddr gpu_src_addr, std::size_t size); /** * ReadBlockUnsafe and WriteBlockUnsafe are special versions of ReadBlock and @@ -93,9 +93,9 @@ public: * WriteBlockUnsafe instead of WriteBlock since it shouldn't invalidate the texture * being flushed. */ - void ReadBlockUnsafe(GPUVAddr src_addr, void* dest_buffer, std::size_t size) const; - void WriteBlockUnsafe(GPUVAddr dest_addr, const void* src_buffer, std::size_t size); - void CopyBlockUnsafe(GPUVAddr dest_addr, GPUVAddr src_addr, std::size_t size); + void ReadBlockUnsafe(GPUVAddr gpu_src_addr, void* dest_buffer, std::size_t size) const; + void WriteBlockUnsafe(GPUVAddr gpu_dest_addr, const void* src_buffer, std::size_t size); + void CopyBlockUnsafe(GPUVAddr gpu_dest_addr, GPUVAddr gpu_src_addr, std::size_t size); /** * IsGranularRange checks if a gpu region can be simply read with a pointer diff --git a/src/video_core/morton.cpp b/src/video_core/morton.cpp index 6d522c318..836b25c1d 100644 --- a/src/video_core/morton.cpp +++ b/src/video_core/morton.cpp @@ -83,6 +83,7 @@ static constexpr ConversionArray morton_to_linear_fns = { MortonCopy<true, PixelFormat::RGBA8_SRGB>, MortonCopy<true, PixelFormat::RG8U>, MortonCopy<true, PixelFormat::RG8S>, + MortonCopy<true, PixelFormat::RG8UI>, MortonCopy<true, PixelFormat::RG32UI>, MortonCopy<true, PixelFormat::RGBX16F>, MortonCopy<true, PixelFormat::R32UI>, @@ -166,6 +167,7 @@ static constexpr ConversionArray linear_to_morton_fns = { MortonCopy<false, PixelFormat::RGBA8_SRGB>, MortonCopy<false, PixelFormat::RG8U>, MortonCopy<false, PixelFormat::RG8S>, + MortonCopy<false, PixelFormat::RG8UI>, MortonCopy<false, PixelFormat::RG32UI>, MortonCopy<false, PixelFormat::RGBX16F>, MortonCopy<false, PixelFormat::R32UI>, diff --git a/src/video_core/query_cache.h b/src/video_core/query_cache.h index 2f75f8801..e12dab899 100644 --- a/src/video_core/query_cache.h +++ b/src/video_core/query_cache.h @@ -220,8 +220,8 @@ private: return cache_begin < addr_end && addr_begin < cache_end; }; - const u64 page_end = addr_end >> PAGE_SHIFT; - for (u64 page = addr_begin >> PAGE_SHIFT; page <= page_end; ++page) { + const u64 page_end = addr_end >> PAGE_BITS; + for (u64 page = addr_begin >> PAGE_BITS; page <= page_end; ++page) { const auto& it = cached_queries.find(page); if (it == std::end(cached_queries)) { continue; @@ -242,14 +242,14 @@ private: /// Registers the passed parameters as cached and returns a pointer to the stored cached query. CachedQuery* Register(VideoCore::QueryType type, VAddr cpu_addr, u8* host_ptr, bool timestamp) { rasterizer.UpdatePagesCachedCount(cpu_addr, CachedQuery::SizeInBytes(timestamp), 1); - const u64 page = static_cast<u64>(cpu_addr) >> PAGE_SHIFT; + const u64 page = static_cast<u64>(cpu_addr) >> PAGE_BITS; return &cached_queries[page].emplace_back(static_cast<QueryCache&>(*this), type, cpu_addr, host_ptr); } /// Tries to a get a cached query. Returns nullptr on failure. CachedQuery* TryGet(VAddr addr) { - const u64 page = static_cast<u64>(addr) >> PAGE_SHIFT; + const u64 page = static_cast<u64>(addr) >> PAGE_BITS; const auto it = cached_queries.find(page); if (it == std::end(cached_queries)) { return nullptr; @@ -268,7 +268,7 @@ private: } static constexpr std::uintptr_t PAGE_SIZE = 4096; - static constexpr unsigned PAGE_SHIFT = 12; + static constexpr unsigned PAGE_BITS = 12; Core::System& system; VideoCore::RasterizerInterface& rasterizer; diff --git a/src/video_core/rasterizer_cache.cpp b/src/video_core/rasterizer_cache.cpp deleted file mode 100644 index 093b2cdf4..000000000 --- a/src/video_core/rasterizer_cache.cpp +++ /dev/null @@ -1,7 +0,0 @@ -// Copyright 2018 yuzu Emulator Project -// Licensed under GPLv2 or any later version -// Refer to the license.txt file included. - -#include "video_core/rasterizer_cache.h" - -RasterizerCacheObject::~RasterizerCacheObject() = default; diff --git a/src/video_core/rasterizer_cache.h b/src/video_core/rasterizer_cache.h deleted file mode 100644 index 22987751e..000000000 --- a/src/video_core/rasterizer_cache.h +++ /dev/null @@ -1,197 +0,0 @@ -// Copyright 2018 yuzu Emulator Project -// Licensed under GPLv2 or any later version -// Refer to the license.txt file included. - -#pragma once - -#include <mutex> -#include <set> -#include <unordered_map> - -#include <boost/icl/interval_map.hpp> -#include <boost/range/iterator_range_core.hpp> - -#include "common/common_types.h" -#include "core/settings.h" -#include "video_core/gpu.h" -#include "video_core/rasterizer_interface.h" - -class RasterizerCacheObject { -public: - explicit RasterizerCacheObject(const VAddr cpu_addr) : cpu_addr{cpu_addr} {} - - virtual ~RasterizerCacheObject(); - - VAddr GetCpuAddr() const { - return cpu_addr; - } - - /// Gets the size of the shader in guest memory, required for cache management - virtual std::size_t GetSizeInBytes() const = 0; - - /// Sets whether the cached object should be considered registered - void SetIsRegistered(bool registered) { - is_registered = registered; - } - - /// Returns true if the cached object is registered - bool IsRegistered() const { - return is_registered; - } - - /// Returns true if the cached object is dirty - bool IsDirty() const { - return is_dirty; - } - - /// Returns ticks from when this cached object was last modified - u64 GetLastModifiedTicks() const { - return last_modified_ticks; - } - - /// Marks an object as recently modified, used to specify whether it is clean or dirty - template <class T> - void MarkAsModified(bool dirty, T& cache) { - is_dirty = dirty; - last_modified_ticks = cache.GetModifiedTicks(); - } - -private: - bool is_registered{}; ///< Whether the object is currently registered with the cache - bool is_dirty{}; ///< Whether the object is dirty (out of sync with guest memory) - u64 last_modified_ticks{}; ///< When the object was last modified, used for in-order flushing - VAddr cpu_addr{}; ///< Cpu address memory, unique from emulated virtual address space -}; - -template <class T> -class RasterizerCache : NonCopyable { - friend class RasterizerCacheObject; - -public: - explicit RasterizerCache(VideoCore::RasterizerInterface& rasterizer) : rasterizer{rasterizer} {} - - /// Write any cached resources overlapping the specified region back to memory - void FlushRegion(VAddr addr, std::size_t size) { - std::lock_guard lock{mutex}; - - const auto& objects{GetSortedObjectsFromRegion(addr, size)}; - for (auto& object : objects) { - FlushObject(object); - } - } - - /// Mark the specified region as being invalidated - void InvalidateRegion(VAddr addr, u64 size) { - std::lock_guard lock{mutex}; - - const auto& objects{GetSortedObjectsFromRegion(addr, size)}; - for (auto& object : objects) { - if (!object->IsRegistered()) { - // Skip duplicates - continue; - } - Unregister(object); - } - } - - /// Invalidates everything in the cache - void InvalidateAll() { - std::lock_guard lock{mutex}; - - while (interval_cache.begin() != interval_cache.end()) { - Unregister(*interval_cache.begin()->second.begin()); - } - } - -protected: - /// Tries to get an object from the cache with the specified cache address - T TryGet(VAddr addr) const { - const auto iter = map_cache.find(addr); - if (iter != map_cache.end()) - return iter->second; - return nullptr; - } - - /// Register an object into the cache - virtual void Register(const T& object) { - std::lock_guard lock{mutex}; - - object->SetIsRegistered(true); - interval_cache.add({GetInterval(object), ObjectSet{object}}); - map_cache.insert({object->GetCpuAddr(), object}); - rasterizer.UpdatePagesCachedCount(object->GetCpuAddr(), object->GetSizeInBytes(), 1); - } - - /// Unregisters an object from the cache - virtual void Unregister(const T& object) { - std::lock_guard lock{mutex}; - - object->SetIsRegistered(false); - rasterizer.UpdatePagesCachedCount(object->GetCpuAddr(), object->GetSizeInBytes(), -1); - const VAddr addr = object->GetCpuAddr(); - interval_cache.subtract({GetInterval(object), ObjectSet{object}}); - map_cache.erase(addr); - } - - /// Returns a ticks counter used for tracking when cached objects were last modified - u64 GetModifiedTicks() { - std::lock_guard lock{mutex}; - - return ++modified_ticks; - } - - virtual void FlushObjectInner(const T& object) = 0; - - /// Flushes the specified object, updating appropriate cache state as needed - void FlushObject(const T& object) { - std::lock_guard lock{mutex}; - - if (!object->IsDirty()) { - return; - } - FlushObjectInner(object); - object->MarkAsModified(false, *this); - } - - std::recursive_mutex mutex; - -private: - /// Returns a list of cached objects from the specified memory region, ordered by access time - std::vector<T> GetSortedObjectsFromRegion(VAddr addr, u64 size) { - if (size == 0) { - return {}; - } - - std::vector<T> objects; - const ObjectInterval interval{addr, addr + size}; - for (auto& pair : boost::make_iterator_range(interval_cache.equal_range(interval))) { - for (auto& cached_object : pair.second) { - if (!cached_object) { - continue; - } - objects.push_back(cached_object); - } - } - - std::sort(objects.begin(), objects.end(), [](const T& a, const T& b) -> bool { - return a->GetLastModifiedTicks() < b->GetLastModifiedTicks(); - }); - - return objects; - } - - using ObjectSet = std::set<T>; - using ObjectCache = std::unordered_map<VAddr, T>; - using IntervalCache = boost::icl::interval_map<VAddr, ObjectSet>; - using ObjectInterval = typename IntervalCache::interval_type; - - static auto GetInterval(const T& object) { - return ObjectInterval::right_open(object->GetCpuAddr(), - object->GetCpuAddr() + object->GetSizeInBytes()); - } - - ObjectCache map_cache; - IntervalCache interval_cache; ///< Cache of objects - u64 modified_ticks{}; ///< Counter of cache state ticks, used for in-order flushing - VideoCore::RasterizerInterface& rasterizer; -}; diff --git a/src/video_core/renderer_opengl/gl_arb_decompiler.cpp b/src/video_core/renderer_opengl/gl_arb_decompiler.cpp new file mode 100644 index 000000000..eb5158407 --- /dev/null +++ b/src/video_core/renderer_opengl/gl_arb_decompiler.cpp @@ -0,0 +1,2073 @@ +// Copyright 2020 yuzu Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include <algorithm> +#include <array> +#include <cstddef> +#include <string> +#include <string_view> +#include <utility> +#include <variant> + +#include <fmt/format.h> + +#include "common/alignment.h" +#include "common/assert.h" +#include "common/common_types.h" +#include "video_core/renderer_opengl/gl_arb_decompiler.h" +#include "video_core/renderer_opengl/gl_device.h" +#include "video_core/shader/registry.h" +#include "video_core/shader/shader_ir.h" + +// Predicates in the decompiled code follow the convention that -1 means true and 0 means false. +// GLASM lacks booleans, so they have to be implemented as integers. +// Using -1 for true is useful because both CMP.S and NOT.U can negate it, and CMP.S can be used to +// select between two values, because -1 will be evaluated as true and 0 as false. + +namespace OpenGL { + +namespace { + +using Tegra::Engines::ShaderType; +using Tegra::Shader::Attribute; +using Tegra::Shader::PixelImap; +using Tegra::Shader::Register; +using namespace VideoCommon::Shader; +using Operation = const OperationNode&; + +constexpr std::array INTERNAL_FLAG_NAMES = {"ZERO", "SIGN", "CARRY", "OVERFLOW"}; + +char Swizzle(std::size_t component) { + ASSERT(component < 4); + return component["xyzw"]; +} + +constexpr bool IsGenericAttribute(Attribute::Index index) { + return index >= Attribute::Index::Attribute_0 && index <= Attribute::Index::Attribute_31; +} + +u32 GetGenericAttributeIndex(Attribute::Index index) { + ASSERT(IsGenericAttribute(index)); + return static_cast<u32>(index) - static_cast<u32>(Attribute::Index::Attribute_0); +} + +std::string_view Modifiers(Operation operation) { + const auto meta = std::get_if<MetaArithmetic>(&operation.GetMeta()); + if (meta && meta->precise) { + return ".PREC"; + } + return ""; +} + +std::string_view GetInputFlags(PixelImap attribute) { + switch (attribute) { + case PixelImap::Perspective: + return ""; + case PixelImap::Constant: + return "FLAT "; + case PixelImap::ScreenLinear: + return "NOPERSPECTIVE "; + case PixelImap::Unused: + break; + } + UNIMPLEMENTED_MSG("Unknown attribute usage index={}", static_cast<int>(attribute)); + return {}; +} + +std::string_view ImageType(Tegra::Shader::ImageType image_type) { + switch (image_type) { + case Tegra::Shader::ImageType::Texture1D: + return "1D"; + case Tegra::Shader::ImageType::TextureBuffer: + return "BUFFER"; + case Tegra::Shader::ImageType::Texture1DArray: + return "ARRAY1D"; + case Tegra::Shader::ImageType::Texture2D: + return "2D"; + case Tegra::Shader::ImageType::Texture2DArray: + return "ARRAY2D"; + case Tegra::Shader::ImageType::Texture3D: + return "3D"; + } + UNREACHABLE(); + return {}; +} + +std::string_view StackName(MetaStackClass stack) { + switch (stack) { + case MetaStackClass::Ssy: + return "SSY"; + case MetaStackClass::Pbk: + return "PBK"; + } + UNREACHABLE(); + return ""; +}; + +std::string_view PrimitiveDescription(Tegra::Engines::Maxwell3D::Regs::PrimitiveTopology topology) { + switch (topology) { + case Tegra::Engines::Maxwell3D::Regs::PrimitiveTopology::Points: + return "POINTS"; + case Tegra::Engines::Maxwell3D::Regs::PrimitiveTopology::Lines: + case Tegra::Engines::Maxwell3D::Regs::PrimitiveTopology::LineStrip: + return "LINES"; + case Tegra::Engines::Maxwell3D::Regs::PrimitiveTopology::LinesAdjacency: + case Tegra::Engines::Maxwell3D::Regs::PrimitiveTopology::LineStripAdjacency: + return "LINES_ADJACENCY"; + case Tegra::Engines::Maxwell3D::Regs::PrimitiveTopology::Triangles: + case Tegra::Engines::Maxwell3D::Regs::PrimitiveTopology::TriangleStrip: + case Tegra::Engines::Maxwell3D::Regs::PrimitiveTopology::TriangleFan: + return "TRIANGLES"; + case Tegra::Engines::Maxwell3D::Regs::PrimitiveTopology::TrianglesAdjacency: + case Tegra::Engines::Maxwell3D::Regs::PrimitiveTopology::TriangleStripAdjacency: + return "TRIANGLES_ADJACENCY"; + default: + UNIMPLEMENTED_MSG("topology={}", static_cast<int>(topology)); + return "POINTS"; + } +} + +std::string_view TopologyName(Tegra::Shader::OutputTopology topology) { + switch (topology) { + case Tegra::Shader::OutputTopology::PointList: + return "POINTS"; + case Tegra::Shader::OutputTopology::LineStrip: + return "LINE_STRIP"; + case Tegra::Shader::OutputTopology::TriangleStrip: + return "TRIANGLE_STRIP"; + default: + UNIMPLEMENTED_MSG("Unknown output topology: {}", static_cast<u32>(topology)); + return "points"; + } +} + +std::string_view StageInputName(ShaderType stage) { + switch (stage) { + case ShaderType::Vertex: + case ShaderType::Geometry: + return "vertex"; + case ShaderType::Fragment: + return "fragment"; + case ShaderType::Compute: + return "invocation"; + default: + UNREACHABLE(); + return ""; + } +} + +std::string TextureType(const MetaTexture& meta) { + if (meta.sampler.is_buffer) { + return "BUFFER"; + } + std::string type; + if (meta.sampler.is_shadow) { + type += "SHADOW"; + } + if (meta.sampler.is_array) { + type += "ARRAY"; + } + type += [&meta] { + switch (meta.sampler.type) { + case Tegra::Shader::TextureType::Texture1D: + return "1D"; + case Tegra::Shader::TextureType::Texture2D: + return "2D"; + case Tegra::Shader::TextureType::Texture3D: + return "3D"; + case Tegra::Shader::TextureType::TextureCube: + return "CUBE"; + } + UNREACHABLE(); + return "2D"; + }(); + return type; +} + +std::string GlobalMemoryName(const GlobalMemoryBase& base) { + return fmt::format("gmem{}_{}", base.cbuf_index, base.cbuf_offset); +} + +class ARBDecompiler final { +public: + explicit ARBDecompiler(const Device& device, const ShaderIR& ir, const Registry& registry, + ShaderType stage, std::string_view identifier); + + std::string Code() const { + return shader_source; + } + +private: + void DeclareHeader(); + void DeclareVertex(); + void DeclareGeometry(); + void DeclareFragment(); + void DeclareCompute(); + void DeclareInputAttributes(); + void DeclareOutputAttributes(); + void DeclareLocalMemory(); + void DeclareGlobalMemory(); + void DeclareConstantBuffers(); + void DeclareRegisters(); + void DeclareTemporaries(); + void DeclarePredicates(); + void DeclareInternalFlags(); + + void InitializeVariables(); + + void DecompileAST(); + void DecompileBranchMode(); + + void VisitAST(const ASTNode& node); + std::string VisitExpression(const Expr& node); + + void VisitBlock(const NodeBlock& bb); + + std::string Visit(const Node& node); + + std::pair<std::string, std::size_t> BuildCoords(Operation); + std::string BuildAoffi(Operation); + void Exit(); + + std::string Assign(Operation); + std::string Select(Operation); + std::string FClamp(Operation); + std::string FCastHalf0(Operation); + std::string FCastHalf1(Operation); + std::string FSqrt(Operation); + std::string FSwizzleAdd(Operation); + std::string HAdd2(Operation); + std::string HMul2(Operation); + std::string HFma2(Operation); + std::string HAbsolute(Operation); + std::string HNegate(Operation); + std::string HClamp(Operation); + std::string HCastFloat(Operation); + std::string HUnpack(Operation); + std::string HMergeF32(Operation); + std::string HMergeH0(Operation); + std::string HMergeH1(Operation); + std::string HPack2(Operation); + std::string LogicalAssign(Operation); + std::string LogicalPick2(Operation); + std::string LogicalAnd2(Operation); + std::string FloatOrdered(Operation); + std::string FloatUnordered(Operation); + std::string LogicalAddCarry(Operation); + std::string Texture(Operation); + std::string TextureGather(Operation); + std::string TextureQueryDimensions(Operation); + std::string TextureQueryLod(Operation); + std::string TexelFetch(Operation); + std::string TextureGradient(Operation); + std::string ImageLoad(Operation); + std::string ImageStore(Operation); + std::string Branch(Operation); + std::string BranchIndirect(Operation); + std::string PushFlowStack(Operation); + std::string PopFlowStack(Operation); + std::string Exit(Operation); + std::string Discard(Operation); + std::string EmitVertex(Operation); + std::string EndPrimitive(Operation); + std::string InvocationId(Operation); + std::string YNegate(Operation); + std::string ThreadId(Operation); + std::string ShuffleIndexed(Operation); + std::string Barrier(Operation); + std::string MemoryBarrierGroup(Operation); + std::string MemoryBarrierGlobal(Operation); + + template <const std::string_view& op> + std::string Unary(Operation operation) { + std::string temporary = AllocTemporary(); + AddLine("{}{} {}, {};", op, Modifiers(operation), temporary, Visit(operation[0])); + return temporary; + } + + template <const std::string_view& op> + std::string Binary(Operation operation) { + std::string temporary = AllocTemporary(); + AddLine("{}{} {}, {}, {};", op, Modifiers(operation), temporary, Visit(operation[0]), + Visit(operation[1])); + return temporary; + } + + template <const std::string_view& op> + std::string Trinary(Operation operation) { + std::string temporary = AllocTemporary(); + AddLine("{}{} {}, {}, {}, {};", op, Modifiers(operation), temporary, Visit(operation[0]), + Visit(operation[1]), Visit(operation[2])); + return temporary; + } + + template <const std::string_view& op, bool unordered> + std::string FloatComparison(Operation operation) { + std::string temporary = AllocTemporary(); + AddLine("TRUNC.U.CC RC.x, {};", Binary<op>(operation)); + AddLine("MOV.S {}, 0;", temporary); + AddLine("MOV.S {} (NE.x), -1;", temporary); + + const std::string op_a = Visit(operation[0]); + const std::string op_b = Visit(operation[1]); + if constexpr (unordered) { + AddLine("SNE.F RC.x, {}, {};", op_a, op_a); + AddLine("TRUNC.U.CC RC.x, RC.x;"); + AddLine("MOV.S {} (NE.x), -1;", temporary); + AddLine("SNE.F RC.x, {}, {};", op_b, op_b); + AddLine("TRUNC.U.CC RC.x, RC.x;"); + AddLine("MOV.S {} (NE.x), -1;", temporary); + } else if (op == SNE_F) { + AddLine("SNE.F RC.x, {}, {};", op_a, op_a); + AddLine("TRUNC.U.CC RC.x, RC.x;"); + AddLine("MOV.S {} (NE.x), 0;", temporary); + AddLine("SNE.F RC.x, {}, {};", op_b, op_b); + AddLine("TRUNC.U.CC RC.x, RC.x;"); + AddLine("MOV.S {} (NE.x), 0;", temporary); + } + return temporary; + } + + template <const std::string_view& op, bool is_nan> + std::string HalfComparison(Operation operation) { + std::string tmp1 = AllocVectorTemporary(); + const std::string tmp2 = AllocVectorTemporary(); + const std::string op_a = Visit(operation[0]); + const std::string op_b = Visit(operation[1]); + AddLine("UP2H.F {}, {};", tmp1, op_a); + AddLine("UP2H.F {}, {};", tmp2, op_b); + AddLine("{} {}, {}, {};", op, tmp1, tmp1, tmp2); + AddLine("TRUNC.U.CC RC.xy, {};", tmp1); + AddLine("MOV.S {}.xy, {{0, 0, 0, 0}};", tmp1); + AddLine("MOV.S {}.x (NE.x), -1;", tmp1); + AddLine("MOV.S {}.y (NE.y), -1;", tmp1); + if constexpr (is_nan) { + AddLine("MOVC.F RC.x, {};", op_a); + AddLine("MOV.S {}.x (NAN.x), -1;", tmp1); + AddLine("MOVC.F RC.x, {};", op_b); + AddLine("MOV.S {}.y (NAN.x), -1;", tmp1); + } + return tmp1; + } + + template <const std::string_view& op, const std::string_view& type> + std::string AtomicImage(Operation operation) { + const auto& meta = std::get<MetaImage>(operation.GetMeta()); + const u32 image_id = device.GetBaseBindings(stage).image + meta.image.index; + const std::size_t num_coords = operation.GetOperandsCount(); + const std::size_t num_values = meta.values.size(); + + const std::string coord = AllocVectorTemporary(); + const std::string value = AllocVectorTemporary(); + for (std::size_t i = 0; i < num_coords; ++i) { + AddLine("MOV.S {}.{}, {};", coord, Swizzle(i), Visit(operation[i])); + } + for (std::size_t i = 0; i < num_values; ++i) { + AddLine("MOV.F {}.{}, {};", value, Swizzle(i), Visit(meta.values[i])); + } + + AddLine("ATOMIM.{}.{} {}.x, {}, {}, image[{}], {};", op, type, coord, value, coord, + image_id, ImageType(meta.image.type)); + return fmt::format("{}.x", coord); + } + + template <const std::string_view& op, const std::string_view& type> + std::string Atomic(Operation operation) { + std::string temporary = AllocTemporary(); + std::string address; + std::string_view opname; + if (const auto gmem = std::get_if<GmemNode>(&*operation[0])) { + AddLine("SUB.U {}, {}, {};", temporary, Visit(gmem->GetRealAddress()), + Visit(gmem->GetBaseAddress())); + address = fmt::format("{}[{}]", GlobalMemoryName(gmem->GetDescriptor()), temporary); + opname = "ATOMB"; + } else if (const auto smem = std::get_if<SmemNode>(&*operation[0])) { + address = fmt::format("shared_mem[{}]", Visit(smem->GetAddress())); + opname = "ATOMS"; + } else { + UNREACHABLE(); + return "{0, 0, 0, 0}"; + } + AddLine("{}.{}.{} {}, {}, {};", opname, op, type, temporary, Visit(operation[1]), address); + return temporary; + } + + template <char type> + std::string Negate(Operation operation) { + std::string temporary = AllocTemporary(); + if constexpr (type == 'F') { + AddLine("MOV.F32 {}, -{};", temporary, Visit(operation[0])); + } else { + AddLine("MOV.{} {}, -{};", type, temporary, Visit(operation[0])); + } + return temporary; + } + + template <char type> + std::string Absolute(Operation operation) { + std::string temporary = AllocTemporary(); + AddLine("MOV.{} {}, |{}|;", type, temporary, Visit(operation[0])); + return temporary; + } + + template <char type> + std::string BitfieldInsert(Operation operation) { + const std::string temporary = AllocVectorTemporary(); + AddLine("MOV.{} {}.x, {};", type, temporary, Visit(operation[3])); + AddLine("MOV.{} {}.y, {};", type, temporary, Visit(operation[2])); + AddLine("BFI.{} {}.x, {}, {}, {};", type, temporary, temporary, Visit(operation[1]), + Visit(operation[0])); + return fmt::format("{}.x", temporary); + } + + template <char type> + std::string BitfieldExtract(Operation operation) { + const std::string temporary = AllocVectorTemporary(); + AddLine("MOV.{} {}.x, {};", type, temporary, Visit(operation[2])); + AddLine("MOV.{} {}.y, {};", type, temporary, Visit(operation[1])); + AddLine("BFE.{} {}.x, {}, {};", type, temporary, temporary, Visit(operation[0])); + return fmt::format("{}.x", temporary); + } + + template <char swizzle> + std::string LocalInvocationId(Operation) { + return fmt::format("invocation.localid.{}", swizzle); + } + + template <char swizzle> + std::string WorkGroupId(Operation) { + return fmt::format("invocation.groupid.{}", swizzle); + } + + template <char c1, char c2> + std::string ThreadMask(Operation) { + return fmt::format("{}.thread{}{}mask", StageInputName(stage), c1, c2); + } + + template <typename... Args> + void AddExpression(std::string_view text, Args&&... args) { + shader_source += fmt::format(text, std::forward<Args>(args)...); + } + + template <typename... Args> + void AddLine(std::string_view text, Args&&... args) { + AddExpression(text, std::forward<Args>(args)...); + shader_source += '\n'; + } + + std::string AllocTemporary() { + max_temporaries = std::max(max_temporaries, num_temporaries + 1); + return fmt::format("T{}.x", num_temporaries++); + } + + std::string AllocVectorTemporary() { + max_temporaries = std::max(max_temporaries, num_temporaries + 1); + return fmt::format("T{}", num_temporaries++); + } + + void ResetTemporaries() noexcept { + num_temporaries = 0; + } + + const Device& device; + const ShaderIR& ir; + const Registry& registry; + const ShaderType stage; + + std::size_t num_temporaries = 0; + std::size_t max_temporaries = 0; + + std::string shader_source; + + static constexpr std::string_view ADD_F32 = "ADD.F32"; + static constexpr std::string_view ADD_S = "ADD.S"; + static constexpr std::string_view ADD_U = "ADD.U"; + static constexpr std::string_view MUL_F32 = "MUL.F32"; + static constexpr std::string_view MUL_S = "MUL.S"; + static constexpr std::string_view MUL_U = "MUL.U"; + static constexpr std::string_view DIV_F32 = "DIV.F32"; + static constexpr std::string_view DIV_S = "DIV.S"; + static constexpr std::string_view DIV_U = "DIV.U"; + static constexpr std::string_view MAD_F32 = "MAD.F32"; + static constexpr std::string_view RSQ_F32 = "RSQ.F32"; + static constexpr std::string_view COS_F32 = "COS.F32"; + static constexpr std::string_view SIN_F32 = "SIN.F32"; + static constexpr std::string_view EX2_F32 = "EX2.F32"; + static constexpr std::string_view LG2_F32 = "LG2.F32"; + static constexpr std::string_view SLT_F = "SLT.F32"; + static constexpr std::string_view SLT_S = "SLT.S"; + static constexpr std::string_view SLT_U = "SLT.U"; + static constexpr std::string_view SEQ_F = "SEQ.F32"; + static constexpr std::string_view SEQ_S = "SEQ.S"; + static constexpr std::string_view SEQ_U = "SEQ.U"; + static constexpr std::string_view SLE_F = "SLE.F32"; + static constexpr std::string_view SLE_S = "SLE.S"; + static constexpr std::string_view SLE_U = "SLE.U"; + static constexpr std::string_view SGT_F = "SGT.F32"; + static constexpr std::string_view SGT_S = "SGT.S"; + static constexpr std::string_view SGT_U = "SGT.U"; + static constexpr std::string_view SNE_F = "SNE.F32"; + static constexpr std::string_view SNE_S = "SNE.S"; + static constexpr std::string_view SNE_U = "SNE.U"; + static constexpr std::string_view SGE_F = "SGE.F32"; + static constexpr std::string_view SGE_S = "SGE.S"; + static constexpr std::string_view SGE_U = "SGE.U"; + static constexpr std::string_view AND_S = "AND.S"; + static constexpr std::string_view AND_U = "AND.U"; + static constexpr std::string_view TRUNC_F = "TRUNC.F"; + static constexpr std::string_view TRUNC_S = "TRUNC.S"; + static constexpr std::string_view TRUNC_U = "TRUNC.U"; + static constexpr std::string_view SHL_S = "SHL.S"; + static constexpr std::string_view SHL_U = "SHL.U"; + static constexpr std::string_view SHR_S = "SHR.S"; + static constexpr std::string_view SHR_U = "SHR.U"; + static constexpr std::string_view OR_S = "OR.S"; + static constexpr std::string_view OR_U = "OR.U"; + static constexpr std::string_view XOR_S = "XOR.S"; + static constexpr std::string_view XOR_U = "XOR.U"; + static constexpr std::string_view NOT_S = "NOT.S"; + static constexpr std::string_view NOT_U = "NOT.U"; + static constexpr std::string_view BTC_S = "BTC.S"; + static constexpr std::string_view BTC_U = "BTC.U"; + static constexpr std::string_view BTFM_S = "BTFM.S"; + static constexpr std::string_view BTFM_U = "BTFM.U"; + static constexpr std::string_view ROUND_F = "ROUND.F"; + static constexpr std::string_view CEIL_F = "CEIL.F"; + static constexpr std::string_view FLR_F = "FLR.F"; + static constexpr std::string_view I2F_S = "I2F.S"; + static constexpr std::string_view I2F_U = "I2F.U"; + static constexpr std::string_view MIN_F = "MIN.F"; + static constexpr std::string_view MIN_S = "MIN.S"; + static constexpr std::string_view MIN_U = "MIN.U"; + static constexpr std::string_view MAX_F = "MAX.F"; + static constexpr std::string_view MAX_S = "MAX.S"; + static constexpr std::string_view MAX_U = "MAX.U"; + static constexpr std::string_view MOV_U = "MOV.U"; + static constexpr std::string_view TGBALLOT_U = "TGBALLOT.U"; + static constexpr std::string_view TGALL_U = "TGALL.U"; + static constexpr std::string_view TGANY_U = "TGANY.U"; + static constexpr std::string_view TGEQ_U = "TGEQ.U"; + static constexpr std::string_view EXCH = "EXCH"; + static constexpr std::string_view ADD = "ADD"; + static constexpr std::string_view MIN = "MIN"; + static constexpr std::string_view MAX = "MAX"; + static constexpr std::string_view AND = "AND"; + static constexpr std::string_view OR = "OR"; + static constexpr std::string_view XOR = "XOR"; + static constexpr std::string_view U32 = "U32"; + static constexpr std::string_view S32 = "S32"; + + static constexpr std::size_t NUM_ENTRIES = static_cast<std::size_t>(OperationCode::Amount); + using DecompilerType = std::string (ARBDecompiler::*)(Operation); + static constexpr std::array<DecompilerType, NUM_ENTRIES> OPERATION_DECOMPILERS = { + &ARBDecompiler::Assign, + + &ARBDecompiler::Select, + + &ARBDecompiler::Binary<ADD_F32>, + &ARBDecompiler::Binary<MUL_F32>, + &ARBDecompiler::Binary<DIV_F32>, + &ARBDecompiler::Trinary<MAD_F32>, + &ARBDecompiler::Negate<'F'>, + &ARBDecompiler::Absolute<'F'>, + &ARBDecompiler::FClamp, + &ARBDecompiler::FCastHalf0, + &ARBDecompiler::FCastHalf1, + &ARBDecompiler::Binary<MIN_F>, + &ARBDecompiler::Binary<MAX_F>, + &ARBDecompiler::Unary<COS_F32>, + &ARBDecompiler::Unary<SIN_F32>, + &ARBDecompiler::Unary<EX2_F32>, + &ARBDecompiler::Unary<LG2_F32>, + &ARBDecompiler::Unary<RSQ_F32>, + &ARBDecompiler::FSqrt, + &ARBDecompiler::Unary<ROUND_F>, + &ARBDecompiler::Unary<FLR_F>, + &ARBDecompiler::Unary<CEIL_F>, + &ARBDecompiler::Unary<TRUNC_F>, + &ARBDecompiler::Unary<I2F_S>, + &ARBDecompiler::Unary<I2F_U>, + &ARBDecompiler::FSwizzleAdd, + + &ARBDecompiler::Binary<ADD_S>, + &ARBDecompiler::Binary<MUL_S>, + &ARBDecompiler::Binary<DIV_S>, + &ARBDecompiler::Negate<'S'>, + &ARBDecompiler::Absolute<'S'>, + &ARBDecompiler::Binary<MIN_S>, + &ARBDecompiler::Binary<MAX_S>, + + &ARBDecompiler::Unary<TRUNC_S>, + &ARBDecompiler::Unary<MOV_U>, + &ARBDecompiler::Binary<SHL_S>, + &ARBDecompiler::Binary<SHR_U>, + &ARBDecompiler::Binary<SHR_S>, + &ARBDecompiler::Binary<AND_S>, + &ARBDecompiler::Binary<OR_S>, + &ARBDecompiler::Binary<XOR_S>, + &ARBDecompiler::Unary<NOT_S>, + &ARBDecompiler::BitfieldInsert<'S'>, + &ARBDecompiler::BitfieldExtract<'S'>, + &ARBDecompiler::Unary<BTC_S>, + &ARBDecompiler::Unary<BTFM_S>, + + &ARBDecompiler::Binary<ADD_U>, + &ARBDecompiler::Binary<MUL_U>, + &ARBDecompiler::Binary<DIV_U>, + &ARBDecompiler::Binary<MIN_U>, + &ARBDecompiler::Binary<MAX_U>, + &ARBDecompiler::Unary<TRUNC_U>, + &ARBDecompiler::Unary<MOV_U>, + &ARBDecompiler::Binary<SHL_U>, + &ARBDecompiler::Binary<SHR_U>, + &ARBDecompiler::Binary<SHR_U>, + &ARBDecompiler::Binary<AND_U>, + &ARBDecompiler::Binary<OR_U>, + &ARBDecompiler::Binary<XOR_U>, + &ARBDecompiler::Unary<NOT_U>, + &ARBDecompiler::BitfieldInsert<'U'>, + &ARBDecompiler::BitfieldExtract<'U'>, + &ARBDecompiler::Unary<BTC_U>, + &ARBDecompiler::Unary<BTFM_U>, + + &ARBDecompiler::HAdd2, + &ARBDecompiler::HMul2, + &ARBDecompiler::HFma2, + &ARBDecompiler::HAbsolute, + &ARBDecompiler::HNegate, + &ARBDecompiler::HClamp, + &ARBDecompiler::HCastFloat, + &ARBDecompiler::HUnpack, + &ARBDecompiler::HMergeF32, + &ARBDecompiler::HMergeH0, + &ARBDecompiler::HMergeH1, + &ARBDecompiler::HPack2, + + &ARBDecompiler::LogicalAssign, + &ARBDecompiler::Binary<AND_U>, + &ARBDecompiler::Binary<OR_U>, + &ARBDecompiler::Binary<XOR_U>, + &ARBDecompiler::Unary<NOT_U>, + &ARBDecompiler::LogicalPick2, + &ARBDecompiler::LogicalAnd2, + + &ARBDecompiler::FloatComparison<SLT_F, false>, + &ARBDecompiler::FloatComparison<SEQ_F, false>, + &ARBDecompiler::FloatComparison<SLE_F, false>, + &ARBDecompiler::FloatComparison<SGT_F, false>, + &ARBDecompiler::FloatComparison<SNE_F, false>, + &ARBDecompiler::FloatComparison<SGE_F, false>, + &ARBDecompiler::FloatOrdered, + &ARBDecompiler::FloatUnordered, + &ARBDecompiler::FloatComparison<SLT_F, true>, + &ARBDecompiler::FloatComparison<SEQ_F, true>, + &ARBDecompiler::FloatComparison<SLE_F, true>, + &ARBDecompiler::FloatComparison<SGT_F, true>, + &ARBDecompiler::FloatComparison<SNE_F, true>, + &ARBDecompiler::FloatComparison<SGE_F, true>, + + &ARBDecompiler::Binary<SLT_S>, + &ARBDecompiler::Binary<SEQ_S>, + &ARBDecompiler::Binary<SLE_S>, + &ARBDecompiler::Binary<SGT_S>, + &ARBDecompiler::Binary<SNE_S>, + &ARBDecompiler::Binary<SGE_S>, + + &ARBDecompiler::Binary<SLT_U>, + &ARBDecompiler::Binary<SEQ_U>, + &ARBDecompiler::Binary<SLE_U>, + &ARBDecompiler::Binary<SGT_U>, + &ARBDecompiler::Binary<SNE_U>, + &ARBDecompiler::Binary<SGE_U>, + + &ARBDecompiler::LogicalAddCarry, + + &ARBDecompiler::HalfComparison<SLT_F, false>, + &ARBDecompiler::HalfComparison<SEQ_F, false>, + &ARBDecompiler::HalfComparison<SLE_F, false>, + &ARBDecompiler::HalfComparison<SGT_F, false>, + &ARBDecompiler::HalfComparison<SNE_F, false>, + &ARBDecompiler::HalfComparison<SGE_F, false>, + &ARBDecompiler::HalfComparison<SLT_F, true>, + &ARBDecompiler::HalfComparison<SEQ_F, true>, + &ARBDecompiler::HalfComparison<SLE_F, true>, + &ARBDecompiler::HalfComparison<SGT_F, true>, + &ARBDecompiler::HalfComparison<SNE_F, true>, + &ARBDecompiler::HalfComparison<SGE_F, true>, + + &ARBDecompiler::Texture, + &ARBDecompiler::Texture, + &ARBDecompiler::TextureGather, + &ARBDecompiler::TextureQueryDimensions, + &ARBDecompiler::TextureQueryLod, + &ARBDecompiler::TexelFetch, + &ARBDecompiler::TextureGradient, + + &ARBDecompiler::ImageLoad, + &ARBDecompiler::ImageStore, + + &ARBDecompiler::AtomicImage<ADD, U32>, + &ARBDecompiler::AtomicImage<AND, U32>, + &ARBDecompiler::AtomicImage<OR, U32>, + &ARBDecompiler::AtomicImage<XOR, U32>, + &ARBDecompiler::AtomicImage<EXCH, U32>, + + &ARBDecompiler::Atomic<EXCH, U32>, + &ARBDecompiler::Atomic<ADD, U32>, + &ARBDecompiler::Atomic<MIN, U32>, + &ARBDecompiler::Atomic<MAX, U32>, + &ARBDecompiler::Atomic<AND, U32>, + &ARBDecompiler::Atomic<OR, U32>, + &ARBDecompiler::Atomic<XOR, U32>, + + &ARBDecompiler::Atomic<EXCH, S32>, + &ARBDecompiler::Atomic<ADD, S32>, + &ARBDecompiler::Atomic<MIN, S32>, + &ARBDecompiler::Atomic<MAX, S32>, + &ARBDecompiler::Atomic<AND, S32>, + &ARBDecompiler::Atomic<OR, S32>, + &ARBDecompiler::Atomic<XOR, S32>, + + &ARBDecompiler::Atomic<ADD, U32>, + &ARBDecompiler::Atomic<MIN, U32>, + &ARBDecompiler::Atomic<MAX, U32>, + &ARBDecompiler::Atomic<AND, U32>, + &ARBDecompiler::Atomic<OR, U32>, + &ARBDecompiler::Atomic<XOR, U32>, + + &ARBDecompiler::Atomic<ADD, S32>, + &ARBDecompiler::Atomic<MIN, S32>, + &ARBDecompiler::Atomic<MAX, S32>, + &ARBDecompiler::Atomic<AND, S32>, + &ARBDecompiler::Atomic<OR, S32>, + &ARBDecompiler::Atomic<XOR, S32>, + + &ARBDecompiler::Branch, + &ARBDecompiler::BranchIndirect, + &ARBDecompiler::PushFlowStack, + &ARBDecompiler::PopFlowStack, + &ARBDecompiler::Exit, + &ARBDecompiler::Discard, + + &ARBDecompiler::EmitVertex, + &ARBDecompiler::EndPrimitive, + + &ARBDecompiler::InvocationId, + &ARBDecompiler::YNegate, + &ARBDecompiler::LocalInvocationId<'x'>, + &ARBDecompiler::LocalInvocationId<'y'>, + &ARBDecompiler::LocalInvocationId<'z'>, + &ARBDecompiler::WorkGroupId<'x'>, + &ARBDecompiler::WorkGroupId<'y'>, + &ARBDecompiler::WorkGroupId<'z'>, + + &ARBDecompiler::Unary<TGBALLOT_U>, + &ARBDecompiler::Unary<TGALL_U>, + &ARBDecompiler::Unary<TGANY_U>, + &ARBDecompiler::Unary<TGEQ_U>, + + &ARBDecompiler::ThreadId, + &ARBDecompiler::ThreadMask<'e', 'q'>, + &ARBDecompiler::ThreadMask<'g', 'e'>, + &ARBDecompiler::ThreadMask<'g', 't'>, + &ARBDecompiler::ThreadMask<'l', 'e'>, + &ARBDecompiler::ThreadMask<'l', 't'>, + &ARBDecompiler::ShuffleIndexed, + + &ARBDecompiler::Barrier, + &ARBDecompiler::MemoryBarrierGroup, + &ARBDecompiler::MemoryBarrierGlobal, + }; +}; + +ARBDecompiler::ARBDecompiler(const Device& device, const ShaderIR& ir, const Registry& registry, + ShaderType stage, std::string_view identifier) + : device{device}, ir{ir}, registry{registry}, stage{stage} { + AddLine("TEMP RC;"); + AddLine("TEMP FSWZA[4];"); + AddLine("TEMP FSWZB[4];"); + if (ir.IsDecompiled()) { + DecompileAST(); + } else { + DecompileBranchMode(); + } + AddLine("END"); + + const std::string code = std::move(shader_source); + DeclareHeader(); + DeclareVertex(); + DeclareGeometry(); + DeclareFragment(); + DeclareCompute(); + DeclareInputAttributes(); + DeclareOutputAttributes(); + DeclareLocalMemory(); + DeclareGlobalMemory(); + DeclareConstantBuffers(); + DeclareRegisters(); + DeclareTemporaries(); + DeclarePredicates(); + DeclareInternalFlags(); + + shader_source += code; +} + +std::string_view HeaderStageName(ShaderType stage) { + switch (stage) { + case ShaderType::Vertex: + return "vp"; + case ShaderType::Geometry: + return "gp"; + case ShaderType::Fragment: + return "fp"; + case ShaderType::Compute: + return "cp"; + default: + UNREACHABLE(); + return ""; + } +} + +void ARBDecompiler::DeclareHeader() { + AddLine("!!NV{}5.0", HeaderStageName(stage)); + // Enabling this allows us to cheat on some instructions like TXL with SHADOWARRAY2D + AddLine("OPTION NV_internal;"); + AddLine("OPTION NV_gpu_program_fp64;"); + AddLine("OPTION NV_shader_storage_buffer;"); + AddLine("OPTION NV_shader_thread_group;"); + if (ir.UsesWarps() && device.HasWarpIntrinsics()) { + AddLine("OPTION NV_shader_thread_shuffle;"); + } + if (stage == ShaderType::Vertex) { + if (device.HasNvViewportArray2()) { + AddLine("OPTION NV_viewport_array2;"); + } + } + if (stage == ShaderType::Fragment) { + AddLine("OPTION ARB_draw_buffers;"); + } + if (device.HasImageLoadFormatted()) { + AddLine("OPTION EXT_shader_image_load_formatted;"); + } +} + +void ARBDecompiler::DeclareVertex() { + if (stage != ShaderType::Vertex) { + return; + } + AddLine("OUTPUT result_clip[] = {{ result.clip[0..7] }};"); +} + +void ARBDecompiler::DeclareGeometry() { + if (stage != ShaderType::Geometry) { + return; + } + const auto& info = registry.GetGraphicsInfo(); + const auto& header = ir.GetHeader(); + AddLine("PRIMITIVE_IN {};", PrimitiveDescription(info.primitive_topology)); + AddLine("PRIMITIVE_OUT {};", TopologyName(header.common3.output_topology)); + AddLine("VERTICES_OUT {};", header.common4.max_output_vertices.Value()); + AddLine("ATTRIB vertex_position = vertex.position;"); +} + +void ARBDecompiler::DeclareFragment() { + if (stage != ShaderType::Fragment) { + return; + } + AddLine("OUTPUT result_color7 = result.color[7];"); + AddLine("OUTPUT result_color6 = result.color[6];"); + AddLine("OUTPUT result_color5 = result.color[5];"); + AddLine("OUTPUT result_color4 = result.color[4];"); + AddLine("OUTPUT result_color3 = result.color[3];"); + AddLine("OUTPUT result_color2 = result.color[2];"); + AddLine("OUTPUT result_color1 = result.color[1];"); + AddLine("OUTPUT result_color0 = result.color;"); +} + +void ARBDecompiler::DeclareCompute() { + if (stage != ShaderType::Compute) { + return; + } + const ComputeInfo& info = registry.GetComputeInfo(); + AddLine("GROUP_SIZE {} {} {};", info.workgroup_size[0], info.workgroup_size[1], + info.workgroup_size[2]); + if (info.shared_memory_size_in_words > 0) { + const u32 size_in_bytes = info.shared_memory_size_in_words * 4; + AddLine("SHARED_MEMORY {};", size_in_bytes); + AddLine("SHARED shared_mem[] = {{program.sharedmem}};"); + } +} + +void ARBDecompiler::DeclareInputAttributes() { + if (stage == ShaderType::Compute) { + return; + } + const std::string_view stage_name = StageInputName(stage); + for (const auto attribute : ir.GetInputAttributes()) { + if (!IsGenericAttribute(attribute)) { + continue; + } + const u32 index = GetGenericAttributeIndex(attribute); + + std::string_view suffix; + if (stage == ShaderType::Fragment) { + const auto input_mode{ir.GetHeader().ps.GetPixelImap(index)}; + if (input_mode == PixelImap::Unused) { + return; + } + suffix = GetInputFlags(input_mode); + } + AddLine("{}ATTRIB in_attr{}[] = {{ {}.attrib[{}..{}] }};", suffix, index, stage_name, index, + index); + } +} + +void ARBDecompiler::DeclareOutputAttributes() { + if (stage == ShaderType::Compute) { + return; + } + for (const auto attribute : ir.GetOutputAttributes()) { + if (!IsGenericAttribute(attribute)) { + continue; + } + const u32 index = GetGenericAttributeIndex(attribute); + AddLine("OUTPUT out_attr{}[] = {{ result.attrib[{}..{}] }};", index, index, index); + } +} + +void ARBDecompiler::DeclareLocalMemory() { + u64 size = 0; + if (stage == ShaderType::Compute) { + size = registry.GetComputeInfo().local_memory_size_in_words * 4ULL; + } else { + size = ir.GetHeader().GetLocalMemorySize(); + } + if (size == 0) { + return; + } + const u64 element_count = Common::AlignUp(size, 4) / 4; + AddLine("TEMP lmem[{}];", element_count); +} + +void ARBDecompiler::DeclareGlobalMemory() { + u32 binding = 0; // device.GetBaseBindings(stage).shader_storage_buffer; + for (const auto& pair : ir.GetGlobalMemory()) { + const auto& base = pair.first; + AddLine("STORAGE {}[] = {{ program.storage[{}] }};", GlobalMemoryName(base), binding); + ++binding; + } +} + +void ARBDecompiler::DeclareConstantBuffers() { + u32 binding = 0; + for (const auto& cbuf : ir.GetConstantBuffers()) { + AddLine("CBUFFER cbuf{}[] = {{ program.buffer[{}] }};", cbuf.first, binding); + ++binding; + } +} + +void ARBDecompiler::DeclareRegisters() { + for (const u32 gpr : ir.GetRegisters()) { + AddLine("TEMP R{};", gpr); + } +} + +void ARBDecompiler::DeclareTemporaries() { + for (std::size_t i = 0; i < max_temporaries; ++i) { + AddLine("TEMP T{};", i); + } +} + +void ARBDecompiler::DeclarePredicates() { + for (const Tegra::Shader::Pred pred : ir.GetPredicates()) { + AddLine("TEMP P{};", static_cast<u64>(pred)); + } +} + +void ARBDecompiler::DeclareInternalFlags() { + for (const char* name : INTERNAL_FLAG_NAMES) { + AddLine("TEMP {};", name); + } +} + +void ARBDecompiler::InitializeVariables() { + AddLine("MOV.F32 FSWZA[0], -1;"); + AddLine("MOV.F32 FSWZA[1], 1;"); + AddLine("MOV.F32 FSWZA[2], -1;"); + AddLine("MOV.F32 FSWZA[3], 0;"); + AddLine("MOV.F32 FSWZB[0], -1;"); + AddLine("MOV.F32 FSWZB[1], -1;"); + AddLine("MOV.F32 FSWZB[2], 1;"); + AddLine("MOV.F32 FSWZB[3], -1;"); + + if (stage == ShaderType::Vertex || stage == ShaderType::Geometry) { + AddLine("MOV.F result.position, {{0, 0, 0, 1}};"); + } + for (const auto attribute : ir.GetOutputAttributes()) { + if (!IsGenericAttribute(attribute)) { + continue; + } + const u32 index = GetGenericAttributeIndex(attribute); + AddLine("MOV.F result.attrib[{}], {{0, 0, 0, 1}};", index); + } + for (const u32 gpr : ir.GetRegisters()) { + AddLine("MOV.F R{}, {{0, 0, 0, 0}};", gpr); + } + for (const Tegra::Shader::Pred pred : ir.GetPredicates()) { + AddLine("MOV.U P{}, {{0, 0, 0, 0}};", static_cast<u64>(pred)); + } +} + +void ARBDecompiler::DecompileAST() { + const u32 num_flow_variables = ir.GetASTNumVariables(); + for (u32 i = 0; i < num_flow_variables; ++i) { + AddLine("TEMP F{};", i); + } + for (u32 i = 0; i < num_flow_variables; ++i) { + AddLine("MOV.U F{}, {{0, 0, 0, 0}};", i); + } + + InitializeVariables(); + + VisitAST(ir.GetASTProgram()); +} + +void ARBDecompiler::DecompileBranchMode() { + static constexpr u32 FLOW_STACK_SIZE = 20; + if (!ir.IsFlowStackDisabled()) { + AddLine("TEMP SSY[{}];", FLOW_STACK_SIZE); + AddLine("TEMP PBK[{}];", FLOW_STACK_SIZE); + AddLine("TEMP SSY_TOP;"); + AddLine("TEMP PBK_TOP;"); + } + + AddLine("TEMP PC;"); + + if (!ir.IsFlowStackDisabled()) { + AddLine("MOV.U SSY_TOP.x, 0;"); + AddLine("MOV.U PBK_TOP.x, 0;"); + } + + InitializeVariables(); + + const auto basic_block_end = ir.GetBasicBlocks().end(); + auto basic_block_it = ir.GetBasicBlocks().begin(); + const u32 first_address = basic_block_it->first; + AddLine("MOV.U PC.x, {};", first_address); + + AddLine("REP;"); + + std::size_t num_blocks = 0; + while (basic_block_it != basic_block_end) { + const auto& [address, bb] = *basic_block_it; + ++num_blocks; + + AddLine("SEQ.S.CC RC.x, PC.x, {};", address); + AddLine("IF NE.x;"); + + VisitBlock(bb); + + ++basic_block_it; + + if (basic_block_it != basic_block_end) { + const auto op = std::get_if<OperationNode>(&*bb[bb.size() - 1]); + if (!op || op->GetCode() != OperationCode::Branch) { + const u32 next_address = basic_block_it->first; + AddLine("MOV.U PC.x, {};", next_address); + AddLine("CONT;"); + } + } + + AddLine("ELSE;"); + } + AddLine("RET;"); + while (num_blocks--) { + AddLine("ENDIF;"); + } + + AddLine("ENDREP;"); +} + +void ARBDecompiler::VisitAST(const ASTNode& node) { + if (const auto ast = std::get_if<ASTProgram>(&*node->GetInnerData())) { + for (ASTNode current = ast->nodes.GetFirst(); current; current = current->GetNext()) { + VisitAST(current); + } + } else if (const auto ast = std::get_if<ASTIfThen>(&*node->GetInnerData())) { + const std::string condition = VisitExpression(ast->condition); + ResetTemporaries(); + + AddLine("MOVC.U RC.x, {};", condition); + AddLine("IF NE.x;"); + for (ASTNode current = ast->nodes.GetFirst(); current; current = current->GetNext()) { + VisitAST(current); + } + AddLine("ENDIF;"); + } else if (const auto ast = std::get_if<ASTIfElse>(&*node->GetInnerData())) { + AddLine("ELSE;"); + for (ASTNode current = ast->nodes.GetFirst(); current; current = current->GetNext()) { + VisitAST(current); + } + } else if (const auto ast = std::get_if<ASTBlockDecoded>(&*node->GetInnerData())) { + VisitBlock(ast->nodes); + } else if (const auto ast = std::get_if<ASTVarSet>(&*node->GetInnerData())) { + AddLine("MOV.U F{}, {};", ast->index, VisitExpression(ast->condition)); + ResetTemporaries(); + } else if (const auto ast = std::get_if<ASTDoWhile>(&*node->GetInnerData())) { + const std::string condition = VisitExpression(ast->condition); + ResetTemporaries(); + AddLine("REP;"); + for (ASTNode current = ast->nodes.GetFirst(); current; current = current->GetNext()) { + VisitAST(current); + } + AddLine("MOVC.U RC.x, {};", condition); + AddLine("BRK (NE.x);"); + AddLine("ENDREP;"); + } else if (const auto ast = std::get_if<ASTReturn>(&*node->GetInnerData())) { + const bool is_true = ExprIsTrue(ast->condition); + if (!is_true) { + AddLine("MOVC.U RC.x, {};", VisitExpression(ast->condition)); + AddLine("IF NE.x;"); + ResetTemporaries(); + } + if (ast->kills) { + AddLine("KIL TR;"); + } else { + Exit(); + } + if (!is_true) { + AddLine("ENDIF;"); + } + } else if (const auto ast = std::get_if<ASTBreak>(&*node->GetInnerData())) { + if (ExprIsTrue(ast->condition)) { + AddLine("BRK;"); + } else { + AddLine("MOVC.U RC.x, {};", VisitExpression(ast->condition)); + AddLine("BRK (NE.x);"); + ResetTemporaries(); + } + } else if (std::holds_alternative<ASTLabel>(*node->GetInnerData())) { + // Nothing to do + } else { + UNREACHABLE(); + } +} + +std::string ARBDecompiler::VisitExpression(const Expr& node) { + if (const auto expr = std::get_if<ExprAnd>(&*node)) { + std::string result = AllocTemporary(); + AddLine("AND.U {}, {}, {};", result, VisitExpression(expr->operand1), + VisitExpression(expr->operand2)); + return result; + } + if (const auto expr = std::get_if<ExprOr>(&*node)) { + std::string result = AllocTemporary(); + AddLine("OR.U {}, {}, {};", result, VisitExpression(expr->operand1), + VisitExpression(expr->operand2)); + return result; + } + if (const auto expr = std::get_if<ExprNot>(&*node)) { + std::string result = AllocTemporary(); + AddLine("CMP.S {}, {}, 0, -1;", result, VisitExpression(expr->operand1)); + return result; + } + if (const auto expr = std::get_if<ExprPredicate>(&*node)) { + return fmt::format("P{}.x", static_cast<u64>(expr->predicate)); + } + if (const auto expr = std::get_if<ExprCondCode>(&*node)) { + return Visit(ir.GetConditionCode(expr->cc)); + } + if (const auto expr = std::get_if<ExprVar>(&*node)) { + return fmt::format("F{}.x", expr->var_index); + } + if (const auto expr = std::get_if<ExprBoolean>(&*node)) { + return expr->value ? "0xffffffff" : "0"; + } + if (const auto expr = std::get_if<ExprGprEqual>(&*node)) { + std::string result = AllocTemporary(); + AddLine("SEQ.U {}, R{}.x, {};", result, expr->gpr, expr->value); + return result; + } + UNREACHABLE(); + return "0"; +} + +void ARBDecompiler::VisitBlock(const NodeBlock& bb) { + for (const auto& node : bb) { + Visit(node); + } +} + +std::string ARBDecompiler::Visit(const Node& node) { + if (const auto operation = std::get_if<OperationNode>(&*node)) { + if (const auto amend_index = operation->GetAmendIndex()) { + Visit(ir.GetAmendNode(*amend_index)); + } + const std::size_t index = static_cast<std::size_t>(operation->GetCode()); + if (index >= OPERATION_DECOMPILERS.size()) { + UNREACHABLE_MSG("Out of bounds operation: {}", index); + return {}; + } + const auto decompiler = OPERATION_DECOMPILERS[index]; + if (decompiler == nullptr) { + UNREACHABLE_MSG("Undefined operation: {}", index); + return {}; + } + return (this->*decompiler)(*operation); + } + + if (const auto gpr = std::get_if<GprNode>(&*node)) { + const u32 index = gpr->GetIndex(); + if (index == Register::ZeroIndex) { + return "{0, 0, 0, 0}.x"; + } + return fmt::format("R{}.x", index); + } + + if (const auto cv = std::get_if<CustomVarNode>(&*node)) { + return fmt::format("CV{}.x", cv->GetIndex()); + } + + if (const auto immediate = std::get_if<ImmediateNode>(&*node)) { + std::string temporary = AllocTemporary(); + AddLine("MOV.U {}, {};", temporary, immediate->GetValue()); + return temporary; + } + + if (const auto predicate = std::get_if<PredicateNode>(&*node)) { + std::string temporary = AllocTemporary(); + switch (const auto index = predicate->GetIndex(); index) { + case Tegra::Shader::Pred::UnusedIndex: + AddLine("MOV.S {}, -1;", temporary); + break; + case Tegra::Shader::Pred::NeverExecute: + AddLine("MOV.S {}, 0;", temporary); + break; + default: + AddLine("MOV.S {}, P{}.x;", temporary, static_cast<u64>(index)); + break; + } + if (predicate->IsNegated()) { + AddLine("CMP.S {}, {}, 0, -1;", temporary, temporary); + } + return temporary; + } + + if (const auto abuf = std::get_if<AbufNode>(&*node)) { + if (abuf->IsPhysicalBuffer()) { + UNIMPLEMENTED_MSG("Physical buffers are not implemented"); + return "{0, 0, 0, 0}.x"; + } + + const auto buffer_index = [this, &abuf]() -> std::string { + if (stage != ShaderType::Geometry) { + return ""; + } + return fmt::format("[{}]", Visit(abuf->GetBuffer())); + }; + + const Attribute::Index index = abuf->GetIndex(); + const u32 element = abuf->GetElement(); + const char swizzle = Swizzle(element); + switch (index) { + case Attribute::Index::Position: { + if (stage == ShaderType::Geometry) { + return fmt::format("{}_position[{}].{}", StageInputName(stage), + Visit(abuf->GetBuffer()), swizzle); + } else { + return fmt::format("{}.position.{}", StageInputName(stage), swizzle); + } + } + case Attribute::Index::TessCoordInstanceIDVertexID: + ASSERT(stage == ShaderType::Vertex); + switch (element) { + case 2: + return "vertex.instance"; + case 3: + return "vertex.id"; + } + UNIMPLEMENTED_MSG("Unmanaged TessCoordInstanceIDVertexID element={}", element); + break; + case Attribute::Index::PointCoord: + switch (element) { + case 0: + return "fragment.pointcoord.x"; + case 1: + return "fragment.pointcoord.y"; + } + UNIMPLEMENTED(); + break; + case Attribute::Index::FrontFacing: { + ASSERT(stage == ShaderType::Fragment); + ASSERT(element == 3); + const std::string temporary = AllocVectorTemporary(); + AddLine("SGT.S RC.x, fragment.facing, {{0, 0, 0, 0}};"); + AddLine("MOV.U.CC RC.x, -RC;"); + AddLine("MOV.S {}.x, 0;", temporary); + AddLine("MOV.S {}.x (NE.x), -1;", temporary); + return fmt::format("{}.x", temporary); + } + default: + if (IsGenericAttribute(index)) { + if (stage == ShaderType::Geometry) { + return fmt::format("in_attr{}[{}][0].{}", GetGenericAttributeIndex(index), + Visit(abuf->GetBuffer()), swizzle); + } else { + return fmt::format("{}.attrib[{}].{}", StageInputName(stage), + GetGenericAttributeIndex(index), swizzle); + } + } + UNIMPLEMENTED_MSG("Unimplemented input attribute={}", static_cast<int>(index)); + break; + } + return "{0, 0, 0, 0}.x"; + } + + if (const auto cbuf = std::get_if<CbufNode>(&*node)) { + std::string offset_string; + const auto& offset = cbuf->GetOffset(); + if (const auto imm = std::get_if<ImmediateNode>(&*offset)) { + offset_string = std::to_string(imm->GetValue()); + } else { + offset_string = Visit(offset); + } + std::string temporary = AllocTemporary(); + AddLine("LDC.F32 {}, cbuf{}[{}];", temporary, cbuf->GetIndex(), offset_string); + return temporary; + } + + if (const auto gmem = std::get_if<GmemNode>(&*node)) { + std::string temporary = AllocTemporary(); + AddLine("SUB.U {}, {}, {};", temporary, Visit(gmem->GetRealAddress()), + Visit(gmem->GetBaseAddress())); + AddLine("LDB.U32 {}, {}[{}];", temporary, GlobalMemoryName(gmem->GetDescriptor()), + temporary); + return temporary; + } + + if (const auto lmem = std::get_if<LmemNode>(&*node)) { + std::string temporary = Visit(lmem->GetAddress()); + AddLine("SHR.U {}, {}, 2;", temporary, temporary); + AddLine("MOV.U {}, lmem[{}].x;", temporary, temporary); + return temporary; + } + + if (const auto smem = std::get_if<SmemNode>(&*node)) { + std::string temporary = Visit(smem->GetAddress()); + AddLine("LDS.U32 {}, shared_mem[{}];", temporary, temporary); + return temporary; + } + + if (const auto internal_flag = std::get_if<InternalFlagNode>(&*node)) { + const std::size_t index = static_cast<std::size_t>(internal_flag->GetFlag()); + return fmt::format("{}.x", INTERNAL_FLAG_NAMES[index]); + } + + if (const auto conditional = std::get_if<ConditionalNode>(&*node)) { + if (const auto amend_index = conditional->GetAmendIndex()) { + Visit(ir.GetAmendNode(*amend_index)); + } + AddLine("MOVC.U RC.x, {};", Visit(conditional->GetCondition())); + AddLine("IF NE.x;"); + VisitBlock(conditional->GetCode()); + AddLine("ENDIF;"); + return {}; + } + + if (const auto cmt = std::get_if<CommentNode>(&*node)) { + // Uncommenting this will generate invalid code. GLASM lacks comments. + // AddLine("// {}", cmt->GetText()); + return {}; + } + + UNIMPLEMENTED(); + return {}; +} + +std::pair<std::string, std::size_t> ARBDecompiler::BuildCoords(Operation operation) { + const auto& meta = std::get<MetaTexture>(operation.GetMeta()); + UNIMPLEMENTED_IF(meta.sampler.is_indexed); + UNIMPLEMENTED_IF(meta.sampler.is_shadow && meta.sampler.is_array && + meta.sampler.type == Tegra::Shader::TextureType::TextureCube); + + const std::size_t count = operation.GetOperandsCount(); + std::string temporary = AllocVectorTemporary(); + std::size_t i = 0; + for (; i < count; ++i) { + AddLine("MOV.F {}.{}, {};", temporary, Swizzle(i), Visit(operation[i])); + } + if (meta.sampler.is_array) { + AddLine("I2F.S {}.{}, {};", temporary, Swizzle(i++), Visit(meta.array)); + } + if (meta.sampler.is_shadow) { + AddLine("MOV.F {}.{}, {};", temporary, Swizzle(i++), Visit(meta.depth_compare)); + } + return {std::move(temporary), i}; +} + +std::string ARBDecompiler::BuildAoffi(Operation operation) { + const auto& meta = std::get<MetaTexture>(operation.GetMeta()); + if (meta.aoffi.empty()) { + return {}; + } + const std::string temporary = AllocVectorTemporary(); + std::size_t i = 0; + for (auto& node : meta.aoffi) { + AddLine("MOV.S {}.{}, {};", temporary, Swizzle(i++), Visit(node)); + } + return fmt::format(", offset({})", temporary); +} + +void ARBDecompiler::Exit() { + if (stage != ShaderType::Fragment) { + AddLine("RET;"); + return; + } + + const auto safe_get_register = [this](u32 reg) -> std::string { + // TODO(Rodrigo): Replace with contains once C++20 releases + const auto& used_registers = ir.GetRegisters(); + if (used_registers.find(reg) != used_registers.end()) { + return fmt::format("R{}.x", reg); + } + return "{0, 0, 0, 0}.x"; + }; + + const auto& header = ir.GetHeader(); + u32 current_reg = 0; + for (u32 rt = 0; rt < Tegra::Engines::Maxwell3D::Regs::NumRenderTargets; ++rt) { + for (u32 component = 0; component < 4; ++component) { + if (!header.ps.IsColorComponentOutputEnabled(rt, component)) { + continue; + } + AddLine("MOV.F result_color{}.{}, {};", rt, Swizzle(component), + safe_get_register(current_reg)); + ++current_reg; + } + } + if (header.ps.omap.depth) { + AddLine("MOV.F result.depth.z, {};", safe_get_register(current_reg + 1)); + } + + AddLine("RET;"); +} + +std::string ARBDecompiler::Assign(Operation operation) { + const Node& dest = operation[0]; + const Node& src = operation[1]; + + std::string dest_name; + if (const auto gpr = std::get_if<GprNode>(&*dest)) { + if (gpr->GetIndex() == Register::ZeroIndex) { + // Writing to Register::ZeroIndex is a no op + return {}; + } + dest_name = fmt::format("R{}.x", gpr->GetIndex()); + } else if (const auto abuf = std::get_if<AbufNode>(&*dest)) { + const u32 element = abuf->GetElement(); + const char swizzle = Swizzle(element); + switch (const Attribute::Index index = abuf->GetIndex()) { + case Attribute::Index::Position: + dest_name = fmt::format("result.position.{}", swizzle); + break; + case Attribute::Index::LayerViewportPointSize: + switch (element) { + case 0: + UNIMPLEMENTED(); + return {}; + case 1: + case 2: + if (!device.HasNvViewportArray2()) { + LOG_ERROR( + Render_OpenGL, + "NV_viewport_array2 is missing. Maxwell gen 2 or better is required."); + return {}; + } + dest_name = element == 1 ? "result.layer.x" : "result.viewport.x"; + break; + case 3: + dest_name = "result.pointsize.x"; + break; + } + break; + case Attribute::Index::ClipDistances0123: + dest_name = fmt::format("result.clip[{}].x", element); + break; + case Attribute::Index::ClipDistances4567: + dest_name = fmt::format("result.clip[{}].x", element + 4); + break; + default: + if (!IsGenericAttribute(index)) { + UNREACHABLE(); + return {}; + } + dest_name = + fmt::format("result.attrib[{}].{}", GetGenericAttributeIndex(index), swizzle); + break; + } + } else if (const auto lmem = std::get_if<LmemNode>(&*dest)) { + const std::string address = Visit(lmem->GetAddress()); + AddLine("SHR.U {}, {}, 2;", address, address); + dest_name = fmt::format("lmem[{}].x", address); + } else if (const auto smem = std::get_if<SmemNode>(&*dest)) { + AddLine("STS.U32 {}, shared_mem[{}];", Visit(src), Visit(smem->GetAddress())); + ResetTemporaries(); + return {}; + } else if (const auto gmem = std::get_if<GmemNode>(&*dest)) { + const std::string temporary = AllocTemporary(); + AddLine("SUB.U {}, {}, {};", temporary, Visit(gmem->GetRealAddress()), + Visit(gmem->GetBaseAddress())); + AddLine("STB.U32 {}, {}[{}];", Visit(src), GlobalMemoryName(gmem->GetDescriptor()), + temporary); + ResetTemporaries(); + return {}; + } else { + UNREACHABLE(); + ResetTemporaries(); + return {}; + } + + AddLine("MOV.U {}, {};", dest_name, Visit(src)); + ResetTemporaries(); + return {}; +} + +std::string ARBDecompiler::Select(Operation operation) { + std::string temporary = AllocTemporary(); + AddLine("CMP.S {}, {}, {}, {};", temporary, Visit(operation[0]), Visit(operation[1]), + Visit(operation[2])); + return temporary; +} + +std::string ARBDecompiler::FClamp(Operation operation) { + // 1.0f in hex, replace with std::bit_cast on C++20 + static constexpr u32 POSITIVE_ONE = 0x3f800000; + + std::string temporary = AllocTemporary(); + const Node& value = operation[0]; + const Node& low = operation[1]; + const Node& high = operation[2]; + const auto* const imm_low = std::get_if<ImmediateNode>(&*low); + const auto* const imm_high = std::get_if<ImmediateNode>(&*high); + if (imm_low && imm_high && imm_low->GetValue() == 0 && imm_high->GetValue() == POSITIVE_ONE) { + AddLine("MOV.F32.SAT {}, {};", temporary, Visit(value)); + } else { + AddLine("MIN.F {}, {}, {};", temporary, Visit(value), Visit(high)); + AddLine("MAX.F {}, {}, {};", temporary, temporary, Visit(low)); + } + return temporary; +} + +std::string ARBDecompiler::FCastHalf0(Operation operation) { + const std::string temporary = AllocVectorTemporary(); + AddLine("UP2H.F {}.x, {};", temporary, Visit(operation[0])); + return fmt::format("{}.x", temporary); +} + +std::string ARBDecompiler::FCastHalf1(Operation operation) { + const std::string temporary = AllocVectorTemporary(); + AddLine("UP2H.F {}.y, {};", temporary, Visit(operation[0])); + AddLine("MOV {}.x, {}.y;", temporary, temporary); + return fmt::format("{}.x", temporary); +} + +std::string ARBDecompiler::FSqrt(Operation operation) { + std::string temporary = AllocTemporary(); + AddLine("RSQ.F32 {}, {};", temporary, Visit(operation[0])); + AddLine("RCP.F32 {}, {};", temporary, temporary); + return temporary; +} + +std::string ARBDecompiler::FSwizzleAdd(Operation operation) { + const std::string temporary = AllocVectorTemporary(); + if (!device.HasWarpIntrinsics()) { + LOG_ERROR(Render_OpenGL, + "NV_shader_thread_shuffle is missing. Kepler or better is required."); + AddLine("ADD.F {}.x, {}, {};", temporary, Visit(operation[0]), Visit(operation[1])); + return fmt::format("{}.x", temporary); + } + + AddLine("AND.U {}.z, {}.threadid, 3;", temporary, StageInputName(stage)); + AddLine("SHL.U {}.z, {}.z, 1;", temporary, temporary); + AddLine("SHR.U {}.z, {}, {}.z;", temporary, Visit(operation[2]), temporary); + AddLine("AND.U {}.z, {}.z, 3;", temporary, temporary); + AddLine("MUL.F32 {}.x, {}, FSWZA[{}.z];", temporary, Visit(operation[0]), temporary); + AddLine("MUL.F32 {}.y, {}, FSWZB[{}.z];", temporary, Visit(operation[1]), temporary); + AddLine("ADD.F32 {}.x, {}.x, {}.y;", temporary, temporary, temporary); + return fmt::format("{}.x", temporary); +} + +std::string ARBDecompiler::HAdd2(Operation operation) { + const std::string tmp1 = AllocVectorTemporary(); + const std::string tmp2 = AllocVectorTemporary(); + AddLine("UP2H.F {}.xy, {};", tmp1, Visit(operation[0])); + AddLine("UP2H.F {}.xy, {};", tmp2, Visit(operation[1])); + AddLine("ADD.F16 {}, {}, {};", tmp1, tmp1, tmp2); + AddLine("PK2H.F {}.x, {};", tmp1, tmp1); + return fmt::format("{}.x", tmp1); +} + +std::string ARBDecompiler::HMul2(Operation operation) { + const std::string tmp1 = AllocVectorTemporary(); + const std::string tmp2 = AllocVectorTemporary(); + AddLine("UP2H.F {}.xy, {};", tmp1, Visit(operation[0])); + AddLine("UP2H.F {}.xy, {};", tmp2, Visit(operation[1])); + AddLine("MUL.F16 {}, {}, {};", tmp1, tmp1, tmp2); + AddLine("PK2H.F {}.x, {};", tmp1, tmp1); + return fmt::format("{}.x", tmp1); +} + +std::string ARBDecompiler::HFma2(Operation operation) { + const std::string tmp1 = AllocVectorTemporary(); + const std::string tmp2 = AllocVectorTemporary(); + const std::string tmp3 = AllocVectorTemporary(); + AddLine("UP2H.F {}.xy, {};", tmp1, Visit(operation[0])); + AddLine("UP2H.F {}.xy, {};", tmp2, Visit(operation[1])); + AddLine("UP2H.F {}.xy, {};", tmp3, Visit(operation[2])); + AddLine("MAD.F16 {}, {}, {}, {};", tmp1, tmp1, tmp2, tmp3); + AddLine("PK2H.F {}.x, {};", tmp1, tmp1); + return fmt::format("{}.x", tmp1); +} + +std::string ARBDecompiler::HAbsolute(Operation operation) { + const std::string temporary = AllocVectorTemporary(); + AddLine("UP2H.F {}.xy, {};", temporary, Visit(operation[0])); + AddLine("PK2H.F {}.x, |{}|;", temporary, temporary); + return fmt::format("{}.x", temporary); +} + +std::string ARBDecompiler::HNegate(Operation operation) { + const std::string temporary = AllocVectorTemporary(); + AddLine("UP2H.F {}.xy, {};", temporary, Visit(operation[0])); + AddLine("MOVC.S RC.x, {};", Visit(operation[1])); + AddLine("MOV.F {}.x (NE.x), -{}.x;", temporary, temporary); + AddLine("MOVC.S RC.x, {};", Visit(operation[2])); + AddLine("MOV.F {}.y (NE.x), -{}.y;", temporary, temporary); + AddLine("PK2H.F {}.x, {};", temporary, temporary); + return fmt::format("{}.x", temporary); +} + +std::string ARBDecompiler::HClamp(Operation operation) { + const std::string tmp1 = AllocVectorTemporary(); + const std::string tmp2 = AllocVectorTemporary(); + AddLine("UP2H.F {}.xy, {};", tmp1, Visit(operation[0])); + AddLine("MOV.U {}.x, {};", tmp2, Visit(operation[1])); + AddLine("MOV.U {}.y, {}.x;", tmp2, tmp2); + AddLine("MAX.F {}, {}, {};", tmp1, tmp1, tmp2); + AddLine("MOV.U {}.x, {};", tmp2, Visit(operation[2])); + AddLine("MOV.U {}.y, {}.x;", tmp2, tmp2); + AddLine("MIN.F {}, {}, {};", tmp1, tmp1, tmp2); + AddLine("PK2H.F {}.x, {};", tmp1, tmp1); + return fmt::format("{}.x", tmp1); +} + +std::string ARBDecompiler::HCastFloat(Operation operation) { + const std::string temporary = AllocVectorTemporary(); + AddLine("MOV.F {}.y, {{0, 0, 0, 0}};", temporary); + AddLine("MOV.F {}.x, {};", temporary, Visit(operation[0])); + AddLine("PK2H.F {}.x, {};", temporary, temporary); + return fmt::format("{}.x", temporary); +} + +std::string ARBDecompiler::HUnpack(Operation operation) { + const std::string operand = Visit(operation[0]); + switch (std::get<Tegra::Shader::HalfType>(operation.GetMeta())) { + case Tegra::Shader::HalfType::H0_H1: + return operand; + case Tegra::Shader::HalfType::F32: { + const std::string temporary = AllocVectorTemporary(); + AddLine("MOV.U {}.x, {};", temporary, operand); + AddLine("MOV.U {}.y, {}.x;", temporary, temporary); + AddLine("PK2H.F {}.x, {};", temporary, temporary); + return fmt::format("{}.x", temporary); + } + case Tegra::Shader::HalfType::H0_H0: { + const std::string temporary = AllocVectorTemporary(); + AddLine("UP2H.F {}.xy, {};", temporary, operand); + AddLine("MOV.U {}.y, {}.x;", temporary, temporary); + AddLine("PK2H.F {}.x, {};", temporary, temporary); + return fmt::format("{}.x", temporary); + } + case Tegra::Shader::HalfType::H1_H1: { + const std::string temporary = AllocVectorTemporary(); + AddLine("UP2H.F {}.xy, {};", temporary, operand); + AddLine("MOV.U {}.x, {}.y;", temporary, temporary); + AddLine("PK2H.F {}.x, {};", temporary, temporary); + return fmt::format("{}.x", temporary); + } + } + UNREACHABLE(); + return "{0, 0, 0, 0}.x"; +} + +std::string ARBDecompiler::HMergeF32(Operation operation) { + const std::string temporary = AllocVectorTemporary(); + AddLine("UP2H.F {}.xy, {};", temporary, Visit(operation[0])); + return fmt::format("{}.x", temporary); +} + +std::string ARBDecompiler::HMergeH0(Operation operation) { + const std::string temporary = AllocVectorTemporary(); + AddLine("UP2H.F {}.xy, {};", temporary, Visit(operation[0])); + AddLine("UP2H.F {}.zw, {};", temporary, Visit(operation[1])); + AddLine("MOV.U {}.x, {}.z;", temporary, temporary); + AddLine("PK2H.F {}.x, {};", temporary, temporary); + return fmt::format("{}.x", temporary); +} + +std::string ARBDecompiler::HMergeH1(Operation operation) { + const std::string temporary = AllocVectorTemporary(); + AddLine("UP2H.F {}.xy, {};", temporary, Visit(operation[0])); + AddLine("UP2H.F {}.zw, {};", temporary, Visit(operation[1])); + AddLine("MOV.U {}.y, {}.w;", temporary, temporary); + AddLine("PK2H.F {}.x, {};", temporary, temporary); + return fmt::format("{}.x", temporary); +} + +std::string ARBDecompiler::HPack2(Operation operation) { + const std::string temporary = AllocVectorTemporary(); + AddLine("MOV.U {}.x, {};", temporary, Visit(operation[0])); + AddLine("MOV.U {}.y, {};", temporary, Visit(operation[1])); + AddLine("PK2H.F {}.x, {};", temporary, temporary); + return fmt::format("{}.x", temporary); +} + +std::string ARBDecompiler::LogicalAssign(Operation operation) { + const Node& dest = operation[0]; + const Node& src = operation[1]; + + std::string target; + + if (const auto pred = std::get_if<PredicateNode>(&*dest)) { + ASSERT_MSG(!pred->IsNegated(), "Negating logical assignment"); + + const Tegra::Shader::Pred index = pred->GetIndex(); + switch (index) { + case Tegra::Shader::Pred::NeverExecute: + case Tegra::Shader::Pred::UnusedIndex: + // Writing to these predicates is a no-op + return {}; + } + target = fmt::format("P{}.x", static_cast<u64>(index)); + } else if (const auto internal_flag = std::get_if<InternalFlagNode>(&*dest)) { + const std::size_t index = static_cast<std::size_t>(internal_flag->GetFlag()); + target = fmt::format("{}.x", INTERNAL_FLAG_NAMES[index]); + } else { + UNREACHABLE(); + ResetTemporaries(); + return {}; + } + + AddLine("MOV.U {}, {};", target, Visit(src)); + ResetTemporaries(); + return {}; +} + +std::string ARBDecompiler::LogicalPick2(Operation operation) { + std::string temporary = AllocTemporary(); + const u32 index = std::get<ImmediateNode>(*operation[1]).GetValue(); + AddLine("MOV.U {}, {}.{};", temporary, Visit(operation[0]), Swizzle(index)); + return temporary; +} + +std::string ARBDecompiler::LogicalAnd2(Operation operation) { + std::string temporary = AllocTemporary(); + const std::string op = Visit(operation[0]); + AddLine("AND.U {}, {}.x, {}.y;", temporary, op, op); + return temporary; +} + +std::string ARBDecompiler::FloatOrdered(Operation operation) { + std::string temporary = AllocTemporary(); + AddLine("MOVC.F32 RC.x, {};", Visit(operation[0])); + AddLine("MOVC.F32 RC.y, {};", Visit(operation[1])); + AddLine("MOV.S {}, -1;", temporary); + AddLine("MOV.S {} (NAN.x), 0;", temporary); + AddLine("MOV.S {} (NAN.y), 0;", temporary); + return temporary; +} + +std::string ARBDecompiler::FloatUnordered(Operation operation) { + std::string temporary = AllocTemporary(); + AddLine("MOVC.F32 RC.x, {};", Visit(operation[0])); + AddLine("MOVC.F32 RC.y, {};", Visit(operation[1])); + AddLine("MOV.S {}, 0;", temporary); + AddLine("MOV.S {} (NAN.x), -1;", temporary); + AddLine("MOV.S {} (NAN.y), -1;", temporary); + return temporary; +} + +std::string ARBDecompiler::LogicalAddCarry(Operation operation) { + std::string temporary = AllocTemporary(); + AddLine("ADDC.U RC, {}, {};", Visit(operation[0]), Visit(operation[1])); + AddLine("MOV.S {}, 0;", temporary); + AddLine("IF CF.x;"); + AddLine("MOV.S {}, -1;", temporary); + AddLine("ENDIF;"); + return temporary; +} + +std::string ARBDecompiler::Texture(Operation operation) { + const auto& meta = std::get<MetaTexture>(operation.GetMeta()); + const u32 sampler_id = device.GetBaseBindings(stage).sampler + meta.sampler.index; + const auto [temporary, swizzle] = BuildCoords(operation); + + std::string_view opcode = "TEX"; + std::string extra; + if (meta.bias) { + ASSERT(!meta.lod); + opcode = "TXB"; + + if (swizzle < 4) { + AddLine("MOV.F {}.w, {};", temporary, Visit(meta.bias)); + } else { + const std::string bias = AllocTemporary(); + AddLine("MOV.F {}, {};", bias, Visit(meta.bias)); + extra = fmt::format(" {},", bias); + } + } + if (meta.lod) { + ASSERT(!meta.bias); + opcode = "TXL"; + + if (swizzle < 4) { + AddLine("MOV.F {}.w, {};", temporary, Visit(meta.lod)); + } else { + const std::string lod = AllocTemporary(); + AddLine("MOV.F {}, {};", lod, Visit(meta.lod)); + extra = fmt::format(" {},", lod); + } + } + + AddLine("{}.F {}, {},{} texture[{}], {}{};", opcode, temporary, temporary, extra, sampler_id, + TextureType(meta), BuildAoffi(operation)); + AddLine("MOV.U {}.x, {}.{};", temporary, temporary, Swizzle(meta.element)); + return fmt::format("{}.x", temporary); +} + +std::string ARBDecompiler::TextureGather(Operation operation) { + const auto& meta = std::get<MetaTexture>(operation.GetMeta()); + const u32 sampler_id = device.GetBaseBindings(stage).sampler + meta.sampler.index; + const auto [temporary, swizzle] = BuildCoords(operation); + + std::string comp; + if (!meta.sampler.is_shadow) { + const auto& immediate = std::get<ImmediateNode>(*meta.component); + comp = fmt::format(".{}", Swizzle(immediate.GetValue())); + } + + AddLine("TXG.F {}, {}, texture[{}]{}, {}{};", temporary, temporary, sampler_id, comp, + TextureType(meta), BuildAoffi(operation)); + AddLine("MOV.U {}.x, {}.{};", temporary, temporary, Swizzle(meta.element)); + return fmt::format("{}.x", temporary); +} + +std::string ARBDecompiler::TextureQueryDimensions(Operation operation) { + const auto& meta = std::get<MetaTexture>(operation.GetMeta()); + const std::string temporary = AllocVectorTemporary(); + const u32 sampler_id = device.GetBaseBindings(stage).sampler + meta.sampler.index; + + ASSERT(!meta.sampler.is_array); + + const std::string lod = operation.GetOperandsCount() > 0 ? Visit(operation[0]) : "0"; + AddLine("TXQ {}, {}, texture[{}], {};", temporary, lod, sampler_id, TextureType(meta)); + AddLine("MOV.U {}.x, {}.{};", temporary, temporary, Swizzle(meta.element)); + return fmt::format("{}.x", temporary); +} + +std::string ARBDecompiler::TextureQueryLod(Operation operation) { + const auto& meta = std::get<MetaTexture>(operation.GetMeta()); + const std::string temporary = AllocVectorTemporary(); + const u32 sampler_id = device.GetBaseBindings(stage).sampler + meta.sampler.index; + + ASSERT(!meta.sampler.is_array); + + const std::size_t count = operation.GetOperandsCount(); + for (std::size_t i = 0; i < count; ++i) { + AddLine("MOV.F {}.{}, {};", temporary, Swizzle(i), Visit(operation[i])); + } + AddLine("LOD.F {}, {}, texture[{}], {};", temporary, temporary, sampler_id, TextureType(meta)); + AddLine("MUL.F32 {}, {}, {{256, 256, 0, 0}};", temporary, temporary); + AddLine("TRUNC.S {}, {};", temporary, temporary); + AddLine("MOV.U {}.x, {}.{};", temporary, temporary, Swizzle(meta.element)); + return fmt::format("{}.x", temporary); +} + +std::string ARBDecompiler::TexelFetch(Operation operation) { + const auto& meta = std::get<MetaTexture>(operation.GetMeta()); + const u32 sampler_id = device.GetBaseBindings(stage).sampler + meta.sampler.index; + const auto [temporary, swizzle] = BuildCoords(operation); + + if (!meta.sampler.is_buffer) { + ASSERT(swizzle < 4); + AddLine("MOV.F {}.w, {};", temporary, Visit(meta.lod)); + } + AddLine("TXF.F {}, {}, texture[{}], {}{};", temporary, temporary, sampler_id, TextureType(meta), + BuildAoffi(operation)); + AddLine("MOV.U {}.x, {}.{};", temporary, temporary, Swizzle(meta.element)); + return fmt::format("{}.x", temporary); +} + +std::string ARBDecompiler::TextureGradient(Operation operation) { + const auto& meta = std::get<MetaTexture>(operation.GetMeta()); + const u32 sampler_id = device.GetBaseBindings(stage).sampler + meta.sampler.index; + const std::string ddx = AllocVectorTemporary(); + const std::string ddy = AllocVectorTemporary(); + const std::string coord = BuildCoords(operation).first; + + const std::size_t num_components = meta.derivates.size() / 2; + for (std::size_t index = 0; index < num_components; ++index) { + const char swizzle = Swizzle(index); + AddLine("MOV.F {}.{}, {};", ddx, swizzle, Visit(meta.derivates[index * 2])); + AddLine("MOV.F {}.{}, {};", ddy, swizzle, Visit(meta.derivates[index * 2 + 1])); + } + + const std::string_view result = coord; + AddLine("TXD.F {}, {}, {}, {}, texture[{}], {}{};", result, coord, ddx, ddy, sampler_id, + TextureType(meta), BuildAoffi(operation)); + AddLine("MOV.F {}.x, {}.{};", result, result, Swizzle(meta.element)); + return fmt::format("{}.x", result); +} + +std::string ARBDecompiler::ImageLoad(Operation operation) { + const auto& meta = std::get<MetaImage>(operation.GetMeta()); + const u32 image_id = device.GetBaseBindings(stage).image + meta.image.index; + const std::size_t count = operation.GetOperandsCount(); + const std::string_view type = ImageType(meta.image.type); + + const std::string temporary = AllocVectorTemporary(); + for (std::size_t i = 0; i < count; ++i) { + AddLine("MOV.S {}.{}, {};", temporary, Swizzle(i), Visit(operation[i])); + } + AddLine("LOADIM.F {}, {}, image[{}], {};", temporary, temporary, image_id, type); + AddLine("MOV.F {}.x, {}.{};", temporary, temporary, Swizzle(meta.element)); + return fmt::format("{}.x", temporary); +} + +std::string ARBDecompiler::ImageStore(Operation operation) { + const auto& meta = std::get<MetaImage>(operation.GetMeta()); + const u32 image_id = device.GetBaseBindings(stage).image + meta.image.index; + const std::size_t num_coords = operation.GetOperandsCount(); + const std::size_t num_values = meta.values.size(); + const std::string_view type = ImageType(meta.image.type); + + const std::string coord = AllocVectorTemporary(); + const std::string value = AllocVectorTemporary(); + for (std::size_t i = 0; i < num_coords; ++i) { + AddLine("MOV.S {}.{}, {};", coord, Swizzle(i), Visit(operation[i])); + } + for (std::size_t i = 0; i < num_values; ++i) { + AddLine("MOV.F {}.{}, {};", value, Swizzle(i), Visit(meta.values[i])); + } + AddLine("STOREIM.F image[{}], {}, {}, {};", image_id, value, coord, type); + return {}; +} + +std::string ARBDecompiler::Branch(Operation operation) { + const auto target = std::get<ImmediateNode>(*operation[0]); + AddLine("MOV.U PC.x, {};", target.GetValue()); + AddLine("CONT;"); + return {}; +} + +std::string ARBDecompiler::BranchIndirect(Operation operation) { + AddLine("MOV.U PC.x, {};", Visit(operation[0])); + AddLine("CONT;"); + return {}; +} + +std::string ARBDecompiler::PushFlowStack(Operation operation) { + const auto stack = std::get<MetaStackClass>(operation.GetMeta()); + const u32 target = std::get<ImmediateNode>(*operation[0]).GetValue(); + const std::string_view stack_name = StackName(stack); + AddLine("MOV.U {}[{}_TOP.x].x, {};", stack_name, stack_name, target); + AddLine("ADD.S {}_TOP.x, {}_TOP.x, 1;", stack_name, stack_name); + return {}; +} + +std::string ARBDecompiler::PopFlowStack(Operation operation) { + const auto stack = std::get<MetaStackClass>(operation.GetMeta()); + const std::string_view stack_name = StackName(stack); + AddLine("SUB.S {}_TOP.x, {}_TOP.x, 1;", stack_name, stack_name); + AddLine("MOV.U PC.x, {}[{}_TOP.x].x;", stack_name, stack_name); + AddLine("CONT;"); + return {}; +} + +std::string ARBDecompiler::Exit(Operation) { + Exit(); + return {}; +} + +std::string ARBDecompiler::Discard(Operation) { + AddLine("KIL TR;"); + return {}; +} + +std::string ARBDecompiler::EmitVertex(Operation) { + AddLine("EMIT;"); + return {}; +} + +std::string ARBDecompiler::EndPrimitive(Operation) { + AddLine("ENDPRIM;"); + return {}; +} + +std::string ARBDecompiler::InvocationId(Operation) { + return "primitive.invocation"; +} + +std::string ARBDecompiler::YNegate(Operation) { + LOG_WARNING(Render_OpenGL, "(STUBBED)"); + const std::string temporary = AllocTemporary(); + AddLine("MOV.F {}, 1;", temporary); + return temporary; +} + +std::string ARBDecompiler::ThreadId(Operation) { + return fmt::format("{}.threadid", StageInputName(stage)); +} + +std::string ARBDecompiler::ShuffleIndexed(Operation operation) { + if (!device.HasWarpIntrinsics()) { + LOG_ERROR(Render_OpenGL, + "NV_shader_thread_shuffle is missing. Kepler or better is required."); + return Visit(operation[0]); + } + const std::string temporary = AllocVectorTemporary(); + AddLine("SHFIDX.U {}, {}, {}, {{31, 0, 0, 0}};", temporary, Visit(operation[0]), + Visit(operation[1])); + AddLine("MOV.U {}.x, {}.y;", temporary, temporary); + return fmt::format("{}.x", temporary); +} + +std::string ARBDecompiler::Barrier(Operation) { + if (!ir.IsDecompiled()) { + LOG_ERROR(Render_OpenGL, "BAR used but shader is not decompiled"); + return {}; + } + AddLine("BAR;"); + return {}; +} + +std::string ARBDecompiler::MemoryBarrierGroup(Operation) { + AddLine("MEMBAR.CTA;"); + return {}; +} + +std::string ARBDecompiler::MemoryBarrierGlobal(Operation) { + AddLine("MEMBAR;"); + return {}; +} + +} // Anonymous namespace + +std::string DecompileAssemblyShader(const Device& device, const VideoCommon::Shader::ShaderIR& ir, + const VideoCommon::Shader::Registry& registry, + Tegra::Engines::ShaderType stage, std::string_view identifier) { + return ARBDecompiler(device, ir, registry, stage, identifier).Code(); +} + +} // namespace OpenGL diff --git a/src/video_core/renderer_opengl/gl_arb_decompiler.h b/src/video_core/renderer_opengl/gl_arb_decompiler.h new file mode 100644 index 000000000..6afc87220 --- /dev/null +++ b/src/video_core/renderer_opengl/gl_arb_decompiler.h @@ -0,0 +1,29 @@ +// Copyright 2020 yuzu Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include <string> +#include <string_view> + +#include "common/common_types.h" + +namespace Tegra::Engines { +enum class ShaderType : u32; +} + +namespace VideoCommon::Shader { +class ShaderIR; +class Registry; +} // namespace VideoCommon::Shader + +namespace OpenGL { + +class Device; + +std::string DecompileAssemblyShader(const Device& device, const VideoCommon::Shader::ShaderIR& ir, + const VideoCommon::Shader::Registry& registry, + Tegra::Engines::ShaderType stage, std::string_view identifier); + +} // namespace OpenGL diff --git a/src/video_core/renderer_opengl/gl_buffer_cache.cpp b/src/video_core/renderer_opengl/gl_buffer_cache.cpp index d2cab50bd..d9f7b4cc6 100644 --- a/src/video_core/renderer_opengl/gl_buffer_cache.cpp +++ b/src/video_core/renderer_opengl/gl_buffer_cache.cpp @@ -8,6 +8,7 @@ #include "common/assert.h" #include "common/microprofile.h" +#include "video_core/buffer_cache/buffer_cache.h" #include "video_core/engines/maxwell_3d.h" #include "video_core/rasterizer_interface.h" #include "video_core/renderer_opengl/gl_buffer_cache.h" @@ -21,22 +22,46 @@ using Maxwell = Tegra::Engines::Maxwell3D::Regs; MICROPROFILE_DEFINE(OpenGL_Buffer_Download, "OpenGL", "Buffer Download", MP_RGB(192, 192, 128)); -CachedBufferBlock::CachedBufferBlock(VAddr cpu_addr, const std::size_t size) +Buffer::Buffer(const Device& device, VAddr cpu_addr, std::size_t size) : VideoCommon::BufferBlock{cpu_addr, size} { gl_buffer.Create(); glNamedBufferData(gl_buffer.handle, static_cast<GLsizeiptr>(size), nullptr, GL_DYNAMIC_DRAW); + if (device.HasVertexBufferUnifiedMemory()) { + glMakeNamedBufferResidentNV(gl_buffer.handle, GL_READ_WRITE); + glGetNamedBufferParameterui64vNV(gl_buffer.handle, GL_BUFFER_GPU_ADDRESS_NV, &gpu_address); + } } -CachedBufferBlock::~CachedBufferBlock() = default; +Buffer::~Buffer() = default; + +void Buffer::Upload(std::size_t offset, std::size_t size, const u8* data) const { + glNamedBufferSubData(Handle(), static_cast<GLintptr>(offset), static_cast<GLsizeiptr>(size), + data); +} + +void Buffer::Download(std::size_t offset, std::size_t size, u8* data) const { + MICROPROFILE_SCOPE(OpenGL_Buffer_Download); + glMemoryBarrier(GL_BUFFER_UPDATE_BARRIER_BIT); + glGetNamedBufferSubData(Handle(), static_cast<GLintptr>(offset), static_cast<GLsizeiptr>(size), + data); +} + +void Buffer::CopyFrom(const Buffer& src, std::size_t src_offset, std::size_t dst_offset, + std::size_t size) const { + glCopyNamedBufferSubData(src.Handle(), Handle(), static_cast<GLintptr>(src_offset), + static_cast<GLintptr>(dst_offset), static_cast<GLsizeiptr>(size)); +} OGLBufferCache::OGLBufferCache(RasterizerOpenGL& rasterizer, Core::System& system, - const Device& device, std::size_t stream_size) - : GenericBufferCache{rasterizer, system, std::make_unique<OGLStreamBuffer>(stream_size, true)} { + const Device& device_, std::size_t stream_size) + : GenericBufferCache{rasterizer, system, + std::make_unique<OGLStreamBuffer>(device_, stream_size, true)}, + device{device_} { if (!device.HasFastBufferSubData()) { return; } - static constexpr auto size = static_cast<GLsizeiptr>(Maxwell::MaxConstBufferSize); + static constexpr GLsizeiptr size = static_cast<GLsizeiptr>(Maxwell::MaxConstBufferSize); glCreateBuffers(static_cast<GLsizei>(std::size(cbufs)), std::data(cbufs)); for (const GLuint cbuf : cbufs) { glNamedBufferData(cbuf, size, nullptr, GL_STREAM_DRAW); @@ -47,44 +72,21 @@ OGLBufferCache::~OGLBufferCache() { glDeleteBuffers(static_cast<GLsizei>(std::size(cbufs)), std::data(cbufs)); } -Buffer OGLBufferCache::CreateBlock(VAddr cpu_addr, std::size_t size) { - return std::make_shared<CachedBufferBlock>(cpu_addr, size); -} - -GLuint OGLBufferCache::ToHandle(const Buffer& buffer) { - return buffer->GetHandle(); -} - -GLuint OGLBufferCache::GetEmptyBuffer(std::size_t) { - return 0; +std::shared_ptr<Buffer> OGLBufferCache::CreateBlock(VAddr cpu_addr, std::size_t size) { + return std::make_shared<Buffer>(device, cpu_addr, size); } -void OGLBufferCache::UploadBlockData(const Buffer& buffer, std::size_t offset, std::size_t size, - const u8* data) { - glNamedBufferSubData(buffer->GetHandle(), static_cast<GLintptr>(offset), - static_cast<GLsizeiptr>(size), data); -} - -void OGLBufferCache::DownloadBlockData(const Buffer& buffer, std::size_t offset, std::size_t size, - u8* data) { - MICROPROFILE_SCOPE(OpenGL_Buffer_Download); - glMemoryBarrier(GL_BUFFER_UPDATE_BARRIER_BIT); - glGetNamedBufferSubData(buffer->GetHandle(), static_cast<GLintptr>(offset), - static_cast<GLsizeiptr>(size), data); -} - -void OGLBufferCache::CopyBlock(const Buffer& src, const Buffer& dst, std::size_t src_offset, - std::size_t dst_offset, std::size_t size) { - glCopyNamedBufferSubData(src->GetHandle(), dst->GetHandle(), static_cast<GLintptr>(src_offset), - static_cast<GLintptr>(dst_offset), static_cast<GLsizeiptr>(size)); +OGLBufferCache::BufferInfo OGLBufferCache::GetEmptyBuffer(std::size_t) { + return {0, 0, 0}; } OGLBufferCache::BufferInfo OGLBufferCache::ConstBufferUpload(const void* raw_pointer, std::size_t size) { DEBUG_ASSERT(cbuf_cursor < std::size(cbufs)); - const GLuint& cbuf = cbufs[cbuf_cursor++]; + const GLuint cbuf = cbufs[cbuf_cursor++]; + glNamedBufferSubData(cbuf, 0, static_cast<GLsizeiptr>(size), raw_pointer); - return {cbuf, 0}; + return {cbuf, 0, 0}; } } // namespace OpenGL diff --git a/src/video_core/renderer_opengl/gl_buffer_cache.h b/src/video_core/renderer_opengl/gl_buffer_cache.h index a9e86cfc7..59d95adbc 100644 --- a/src/video_core/renderer_opengl/gl_buffer_cache.h +++ b/src/video_core/renderer_opengl/gl_buffer_cache.h @@ -10,7 +10,6 @@ #include "common/common_types.h" #include "video_core/buffer_cache/buffer_cache.h" #include "video_core/engines/maxwell_3d.h" -#include "video_core/rasterizer_cache.h" #include "video_core/renderer_opengl/gl_resource_manager.h" #include "video_core/renderer_opengl/gl_stream_buffer.h" @@ -24,57 +23,57 @@ class Device; class OGLStreamBuffer; class RasterizerOpenGL; -class CachedBufferBlock; +class Buffer : public VideoCommon::BufferBlock { +public: + explicit Buffer(const Device& device, VAddr cpu_addr, std::size_t size); + ~Buffer(); -using Buffer = std::shared_ptr<CachedBufferBlock>; -using GenericBufferCache = VideoCommon::BufferCache<Buffer, GLuint, OGLStreamBuffer>; + void Upload(std::size_t offset, std::size_t size, const u8* data) const; -class CachedBufferBlock : public VideoCommon::BufferBlock { -public: - explicit CachedBufferBlock(VAddr cpu_addr, const std::size_t size); - ~CachedBufferBlock(); + void Download(std::size_t offset, std::size_t size, u8* data) const; + + void CopyFrom(const Buffer& src, std::size_t src_offset, std::size_t dst_offset, + std::size_t size) const; - GLuint GetHandle() const { + GLuint Handle() const noexcept { return gl_buffer.handle; } + u64 Address() const noexcept { + return gpu_address; + } + private: OGLBuffer gl_buffer; + u64 gpu_address = 0; }; +using GenericBufferCache = VideoCommon::BufferCache<Buffer, GLuint, OGLStreamBuffer>; class OGLBufferCache final : public GenericBufferCache { public: explicit OGLBufferCache(RasterizerOpenGL& rasterizer, Core::System& system, const Device& device, std::size_t stream_size); ~OGLBufferCache(); - GLuint GetEmptyBuffer(std::size_t) override; + BufferInfo GetEmptyBuffer(std::size_t) override; void Acquire() noexcept { cbuf_cursor = 0; } protected: - Buffer CreateBlock(VAddr cpu_addr, std::size_t size) override; - - GLuint ToHandle(const Buffer& buffer) override; - - void UploadBlockData(const Buffer& buffer, std::size_t offset, std::size_t size, - const u8* data) override; - - void DownloadBlockData(const Buffer& buffer, std::size_t offset, std::size_t size, - u8* data) override; - - void CopyBlock(const Buffer& src, const Buffer& dst, std::size_t src_offset, - std::size_t dst_offset, std::size_t size) override; + std::shared_ptr<Buffer> CreateBlock(VAddr cpu_addr, std::size_t size) override; BufferInfo ConstBufferUpload(const void* raw_pointer, std::size_t size) override; private: + static constexpr std::size_t NUM_CBUFS = Tegra::Engines::Maxwell3D::Regs::MaxConstBuffers * + Tegra::Engines::Maxwell3D::Regs::MaxShaderProgram; + + const Device& device; + std::size_t cbuf_cursor = 0; - std::array<GLuint, Tegra::Engines::Maxwell3D::Regs::MaxConstBuffers * - Tegra::Engines::Maxwell3D::Regs::MaxShaderProgram> - cbufs; + std::array<GLuint, NUM_CBUFS> cbufs{}; }; } // namespace OpenGL diff --git a/src/video_core/renderer_opengl/gl_device.cpp b/src/video_core/renderer_opengl/gl_device.cpp index d83dca25a..208fc6167 100644 --- a/src/video_core/renderer_opengl/gl_device.cpp +++ b/src/video_core/renderer_opengl/gl_device.cpp @@ -6,6 +6,7 @@ #include <array> #include <cstddef> #include <cstring> +#include <limits> #include <optional> #include <vector> @@ -13,6 +14,7 @@ #include "common/logging/log.h" #include "common/scope_exit.h" +#include "core/settings.h" #include "video_core/renderer_opengl/gl_device.h" #include "video_core/renderer_opengl/gl_resource_manager.h" @@ -25,24 +27,27 @@ constexpr u32 ReservedUniformBlocks = 1; constexpr u32 NumStages = 5; -constexpr std::array LimitUBOs = {GL_MAX_VERTEX_UNIFORM_BLOCKS, GL_MAX_TESS_CONTROL_UNIFORM_BLOCKS, - GL_MAX_TESS_EVALUATION_UNIFORM_BLOCKS, - GL_MAX_GEOMETRY_UNIFORM_BLOCKS, GL_MAX_FRAGMENT_UNIFORM_BLOCKS}; +constexpr std::array LimitUBOs = { + GL_MAX_VERTEX_UNIFORM_BLOCKS, GL_MAX_TESS_CONTROL_UNIFORM_BLOCKS, + GL_MAX_TESS_EVALUATION_UNIFORM_BLOCKS, GL_MAX_GEOMETRY_UNIFORM_BLOCKS, + GL_MAX_FRAGMENT_UNIFORM_BLOCKS, GL_MAX_COMPUTE_UNIFORM_BLOCKS}; constexpr std::array LimitSSBOs = { - GL_MAX_VERTEX_SHADER_STORAGE_BLOCKS, GL_MAX_TESS_CONTROL_SHADER_STORAGE_BLOCKS, + GL_MAX_VERTEX_SHADER_STORAGE_BLOCKS, GL_MAX_TESS_CONTROL_SHADER_STORAGE_BLOCKS, GL_MAX_TESS_EVALUATION_SHADER_STORAGE_BLOCKS, GL_MAX_GEOMETRY_SHADER_STORAGE_BLOCKS, - GL_MAX_FRAGMENT_SHADER_STORAGE_BLOCKS}; + GL_MAX_FRAGMENT_SHADER_STORAGE_BLOCKS, GL_MAX_COMPUTE_SHADER_STORAGE_BLOCKS}; -constexpr std::array LimitSamplers = { - GL_MAX_VERTEX_TEXTURE_IMAGE_UNITS, GL_MAX_TESS_CONTROL_TEXTURE_IMAGE_UNITS, - GL_MAX_TESS_EVALUATION_TEXTURE_IMAGE_UNITS, GL_MAX_GEOMETRY_TEXTURE_IMAGE_UNITS, - GL_MAX_TEXTURE_IMAGE_UNITS}; +constexpr std::array LimitSamplers = {GL_MAX_VERTEX_TEXTURE_IMAGE_UNITS, + GL_MAX_TESS_CONTROL_TEXTURE_IMAGE_UNITS, + GL_MAX_TESS_EVALUATION_TEXTURE_IMAGE_UNITS, + GL_MAX_GEOMETRY_TEXTURE_IMAGE_UNITS, + GL_MAX_TEXTURE_IMAGE_UNITS, + GL_MAX_COMPUTE_TEXTURE_IMAGE_UNITS}; -constexpr std::array LimitImages = {GL_MAX_VERTEX_IMAGE_UNIFORMS, - GL_MAX_TESS_CONTROL_IMAGE_UNIFORMS, - GL_MAX_TESS_EVALUATION_IMAGE_UNIFORMS, - GL_MAX_GEOMETRY_IMAGE_UNIFORMS, GL_MAX_FRAGMENT_IMAGE_UNIFORMS}; +constexpr std::array LimitImages = { + GL_MAX_VERTEX_IMAGE_UNIFORMS, GL_MAX_TESS_CONTROL_IMAGE_UNIFORMS, + GL_MAX_TESS_EVALUATION_IMAGE_UNIFORMS, GL_MAX_GEOMETRY_IMAGE_UNIFORMS, + GL_MAX_FRAGMENT_IMAGE_UNIFORMS, GL_MAX_COMPUTE_IMAGE_UNIFORMS}; template <typename T> T GetInteger(GLenum pname) { @@ -84,6 +89,13 @@ u32 Extract(u32& base, u32& num, u32 amount, std::optional<GLenum> limit = {}) { return std::exchange(base, base + amount); } +std::array<u32, Tegra::Engines::MaxShaderTypes> BuildMaxUniformBuffers() noexcept { + std::array<u32, Tegra::Engines::MaxShaderTypes> max; + std::transform(LimitUBOs.begin(), LimitUBOs.end(), max.begin(), + [](GLenum pname) { return GetInteger<u32>(pname); }); + return max; +} + std::array<Device::BaseBindings, Tegra::Engines::MaxShaderTypes> BuildBaseBindings() noexcept { std::array<Device::BaseBindings, Tegra::Engines::MaxShaderTypes> bindings; @@ -111,16 +123,24 @@ std::array<Device::BaseBindings, Tegra::Engines::MaxShaderTypes> BuildBaseBindin u32 num_images = GetInteger<u32>(GL_MAX_IMAGE_UNITS); u32 base_images = 0; - // Reserve more image bindings on fragment and vertex stages. + // GL_MAX_IMAGE_UNITS is guaranteed by the spec to have a minimum value of 8. + // Due to the limitation of GL_MAX_IMAGE_UNITS, reserve at least 4 image bindings on the + // fragment stage, and at least 1 for the rest of the stages. + // So far games are observed to use 1 image binding on vertex and 4 on fragment stages. + + // Reserve at least 4 image bindings on the fragment stage. bindings[4].image = - Extract(base_images, num_images, num_images / NumStages + 2, LimitImages[4]); - bindings[0].image = - Extract(base_images, num_images, num_images / NumStages + 1, LimitImages[0]); + Extract(base_images, num_images, std::max(4U, num_images / NumStages), LimitImages[4]); + + // This is guaranteed to be at least 1. + const u32 total_extracted_images = num_images / (NumStages - 1); // Reserve the other image bindings. - const u32 total_extracted_images = num_images / (NumStages - 2); - for (std::size_t i = 2; i < NumStages; ++i) { + for (std::size_t i = 0; i < NumStages; ++i) { const std::size_t stage = stage_swizzle[i]; + if (stage == 4) { + continue; + } bindings[stage].image = Extract(base_images, num_images, total_extracted_images, LimitImages[stage]); } @@ -132,6 +152,7 @@ std::array<Device::BaseBindings, Tegra::Engines::MaxShaderTypes> BuildBaseBindin } bool IsASTCSupported() { + static constexpr std::array targets = {GL_TEXTURE_2D, GL_TEXTURE_2D_ARRAY}; static constexpr std::array formats = { GL_COMPRESSED_RGBA_ASTC_4x4_KHR, GL_COMPRESSED_RGBA_ASTC_5x4_KHR, GL_COMPRESSED_RGBA_ASTC_5x5_KHR, GL_COMPRESSED_RGBA_ASTC_6x5_KHR, @@ -148,25 +169,44 @@ bool IsASTCSupported() { GL_COMPRESSED_SRGB8_ALPHA8_ASTC_10x8_KHR, GL_COMPRESSED_SRGB8_ALPHA8_ASTC_10x10_KHR, GL_COMPRESSED_SRGB8_ALPHA8_ASTC_12x10_KHR, GL_COMPRESSED_SRGB8_ALPHA8_ASTC_12x12_KHR, }; - return std::find_if_not(formats.begin(), formats.end(), [](GLenum format) { - GLint supported; - glGetInternalformativ(GL_TEXTURE_2D, format, GL_INTERNALFORMAT_SUPPORTED, 1, - &supported); - return supported == GL_TRUE; - }) == formats.end(); + static constexpr std::array required_support = { + GL_VERTEX_TEXTURE, GL_TESS_CONTROL_TEXTURE, GL_TESS_EVALUATION_TEXTURE, + GL_GEOMETRY_TEXTURE, GL_FRAGMENT_TEXTURE, GL_COMPUTE_TEXTURE, + }; + + for (const GLenum target : targets) { + for (const GLenum format : formats) { + for (const GLenum support : required_support) { + GLint value; + glGetInternalformativ(target, format, support, 1, &value); + if (value != GL_FULL_SUPPORT) { + return false; + } + } + } + } + return true; } } // Anonymous namespace -Device::Device() : base_bindings{BuildBaseBindings()} { +Device::Device() + : max_uniform_buffers{BuildMaxUniformBuffers()}, base_bindings{BuildBaseBindings()} { const std::string_view vendor = reinterpret_cast<const char*>(glGetString(GL_VENDOR)); - const auto renderer = reinterpret_cast<const char*>(glGetString(GL_RENDERER)); + const std::string_view renderer = reinterpret_cast<const char*>(glGetString(GL_RENDERER)); + const std::string_view version = reinterpret_cast<const char*>(glGetString(GL_VERSION)); const std::vector extensions = GetExtensions(); const bool is_nvidia = vendor == "NVIDIA Corporation"; const bool is_amd = vendor == "ATI Technologies Inc."; - const bool is_intel = vendor == "Intel"; - const bool is_intel_proprietary = is_intel && std::strstr(renderer, "Mesa") == nullptr; + + bool disable_fast_buffer_sub_data = false; + if (is_nvidia && version == "4.6.0 NVIDIA 443.24") { + LOG_WARNING( + Render_OpenGL, + "Beta driver 443.24 is known to have issues. There might be performance issues."); + disable_fast_buffer_sub_data = true; + } uniform_buffer_alignment = GetInteger<std::size_t>(GL_UNIFORM_BUFFER_OFFSET_ALIGNMENT); shader_storage_alignment = GetInteger<std::size_t>(GL_SHADER_STORAGE_BUFFER_OFFSET_ALIGNMENT); @@ -177,30 +217,43 @@ Device::Device() : base_bindings{BuildBaseBindings()} { has_shader_ballot = GLAD_GL_ARB_shader_ballot; has_vertex_viewport_layer = GLAD_GL_ARB_shader_viewport_layer_array; has_image_load_formatted = HasExtension(extensions, "GL_EXT_shader_image_load_formatted"); + has_texture_shadow_lod = HasExtension(extensions, "GL_EXT_texture_shadow_lod"); has_astc = IsASTCSupported(); has_variable_aoffi = TestVariableAoffi(); has_component_indexing_bug = is_amd; has_precise_bug = TestPreciseBug(); - has_broken_compute = is_intel_proprietary; - has_fast_buffer_sub_data = is_nvidia; + has_nv_viewport_array2 = GLAD_GL_NV_viewport_array2; + has_vertex_buffer_unified_memory = GLAD_GL_NV_vertex_buffer_unified_memory; + + // At the moment of writing this, only Nvidia's driver optimizes BufferSubData on exclusive + // uniform buffers as "push constants" + has_fast_buffer_sub_data = is_nvidia && !disable_fast_buffer_sub_data; + + use_assembly_shaders = Settings::values.use_assembly_shaders && GLAD_GL_NV_gpu_program5 && + GLAD_GL_NV_compute_program5 && GLAD_GL_NV_transform_feedback && + GLAD_GL_NV_transform_feedback2; LOG_INFO(Render_OpenGL, "Renderer_VariableAOFFI: {}", has_variable_aoffi); LOG_INFO(Render_OpenGL, "Renderer_ComponentIndexingBug: {}", has_component_indexing_bug); LOG_INFO(Render_OpenGL, "Renderer_PreciseBug: {}", has_precise_bug); + + if (Settings::values.use_assembly_shaders && !use_assembly_shaders) { + LOG_ERROR(Render_OpenGL, "Assembly shaders enabled but not supported"); + } } Device::Device(std::nullptr_t) { - uniform_buffer_alignment = 0; + max_uniform_buffers.fill(std::numeric_limits<u32>::max()); + uniform_buffer_alignment = 4; + shader_storage_alignment = 4; max_vertex_attributes = 16; max_varyings = 15; has_warp_intrinsics = true; has_shader_ballot = true; has_vertex_viewport_layer = true; has_image_load_formatted = true; + has_texture_shadow_lod = true; has_variable_aoffi = true; - has_component_indexing_bug = false; - has_broken_compute = false; - has_precise_bug = false; } bool Device::TestVariableAoffi() { diff --git a/src/video_core/renderer_opengl/gl_device.h b/src/video_core/renderer_opengl/gl_device.h index a55050cb5..e1d811966 100644 --- a/src/video_core/renderer_opengl/gl_device.h +++ b/src/video_core/renderer_opengl/gl_device.h @@ -24,6 +24,10 @@ public: explicit Device(); explicit Device(std::nullptr_t); + u32 GetMaxUniformBuffers(Tegra::Engines::ShaderType shader_type) const noexcept { + return max_uniform_buffers[static_cast<std::size_t>(shader_type)]; + } + const BaseBindings& GetBaseBindings(std::size_t stage_index) const noexcept { return base_bindings[stage_index]; } @@ -64,6 +68,14 @@ public: return has_image_load_formatted; } + bool HasTextureShadowLod() const { + return has_texture_shadow_lod; + } + + bool HasVertexBufferUnifiedMemory() const { + return has_vertex_buffer_unified_memory; + } + bool HasASTC() const { return has_astc; } @@ -80,19 +92,24 @@ public: return has_precise_bug; } - bool HasBrokenCompute() const { - return has_broken_compute; - } - bool HasFastBufferSubData() const { return has_fast_buffer_sub_data; } + bool HasNvViewportArray2() const { + return has_nv_viewport_array2; + } + + bool UseAssemblyShaders() const { + return use_assembly_shaders; + } + private: static bool TestVariableAoffi(); static bool TestPreciseBug(); - std::array<BaseBindings, Tegra::Engines::MaxShaderTypes> base_bindings; + std::array<u32, Tegra::Engines::MaxShaderTypes> max_uniform_buffers{}; + std::array<BaseBindings, Tegra::Engines::MaxShaderTypes> base_bindings{}; std::size_t uniform_buffer_alignment{}; std::size_t shader_storage_alignment{}; u32 max_vertex_attributes{}; @@ -101,12 +118,15 @@ private: bool has_shader_ballot{}; bool has_vertex_viewport_layer{}; bool has_image_load_formatted{}; + bool has_texture_shadow_lod{}; + bool has_vertex_buffer_unified_memory{}; bool has_astc{}; bool has_variable_aoffi{}; bool has_component_indexing_bug{}; bool has_precise_bug{}; - bool has_broken_compute{}; bool has_fast_buffer_sub_data{}; + bool has_nv_viewport_array2{}; + bool use_assembly_shaders{}; }; } // namespace OpenGL diff --git a/src/video_core/renderer_opengl/gl_fence_manager.cpp b/src/video_core/renderer_opengl/gl_fence_manager.cpp index 99ddcb3f8..ec5421afa 100644 --- a/src/video_core/renderer_opengl/gl_fence_manager.cpp +++ b/src/video_core/renderer_opengl/gl_fence_manager.cpp @@ -4,6 +4,7 @@ #include "common/assert.h" +#include "video_core/renderer_opengl/gl_buffer_cache.h" #include "video_core/renderer_opengl/gl_fence_manager.h" namespace OpenGL { diff --git a/src/video_core/renderer_opengl/gl_rasterizer.cpp b/src/video_core/renderer_opengl/gl_rasterizer.cpp index 8b3b3ce92..e960a0ef1 100644 --- a/src/video_core/renderer_opengl/gl_rasterizer.cpp +++ b/src/video_core/renderer_opengl/gl_rasterizer.cpp @@ -30,6 +30,7 @@ #include "video_core/renderer_opengl/gl_shader_cache.h" #include "video_core/renderer_opengl/maxwell_to_gl.h" #include "video_core/renderer_opengl/renderer_opengl.h" +#include "video_core/shader_cache.h" namespace OpenGL { @@ -54,15 +55,34 @@ MICROPROFILE_DEFINE(OpenGL_PrimitiveAssembly, "OpenGL", "Prim Asmbl", MP_RGB(255 namespace { -constexpr std::size_t NumSupportedVertexAttributes = 16; +constexpr std::size_t NUM_CONST_BUFFERS_PER_STAGE = 18; +constexpr std::size_t NUM_CONST_BUFFERS_BYTES_PER_STAGE = + NUM_CONST_BUFFERS_PER_STAGE * Maxwell::MaxConstBufferSize; +constexpr std::size_t TOTAL_CONST_BUFFER_BYTES = + NUM_CONST_BUFFERS_BYTES_PER_STAGE * Maxwell::MaxShaderStage; + +constexpr std::size_t NUM_SUPPORTED_VERTEX_ATTRIBUTES = 16; +constexpr std::size_t NUM_SUPPORTED_VERTEX_BINDINGS = 16; template <typename Engine, typename Entry> Tegra::Texture::FullTextureInfo GetTextureInfo(const Engine& engine, const Entry& entry, ShaderType shader_type, std::size_t index = 0) { + if constexpr (std::is_same_v<Entry, SamplerEntry>) { + if (entry.is_separated) { + const u32 buffer_1 = entry.buffer; + const u32 buffer_2 = entry.secondary_buffer; + const u32 offset_1 = entry.offset; + const u32 offset_2 = entry.secondary_offset; + const u32 handle_1 = engine.AccessConstBuffer32(shader_type, buffer_1, offset_1); + const u32 handle_2 = engine.AccessConstBuffer32(shader_type, buffer_2, offset_2); + return engine.GetTextureInfo(handle_1 | handle_2); + } + } if (entry.is_bindless) { - const auto tex_handle = engine.AccessConstBuffer32(shader_type, entry.buffer, entry.offset); - return engine.GetTextureInfo(tex_handle); + const u32 handle = engine.AccessConstBuffer32(shader_type, entry.buffer, entry.offset); + return engine.GetTextureInfo(handle); } + const auto& gpu_profile = engine.AccessGuestDriverProfile(); const u32 offset = entry.offset + static_cast<u32>(index * gpu_profile.GetTextureHandlerSize()); if constexpr (std::is_same_v<Engine, Tegra::Engines::Maxwell3D>) { @@ -87,6 +107,34 @@ std::size_t GetConstBufferSize(const Tegra::Engines::ConstBufferInfo& buffer, return buffer.size; } +/// Translates hardware transform feedback indices +/// @param location Hardware location +/// @return Pair of ARB_transform_feedback3 token stream first and third arguments +/// @note Read https://www.khronos.org/registry/OpenGL/extensions/ARB/ARB_transform_feedback3.txt +std::pair<GLint, GLint> TransformFeedbackEnum(u8 location) { + const u8 index = location / 4; + if (index >= 8 && index <= 39) { + return {GL_GENERIC_ATTRIB_NV, index - 8}; + } + if (index >= 48 && index <= 55) { + return {GL_TEXTURE_COORD_NV, index - 48}; + } + switch (index) { + case 7: + return {GL_POSITION, 0}; + case 40: + return {GL_PRIMARY_COLOR_NV, 0}; + case 41: + return {GL_SECONDARY_COLOR_NV, 0}; + case 42: + return {GL_BACK_PRIMARY_COLOR_NV, 0}; + case 43: + return {GL_BACK_SECONDARY_COLOR_NV, 0}; + } + UNIMPLEMENTED_MSG("index={}", static_cast<int>(index)); + return {GL_POSITION, 0}; +} + void oglEnable(GLenum cap, bool state) { (state ? glEnable : glDisable)(cap); } @@ -94,17 +142,33 @@ void oglEnable(GLenum cap, bool state) { } // Anonymous namespace RasterizerOpenGL::RasterizerOpenGL(Core::System& system, Core::Frontend::EmuWindow& emu_window, - ScreenInfo& info, GLShader::ProgramManager& program_manager, - StateTracker& state_tracker) - : RasterizerAccelerated{system.Memory()}, texture_cache{system, *this, device, state_tracker}, + const Device& device, ScreenInfo& info, + ProgramManager& program_manager, StateTracker& state_tracker) + : RasterizerAccelerated{system.Memory()}, device{device}, texture_cache{system, *this, device, + state_tracker}, shader_cache{*this, system, emu_window, device}, query_cache{system, *this}, buffer_cache{*this, system, device, STREAM_BUFFER_SIZE}, fence_manager{system, *this, texture_cache, buffer_cache, query_cache}, system{system}, screen_info{info}, program_manager{program_manager}, state_tracker{state_tracker} { CheckExtensions(); + + unified_uniform_buffer.Create(); + glNamedBufferStorage(unified_uniform_buffer.handle, TOTAL_CONST_BUFFER_BYTES, nullptr, 0); + + if (device.UseAssemblyShaders()) { + glCreateBuffers(static_cast<GLsizei>(staging_cbufs.size()), staging_cbufs.data()); + for (const GLuint cbuf : staging_cbufs) { + glNamedBufferStorage(cbuf, static_cast<GLsizeiptr>(Maxwell::MaxConstBufferSize), + nullptr, 0); + } + } } -RasterizerOpenGL::~RasterizerOpenGL() {} +RasterizerOpenGL::~RasterizerOpenGL() { + if (device.UseAssemblyShaders()) { + glDeleteBuffers(static_cast<GLsizei>(staging_cbufs.size()), staging_cbufs.data()); + } +} void RasterizerOpenGL::CheckExtensions() { if (!GLAD_GL_ARB_texture_filter_anisotropic && !GLAD_GL_EXT_texture_filter_anisotropic) { @@ -130,7 +194,7 @@ void RasterizerOpenGL::SetupVertexFormat() { // avoid OpenGL errors. // TODO(Subv): Analyze the shader to identify which attributes are actually used and don't // assume every shader uses them all. - for (std::size_t index = 0; index < NumSupportedVertexAttributes; ++index) { + for (std::size_t index = 0; index < NUM_SUPPORTED_VERTEX_ATTRIBUTES; ++index) { if (!flags[Dirty::VertexFormat0 + index]) { continue; } @@ -149,9 +213,10 @@ void RasterizerOpenGL::SetupVertexFormat() { if (attrib.type == Maxwell::VertexAttribute::Type::SignedInt || attrib.type == Maxwell::VertexAttribute::Type::UnsignedInt) { glVertexAttribIFormat(gl_index, attrib.ComponentCount(), - MaxwellToGL::VertexType(attrib), attrib.offset); + MaxwellToGL::VertexFormat(attrib), attrib.offset); } else { - glVertexAttribFormat(gl_index, attrib.ComponentCount(), MaxwellToGL::VertexType(attrib), + glVertexAttribFormat(gl_index, attrib.ComponentCount(), + MaxwellToGL::VertexFormat(attrib), attrib.IsNormalized() ? GL_TRUE : GL_FALSE, attrib.offset); } glVertexAttribBinding(gl_index, attrib.buffer); @@ -168,9 +233,11 @@ void RasterizerOpenGL::SetupVertexBuffer() { MICROPROFILE_SCOPE(OpenGL_VB); + const bool use_unified_memory = device.HasVertexBufferUnifiedMemory(); + // Upload all guest vertex arrays sequentially to our buffer const auto& regs = gpu.regs; - for (std::size_t index = 0; index < Maxwell::NumVertexArrays; ++index) { + for (std::size_t index = 0; index < NUM_SUPPORTED_VERTEX_BINDINGS; ++index) { if (!flags[Dirty::VertexBuffer0 + index]) { continue; } @@ -183,16 +250,25 @@ void RasterizerOpenGL::SetupVertexBuffer() { const GPUVAddr start = vertex_array.StartAddress(); const GPUVAddr end = regs.vertex_array_limit[index].LimitAddress(); - ASSERT(end >= start); + + const GLuint gl_index = static_cast<GLuint>(index); const u64 size = end - start; if (size == 0) { - glBindVertexBuffer(static_cast<GLuint>(index), 0, 0, vertex_array.stride); + glBindVertexBuffer(gl_index, 0, 0, vertex_array.stride); + if (use_unified_memory) { + glBufferAddressRangeNV(GL_VERTEX_ATTRIB_ARRAY_ADDRESS_NV, gl_index, 0, 0); + } continue; } - const auto [vertex_buffer, vertex_buffer_offset] = buffer_cache.UploadMemory(start, size); - glBindVertexBuffer(static_cast<GLuint>(index), vertex_buffer, vertex_buffer_offset, - vertex_array.stride); + const auto info = buffer_cache.UploadMemory(start, size); + if (use_unified_memory) { + glBindVertexBuffer(gl_index, 0, 0, vertex_array.stride); + glBufferAddressRangeNV(GL_VERTEX_ATTRIB_ARRAY_ADDRESS_NV, gl_index, + info.address + info.offset, size); + } else { + glBindVertexBuffer(gl_index, info.handle, info.offset, vertex_array.stride); + } } } @@ -205,7 +281,7 @@ void RasterizerOpenGL::SetupVertexInstances() { flags[Dirty::VertexInstances] = false; const auto& regs = gpu.regs; - for (std::size_t index = 0; index < NumSupportedVertexAttributes; ++index) { + for (std::size_t index = 0; index < NUM_SUPPORTED_VERTEX_ATTRIBUTES; ++index) { if (!flags[Dirty::VertexInstance0 + index]) { continue; } @@ -222,14 +298,15 @@ GLintptr RasterizerOpenGL::SetupIndexBuffer() { MICROPROFILE_SCOPE(OpenGL_Index); const auto& regs = system.GPU().Maxwell3D().regs; const std::size_t size = CalculateIndexBufferSize(); - const auto [buffer, offset] = buffer_cache.UploadMemory(regs.index_array.IndexStart(), size); - glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, buffer); - return offset; + const auto info = buffer_cache.UploadMemory(regs.index_array.IndexStart(), size); + glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, info.handle); + return info.offset; } void RasterizerOpenGL::SetupShaders(GLenum primitive_mode) { MICROPROFILE_SCOPE(OpenGL_Shader); auto& gpu = system.GPU().Maxwell3D(); + std::size_t num_ssbos = 0; u32 clip_distances = 0; for (std::size_t index = 0; index < Maxwell::MaxShaderProgram; ++index) { @@ -259,7 +336,15 @@ void RasterizerOpenGL::SetupShaders(GLenum primitive_mode) { continue; } - Shader shader{shader_cache.GetStageProgram(program)}; + Shader* const shader = shader_cache.GetStageProgram(program); + + if (device.UseAssemblyShaders()) { + // Check for ARB limitation. We only have 16 SSBOs per context state. To workaround this + // all stages share the same bindings. + const std::size_t num_stage_ssbos = shader->GetEntries().global_memory_entries.size(); + ASSERT_MSG(num_stage_ssbos == 0 || num_ssbos == 0, "SSBOs on more than one stage"); + num_ssbos += num_stage_ssbos; + } // Stage indices are 0 - 5 const std::size_t stage = index == 0 ? 0 : index - 1; @@ -526,6 +611,7 @@ void RasterizerOpenGL::Draw(bool is_indexed, bool is_instanced) { SyncFramebufferSRGB(); buffer_cache.Acquire(); + current_cbuf = 0; std::size_t buffer_size = CalculateVertexArraysSize(); @@ -535,16 +621,25 @@ void RasterizerOpenGL::Draw(bool is_indexed, bool is_instanced) { } // Uniform space for the 5 shader stages - buffer_size = Common::AlignUp<std::size_t>(buffer_size, 4) + - (sizeof(GLShader::MaxwellUniformData) + device.GetUniformBufferAlignment()) * - Maxwell::MaxShaderStage; + buffer_size = + Common::AlignUp<std::size_t>(buffer_size, 4) + + (sizeof(MaxwellUniformData) + device.GetUniformBufferAlignment()) * Maxwell::MaxShaderStage; // Add space for at least 18 constant buffers buffer_size += Maxwell::MaxConstBuffers * (Maxwell::MaxConstBufferSize + device.GetUniformBufferAlignment()); // Prepare the vertex array. - buffer_cache.Map(buffer_size); + const bool invalidated = buffer_cache.Map(buffer_size); + + if (invalidated) { + // When the stream buffer has been invalidated, we have to consider vertex buffers as dirty + auto& dirty = gpu.dirty.flags; + dirty[Dirty::VertexBuffers] = true; + for (int index = Dirty::VertexBuffer0; index <= Dirty::VertexBuffer31; ++index) { + dirty[index] = true; + } + } // Prepare vertex array format. SetupVertexFormat(); @@ -558,12 +653,14 @@ void RasterizerOpenGL::Draw(bool is_indexed, bool is_instanced) { } // Setup emulation uniform buffer. - GLShader::MaxwellUniformData ubo; - ubo.SetFromRegs(gpu); - const auto [buffer, offset] = - buffer_cache.UploadHostMemory(&ubo, sizeof(ubo), device.GetUniformBufferAlignment()); - glBindBufferRange(GL_UNIFORM_BUFFER, EmulationUniformBlockBinding, buffer, offset, - static_cast<GLsizeiptr>(sizeof(ubo))); + if (!device.UseAssemblyShaders()) { + MaxwellUniformData ubo; + ubo.SetFromRegs(gpu); + const auto info = + buffer_cache.UploadHostMemory(&ubo, sizeof(ubo), device.GetUniformBufferAlignment()); + glBindBufferRange(GL_UNIFORM_BUFFER, EmulationUniformBlockBinding, info.handle, info.offset, + static_cast<GLsizeiptr>(sizeof(ubo))); + } // Setup shaders and their used resources. texture_cache.GuardSamplers(true); @@ -630,16 +727,12 @@ void RasterizerOpenGL::Draw(bool is_indexed, bool is_instanced) { } void RasterizerOpenGL::DispatchCompute(GPUVAddr code_addr) { - if (device.HasBrokenCompute()) { - return; - } - buffer_cache.Acquire(); + current_cbuf = 0; auto kernel = shader_cache.GetComputeKernel(code_addr); SetupComputeTextures(kernel); SetupComputeImages(kernel); - program_manager.BindComputeShader(kernel->GetHandle()); const std::size_t buffer_size = Tegra::Engines::KeplerCompute::NumConstBuffers * @@ -652,6 +745,7 @@ void RasterizerOpenGL::DispatchCompute(GPUVAddr code_addr) { buffer_cache.Unmap(); const auto& launch_desc = system.GPU().KeplerCompute().launch_description; + program_manager.BindCompute(kernel->GetHandle()); glDispatchCompute(launch_desc.grid_dim_x, launch_desc.grid_dim_y, launch_desc.grid_dim_z); ++num_queued_commands; } @@ -701,15 +795,15 @@ void RasterizerOpenGL::OnCPUWrite(VAddr addr, u64 size) { return; } texture_cache.OnCPUWrite(addr, size); - shader_cache.InvalidateRegion(addr, size); + shader_cache.OnCPUWrite(addr, size); buffer_cache.OnCPUWrite(addr, size); - query_cache.InvalidateRegion(addr, size); } void RasterizerOpenGL::SyncGuestHost() { MICROPROFILE_SCOPE(OpenGL_CacheManagement); texture_cache.SyncGuestHost(); buffer_cache.SyncGuestHost(); + shader_cache.SyncGuestHost(); } void RasterizerOpenGL::SignalSemaphore(GPUVAddr addr, u32 value) { @@ -811,40 +905,72 @@ bool RasterizerOpenGL::AccelerateDisplay(const Tegra::FramebufferConfig& config, return true; } -void RasterizerOpenGL::SetupDrawConstBuffers(std::size_t stage_index, const Shader& shader) { +void RasterizerOpenGL::SetupDrawConstBuffers(std::size_t stage_index, Shader* shader) { + static constexpr std::array PARAMETER_LUT = { + GL_VERTEX_PROGRAM_PARAMETER_BUFFER_NV, GL_TESS_CONTROL_PROGRAM_PARAMETER_BUFFER_NV, + GL_TESS_EVALUATION_PROGRAM_PARAMETER_BUFFER_NV, GL_GEOMETRY_PROGRAM_PARAMETER_BUFFER_NV, + GL_FRAGMENT_PROGRAM_PARAMETER_BUFFER_NV}; + MICROPROFILE_SCOPE(OpenGL_UBO); const auto& stages = system.GPU().Maxwell3D().state.shader_stages; const auto& shader_stage = stages[stage_index]; + const auto& entries = shader->GetEntries(); + const bool use_unified = entries.use_unified_uniforms; + const std::size_t base_unified_offset = stage_index * NUM_CONST_BUFFERS_BYTES_PER_STAGE; - u32 binding = device.GetBaseBindings(stage_index).uniform_buffer; - for (const auto& entry : shader->GetEntries().const_buffers) { - const auto& buffer = shader_stage.const_buffers[entry.GetIndex()]; - SetupConstBuffer(binding++, buffer, entry); + const auto base_bindings = device.GetBaseBindings(stage_index); + u32 binding = device.UseAssemblyShaders() ? 0 : base_bindings.uniform_buffer; + for (const auto& entry : entries.const_buffers) { + const u32 index = entry.GetIndex(); + const auto& buffer = shader_stage.const_buffers[index]; + SetupConstBuffer(PARAMETER_LUT[stage_index], binding, buffer, entry, use_unified, + base_unified_offset + index * Maxwell::MaxConstBufferSize); + ++binding; + } + if (use_unified) { + const u32 index = static_cast<u32>(base_bindings.shader_storage_buffer + + entries.global_memory_entries.size()); + glBindBufferRange(GL_SHADER_STORAGE_BUFFER, index, unified_uniform_buffer.handle, + base_unified_offset, NUM_CONST_BUFFERS_BYTES_PER_STAGE); } } -void RasterizerOpenGL::SetupComputeConstBuffers(const Shader& kernel) { +void RasterizerOpenGL::SetupComputeConstBuffers(Shader* kernel) { MICROPROFILE_SCOPE(OpenGL_UBO); const auto& launch_desc = system.GPU().KeplerCompute().launch_description; + const auto& entries = kernel->GetEntries(); + const bool use_unified = entries.use_unified_uniforms; u32 binding = 0; - for (const auto& entry : kernel->GetEntries().const_buffers) { + for (const auto& entry : entries.const_buffers) { const auto& config = launch_desc.const_buffer_config[entry.GetIndex()]; const std::bitset<8> mask = launch_desc.const_buffer_enable_mask.Value(); Tegra::Engines::ConstBufferInfo buffer; buffer.address = config.Address(); buffer.size = config.size; buffer.enabled = mask[entry.GetIndex()]; - SetupConstBuffer(binding++, buffer, entry); + SetupConstBuffer(GL_COMPUTE_PROGRAM_PARAMETER_BUFFER_NV, binding, buffer, entry, + use_unified, entry.GetIndex() * Maxwell::MaxConstBufferSize); + ++binding; + } + if (use_unified) { + const GLuint index = static_cast<GLuint>(entries.global_memory_entries.size()); + glBindBufferRange(GL_SHADER_STORAGE_BUFFER, index, unified_uniform_buffer.handle, 0, + NUM_CONST_BUFFERS_BYTES_PER_STAGE); } } -void RasterizerOpenGL::SetupConstBuffer(u32 binding, const Tegra::Engines::ConstBufferInfo& buffer, - const ConstBufferEntry& entry) { +void RasterizerOpenGL::SetupConstBuffer(GLenum stage, u32 binding, + const Tegra::Engines::ConstBufferInfo& buffer, + const ConstBufferEntry& entry, bool use_unified, + std::size_t unified_offset) { if (!buffer.enabled) { // Set values to zero to unbind buffers - glBindBufferRange(GL_UNIFORM_BUFFER, binding, buffer_cache.GetEmptyBuffer(sizeof(float)), 0, - sizeof(float)); + if (device.UseAssemblyShaders()) { + glBindBufferRangeNV(stage, entry.GetIndex(), 0, 0, 0); + } else { + glBindBufferRange(GL_UNIFORM_BUFFER, binding, 0, 0, sizeof(float)); + } return; } @@ -852,18 +978,39 @@ void RasterizerOpenGL::SetupConstBuffer(u32 binding, const Tegra::Engines::Const // UBO alignment requirements. const std::size_t size = Common::AlignUp(GetConstBufferSize(buffer, entry), sizeof(GLvec4)); - const auto alignment = device.GetUniformBufferAlignment(); - const auto [cbuf, offset] = buffer_cache.UploadMemory(buffer.address, size, alignment, false, - device.HasFastBufferSubData()); - glBindBufferRange(GL_UNIFORM_BUFFER, binding, cbuf, offset, size); + const bool fast_upload = !use_unified && device.HasFastBufferSubData(); + + const std::size_t alignment = use_unified ? 4 : device.GetUniformBufferAlignment(); + const GPUVAddr gpu_addr = buffer.address; + auto info = buffer_cache.UploadMemory(gpu_addr, size, alignment, false, fast_upload); + + if (device.UseAssemblyShaders()) { + UNIMPLEMENTED_IF(use_unified); + if (info.offset != 0) { + const GLuint staging_cbuf = staging_cbufs[current_cbuf++]; + glCopyNamedBufferSubData(info.handle, staging_cbuf, info.offset, 0, size); + info.handle = staging_cbuf; + info.offset = 0; + } + glBindBufferRangeNV(stage, binding, info.handle, info.offset, size); + return; + } + + if (use_unified) { + glCopyNamedBufferSubData(info.handle, unified_uniform_buffer.handle, info.offset, + unified_offset, size); + } else { + glBindBufferRange(GL_UNIFORM_BUFFER, binding, info.handle, info.offset, size); + } } -void RasterizerOpenGL::SetupDrawGlobalMemory(std::size_t stage_index, const Shader& shader) { +void RasterizerOpenGL::SetupDrawGlobalMemory(std::size_t stage_index, Shader* shader) { auto& gpu{system.GPU()}; auto& memory_manager{gpu.MemoryManager()}; const auto cbufs{gpu.Maxwell3D().state.shader_stages[stage_index]}; - u32 binding = device.GetBaseBindings(stage_index).shader_storage_buffer; + u32 binding = + device.UseAssemblyShaders() ? 0 : device.GetBaseBindings(stage_index).shader_storage_buffer; for (const auto& entry : shader->GetEntries().global_memory_entries) { const GPUVAddr addr{cbufs.const_buffers[entry.cbuf_index].address + entry.cbuf_offset}; const GPUVAddr gpu_addr{memory_manager.Read<u64>(addr)}; @@ -872,7 +1019,7 @@ void RasterizerOpenGL::SetupDrawGlobalMemory(std::size_t stage_index, const Shad } } -void RasterizerOpenGL::SetupComputeGlobalMemory(const Shader& kernel) { +void RasterizerOpenGL::SetupComputeGlobalMemory(Shader* kernel) { auto& gpu{system.GPU()}; auto& memory_manager{gpu.MemoryManager()}; const auto cbufs{gpu.KeplerCompute().launch_description.const_buffer_config}; @@ -889,13 +1036,12 @@ void RasterizerOpenGL::SetupComputeGlobalMemory(const Shader& kernel) { void RasterizerOpenGL::SetupGlobalMemory(u32 binding, const GlobalMemoryEntry& entry, GPUVAddr gpu_addr, std::size_t size) { const auto alignment{device.GetShaderStorageBufferAlignment()}; - const auto [ssbo, buffer_offset] = - buffer_cache.UploadMemory(gpu_addr, size, alignment, entry.is_written); - glBindBufferRange(GL_SHADER_STORAGE_BUFFER, binding, ssbo, buffer_offset, + const auto info = buffer_cache.UploadMemory(gpu_addr, size, alignment, entry.is_written); + glBindBufferRange(GL_SHADER_STORAGE_BUFFER, binding, info.handle, info.offset, static_cast<GLsizeiptr>(size)); } -void RasterizerOpenGL::SetupDrawTextures(std::size_t stage_index, const Shader& shader) { +void RasterizerOpenGL::SetupDrawTextures(std::size_t stage_index, Shader* shader) { MICROPROFILE_SCOPE(OpenGL_Texture); const auto& maxwell3d = system.GPU().Maxwell3D(); u32 binding = device.GetBaseBindings(stage_index).sampler; @@ -908,7 +1054,7 @@ void RasterizerOpenGL::SetupDrawTextures(std::size_t stage_index, const Shader& } } -void RasterizerOpenGL::SetupComputeTextures(const Shader& kernel) { +void RasterizerOpenGL::SetupComputeTextures(Shader* kernel) { MICROPROFILE_SCOPE(OpenGL_Texture); const auto& compute = system.GPU().KeplerCompute(); u32 binding = 0; @@ -929,19 +1075,15 @@ void RasterizerOpenGL::SetupTexture(u32 binding, const Tegra::Texture::FullTextu glBindTextureUnit(binding, 0); return; } - glBindTextureUnit(binding, view->GetTexture()); - - if (view->GetSurfaceParams().IsBuffer()) { - return; + const GLuint handle = view->GetTexture(texture.tic.x_source, texture.tic.y_source, + texture.tic.z_source, texture.tic.w_source); + glBindTextureUnit(binding, handle); + if (!view->GetSurfaceParams().IsBuffer()) { + glBindSampler(binding, sampler_cache.GetSampler(texture.tsc)); } - // Apply swizzle to textures that are not buffers. - view->ApplySwizzle(texture.tic.x_source, texture.tic.y_source, texture.tic.z_source, - texture.tic.w_source); - - glBindSampler(binding, sampler_cache.GetSampler(texture.tsc)); } -void RasterizerOpenGL::SetupDrawImages(std::size_t stage_index, const Shader& shader) { +void RasterizerOpenGL::SetupDrawImages(std::size_t stage_index, Shader* shader) { const auto& maxwell3d = system.GPU().Maxwell3D(); u32 binding = device.GetBaseBindings(stage_index).image; for (const auto& entry : shader->GetEntries().images) { @@ -951,7 +1093,7 @@ void RasterizerOpenGL::SetupDrawImages(std::size_t stage_index, const Shader& sh } } -void RasterizerOpenGL::SetupComputeImages(const Shader& shader) { +void RasterizerOpenGL::SetupComputeImages(Shader* shader) { const auto& compute = system.GPU().KeplerCompute(); u32 binding = 0; for (const auto& entry : shader->GetEntries().images) { @@ -967,14 +1109,11 @@ void RasterizerOpenGL::SetupImage(u32 binding, const Tegra::Texture::TICEntry& t glBindImageTexture(binding, 0, 0, GL_FALSE, 0, GL_READ_ONLY, GL_R8); return; } - if (!tic.IsBuffer()) { - view->ApplySwizzle(tic.x_source, tic.y_source, tic.z_source, tic.w_source); - } if (entry.is_written) { view->MarkAsModified(texture_cache.Tick()); } - glBindImageTexture(binding, view->GetTexture(), 0, GL_TRUE, 0, GL_READ_WRITE, - view->GetFormat()); + const GLuint handle = view->GetTexture(tic.x_source, tic.y_source, tic.z_source, tic.w_source); + glBindImageTexture(binding, handle, 0, GL_TRUE, 0, GL_READ_WRITE, view->GetFormat()); } void RasterizerOpenGL::SyncViewport() { @@ -983,6 +1122,26 @@ void RasterizerOpenGL::SyncViewport() { const auto& regs = gpu.regs; const bool dirty_viewport = flags[Dirty::Viewports]; + const bool dirty_clip_control = flags[Dirty::ClipControl]; + + if (dirty_clip_control || flags[Dirty::FrontFace]) { + flags[Dirty::FrontFace] = false; + + GLenum mode = MaxwellToGL::FrontFace(regs.front_face); + if (regs.screen_y_control.triangle_rast_flip != 0 && + regs.viewport_transform[0].scale_y < 0.0f) { + switch (mode) { + case GL_CW: + mode = GL_CCW; + break; + case GL_CCW: + mode = GL_CW; + break; + } + } + glFrontFace(mode); + } + if (dirty_viewport || flags[Dirty::ClipControl]) { flags[Dirty::ClipControl] = false; @@ -1019,6 +1178,14 @@ void RasterizerOpenGL::SyncViewport() { const GLdouble near_depth = src.translate_z - src.scale_z * reduce_z; const GLdouble far_depth = src.translate_z + src.scale_z; glDepthRangeIndexed(static_cast<GLuint>(i), near_depth, far_depth); + + if (!GLAD_GL_NV_viewport_swizzle) { + continue; + } + glViewportSwizzleNV(static_cast<GLuint>(i), MaxwellToGL::ViewportSwizzle(src.swizzle.x), + MaxwellToGL::ViewportSwizzle(src.swizzle.y), + MaxwellToGL::ViewportSwizzle(src.swizzle.z), + MaxwellToGL::ViewportSwizzle(src.swizzle.w)); } } } @@ -1072,11 +1239,6 @@ void RasterizerOpenGL::SyncCullMode() { glDisable(GL_CULL_FACE); } } - - if (flags[Dirty::FrontFace]) { - flags[Dirty::FrontFace] = false; - glFrontFace(MaxwellToGL::FrontFace(regs.front_face)); - } } void RasterizerOpenGL::SyncPrimitiveRestart() { @@ -1447,12 +1609,70 @@ void RasterizerOpenGL::SyncFramebufferSRGB() { oglEnable(GL_FRAMEBUFFER_SRGB, gpu.regs.framebuffer_srgb); } +void RasterizerOpenGL::SyncTransformFeedback() { + // TODO(Rodrigo): Inject SKIP_COMPONENTS*_NV when required. An unimplemented message will signal + // when this is required. + const auto& regs = system.GPU().Maxwell3D().regs; + + static constexpr std::size_t STRIDE = 3; + std::array<GLint, 128 * STRIDE * Maxwell::NumTransformFeedbackBuffers> attribs; + std::array<GLint, Maxwell::NumTransformFeedbackBuffers> streams; + + GLint* cursor = attribs.data(); + GLint* current_stream = streams.data(); + + for (std::size_t feedback = 0; feedback < Maxwell::NumTransformFeedbackBuffers; ++feedback) { + const auto& layout = regs.tfb_layouts[feedback]; + UNIMPLEMENTED_IF_MSG(layout.stride != layout.varying_count * 4, "Stride padding"); + if (layout.varying_count == 0) { + continue; + } + + *current_stream = static_cast<GLint>(feedback); + if (current_stream != streams.data()) { + // When stepping one stream, push the expected token + cursor[0] = GL_NEXT_BUFFER_NV; + cursor[1] = 0; + cursor[2] = 0; + cursor += STRIDE; + } + ++current_stream; + + const auto& locations = regs.tfb_varying_locs[feedback]; + std::optional<u8> current_index; + for (u32 offset = 0; offset < layout.varying_count; ++offset) { + const u8 location = locations[offset]; + const u8 index = location / 4; + + if (current_index == index) { + // Increase number of components of the previous attachment + ++cursor[-2]; + continue; + } + current_index = index; + + std::tie(cursor[0], cursor[2]) = TransformFeedbackEnum(location); + cursor[1] = 1; + cursor += STRIDE; + } + } + + const GLsizei num_attribs = static_cast<GLsizei>((cursor - attribs.data()) / STRIDE); + const GLsizei num_strides = static_cast<GLsizei>(current_stream - streams.data()); + glTransformFeedbackStreamAttribsNV(num_attribs, attribs.data(), num_strides, streams.data(), + GL_INTERLEAVED_ATTRIBS); +} + void RasterizerOpenGL::BeginTransformFeedback(GLenum primitive_mode) { const auto& regs = system.GPU().Maxwell3D().regs; if (regs.tfb_enabled == 0) { return; } + if (device.UseAssemblyShaders()) { + SyncTransformFeedback(); + } + UNIMPLEMENTED_IF(regs.IsShaderConfigEnabled(Maxwell::ShaderProgram::TesselationControl) || regs.IsShaderConfigEnabled(Maxwell::ShaderProgram::TesselationEval) || regs.IsShaderConfigEnabled(Maxwell::ShaderProgram::Geometry)); @@ -1479,6 +1699,10 @@ void RasterizerOpenGL::BeginTransformFeedback(GLenum primitive_mode) { static_cast<GLsizeiptr>(size)); } + // We may have to call BeginTransformFeedbackNV here since they seem to call different + // implementations on Nvidia's driver (the pointer is different) but we are using + // ARB_transform_feedback3 features with NV_transform_feedback interactions and the ARB + // extension doesn't define BeginTransformFeedback (without NV) interactions. It just works. glBeginTransformFeedback(GL_POINTS); } @@ -1500,8 +1724,9 @@ void RasterizerOpenGL::EndTransformFeedback() { const GLuint handle = transform_feedback_buffers[index].handle; const GPUVAddr gpu_addr = binding.Address(); const std::size_t size = binding.buffer_size; - const auto [dest_buffer, offset] = buffer_cache.UploadMemory(gpu_addr, size, 4, true); - glCopyNamedBufferSubData(handle, dest_buffer, 0, offset, static_cast<GLsizeiptr>(size)); + const auto info = buffer_cache.UploadMemory(gpu_addr, size, 4, true); + glCopyNamedBufferSubData(handle, info.handle, 0, info.offset, + static_cast<GLsizeiptr>(size)); } } diff --git a/src/video_core/renderer_opengl/gl_rasterizer.h b/src/video_core/renderer_opengl/gl_rasterizer.h index b94c65907..4f082592f 100644 --- a/src/video_core/renderer_opengl/gl_rasterizer.h +++ b/src/video_core/renderer_opengl/gl_rasterizer.h @@ -19,7 +19,6 @@ #include "video_core/engines/const_buffer_info.h" #include "video_core/engines/maxwell_3d.h" #include "video_core/rasterizer_accelerated.h" -#include "video_core/rasterizer_cache.h" #include "video_core/rasterizer_interface.h" #include "video_core/renderer_opengl/gl_buffer_cache.h" #include "video_core/renderer_opengl/gl_device.h" @@ -56,8 +55,8 @@ struct DrawParameters; class RasterizerOpenGL : public VideoCore::RasterizerAccelerated { public: explicit RasterizerOpenGL(Core::System& system, Core::Frontend::EmuWindow& emu_window, - ScreenInfo& info, GLShader::ProgramManager& program_manager, - StateTracker& state_tracker); + const Device& device, ScreenInfo& info, + ProgramManager& program_manager, StateTracker& state_tracker); ~RasterizerOpenGL() override; void Draw(bool is_indexed, bool is_instanced) override; @@ -100,40 +99,41 @@ private: void ConfigureClearFramebuffer(bool using_color, bool using_depth_stencil); /// Configures the current constbuffers to use for the draw command. - void SetupDrawConstBuffers(std::size_t stage_index, const Shader& shader); + void SetupDrawConstBuffers(std::size_t stage_index, Shader* shader); /// Configures the current constbuffers to use for the kernel invocation. - void SetupComputeConstBuffers(const Shader& kernel); + void SetupComputeConstBuffers(Shader* kernel); /// Configures a constant buffer. - void SetupConstBuffer(u32 binding, const Tegra::Engines::ConstBufferInfo& buffer, - const ConstBufferEntry& entry); + void SetupConstBuffer(GLenum stage, u32 binding, const Tegra::Engines::ConstBufferInfo& buffer, + const ConstBufferEntry& entry, bool use_unified, + std::size_t unified_offset); /// Configures the current global memory entries to use for the draw command. - void SetupDrawGlobalMemory(std::size_t stage_index, const Shader& shader); + void SetupDrawGlobalMemory(std::size_t stage_index, Shader* shader); /// Configures the current global memory entries to use for the kernel invocation. - void SetupComputeGlobalMemory(const Shader& kernel); + void SetupComputeGlobalMemory(Shader* kernel); /// Configures a constant buffer. void SetupGlobalMemory(u32 binding, const GlobalMemoryEntry& entry, GPUVAddr gpu_addr, std::size_t size); /// Configures the current textures to use for the draw command. - void SetupDrawTextures(std::size_t stage_index, const Shader& shader); + void SetupDrawTextures(std::size_t stage_index, Shader* shader); /// Configures the textures used in a compute shader. - void SetupComputeTextures(const Shader& kernel); + void SetupComputeTextures(Shader* kernel); /// Configures a texture. void SetupTexture(u32 binding, const Tegra::Texture::FullTextureInfo& texture, const SamplerEntry& entry); /// Configures images in a graphics shader. - void SetupDrawImages(std::size_t stage_index, const Shader& shader); + void SetupDrawImages(std::size_t stage_index, Shader* shader); /// Configures images in a compute shader. - void SetupComputeImages(const Shader& shader); + void SetupComputeImages(Shader* shader); /// Configures an image. void SetupImage(u32 binding, const Tegra::Texture::TICEntry& tic, const ImageEntry& entry); @@ -201,6 +201,10 @@ private: /// Syncs the framebuffer sRGB state to match the guest state void SyncFramebufferSRGB(); + /// Syncs transform feedback state to match guest state + /// @note Only valid on assembly shaders + void SyncTransformFeedback(); + /// Begin a transform feedback void BeginTransformFeedback(GLenum primitive_mode); @@ -224,7 +228,7 @@ private: void SetupShaders(GLenum primitive_mode); - const Device device; + const Device& device; TextureCacheOpenGL texture_cache; ShaderCacheOpenGL shader_cache; @@ -236,7 +240,7 @@ private: Core::System& system; ScreenInfo& screen_info; - GLShader::ProgramManager& program_manager; + ProgramManager& program_manager; StateTracker& state_tracker; static constexpr std::size_t STREAM_BUFFER_SIZE = 128 * 1024 * 1024; @@ -248,6 +252,13 @@ private: std::bitset<Tegra::Engines::Maxwell3D::Regs::NumTransformFeedbackBuffers> enabled_transform_feedback_buffers; + static constexpr std::size_t NUM_CONSTANT_BUFFERS = + Tegra::Engines::Maxwell3D::Regs::MaxConstBuffers * + Tegra::Engines::Maxwell3D::Regs::MaxShaderProgram; + std::array<GLuint, NUM_CONSTANT_BUFFERS> staging_cbufs{}; + std::size_t current_cbuf = 0; + OGLBuffer unified_uniform_buffer; + /// Number of commands queued to the OpenGL driver. Reseted on flush. std::size_t num_queued_commands = 0; diff --git a/src/video_core/renderer_opengl/gl_resource_manager.cpp b/src/video_core/renderer_opengl/gl_resource_manager.cpp index 97803d480..a787e27d2 100644 --- a/src/video_core/renderer_opengl/gl_resource_manager.cpp +++ b/src/video_core/renderer_opengl/gl_resource_manager.cpp @@ -125,6 +125,15 @@ void OGLProgram::Release() { handle = 0; } +void OGLAssemblyProgram::Release() { + if (handle == 0) { + return; + } + MICROPROFILE_SCOPE(OpenGL_ResourceDeletion); + glDeleteProgramsARB(1, &handle); + handle = 0; +} + void OGLPipeline::Create() { if (handle != 0) return; diff --git a/src/video_core/renderer_opengl/gl_resource_manager.h b/src/video_core/renderer_opengl/gl_resource_manager.h index de93f4212..f8b322227 100644 --- a/src/video_core/renderer_opengl/gl_resource_manager.h +++ b/src/video_core/renderer_opengl/gl_resource_manager.h @@ -167,6 +167,22 @@ public: GLuint handle = 0; }; +class OGLAssemblyProgram : private NonCopyable { +public: + OGLAssemblyProgram() = default; + + OGLAssemblyProgram(OGLAssemblyProgram&& o) noexcept : handle(std::exchange(o.handle, 0)) {} + + ~OGLAssemblyProgram() { + Release(); + } + + /// Deletes the internal OpenGL resource + void Release(); + + GLuint handle = 0; +}; + class OGLPipeline : private NonCopyable { public: OGLPipeline() = default; diff --git a/src/video_core/renderer_opengl/gl_shader_cache.cpp b/src/video_core/renderer_opengl/gl_shader_cache.cpp index 9759a7078..c6a3bf3a1 100644 --- a/src/video_core/renderer_opengl/gl_shader_cache.cpp +++ b/src/video_core/renderer_opengl/gl_shader_cache.cpp @@ -20,6 +20,7 @@ #include "video_core/engines/maxwell_3d.h" #include "video_core/engines/shader_type.h" #include "video_core/memory_manager.h" +#include "video_core/renderer_opengl/gl_arb_decompiler.h" #include "video_core/renderer_opengl/gl_rasterizer.h" #include "video_core/renderer_opengl/gl_shader_cache.h" #include "video_core/renderer_opengl/gl_shader_decompiler.h" @@ -29,6 +30,7 @@ #include "video_core/shader/memory_util.h" #include "video_core/shader/registry.h" #include "video_core/shader/shader_ir.h" +#include "video_core/shader_cache.h" namespace OpenGL { @@ -97,6 +99,24 @@ constexpr ShaderType GetShaderType(Maxwell::ShaderProgram program_type) { return {}; } +constexpr GLenum AssemblyEnum(ShaderType shader_type) { + switch (shader_type) { + case ShaderType::Vertex: + return GL_VERTEX_PROGRAM_NV; + case ShaderType::TesselationControl: + return GL_TESS_CONTROL_PROGRAM_NV; + case ShaderType::TesselationEval: + return GL_TESS_EVALUATION_PROGRAM_NV; + case ShaderType::Geometry: + return GL_GEOMETRY_PROGRAM_NV; + case ShaderType::Fragment: + return GL_FRAGMENT_PROGRAM_NV; + case ShaderType::Compute: + return GL_COMPUTE_PROGRAM_NV; + } + return {}; +} + std::string MakeShaderID(u64 unique_identifier, ShaderType shader_type) { return fmt::format("{}{:016X}", GetShaderTypeName(shader_type), unique_identifier); } @@ -120,18 +140,44 @@ std::shared_ptr<Registry> MakeRegistry(const ShaderDiskCacheEntry& entry) { return registry; } -std::shared_ptr<OGLProgram> BuildShader(const Device& device, ShaderType shader_type, - u64 unique_identifier, const ShaderIR& ir, - const Registry& registry, bool hint_retrievable = false) { +ProgramSharedPtr BuildShader(const Device& device, ShaderType shader_type, u64 unique_identifier, + const ShaderIR& ir, const Registry& registry, + bool hint_retrievable = false) { const std::string shader_id = MakeShaderID(unique_identifier, shader_type); LOG_INFO(Render_OpenGL, "{}", shader_id); - const std::string glsl = DecompileShader(device, ir, registry, shader_type, shader_id); - OGLShader shader; - shader.Create(glsl.c_str(), GetGLShaderType(shader_type)); + auto program = std::make_shared<ProgramHandle>(); + + if (device.UseAssemblyShaders()) { + const std::string arb = + DecompileAssemblyShader(device, ir, registry, shader_type, shader_id); + + GLuint& arb_prog = program->assembly_program.handle; + +// Commented out functions signal OpenGL errors but are compatible with apitrace. +// Use them only to capture and replay on apitrace. +#if 0 + glGenProgramsNV(1, &arb_prog); + glLoadProgramNV(AssemblyEnum(shader_type), arb_prog, static_cast<GLsizei>(arb.size()), + reinterpret_cast<const GLubyte*>(arb.data())); +#else + glGenProgramsARB(1, &arb_prog); + glNamedProgramStringEXT(arb_prog, AssemblyEnum(shader_type), GL_PROGRAM_FORMAT_ASCII_ARB, + static_cast<GLsizei>(arb.size()), arb.data()); +#endif + const auto err = reinterpret_cast<const char*>(glGetString(GL_PROGRAM_ERROR_STRING_NV)); + if (err && *err) { + LOG_CRITICAL(Render_OpenGL, "{}", err); + LOG_INFO(Render_OpenGL, "\n{}", arb); + } + } else { + const std::string glsl = DecompileShader(device, ir, registry, shader_type, shader_id); + OGLShader shader; + shader.Create(glsl.c_str(), GetGLShaderType(shader_type)); + + program->source_program.Create(true, hint_retrievable, shader.handle); + } - auto program = std::make_shared<OGLProgram>(); - program->Create(true, hint_retrievable, shader.handle); return program; } @@ -151,22 +197,26 @@ std::unordered_set<GLenum> GetSupportedFormats() { } // Anonymous namespace -CachedShader::CachedShader(VAddr cpu_addr, std::size_t size_in_bytes, - std::shared_ptr<VideoCommon::Shader::Registry> registry, - ShaderEntries entries, std::shared_ptr<OGLProgram> program) - : RasterizerCacheObject{cpu_addr}, registry{std::move(registry)}, entries{std::move(entries)}, - size_in_bytes{size_in_bytes}, program{std::move(program)} {} +Shader::Shader(std::shared_ptr<VideoCommon::Shader::Registry> registry_, ShaderEntries entries_, + ProgramSharedPtr program_) + : registry{std::move(registry_)}, entries{std::move(entries_)}, program{std::move(program_)} { + handle = program->assembly_program.handle; + if (handle == 0) { + handle = program->source_program.handle; + } + ASSERT(handle != 0); +} -CachedShader::~CachedShader() = default; +Shader::~Shader() = default; -GLuint CachedShader::GetHandle() const { +GLuint Shader::GetHandle() const { DEBUG_ASSERT(registry->IsConsistent()); - return program->handle; + return handle; } -Shader CachedShader::CreateStageFromMemory(const ShaderParameters& params, - Maxwell::ShaderProgram program_type, ProgramCode code, - ProgramCode code_b) { +std::unique_ptr<Shader> Shader::CreateStageFromMemory(const ShaderParameters& params, + Maxwell::ShaderProgram program_type, + ProgramCode code, ProgramCode code_b) { const auto shader_type = GetShaderType(program_type); const std::size_t size_in_bytes = code.size() * sizeof(u64); @@ -191,11 +241,12 @@ Shader CachedShader::CreateStageFromMemory(const ShaderParameters& params, entry.bindless_samplers = registry->GetBindlessSamplers(); params.disk_cache.SaveEntry(std::move(entry)); - return std::shared_ptr<CachedShader>(new CachedShader( - params.cpu_addr, size_in_bytes, std::move(registry), MakeEntries(ir), std::move(program))); + return std::unique_ptr<Shader>(new Shader( + std::move(registry), MakeEntries(params.device, ir, shader_type), std::move(program))); } -Shader CachedShader::CreateKernelFromMemory(const ShaderParameters& params, ProgramCode code) { +std::unique_ptr<Shader> Shader::CreateKernelFromMemory(const ShaderParameters& params, + ProgramCode code) { const std::size_t size_in_bytes = code.size() * sizeof(u64); auto& engine = params.system.GPU().KeplerCompute(); @@ -215,22 +266,23 @@ Shader CachedShader::CreateKernelFromMemory(const ShaderParameters& params, Prog entry.bindless_samplers = registry->GetBindlessSamplers(); params.disk_cache.SaveEntry(std::move(entry)); - return std::shared_ptr<CachedShader>(new CachedShader( - params.cpu_addr, size_in_bytes, std::move(registry), MakeEntries(ir), std::move(program))); + return std::unique_ptr<Shader>(new Shader(std::move(registry), + MakeEntries(params.device, ir, ShaderType::Compute), + std::move(program))); } -Shader CachedShader::CreateFromCache(const ShaderParameters& params, - const PrecompiledShader& precompiled_shader, - std::size_t size_in_bytes) { - return std::shared_ptr<CachedShader>( - new CachedShader(params.cpu_addr, size_in_bytes, precompiled_shader.registry, - precompiled_shader.entries, precompiled_shader.program)); +std::unique_ptr<Shader> Shader::CreateFromCache(const ShaderParameters& params, + const PrecompiledShader& precompiled_shader) { + return std::unique_ptr<Shader>(new Shader( + precompiled_shader.registry, precompiled_shader.entries, precompiled_shader.program)); } ShaderCacheOpenGL::ShaderCacheOpenGL(RasterizerOpenGL& rasterizer, Core::System& system, Core::Frontend::EmuWindow& emu_window, const Device& device) - : RasterizerCache{rasterizer}, system{system}, emu_window{emu_window}, device{device}, - disk_cache{system} {} + : VideoCommon::ShaderCache<Shader>{rasterizer}, system{system}, + emu_window{emu_window}, device{device}, disk_cache{system} {} + +ShaderCacheOpenGL::~ShaderCacheOpenGL() = default; void ShaderCacheOpenGL::LoadDiskCache(const std::atomic_bool& stop_loading, const VideoCore::DiskResourceLoadCallback& callback) { @@ -239,7 +291,11 @@ void ShaderCacheOpenGL::LoadDiskCache(const std::atomic_bool& stop_loading, return; } - const std::vector gl_cache = disk_cache.LoadPrecompiled(); + std::vector<ShaderDiskCachePrecompiled> gl_cache; + if (!device.UseAssemblyShaders()) { + // Only load precompiled cache when we are not using assembly shaders + gl_cache = disk_cache.LoadPrecompiled(); + } const auto supported_formats = GetSupportedFormats(); // Track if precompiled cache was altered during loading to know if we have to @@ -278,7 +334,7 @@ void ShaderCacheOpenGL::LoadDiskCache(const std::atomic_bool& stop_loading, auto registry = MakeRegistry(entry); const ShaderIR ir(entry.code, main_offset, COMPILER_SETTINGS, *registry); - std::shared_ptr<OGLProgram> program; + ProgramSharedPtr program; if (precompiled_entry) { // If the shader is precompiled, attempt to load it with program = GeneratePrecompiledProgram(entry, *precompiled_entry, supported_formats); @@ -294,7 +350,7 @@ void ShaderCacheOpenGL::LoadDiskCache(const std::atomic_bool& stop_loading, PrecompiledShader shader; shader.program = std::move(program); shader.registry = std::move(registry); - shader.entries = MakeEntries(ir); + shader.entries = MakeEntries(device, ir, entry.type); std::scoped_lock lock{mutex}; if (callback) { @@ -332,6 +388,11 @@ void ShaderCacheOpenGL::LoadDiskCache(const std::atomic_bool& stop_loading, return; } + if (device.UseAssemblyShaders()) { + // Don't store precompiled binaries for assembly shaders. + return; + } + // TODO(Rodrigo): Do state tracking for transferable shaders and do a dummy draw // before precompiling them @@ -339,7 +400,7 @@ void ShaderCacheOpenGL::LoadDiskCache(const std::atomic_bool& stop_loading, const u64 id = (*transferable)[i].unique_identifier; const auto it = find_precompiled(id); if (it == gl_cache.end()) { - const GLuint program = runtime_cache.at(id).program->handle; + const GLuint program = runtime_cache.at(id).program->source_program.handle; disk_cache.SavePrecompiled(id, program); precompiled_cache_altered = true; } @@ -350,7 +411,7 @@ void ShaderCacheOpenGL::LoadDiskCache(const std::atomic_bool& stop_loading, } } -std::shared_ptr<OGLProgram> ShaderCacheOpenGL::GeneratePrecompiledProgram( +ProgramSharedPtr ShaderCacheOpenGL::GeneratePrecompiledProgram( const ShaderDiskCacheEntry& entry, const ShaderDiskCachePrecompiled& precompiled_entry, const std::unordered_set<GLenum>& supported_formats) { if (supported_formats.find(precompiled_entry.binary_format) == supported_formats.end()) { @@ -358,15 +419,15 @@ std::shared_ptr<OGLProgram> ShaderCacheOpenGL::GeneratePrecompiledProgram( return {}; } - auto program = std::make_shared<OGLProgram>(); - program->handle = glCreateProgram(); - glProgramParameteri(program->handle, GL_PROGRAM_SEPARABLE, GL_TRUE); - glProgramBinary(program->handle, precompiled_entry.binary_format, - precompiled_entry.binary.data(), + auto program = std::make_shared<ProgramHandle>(); + GLuint& handle = program->source_program.handle; + handle = glCreateProgram(); + glProgramParameteri(handle, GL_PROGRAM_SEPARABLE, GL_TRUE); + glProgramBinary(handle, precompiled_entry.binary_format, precompiled_entry.binary.data(), static_cast<GLsizei>(precompiled_entry.binary.size())); GLint link_status; - glGetProgramiv(program->handle, GL_LINK_STATUS, &link_status); + glGetProgramiv(handle, GL_LINK_STATUS, &link_status); if (link_status == GL_FALSE) { LOG_INFO(Render_OpenGL, "Precompiled cache rejected by the driver, removing"); return {}; @@ -375,7 +436,7 @@ std::shared_ptr<OGLProgram> ShaderCacheOpenGL::GeneratePrecompiledProgram( return program; } -Shader ShaderCacheOpenGL::GetStageProgram(Maxwell::ShaderProgram program) { +Shader* ShaderCacheOpenGL::GetStageProgram(Maxwell::ShaderProgram program) { if (!system.GPU().Maxwell3D().dirty.flags[Dirty::Shaders]) { return last_shaders[static_cast<std::size_t>(program)]; } @@ -385,8 +446,7 @@ Shader ShaderCacheOpenGL::GetStageProgram(Maxwell::ShaderProgram program) { // Look up shader in the cache based on address const auto cpu_addr{memory_manager.GpuToCpuAddress(address)}; - Shader shader{cpu_addr ? TryGet(*cpu_addr) : null_shader}; - if (shader) { + if (Shader* const shader{cpu_addr ? TryGet(*cpu_addr) : null_shader.get()}) { return last_shaders[static_cast<std::size_t>(program)] = shader; } @@ -400,62 +460,64 @@ Shader ShaderCacheOpenGL::GetStageProgram(Maxwell::ShaderProgram program) { const u8* host_ptr_b = memory_manager.GetPointer(address_b); code_b = GetShaderCode(memory_manager, address_b, host_ptr_b, false); } + const std::size_t code_size = code.size() * sizeof(u64); - const auto unique_identifier = GetUniqueIdentifier( + const u64 unique_identifier = GetUniqueIdentifier( GetShaderType(program), program == Maxwell::ShaderProgram::VertexA, code, code_b); const ShaderParameters params{system, disk_cache, device, *cpu_addr, host_ptr, unique_identifier}; + std::unique_ptr<Shader> shader; const auto found = runtime_cache.find(unique_identifier); if (found == runtime_cache.end()) { - shader = CachedShader::CreateStageFromMemory(params, program, std::move(code), - std::move(code_b)); + shader = Shader::CreateStageFromMemory(params, program, std::move(code), std::move(code_b)); } else { - const std::size_t size_in_bytes = code.size() * sizeof(u64); - shader = CachedShader::CreateFromCache(params, found->second, size_in_bytes); + shader = Shader::CreateFromCache(params, found->second); } + Shader* const result = shader.get(); if (cpu_addr) { - Register(shader); + Register(std::move(shader), *cpu_addr, code_size); } else { - null_shader = shader; + null_shader = std::move(shader); } - return last_shaders[static_cast<std::size_t>(program)] = shader; + return last_shaders[static_cast<std::size_t>(program)] = result; } -Shader ShaderCacheOpenGL::GetComputeKernel(GPUVAddr code_addr) { +Shader* ShaderCacheOpenGL::GetComputeKernel(GPUVAddr code_addr) { auto& memory_manager{system.GPU().MemoryManager()}; const auto cpu_addr{memory_manager.GpuToCpuAddress(code_addr)}; - auto kernel = cpu_addr ? TryGet(*cpu_addr) : null_kernel; - if (kernel) { + if (Shader* const kernel = cpu_addr ? TryGet(*cpu_addr) : null_kernel.get()) { return kernel; } const auto host_ptr{memory_manager.GetPointer(code_addr)}; // No kernel found, create a new one - auto code{GetShaderCode(memory_manager, code_addr, host_ptr, true)}; - const auto unique_identifier{GetUniqueIdentifier(ShaderType::Compute, false, code)}; + ProgramCode code{GetShaderCode(memory_manager, code_addr, host_ptr, true)}; + const std::size_t code_size{code.size() * sizeof(u64)}; + const u64 unique_identifier{GetUniqueIdentifier(ShaderType::Compute, false, code)}; const ShaderParameters params{system, disk_cache, device, *cpu_addr, host_ptr, unique_identifier}; + std::unique_ptr<Shader> kernel; const auto found = runtime_cache.find(unique_identifier); if (found == runtime_cache.end()) { - kernel = CachedShader::CreateKernelFromMemory(params, std::move(code)); + kernel = Shader::CreateKernelFromMemory(params, std::move(code)); } else { - const std::size_t size_in_bytes = code.size() * sizeof(u64); - kernel = CachedShader::CreateFromCache(params, found->second, size_in_bytes); + kernel = Shader::CreateFromCache(params, found->second); } + Shader* const result = kernel.get(); if (cpu_addr) { - Register(kernel); + Register(std::move(kernel), *cpu_addr, code_size); } else { - null_kernel = kernel; + null_kernel = std::move(kernel); } - return kernel; + return result; } } // namespace OpenGL diff --git a/src/video_core/renderer_opengl/gl_shader_cache.h b/src/video_core/renderer_opengl/gl_shader_cache.h index 91690b470..994aaeaf2 100644 --- a/src/video_core/renderer_opengl/gl_shader_cache.h +++ b/src/video_core/renderer_opengl/gl_shader_cache.h @@ -18,12 +18,12 @@ #include "common/common_types.h" #include "video_core/engines/shader_type.h" -#include "video_core/rasterizer_cache.h" #include "video_core/renderer_opengl/gl_resource_manager.h" #include "video_core/renderer_opengl/gl_shader_decompiler.h" #include "video_core/renderer_opengl/gl_shader_disk_cache.h" #include "video_core/shader/registry.h" #include "video_core/shader/shader_ir.h" +#include "video_core/shader_cache.h" namespace Core { class System; @@ -35,16 +35,19 @@ class EmuWindow; namespace OpenGL { -class CachedShader; class Device; class RasterizerOpenGL; -struct UnspecializedShader; -using Shader = std::shared_ptr<CachedShader>; using Maxwell = Tegra::Engines::Maxwell3D::Regs; +struct ProgramHandle { + OGLProgram source_program; + OGLAssemblyProgram assembly_program; +}; +using ProgramSharedPtr = std::shared_ptr<ProgramHandle>; + struct PrecompiledShader { - std::shared_ptr<OGLProgram> program; + ProgramSharedPtr program; std::shared_ptr<VideoCommon::Shader::Registry> registry; ShaderEntries entries; }; @@ -58,64 +61,56 @@ struct ShaderParameters { u64 unique_identifier; }; -class CachedShader final : public RasterizerCacheObject { +class Shader final { public: - ~CachedShader(); + ~Shader(); /// Gets the GL program handle for the shader GLuint GetHandle() const; - /// Returns the size in bytes of the shader - std::size_t GetSizeInBytes() const override { - return size_in_bytes; - } - /// Gets the shader entries for the shader const ShaderEntries& GetEntries() const { return entries; } - static Shader CreateStageFromMemory(const ShaderParameters& params, - Maxwell::ShaderProgram program_type, - ProgramCode program_code, ProgramCode program_code_b); - static Shader CreateKernelFromMemory(const ShaderParameters& params, ProgramCode code); + static std::unique_ptr<Shader> CreateStageFromMemory(const ShaderParameters& params, + Maxwell::ShaderProgram program_type, + ProgramCode program_code, + ProgramCode program_code_b); + static std::unique_ptr<Shader> CreateKernelFromMemory(const ShaderParameters& params, + ProgramCode code); - static Shader CreateFromCache(const ShaderParameters& params, - const PrecompiledShader& precompiled_shader, - std::size_t size_in_bytes); + static std::unique_ptr<Shader> CreateFromCache(const ShaderParameters& params, + const PrecompiledShader& precompiled_shader); private: - explicit CachedShader(VAddr cpu_addr, std::size_t size_in_bytes, - std::shared_ptr<VideoCommon::Shader::Registry> registry, - ShaderEntries entries, std::shared_ptr<OGLProgram> program); + explicit Shader(std::shared_ptr<VideoCommon::Shader::Registry> registry, ShaderEntries entries, + ProgramSharedPtr program); std::shared_ptr<VideoCommon::Shader::Registry> registry; ShaderEntries entries; - std::size_t size_in_bytes = 0; - std::shared_ptr<OGLProgram> program; + ProgramSharedPtr program; + GLuint handle = 0; }; -class ShaderCacheOpenGL final : public RasterizerCache<Shader> { +class ShaderCacheOpenGL final : public VideoCommon::ShaderCache<Shader> { public: explicit ShaderCacheOpenGL(RasterizerOpenGL& rasterizer, Core::System& system, Core::Frontend::EmuWindow& emu_window, const Device& device); + ~ShaderCacheOpenGL() override; /// Loads disk cache for the current game void LoadDiskCache(const std::atomic_bool& stop_loading, const VideoCore::DiskResourceLoadCallback& callback); /// Gets the current specified shader stage program - Shader GetStageProgram(Maxwell::ShaderProgram program); + Shader* GetStageProgram(Maxwell::ShaderProgram program); /// Gets a compute kernel in the passed address - Shader GetComputeKernel(GPUVAddr code_addr); - -protected: - // We do not have to flush this cache as things in it are never modified by us. - void FlushObjectInner(const Shader& object) override {} + Shader* GetComputeKernel(GPUVAddr code_addr); private: - std::shared_ptr<OGLProgram> GeneratePrecompiledProgram( + ProgramSharedPtr GeneratePrecompiledProgram( const ShaderDiskCacheEntry& entry, const ShaderDiskCachePrecompiled& precompiled_entry, const std::unordered_set<GLenum>& supported_formats); @@ -125,10 +120,10 @@ private: ShaderDiskCacheOpenGL disk_cache; std::unordered_map<u64, PrecompiledShader> runtime_cache; - Shader null_shader{}; - Shader null_kernel{}; + std::unique_ptr<Shader> null_shader; + std::unique_ptr<Shader> null_kernel; - std::array<Shader, Maxwell::MaxShaderProgram> last_shaders; + std::array<Shader*, Maxwell::MaxShaderProgram> last_shaders{}; }; } // namespace OpenGL diff --git a/src/video_core/renderer_opengl/gl_shader_decompiler.cpp b/src/video_core/renderer_opengl/gl_shader_decompiler.cpp index 99fd4ae2c..2c49aeaac 100644 --- a/src/video_core/renderer_opengl/gl_shader_decompiler.cpp +++ b/src/video_core/renderer_opengl/gl_shader_decompiler.cpp @@ -37,6 +37,7 @@ using Tegra::Shader::IpaMode; using Tegra::Shader::IpaSampleMode; using Tegra::Shader::PixelImap; using Tegra::Shader::Register; +using Tegra::Shader::TextureType; using VideoCommon::Shader::BuildTransformFeedback; using VideoCommon::Shader::Registry; @@ -61,8 +62,8 @@ struct TextureDerivates {}; using TextureArgument = std::pair<Type, Node>; using TextureIR = std::variant<TextureOffset, TextureDerivates, TextureArgument>; -constexpr u32 MAX_CONSTBUFFER_ELEMENTS = - static_cast<u32>(Maxwell::MaxConstBufferSize) / (4 * sizeof(float)); +constexpr u32 MAX_CONSTBUFFER_SCALARS = static_cast<u32>(Maxwell::MaxConstBufferSize) / sizeof(u32); +constexpr u32 MAX_CONSTBUFFER_ELEMENTS = MAX_CONSTBUFFER_SCALARS / sizeof(u32); constexpr std::string_view CommonDeclarations = R"(#define ftoi floatBitsToInt #define ftou floatBitsToUint @@ -402,6 +403,13 @@ std::string FlowStackTopName(MetaStackClass stack) { return fmt::format("{}_flow_stack_top", GetFlowStackPrefix(stack)); } +bool UseUnifiedUniforms(const Device& device, const ShaderIR& ir, ShaderType stage) { + const u32 num_ubos = static_cast<u32>(ir.GetConstantBuffers().size()); + // We waste one UBO for emulation + const u32 num_available_ubos = device.GetMaxUniformBuffers(stage) - 1; + return num_ubos > num_available_ubos; +} + struct GenericVaryingDescription { std::string name; u8 first_element = 0; @@ -412,8 +420,9 @@ class GLSLDecompiler final { public: explicit GLSLDecompiler(const Device& device, const ShaderIR& ir, const Registry& registry, ShaderType stage, std::string_view identifier, std::string_view suffix) - : device{device}, ir{ir}, registry{registry}, stage{stage}, - identifier{identifier}, suffix{suffix}, header{ir.GetHeader()} { + : device{device}, ir{ir}, registry{registry}, stage{stage}, identifier{identifier}, + suffix{suffix}, header{ir.GetHeader()}, use_unified_uniforms{ + UseUnifiedUniforms(device, ir, stage)} { if (stage != ShaderType::Compute) { transform_feedback = BuildTransformFeedback(registry.GetGraphicsInfo()); } @@ -518,6 +527,9 @@ private: if (device.HasImageLoadFormatted()) { code.AddLine("#extension GL_EXT_shader_image_load_formatted : require"); } + if (device.HasTextureShadowLod()) { + code.AddLine("#extension GL_EXT_texture_shadow_lod : require"); + } if (device.HasWarpIntrinsics()) { code.AddLine("#extension GL_NV_gpu_shader5 : require"); code.AddLine("#extension GL_NV_shader_thread_group : require"); @@ -618,7 +630,9 @@ private: break; } } - if (stage != ShaderType::Vertex || device.HasVertexViewportLayer()) { + + if (stage != ShaderType::Geometry && + (stage != ShaderType::Vertex || device.HasVertexViewportLayer())) { if (ir.UsesLayer()) { code.AddLine("int gl_Layer;"); } @@ -647,6 +661,16 @@ private: --code.scope; code.AddLine("}};"); code.AddNewLine(); + + if (stage == ShaderType::Geometry) { + if (ir.UsesLayer()) { + code.AddLine("out int gl_Layer;"); + } + if (ir.UsesViewportIndex()) { + code.AddLine("out int gl_ViewportIndex;"); + } + } + code.AddNewLine(); } void DeclareRegisters() { @@ -834,12 +858,24 @@ private: } void DeclareConstantBuffers() { + if (use_unified_uniforms) { + const u32 binding = device.GetBaseBindings(stage).shader_storage_buffer + + static_cast<u32>(ir.GetGlobalMemory().size()); + code.AddLine("layout (std430, binding = {}) readonly buffer UnifiedUniforms {{", + binding); + code.AddLine(" uint cbufs[];"); + code.AddLine("}};"); + code.AddNewLine(); + return; + } + u32 binding = device.GetBaseBindings(stage).uniform_buffer; - for (const auto& buffers : ir.GetConstantBuffers()) { - const auto index = buffers.first; + for (const auto [index, info] : ir.GetConstantBuffers()) { + const u32 num_elements = Common::AlignUp(info.GetSize(), 4) / 4; + const u32 size = info.IsIndirect() ? MAX_CONSTBUFFER_ELEMENTS : num_elements; code.AddLine("layout (std140, binding = {}) uniform {} {{", binding++, GetConstBufferBlock(index)); - code.AddLine(" uvec4 {}[{}];", GetConstBuffer(index), MAX_CONSTBUFFER_ELEMENTS); + code.AddLine(" uvec4 {}[{}];", GetConstBuffer(index), size); code.AddLine("}};"); code.AddNewLine(); } @@ -877,13 +913,13 @@ private: return "samplerBuffer"; } switch (sampler.type) { - case Tegra::Shader::TextureType::Texture1D: + case TextureType::Texture1D: return "sampler1D"; - case Tegra::Shader::TextureType::Texture2D: + case TextureType::Texture2D: return "sampler2D"; - case Tegra::Shader::TextureType::Texture3D: + case TextureType::Texture3D: return "sampler3D"; - case Tegra::Shader::TextureType::TextureCube: + case TextureType::TextureCube: return "samplerCube"; default: UNREACHABLE(); @@ -1038,42 +1074,51 @@ private: if (const auto cbuf = std::get_if<CbufNode>(&*node)) { const Node offset = cbuf->GetOffset(); + const u32 base_unified_offset = cbuf->GetIndex() * MAX_CONSTBUFFER_SCALARS; + if (const auto immediate = std::get_if<ImmediateNode>(&*offset)) { // Direct access const u32 offset_imm = immediate->GetValue(); ASSERT_MSG(offset_imm % 4 == 0, "Unaligned cbuf direct access"); - return {fmt::format("{}[{}][{}]", GetConstBuffer(cbuf->GetIndex()), - offset_imm / (4 * 4), (offset_imm / 4) % 4), - Type::Uint}; + if (use_unified_uniforms) { + return {fmt::format("cbufs[{}]", base_unified_offset + offset_imm / 4), + Type::Uint}; + } else { + return {fmt::format("{}[{}][{}]", GetConstBuffer(cbuf->GetIndex()), + offset_imm / (4 * 4), (offset_imm / 4) % 4), + Type::Uint}; + } } - if (std::holds_alternative<OperationNode>(*offset)) { - // Indirect access - const std::string final_offset = code.GenerateTemporary(); - code.AddLine("uint {} = {} >> 2;", final_offset, Visit(offset).AsUint()); + // Indirect access + if (use_unified_uniforms) { + return {fmt::format("cbufs[{} + ({} >> 2)]", base_unified_offset, + Visit(offset).AsUint()), + Type::Uint}; + } - if (!device.HasComponentIndexingBug()) { - return {fmt::format("{}[{} >> 2][{} & 3]", GetConstBuffer(cbuf->GetIndex()), - final_offset, final_offset), - Type::Uint}; - } + const std::string final_offset = code.GenerateTemporary(); + code.AddLine("uint {} = {} >> 2;", final_offset, Visit(offset).AsUint()); - // AMD's proprietary GLSL compiler emits ill code for variable component access. - // To bypass this driver bug generate 4 ifs, one per each component. - const std::string pack = code.GenerateTemporary(); - code.AddLine("uvec4 {} = {}[{} >> 2];", pack, GetConstBuffer(cbuf->GetIndex()), - final_offset); - - const std::string result = code.GenerateTemporary(); - code.AddLine("uint {};", result); - for (u32 swizzle = 0; swizzle < 4; ++swizzle) { - code.AddLine("if (({} & 3) == {}) {} = {}{};", final_offset, swizzle, result, - pack, GetSwizzle(swizzle)); - } - return {result, Type::Uint}; + if (!device.HasComponentIndexingBug()) { + return {fmt::format("{}[{} >> 2][{} & 3]", GetConstBuffer(cbuf->GetIndex()), + final_offset, final_offset), + Type::Uint}; } - UNREACHABLE_MSG("Unmanaged offset node type"); + // AMD's proprietary GLSL compiler emits ill code for variable component access. + // To bypass this driver bug generate 4 ifs, one per each component. + const std::string pack = code.GenerateTemporary(); + code.AddLine("uvec4 {} = {}[{} >> 2];", pack, GetConstBuffer(cbuf->GetIndex()), + final_offset); + + const std::string result = code.GenerateTemporary(); + code.AddLine("uint {};", result); + for (u32 swizzle = 0; swizzle < 4; ++swizzle) { + code.AddLine("if (({} & 3) == {}) {} = {}{};", final_offset, swizzle, result, pack, + GetSwizzle(swizzle)); + } + return {result, Type::Uint}; } if (const auto gmem = std::get_if<GmemNode>(&*node)) { @@ -1339,8 +1384,19 @@ private: const std::size_t count = operation.GetOperandsCount(); const bool has_array = meta->sampler.is_array; const bool has_shadow = meta->sampler.is_shadow; + const bool workaround_lod_array_shadow_as_grad = + !device.HasTextureShadowLod() && function_suffix == "Lod" && meta->sampler.is_shadow && + ((meta->sampler.type == TextureType::Texture2D && meta->sampler.is_array) || + meta->sampler.type == TextureType::TextureCube); + + std::string expr = "texture"; + + if (workaround_lod_array_shadow_as_grad) { + expr += "Grad"; + } else { + expr += function_suffix; + } - std::string expr = "texture" + function_suffix; if (!meta->aoffi.empty()) { expr += "Offset"; } else if (!meta->ptp.empty()) { @@ -1374,6 +1430,16 @@ private: expr += ')'; } + if (workaround_lod_array_shadow_as_grad) { + switch (meta->sampler.type) { + case TextureType::Texture2D: + return expr + ", vec2(0.0), vec2(0.0))"; + case TextureType::TextureCube: + return expr + ", vec3(0.0), vec3(0.0))"; + } + UNREACHABLE(); + } + for (const auto& variant : extras) { if (const auto argument = std::get_if<TextureArgument>(&variant)) { expr += GenerateTextureArgument(*argument); @@ -1538,7 +1604,9 @@ private: Expression target; if (const auto gpr = std::get_if<GprNode>(&*dest)) { if (gpr->GetIndex() == Register::ZeroIndex) { - // Writing to Register::ZeroIndex is a no op + // Writing to Register::ZeroIndex is a no op but we still have to visit the source + // as it might have side effects. + code.AddLine("{};", Visit(src).GetCode()); return {}; } target = {GetRegister(gpr->GetIndex()), Type::Float}; @@ -1840,34 +1908,40 @@ private: Type::HalfFloat}; } - template <Type type> - Expression LogicalLessThan(Operation operation) { - return GenerateBinaryInfix(operation, "<", Type::Bool, type, type); - } + template <const std::string_view& op, Type type, bool unordered = false> + Expression Comparison(Operation operation) { + static_assert(!unordered || type == Type::Float); - template <Type type> - Expression LogicalEqual(Operation operation) { - return GenerateBinaryInfix(operation, "==", Type::Bool, type, type); - } - - template <Type type> - Expression LogicalLessEqual(Operation operation) { - return GenerateBinaryInfix(operation, "<=", Type::Bool, type, type); - } + const Expression expr = GenerateBinaryInfix(operation, op, Type::Bool, type, type); - template <Type type> - Expression LogicalGreaterThan(Operation operation) { - return GenerateBinaryInfix(operation, ">", Type::Bool, type, type); + if constexpr (op.compare("!=") == 0 && type == Type::Float && !unordered) { + // GLSL's operator!=(float, float) doesn't seem be ordered. This happens on both AMD's + // and Nvidia's proprietary stacks. Manually force an ordered comparison. + return {fmt::format("({} && !isnan({}) && !isnan({}))", expr.AsBool(), + VisitOperand(operation, 0).AsFloat(), + VisitOperand(operation, 1).AsFloat()), + Type::Bool}; + } + if constexpr (!unordered) { + return expr; + } + // Unordered comparisons are always true for NaN operands. + return {fmt::format("({} || isnan({}) || isnan({}))", expr.AsBool(), + VisitOperand(operation, 0).AsFloat(), + VisitOperand(operation, 1).AsFloat()), + Type::Bool}; } - template <Type type> - Expression LogicalNotEqual(Operation operation) { - return GenerateBinaryInfix(operation, "!=", Type::Bool, type, type); + Expression FOrdered(Operation operation) { + return {fmt::format("(!isnan({}) && !isnan({}))", VisitOperand(operation, 0).AsFloat(), + VisitOperand(operation, 1).AsFloat()), + Type::Bool}; } - template <Type type> - Expression LogicalGreaterEqual(Operation operation) { - return GenerateBinaryInfix(operation, ">=", Type::Bool, type, type); + Expression FUnordered(Operation operation) { + return {fmt::format("(isnan({}) || isnan({}))", VisitOperand(operation, 0).AsFloat(), + VisitOperand(operation, 1).AsFloat()), + Type::Bool}; } Expression LogicalAddCarry(Operation operation) { @@ -1992,8 +2066,19 @@ private: const auto meta = std::get_if<MetaTexture>(&operation.GetMeta()); ASSERT(meta); - std::string expr = GenerateTexture( - operation, "Lod", {TextureArgument{Type::Float, meta->lod}, TextureOffset{}}); + std::string expr{}; + + if (!device.HasTextureShadowLod() && meta->sampler.is_shadow && + ((meta->sampler.type == TextureType::Texture2D && meta->sampler.is_array) || + meta->sampler.type == TextureType::TextureCube)) { + LOG_ERROR(Render_OpenGL, + "Device lacks GL_EXT_texture_shadow_lod, using textureGrad as a workaround"); + expr = GenerateTexture(operation, "Lod", {}); + } else { + expr = GenerateTexture(operation, "Lod", + {TextureArgument{Type::Float, meta->lod}, TextureOffset{}}); + } + if (meta->sampler.is_shadow) { expr = "vec4(" + expr + ')'; } @@ -2303,6 +2388,18 @@ private: return {"gl_SubGroupInvocationARB", Type::Uint}; } + template <const std::string_view& comparison> + Expression ThreadMask(Operation) { + if (device.HasWarpIntrinsics()) { + return {fmt::format("gl_Thread{}MaskNV", comparison), Type::Uint}; + } + if (device.HasShaderBallot()) { + return {fmt::format("uint(gl_SubGroup{}MaskARB)", comparison), Type::Uint}; + } + LOG_ERROR(Render_OpenGL, "Thread mask intrinsics are required by the shader"); + return {"0U", Type::Uint}; + } + Expression ShuffleIndexed(Operation operation) { std::string value = VisitOperand(operation, 0).AsFloat(); @@ -2315,7 +2412,21 @@ private: return {fmt::format("readInvocationARB({}, {})", value, index), Type::Float}; } - Expression MemoryBarrierGL(Operation) { + Expression Barrier(Operation) { + if (!ir.IsDecompiled()) { + LOG_ERROR(Render_OpenGL, "barrier() used but shader is not decompiled"); + return {}; + } + code.AddLine("barrier();"); + return {}; + } + + Expression MemoryBarrierGroup(Operation) { + code.AddLine("groupMemoryBarrier();"); + return {}; + } + + Expression MemoryBarrierGlobal(Operation) { code.AddLine("memoryBarrier();"); return {}; } @@ -2324,6 +2435,19 @@ private: Func() = delete; ~Func() = delete; + static constexpr std::string_view LessThan = "<"; + static constexpr std::string_view Equal = "=="; + static constexpr std::string_view LessEqual = "<="; + static constexpr std::string_view GreaterThan = ">"; + static constexpr std::string_view NotEqual = "!="; + static constexpr std::string_view GreaterEqual = ">="; + + static constexpr std::string_view Eq = "Eq"; + static constexpr std::string_view Ge = "Ge"; + static constexpr std::string_view Gt = "Gt"; + static constexpr std::string_view Le = "Le"; + static constexpr std::string_view Lt = "Lt"; + static constexpr std::string_view Add = "Add"; static constexpr std::string_view Min = "Min"; static constexpr std::string_view Max = "Max"; @@ -2425,27 +2549,34 @@ private: &GLSLDecompiler::LogicalPick2, &GLSLDecompiler::LogicalAnd2, - &GLSLDecompiler::LogicalLessThan<Type::Float>, - &GLSLDecompiler::LogicalEqual<Type::Float>, - &GLSLDecompiler::LogicalLessEqual<Type::Float>, - &GLSLDecompiler::LogicalGreaterThan<Type::Float>, - &GLSLDecompiler::LogicalNotEqual<Type::Float>, - &GLSLDecompiler::LogicalGreaterEqual<Type::Float>, - &GLSLDecompiler::LogicalFIsNan, - - &GLSLDecompiler::LogicalLessThan<Type::Int>, - &GLSLDecompiler::LogicalEqual<Type::Int>, - &GLSLDecompiler::LogicalLessEqual<Type::Int>, - &GLSLDecompiler::LogicalGreaterThan<Type::Int>, - &GLSLDecompiler::LogicalNotEqual<Type::Int>, - &GLSLDecompiler::LogicalGreaterEqual<Type::Int>, - - &GLSLDecompiler::LogicalLessThan<Type::Uint>, - &GLSLDecompiler::LogicalEqual<Type::Uint>, - &GLSLDecompiler::LogicalLessEqual<Type::Uint>, - &GLSLDecompiler::LogicalGreaterThan<Type::Uint>, - &GLSLDecompiler::LogicalNotEqual<Type::Uint>, - &GLSLDecompiler::LogicalGreaterEqual<Type::Uint>, + &GLSLDecompiler::Comparison<Func::LessThan, Type::Float, false>, + &GLSLDecompiler::Comparison<Func::Equal, Type::Float, false>, + &GLSLDecompiler::Comparison<Func::LessEqual, Type::Float, false>, + &GLSLDecompiler::Comparison<Func::GreaterThan, Type::Float, false>, + &GLSLDecompiler::Comparison<Func::NotEqual, Type::Float, false>, + &GLSLDecompiler::Comparison<Func::GreaterEqual, Type::Float, false>, + &GLSLDecompiler::FOrdered, + &GLSLDecompiler::FUnordered, + &GLSLDecompiler::Comparison<Func::LessThan, Type::Float, true>, + &GLSLDecompiler::Comparison<Func::Equal, Type::Float, true>, + &GLSLDecompiler::Comparison<Func::LessEqual, Type::Float, true>, + &GLSLDecompiler::Comparison<Func::GreaterThan, Type::Float, true>, + &GLSLDecompiler::Comparison<Func::NotEqual, Type::Float, true>, + &GLSLDecompiler::Comparison<Func::GreaterEqual, Type::Float, true>, + + &GLSLDecompiler::Comparison<Func::LessThan, Type::Int>, + &GLSLDecompiler::Comparison<Func::Equal, Type::Int>, + &GLSLDecompiler::Comparison<Func::LessEqual, Type::Int>, + &GLSLDecompiler::Comparison<Func::GreaterThan, Type::Int>, + &GLSLDecompiler::Comparison<Func::NotEqual, Type::Int>, + &GLSLDecompiler::Comparison<Func::GreaterEqual, Type::Int>, + + &GLSLDecompiler::Comparison<Func::LessThan, Type::Uint>, + &GLSLDecompiler::Comparison<Func::Equal, Type::Uint>, + &GLSLDecompiler::Comparison<Func::LessEqual, Type::Uint>, + &GLSLDecompiler::Comparison<Func::GreaterThan, Type::Uint>, + &GLSLDecompiler::Comparison<Func::NotEqual, Type::Uint>, + &GLSLDecompiler::Comparison<Func::GreaterEqual, Type::Uint>, &GLSLDecompiler::LogicalAddCarry, @@ -2534,9 +2665,16 @@ private: &GLSLDecompiler::VoteEqual, &GLSLDecompiler::ThreadId, + &GLSLDecompiler::ThreadMask<Func::Eq>, + &GLSLDecompiler::ThreadMask<Func::Ge>, + &GLSLDecompiler::ThreadMask<Func::Gt>, + &GLSLDecompiler::ThreadMask<Func::Le>, + &GLSLDecompiler::ThreadMask<Func::Lt>, &GLSLDecompiler::ShuffleIndexed, - &GLSLDecompiler::MemoryBarrierGL, + &GLSLDecompiler::Barrier, + &GLSLDecompiler::MemoryBarrierGroup, + &GLSLDecompiler::MemoryBarrierGlobal, }; static_assert(operation_decompilers.size() == static_cast<std::size_t>(OperationCode::Amount)); @@ -2649,6 +2787,7 @@ private: const std::string_view identifier; const std::string_view suffix; const Header header; + const bool use_unified_uniforms; std::unordered_map<u8, VaryingTFB> transform_feedback; ShaderWriter code; @@ -2844,7 +2983,7 @@ void GLSLDecompiler::DecompileAST() { } // Anonymous namespace -ShaderEntries MakeEntries(const VideoCommon::Shader::ShaderIR& ir) { +ShaderEntries MakeEntries(const Device& device, const ShaderIR& ir, ShaderType stage) { ShaderEntries entries; for (const auto& cbuf : ir.GetConstantBuffers()) { entries.const_buffers.emplace_back(cbuf.second.GetMaxOffset(), cbuf.second.IsIndirect(), @@ -2865,6 +3004,7 @@ ShaderEntries MakeEntries(const VideoCommon::Shader::ShaderIR& ir) { entries.clip_distances = (clip_distances[i] ? 1U : 0U) << i; } entries.shader_length = ir.GetLength(); + entries.use_unified_uniforms = UseUnifiedUniforms(device, ir, stage); return entries; } diff --git a/src/video_core/renderer_opengl/gl_shader_decompiler.h b/src/video_core/renderer_opengl/gl_shader_decompiler.h index e8a178764..451c9689a 100644 --- a/src/video_core/renderer_opengl/gl_shader_decompiler.h +++ b/src/video_core/renderer_opengl/gl_shader_decompiler.h @@ -53,11 +53,13 @@ struct ShaderEntries { std::vector<GlobalMemoryEntry> global_memory_entries; std::vector<SamplerEntry> samplers; std::vector<ImageEntry> images; - u32 clip_distances{}; std::size_t shader_length{}; + u32 clip_distances{}; + bool use_unified_uniforms{}; }; -ShaderEntries MakeEntries(const VideoCommon::Shader::ShaderIR& ir); +ShaderEntries MakeEntries(const Device& device, const VideoCommon::Shader::ShaderIR& ir, + Tegra::Engines::ShaderType stage); std::string DecompileShader(const Device& device, const VideoCommon::Shader::ShaderIR& ir, const VideoCommon::Shader::Registry& registry, diff --git a/src/video_core/renderer_opengl/gl_shader_disk_cache.cpp b/src/video_core/renderer_opengl/gl_shader_disk_cache.cpp index 9e95a122b..653c3f2f9 100644 --- a/src/video_core/renderer_opengl/gl_shader_disk_cache.cpp +++ b/src/video_core/renderer_opengl/gl_shader_disk_cache.cpp @@ -29,6 +29,8 @@ using VideoCommon::Shader::KeyMap; namespace { +using VideoCommon::Shader::SeparateSamplerKey; + using ShaderCacheVersionHash = std::array<u8, 64>; struct ConstBufferKey { @@ -37,18 +39,26 @@ struct ConstBufferKey { u32 value = 0; }; -struct BoundSamplerKey { +struct BoundSamplerEntry { u32 offset = 0; Tegra::Engines::SamplerDescriptor sampler; }; -struct BindlessSamplerKey { +struct SeparateSamplerEntry { + u32 cbuf1 = 0; + u32 cbuf2 = 0; + u32 offset1 = 0; + u32 offset2 = 0; + Tegra::Engines::SamplerDescriptor sampler; +}; + +struct BindlessSamplerEntry { u32 cbuf = 0; u32 offset = 0; Tegra::Engines::SamplerDescriptor sampler; }; -constexpr u32 NativeVersion = 20; +constexpr u32 NativeVersion = 21; ShaderCacheVersionHash GetShaderCacheVersionHash() { ShaderCacheVersionHash hash{}; @@ -87,12 +97,14 @@ bool ShaderDiskCacheEntry::Load(FileUtil::IOFile& file) { u32 texture_handler_size_value; u32 num_keys; u32 num_bound_samplers; + u32 num_separate_samplers; u32 num_bindless_samplers; if (file.ReadArray(&unique_identifier, 1) != 1 || file.ReadArray(&bound_buffer, 1) != 1 || file.ReadArray(&is_texture_handler_size_known, 1) != 1 || file.ReadArray(&texture_handler_size_value, 1) != 1 || file.ReadArray(&graphics_info, 1) != 1 || file.ReadArray(&compute_info, 1) != 1 || file.ReadArray(&num_keys, 1) != 1 || file.ReadArray(&num_bound_samplers, 1) != 1 || + file.ReadArray(&num_separate_samplers, 1) != 1 || file.ReadArray(&num_bindless_samplers, 1) != 1) { return false; } @@ -101,23 +113,32 @@ bool ShaderDiskCacheEntry::Load(FileUtil::IOFile& file) { } std::vector<ConstBufferKey> flat_keys(num_keys); - std::vector<BoundSamplerKey> flat_bound_samplers(num_bound_samplers); - std::vector<BindlessSamplerKey> flat_bindless_samplers(num_bindless_samplers); + std::vector<BoundSamplerEntry> flat_bound_samplers(num_bound_samplers); + std::vector<SeparateSamplerEntry> flat_separate_samplers(num_separate_samplers); + std::vector<BindlessSamplerEntry> flat_bindless_samplers(num_bindless_samplers); if (file.ReadArray(flat_keys.data(), flat_keys.size()) != flat_keys.size() || file.ReadArray(flat_bound_samplers.data(), flat_bound_samplers.size()) != flat_bound_samplers.size() || + file.ReadArray(flat_separate_samplers.data(), flat_separate_samplers.size()) != + flat_separate_samplers.size() || file.ReadArray(flat_bindless_samplers.data(), flat_bindless_samplers.size()) != flat_bindless_samplers.size()) { return false; } - for (const auto& key : flat_keys) { - keys.insert({{key.cbuf, key.offset}, key.value}); + for (const auto& entry : flat_keys) { + keys.insert({{entry.cbuf, entry.offset}, entry.value}); } - for (const auto& key : flat_bound_samplers) { - bound_samplers.emplace(key.offset, key.sampler); + for (const auto& entry : flat_bound_samplers) { + bound_samplers.emplace(entry.offset, entry.sampler); } - for (const auto& key : flat_bindless_samplers) { - bindless_samplers.insert({{key.cbuf, key.offset}, key.sampler}); + for (const auto& entry : flat_separate_samplers) { + SeparateSamplerKey key; + key.buffers = {entry.cbuf1, entry.cbuf2}; + key.offsets = {entry.offset1, entry.offset2}; + separate_samplers.emplace(key, entry.sampler); + } + for (const auto& entry : flat_bindless_samplers) { + bindless_samplers.insert({{entry.cbuf, entry.offset}, entry.sampler}); } return true; @@ -142,6 +163,7 @@ bool ShaderDiskCacheEntry::Save(FileUtil::IOFile& file) const { file.WriteObject(graphics_info) != 1 || file.WriteObject(compute_info) != 1 || file.WriteObject(static_cast<u32>(keys.size())) != 1 || file.WriteObject(static_cast<u32>(bound_samplers.size())) != 1 || + file.WriteObject(static_cast<u32>(separate_samplers.size())) != 1 || file.WriteObject(static_cast<u32>(bindless_samplers.size())) != 1) { return false; } @@ -152,22 +174,34 @@ bool ShaderDiskCacheEntry::Save(FileUtil::IOFile& file) const { flat_keys.push_back(ConstBufferKey{address.first, address.second, value}); } - std::vector<BoundSamplerKey> flat_bound_samplers; + std::vector<BoundSamplerEntry> flat_bound_samplers; flat_bound_samplers.reserve(bound_samplers.size()); for (const auto& [address, sampler] : bound_samplers) { - flat_bound_samplers.push_back(BoundSamplerKey{address, sampler}); + flat_bound_samplers.push_back(BoundSamplerEntry{address, sampler}); + } + + std::vector<SeparateSamplerEntry> flat_separate_samplers; + flat_separate_samplers.reserve(separate_samplers.size()); + for (const auto& [key, sampler] : separate_samplers) { + SeparateSamplerEntry entry; + std::tie(entry.cbuf1, entry.cbuf2) = key.buffers; + std::tie(entry.offset1, entry.offset2) = key.offsets; + entry.sampler = sampler; + flat_separate_samplers.push_back(entry); } - std::vector<BindlessSamplerKey> flat_bindless_samplers; + std::vector<BindlessSamplerEntry> flat_bindless_samplers; flat_bindless_samplers.reserve(bindless_samplers.size()); for (const auto& [address, sampler] : bindless_samplers) { flat_bindless_samplers.push_back( - BindlessSamplerKey{address.first, address.second, sampler}); + BindlessSamplerEntry{address.first, address.second, sampler}); } return file.WriteArray(flat_keys.data(), flat_keys.size()) == flat_keys.size() && file.WriteArray(flat_bound_samplers.data(), flat_bound_samplers.size()) == flat_bound_samplers.size() && + file.WriteArray(flat_separate_samplers.data(), flat_separate_samplers.size()) == + flat_separate_samplers.size() && file.WriteArray(flat_bindless_samplers.data(), flat_bindless_samplers.size()) == flat_bindless_samplers.size(); } diff --git a/src/video_core/renderer_opengl/gl_shader_disk_cache.h b/src/video_core/renderer_opengl/gl_shader_disk_cache.h index d5be52e40..a79cef0e9 100644 --- a/src/video_core/renderer_opengl/gl_shader_disk_cache.h +++ b/src/video_core/renderer_opengl/gl_shader_disk_cache.h @@ -57,6 +57,7 @@ struct ShaderDiskCacheEntry { VideoCommon::Shader::ComputeInfo compute_info; VideoCommon::Shader::KeyMap keys; VideoCommon::Shader::BoundSamplerMap bound_samplers; + VideoCommon::Shader::SeparateSamplerMap separate_samplers; VideoCommon::Shader::BindlessSamplerMap bindless_samplers; }; diff --git a/src/video_core/renderer_opengl/gl_shader_manager.cpp b/src/video_core/renderer_opengl/gl_shader_manager.cpp index 9c7b0adbd..8e754fa90 100644 --- a/src/video_core/renderer_opengl/gl_shader_manager.cpp +++ b/src/video_core/renderer_opengl/gl_shader_manager.cpp @@ -6,45 +6,109 @@ #include "common/common_types.h" #include "video_core/engines/maxwell_3d.h" +#include "video_core/renderer_opengl/gl_device.h" #include "video_core/renderer_opengl/gl_shader_manager.h" -namespace OpenGL::GLShader { +namespace OpenGL { -ProgramManager::ProgramManager() = default; +ProgramManager::ProgramManager(const Device& device) { + use_assembly_programs = device.UseAssemblyShaders(); + if (use_assembly_programs) { + glEnable(GL_COMPUTE_PROGRAM_NV); + } else { + graphics_pipeline.Create(); + glBindProgramPipeline(graphics_pipeline.handle); + } +} ProgramManager::~ProgramManager() = default; -void ProgramManager::Create() { - graphics_pipeline.Create(); - glBindProgramPipeline(graphics_pipeline.handle); +void ProgramManager::BindCompute(GLuint program) { + if (use_assembly_programs) { + glBindProgramARB(GL_COMPUTE_PROGRAM_NV, program); + } else { + is_graphics_bound = false; + glUseProgram(program); + } } void ProgramManager::BindGraphicsPipeline() { - if (!is_graphics_bound) { - is_graphics_bound = true; - glUseProgram(0); + if (use_assembly_programs) { + UpdateAssemblyPrograms(); + } else { + UpdateSourcePrograms(); } +} - // Avoid updating the pipeline when values have no changed - if (old_state == current_state) { - return; +void ProgramManager::BindHostPipeline(GLuint pipeline) { + if (use_assembly_programs) { + if (geometry_enabled) { + geometry_enabled = false; + old_state.geometry = 0; + glDisable(GL_GEOMETRY_PROGRAM_NV); + } + } else { + if (!is_graphics_bound) { + glUseProgram(0); + } } + glBindProgramPipeline(pipeline); +} - // Workaround for AMD bug - static constexpr GLenum all_used_stages{GL_VERTEX_SHADER_BIT | GL_GEOMETRY_SHADER_BIT | - GL_FRAGMENT_SHADER_BIT}; - const GLuint handle = graphics_pipeline.handle; - glUseProgramStages(handle, all_used_stages, 0); - glUseProgramStages(handle, GL_VERTEX_SHADER_BIT, current_state.vertex_shader); - glUseProgramStages(handle, GL_GEOMETRY_SHADER_BIT, current_state.geometry_shader); - glUseProgramStages(handle, GL_FRAGMENT_SHADER_BIT, current_state.fragment_shader); +void ProgramManager::RestoreGuestPipeline() { + if (use_assembly_programs) { + glBindProgramPipeline(0); + } else { + glBindProgramPipeline(graphics_pipeline.handle); + } +} + +void ProgramManager::UpdateAssemblyPrograms() { + const auto update_state = [](GLenum stage, bool& enabled, GLuint current, GLuint old) { + if (current == old) { + return; + } + if (current == 0) { + if (enabled) { + enabled = false; + glDisable(stage); + } + return; + } + if (!enabled) { + enabled = true; + glEnable(stage); + } + glBindProgramARB(stage, current); + }; + + update_state(GL_VERTEX_PROGRAM_NV, vertex_enabled, current_state.vertex, old_state.vertex); + update_state(GL_GEOMETRY_PROGRAM_NV, geometry_enabled, current_state.geometry, + old_state.geometry); + update_state(GL_FRAGMENT_PROGRAM_NV, fragment_enabled, current_state.fragment, + old_state.fragment); old_state = current_state; } -void ProgramManager::BindComputeShader(GLuint program) { - is_graphics_bound = false; - glUseProgram(program); +void ProgramManager::UpdateSourcePrograms() { + if (!is_graphics_bound) { + is_graphics_bound = true; + glUseProgram(0); + } + + const GLuint handle = graphics_pipeline.handle; + const auto update_state = [handle](GLenum stage, GLuint current, GLuint old) { + if (current == old) { + return; + } + glUseProgramStages(handle, stage, current); + }; + update_state(GL_VERTEX_SHADER_BIT, current_state.vertex, old_state.vertex); + update_state(GL_GEOMETRY_SHADER_BIT, current_state.geometry, old_state.geometry); + update_state(GL_FRAGMENT_SHADER_BIT, current_state.fragment, old_state.fragment); + + old_state = current_state; } void MaxwellUniformData::SetFromRegs(const Tegra::Engines::Maxwell3D& maxwell) { @@ -54,4 +118,4 @@ void MaxwellUniformData::SetFromRegs(const Tegra::Engines::Maxwell3D& maxwell) { y_direction = regs.screen_y_control.y_negate == 0 ? 1.0f : -1.0f; } -} // namespace OpenGL::GLShader +} // namespace OpenGL diff --git a/src/video_core/renderer_opengl/gl_shader_manager.h b/src/video_core/renderer_opengl/gl_shader_manager.h index d2e47f2a9..0f03b4f12 100644 --- a/src/video_core/renderer_opengl/gl_shader_manager.h +++ b/src/video_core/renderer_opengl/gl_shader_manager.h @@ -11,7 +11,9 @@ #include "video_core/renderer_opengl/gl_resource_manager.h" #include "video_core/renderer_opengl/maxwell_to_gl.h" -namespace OpenGL::GLShader { +namespace OpenGL { + +class Device; /// Uniform structure for the Uniform Buffer Object, all vectors must be 16-byte aligned /// @note Always keep a vec4 at the end. The GL spec is not clear whether the alignment at @@ -28,50 +30,58 @@ static_assert(sizeof(MaxwellUniformData) < 16384, class ProgramManager { public: - explicit ProgramManager(); + explicit ProgramManager(const Device& device); ~ProgramManager(); - void Create(); + /// Binds a compute program + void BindCompute(GLuint program); - /// Updates the graphics pipeline and binds it. + /// Updates bound programs. void BindGraphicsPipeline(); - /// Binds a compute shader. - void BindComputeShader(GLuint program); + /// Binds an OpenGL pipeline object unsynchronized with the guest state. + void BindHostPipeline(GLuint pipeline); + + /// Rewinds BindHostPipeline state changes. + void RestoreGuestPipeline(); void UseVertexShader(GLuint program) { - current_state.vertex_shader = program; + current_state.vertex = program; } void UseGeometryShader(GLuint program) { - current_state.geometry_shader = program; + current_state.geometry = program; } void UseFragmentShader(GLuint program) { - current_state.fragment_shader = program; + current_state.fragment = program; } private: struct PipelineState { - bool operator==(const PipelineState& rhs) const noexcept { - return vertex_shader == rhs.vertex_shader && fragment_shader == rhs.fragment_shader && - geometry_shader == rhs.geometry_shader; - } - - bool operator!=(const PipelineState& rhs) const noexcept { - return !operator==(rhs); - } - - GLuint vertex_shader = 0; - GLuint fragment_shader = 0; - GLuint geometry_shader = 0; + GLuint vertex = 0; + GLuint geometry = 0; + GLuint fragment = 0; }; + /// Update NV_gpu_program5 programs. + void UpdateAssemblyPrograms(); + + /// Update GLSL programs. + void UpdateSourcePrograms(); + OGLPipeline graphics_pipeline; - OGLPipeline compute_pipeline; + PipelineState current_state; PipelineState old_state; + + bool use_assembly_programs = false; + bool is_graphics_bound = true; + + bool vertex_enabled = false; + bool geometry_enabled = false; + bool fragment_enabled = false; }; -} // namespace OpenGL::GLShader +} // namespace OpenGL diff --git a/src/video_core/renderer_opengl/gl_stream_buffer.cpp b/src/video_core/renderer_opengl/gl_stream_buffer.cpp index 6ec328c53..3655ff629 100644 --- a/src/video_core/renderer_opengl/gl_stream_buffer.cpp +++ b/src/video_core/renderer_opengl/gl_stream_buffer.cpp @@ -2,11 +2,13 @@ // Licensed under GPLv2 or any later version // Refer to the license.txt file included. -#include <deque> +#include <tuple> #include <vector> + #include "common/alignment.h" #include "common/assert.h" #include "common/microprofile.h" +#include "video_core/renderer_opengl/gl_device.h" #include "video_core/renderer_opengl/gl_stream_buffer.h" MICROPROFILE_DEFINE(OpenGL_StreamBuffer, "OpenGL", "Stream Buffer Orphaning", @@ -14,8 +16,7 @@ MICROPROFILE_DEFINE(OpenGL_StreamBuffer, "OpenGL", "Stream Buffer Orphaning", namespace OpenGL { -OGLStreamBuffer::OGLStreamBuffer(GLsizeiptr size, bool vertex_data_usage, bool prefer_coherent, - bool use_persistent) +OGLStreamBuffer::OGLStreamBuffer(const Device& device, GLsizeiptr size, bool vertex_data_usage) : buffer_size(size) { gl_buffer.Create(); @@ -29,34 +30,22 @@ OGLStreamBuffer::OGLStreamBuffer(GLsizeiptr size, bool vertex_data_usage, bool p allocate_size *= 2; } - if (use_persistent) { - persistent = true; - coherent = prefer_coherent; - const GLbitfield flags = - GL_MAP_WRITE_BIT | GL_MAP_PERSISTENT_BIT | (coherent ? GL_MAP_COHERENT_BIT : 0); - glNamedBufferStorage(gl_buffer.handle, allocate_size, nullptr, flags); - mapped_ptr = static_cast<u8*>(glMapNamedBufferRange( - gl_buffer.handle, 0, buffer_size, flags | (coherent ? 0 : GL_MAP_FLUSH_EXPLICIT_BIT))); - } else { - glNamedBufferData(gl_buffer.handle, allocate_size, nullptr, GL_STREAM_DRAW); + static constexpr GLbitfield flags = GL_MAP_WRITE_BIT | GL_MAP_PERSISTENT_BIT; + glNamedBufferStorage(gl_buffer.handle, allocate_size, nullptr, flags); + mapped_ptr = static_cast<u8*>( + glMapNamedBufferRange(gl_buffer.handle, 0, buffer_size, flags | GL_MAP_FLUSH_EXPLICIT_BIT)); + + if (device.HasVertexBufferUnifiedMemory()) { + glMakeNamedBufferResidentNV(gl_buffer.handle, GL_READ_ONLY); + glGetNamedBufferParameterui64vNV(gl_buffer.handle, GL_BUFFER_GPU_ADDRESS_NV, &gpu_address); } } OGLStreamBuffer::~OGLStreamBuffer() { - if (persistent) { - glUnmapNamedBuffer(gl_buffer.handle); - } + glUnmapNamedBuffer(gl_buffer.handle); gl_buffer.Release(); } -GLuint OGLStreamBuffer::GetHandle() const { - return gl_buffer.handle; -} - -GLsizeiptr OGLStreamBuffer::GetSize() const { - return buffer_size; -} - std::tuple<u8*, GLintptr, bool> OGLStreamBuffer::Map(GLsizeiptr size, GLintptr alignment) { ASSERT(size <= buffer_size); ASSERT(alignment <= buffer_size); @@ -68,36 +57,21 @@ std::tuple<u8*, GLintptr, bool> OGLStreamBuffer::Map(GLsizeiptr size, GLintptr a bool invalidate = false; if (buffer_pos + size > buffer_size) { + MICROPROFILE_SCOPE(OpenGL_StreamBuffer); + glInvalidateBufferData(gl_buffer.handle); + buffer_pos = 0; invalidate = true; - - if (persistent) { - glUnmapNamedBuffer(gl_buffer.handle); - } } - if (invalidate || !persistent) { - MICROPROFILE_SCOPE(OpenGL_StreamBuffer); - GLbitfield flags = GL_MAP_WRITE_BIT | (persistent ? GL_MAP_PERSISTENT_BIT : 0) | - (coherent ? GL_MAP_COHERENT_BIT : GL_MAP_FLUSH_EXPLICIT_BIT) | - (invalidate ? GL_MAP_INVALIDATE_BUFFER_BIT : GL_MAP_UNSYNCHRONIZED_BIT); - mapped_ptr = static_cast<u8*>( - glMapNamedBufferRange(gl_buffer.handle, buffer_pos, buffer_size - buffer_pos, flags)); - mapped_offset = buffer_pos; - } - - return std::make_tuple(mapped_ptr + buffer_pos - mapped_offset, buffer_pos, invalidate); + return std::make_tuple(mapped_ptr + buffer_pos, buffer_pos, invalidate); } void OGLStreamBuffer::Unmap(GLsizeiptr size) { ASSERT(size <= mapped_size); - if (!coherent && size > 0) { - glFlushMappedNamedBufferRange(gl_buffer.handle, buffer_pos - mapped_offset, size); - } - - if (!persistent) { - glUnmapNamedBuffer(gl_buffer.handle); + if (size > 0) { + glFlushMappedNamedBufferRange(gl_buffer.handle, buffer_pos, size); } buffer_pos += size; diff --git a/src/video_core/renderer_opengl/gl_stream_buffer.h b/src/video_core/renderer_opengl/gl_stream_buffer.h index f8383cbd4..307a67113 100644 --- a/src/video_core/renderer_opengl/gl_stream_buffer.h +++ b/src/video_core/renderer_opengl/gl_stream_buffer.h @@ -11,15 +11,13 @@ namespace OpenGL { +class Device; + class OGLStreamBuffer : private NonCopyable { public: - explicit OGLStreamBuffer(GLsizeiptr size, bool vertex_data_usage, bool prefer_coherent = false, - bool use_persistent = true); + explicit OGLStreamBuffer(const Device& device, GLsizeiptr size, bool vertex_data_usage); ~OGLStreamBuffer(); - GLuint GetHandle() const; - GLsizeiptr GetSize() const; - /* * Allocates a linear chunk of memory in the GPU buffer with at least "size" bytes * and the optional alignment requirement. @@ -32,15 +30,24 @@ public: void Unmap(GLsizeiptr size); + GLuint Handle() const { + return gl_buffer.handle; + } + + u64 Address() const { + return gpu_address; + } + + GLsizeiptr Size() const noexcept { + return buffer_size; + } + private: OGLBuffer gl_buffer; - bool coherent = false; - bool persistent = false; - + GLuint64EXT gpu_address = 0; GLintptr buffer_pos = 0; GLsizeiptr buffer_size = 0; - GLintptr mapped_offset = 0; GLsizeiptr mapped_size = 0; u8* mapped_ptr = nullptr; }; diff --git a/src/video_core/renderer_opengl/gl_texture_cache.cpp b/src/video_core/renderer_opengl/gl_texture_cache.cpp index 2729d1265..61505879b 100644 --- a/src/video_core/renderer_opengl/gl_texture_cache.cpp +++ b/src/video_core/renderer_opengl/gl_texture_cache.cpp @@ -35,7 +35,7 @@ MICROPROFILE_DEFINE(OpenGL_Texture_Buffer_Copy, "OpenGL", "Texture Buffer Copy", namespace { struct FormatTuple { - GLint internal_format; + GLenum internal_format; GLenum format = GL_NONE; GLenum type = GL_NONE; }; @@ -83,6 +83,7 @@ constexpr std::array<FormatTuple, VideoCore::Surface::MaxPixelFormat> tex_format {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8_REV}, // RGBA8_SRGB {GL_RG8, GL_RG, GL_UNSIGNED_BYTE}, // RG8U {GL_RG8_SNORM, GL_RG, GL_BYTE}, // RG8S + {GL_RG8UI, GL_RG_INTEGER, GL_UNSIGNED_INT}, // RG8UI {GL_RG32UI, GL_RG_INTEGER, GL_UNSIGNED_INT}, // RG32UI {GL_RGB16F, GL_RGBA, GL_HALF_FLOAT}, // RGBX16F {GL_R32UI, GL_RED_INTEGER, GL_UNSIGNED_INT}, // R32UI @@ -237,6 +238,12 @@ OGLTexture CreateTexture(const SurfaceParams& params, GLenum target, GLenum inte return texture; } +constexpr u32 EncodeSwizzle(SwizzleSource x_source, SwizzleSource y_source, SwizzleSource z_source, + SwizzleSource w_source) { + return (static_cast<u32>(x_source) << 24) | (static_cast<u32>(y_source) << 16) | + (static_cast<u32>(z_source) << 8) | static_cast<u32>(w_source); +} + } // Anonymous namespace CachedSurface::CachedSurface(const GPUVAddr gpu_addr, const SurfaceParams& params, @@ -256,9 +263,14 @@ CachedSurface::CachedSurface(const GPUVAddr gpu_addr, const SurfaceParams& param target = GetTextureTarget(params.target); texture = CreateTexture(params, target, internal_format, texture_buffer); DecorateSurfaceName(); - main_view = CreateViewInner( - ViewParams(params.target, 0, params.is_layered ? params.depth : 1, 0, params.num_levels), - true); + + u32 num_layers = 1; + if (params.is_layered || params.target == SurfaceTarget::Texture3D) { + num_layers = params.depth; + } + + main_view = + CreateViewInner(ViewParams(params.target, 0, num_layers, 0, params.num_levels), true); } CachedSurface::~CachedSurface() = default; @@ -380,7 +392,7 @@ void CachedSurface::DecorateSurfaceName() { } void CachedSurfaceView::DecorateViewName(GPUVAddr gpu_addr, std::string prefix) { - LabelGLObject(GL_TEXTURE, texture_view.handle, gpu_addr, prefix); + LabelGLObject(GL_TEXTURE, main_view.handle, gpu_addr, prefix); } View CachedSurface::CreateView(const ViewParams& view_key) { @@ -396,32 +408,33 @@ View CachedSurface::CreateViewInner(const ViewParams& view_key, const bool is_pr } CachedSurfaceView::CachedSurfaceView(CachedSurface& surface, const ViewParams& params, - const bool is_proxy) - : VideoCommon::ViewBase(params), surface{surface}, is_proxy{is_proxy} { - target = GetTextureTarget(params.target); - format = GetFormatTuple(surface.GetSurfaceParams().pixel_format).internal_format; + bool is_proxy) + : VideoCommon::ViewBase(params), surface{surface}, format{surface.internal_format}, + target{GetTextureTarget(params.target)}, is_proxy{is_proxy} { if (!is_proxy) { - texture_view = CreateTextureView(); + main_view = CreateTextureView(); } - swizzle = EncodeSwizzle(SwizzleSource::R, SwizzleSource::G, SwizzleSource::B, SwizzleSource::A); } CachedSurfaceView::~CachedSurfaceView() = default; -void CachedSurfaceView::Attach(GLenum attachment, GLenum target) const { +void CachedSurfaceView::Attach(GLenum attachment, GLenum fb_target) const { ASSERT(params.num_levels == 1); + if (params.target == SurfaceTarget::Texture3D) { + if (params.num_layers > 1) { + ASSERT(params.base_layer == 0); + glFramebufferTexture(fb_target, attachment, surface.texture.handle, params.base_level); + } else { + glFramebufferTexture3D(fb_target, attachment, target, surface.texture.handle, + params.base_level, params.base_layer); + } + return; + } + if (params.num_layers > 1) { - // Layered framebuffer attachments UNIMPLEMENTED_IF(params.base_layer != 0); - - switch (params.target) { - case SurfaceTarget::Texture2DArray: - glFramebufferTexture(target, attachment, GetTexture(), 0); - break; - default: - UNIMPLEMENTED(); - } + glFramebufferTexture(fb_target, attachment, GetTexture(), 0); return; } @@ -429,16 +442,16 @@ void CachedSurfaceView::Attach(GLenum attachment, GLenum target) const { const GLuint texture = surface.GetTexture(); switch (surface.GetSurfaceParams().target) { case SurfaceTarget::Texture1D: - glFramebufferTexture1D(target, attachment, view_target, texture, params.base_level); + glFramebufferTexture1D(fb_target, attachment, view_target, texture, params.base_level); break; case SurfaceTarget::Texture2D: - glFramebufferTexture2D(target, attachment, view_target, texture, params.base_level); + glFramebufferTexture2D(fb_target, attachment, view_target, texture, params.base_level); break; case SurfaceTarget::Texture1DArray: case SurfaceTarget::Texture2DArray: case SurfaceTarget::TextureCubemap: case SurfaceTarget::TextureCubeArray: - glFramebufferTextureLayer(target, attachment, texture, params.base_level, + glFramebufferTextureLayer(fb_target, attachment, texture, params.base_level, params.base_layer); break; default: @@ -446,35 +459,62 @@ void CachedSurfaceView::Attach(GLenum attachment, GLenum target) const { } } -void CachedSurfaceView::ApplySwizzle(SwizzleSource x_source, SwizzleSource y_source, +GLuint CachedSurfaceView::GetTexture(SwizzleSource x_source, SwizzleSource y_source, SwizzleSource z_source, SwizzleSource w_source) { - u32 new_swizzle = EncodeSwizzle(x_source, y_source, z_source, w_source); - if (new_swizzle == swizzle) - return; - swizzle = new_swizzle; - const std::array gl_swizzle = {GetSwizzleSource(x_source), GetSwizzleSource(y_source), - GetSwizzleSource(z_source), GetSwizzleSource(w_source)}; - const GLuint handle = GetTexture(); - const PixelFormat format = surface.GetSurfaceParams().pixel_format; - switch (format) { + if (GetSurfaceParams().IsBuffer()) { + return GetTexture(); + } + const u32 new_swizzle = EncodeSwizzle(x_source, y_source, z_source, w_source); + if (current_swizzle == new_swizzle) { + return current_view; + } + current_swizzle = new_swizzle; + + const auto [entry, is_cache_miss] = view_cache.try_emplace(new_swizzle); + OGLTextureView& view = entry->second; + if (!is_cache_miss) { + current_view = view.handle; + return view.handle; + } + view = CreateTextureView(); + current_view = view.handle; + + std::array swizzle{x_source, y_source, z_source, w_source}; + + switch (const PixelFormat format = GetSurfaceParams().pixel_format) { case PixelFormat::Z24S8: case PixelFormat::Z32FS8: case PixelFormat::S8Z24: - glTextureParameteri(handle, GL_DEPTH_STENCIL_TEXTURE_MODE, + UNIMPLEMENTED_IF(x_source != SwizzleSource::R && x_source != SwizzleSource::G); + glTextureParameteri(view.handle, GL_DEPTH_STENCIL_TEXTURE_MODE, GetComponent(format, x_source == SwizzleSource::R)); - break; - default: - glTextureParameteriv(handle, GL_TEXTURE_SWIZZLE_RGBA, gl_swizzle.data()); + + // Make sure we sample the first component + std::transform(swizzle.begin(), swizzle.end(), swizzle.begin(), [](SwizzleSource value) { + return value == SwizzleSource::G ? SwizzleSource::R : value; + }); + [[fallthrough]]; + default: { + const std::array gl_swizzle = {GetSwizzleSource(swizzle[0]), GetSwizzleSource(swizzle[1]), + GetSwizzleSource(swizzle[2]), GetSwizzleSource(swizzle[3])}; + glTextureParameteriv(view.handle, GL_TEXTURE_SWIZZLE_RGBA, gl_swizzle.data()); break; } + } + return view.handle; } OGLTextureView CachedSurfaceView::CreateTextureView() const { OGLTextureView texture_view; texture_view.Create(); - glTextureView(texture_view.handle, target, surface.texture.handle, format, params.base_level, - params.num_levels, params.base_layer, params.num_layers); + if (target == GL_TEXTURE_3D) { + glTextureView(texture_view.handle, target, surface.texture.handle, format, + params.base_level, params.num_levels, 0, 1); + } else { + glTextureView(texture_view.handle, target, surface.texture.handle, format, + params.base_level, params.num_levels, params.base_layer, params.num_layers); + } ApplyTextureDefaults(surface.GetSurfaceParams(), texture_view.handle); return texture_view; @@ -517,8 +557,8 @@ void TextureCacheOpenGL::ImageBlit(View& src_view, View& dst_view, const Tegra::Engines::Fermi2D::Config& copy_config) { const auto& src_params{src_view->GetSurfaceParams()}; const auto& dst_params{dst_view->GetSurfaceParams()}; - UNIMPLEMENTED_IF(src_params.target == SurfaceTarget::Texture3D); - UNIMPLEMENTED_IF(dst_params.target == SurfaceTarget::Texture3D); + UNIMPLEMENTED_IF(src_params.depth != 1); + UNIMPLEMENTED_IF(dst_params.depth != 1); state_tracker.NotifyScissor0(); state_tracker.NotifyFramebuffer(); diff --git a/src/video_core/renderer_opengl/gl_texture_cache.h b/src/video_core/renderer_opengl/gl_texture_cache.h index 02d9981a1..bfc4ddf5d 100644 --- a/src/video_core/renderer_opengl/gl_texture_cache.h +++ b/src/video_core/renderer_opengl/gl_texture_cache.h @@ -80,10 +80,12 @@ public: explicit CachedSurfaceView(CachedSurface& surface, const ViewParams& params, bool is_proxy); ~CachedSurfaceView(); - /// Attaches this texture view to the current bound GL_DRAW_FRAMEBUFFER - void Attach(GLenum attachment, GLenum target) const; + /// @brief Attaches this texture view to the currently bound fb_target framebuffer + /// @param attachment Attachment to bind textures to + /// @param fb_target Framebuffer target to attach to (e.g. DRAW_FRAMEBUFFER) + void Attach(GLenum attachment, GLenum fb_target) const; - void ApplySwizzle(Tegra::Texture::SwizzleSource x_source, + GLuint GetTexture(Tegra::Texture::SwizzleSource x_source, Tegra::Texture::SwizzleSource y_source, Tegra::Texture::SwizzleSource z_source, Tegra::Texture::SwizzleSource w_source); @@ -98,7 +100,7 @@ public: if (is_proxy) { return surface.GetTexture(); } - return texture_view.handle; + return main_view.handle; } GLenum GetFormat() const { @@ -110,23 +112,19 @@ public: } private: - u32 EncodeSwizzle(Tegra::Texture::SwizzleSource x_source, - Tegra::Texture::SwizzleSource y_source, - Tegra::Texture::SwizzleSource z_source, - Tegra::Texture::SwizzleSource w_source) const { - return (static_cast<u32>(x_source) << 24) | (static_cast<u32>(y_source) << 16) | - (static_cast<u32>(z_source) << 8) | static_cast<u32>(w_source); - } - OGLTextureView CreateTextureView() const; CachedSurface& surface; - GLenum target{}; - GLenum format{}; + const GLenum format; + const GLenum target; + const bool is_proxy; + + std::unordered_map<u32, OGLTextureView> view_cache; + OGLTextureView main_view; - OGLTextureView texture_view; - u32 swizzle{}; - bool is_proxy{}; + // Use an invalid default so it always fails the comparison test + u32 current_swizzle = 0xffffffff; + GLuint current_view = 0; }; class TextureCacheOpenGL final : public TextureCacheBase { diff --git a/src/video_core/renderer_opengl/maxwell_to_gl.h b/src/video_core/renderer_opengl/maxwell_to_gl.h index 2c0c77c28..774e70a5b 100644 --- a/src/video_core/renderer_opengl/maxwell_to_gl.h +++ b/src/video_core/renderer_opengl/maxwell_to_gl.h @@ -24,10 +24,11 @@ namespace MaxwellToGL { using Maxwell = Tegra::Engines::Maxwell3D::Regs; -inline GLenum VertexType(Maxwell::VertexAttribute attrib) { +inline GLenum VertexFormat(Maxwell::VertexAttribute attrib) { switch (attrib.type) { - case Maxwell::VertexAttribute::Type::UnsignedInt: case Maxwell::VertexAttribute::Type::UnsignedNorm: + case Maxwell::VertexAttribute::Type::UnsignedScaled: + case Maxwell::VertexAttribute::Type::UnsignedInt: switch (attrib.size) { case Maxwell::VertexAttribute::Size::Size_8: case Maxwell::VertexAttribute::Size::Size_8_8: @@ -46,12 +47,11 @@ inline GLenum VertexType(Maxwell::VertexAttribute attrib) { return GL_UNSIGNED_INT; case Maxwell::VertexAttribute::Size::Size_10_10_10_2: return GL_UNSIGNED_INT_2_10_10_10_REV; - default: - LOG_ERROR(Render_OpenGL, "Unimplemented vertex size={}", attrib.SizeString()); - return {}; } - case Maxwell::VertexAttribute::Type::SignedInt: + break; case Maxwell::VertexAttribute::Type::SignedNorm: + case Maxwell::VertexAttribute::Type::SignedScaled: + case Maxwell::VertexAttribute::Type::SignedInt: switch (attrib.size) { case Maxwell::VertexAttribute::Size::Size_8: case Maxwell::VertexAttribute::Size::Size_8_8: @@ -70,10 +70,8 @@ inline GLenum VertexType(Maxwell::VertexAttribute attrib) { return GL_INT; case Maxwell::VertexAttribute::Size::Size_10_10_10_2: return GL_INT_2_10_10_10_REV; - default: - LOG_ERROR(Render_OpenGL, "Unimplemented vertex size={}", attrib.SizeString()); - return {}; } + break; case Maxwell::VertexAttribute::Type::Float: switch (attrib.size) { case Maxwell::VertexAttribute::Size::Size_16: @@ -86,46 +84,12 @@ inline GLenum VertexType(Maxwell::VertexAttribute attrib) { case Maxwell::VertexAttribute::Size::Size_32_32_32: case Maxwell::VertexAttribute::Size::Size_32_32_32_32: return GL_FLOAT; - default: - LOG_ERROR(Render_OpenGL, "Unimplemented vertex size={}", attrib.SizeString()); - return {}; } - case Maxwell::VertexAttribute::Type::UnsignedScaled: - switch (attrib.size) { - case Maxwell::VertexAttribute::Size::Size_8: - case Maxwell::VertexAttribute::Size::Size_8_8: - case Maxwell::VertexAttribute::Size::Size_8_8_8: - case Maxwell::VertexAttribute::Size::Size_8_8_8_8: - return GL_UNSIGNED_BYTE; - case Maxwell::VertexAttribute::Size::Size_16: - case Maxwell::VertexAttribute::Size::Size_16_16: - case Maxwell::VertexAttribute::Size::Size_16_16_16: - case Maxwell::VertexAttribute::Size::Size_16_16_16_16: - return GL_UNSIGNED_SHORT; - default: - LOG_ERROR(Render_OpenGL, "Unimplemented vertex size={}", attrib.SizeString()); - return {}; - } - case Maxwell::VertexAttribute::Type::SignedScaled: - switch (attrib.size) { - case Maxwell::VertexAttribute::Size::Size_8: - case Maxwell::VertexAttribute::Size::Size_8_8: - case Maxwell::VertexAttribute::Size::Size_8_8_8: - case Maxwell::VertexAttribute::Size::Size_8_8_8_8: - return GL_BYTE; - case Maxwell::VertexAttribute::Size::Size_16: - case Maxwell::VertexAttribute::Size::Size_16_16: - case Maxwell::VertexAttribute::Size::Size_16_16_16: - case Maxwell::VertexAttribute::Size::Size_16_16_16_16: - return GL_SHORT; - default: - LOG_ERROR(Render_OpenGL, "Unimplemented vertex size={}", attrib.SizeString()); - return {}; - } - default: - LOG_ERROR(Render_OpenGL, "Unimplemented vertex type={}", attrib.TypeString()); - return {}; + break; } + UNIMPLEMENTED_MSG("Unimplemented vertex format of type={} and size={}", attrib.TypeString(), + attrib.SizeString()); + return {}; } inline GLenum IndexFormat(Maxwell::IndexFormat index_format) { @@ -137,8 +101,7 @@ inline GLenum IndexFormat(Maxwell::IndexFormat index_format) { case Maxwell::IndexFormat::UnsignedInt: return GL_UNSIGNED_INT; } - LOG_CRITICAL(Render_OpenGL, "Unimplemented index_format={}", static_cast<u32>(index_format)); - UNREACHABLE(); + UNREACHABLE_MSG("Invalid index_format={}", static_cast<u32>(index_format)); return {}; } @@ -180,33 +143,32 @@ inline GLenum PrimitiveTopology(Maxwell::PrimitiveTopology topology) { } inline GLenum TextureFilterMode(Tegra::Texture::TextureFilter filter_mode, - Tegra::Texture::TextureMipmapFilter mip_filter_mode) { + Tegra::Texture::TextureMipmapFilter mipmap_filter_mode) { switch (filter_mode) { - case Tegra::Texture::TextureFilter::Linear: { - switch (mip_filter_mode) { + case Tegra::Texture::TextureFilter::Nearest: + switch (mipmap_filter_mode) { case Tegra::Texture::TextureMipmapFilter::None: - return GL_LINEAR; + return GL_NEAREST; case Tegra::Texture::TextureMipmapFilter::Nearest: - return GL_LINEAR_MIPMAP_NEAREST; + return GL_NEAREST_MIPMAP_NEAREST; case Tegra::Texture::TextureMipmapFilter::Linear: - return GL_LINEAR_MIPMAP_LINEAR; + return GL_NEAREST_MIPMAP_LINEAR; } break; - } - case Tegra::Texture::TextureFilter::Nearest: { - switch (mip_filter_mode) { + case Tegra::Texture::TextureFilter::Linear: + switch (mipmap_filter_mode) { case Tegra::Texture::TextureMipmapFilter::None: - return GL_NEAREST; + return GL_LINEAR; case Tegra::Texture::TextureMipmapFilter::Nearest: - return GL_NEAREST_MIPMAP_NEAREST; + return GL_LINEAR_MIPMAP_NEAREST; case Tegra::Texture::TextureMipmapFilter::Linear: - return GL_NEAREST_MIPMAP_LINEAR; + return GL_LINEAR_MIPMAP_LINEAR; } break; } - } - LOG_ERROR(Render_OpenGL, "Unimplemented texture filter mode={}", static_cast<u32>(filter_mode)); - return GL_LINEAR; + UNREACHABLE_MSG("Invalid texture filter mode={} and mipmap filter mode={}", + static_cast<u32>(filter_mode), static_cast<u32>(mipmap_filter_mode)); + return GL_NEAREST; } inline GLenum WrapMode(Tegra::Texture::WrapMode wrap_mode) { @@ -229,10 +191,9 @@ inline GLenum WrapMode(Tegra::Texture::WrapMode wrap_mode) { } else { return GL_MIRROR_CLAMP_TO_EDGE; } - default: - LOG_ERROR(Render_OpenGL, "Unimplemented texture wrap mode={}", static_cast<u32>(wrap_mode)); - return GL_REPEAT; } + UNIMPLEMENTED_MSG("Unimplemented texture wrap mode={}", static_cast<u32>(wrap_mode)); + return GL_REPEAT; } inline GLenum DepthCompareFunc(Tegra::Texture::DepthCompareFunc func) { @@ -254,8 +215,7 @@ inline GLenum DepthCompareFunc(Tegra::Texture::DepthCompareFunc func) { case Tegra::Texture::DepthCompareFunc::Always: return GL_ALWAYS; } - LOG_ERROR(Render_OpenGL, "Unimplemented texture depth compare function ={}", - static_cast<u32>(func)); + UNIMPLEMENTED_MSG("Unimplemented texture depth compare function={}", static_cast<u32>(func)); return GL_GREATER; } @@ -277,7 +237,7 @@ inline GLenum BlendEquation(Maxwell::Blend::Equation equation) { case Maxwell::Blend::Equation::MaxGL: return GL_MAX; } - LOG_ERROR(Render_OpenGL, "Unimplemented blend equation={}", static_cast<u32>(equation)); + UNIMPLEMENTED_MSG("Unimplemented blend equation={}", static_cast<u32>(equation)); return GL_FUNC_ADD; } @@ -341,7 +301,7 @@ inline GLenum BlendFunc(Maxwell::Blend::Factor factor) { case Maxwell::Blend::Factor::OneMinusConstantAlphaGL: return GL_ONE_MINUS_CONSTANT_ALPHA; } - LOG_ERROR(Render_OpenGL, "Unimplemented blend factor={}", static_cast<u32>(factor)); + UNIMPLEMENTED_MSG("Unimplemented blend factor={}", static_cast<u32>(factor)); return GL_ZERO; } @@ -361,7 +321,7 @@ inline GLenum SwizzleSource(Tegra::Texture::SwizzleSource source) { case Tegra::Texture::SwizzleSource::OneFloat: return GL_ONE; } - LOG_ERROR(Render_OpenGL, "Unimplemented swizzle source={}", static_cast<u32>(source)); + UNIMPLEMENTED_MSG("Unimplemented swizzle source={}", static_cast<u32>(source)); return GL_ZERO; } @@ -392,7 +352,7 @@ inline GLenum ComparisonOp(Maxwell::ComparisonOp comparison) { case Maxwell::ComparisonOp::AlwaysOld: return GL_ALWAYS; } - LOG_ERROR(Render_OpenGL, "Unimplemented comparison op={}", static_cast<u32>(comparison)); + UNIMPLEMENTED_MSG("Unimplemented comparison op={}", static_cast<u32>(comparison)); return GL_ALWAYS; } @@ -423,7 +383,7 @@ inline GLenum StencilOp(Maxwell::StencilOp stencil) { case Maxwell::StencilOp::DecrWrapOGL: return GL_DECR_WRAP; } - LOG_ERROR(Render_OpenGL, "Unimplemented stencil op={}", static_cast<u32>(stencil)); + UNIMPLEMENTED_MSG("Unimplemented stencil op={}", static_cast<u32>(stencil)); return GL_KEEP; } @@ -434,7 +394,7 @@ inline GLenum FrontFace(Maxwell::FrontFace front_face) { case Maxwell::FrontFace::CounterClockWise: return GL_CCW; } - LOG_ERROR(Render_OpenGL, "Unimplemented front face cull={}", static_cast<u32>(front_face)); + UNIMPLEMENTED_MSG("Unimplemented front face cull={}", static_cast<u32>(front_face)); return GL_CCW; } @@ -447,7 +407,7 @@ inline GLenum CullFace(Maxwell::CullFace cull_face) { case Maxwell::CullFace::FrontAndBack: return GL_FRONT_AND_BACK; } - LOG_ERROR(Render_OpenGL, "Unimplemented cull face={}", static_cast<u32>(cull_face)); + UNIMPLEMENTED_MSG("Unimplemented cull face={}", static_cast<u32>(cull_face)); return GL_BACK; } @@ -486,7 +446,7 @@ inline GLenum LogicOp(Maxwell::LogicOperation operation) { case Maxwell::LogicOperation::Set: return GL_SET; } - LOG_ERROR(Render_OpenGL, "Unimplemented logic operation={}", static_cast<u32>(operation)); + UNIMPLEMENTED_MSG("Unimplemented logic operation={}", static_cast<u32>(operation)); return GL_COPY; } @@ -503,5 +463,10 @@ inline GLenum PolygonMode(Maxwell::PolygonMode polygon_mode) { return GL_FILL; } +inline GLenum ViewportSwizzle(Maxwell::ViewportSwizzle swizzle) { + // Enumeration order matches register order. We can convert it arithmetically. + return GL_VIEWPORT_SWIZZLE_POSITIVE_X_NV + static_cast<GLenum>(swizzle); +} + } // namespace MaxwellToGL } // namespace OpenGL diff --git a/src/video_core/renderer_opengl/renderer_opengl.cpp b/src/video_core/renderer_opengl/renderer_opengl.cpp index b2a179746..c40adb6e7 100644 --- a/src/video_core/renderer_opengl/renderer_opengl.cpp +++ b/src/video_core/renderer_opengl/renderer_opengl.cpp @@ -316,7 +316,7 @@ public: RendererOpenGL::RendererOpenGL(Core::Frontend::EmuWindow& emu_window, Core::System& system, Core::Frontend::GraphicsContext& context) : RendererBase{emu_window}, emu_window{emu_window}, system{system}, context{context}, - has_debug_tool{HasDebugTool()} {} + program_manager{device}, has_debug_tool{HasDebugTool()} {} RendererOpenGL::~RendererOpenGL() = default; @@ -468,8 +468,9 @@ void RendererOpenGL::InitOpenGLObjects() { vertex_program.Create(true, false, vertex_shader.handle); fragment_program.Create(true, false, fragment_shader.handle); - // Create program pipeline - program_manager.Create(); + pipeline.Create(); + glUseProgramStages(pipeline.handle, GL_VERTEX_SHADER_BIT, vertex_program.handle); + glUseProgramStages(pipeline.handle, GL_FRAGMENT_SHADER_BIT, fragment_program.handle); // Generate VBO handle for drawing vertex_buffer.Create(); @@ -487,6 +488,15 @@ void RendererOpenGL::InitOpenGLObjects() { // Clear screen to black LoadColorToActiveGLTexture(0, 0, 0, 0, screen_info.texture); + + // Enable unified vertex attributes and query vertex buffer address when the driver supports it + if (device.HasVertexBufferUnifiedMemory()) { + glEnableClientState(GL_VERTEX_ATTRIB_ARRAY_UNIFIED_NV); + + glMakeNamedBufferResidentNV(vertex_buffer.handle, GL_READ_ONLY); + glGetNamedBufferParameterui64vNV(vertex_buffer.handle, GL_BUFFER_GPU_ADDRESS_NV, + &vertex_buffer_address); + } } void RendererOpenGL::AddTelemetryFields() { @@ -508,7 +518,7 @@ void RendererOpenGL::CreateRasterizer() { if (rasterizer) { return; } - rasterizer = std::make_unique<RasterizerOpenGL>(system, emu_window, screen_info, + rasterizer = std::make_unique<RasterizerOpenGL>(system, emu_window, device, screen_info, program_manager, state_tracker); } @@ -620,10 +630,7 @@ void RendererOpenGL::DrawScreen(const Layout::FramebufferLayout& layout) { state_tracker.NotifyClipControl(); state_tracker.NotifyAlphaTest(); - program_manager.UseVertexShader(vertex_program.handle); - program_manager.UseGeometryShader(0); - program_manager.UseFragmentShader(fragment_program.handle); - program_manager.BindGraphicsPipeline(); + program_manager.BindHostPipeline(pipeline.handle); glEnable(GL_CULL_FACE); if (screen_info.display_srgb) { @@ -658,13 +665,21 @@ void RendererOpenGL::DrawScreen(const Layout::FramebufferLayout& layout) { offsetof(ScreenRectVertex, tex_coord)); glVertexAttribBinding(PositionLocation, 0); glVertexAttribBinding(TexCoordLocation, 0); - glBindVertexBuffer(0, vertex_buffer.handle, 0, sizeof(ScreenRectVertex)); + if (device.HasVertexBufferUnifiedMemory()) { + glBindVertexBuffer(0, 0, 0, sizeof(ScreenRectVertex)); + glBufferAddressRangeNV(GL_VERTEX_ATTRIB_ARRAY_ADDRESS_NV, 0, vertex_buffer_address, + sizeof(vertices)); + } else { + glBindVertexBuffer(0, vertex_buffer.handle, 0, sizeof(ScreenRectVertex)); + } glBindTextureUnit(0, screen_info.display_texture); glBindSampler(0, 0); glClear(GL_COLOR_BUFFER_BIT); glDrawArrays(GL_TRIANGLE_STRIP, 0, 4); + + program_manager.RestoreGuestPipeline(); } bool RendererOpenGL::TryPresent(int timeout_ms) { @@ -751,8 +766,9 @@ void RendererOpenGL::RenderScreenshot() { } bool RendererOpenGL::Init() { - if (GLAD_GL_KHR_debug) { + if (Settings::values.renderer_debug && GLAD_GL_KHR_debug) { glEnable(GL_DEBUG_OUTPUT); + glEnable(GL_DEBUG_OUTPUT_SYNCHRONOUS); glDebugMessageCallback(DebugHandler, nullptr); } diff --git a/src/video_core/renderer_opengl/renderer_opengl.h b/src/video_core/renderer_opengl/renderer_opengl.h index 50b647661..8b18d32e6 100644 --- a/src/video_core/renderer_opengl/renderer_opengl.h +++ b/src/video_core/renderer_opengl/renderer_opengl.h @@ -9,6 +9,7 @@ #include "common/common_types.h" #include "common/math_util.h" #include "video_core/renderer_base.h" +#include "video_core/renderer_opengl/gl_device.h" #include "video_core/renderer_opengl/gl_resource_manager.h" #include "video_core/renderer_opengl/gl_shader_manager.h" #include "video_core/renderer_opengl/gl_state_tracker.h" @@ -95,6 +96,7 @@ private: Core::Frontend::EmuWindow& emu_window; Core::System& system; Core::Frontend::GraphicsContext& context; + const Device device; StateTracker state_tracker{system}; @@ -102,13 +104,17 @@ private: OGLBuffer vertex_buffer; OGLProgram vertex_program; OGLProgram fragment_program; + OGLPipeline pipeline; OGLFramebuffer screenshot_framebuffer; + // GPU address of the vertex buffer + GLuint64EXT vertex_buffer_address = 0; + /// Display information for Switch screen ScreenInfo screen_info; /// Global dummy shader pipeline - GLShader::ProgramManager program_manager; + ProgramManager program_manager; /// OpenGL framebuffer data std::vector<u8> gl_framebuffer_data; diff --git a/src/video_core/renderer_vulkan/fixed_pipeline_state.cpp b/src/video_core/renderer_vulkan/fixed_pipeline_state.cpp index 648b1e71b..424278816 100644 --- a/src/video_core/renderer_vulkan/fixed_pipeline_state.cpp +++ b/src/video_core/renderer_vulkan/fixed_pipeline_state.cpp @@ -2,6 +2,7 @@ // Licensed under GPLv2 or any later version // Refer to the license.txt file included. +#include <algorithm> #include <cstring> #include <tuple> @@ -70,8 +71,7 @@ void FixedPipelineState::Rasterizer::Fill(const Maxwell& regs) noexcept { const u32 topology_index = static_cast<u32>(regs.draw.topology.Value()); u32 packed_front_face = PackFrontFace(regs.front_face); - if (regs.screen_y_control.triangle_rast_flip != 0 && - regs.viewport_transform[0].scale_y > 0.0f) { + if (regs.screen_y_control.triangle_rast_flip != 0) { // Flip front face packed_front_face = 1 - packed_front_face; } @@ -92,6 +92,7 @@ void FixedPipelineState::Rasterizer::Fill(const Maxwell& regs) noexcept { tessellation_clockwise.Assign(regs.tess_mode.cw.Value()); logic_op_enable.Assign(regs.logic_op.enable != 0 ? 1 : 0); logic_op.Assign(PackLogicOp(regs.logic_op.operation)); + rasterize_enable.Assign(regs.rasterize_enable != 0 ? 1 : 0); std::memcpy(&point_size, ®s.point_size, sizeof(point_size)); // TODO: C++20 std::bit_cast } @@ -101,6 +102,12 @@ void FixedPipelineState::ColorBlending::Fill(const Maxwell& regs) noexcept { } } +void FixedPipelineState::ViewportSwizzles::Fill(const Maxwell& regs) noexcept { + const auto& transform = regs.viewport_transform; + std::transform(transform.begin(), transform.end(), swizzles.begin(), + [](const auto& viewport) { return static_cast<u16>(viewport.swizzle.raw); }); +} + void FixedPipelineState::BlendingAttachment::Fill(const Maxwell& regs, std::size_t index) { const auto& mask = regs.color_mask[regs.color_mask_common ? 0 : index]; @@ -144,6 +151,7 @@ void FixedPipelineState::Fill(const Maxwell& regs) { rasterizer.Fill(regs); depth_stencil.Fill(regs); color_blending.Fill(regs); + viewport_swizzles.Fill(regs); } std::size_t FixedPipelineState::Hash() const noexcept { diff --git a/src/video_core/renderer_vulkan/fixed_pipeline_state.h b/src/video_core/renderer_vulkan/fixed_pipeline_state.h index 8652067a7..31a6398f2 100644 --- a/src/video_core/renderer_vulkan/fixed_pipeline_state.h +++ b/src/video_core/renderer_vulkan/fixed_pipeline_state.h @@ -164,6 +164,7 @@ struct FixedPipelineState { BitField<23, 1, u32> tessellation_clockwise; BitField<24, 1, u32> logic_op_enable; BitField<25, 4, u32> logic_op; + BitField<29, 1, u32> rasterize_enable; }; // TODO(Rodrigo): Move this to push constants @@ -233,10 +234,17 @@ struct FixedPipelineState { void Fill(const Maxwell& regs) noexcept; }; + struct ViewportSwizzles { + std::array<u16, Maxwell::NumViewports> swizzles; + + void Fill(const Maxwell& regs) noexcept; + }; + VertexInput vertex_input; Rasterizer rasterizer; DepthStencil depth_stencil; ColorBlending color_blending; + ViewportSwizzles viewport_swizzles; void Fill(const Maxwell& regs); diff --git a/src/video_core/renderer_vulkan/maxwell_to_vk.cpp b/src/video_core/renderer_vulkan/maxwell_to_vk.cpp index 8681b821f..d7f1ae89f 100644 --- a/src/video_core/renderer_vulkan/maxwell_to_vk.cpp +++ b/src/video_core/renderer_vulkan/maxwell_to_vk.cpp @@ -21,29 +21,29 @@ namespace Sampler { VkFilter Filter(Tegra::Texture::TextureFilter filter) { switch (filter) { - case Tegra::Texture::TextureFilter::Linear: - return VK_FILTER_LINEAR; case Tegra::Texture::TextureFilter::Nearest: return VK_FILTER_NEAREST; + case Tegra::Texture::TextureFilter::Linear: + return VK_FILTER_LINEAR; } - UNIMPLEMENTED_MSG("Unimplemented sampler filter={}", static_cast<u32>(filter)); + UNREACHABLE_MSG("Invalid sampler filter={}", static_cast<u32>(filter)); return {}; } VkSamplerMipmapMode MipmapMode(Tegra::Texture::TextureMipmapFilter mipmap_filter) { switch (mipmap_filter) { case Tegra::Texture::TextureMipmapFilter::None: - // TODO(Rodrigo): None seems to be mapped to OpenGL's mag and min filters without mipmapping - // (e.g. GL_NEAREST and GL_LINEAR). Vulkan doesn't have such a thing, find out if we have to - // use an image view with a single mipmap level to emulate this. - return VK_SAMPLER_MIPMAP_MODE_LINEAR; - ; - case Tegra::Texture::TextureMipmapFilter::Linear: - return VK_SAMPLER_MIPMAP_MODE_LINEAR; + // There are no Vulkan filter modes that directly correspond to OpenGL minification filters + // of GL_LINEAR or GL_NEAREST, but they can be emulated using + // VK_SAMPLER_MIPMAP_MODE_NEAREST, minLod = 0, and maxLod = 0.25, and using minFilter = + // VK_FILTER_LINEAR or minFilter = VK_FILTER_NEAREST, respectively. + return VK_SAMPLER_MIPMAP_MODE_NEAREST; case Tegra::Texture::TextureMipmapFilter::Nearest: return VK_SAMPLER_MIPMAP_MODE_NEAREST; + case Tegra::Texture::TextureMipmapFilter::Linear: + return VK_SAMPLER_MIPMAP_MODE_LINEAR; } - UNIMPLEMENTED_MSG("Unimplemented sampler mipmap mode={}", static_cast<u32>(mipmap_filter)); + UNREACHABLE_MSG("Invalid sampler mipmap mode={}", static_cast<u32>(mipmap_filter)); return {}; } @@ -78,10 +78,9 @@ VkSamplerAddressMode WrapMode(const VKDevice& device, Tegra::Texture::WrapMode w case Tegra::Texture::WrapMode::MirrorOnceBorder: UNIMPLEMENTED(); return VK_SAMPLER_ADDRESS_MODE_MIRROR_CLAMP_TO_EDGE; - default: - UNIMPLEMENTED_MSG("Unimplemented wrap mode={}", static_cast<u32>(wrap_mode)); - return {}; } + UNIMPLEMENTED_MSG("Unimplemented wrap mode={}", static_cast<u32>(wrap_mode)); + return {}; } VkCompareOp DepthCompareFunction(Tegra::Texture::DepthCompareFunc depth_compare_func) { @@ -142,14 +141,14 @@ struct FormatTuple { {VK_FORMAT_BC6H_UFLOAT_BLOCK}, // BC6H_UF16 {VK_FORMAT_BC6H_SFLOAT_BLOCK}, // BC6H_SF16 {VK_FORMAT_ASTC_4x4_UNORM_BLOCK}, // ASTC_2D_4X4 - {VK_FORMAT_B8G8R8A8_UNORM}, // BGRA8 + {VK_FORMAT_B8G8R8A8_UNORM, Attachable}, // BGRA8 {VK_FORMAT_R32G32B32A32_SFLOAT, Attachable | Storage}, // RGBA32F {VK_FORMAT_R32G32_SFLOAT, Attachable | Storage}, // RG32F {VK_FORMAT_R32_SFLOAT, Attachable | Storage}, // R32F {VK_FORMAT_R16_SFLOAT, Attachable | Storage}, // R16F {VK_FORMAT_R16_UNORM, Attachable | Storage}, // R16U {VK_FORMAT_UNDEFINED}, // R16S - {VK_FORMAT_UNDEFINED}, // R16UI + {VK_FORMAT_R16_UINT, Attachable | Storage}, // R16UI {VK_FORMAT_UNDEFINED}, // R16I {VK_FORMAT_R16G16_UNORM, Attachable | Storage}, // RG16 {VK_FORMAT_R16G16_SFLOAT, Attachable | Storage}, // RG16F @@ -160,6 +159,7 @@ struct FormatTuple { {VK_FORMAT_R8G8B8A8_SRGB, Attachable}, // RGBA8_SRGB {VK_FORMAT_R8G8_UNORM, Attachable | Storage}, // RG8U {VK_FORMAT_R8G8_SNORM, Attachable | Storage}, // RG8S + {VK_FORMAT_R8G8_UINT, Attachable | Storage}, // RG8UI {VK_FORMAT_R32G32_UINT, Attachable | Storage}, // RG32UI {VK_FORMAT_UNDEFINED}, // RGBX16F {VK_FORMAT_R32_UINT, Attachable | Storage}, // R32UI @@ -167,7 +167,7 @@ struct FormatTuple { {VK_FORMAT_ASTC_8x8_UNORM_BLOCK}, // ASTC_2D_8X8 {VK_FORMAT_UNDEFINED}, // ASTC_2D_8X5 {VK_FORMAT_UNDEFINED}, // ASTC_2D_5X4 - {VK_FORMAT_UNDEFINED}, // BGRA8_SRGB + {VK_FORMAT_B8G8R8A8_SRGB, Attachable}, // BGRA8_SRGB {VK_FORMAT_BC1_RGBA_SRGB_BLOCK}, // DXT1_SRGB {VK_FORMAT_BC2_SRGB_BLOCK}, // DXT23_SRGB {VK_FORMAT_BC3_SRGB_BLOCK}, // DXT45_SRGB @@ -287,14 +287,35 @@ VkPrimitiveTopology PrimitiveTopology([[maybe_unused]] const VKDevice& device, return VK_PRIMITIVE_TOPOLOGY_TRIANGLE_LIST; case Maxwell::PrimitiveTopology::Patches: return VK_PRIMITIVE_TOPOLOGY_PATCH_LIST; - default: - UNIMPLEMENTED_MSG("Unimplemented topology={}", static_cast<u32>(topology)); - return {}; } + UNIMPLEMENTED_MSG("Unimplemented topology={}", static_cast<u32>(topology)); + return {}; } VkFormat VertexFormat(Maxwell::VertexAttribute::Type type, Maxwell::VertexAttribute::Size size) { switch (type) { + case Maxwell::VertexAttribute::Type::UnsignedNorm: + switch (size) { + case Maxwell::VertexAttribute::Size::Size_8: + return VK_FORMAT_R8_UNORM; + case Maxwell::VertexAttribute::Size::Size_8_8: + return VK_FORMAT_R8G8_UNORM; + case Maxwell::VertexAttribute::Size::Size_8_8_8: + return VK_FORMAT_R8G8B8_UNORM; + case Maxwell::VertexAttribute::Size::Size_8_8_8_8: + return VK_FORMAT_R8G8B8A8_UNORM; + case Maxwell::VertexAttribute::Size::Size_16: + return VK_FORMAT_R16_UNORM; + case Maxwell::VertexAttribute::Size::Size_16_16: + return VK_FORMAT_R16G16_UNORM; + case Maxwell::VertexAttribute::Size::Size_16_16_16: + return VK_FORMAT_R16G16B16_UNORM; + case Maxwell::VertexAttribute::Size::Size_16_16_16_16: + return VK_FORMAT_R16G16B16A16_UNORM; + case Maxwell::VertexAttribute::Size::Size_10_10_10_2: + return VK_FORMAT_A2B10G10R10_UNORM_PACK32; + } + break; case Maxwell::VertexAttribute::Type::SignedNorm: switch (size) { case Maxwell::VertexAttribute::Size::Size_8: @@ -315,50 +336,50 @@ VkFormat VertexFormat(Maxwell::VertexAttribute::Type type, Maxwell::VertexAttrib return VK_FORMAT_R16G16B16A16_SNORM; case Maxwell::VertexAttribute::Size::Size_10_10_10_2: return VK_FORMAT_A2B10G10R10_SNORM_PACK32; - default: - break; } break; - case Maxwell::VertexAttribute::Type::UnsignedNorm: + case Maxwell::VertexAttribute::Type::UnsignedScaled: switch (size) { case Maxwell::VertexAttribute::Size::Size_8: - return VK_FORMAT_R8_UNORM; + return VK_FORMAT_R8_USCALED; case Maxwell::VertexAttribute::Size::Size_8_8: - return VK_FORMAT_R8G8_UNORM; + return VK_FORMAT_R8G8_USCALED; case Maxwell::VertexAttribute::Size::Size_8_8_8: - return VK_FORMAT_R8G8B8_UNORM; + return VK_FORMAT_R8G8B8_USCALED; case Maxwell::VertexAttribute::Size::Size_8_8_8_8: - return VK_FORMAT_R8G8B8A8_UNORM; + return VK_FORMAT_R8G8B8A8_USCALED; case Maxwell::VertexAttribute::Size::Size_16: - return VK_FORMAT_R16_UNORM; + return VK_FORMAT_R16_USCALED; case Maxwell::VertexAttribute::Size::Size_16_16: - return VK_FORMAT_R16G16_UNORM; + return VK_FORMAT_R16G16_USCALED; case Maxwell::VertexAttribute::Size::Size_16_16_16: - return VK_FORMAT_R16G16B16_UNORM; + return VK_FORMAT_R16G16B16_USCALED; case Maxwell::VertexAttribute::Size::Size_16_16_16_16: - return VK_FORMAT_R16G16B16A16_UNORM; + return VK_FORMAT_R16G16B16A16_USCALED; case Maxwell::VertexAttribute::Size::Size_10_10_10_2: - return VK_FORMAT_A2B10G10R10_UNORM_PACK32; - default: - break; + return VK_FORMAT_A2B10G10R10_USCALED_PACK32; } break; - case Maxwell::VertexAttribute::Type::SignedInt: + case Maxwell::VertexAttribute::Type::SignedScaled: switch (size) { - case Maxwell::VertexAttribute::Size::Size_16_16_16_16: - return VK_FORMAT_R16G16B16A16_SINT; case Maxwell::VertexAttribute::Size::Size_8: - return VK_FORMAT_R8_SINT; + return VK_FORMAT_R8_SSCALED; case Maxwell::VertexAttribute::Size::Size_8_8: - return VK_FORMAT_R8G8_SINT; + return VK_FORMAT_R8G8_SSCALED; case Maxwell::VertexAttribute::Size::Size_8_8_8: - return VK_FORMAT_R8G8B8_SINT; + return VK_FORMAT_R8G8B8_SSCALED; case Maxwell::VertexAttribute::Size::Size_8_8_8_8: - return VK_FORMAT_R8G8B8A8_SINT; - case Maxwell::VertexAttribute::Size::Size_32: - return VK_FORMAT_R32_SINT; - default: - break; + return VK_FORMAT_R8G8B8A8_SSCALED; + case Maxwell::VertexAttribute::Size::Size_16: + return VK_FORMAT_R16_SSCALED; + case Maxwell::VertexAttribute::Size::Size_16_16: + return VK_FORMAT_R16G16_SSCALED; + case Maxwell::VertexAttribute::Size::Size_16_16_16: + return VK_FORMAT_R16G16B16_SSCALED; + case Maxwell::VertexAttribute::Size::Size_16_16_16_16: + return VK_FORMAT_R16G16B16A16_SSCALED; + case Maxwell::VertexAttribute::Size::Size_10_10_10_2: + return VK_FORMAT_A2B10G10R10_SSCALED_PACK32; } break; case Maxwell::VertexAttribute::Type::UnsignedInt: @@ -387,56 +408,50 @@ VkFormat VertexFormat(Maxwell::VertexAttribute::Type type, Maxwell::VertexAttrib return VK_FORMAT_R32G32B32_UINT; case Maxwell::VertexAttribute::Size::Size_32_32_32_32: return VK_FORMAT_R32G32B32A32_UINT; - default: - break; + case Maxwell::VertexAttribute::Size::Size_10_10_10_2: + return VK_FORMAT_A2B10G10R10_UINT_PACK32; } break; - case Maxwell::VertexAttribute::Type::UnsignedScaled: + case Maxwell::VertexAttribute::Type::SignedInt: switch (size) { case Maxwell::VertexAttribute::Size::Size_8: - return VK_FORMAT_R8_USCALED; + return VK_FORMAT_R8_SINT; case Maxwell::VertexAttribute::Size::Size_8_8: - return VK_FORMAT_R8G8_USCALED; + return VK_FORMAT_R8G8_SINT; case Maxwell::VertexAttribute::Size::Size_8_8_8: - return VK_FORMAT_R8G8B8_USCALED; + return VK_FORMAT_R8G8B8_SINT; case Maxwell::VertexAttribute::Size::Size_8_8_8_8: - return VK_FORMAT_R8G8B8A8_USCALED; + return VK_FORMAT_R8G8B8A8_SINT; case Maxwell::VertexAttribute::Size::Size_16: - return VK_FORMAT_R16_USCALED; + return VK_FORMAT_R16_SINT; case Maxwell::VertexAttribute::Size::Size_16_16: - return VK_FORMAT_R16G16_USCALED; + return VK_FORMAT_R16G16_SINT; case Maxwell::VertexAttribute::Size::Size_16_16_16: - return VK_FORMAT_R16G16B16_USCALED; + return VK_FORMAT_R16G16B16_SINT; case Maxwell::VertexAttribute::Size::Size_16_16_16_16: - return VK_FORMAT_R16G16B16A16_USCALED; - default: - break; + return VK_FORMAT_R16G16B16A16_SINT; + case Maxwell::VertexAttribute::Size::Size_32: + return VK_FORMAT_R32_SINT; + case Maxwell::VertexAttribute::Size::Size_32_32: + return VK_FORMAT_R32G32_SINT; + case Maxwell::VertexAttribute::Size::Size_32_32_32: + return VK_FORMAT_R32G32B32_SINT; + case Maxwell::VertexAttribute::Size::Size_32_32_32_32: + return VK_FORMAT_R32G32B32A32_SINT; + case Maxwell::VertexAttribute::Size::Size_10_10_10_2: + return VK_FORMAT_A2B10G10R10_SINT_PACK32; } break; - case Maxwell::VertexAttribute::Type::SignedScaled: + case Maxwell::VertexAttribute::Type::Float: switch (size) { - case Maxwell::VertexAttribute::Size::Size_8: - return VK_FORMAT_R8_SSCALED; - case Maxwell::VertexAttribute::Size::Size_8_8: - return VK_FORMAT_R8G8_SSCALED; - case Maxwell::VertexAttribute::Size::Size_8_8_8: - return VK_FORMAT_R8G8B8_SSCALED; - case Maxwell::VertexAttribute::Size::Size_8_8_8_8: - return VK_FORMAT_R8G8B8A8_SSCALED; case Maxwell::VertexAttribute::Size::Size_16: - return VK_FORMAT_R16_SSCALED; + return VK_FORMAT_R16_SFLOAT; case Maxwell::VertexAttribute::Size::Size_16_16: - return VK_FORMAT_R16G16_SSCALED; + return VK_FORMAT_R16G16_SFLOAT; case Maxwell::VertexAttribute::Size::Size_16_16_16: - return VK_FORMAT_R16G16B16_SSCALED; + return VK_FORMAT_R16G16B16_SFLOAT; case Maxwell::VertexAttribute::Size::Size_16_16_16_16: - return VK_FORMAT_R16G16B16A16_SSCALED; - default: - break; - } - break; - case Maxwell::VertexAttribute::Type::Float: - switch (size) { + return VK_FORMAT_R16G16B16A16_SFLOAT; case Maxwell::VertexAttribute::Size::Size_32: return VK_FORMAT_R32_SFLOAT; case Maxwell::VertexAttribute::Size::Size_32_32: @@ -445,16 +460,6 @@ VkFormat VertexFormat(Maxwell::VertexAttribute::Type type, Maxwell::VertexAttrib return VK_FORMAT_R32G32B32_SFLOAT; case Maxwell::VertexAttribute::Size::Size_32_32_32_32: return VK_FORMAT_R32G32B32A32_SFLOAT; - case Maxwell::VertexAttribute::Size::Size_16: - return VK_FORMAT_R16_SFLOAT; - case Maxwell::VertexAttribute::Size::Size_16_16: - return VK_FORMAT_R16G16_SFLOAT; - case Maxwell::VertexAttribute::Size::Size_16_16_16: - return VK_FORMAT_R16G16B16_SFLOAT; - case Maxwell::VertexAttribute::Size::Size_16_16_16_16: - return VK_FORMAT_R16G16B16A16_SFLOAT; - default: - break; } break; } @@ -672,4 +677,27 @@ VkComponentSwizzle SwizzleSource(Tegra::Texture::SwizzleSource swizzle) { return {}; } +VkViewportCoordinateSwizzleNV ViewportSwizzle(Maxwell::ViewportSwizzle swizzle) { + switch (swizzle) { + case Maxwell::ViewportSwizzle::PositiveX: + return VK_VIEWPORT_COORDINATE_SWIZZLE_POSITIVE_X_NV; + case Maxwell::ViewportSwizzle::NegativeX: + return VK_VIEWPORT_COORDINATE_SWIZZLE_NEGATIVE_X_NV; + case Maxwell::ViewportSwizzle::PositiveY: + return VK_VIEWPORT_COORDINATE_SWIZZLE_POSITIVE_Y_NV; + case Maxwell::ViewportSwizzle::NegativeY: + return VK_VIEWPORT_COORDINATE_SWIZZLE_NEGATIVE_Y_NV; + case Maxwell::ViewportSwizzle::PositiveZ: + return VK_VIEWPORT_COORDINATE_SWIZZLE_POSITIVE_Z_NV; + case Maxwell::ViewportSwizzle::NegativeZ: + return VK_VIEWPORT_COORDINATE_SWIZZLE_NEGATIVE_Z_NV; + case Maxwell::ViewportSwizzle::PositiveW: + return VK_VIEWPORT_COORDINATE_SWIZZLE_POSITIVE_W_NV; + case Maxwell::ViewportSwizzle::NegativeW: + return VK_VIEWPORT_COORDINATE_SWIZZLE_NEGATIVE_W_NV; + } + UNREACHABLE_MSG("Invalid swizzle={}", static_cast<int>(swizzle)); + return {}; +} + } // namespace Vulkan::MaxwellToVK diff --git a/src/video_core/renderer_vulkan/maxwell_to_vk.h b/src/video_core/renderer_vulkan/maxwell_to_vk.h index 81bce4c6c..7e213452f 100644 --- a/src/video_core/renderer_vulkan/maxwell_to_vk.h +++ b/src/video_core/renderer_vulkan/maxwell_to_vk.h @@ -59,4 +59,6 @@ VkCullModeFlags CullFace(Maxwell::CullFace cull_face); VkComponentSwizzle SwizzleSource(Tegra::Texture::SwizzleSource swizzle); +VkViewportCoordinateSwizzleNV ViewportSwizzle(Maxwell::ViewportSwizzle swizzle); + } // namespace Vulkan::MaxwellToVK diff --git a/src/video_core/renderer_vulkan/renderer_vulkan.cpp b/src/video_core/renderer_vulkan/renderer_vulkan.cpp index 59b441943..2d9b18ed9 100644 --- a/src/video_core/renderer_vulkan/renderer_vulkan.cpp +++ b/src/video_core/renderer_vulkan/renderer_vulkan.cpp @@ -13,6 +13,7 @@ #include <fmt/format.h> #include "common/dynamic_library.h" +#include "common/file_util.h" #include "common/logging/log.h" #include "common/telemetry.h" #include "core/core.h" @@ -76,7 +77,8 @@ Common::DynamicLibrary OpenVulkanLibrary() { char* libvulkan_env = getenv("LIBVULKAN_PATH"); if (!libvulkan_env || !library.Open(libvulkan_env)) { // Use the libvulkan.dylib from the application bundle. - std::string filename = File::GetBundleDirectory() + "/Contents/Frameworks/libvulkan.dylib"; + const std::string filename = + FileUtil::GetBundleDirectory() + "/Contents/Frameworks/libvulkan.dylib"; library.Open(filename.c_str()); } #else @@ -153,11 +155,31 @@ vk::Instance CreateInstance(Common::DynamicLibrary& library, vk::InstanceDispatc } } - static constexpr std::array layers_data{"VK_LAYER_LUNARG_standard_validation"}; - vk::Span<const char*> layers = layers_data; - if (!enable_layers) { - layers = {}; + std::vector<const char*> layers; + layers.reserve(1); + if (enable_layers) { + layers.push_back("VK_LAYER_KHRONOS_validation"); + } + + const std::optional layer_properties = vk::EnumerateInstanceLayerProperties(dld); + if (!layer_properties) { + LOG_ERROR(Render_Vulkan, "Failed to query layer properties, disabling layers"); + layers.clear(); + } + + for (auto layer_it = layers.begin(); layer_it != layers.end();) { + const char* const layer = *layer_it; + const auto it = std::find_if( + layer_properties->begin(), layer_properties->end(), + [layer](const VkLayerProperties& prop) { return !std::strcmp(layer, prop.layerName); }); + if (it == layer_properties->end()) { + LOG_ERROR(Render_Vulkan, "Layer {} not available, removing it", layer); + layer_it = layers.erase(layer_it); + } else { + ++layer_it; + } } + vk::Instance instance = vk::Instance::Create(layers, extensions, dld); if (!instance) { LOG_ERROR(Render_Vulkan, "Failed to create Vulkan instance"); diff --git a/src/video_core/renderer_vulkan/vk_buffer_cache.cpp b/src/video_core/renderer_vulkan/vk_buffer_cache.cpp index 5b494da8c..f10f96cd8 100644 --- a/src/video_core/renderer_vulkan/vk_buffer_cache.cpp +++ b/src/video_core/renderer_vulkan/vk_buffer_cache.cpp @@ -7,6 +7,7 @@ #include <memory> #include "core/core.h" +#include "video_core/buffer_cache/buffer_cache.h" #include "video_core/renderer_vulkan/vk_buffer_cache.h" #include "video_core/renderer_vulkan/vk_device.h" #include "video_core/renderer_vulkan/vk_scheduler.h" @@ -36,9 +37,9 @@ std::unique_ptr<VKStreamBuffer> CreateStreamBuffer(const VKDevice& device, VKSch } // Anonymous namespace -CachedBufferBlock::CachedBufferBlock(const VKDevice& device, VKMemoryManager& memory_manager, - VAddr cpu_addr, std::size_t size) - : VideoCommon::BufferBlock{cpu_addr, size} { +Buffer::Buffer(const VKDevice& device, VKMemoryManager& memory_manager, VKScheduler& scheduler_, + VKStagingBufferPool& staging_pool_, VAddr cpu_addr, std::size_t size) + : VideoCommon::BufferBlock{cpu_addr, size}, scheduler{scheduler_}, staging_pool{staging_pool_} { VkBufferCreateInfo ci; ci.sType = VK_STRUCTURE_TYPE_BUFFER_CREATE_INFO; ci.pNext = nullptr; @@ -53,46 +54,17 @@ CachedBufferBlock::CachedBufferBlock(const VKDevice& device, VKMemoryManager& me buffer.commit = memory_manager.Commit(buffer.handle, false); } -CachedBufferBlock::~CachedBufferBlock() = default; +Buffer::~Buffer() = default; -VKBufferCache::VKBufferCache(VideoCore::RasterizerInterface& rasterizer, Core::System& system, - const VKDevice& device, VKMemoryManager& memory_manager, - VKScheduler& scheduler, VKStagingBufferPool& staging_pool) - : VideoCommon::BufferCache<Buffer, VkBuffer, VKStreamBuffer>{rasterizer, system, - CreateStreamBuffer(device, - scheduler)}, - device{device}, memory_manager{memory_manager}, scheduler{scheduler}, staging_pool{ - staging_pool} {} - -VKBufferCache::~VKBufferCache() = default; - -Buffer VKBufferCache::CreateBlock(VAddr cpu_addr, std::size_t size) { - return std::make_shared<CachedBufferBlock>(device, memory_manager, cpu_addr, size); -} - -VkBuffer VKBufferCache::ToHandle(const Buffer& buffer) { - return buffer->GetHandle(); -} - -VkBuffer VKBufferCache::GetEmptyBuffer(std::size_t size) { - size = std::max(size, std::size_t(4)); - const auto& empty = staging_pool.GetUnusedBuffer(size, false); - scheduler.RequestOutsideRenderPassOperationContext(); - scheduler.Record([size, buffer = *empty.handle](vk::CommandBuffer cmdbuf) { - cmdbuf.FillBuffer(buffer, 0, size, 0); - }); - return *empty.handle; -} - -void VKBufferCache::UploadBlockData(const Buffer& buffer, std::size_t offset, std::size_t size, - const u8* data) { +void Buffer::Upload(std::size_t offset, std::size_t size, const u8* data) const { const auto& staging = staging_pool.GetUnusedBuffer(size, true); std::memcpy(staging.commit->Map(size), data, size); scheduler.RequestOutsideRenderPassOperationContext(); - scheduler.Record([staging = *staging.handle, buffer = buffer->GetHandle(), offset, - size](vk::CommandBuffer cmdbuf) { - cmdbuf.CopyBuffer(staging, buffer, VkBufferCopy{0, offset, size}); + + const VkBuffer handle = Handle(); + scheduler.Record([staging = *staging.handle, handle, offset, size](vk::CommandBuffer cmdbuf) { + cmdbuf.CopyBuffer(staging, handle, VkBufferCopy{0, offset, size}); VkBufferMemoryBarrier barrier; barrier.sType = VK_STRUCTURE_TYPE_BUFFER_MEMORY_BARRIER; @@ -101,7 +73,7 @@ void VKBufferCache::UploadBlockData(const Buffer& buffer, std::size_t offset, st barrier.dstAccessMask = UPLOAD_ACCESS_BARRIERS; barrier.srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED; barrier.dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED; - barrier.buffer = buffer; + barrier.buffer = handle; barrier.offset = offset; barrier.size = size; cmdbuf.PipelineBarrier(VK_PIPELINE_STAGE_TRANSFER_BIT, UPLOAD_PIPELINE_STAGE, 0, {}, @@ -109,12 +81,12 @@ void VKBufferCache::UploadBlockData(const Buffer& buffer, std::size_t offset, st }); } -void VKBufferCache::DownloadBlockData(const Buffer& buffer, std::size_t offset, std::size_t size, - u8* data) { +void Buffer::Download(std::size_t offset, std::size_t size, u8* data) const { const auto& staging = staging_pool.GetUnusedBuffer(size, true); scheduler.RequestOutsideRenderPassOperationContext(); - scheduler.Record([staging = *staging.handle, buffer = buffer->GetHandle(), offset, - size](vk::CommandBuffer cmdbuf) { + + const VkBuffer handle = Handle(); + scheduler.Record([staging = *staging.handle, handle, offset, size](vk::CommandBuffer cmdbuf) { VkBufferMemoryBarrier barrier; barrier.sType = VK_STRUCTURE_TYPE_BUFFER_MEMORY_BARRIER; barrier.pNext = nullptr; @@ -122,7 +94,7 @@ void VKBufferCache::DownloadBlockData(const Buffer& buffer, std::size_t offset, barrier.dstAccessMask = VK_ACCESS_TRANSFER_READ_BIT; barrier.srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED; barrier.dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED; - barrier.buffer = buffer; + barrier.buffer = handle; barrier.offset = offset; barrier.size = size; @@ -130,18 +102,20 @@ void VKBufferCache::DownloadBlockData(const Buffer& buffer, std::size_t offset, VK_PIPELINE_STAGE_FRAGMENT_SHADER_BIT | VK_PIPELINE_STAGE_COMPUTE_SHADER_BIT, VK_PIPELINE_STAGE_TRANSFER_BIT, 0, {}, barrier, {}); - cmdbuf.CopyBuffer(buffer, staging, VkBufferCopy{offset, 0, size}); + cmdbuf.CopyBuffer(handle, staging, VkBufferCopy{offset, 0, size}); }); scheduler.Finish(); std::memcpy(data, staging.commit->Map(size), size); } -void VKBufferCache::CopyBlock(const Buffer& src, const Buffer& dst, std::size_t src_offset, - std::size_t dst_offset, std::size_t size) { +void Buffer::CopyFrom(const Buffer& src, std::size_t src_offset, std::size_t dst_offset, + std::size_t size) const { scheduler.RequestOutsideRenderPassOperationContext(); - scheduler.Record([src_buffer = src->GetHandle(), dst_buffer = dst->GetHandle(), src_offset, - dst_offset, size](vk::CommandBuffer cmdbuf) { + + const VkBuffer dst_buffer = Handle(); + scheduler.Record([src_buffer = src.Handle(), dst_buffer, src_offset, dst_offset, + size](vk::CommandBuffer cmdbuf) { cmdbuf.CopyBuffer(src_buffer, dst_buffer, VkBufferCopy{src_offset, dst_offset, size}); std::array<VkBufferMemoryBarrier, 2> barriers; @@ -168,4 +142,30 @@ void VKBufferCache::CopyBlock(const Buffer& src, const Buffer& dst, std::size_t }); } +VKBufferCache::VKBufferCache(VideoCore::RasterizerInterface& rasterizer, Core::System& system, + const VKDevice& device, VKMemoryManager& memory_manager, + VKScheduler& scheduler, VKStagingBufferPool& staging_pool) + : VideoCommon::BufferCache<Buffer, VkBuffer, VKStreamBuffer>{rasterizer, system, + CreateStreamBuffer(device, + scheduler)}, + device{device}, memory_manager{memory_manager}, scheduler{scheduler}, staging_pool{ + staging_pool} {} + +VKBufferCache::~VKBufferCache() = default; + +std::shared_ptr<Buffer> VKBufferCache::CreateBlock(VAddr cpu_addr, std::size_t size) { + return std::make_shared<Buffer>(device, memory_manager, scheduler, staging_pool, cpu_addr, + size); +} + +VKBufferCache::BufferInfo VKBufferCache::GetEmptyBuffer(std::size_t size) { + size = std::max(size, std::size_t(4)); + const auto& empty = staging_pool.GetUnusedBuffer(size, false); + scheduler.RequestOutsideRenderPassOperationContext(); + scheduler.Record([size, buffer = *empty.handle](vk::CommandBuffer cmdbuf) { + cmdbuf.FillBuffer(buffer, 0, size, 0); + }); + return {*empty.handle, 0, 0}; +} + } // namespace Vulkan diff --git a/src/video_core/renderer_vulkan/vk_buffer_cache.h b/src/video_core/renderer_vulkan/vk_buffer_cache.h index a54583e7d..3630aca77 100644 --- a/src/video_core/renderer_vulkan/vk_buffer_cache.h +++ b/src/video_core/renderer_vulkan/vk_buffer_cache.h @@ -8,7 +8,6 @@ #include "common/common_types.h" #include "video_core/buffer_cache/buffer_cache.h" -#include "video_core/rasterizer_cache.h" #include "video_core/renderer_vulkan/vk_memory_manager.h" #include "video_core/renderer_vulkan/vk_staging_buffer_pool.h" #include "video_core/renderer_vulkan/vk_stream_buffer.h" @@ -24,22 +23,34 @@ class VKDevice; class VKMemoryManager; class VKScheduler; -class CachedBufferBlock final : public VideoCommon::BufferBlock { +class Buffer final : public VideoCommon::BufferBlock { public: - explicit CachedBufferBlock(const VKDevice& device, VKMemoryManager& memory_manager, - VAddr cpu_addr, std::size_t size); - ~CachedBufferBlock(); + explicit Buffer(const VKDevice& device, VKMemoryManager& memory_manager, VKScheduler& scheduler, + VKStagingBufferPool& staging_pool, VAddr cpu_addr, std::size_t size); + ~Buffer(); - VkBuffer GetHandle() const { + void Upload(std::size_t offset, std::size_t size, const u8* data) const; + + void Download(std::size_t offset, std::size_t size, u8* data) const; + + void CopyFrom(const Buffer& src, std::size_t src_offset, std::size_t dst_offset, + std::size_t size) const; + + VkBuffer Handle() const { return *buffer.handle; } + u64 Address() const { + return 0; + } + private: + VKScheduler& scheduler; + VKStagingBufferPool& staging_pool; + VKBuffer buffer; }; -using Buffer = std::shared_ptr<CachedBufferBlock>; - class VKBufferCache final : public VideoCommon::BufferCache<Buffer, VkBuffer, VKStreamBuffer> { public: explicit VKBufferCache(VideoCore::RasterizerInterface& rasterizer, Core::System& system, @@ -47,21 +58,10 @@ public: VKScheduler& scheduler, VKStagingBufferPool& staging_pool); ~VKBufferCache(); - VkBuffer GetEmptyBuffer(std::size_t size) override; + BufferInfo GetEmptyBuffer(std::size_t size) override; protected: - VkBuffer ToHandle(const Buffer& buffer) override; - - Buffer CreateBlock(VAddr cpu_addr, std::size_t size) override; - - void UploadBlockData(const Buffer& buffer, std::size_t offset, std::size_t size, - const u8* data) override; - - void DownloadBlockData(const Buffer& buffer, std::size_t offset, std::size_t size, - u8* data) override; - - void CopyBlock(const Buffer& src, const Buffer& dst, std::size_t src_offset, - std::size_t dst_offset, std::size_t size) override; + std::shared_ptr<Buffer> CreateBlock(VAddr cpu_addr, std::size_t size) override; private: const VKDevice& device; diff --git a/src/video_core/renderer_vulkan/vk_compute_pipeline.cpp b/src/video_core/renderer_vulkan/vk_compute_pipeline.cpp index 8e1b46277..281bf9ac3 100644 --- a/src/video_core/renderer_vulkan/vk_compute_pipeline.cpp +++ b/src/video_core/renderer_vulkan/vk_compute_pipeline.cpp @@ -53,8 +53,9 @@ vk::DescriptorSetLayout VKComputePipeline::CreateDescriptorSetLayout() const { }; add_bindings(VK_DESCRIPTOR_TYPE_UNIFORM_BUFFER, entries.const_buffers.size()); add_bindings(VK_DESCRIPTOR_TYPE_STORAGE_BUFFER, entries.global_buffers.size()); - add_bindings(VK_DESCRIPTOR_TYPE_UNIFORM_TEXEL_BUFFER, entries.texel_buffers.size()); + add_bindings(VK_DESCRIPTOR_TYPE_UNIFORM_TEXEL_BUFFER, entries.uniform_texels.size()); add_bindings(VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER, entries.samplers.size()); + add_bindings(VK_DESCRIPTOR_TYPE_STORAGE_TEXEL_BUFFER, entries.storage_texels.size()); add_bindings(VK_DESCRIPTOR_TYPE_STORAGE_IMAGE, entries.images.size()); VkDescriptorSetLayoutCreateInfo ci; diff --git a/src/video_core/renderer_vulkan/vk_descriptor_pool.cpp b/src/video_core/renderer_vulkan/vk_descriptor_pool.cpp index 890fd52cf..9259b618d 100644 --- a/src/video_core/renderer_vulkan/vk_descriptor_pool.cpp +++ b/src/video_core/renderer_vulkan/vk_descriptor_pool.cpp @@ -42,6 +42,7 @@ vk::DescriptorPool* VKDescriptorPool::AllocateNewPool() { {VK_DESCRIPTOR_TYPE_STORAGE_BUFFER, num_sets * 60}, {VK_DESCRIPTOR_TYPE_UNIFORM_TEXEL_BUFFER, num_sets * 64}, {VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER, num_sets * 64}, + {VK_DESCRIPTOR_TYPE_STORAGE_TEXEL_BUFFER, num_sets * 64}, {VK_DESCRIPTOR_TYPE_STORAGE_IMAGE, num_sets * 40}}; VkDescriptorPoolCreateInfo ci; diff --git a/src/video_core/renderer_vulkan/vk_device.cpp b/src/video_core/renderer_vulkan/vk_device.cpp index 0e4bbca97..9fd8ac3f6 100644 --- a/src/video_core/renderer_vulkan/vk_device.cpp +++ b/src/video_core/renderer_vulkan/vk_device.cpp @@ -73,74 +73,79 @@ VkFormatFeatureFlags GetFormatFeatures(VkFormatProperties properties, FormatType std::unordered_map<VkFormat, VkFormatProperties> GetFormatProperties( vk::PhysicalDevice physical, const vk::InstanceDispatch& dld) { - static constexpr std::array formats{VK_FORMAT_A8B8G8R8_UNORM_PACK32, - VK_FORMAT_A8B8G8R8_UINT_PACK32, - VK_FORMAT_A8B8G8R8_SNORM_PACK32, - VK_FORMAT_A8B8G8R8_SRGB_PACK32, - VK_FORMAT_B5G6R5_UNORM_PACK16, - VK_FORMAT_A2B10G10R10_UNORM_PACK32, - VK_FORMAT_A1R5G5B5_UNORM_PACK16, - VK_FORMAT_R32G32B32A32_SFLOAT, - VK_FORMAT_R32G32B32A32_UINT, - VK_FORMAT_R32G32_SFLOAT, - VK_FORMAT_R32G32_UINT, - VK_FORMAT_R16G16B16A16_UINT, - VK_FORMAT_R16G16B16A16_SNORM, - VK_FORMAT_R16G16B16A16_UNORM, - VK_FORMAT_R16G16_UNORM, - VK_FORMAT_R16G16_SNORM, - VK_FORMAT_R16G16_SFLOAT, - VK_FORMAT_R16_UNORM, - VK_FORMAT_R8G8B8A8_SRGB, - VK_FORMAT_R8G8_UNORM, - VK_FORMAT_R8G8_SNORM, - VK_FORMAT_R8_UNORM, - VK_FORMAT_R8_UINT, - VK_FORMAT_B10G11R11_UFLOAT_PACK32, - VK_FORMAT_R32_SFLOAT, - VK_FORMAT_R32_UINT, - VK_FORMAT_R32_SINT, - VK_FORMAT_R16_SFLOAT, - VK_FORMAT_R16G16B16A16_SFLOAT, - VK_FORMAT_B8G8R8A8_UNORM, - VK_FORMAT_R4G4B4A4_UNORM_PACK16, - VK_FORMAT_D32_SFLOAT, - VK_FORMAT_D16_UNORM, - VK_FORMAT_D16_UNORM_S8_UINT, - VK_FORMAT_D24_UNORM_S8_UINT, - VK_FORMAT_D32_SFLOAT_S8_UINT, - VK_FORMAT_BC1_RGBA_UNORM_BLOCK, - VK_FORMAT_BC2_UNORM_BLOCK, - VK_FORMAT_BC3_UNORM_BLOCK, - VK_FORMAT_BC4_UNORM_BLOCK, - VK_FORMAT_BC5_UNORM_BLOCK, - VK_FORMAT_BC5_SNORM_BLOCK, - VK_FORMAT_BC7_UNORM_BLOCK, - VK_FORMAT_BC6H_UFLOAT_BLOCK, - VK_FORMAT_BC6H_SFLOAT_BLOCK, - VK_FORMAT_BC1_RGBA_SRGB_BLOCK, - VK_FORMAT_BC2_SRGB_BLOCK, - VK_FORMAT_BC3_SRGB_BLOCK, - VK_FORMAT_BC7_SRGB_BLOCK, - VK_FORMAT_ASTC_4x4_SRGB_BLOCK, - VK_FORMAT_ASTC_8x8_SRGB_BLOCK, - VK_FORMAT_ASTC_8x5_SRGB_BLOCK, - VK_FORMAT_ASTC_5x4_SRGB_BLOCK, - VK_FORMAT_ASTC_5x5_UNORM_BLOCK, - VK_FORMAT_ASTC_5x5_SRGB_BLOCK, - VK_FORMAT_ASTC_10x8_UNORM_BLOCK, - VK_FORMAT_ASTC_10x8_SRGB_BLOCK, - VK_FORMAT_ASTC_6x6_UNORM_BLOCK, - VK_FORMAT_ASTC_6x6_SRGB_BLOCK, - VK_FORMAT_ASTC_10x10_UNORM_BLOCK, - VK_FORMAT_ASTC_10x10_SRGB_BLOCK, - VK_FORMAT_ASTC_12x12_UNORM_BLOCK, - VK_FORMAT_ASTC_12x12_SRGB_BLOCK, - VK_FORMAT_ASTC_8x6_UNORM_BLOCK, - VK_FORMAT_ASTC_8x6_SRGB_BLOCK, - VK_FORMAT_ASTC_6x5_UNORM_BLOCK, - VK_FORMAT_ASTC_6x5_SRGB_BLOCK, - VK_FORMAT_E5B9G9R9_UFLOAT_PACK32}; + static constexpr std::array formats{ + VK_FORMAT_A8B8G8R8_UNORM_PACK32, + VK_FORMAT_A8B8G8R8_UINT_PACK32, + VK_FORMAT_A8B8G8R8_SNORM_PACK32, + VK_FORMAT_A8B8G8R8_SRGB_PACK32, + VK_FORMAT_B5G6R5_UNORM_PACK16, + VK_FORMAT_A2B10G10R10_UNORM_PACK32, + VK_FORMAT_A1R5G5B5_UNORM_PACK16, + VK_FORMAT_R32G32B32A32_SFLOAT, + VK_FORMAT_R32G32B32A32_UINT, + VK_FORMAT_R32G32_SFLOAT, + VK_FORMAT_R32G32_UINT, + VK_FORMAT_R16G16B16A16_UINT, + VK_FORMAT_R16G16B16A16_SNORM, + VK_FORMAT_R16G16B16A16_UNORM, + VK_FORMAT_R16G16_UNORM, + VK_FORMAT_R16G16_SNORM, + VK_FORMAT_R16G16_SFLOAT, + VK_FORMAT_R16_UNORM, + VK_FORMAT_R16_UINT, + VK_FORMAT_R8G8B8A8_SRGB, + VK_FORMAT_R8G8_UNORM, + VK_FORMAT_R8G8_SNORM, + VK_FORMAT_R8G8_UINT, + VK_FORMAT_R8_UNORM, + VK_FORMAT_R8_UINT, + VK_FORMAT_B10G11R11_UFLOAT_PACK32, + VK_FORMAT_R32_SFLOAT, + VK_FORMAT_R32_UINT, + VK_FORMAT_R32_SINT, + VK_FORMAT_R16_SFLOAT, + VK_FORMAT_R16G16B16A16_SFLOAT, + VK_FORMAT_B8G8R8A8_UNORM, + VK_FORMAT_B8G8R8A8_SRGB, + VK_FORMAT_R4G4B4A4_UNORM_PACK16, + VK_FORMAT_D32_SFLOAT, + VK_FORMAT_D16_UNORM, + VK_FORMAT_D16_UNORM_S8_UINT, + VK_FORMAT_D24_UNORM_S8_UINT, + VK_FORMAT_D32_SFLOAT_S8_UINT, + VK_FORMAT_BC1_RGBA_UNORM_BLOCK, + VK_FORMAT_BC2_UNORM_BLOCK, + VK_FORMAT_BC3_UNORM_BLOCK, + VK_FORMAT_BC4_UNORM_BLOCK, + VK_FORMAT_BC5_UNORM_BLOCK, + VK_FORMAT_BC5_SNORM_BLOCK, + VK_FORMAT_BC7_UNORM_BLOCK, + VK_FORMAT_BC6H_UFLOAT_BLOCK, + VK_FORMAT_BC6H_SFLOAT_BLOCK, + VK_FORMAT_BC1_RGBA_SRGB_BLOCK, + VK_FORMAT_BC2_SRGB_BLOCK, + VK_FORMAT_BC3_SRGB_BLOCK, + VK_FORMAT_BC7_SRGB_BLOCK, + VK_FORMAT_ASTC_4x4_SRGB_BLOCK, + VK_FORMAT_ASTC_8x8_SRGB_BLOCK, + VK_FORMAT_ASTC_8x5_SRGB_BLOCK, + VK_FORMAT_ASTC_5x4_SRGB_BLOCK, + VK_FORMAT_ASTC_5x5_UNORM_BLOCK, + VK_FORMAT_ASTC_5x5_SRGB_BLOCK, + VK_FORMAT_ASTC_10x8_UNORM_BLOCK, + VK_FORMAT_ASTC_10x8_SRGB_BLOCK, + VK_FORMAT_ASTC_6x6_UNORM_BLOCK, + VK_FORMAT_ASTC_6x6_SRGB_BLOCK, + VK_FORMAT_ASTC_10x10_UNORM_BLOCK, + VK_FORMAT_ASTC_10x10_SRGB_BLOCK, + VK_FORMAT_ASTC_12x12_UNORM_BLOCK, + VK_FORMAT_ASTC_12x12_SRGB_BLOCK, + VK_FORMAT_ASTC_8x6_UNORM_BLOCK, + VK_FORMAT_ASTC_8x6_SRGB_BLOCK, + VK_FORMAT_ASTC_6x5_UNORM_BLOCK, + VK_FORMAT_ASTC_6x5_SRGB_BLOCK, + VK_FORMAT_E5B9G9R9_UFLOAT_PACK32, + }; std::unordered_map<VkFormat, VkFormatProperties> format_properties; for (const auto format : formats) { format_properties.emplace(format, physical.GetFormatProperties(format)); @@ -260,6 +265,10 @@ bool VKDevice::Create() { LOG_INFO(Render_Vulkan, "Device doesn't support float16 natively"); } + if (!nv_viewport_swizzle) { + LOG_INFO(Render_Vulkan, "Device doesn't support viewport swizzles"); + } + VkPhysicalDeviceUniformBufferStandardLayoutFeaturesKHR std430_layout; if (khr_uniform_buffer_standard_layout) { std430_layout.sType = @@ -293,6 +302,17 @@ bool VKDevice::Create() { LOG_INFO(Render_Vulkan, "Device doesn't support transform feedbacks"); } + VkPhysicalDeviceCustomBorderColorFeaturesEXT custom_border; + if (ext_custom_border_color) { + custom_border.sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_CUSTOM_BORDER_COLOR_FEATURES_EXT; + custom_border.pNext = nullptr; + custom_border.customBorderColors = VK_TRUE; + custom_border.customBorderColorWithoutFormat = VK_TRUE; + SetNext(next, custom_border); + } else { + LOG_INFO(Render_Vulkan, "Device doesn't support custom border colors"); + } + if (!ext_depth_range_unrestricted) { LOG_INFO(Render_Vulkan, "Device doesn't support depth range unrestricted"); } @@ -520,7 +540,9 @@ std::vector<const char*> VKDevice::LoadExtensions() { bool has_khr_shader_float16_int8{}; bool has_ext_subgroup_size_control{}; bool has_ext_transform_feedback{}; + bool has_ext_custom_border_color{}; for (const auto& extension : physical.EnumerateDeviceExtensionProperties()) { + Test(extension, nv_viewport_swizzle, VK_NV_VIEWPORT_SWIZZLE_EXTENSION_NAME, true); Test(extension, khr_uniform_buffer_standard_layout, VK_KHR_UNIFORM_BUFFER_STANDARD_LAYOUT_EXTENSION_NAME, true); Test(extension, has_khr_shader_float16_int8, VK_KHR_SHADER_FLOAT16_INT8_EXTENSION_NAME, @@ -534,6 +556,8 @@ std::vector<const char*> VKDevice::LoadExtensions() { false); Test(extension, has_ext_transform_feedback, VK_EXT_TRANSFORM_FEEDBACK_EXTENSION_NAME, false); + Test(extension, has_ext_custom_border_color, VK_EXT_CUSTOM_BORDER_COLOR_EXTENSION_NAME, + false); if (Settings::values.renderer_debug) { Test(extension, nv_device_diagnostics_config, VK_NV_DEVICE_DIAGNOSTICS_CONFIG_EXTENSION_NAME, true); @@ -606,6 +630,19 @@ std::vector<const char*> VKDevice::LoadExtensions() { } } + if (has_ext_custom_border_color) { + VkPhysicalDeviceCustomBorderColorFeaturesEXT border_features; + border_features.sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_CUSTOM_BORDER_COLOR_FEATURES_EXT; + border_features.pNext = nullptr; + features.pNext = &border_features; + physical.GetFeatures2KHR(features); + + if (border_features.customBorderColors && border_features.customBorderColorWithoutFormat) { + extensions.push_back(VK_EXT_CUSTOM_BORDER_COLOR_EXTENSION_NAME); + ext_custom_border_color = true; + } + } + return extensions; } diff --git a/src/video_core/renderer_vulkan/vk_device.h b/src/video_core/renderer_vulkan/vk_device.h index c8640762d..6b9227b09 100644 --- a/src/video_core/renderer_vulkan/vk_device.h +++ b/src/video_core/renderer_vulkan/vk_device.h @@ -147,6 +147,11 @@ public: return is_formatless_image_load_supported; } + /// Returns true if the device supports VK_NV_viewport_swizzle. + bool IsNvViewportSwizzleSupported() const { + return nv_viewport_swizzle; + } + /// Returns true if the device supports VK_EXT_scalar_block_layout. bool IsKhrUniformBufferStandardLayoutSupported() const { return khr_uniform_buffer_standard_layout; @@ -172,6 +177,11 @@ public: return ext_transform_feedback; } + /// Returns true if the device supports VK_EXT_custom_border_color. + bool IsExtCustomBorderColorSupported() const { + return ext_custom_border_color; + } + /// Returns the vendor name reported from Vulkan. std::string_view GetVendorName() const { return vendor_name; @@ -222,11 +232,13 @@ private: bool is_float16_supported{}; ///< Support for float16 arithmetics. bool is_warp_potentially_bigger{}; ///< Host warp size can be bigger than guest. bool is_formatless_image_load_supported{}; ///< Support for shader image read without format. + bool nv_viewport_swizzle{}; ///< Support for VK_NV_viewport_swizzle. bool khr_uniform_buffer_standard_layout{}; ///< Support for std430 on UBOs. bool ext_index_type_uint8{}; ///< Support for VK_EXT_index_type_uint8. bool ext_depth_range_unrestricted{}; ///< Support for VK_EXT_depth_range_unrestricted. bool ext_shader_viewport_index_layer{}; ///< Support for VK_EXT_shader_viewport_index_layer. bool ext_transform_feedback{}; ///< Support for VK_EXT_transform_feedback. + bool ext_custom_border_color{}; ///< Support for VK_EXT_custom_border_color. bool nv_device_diagnostics_config{}; ///< Support for VK_NV_device_diagnostics_config. // Telemetry parameters diff --git a/src/video_core/renderer_vulkan/vk_fence_manager.h b/src/video_core/renderer_vulkan/vk_fence_manager.h index 04d07fe6a..043fe7947 100644 --- a/src/video_core/renderer_vulkan/vk_fence_manager.h +++ b/src/video_core/renderer_vulkan/vk_fence_manager.h @@ -7,6 +7,7 @@ #include <memory> #include "video_core/fence_manager.h" +#include "video_core/renderer_vulkan/vk_buffer_cache.h" #include "video_core/renderer_vulkan/wrapper.h" namespace Core { diff --git a/src/video_core/renderer_vulkan/vk_graphics_pipeline.cpp b/src/video_core/renderer_vulkan/vk_graphics_pipeline.cpp index 1ac981974..69b6bba00 100644 --- a/src/video_core/renderer_vulkan/vk_graphics_pipeline.cpp +++ b/src/video_core/renderer_vulkan/vk_graphics_pipeline.cpp @@ -2,6 +2,7 @@ // Licensed under GPLv2 or any later version // Refer to the license.txt file included. +#include <algorithm> #include <array> #include <cstring> #include <vector> @@ -50,6 +51,23 @@ bool SupportsPrimitiveRestart(VkPrimitiveTopology topology) { topology) == std::end(unsupported_topologies); } +VkViewportSwizzleNV UnpackViewportSwizzle(u16 swizzle) { + union { + u32 raw; + BitField<0, 3, Maxwell::ViewportSwizzle> x; + BitField<4, 3, Maxwell::ViewportSwizzle> y; + BitField<8, 3, Maxwell::ViewportSwizzle> z; + BitField<12, 3, Maxwell::ViewportSwizzle> w; + } const unpacked{swizzle}; + + VkViewportSwizzleNV result; + result.x = MaxwellToVK::ViewportSwizzle(unpacked.x); + result.y = MaxwellToVK::ViewportSwizzle(unpacked.y); + result.z = MaxwellToVK::ViewportSwizzle(unpacked.z); + result.w = MaxwellToVK::ViewportSwizzle(unpacked.w); + return result; +} + } // Anonymous namespace VKGraphicsPipeline::VKGraphicsPipeline(const VKDevice& device, VKScheduler& scheduler, @@ -162,6 +180,7 @@ vk::Pipeline VKGraphicsPipeline::CreatePipeline(const RenderPassParams& renderpa const auto& ds = fixed_state.depth_stencil; const auto& cd = fixed_state.color_blending; const auto& rs = fixed_state.rasterizer; + const auto& viewport_swizzles = fixed_state.viewport_swizzles.swizzles; std::vector<VkVertexInputBindingDescription> vertex_bindings; std::vector<VkVertexInputBindingDivisorDescriptionEXT> vertex_binding_divisors; @@ -244,12 +263,25 @@ vk::Pipeline VKGraphicsPipeline::CreatePipeline(const RenderPassParams& renderpa viewport_ci.scissorCount = Maxwell::NumViewports; viewport_ci.pScissors = nullptr; + std::array<VkViewportSwizzleNV, Maxwell::NumViewports> swizzles; + std::transform(viewport_swizzles.begin(), viewport_swizzles.end(), swizzles.begin(), + UnpackViewportSwizzle); + VkPipelineViewportSwizzleStateCreateInfoNV swizzle_ci; + swizzle_ci.sType = VK_STRUCTURE_TYPE_PIPELINE_VIEWPORT_SWIZZLE_STATE_CREATE_INFO_NV; + swizzle_ci.pNext = nullptr; + swizzle_ci.flags = 0; + swizzle_ci.viewportCount = Maxwell::NumViewports; + swizzle_ci.pViewportSwizzles = swizzles.data(); + if (device.IsNvViewportSwizzleSupported()) { + viewport_ci.pNext = &swizzle_ci; + } + VkPipelineRasterizationStateCreateInfo rasterization_ci; rasterization_ci.sType = VK_STRUCTURE_TYPE_PIPELINE_RASTERIZATION_STATE_CREATE_INFO; rasterization_ci.pNext = nullptr; rasterization_ci.flags = 0; rasterization_ci.depthClampEnable = rs.depth_clamp_disabled == 0 ? VK_TRUE : VK_FALSE; - rasterization_ci.rasterizerDiscardEnable = VK_FALSE; + rasterization_ci.rasterizerDiscardEnable = rs.rasterize_enable == 0 ? VK_TRUE : VK_FALSE; rasterization_ci.polygonMode = VK_POLYGON_MODE_FILL; rasterization_ci.cullMode = rs.cull_enable ? MaxwellToVK::CullFace(rs.CullFace()) : VK_CULL_MODE_NONE; diff --git a/src/video_core/renderer_vulkan/vk_pipeline_cache.cpp b/src/video_core/renderer_vulkan/vk_pipeline_cache.cpp index fe45ed269..ea66e621e 100644 --- a/src/video_core/renderer_vulkan/vk_pipeline_cache.cpp +++ b/src/video_core/renderer_vulkan/vk_pipeline_cache.cpp @@ -27,6 +27,7 @@ #include "video_core/renderer_vulkan/wrapper.h" #include "video_core/shader/compiler_settings.h" #include "video_core/shader/memory_util.h" +#include "video_core/shader_cache.h" namespace Vulkan { @@ -45,6 +46,7 @@ constexpr VkDescriptorType UNIFORM_BUFFER = VK_DESCRIPTOR_TYPE_UNIFORM_BUFFER; constexpr VkDescriptorType STORAGE_BUFFER = VK_DESCRIPTOR_TYPE_STORAGE_BUFFER; constexpr VkDescriptorType UNIFORM_TEXEL_BUFFER = VK_DESCRIPTOR_TYPE_UNIFORM_TEXEL_BUFFER; constexpr VkDescriptorType COMBINED_IMAGE_SAMPLER = VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER; +constexpr VkDescriptorType STORAGE_TEXEL_BUFFER = VK_DESCRIPTOR_TYPE_STORAGE_TEXEL_BUFFER; constexpr VkDescriptorType STORAGE_IMAGE = VK_DESCRIPTOR_TYPE_STORAGE_IMAGE; constexpr VideoCommon::Shader::CompilerSettings compiler_settings{ @@ -104,8 +106,9 @@ u32 FillDescriptorLayout(const ShaderEntries& entries, u32 binding = base_binding; AddBindings<UNIFORM_BUFFER>(bindings, binding, flags, entries.const_buffers); AddBindings<STORAGE_BUFFER>(bindings, binding, flags, entries.global_buffers); - AddBindings<UNIFORM_TEXEL_BUFFER>(bindings, binding, flags, entries.texel_buffers); + AddBindings<UNIFORM_TEXEL_BUFFER>(bindings, binding, flags, entries.uniform_texels); AddBindings<COMBINED_IMAGE_SAMPLER>(bindings, binding, flags, entries.samplers); + AddBindings<STORAGE_TEXEL_BUFFER>(bindings, binding, flags, entries.storage_texels); AddBindings<STORAGE_IMAGE>(bindings, binding, flags, entries.images); return binding; } @@ -130,19 +133,18 @@ bool ComputePipelineCacheKey::operator==(const ComputePipelineCacheKey& rhs) con return std::memcmp(&rhs, this, sizeof *this) == 0; } -CachedShader::CachedShader(Core::System& system, Tegra::Engines::ShaderType stage, - GPUVAddr gpu_addr, VAddr cpu_addr, ProgramCode program_code, - u32 main_offset) - : RasterizerCacheObject{cpu_addr}, gpu_addr{gpu_addr}, program_code{std::move(program_code)}, +Shader::Shader(Core::System& system, Tegra::Engines::ShaderType stage, GPUVAddr gpu_addr, + VideoCommon::Shader::ProgramCode program_code, u32 main_offset) + : gpu_addr{gpu_addr}, program_code{std::move(program_code)}, registry{stage, GetEngine(system, stage)}, shader_ir{this->program_code, main_offset, compiler_settings, registry}, entries{GenerateShaderEntries(shader_ir)} {} -CachedShader::~CachedShader() = default; +Shader::~Shader() = default; -Tegra::Engines::ConstBufferEngineInterface& CachedShader::GetEngine( - Core::System& system, Tegra::Engines::ShaderType stage) { - if (stage == Tegra::Engines::ShaderType::Compute) { +Tegra::Engines::ConstBufferEngineInterface& Shader::GetEngine(Core::System& system, + Tegra::Engines::ShaderType stage) { + if (stage == ShaderType::Compute) { return system.GPU().KeplerCompute(); } else { return system.GPU().Maxwell3D(); @@ -154,16 +156,16 @@ VKPipelineCache::VKPipelineCache(Core::System& system, RasterizerVulkan& rasteri VKDescriptorPool& descriptor_pool, VKUpdateDescriptorQueue& update_descriptor_queue, VKRenderPassCache& renderpass_cache) - : RasterizerCache{rasterizer}, system{system}, device{device}, scheduler{scheduler}, - descriptor_pool{descriptor_pool}, update_descriptor_queue{update_descriptor_queue}, - renderpass_cache{renderpass_cache} {} + : VideoCommon::ShaderCache<Shader>{rasterizer}, system{system}, device{device}, + scheduler{scheduler}, descriptor_pool{descriptor_pool}, + update_descriptor_queue{update_descriptor_queue}, renderpass_cache{renderpass_cache} {} VKPipelineCache::~VKPipelineCache() = default; -std::array<Shader, Maxwell::MaxShaderProgram> VKPipelineCache::GetShaders() { +std::array<Shader*, Maxwell::MaxShaderProgram> VKPipelineCache::GetShaders() { const auto& gpu = system.GPU().Maxwell3D(); - std::array<Shader, Maxwell::MaxShaderProgram> shaders; + std::array<Shader*, Maxwell::MaxShaderProgram> shaders{}; for (std::size_t index = 0; index < Maxwell::MaxShaderProgram; ++index) { const auto program{static_cast<Maxwell::ShaderProgram>(index)}; @@ -176,24 +178,28 @@ std::array<Shader, Maxwell::MaxShaderProgram> VKPipelineCache::GetShaders() { const GPUVAddr program_addr{GetShaderAddress(system, program)}; const std::optional cpu_addr = memory_manager.GpuToCpuAddress(program_addr); ASSERT(cpu_addr); - auto shader = cpu_addr ? TryGet(*cpu_addr) : null_shader; - if (!shader) { + + Shader* result = cpu_addr ? TryGet(*cpu_addr) : null_shader.get(); + if (!result) { const auto host_ptr{memory_manager.GetPointer(program_addr)}; // No shader found - create a new one constexpr u32 stage_offset = STAGE_MAIN_OFFSET; - const auto stage = static_cast<Tegra::Engines::ShaderType>(index == 0 ? 0 : index - 1); + const auto stage = static_cast<ShaderType>(index == 0 ? 0 : index - 1); ProgramCode code = GetShaderCode(memory_manager, program_addr, host_ptr, false); + const std::size_t size_in_bytes = code.size() * sizeof(u64); + + auto shader = std::make_unique<Shader>(system, stage, program_addr, std::move(code), + stage_offset); + result = shader.get(); - shader = std::make_shared<CachedShader>(system, stage, program_addr, *cpu_addr, - std::move(code), stage_offset); if (cpu_addr) { - Register(shader); + Register(std::move(shader), *cpu_addr, size_in_bytes); } else { - null_shader = shader; + null_shader = std::move(shader); } } - shaders[index] = std::move(shader); + shaders[index] = result; } return last_shaders = shaders; } @@ -234,19 +240,22 @@ VKComputePipeline& VKPipelineCache::GetComputePipeline(const ComputePipelineCach const auto cpu_addr = memory_manager.GpuToCpuAddress(program_addr); ASSERT(cpu_addr); - auto shader = cpu_addr ? TryGet(*cpu_addr) : null_kernel; + Shader* shader = cpu_addr ? TryGet(*cpu_addr) : null_kernel.get(); if (!shader) { // No shader found - create a new one const auto host_ptr = memory_manager.GetPointer(program_addr); ProgramCode code = GetShaderCode(memory_manager, program_addr, host_ptr, true); - shader = std::make_shared<CachedShader>(system, Tegra::Engines::ShaderType::Compute, - program_addr, *cpu_addr, std::move(code), - KERNEL_MAIN_OFFSET); + const std::size_t size_in_bytes = code.size() * sizeof(u64); + + auto shader_info = std::make_unique<Shader>(system, ShaderType::Compute, program_addr, + std::move(code), KERNEL_MAIN_OFFSET); + shader = shader_info.get(); + if (cpu_addr) { - Register(shader); + Register(std::move(shader_info), *cpu_addr, size_in_bytes); } else { - null_kernel = shader; + null_kernel = std::move(shader_info); } } @@ -262,7 +271,7 @@ VKComputePipeline& VKPipelineCache::GetComputePipeline(const ComputePipelineCach return *entry; } -void VKPipelineCache::Unregister(const Shader& shader) { +void VKPipelineCache::OnShaderRemoval(Shader* shader) { bool finished = false; const auto Finish = [&] { // TODO(Rodrigo): Instead of finishing here, wait for the fences that use this pipeline and @@ -294,8 +303,6 @@ void VKPipelineCache::Unregister(const Shader& shader) { Finish(); it = compute_cache.erase(it); } - - RasterizerCache::Unregister(shader); } std::pair<SPIRVProgram, std::vector<VkDescriptorSetLayoutBinding>> @@ -312,7 +319,9 @@ VKPipelineCache::DecompileShaders(const GraphicsPipelineCacheKey& key) { ASSERT(point_size != 0.0f); } for (std::size_t i = 0; i < Maxwell::NumVertexAttributes; ++i) { - specialization.attribute_types[i] = fixed_state.vertex_input.attributes[i].Type(); + const auto& attribute = fixed_state.vertex_input.attributes[i]; + specialization.enabled_attributes[i] = attribute.enabled.Value() != 0; + specialization.attribute_types[i] = attribute.Type(); } specialization.ndc_minus_one_to_one = fixed_state.rasterizer.ndc_minus_one_to_one; @@ -328,13 +337,11 @@ VKPipelineCache::DecompileShaders(const GraphicsPipelineCacheKey& key) { } const GPUVAddr gpu_addr = GetShaderAddress(system, program_enum); - const auto cpu_addr = memory_manager.GpuToCpuAddress(gpu_addr); - ASSERT(cpu_addr); - const auto shader = TryGet(*cpu_addr); - ASSERT(shader); + const std::optional<VAddr> cpu_addr = memory_manager.GpuToCpuAddress(gpu_addr); + Shader* const shader = cpu_addr ? TryGet(*cpu_addr) : null_shader.get(); const std::size_t stage = index == 0 ? 0 : index - 1; // Stage indices are 0 - 5 - const auto program_type = GetShaderType(program_enum); + const ShaderType program_type = GetShaderType(program_enum); const auto& entries = shader->GetEntries(); program[stage] = { Decompile(device, shader->GetIR(), program_type, shader->GetRegistry(), specialization), @@ -376,16 +383,17 @@ void AddEntry(std::vector<VkDescriptorUpdateTemplateEntry>& template_entries, u3 return; } - if constexpr (descriptor_type == UNIFORM_TEXEL_BUFFER) { - // Nvidia has a bug where updating multiple uniform texels at once causes the driver to - // crash. + if constexpr (descriptor_type == UNIFORM_TEXEL_BUFFER || + descriptor_type == STORAGE_TEXEL_BUFFER) { + // Nvidia has a bug where updating multiple texels at once causes the driver to crash. + // Note: Fixed in driver Windows 443.24, Linux 440.66.15 for (u32 i = 0; i < count; ++i) { VkDescriptorUpdateTemplateEntry& entry = template_entries.emplace_back(); entry.dstBinding = binding + i; entry.dstArrayElement = 0; entry.descriptorCount = 1; entry.descriptorType = descriptor_type; - entry.offset = offset + i * entry_size; + entry.offset = static_cast<std::size_t>(offset + i * entry_size); entry.stride = entry_size; } } else if (count > 0) { @@ -406,8 +414,9 @@ void FillDescriptorUpdateTemplateEntries( std::vector<VkDescriptorUpdateTemplateEntryKHR>& template_entries) { AddEntry<UNIFORM_BUFFER>(template_entries, offset, binding, entries.const_buffers); AddEntry<STORAGE_BUFFER>(template_entries, offset, binding, entries.global_buffers); - AddEntry<UNIFORM_TEXEL_BUFFER>(template_entries, offset, binding, entries.texel_buffers); + AddEntry<UNIFORM_TEXEL_BUFFER>(template_entries, offset, binding, entries.uniform_texels); AddEntry<COMBINED_IMAGE_SAMPLER>(template_entries, offset, binding, entries.samplers); + AddEntry<STORAGE_TEXEL_BUFFER>(template_entries, offset, binding, entries.storage_texels); AddEntry<STORAGE_IMAGE>(template_entries, offset, binding, entries.images); } diff --git a/src/video_core/renderer_vulkan/vk_pipeline_cache.h b/src/video_core/renderer_vulkan/vk_pipeline_cache.h index 0b5796fef..0a36e5112 100644 --- a/src/video_core/renderer_vulkan/vk_pipeline_cache.h +++ b/src/video_core/renderer_vulkan/vk_pipeline_cache.h @@ -17,7 +17,6 @@ #include "common/common_types.h" #include "video_core/engines/const_buffer_engine_interface.h" #include "video_core/engines/maxwell_3d.h" -#include "video_core/rasterizer_cache.h" #include "video_core/renderer_vulkan/fixed_pipeline_state.h" #include "video_core/renderer_vulkan/vk_graphics_pipeline.h" #include "video_core/renderer_vulkan/vk_renderpass_cache.h" @@ -26,6 +25,7 @@ #include "video_core/shader/memory_util.h" #include "video_core/shader/registry.h" #include "video_core/shader/shader_ir.h" +#include "video_core/shader_cache.h" namespace Core { class System; @@ -41,8 +41,6 @@ class VKFence; class VKScheduler; class VKUpdateDescriptorQueue; -class CachedShader; -using Shader = std::shared_ptr<CachedShader>; using Maxwell = Tegra::Engines::Maxwell3D::Regs; struct GraphicsPipelineCacheKey { @@ -102,21 +100,16 @@ struct hash<Vulkan::ComputePipelineCacheKey> { namespace Vulkan { -class CachedShader final : public RasterizerCacheObject { +class Shader { public: - explicit CachedShader(Core::System& system, Tegra::Engines::ShaderType stage, GPUVAddr gpu_addr, - VAddr cpu_addr, VideoCommon::Shader::ProgramCode program_code, - u32 main_offset); - ~CachedShader(); + explicit Shader(Core::System& system, Tegra::Engines::ShaderType stage, GPUVAddr gpu_addr, + VideoCommon::Shader::ProgramCode program_code, u32 main_offset); + ~Shader(); GPUVAddr GetGpuAddr() const { return gpu_addr; } - std::size_t GetSizeInBytes() const override { - return program_code.size() * sizeof(u64); - } - VideoCommon::Shader::ShaderIR& GetIR() { return shader_ir; } @@ -144,25 +137,23 @@ private: ShaderEntries entries; }; -class VKPipelineCache final : public RasterizerCache<Shader> { +class VKPipelineCache final : public VideoCommon::ShaderCache<Shader> { public: explicit VKPipelineCache(Core::System& system, RasterizerVulkan& rasterizer, const VKDevice& device, VKScheduler& scheduler, VKDescriptorPool& descriptor_pool, VKUpdateDescriptorQueue& update_descriptor_queue, VKRenderPassCache& renderpass_cache); - ~VKPipelineCache(); + ~VKPipelineCache() override; - std::array<Shader, Maxwell::MaxShaderProgram> GetShaders(); + std::array<Shader*, Maxwell::MaxShaderProgram> GetShaders(); VKGraphicsPipeline& GetGraphicsPipeline(const GraphicsPipelineCacheKey& key); VKComputePipeline& GetComputePipeline(const ComputePipelineCacheKey& key); protected: - void Unregister(const Shader& shader) override; - - void FlushObjectInner(const Shader& object) override {} + void OnShaderRemoval(Shader* shader) final; private: std::pair<SPIRVProgram, std::vector<VkDescriptorSetLayoutBinding>> DecompileShaders( @@ -175,10 +166,10 @@ private: VKUpdateDescriptorQueue& update_descriptor_queue; VKRenderPassCache& renderpass_cache; - Shader null_shader{}; - Shader null_kernel{}; + std::unique_ptr<Shader> null_shader; + std::unique_ptr<Shader> null_kernel; - std::array<Shader, Maxwell::MaxShaderProgram> last_shaders; + std::array<Shader*, Maxwell::MaxShaderProgram> last_shaders{}; GraphicsPipelineCacheKey last_graphics_key; VKGraphicsPipeline* last_graphics_pipeline = nullptr; diff --git a/src/video_core/renderer_vulkan/vk_rasterizer.cpp b/src/video_core/renderer_vulkan/vk_rasterizer.cpp index 8b009fc22..a8d94eac3 100644 --- a/src/video_core/renderer_vulkan/vk_rasterizer.cpp +++ b/src/video_core/renderer_vulkan/vk_rasterizer.cpp @@ -38,6 +38,7 @@ #include "video_core/renderer_vulkan/vk_texture_cache.h" #include "video_core/renderer_vulkan/vk_update_descriptor.h" #include "video_core/renderer_vulkan/wrapper.h" +#include "video_core/shader_cache.h" namespace Vulkan { @@ -98,7 +99,7 @@ VkRect2D GetScissorState(const Maxwell& regs, std::size_t index) { } std::array<GPUVAddr, Maxwell::MaxShaderProgram> GetShaderAddresses( - const std::array<Shader, Maxwell::MaxShaderProgram>& shaders) { + const std::array<Shader*, Maxwell::MaxShaderProgram>& shaders) { std::array<GPUVAddr, Maxwell::MaxShaderProgram> addresses; for (std::size_t i = 0; i < std::size(addresses); ++i) { addresses[i] = shaders[i] ? shaders[i]->GetGpuAddr() : 0; @@ -117,6 +118,17 @@ template <typename Engine, typename Entry> Tegra::Texture::FullTextureInfo GetTextureInfo(const Engine& engine, const Entry& entry, std::size_t stage, std::size_t index = 0) { const auto stage_type = static_cast<Tegra::Engines::ShaderType>(stage); + if constexpr (std::is_same_v<Entry, SamplerEntry>) { + if (entry.is_separated) { + const u32 buffer_1 = entry.buffer; + const u32 buffer_2 = entry.secondary_buffer; + const u32 offset_1 = entry.offset; + const u32 offset_2 = entry.secondary_offset; + const u32 handle_1 = engine.AccessConstBuffer32(stage_type, buffer_1, offset_1); + const u32 handle_2 = engine.AccessConstBuffer32(stage_type, buffer_2, offset_2); + return engine.GetTextureInfo(handle_1 | handle_2); + } + } if (entry.is_bindless) { const auto tex_handle = engine.AccessConstBuffer32(stage_type, entry.buffer, entry.offset); return engine.GetTextureInfo(tex_handle); @@ -131,6 +143,49 @@ Tegra::Texture::FullTextureInfo GetTextureInfo(const Engine& engine, const Entry } } +/// @brief Determine if an attachment to be updated has to preserve contents +/// @param is_clear True when a clear is being executed +/// @param regs 3D registers +/// @return True when the contents have to be preserved +bool HasToPreserveColorContents(bool is_clear, const Maxwell& regs) { + if (!is_clear) { + return true; + } + // First we have to make sure all clear masks are enabled. + if (!regs.clear_buffers.R || !regs.clear_buffers.G || !regs.clear_buffers.B || + !regs.clear_buffers.A) { + return true; + } + // If scissors are disabled, the whole screen is cleared + if (!regs.clear_flags.scissor) { + return false; + } + // Then we have to confirm scissor testing clears the whole image + const std::size_t index = regs.clear_buffers.RT; + const auto& scissor = regs.scissor_test[0]; + return scissor.min_x > 0 || scissor.min_y > 0 || scissor.max_x < regs.rt[index].width || + scissor.max_y < regs.rt[index].height; +} + +/// @brief Determine if an attachment to be updated has to preserve contents +/// @param is_clear True when a clear is being executed +/// @param regs 3D registers +/// @return True when the contents have to be preserved +bool HasToPreserveDepthContents(bool is_clear, const Maxwell& regs) { + // If we are not clearing, the contents have to be preserved + if (!is_clear) { + return true; + } + // For depth stencil clears we only have to confirm scissor test covers the whole image + if (!regs.clear_flags.scissor) { + return false; + } + // Make sure the clear cover the whole image + const auto& scissor = regs.scissor_test[0]; + return scissor.min_x > 0 || scissor.min_y > 0 || scissor.max_x < regs.zeta_width || + scissor.max_y < regs.zeta_height; +} + } // Anonymous namespace class BufferBindings final { @@ -332,7 +387,7 @@ void RasterizerVulkan::Draw(bool is_indexed, bool is_instanced) { buffer_cache.Unmap(); - const Texceptions texceptions = UpdateAttachments(); + const Texceptions texceptions = UpdateAttachments(false); SetupImageTransitions(texceptions, color_attachments, zeta_attachment); key.renderpass_params = GetRenderPassParams(texceptions); @@ -388,7 +443,7 @@ void RasterizerVulkan::Clear() { return; } - [[maybe_unused]] const auto texceptions = UpdateAttachments(); + [[maybe_unused]] const auto texceptions = UpdateAttachments(true); DEBUG_ASSERT(texceptions.none()); SetupImageTransitions(0, color_attachments, zeta_attachment); @@ -468,8 +523,9 @@ void RasterizerVulkan::DispatchCompute(GPUVAddr code_addr) { const auto& entries = pipeline.GetEntries(); SetupComputeConstBuffers(entries); SetupComputeGlobalBuffers(entries); - SetupComputeTexelBuffers(entries); + SetupComputeUniformTexels(entries); SetupComputeTextures(entries); + SetupComputeStorageTexels(entries); SetupComputeImages(entries); buffer_cache.Unmap(); @@ -532,14 +588,14 @@ void RasterizerVulkan::OnCPUWrite(VAddr addr, u64 size) { return; } texture_cache.OnCPUWrite(addr, size); - pipeline_cache.InvalidateRegion(addr, size); + pipeline_cache.OnCPUWrite(addr, size); buffer_cache.OnCPUWrite(addr, size); - query_cache.InvalidateRegion(addr, size); } void RasterizerVulkan::SyncGuestHost() { texture_cache.SyncGuestHost(); buffer_cache.SyncGuestHost(); + pipeline_cache.SyncGuestHost(); } void RasterizerVulkan::SignalSemaphore(GPUVAddr addr, u32 value) { @@ -569,7 +625,9 @@ void RasterizerVulkan::ReleaseFences() { } void RasterizerVulkan::FlushAndInvalidateRegion(VAddr addr, u64 size) { - FlushRegion(addr, size); + if (Settings::IsGPULevelExtreme()) { + FlushRegion(addr, size); + } InvalidateRegion(addr, size); } @@ -662,9 +720,12 @@ void RasterizerVulkan::FlushWork() { draw_counter = 0; } -RasterizerVulkan::Texceptions RasterizerVulkan::UpdateAttachments() { +RasterizerVulkan::Texceptions RasterizerVulkan::UpdateAttachments(bool is_clear) { MICROPROFILE_SCOPE(Vulkan_RenderTargets); - auto& dirty = system.GPU().Maxwell3D().dirty.flags; + auto& maxwell3d = system.GPU().Maxwell3D(); + auto& dirty = maxwell3d.dirty.flags; + auto& regs = maxwell3d.regs; + const bool update_rendertargets = dirty[VideoCommon::Dirty::RenderTargets]; dirty[VideoCommon::Dirty::RenderTargets] = false; @@ -673,7 +734,8 @@ RasterizerVulkan::Texceptions RasterizerVulkan::UpdateAttachments() { Texceptions texceptions; for (std::size_t rt = 0; rt < Maxwell::NumRenderTargets; ++rt) { if (update_rendertargets) { - color_attachments[rt] = texture_cache.GetColorBufferSurface(rt, true); + const bool preserve_contents = HasToPreserveColorContents(is_clear, regs); + color_attachments[rt] = texture_cache.GetColorBufferSurface(rt, preserve_contents); } if (color_attachments[rt] && WalkAttachmentOverlaps(*color_attachments[rt])) { texceptions[rt] = true; @@ -681,7 +743,8 @@ RasterizerVulkan::Texceptions RasterizerVulkan::UpdateAttachments() { } if (update_rendertargets) { - zeta_attachment = texture_cache.GetDepthBufferSurface(true); + const bool preserve_contents = HasToPreserveDepthContents(is_clear, regs); + zeta_attachment = texture_cache.GetDepthBufferSurface(preserve_contents); } if (zeta_attachment && WalkAttachmentOverlaps(*zeta_attachment)) { texceptions[ZETA_TEXCEPTION_INDEX] = true; @@ -713,7 +776,7 @@ std::tuple<VkFramebuffer, VkExtent2D> RasterizerVulkan::ConfigureFramebuffers( if (!view) { return false; } - key.views.push_back(view->GetHandle()); + key.views.push_back(view->GetAttachment()); key.width = std::min(key.width, view->GetWidth()); key.height = std::min(key.height, view->GetHeight()); key.layers = std::min(key.layers, view->GetNumLayers()); @@ -773,20 +836,21 @@ RasterizerVulkan::DrawParameters RasterizerVulkan::SetupGeometry(FixedPipelineSt } void RasterizerVulkan::SetupShaderDescriptors( - const std::array<Shader, Maxwell::MaxShaderProgram>& shaders) { + const std::array<Shader*, Maxwell::MaxShaderProgram>& shaders) { texture_cache.GuardSamplers(true); for (std::size_t stage = 0; stage < Maxwell::MaxShaderStage; ++stage) { // Skip VertexA stage - const auto& shader = shaders[stage + 1]; + Shader* const shader = shaders[stage + 1]; if (!shader) { continue; } const auto& entries = shader->GetEntries(); SetupGraphicsConstBuffers(entries, stage); SetupGraphicsGlobalBuffers(entries, stage); - SetupGraphicsTexelBuffers(entries, stage); + SetupGraphicsUniformTexels(entries, stage); SetupGraphicsTextures(entries, stage); + SetupGraphicsStorageTexels(entries, stage); SetupGraphicsImages(entries, stage); } texture_cache.GuardSamplers(false); @@ -836,6 +900,10 @@ void RasterizerVulkan::BeginTransformFeedback() { if (regs.tfb_enabled == 0) { return; } + if (!device.IsExtTransformFeedbackSupported()) { + LOG_ERROR(Render_Vulkan, "Transform feedbacks used but not supported"); + return; + } UNIMPLEMENTED_IF(regs.IsShaderConfigEnabled(Maxwell::ShaderProgram::TesselationControl) || regs.IsShaderConfigEnabled(Maxwell::ShaderProgram::TesselationEval) || @@ -850,10 +918,10 @@ void RasterizerVulkan::BeginTransformFeedback() { UNIMPLEMENTED_IF(binding.buffer_offset != 0); const GPUVAddr gpu_addr = binding.Address(); - const std::size_t size = binding.buffer_size; - const auto [buffer, offset] = buffer_cache.UploadMemory(gpu_addr, size, 4, true); + const VkDeviceSize size = static_cast<VkDeviceSize>(binding.buffer_size); + const auto info = buffer_cache.UploadMemory(gpu_addr, size, 4, true); - scheduler.Record([buffer = buffer, offset = offset, size](vk::CommandBuffer cmdbuf) { + scheduler.Record([buffer = info.handle, offset = info.offset, size](vk::CommandBuffer cmdbuf) { cmdbuf.BindTransformFeedbackBuffersEXT(0, 1, &buffer, &offset, &size); cmdbuf.BeginTransformFeedbackEXT(0, 0, nullptr, nullptr); }); @@ -864,6 +932,9 @@ void RasterizerVulkan::EndTransformFeedback() { if (regs.tfb_enabled == 0) { return; } + if (!device.IsExtTransformFeedbackSupported()) { + return; + } scheduler.Record( [](vk::CommandBuffer cmdbuf) { cmdbuf.EndTransformFeedbackEXT(0, 0, nullptr, nullptr); }); @@ -875,14 +946,10 @@ void RasterizerVulkan::SetupVertexArrays(FixedPipelineState::VertexInput& vertex for (std::size_t index = 0; index < Maxwell::NumVertexAttributes; ++index) { const auto& attrib = regs.vertex_attrib_format[index]; - if (!attrib.IsValid()) { + if (attrib.IsConstant()) { vertex_input.SetAttribute(index, false, 0, 0, {}, {}); continue; } - - [[maybe_unused]] const auto& buffer = regs.vertex_array[attrib.buffer]; - ASSERT(buffer.IsEnabled()); - vertex_input.SetAttribute(index, true, attrib.buffer, attrib.offset, attrib.type.Value(), attrib.size.Value()); } @@ -906,8 +973,8 @@ void RasterizerVulkan::SetupVertexArrays(FixedPipelineState::VertexInput& vertex buffer_bindings.AddVertexBinding(DefaultBuffer(), 0); continue; } - const auto [buffer, offset] = buffer_cache.UploadMemory(start, size); - buffer_bindings.AddVertexBinding(buffer, offset); + const auto info = buffer_cache.UploadMemory(start, size); + buffer_bindings.AddVertexBinding(info.handle, info.offset); } } @@ -929,7 +996,9 @@ void RasterizerVulkan::SetupIndexBuffer(BufferBindings& buffer_bindings, DrawPar break; } const GPUVAddr gpu_addr = regs.index_array.IndexStart(); - auto [buffer, offset] = buffer_cache.UploadMemory(gpu_addr, CalculateIndexBufferSize()); + const auto info = buffer_cache.UploadMemory(gpu_addr, CalculateIndexBufferSize()); + VkBuffer buffer = info.handle; + u64 offset = info.offset; std::tie(buffer, offset) = quad_indexed_pass.Assemble( regs.index_array.format, params.num_vertices, params.base_vertex, buffer, offset); @@ -943,7 +1012,9 @@ void RasterizerVulkan::SetupIndexBuffer(BufferBindings& buffer_bindings, DrawPar break; } const GPUVAddr gpu_addr = regs.index_array.IndexStart(); - auto [buffer, offset] = buffer_cache.UploadMemory(gpu_addr, CalculateIndexBufferSize()); + const auto info = buffer_cache.UploadMemory(gpu_addr, CalculateIndexBufferSize()); + VkBuffer buffer = info.handle; + u64 offset = info.offset; auto format = regs.index_array.format; const bool is_uint8 = format == Maxwell::IndexFormat::UnsignedByte; @@ -978,12 +1049,12 @@ void RasterizerVulkan::SetupGraphicsGlobalBuffers(const ShaderEntries& entries, } } -void RasterizerVulkan::SetupGraphicsTexelBuffers(const ShaderEntries& entries, std::size_t stage) { +void RasterizerVulkan::SetupGraphicsUniformTexels(const ShaderEntries& entries, std::size_t stage) { MICROPROFILE_SCOPE(Vulkan_Textures); const auto& gpu = system.GPU().Maxwell3D(); - for (const auto& entry : entries.texel_buffers) { + for (const auto& entry : entries.uniform_texels) { const auto image = GetTextureInfo(gpu, entry, stage).tic; - SetupTexelBuffer(image, entry); + SetupUniformTexels(image, entry); } } @@ -998,6 +1069,15 @@ void RasterizerVulkan::SetupGraphicsTextures(const ShaderEntries& entries, std:: } } +void RasterizerVulkan::SetupGraphicsStorageTexels(const ShaderEntries& entries, std::size_t stage) { + MICROPROFILE_SCOPE(Vulkan_Textures); + const auto& gpu = system.GPU().Maxwell3D(); + for (const auto& entry : entries.storage_texels) { + const auto image = GetTextureInfo(gpu, entry, stage).tic; + SetupStorageTexel(image, entry); + } +} + void RasterizerVulkan::SetupGraphicsImages(const ShaderEntries& entries, std::size_t stage) { MICROPROFILE_SCOPE(Vulkan_Images); const auto& gpu = system.GPU().Maxwell3D(); @@ -1030,12 +1110,12 @@ void RasterizerVulkan::SetupComputeGlobalBuffers(const ShaderEntries& entries) { } } -void RasterizerVulkan::SetupComputeTexelBuffers(const ShaderEntries& entries) { +void RasterizerVulkan::SetupComputeUniformTexels(const ShaderEntries& entries) { MICROPROFILE_SCOPE(Vulkan_Textures); const auto& gpu = system.GPU().KeplerCompute(); - for (const auto& entry : entries.texel_buffers) { + for (const auto& entry : entries.uniform_texels) { const auto image = GetTextureInfo(gpu, entry, ComputeShaderIndex).tic; - SetupTexelBuffer(image, entry); + SetupUniformTexels(image, entry); } } @@ -1050,6 +1130,15 @@ void RasterizerVulkan::SetupComputeTextures(const ShaderEntries& entries) { } } +void RasterizerVulkan::SetupComputeStorageTexels(const ShaderEntries& entries) { + MICROPROFILE_SCOPE(Vulkan_Textures); + const auto& gpu = system.GPU().KeplerCompute(); + for (const auto& entry : entries.storage_texels) { + const auto image = GetTextureInfo(gpu, entry, ComputeShaderIndex).tic; + SetupStorageTexel(image, entry); + } +} + void RasterizerVulkan::SetupComputeImages(const ShaderEntries& entries) { MICROPROFILE_SCOPE(Vulkan_Images); const auto& gpu = system.GPU().KeplerCompute(); @@ -1072,10 +1161,9 @@ void RasterizerVulkan::SetupConstBuffer(const ConstBufferEntry& entry, Common::AlignUp(CalculateConstBufferSize(entry, buffer), 4 * sizeof(float)); ASSERT(size <= MaxConstbufferSize); - const auto [buffer_handle, offset] = + const auto info = buffer_cache.UploadMemory(buffer.address, size, device.GetUniformBufferAlignment()); - - update_descriptor_queue.AddBuffer(buffer_handle, offset, size); + update_descriptor_queue.AddBuffer(info.handle, info.offset, size); } void RasterizerVulkan::SetupGlobalBuffer(const GlobalBufferEntry& entry, GPUVAddr address) { @@ -1089,18 +1177,18 @@ void RasterizerVulkan::SetupGlobalBuffer(const GlobalBufferEntry& entry, GPUVAdd // Note: Do *not* use DefaultBuffer() here, storage buffers can be written breaking the // default buffer. static constexpr std::size_t dummy_size = 4; - const auto buffer = buffer_cache.GetEmptyBuffer(dummy_size); - update_descriptor_queue.AddBuffer(buffer, 0, dummy_size); + const auto info = buffer_cache.GetEmptyBuffer(dummy_size); + update_descriptor_queue.AddBuffer(info.handle, info.offset, dummy_size); return; } - const auto [buffer, offset] = buffer_cache.UploadMemory( + const auto info = buffer_cache.UploadMemory( actual_addr, size, device.GetStorageBufferAlignment(), entry.IsWritten()); - update_descriptor_queue.AddBuffer(buffer, offset, size); + update_descriptor_queue.AddBuffer(info.handle, info.offset, size); } -void RasterizerVulkan::SetupTexelBuffer(const Tegra::Texture::TICEntry& tic, - const TexelBufferEntry& entry) { +void RasterizerVulkan::SetupUniformTexels(const Tegra::Texture::TICEntry& tic, + const UniformTexelEntry& entry) { const auto view = texture_cache.GetTextureSurface(tic, entry); ASSERT(view->IsBufferView()); @@ -1112,16 +1200,24 @@ void RasterizerVulkan::SetupTexture(const Tegra::Texture::FullTextureInfo& textu auto view = texture_cache.GetTextureSurface(texture.tic, entry); ASSERT(!view->IsBufferView()); - const auto image_view = view->GetHandle(texture.tic.x_source, texture.tic.y_source, - texture.tic.z_source, texture.tic.w_source); + const VkImageView image_view = view->GetImageView(texture.tic.x_source, texture.tic.y_source, + texture.tic.z_source, texture.tic.w_source); const auto sampler = sampler_cache.GetSampler(texture.tsc); update_descriptor_queue.AddSampledImage(sampler, image_view); - const auto image_layout = update_descriptor_queue.GetLastImageLayout(); + VkImageLayout* const image_layout = update_descriptor_queue.LastImageLayout(); *image_layout = VK_IMAGE_LAYOUT_SHADER_READ_ONLY_OPTIMAL; sampled_views.push_back(ImageView{std::move(view), image_layout}); } +void RasterizerVulkan::SetupStorageTexel(const Tegra::Texture::TICEntry& tic, + const StorageTexelEntry& entry) { + const auto view = texture_cache.GetImageSurface(tic, entry); + ASSERT(view->IsBufferView()); + + update_descriptor_queue.AddTexelBuffer(view->GetBufferView()); +} + void RasterizerVulkan::SetupImage(const Tegra::Texture::TICEntry& tic, const ImageEntry& entry) { auto view = texture_cache.GetImageSurface(tic, entry); @@ -1131,10 +1227,11 @@ void RasterizerVulkan::SetupImage(const Tegra::Texture::TICEntry& tic, const Ima UNIMPLEMENTED_IF(tic.IsBuffer()); - const auto image_view = view->GetHandle(tic.x_source, tic.y_source, tic.z_source, tic.w_source); + const VkImageView image_view = + view->GetImageView(tic.x_source, tic.y_source, tic.z_source, tic.w_source); update_descriptor_queue.AddImage(image_view); - const auto image_layout = update_descriptor_queue.GetLastImageLayout(); + VkImageLayout* const image_layout = update_descriptor_queue.LastImageLayout(); *image_layout = VK_IMAGE_LAYOUT_GENERAL; image_views.push_back(ImageView{std::move(view), image_layout}); } diff --git a/src/video_core/renderer_vulkan/vk_rasterizer.h b/src/video_core/renderer_vulkan/vk_rasterizer.h index 0ed0e48c6..83e00e7e9 100644 --- a/src/video_core/renderer_vulkan/vk_rasterizer.h +++ b/src/video_core/renderer_vulkan/vk_rasterizer.h @@ -159,7 +159,10 @@ private: void FlushWork(); - Texceptions UpdateAttachments(); + /// @brief Updates the currently bound attachments + /// @param is_clear True when the framebuffer is updated as a clear + /// @return Bitfield of attachments being used as sampled textures + Texceptions UpdateAttachments(bool is_clear); std::tuple<VkFramebuffer, VkExtent2D> ConfigureFramebuffers(VkRenderPass renderpass); @@ -168,7 +171,7 @@ private: bool is_indexed, bool is_instanced); /// Setup descriptors in the graphics pipeline. - void SetupShaderDescriptors(const std::array<Shader, Maxwell::MaxShaderProgram>& shaders); + void SetupShaderDescriptors(const std::array<Shader*, Maxwell::MaxShaderProgram>& shaders); void SetupImageTransitions(Texceptions texceptions, const std::array<View, Maxwell::NumRenderTargets>& color_attachments, @@ -193,12 +196,15 @@ private: /// Setup global buffers in the graphics pipeline. void SetupGraphicsGlobalBuffers(const ShaderEntries& entries, std::size_t stage); - /// Setup texel buffers in the graphics pipeline. - void SetupGraphicsTexelBuffers(const ShaderEntries& entries, std::size_t stage); + /// Setup uniform texels in the graphics pipeline. + void SetupGraphicsUniformTexels(const ShaderEntries& entries, std::size_t stage); /// Setup textures in the graphics pipeline. void SetupGraphicsTextures(const ShaderEntries& entries, std::size_t stage); + /// Setup storage texels in the graphics pipeline. + void SetupGraphicsStorageTexels(const ShaderEntries& entries, std::size_t stage); + /// Setup images in the graphics pipeline. void SetupGraphicsImages(const ShaderEntries& entries, std::size_t stage); @@ -209,11 +215,14 @@ private: void SetupComputeGlobalBuffers(const ShaderEntries& entries); /// Setup texel buffers in the compute pipeline. - void SetupComputeTexelBuffers(const ShaderEntries& entries); + void SetupComputeUniformTexels(const ShaderEntries& entries); /// Setup textures in the compute pipeline. void SetupComputeTextures(const ShaderEntries& entries); + /// Setup storage texels in the compute pipeline. + void SetupComputeStorageTexels(const ShaderEntries& entries); + /// Setup images in the compute pipeline. void SetupComputeImages(const ShaderEntries& entries); @@ -222,10 +231,12 @@ private: void SetupGlobalBuffer(const GlobalBufferEntry& entry, GPUVAddr address); - void SetupTexelBuffer(const Tegra::Texture::TICEntry& image, const TexelBufferEntry& entry); + void SetupUniformTexels(const Tegra::Texture::TICEntry& image, const UniformTexelEntry& entry); void SetupTexture(const Tegra::Texture::FullTextureInfo& texture, const SamplerEntry& entry); + void SetupStorageTexel(const Tegra::Texture::TICEntry& tic, const StorageTexelEntry& entry); + void SetupImage(const Tegra::Texture::TICEntry& tic, const ImageEntry& entry); void UpdateViewportsState(Tegra::Engines::Maxwell3D::Regs& regs); diff --git a/src/video_core/renderer_vulkan/vk_sampler_cache.cpp b/src/video_core/renderer_vulkan/vk_sampler_cache.cpp index 2687d8d95..616eacc36 100644 --- a/src/video_core/renderer_vulkan/vk_sampler_cache.cpp +++ b/src/video_core/renderer_vulkan/vk_sampler_cache.cpp @@ -9,6 +9,8 @@ #include "video_core/renderer_vulkan/wrapper.h" #include "video_core/textures/texture.h" +using Tegra::Texture::TextureMipmapFilter; + namespace Vulkan { namespace { @@ -39,9 +41,18 @@ VKSamplerCache::VKSamplerCache(const VKDevice& device) : device{device} {} VKSamplerCache::~VKSamplerCache() = default; vk::Sampler VKSamplerCache::CreateSampler(const Tegra::Texture::TSCEntry& tsc) const { + const bool arbitrary_borders = device.IsExtCustomBorderColorSupported(); + const std::array color = tsc.GetBorderColor(); + + VkSamplerCustomBorderColorCreateInfoEXT border; + border.sType = VK_STRUCTURE_TYPE_SAMPLER_CUSTOM_BORDER_COLOR_CREATE_INFO_EXT; + border.pNext = nullptr; + border.format = VK_FORMAT_UNDEFINED; + std::memcpy(&border.customBorderColor, color.data(), sizeof(color)); + VkSamplerCreateInfo ci; ci.sType = VK_STRUCTURE_TYPE_SAMPLER_CREATE_INFO; - ci.pNext = nullptr; + ci.pNext = arbitrary_borders ? &border : nullptr; ci.flags = 0; ci.magFilter = MaxwellToVK::Sampler::Filter(tsc.mag_filter); ci.minFilter = MaxwellToVK::Sampler::Filter(tsc.min_filter); @@ -54,9 +65,9 @@ vk::Sampler VKSamplerCache::CreateSampler(const Tegra::Texture::TSCEntry& tsc) c ci.maxAnisotropy = tsc.GetMaxAnisotropy(); ci.compareEnable = tsc.depth_compare_enabled; ci.compareOp = MaxwellToVK::Sampler::DepthCompareFunction(tsc.depth_compare_func); - ci.minLod = tsc.GetMinLod(); - ci.maxLod = tsc.GetMaxLod(); - ci.borderColor = ConvertBorderColor(tsc.GetBorderColor()); + ci.minLod = tsc.mipmap_filter == TextureMipmapFilter::None ? 0.0f : tsc.GetMinLod(); + ci.maxLod = tsc.mipmap_filter == TextureMipmapFilter::None ? 0.25f : tsc.GetMaxLod(); + ci.borderColor = arbitrary_borders ? VK_BORDER_COLOR_INT_CUSTOM_EXT : ConvertBorderColor(color); ci.unnormalizedCoordinates = VK_FALSE; return device.GetLogical().CreateSampler(ci); } diff --git a/src/video_core/renderer_vulkan/vk_scheduler.cpp b/src/video_core/renderer_vulkan/vk_scheduler.cpp index 82ec9180e..56524e6f3 100644 --- a/src/video_core/renderer_vulkan/vk_scheduler.cpp +++ b/src/video_core/renderer_vulkan/vk_scheduler.cpp @@ -9,6 +9,7 @@ #include <utility> #include "common/microprofile.h" +#include "common/thread.h" #include "video_core/renderer_vulkan/vk_device.h" #include "video_core/renderer_vulkan/vk_query_cache.h" #include "video_core/renderer_vulkan/vk_resource_manager.h" @@ -133,6 +134,7 @@ void VKScheduler::BindGraphicsPipeline(VkPipeline pipeline) { } void VKScheduler::WorkerThread() { + Common::SetCurrentThreadPriority(Common::ThreadPriority::High); std::unique_lock lock{mutex}; do { cv.wait(lock, [this] { return !chunk_queue.Empty() || quit; }); diff --git a/src/video_core/renderer_vulkan/vk_shader_decompiler.cpp b/src/video_core/renderer_vulkan/vk_shader_decompiler.cpp index 18678968c..97429cc59 100644 --- a/src/video_core/renderer_vulkan/vk_shader_decompiler.cpp +++ b/src/video_core/renderer_vulkan/vk_shader_decompiler.cpp @@ -400,8 +400,9 @@ private: u32 binding = specialization.base_binding; binding = DeclareConstantBuffers(binding); binding = DeclareGlobalBuffers(binding); - binding = DeclareTexelBuffers(binding); + binding = DeclareUniformTexels(binding); binding = DeclareSamplers(binding); + binding = DeclareStorageTexels(binding); binding = DeclareImages(binding); const Id main = OpFunction(t_void, {}, TypeFunction(t_void)); @@ -515,6 +516,16 @@ private: void DeclareCommon() { thread_id = DeclareInputBuiltIn(spv::BuiltIn::SubgroupLocalInvocationId, t_in_uint, "thread_id"); + thread_masks[0] = + DeclareInputBuiltIn(spv::BuiltIn::SubgroupEqMask, t_in_uint4, "thread_eq_mask"); + thread_masks[1] = + DeclareInputBuiltIn(spv::BuiltIn::SubgroupGeMask, t_in_uint4, "thread_ge_mask"); + thread_masks[2] = + DeclareInputBuiltIn(spv::BuiltIn::SubgroupGtMask, t_in_uint4, "thread_gt_mask"); + thread_masks[3] = + DeclareInputBuiltIn(spv::BuiltIn::SubgroupLeMask, t_in_uint4, "thread_le_mask"); + thread_masks[4] = + DeclareInputBuiltIn(spv::BuiltIn::SubgroupLtMask, t_in_uint4, "thread_lt_mask"); } void DeclareVertex() { @@ -731,8 +742,10 @@ private: if (!IsGenericAttribute(index)) { continue; } - const u32 location = GetGenericAttributeLocation(index); + if (!IsAttributeEnabled(location)) { + continue; + } const auto type_descriptor = GetAttributeType(location); Id type; if (IsInputAttributeArray()) { @@ -877,7 +890,7 @@ private: return binding; } - u32 DeclareTexelBuffers(u32 binding) { + u32 DeclareUniformTexels(u32 binding) { for (const auto& sampler : ir.GetSamplers()) { if (!sampler.is_buffer) { continue; @@ -898,7 +911,7 @@ private: Decorate(id, spv::Decoration::Binding, binding++); Decorate(id, spv::Decoration::DescriptorSet, DESCRIPTOR_SET); - texel_buffers.emplace(sampler.index, TexelBuffer{image_type, id}); + uniform_texels.emplace(sampler.index, TexelBuffer{image_type, id}); } return binding; } @@ -933,31 +946,48 @@ private: return binding; } - u32 DeclareImages(u32 binding) { + u32 DeclareStorageTexels(u32 binding) { for (const auto& image : ir.GetImages()) { - const auto [dim, arrayed] = GetImageDim(image); - constexpr int depth = 0; - constexpr bool ms = false; - constexpr int sampled = 2; // This won't be accessed with a sampler - constexpr auto format = spv::ImageFormat::Unknown; - const Id image_type = TypeImage(t_uint, dim, depth, arrayed, ms, sampled, format, {}); - const Id pointer_type = TypePointer(spv::StorageClass::UniformConstant, image_type); - const Id id = OpVariable(pointer_type, spv::StorageClass::UniformConstant); - AddGlobalVariable(Name(id, fmt::format("image_{}", image.index))); - - Decorate(id, spv::Decoration::Binding, binding++); - Decorate(id, spv::Decoration::DescriptorSet, DESCRIPTOR_SET); - if (image.is_read && !image.is_written) { - Decorate(id, spv::Decoration::NonWritable); - } else if (image.is_written && !image.is_read) { - Decorate(id, spv::Decoration::NonReadable); + if (image.type != Tegra::Shader::ImageType::TextureBuffer) { + continue; } + DeclareImage(image, binding); + } + return binding; + } - images.emplace(image.index, StorageImage{image_type, id}); + u32 DeclareImages(u32 binding) { + for (const auto& image : ir.GetImages()) { + if (image.type == Tegra::Shader::ImageType::TextureBuffer) { + continue; + } + DeclareImage(image, binding); } return binding; } + void DeclareImage(const Image& image, u32& binding) { + const auto [dim, arrayed] = GetImageDim(image); + constexpr int depth = 0; + constexpr bool ms = false; + constexpr int sampled = 2; // This won't be accessed with a sampler + const auto format = image.is_atomic ? spv::ImageFormat::R32ui : spv::ImageFormat::Unknown; + const Id image_type = TypeImage(t_uint, dim, depth, arrayed, ms, sampled, format, {}); + const Id pointer_type = TypePointer(spv::StorageClass::UniformConstant, image_type); + const Id id = OpVariable(pointer_type, spv::StorageClass::UniformConstant); + AddGlobalVariable(Name(id, fmt::format("image_{}", image.index))); + + Decorate(id, spv::Decoration::Binding, binding++); + Decorate(id, spv::Decoration::DescriptorSet, DESCRIPTOR_SET); + if (image.is_read && !image.is_written) { + Decorate(id, spv::Decoration::NonWritable); + } else if (image.is_written && !image.is_read) { + Decorate(id, spv::Decoration::NonReadable); + } + + images.emplace(image.index, StorageImage{image_type, id}); + } + bool IsRenderTargetEnabled(u32 rt) const { for (u32 component = 0; component < 4; ++component) { if (header.ps.IsColorComponentOutputEnabled(rt, component)) { @@ -976,6 +1006,10 @@ private: return stage == ShaderType::TesselationControl; } + bool IsAttributeEnabled(u32 location) const { + return stage != ShaderType::Vertex || specialization.enabled_attributes[location]; + } + u32 GetNumInputVertices() const { switch (stage) { case ShaderType::Geometry: @@ -1071,8 +1105,7 @@ private: void VisitBasicBlock(const NodeBlock& bb) { for (const auto& node : bb) { - [[maybe_unused]] const Type type = Visit(node).type; - ASSERT(type == Type::Void); + Visit(node); } } @@ -1192,16 +1225,20 @@ private: UNIMPLEMENTED_MSG("Unmanaged FrontFacing element={}", element); return {v_float_zero, Type::Float}; default: - if (IsGenericAttribute(attribute)) { - const u32 location = GetGenericAttributeLocation(attribute); - const auto type_descriptor = GetAttributeType(location); - const Type type = type_descriptor.type; - const Id attribute_id = input_attributes.at(attribute); - const std::vector elements = {element}; - const Id pointer = ArrayPass(type_descriptor.scalar, attribute_id, elements); - return {OpLoad(GetTypeDefinition(type), pointer), type}; + if (!IsGenericAttribute(attribute)) { + break; } - break; + const u32 location = GetGenericAttributeLocation(attribute); + if (!IsAttributeEnabled(location)) { + // Disabled attributes (also known as constant attributes) always return zero. + return {v_float_zero, Type::Float}; + } + const auto type_descriptor = GetAttributeType(location); + const Type type = type_descriptor.type; + const Id attribute_id = input_attributes.at(attribute); + const std::vector elements = {element}; + const Id pointer = ArrayPass(type_descriptor.scalar, attribute_id, elements); + return {OpLoad(GetTypeDefinition(type), pointer), type}; } UNIMPLEMENTED_MSG("Unhandled input attribute: {}", static_cast<u32>(attribute)); return {v_float_zero, Type::Float}; @@ -1237,7 +1274,7 @@ private: } else { UNREACHABLE_MSG("Unmanaged offset node type"); } - pointer = OpAccessChain(t_cbuf_float, buffer_id, Constant(t_uint, 0), buffer_index, + pointer = OpAccessChain(t_cbuf_float, buffer_id, v_uint_zero, buffer_index, buffer_element); } return {OpLoad(t_float, pointer), Type::Float}; @@ -1362,7 +1399,9 @@ private: Expression target{}; if (const auto gpr = std::get_if<GprNode>(&*dest)) { if (gpr->GetIndex() == Register::ZeroIndex) { - // Writing to Register::ZeroIndex is a no op + // Writing to Register::ZeroIndex is a no op but we still have to visit its source + // because it might have side effects. + Visit(src); return {}; } target = {registers.at(gpr->GetIndex()), Type::Float}; @@ -1590,7 +1629,7 @@ private: const Id result = OpIAddCarry(TypeStruct({t_uint, t_uint}), op_a, op_b); const Id carry = OpCompositeExtract(t_uint, result, 1); - return {OpINotEqual(t_bool, carry, Constant(t_uint, 0)), Type::Bool}; + return {OpINotEqual(t_bool, carry, v_uint_zero), Type::Bool}; } Expression LogicalAssign(Operation operation) { @@ -1618,6 +1657,24 @@ private: return {}; } + Expression LogicalFOrdered(Operation operation) { + // Emulate SPIR-V's OpOrdered + const Id op_a = AsFloat(Visit(operation[0])); + const Id op_b = AsFloat(Visit(operation[1])); + const Id is_num_a = OpFOrdEqual(t_bool, op_a, op_a); + const Id is_num_b = OpFOrdEqual(t_bool, op_b, op_b); + return {OpLogicalAnd(t_bool, is_num_a, is_num_b), Type::Bool}; + } + + Expression LogicalFUnordered(Operation operation) { + // Emulate SPIR-V's OpUnordered + const Id op_a = AsFloat(Visit(operation[0])); + const Id op_b = AsFloat(Visit(operation[1])); + const Id is_nan_a = OpIsNan(t_bool, op_a); + const Id is_nan_b = OpIsNan(t_bool, op_b); + return {OpLogicalOr(t_bool, is_nan_a, is_nan_b), Type::Bool}; + } + Id GetTextureSampler(Operation operation) { const auto& meta = std::get<MetaTexture>(operation.GetMeta()); ASSERT(!meta.sampler.is_buffer); @@ -1635,7 +1692,7 @@ private: const auto& meta = std::get<MetaTexture>(operation.GetMeta()); const u32 index = meta.sampler.index; if (meta.sampler.is_buffer) { - const auto& entry = texel_buffers.at(index); + const auto& entry = uniform_texels.at(index); return OpLoad(entry.image_type, entry.image); } else { const auto& entry = sampled_images.at(index); @@ -1912,39 +1969,20 @@ private: return {}; } - Expression AtomicImageAdd(Operation operation) { - UNIMPLEMENTED(); - return {}; - } - - Expression AtomicImageMin(Operation operation) { - UNIMPLEMENTED(); - return {}; - } - - Expression AtomicImageMax(Operation operation) { - UNIMPLEMENTED(); - return {}; - } - - Expression AtomicImageAnd(Operation operation) { - UNIMPLEMENTED(); - return {}; - } - - Expression AtomicImageOr(Operation operation) { - UNIMPLEMENTED(); - return {}; - } + template <Id (Module::*func)(Id, Id, Id, Id, Id)> + Expression AtomicImage(Operation operation) { + const auto& meta{std::get<MetaImage>(operation.GetMeta())}; + ASSERT(meta.values.size() == 1); - Expression AtomicImageXor(Operation operation) { - UNIMPLEMENTED(); - return {}; - } + const Id coordinate = GetCoordinates(operation, Type::Int); + const Id image = images.at(meta.image.index).image; + const Id sample = v_uint_zero; + const Id pointer = OpImageTexelPointer(t_image_uint, image, coordinate, sample); - Expression AtomicImageExchange(Operation operation) { - UNIMPLEMENTED(); - return {}; + const Id scope = Constant(t_uint, static_cast<u32>(spv::Scope::Device)); + const Id semantics = v_uint_zero; + const Id value = AsUint(Visit(meta.values[0])); + return {(this->*func)(t_uint, pointer, scope, semantics, value), Type::Uint}; } template <Id (Module::*func)(Id, Id, Id, Id, Id)> @@ -1959,7 +1997,7 @@ private: return {v_float_zero, Type::Float}; } const Id scope = Constant(t_uint, static_cast<u32>(spv::Scope::Device)); - const Id semantics = Constant(t_uint, 0); + const Id semantics = v_uint_zero; const Id value = AsUint(Visit(operation[1])); return {(this->*func)(t_uint, pointer, scope, semantics, value), Type::Uint}; @@ -2157,14 +2195,37 @@ private: return {OpLoad(t_uint, thread_id), Type::Uint}; } + template <std::size_t index> + Expression ThreadMask(Operation) { + // TODO(Rodrigo): Handle devices with different warp sizes + const Id mask = thread_masks[index]; + return {OpLoad(t_uint, AccessElement(t_in_uint, mask, 0)), Type::Uint}; + } + Expression ShuffleIndexed(Operation operation) { const Id value = AsFloat(Visit(operation[0])); const Id index = AsUint(Visit(operation[1])); return {OpSubgroupReadInvocationKHR(t_float, value, index), Type::Float}; } - Expression MemoryBarrierGL(Operation) { - const auto scope = spv::Scope::Device; + Expression Barrier(Operation) { + if (!ir.IsDecompiled()) { + LOG_ERROR(Render_Vulkan, "OpBarrier used by shader is not decompiled"); + return {}; + } + + const auto scope = spv::Scope::Workgroup; + const auto memory = spv::Scope::Workgroup; + const auto semantics = + spv::MemorySemanticsMask::WorkgroupMemory | spv::MemorySemanticsMask::AcquireRelease; + OpControlBarrier(Constant(t_uint, static_cast<u32>(scope)), + Constant(t_uint, static_cast<u32>(memory)), + Constant(t_uint, static_cast<u32>(semantics))); + return {}; + } + + template <spv::Scope scope> + Expression MemoryBarrier(Operation) { const auto semantics = spv::MemorySemanticsMask::AcquireRelease | spv::MemorySemanticsMask::UniformMemory | spv::MemorySemanticsMask::WorkgroupMemory | @@ -2511,7 +2572,14 @@ private: &SPIRVDecompiler::Binary<&Module::OpFOrdGreaterThan, Type::Bool, Type::Float>, &SPIRVDecompiler::Binary<&Module::OpFOrdNotEqual, Type::Bool, Type::Float>, &SPIRVDecompiler::Binary<&Module::OpFOrdGreaterThanEqual, Type::Bool, Type::Float>, - &SPIRVDecompiler::Unary<&Module::OpIsNan, Type::Bool, Type::Float>, + &SPIRVDecompiler::LogicalFOrdered, + &SPIRVDecompiler::LogicalFUnordered, + &SPIRVDecompiler::Binary<&Module::OpFUnordLessThan, Type::Bool, Type::Float>, + &SPIRVDecompiler::Binary<&Module::OpFUnordEqual, Type::Bool, Type::Float>, + &SPIRVDecompiler::Binary<&Module::OpFUnordLessThanEqual, Type::Bool, Type::Float>, + &SPIRVDecompiler::Binary<&Module::OpFUnordGreaterThan, Type::Bool, Type::Float>, + &SPIRVDecompiler::Binary<&Module::OpFUnordNotEqual, Type::Bool, Type::Float>, + &SPIRVDecompiler::Binary<&Module::OpFUnordGreaterThanEqual, Type::Bool, Type::Float>, &SPIRVDecompiler::Binary<&Module::OpSLessThan, Type::Bool, Type::Int>, &SPIRVDecompiler::Binary<&Module::OpIEqual, Type::Bool, Type::Int>, @@ -2553,11 +2621,11 @@ private: &SPIRVDecompiler::ImageLoad, &SPIRVDecompiler::ImageStore, - &SPIRVDecompiler::AtomicImageAdd, - &SPIRVDecompiler::AtomicImageAnd, - &SPIRVDecompiler::AtomicImageOr, - &SPIRVDecompiler::AtomicImageXor, - &SPIRVDecompiler::AtomicImageExchange, + &SPIRVDecompiler::AtomicImage<&Module::OpAtomicIAdd>, + &SPIRVDecompiler::AtomicImage<&Module::OpAtomicAnd>, + &SPIRVDecompiler::AtomicImage<&Module::OpAtomicOr>, + &SPIRVDecompiler::AtomicImage<&Module::OpAtomicXor>, + &SPIRVDecompiler::AtomicImage<&Module::OpAtomicExchange>, &SPIRVDecompiler::Atomic<&Module::OpAtomicExchange>, &SPIRVDecompiler::Atomic<&Module::OpAtomicIAdd>, @@ -2614,9 +2682,16 @@ private: &SPIRVDecompiler::Vote<&Module::OpSubgroupAllEqualKHR>, &SPIRVDecompiler::ThreadId, + &SPIRVDecompiler::ThreadMask<0>, // Eq + &SPIRVDecompiler::ThreadMask<1>, // Ge + &SPIRVDecompiler::ThreadMask<2>, // Gt + &SPIRVDecompiler::ThreadMask<3>, // Le + &SPIRVDecompiler::ThreadMask<4>, // Lt &SPIRVDecompiler::ShuffleIndexed, - &SPIRVDecompiler::MemoryBarrierGL, + &SPIRVDecompiler::Barrier, + &SPIRVDecompiler::MemoryBarrier<spv::Scope::Workgroup>, + &SPIRVDecompiler::MemoryBarrier<spv::Scope::Device>, }; static_assert(operation_decompilers.size() == static_cast<std::size_t>(OperationCode::Amount)); @@ -2692,8 +2767,11 @@ private: Decorate(TypeStruct(t_gmem_array), spv::Decoration::Block), 0, spv::Decoration::Offset, 0); const Id t_gmem_ssbo = TypePointer(spv::StorageClass::StorageBuffer, t_gmem_struct); + const Id t_image_uint = TypePointer(spv::StorageClass::Image, t_uint); + const Id v_float_zero = Constant(t_float, 0.0f); const Id v_float_one = Constant(t_float, 1.0f); + const Id v_uint_zero = Constant(t_uint, 0); // Nvidia uses these defaults for varyings (e.g. position and generic attributes) const Id v_varying_default = @@ -2718,15 +2796,16 @@ private: std::unordered_map<u8, GenericVaryingDescription> output_attributes; std::map<u32, Id> constant_buffers; std::map<GlobalMemoryBase, Id> global_buffers; - std::map<u32, TexelBuffer> texel_buffers; + std::map<u32, TexelBuffer> uniform_texels; std::map<u32, SampledImage> sampled_images; + std::map<u32, TexelBuffer> storage_texels; std::map<u32, StorageImage> images; + std::array<Id, Maxwell::NumRenderTargets> frag_colors{}; Id instance_index{}; Id vertex_index{}; Id base_instance{}; Id base_vertex{}; - std::array<Id, Maxwell::NumRenderTargets> frag_colors{}; Id frag_depth{}; Id frag_coord{}; Id front_facing{}; @@ -2738,6 +2817,7 @@ private: Id workgroup_id{}; Id local_invocation_id{}; Id thread_id{}; + std::array<Id, 5> thread_masks{}; // eq, ge, gt, le, lt VertexIndices in_indices; VertexIndices out_indices; @@ -2981,13 +3061,17 @@ ShaderEntries GenerateShaderEntries(const VideoCommon::Shader::ShaderIR& ir) { } for (const auto& sampler : ir.GetSamplers()) { if (sampler.is_buffer) { - entries.texel_buffers.emplace_back(sampler); + entries.uniform_texels.emplace_back(sampler); } else { entries.samplers.emplace_back(sampler); } } for (const auto& image : ir.GetImages()) { - entries.images.emplace_back(image); + if (image.type == Tegra::Shader::ImageType::TextureBuffer) { + entries.storage_texels.emplace_back(image); + } else { + entries.images.emplace_back(image); + } } for (const auto& attribute : ir.GetInputAttributes()) { if (IsGenericAttribute(attribute)) { diff --git a/src/video_core/renderer_vulkan/vk_shader_decompiler.h b/src/video_core/renderer_vulkan/vk_shader_decompiler.h index f4c05ac3c..2b0e90396 100644 --- a/src/video_core/renderer_vulkan/vk_shader_decompiler.h +++ b/src/video_core/renderer_vulkan/vk_shader_decompiler.h @@ -21,8 +21,9 @@ class VKDevice; namespace Vulkan { using Maxwell = Tegra::Engines::Maxwell3D::Regs; -using TexelBufferEntry = VideoCommon::Shader::Sampler; +using UniformTexelEntry = VideoCommon::Shader::Sampler; using SamplerEntry = VideoCommon::Shader::Sampler; +using StorageTexelEntry = VideoCommon::Shader::Image; using ImageEntry = VideoCommon::Shader::Image; constexpr u32 DESCRIPTOR_SET = 0; @@ -66,13 +67,15 @@ private: struct ShaderEntries { u32 NumBindings() const { return static_cast<u32>(const_buffers.size() + global_buffers.size() + - texel_buffers.size() + samplers.size() + images.size()); + uniform_texels.size() + samplers.size() + storage_texels.size() + + images.size()); } std::vector<ConstBufferEntry> const_buffers; std::vector<GlobalBufferEntry> global_buffers; - std::vector<TexelBufferEntry> texel_buffers; + std::vector<UniformTexelEntry> uniform_texels; std::vector<SamplerEntry> samplers; + std::vector<StorageTexelEntry> storage_texels; std::vector<ImageEntry> images; std::set<u32> attributes; std::array<bool, Maxwell::NumClipDistances> clip_distances{}; @@ -88,7 +91,8 @@ struct Specialization final { u32 shared_memory_size{}; // Graphics specific - std::optional<float> point_size{}; + std::optional<float> point_size; + std::bitset<Maxwell::NumVertexAttributes> enabled_attributes; std::array<Maxwell::VertexAttribute::Type, Maxwell::NumVertexAttributes> attribute_types{}; bool ndc_minus_one_to_one{}; }; diff --git a/src/video_core/renderer_vulkan/vk_stream_buffer.h b/src/video_core/renderer_vulkan/vk_stream_buffer.h index dfddf7ad6..689f0d276 100644 --- a/src/video_core/renderer_vulkan/vk_stream_buffer.h +++ b/src/video_core/renderer_vulkan/vk_stream_buffer.h @@ -35,10 +35,14 @@ public: /// Ensures that "size" bytes of memory are available to the GPU, potentially recording a copy. void Unmap(u64 size); - VkBuffer GetHandle() const { + VkBuffer Handle() const noexcept { return *buffer; } + u64 Address() const noexcept { + return 0; + } + private: struct Watch final { VKFenceWatch fence; diff --git a/src/video_core/renderer_vulkan/vk_texture_cache.cpp b/src/video_core/renderer_vulkan/vk_texture_cache.cpp index 55f43e61b..430031665 100644 --- a/src/video_core/renderer_vulkan/vk_texture_cache.cpp +++ b/src/video_core/renderer_vulkan/vk_texture_cache.cpp @@ -100,8 +100,8 @@ vk::Buffer CreateBuffer(const VKDevice& device, const SurfaceParams& params, ci.pNext = nullptr; ci.flags = 0; ci.size = static_cast<VkDeviceSize>(host_memory_size); - ci.usage = VK_BUFFER_USAGE_UNIFORM_TEXEL_BUFFER_BIT | VK_BUFFER_USAGE_TRANSFER_SRC_BIT | - VK_BUFFER_USAGE_TRANSFER_DST_BIT; + ci.usage = VK_BUFFER_USAGE_UNIFORM_TEXEL_BUFFER_BIT | VK_BUFFER_USAGE_STORAGE_TEXEL_BUFFER_BIT | + VK_BUFFER_USAGE_TRANSFER_SRC_BIT | VK_BUFFER_USAGE_TRANSFER_DST_BIT; ci.sharingMode = VK_SHARING_MODE_EXCLUSIVE; ci.queueFamilyIndexCount = 0; ci.pQueueFamilyIndices = nullptr; @@ -167,6 +167,7 @@ VkImageCreateInfo GenerateImageCreateInfo(const VKDevice& device, const SurfaceP ci.extent = {params.width, params.height, 1}; break; case SurfaceTarget::Texture3D: + ci.flags |= VK_IMAGE_CREATE_2D_ARRAY_COMPATIBLE_BIT; ci.extent = {params.width, params.height, params.depth}; break; case SurfaceTarget::TextureBuffer: @@ -176,6 +177,12 @@ VkImageCreateInfo GenerateImageCreateInfo(const VKDevice& device, const SurfaceP return ci; } +u32 EncodeSwizzle(Tegra::Texture::SwizzleSource x_source, Tegra::Texture::SwizzleSource y_source, + Tegra::Texture::SwizzleSource z_source, Tegra::Texture::SwizzleSource w_source) { + return (static_cast<u32>(x_source) << 24) | (static_cast<u32>(y_source) << 16) | + (static_cast<u32>(z_source) << 8) | static_cast<u32>(w_source); +} + } // Anonymous namespace CachedSurface::CachedSurface(Core::System& system, const VKDevice& device, @@ -203,9 +210,11 @@ CachedSurface::CachedSurface(Core::System& system, const VKDevice& device, } // TODO(Rodrigo): Move this to a virtual function. - main_view = CreateViewInner( - ViewParams(params.target, 0, static_cast<u32>(params.GetNumLayers()), 0, params.num_levels), - true); + u32 num_layers = 1; + if (params.is_layered || params.target == SurfaceTarget::Texture3D) { + num_layers = params.depth; + } + main_view = CreateView(ViewParams(params.target, 0, num_layers, 0, params.num_levels)); } CachedSurface::~CachedSurface() = default; @@ -253,12 +262,8 @@ void CachedSurface::DecorateSurfaceName() { } View CachedSurface::CreateView(const ViewParams& params) { - return CreateViewInner(params, false); -} - -View CachedSurface::CreateViewInner(const ViewParams& params, bool is_proxy) { // TODO(Rodrigo): Add name decorations - return views[params] = std::make_shared<CachedSurfaceView>(device, *this, params, is_proxy); + return views[params] = std::make_shared<CachedSurfaceView>(device, *this, params); } void CachedSurface::UploadBuffer(const std::vector<u8>& staging_buffer) { @@ -342,38 +347,44 @@ VkImageSubresourceRange CachedSurface::GetImageSubresourceRange() const { } CachedSurfaceView::CachedSurfaceView(const VKDevice& device, CachedSurface& surface, - const ViewParams& params, bool is_proxy) + const ViewParams& params) : VideoCommon::ViewBase{params}, params{surface.GetSurfaceParams()}, image{surface.GetImageHandle()}, buffer_view{surface.GetBufferViewHandle()}, aspect_mask{surface.GetAspectMask()}, device{device}, surface{surface}, - base_layer{params.base_layer}, num_layers{params.num_layers}, base_level{params.base_level}, - num_levels{params.num_levels}, image_view_type{image ? GetImageViewType(params.target) - : VK_IMAGE_VIEW_TYPE_1D} {} + base_level{params.base_level}, num_levels{params.num_levels}, + image_view_type{image ? GetImageViewType(params.target) : VK_IMAGE_VIEW_TYPE_1D} { + if (image_view_type == VK_IMAGE_VIEW_TYPE_3D) { + base_layer = 0; + num_layers = 1; + base_slice = params.base_layer; + num_slices = params.num_layers; + } else { + base_layer = params.base_layer; + num_layers = params.num_layers; + } +} CachedSurfaceView::~CachedSurfaceView() = default; -VkImageView CachedSurfaceView::GetHandle(SwizzleSource x_source, SwizzleSource y_source, - SwizzleSource z_source, SwizzleSource w_source) { - const u32 swizzle = EncodeSwizzle(x_source, y_source, z_source, w_source); - if (last_image_view && last_swizzle == swizzle) { +VkImageView CachedSurfaceView::GetImageView(SwizzleSource x_source, SwizzleSource y_source, + SwizzleSource z_source, SwizzleSource w_source) { + const u32 new_swizzle = EncodeSwizzle(x_source, y_source, z_source, w_source); + if (last_image_view && last_swizzle == new_swizzle) { return last_image_view; } - last_swizzle = swizzle; + last_swizzle = new_swizzle; - const auto [entry, is_cache_miss] = view_cache.try_emplace(swizzle); + const auto [entry, is_cache_miss] = view_cache.try_emplace(new_swizzle); auto& image_view = entry->second; if (!is_cache_miss) { return last_image_view = *image_view; } - auto swizzle_x = MaxwellToVK::SwizzleSource(x_source); - auto swizzle_y = MaxwellToVK::SwizzleSource(y_source); - auto swizzle_z = MaxwellToVK::SwizzleSource(z_source); - auto swizzle_w = MaxwellToVK::SwizzleSource(w_source); - + std::array swizzle{MaxwellToVK::SwizzleSource(x_source), MaxwellToVK::SwizzleSource(y_source), + MaxwellToVK::SwizzleSource(z_source), MaxwellToVK::SwizzleSource(w_source)}; if (params.pixel_format == VideoCore::Surface::PixelFormat::A1B5G5R5U) { // A1B5G5R5 is implemented as A1R5G5B5, we have to change the swizzle here. - std::swap(swizzle_x, swizzle_z); + std::swap(swizzle[0], swizzle[2]); } // Games can sample depth or stencil values on textures. This is decided by the swizzle value on @@ -395,11 +406,16 @@ VkImageView CachedSurfaceView::GetHandle(SwizzleSource x_source, SwizzleSource y UNIMPLEMENTED(); } - // Vulkan doesn't seem to understand swizzling of a depth stencil image, use identity - swizzle_x = VK_COMPONENT_SWIZZLE_R; - swizzle_y = VK_COMPONENT_SWIZZLE_G; - swizzle_z = VK_COMPONENT_SWIZZLE_B; - swizzle_w = VK_COMPONENT_SWIZZLE_A; + // Make sure we sample the first component + std::transform( + swizzle.begin(), swizzle.end(), swizzle.begin(), [](VkComponentSwizzle component) { + return component == VK_COMPONENT_SWIZZLE_G ? VK_COMPONENT_SWIZZLE_R : component; + }); + } + + if (image_view_type == VK_IMAGE_VIEW_TYPE_3D) { + ASSERT(base_slice == 0); + ASSERT(num_slices == params.depth); } VkImageViewCreateInfo ci; @@ -409,7 +425,7 @@ VkImageView CachedSurfaceView::GetHandle(SwizzleSource x_source, SwizzleSource y ci.image = surface.GetImageHandle(); ci.viewType = image_view_type; ci.format = surface.GetImage().GetFormat(); - ci.components = {swizzle_x, swizzle_y, swizzle_z, swizzle_w}; + ci.components = {swizzle[0], swizzle[1], swizzle[2], swizzle[3]}; ci.subresourceRange.aspectMask = aspect; ci.subresourceRange.baseMipLevel = base_level; ci.subresourceRange.levelCount = num_levels; @@ -420,6 +436,35 @@ VkImageView CachedSurfaceView::GetHandle(SwizzleSource x_source, SwizzleSource y return last_image_view = *image_view; } +VkImageView CachedSurfaceView::GetAttachment() { + if (render_target) { + return *render_target; + } + + VkImageViewCreateInfo ci; + ci.sType = VK_STRUCTURE_TYPE_IMAGE_VIEW_CREATE_INFO; + ci.pNext = nullptr; + ci.flags = 0; + ci.image = surface.GetImageHandle(); + ci.format = surface.GetImage().GetFormat(); + ci.components = {VK_COMPONENT_SWIZZLE_IDENTITY, VK_COMPONENT_SWIZZLE_IDENTITY, + VK_COMPONENT_SWIZZLE_IDENTITY, VK_COMPONENT_SWIZZLE_IDENTITY}; + ci.subresourceRange.aspectMask = aspect_mask; + ci.subresourceRange.baseMipLevel = base_level; + ci.subresourceRange.levelCount = num_levels; + if (image_view_type == VK_IMAGE_VIEW_TYPE_3D) { + ci.viewType = num_slices > 1 ? VK_IMAGE_VIEW_TYPE_2D_ARRAY : VK_IMAGE_VIEW_TYPE_2D; + ci.subresourceRange.baseArrayLayer = base_slice; + ci.subresourceRange.layerCount = num_slices; + } else { + ci.viewType = image_view_type; + ci.subresourceRange.baseArrayLayer = base_layer; + ci.subresourceRange.layerCount = num_layers; + } + render_target = device.GetLogical().CreateImageView(ci); + return *render_target; +} + VKTextureCache::VKTextureCache(Core::System& system, VideoCore::RasterizerInterface& rasterizer, const VKDevice& device, VKResourceManager& resource_manager, VKMemoryManager& memory_manager, VKScheduler& scheduler, diff --git a/src/video_core/renderer_vulkan/vk_texture_cache.h b/src/video_core/renderer_vulkan/vk_texture_cache.h index f211ccb1e..807e26c8a 100644 --- a/src/video_core/renderer_vulkan/vk_texture_cache.h +++ b/src/video_core/renderer_vulkan/vk_texture_cache.h @@ -91,7 +91,6 @@ protected: void DecorateSurfaceName(); View CreateView(const ViewParams& params) override; - View CreateViewInner(const ViewParams& params, bool is_proxy); private: void UploadBuffer(const std::vector<u8>& staging_buffer); @@ -120,23 +119,20 @@ private: class CachedSurfaceView final : public VideoCommon::ViewBase { public: explicit CachedSurfaceView(const VKDevice& device, CachedSurface& surface, - const ViewParams& params, bool is_proxy); + const ViewParams& params); ~CachedSurfaceView(); - VkImageView GetHandle(Tegra::Texture::SwizzleSource x_source, - Tegra::Texture::SwizzleSource y_source, - Tegra::Texture::SwizzleSource z_source, - Tegra::Texture::SwizzleSource w_source); + VkImageView GetImageView(Tegra::Texture::SwizzleSource x_source, + Tegra::Texture::SwizzleSource y_source, + Tegra::Texture::SwizzleSource z_source, + Tegra::Texture::SwizzleSource w_source); + + VkImageView GetAttachment(); bool IsSameSurface(const CachedSurfaceView& rhs) const { return &surface == &rhs.surface; } - VkImageView GetHandle() { - return GetHandle(Tegra::Texture::SwizzleSource::R, Tegra::Texture::SwizzleSource::G, - Tegra::Texture::SwizzleSource::B, Tegra::Texture::SwizzleSource::A); - } - u32 GetWidth() const { return params.GetMipWidth(base_level); } @@ -180,14 +176,6 @@ public: } private: - static u32 EncodeSwizzle(Tegra::Texture::SwizzleSource x_source, - Tegra::Texture::SwizzleSource y_source, - Tegra::Texture::SwizzleSource z_source, - Tegra::Texture::SwizzleSource w_source) { - return (static_cast<u32>(x_source) << 24) | (static_cast<u32>(y_source) << 16) | - (static_cast<u32>(z_source) << 8) | static_cast<u32>(w_source); - } - // Store a copy of these values to avoid double dereference when reading them const SurfaceParams params; const VkImage image; @@ -196,15 +184,18 @@ private: const VKDevice& device; CachedSurface& surface; - const u32 base_layer; - const u32 num_layers; const u32 base_level; const u32 num_levels; const VkImageViewType image_view_type; + u32 base_layer = 0; + u32 num_layers = 0; + u32 base_slice = 0; + u32 num_slices = 0; VkImageView last_image_view = nullptr; u32 last_swizzle = 0; + vk::ImageView render_target; std::unordered_map<u32, vk::ImageView> view_cache; }; diff --git a/src/video_core/renderer_vulkan/vk_update_descriptor.cpp b/src/video_core/renderer_vulkan/vk_update_descriptor.cpp index 681ecde98..351c048d2 100644 --- a/src/video_core/renderer_vulkan/vk_update_descriptor.cpp +++ b/src/video_core/renderer_vulkan/vk_update_descriptor.cpp @@ -24,35 +24,25 @@ void VKUpdateDescriptorQueue::TickFrame() { } void VKUpdateDescriptorQueue::Acquire() { - entries.clear(); -} + // Minimum number of entries required. + // This is the maximum number of entries a single draw call migth use. + static constexpr std::size_t MIN_ENTRIES = 0x400; -void VKUpdateDescriptorQueue::Send(VkDescriptorUpdateTemplateKHR update_template, - VkDescriptorSet set) { - if (payload.size() + entries.size() >= payload.max_size()) { + if (payload.size() + MIN_ENTRIES >= payload.max_size()) { LOG_WARNING(Render_Vulkan, "Payload overflow, waiting for worker thread"); scheduler.WaitWorker(); payload.clear(); } + upload_start = &*payload.end(); +} - // TODO(Rodrigo): Rework to write the payload directly - const auto payload_start = payload.data() + payload.size(); - for (const auto& entry : entries) { - if (const auto image = std::get_if<VkDescriptorImageInfo>(&entry)) { - payload.push_back(*image); - } else if (const auto buffer = std::get_if<VkDescriptorBufferInfo>(&entry)) { - payload.push_back(*buffer); - } else if (const auto texel = std::get_if<VkBufferView>(&entry)) { - payload.push_back(*texel); - } else { - UNREACHABLE(); - } - } - - scheduler.Record( - [payload_start, set, update_template, logical = &device.GetLogical()](vk::CommandBuffer) { - logical->UpdateDescriptorSet(set, update_template, payload_start); - }); +void VKUpdateDescriptorQueue::Send(VkDescriptorUpdateTemplateKHR update_template, + VkDescriptorSet set) { + const void* const data = upload_start; + const vk::Device* const logical = &device.GetLogical(); + scheduler.Record([data, logical, set, update_template](vk::CommandBuffer) { + logical->UpdateDescriptorSet(set, update_template, data); + }); } } // namespace Vulkan diff --git a/src/video_core/renderer_vulkan/vk_update_descriptor.h b/src/video_core/renderer_vulkan/vk_update_descriptor.h index cc7e3dff4..945320c72 100644 --- a/src/video_core/renderer_vulkan/vk_update_descriptor.h +++ b/src/video_core/renderer_vulkan/vk_update_descriptor.h @@ -15,17 +15,13 @@ namespace Vulkan { class VKDevice; class VKScheduler; -class DescriptorUpdateEntry { -public: - explicit DescriptorUpdateEntry() {} - - DescriptorUpdateEntry(VkDescriptorImageInfo image) : image{image} {} +struct DescriptorUpdateEntry { + DescriptorUpdateEntry(VkDescriptorImageInfo image_) : image{image_} {} - DescriptorUpdateEntry(VkDescriptorBufferInfo buffer) : buffer{buffer} {} + DescriptorUpdateEntry(VkDescriptorBufferInfo buffer_) : buffer{buffer_} {} - DescriptorUpdateEntry(VkBufferView texel_buffer) : texel_buffer{texel_buffer} {} + DescriptorUpdateEntry(VkBufferView texel_buffer_) : texel_buffer{texel_buffer_} {} -private: union { VkDescriptorImageInfo image; VkDescriptorBufferInfo buffer; @@ -45,32 +41,34 @@ public: void Send(VkDescriptorUpdateTemplateKHR update_template, VkDescriptorSet set); void AddSampledImage(VkSampler sampler, VkImageView image_view) { - entries.emplace_back(VkDescriptorImageInfo{sampler, image_view, {}}); + payload.emplace_back(VkDescriptorImageInfo{sampler, image_view, {}}); } void AddImage(VkImageView image_view) { - entries.emplace_back(VkDescriptorImageInfo{{}, image_view, {}}); + payload.emplace_back(VkDescriptorImageInfo{{}, image_view, {}}); } void AddBuffer(VkBuffer buffer, u64 offset, std::size_t size) { - entries.emplace_back(VkDescriptorBufferInfo{buffer, offset, size}); + payload.emplace_back(VkDescriptorBufferInfo{buffer, offset, size}); } void AddTexelBuffer(VkBufferView texel_buffer) { - entries.emplace_back(texel_buffer); + payload.emplace_back(texel_buffer); } - VkImageLayout* GetLastImageLayout() { - return &std::get<VkDescriptorImageInfo>(entries.back()).imageLayout; + VkImageLayout* LastImageLayout() { + return &payload.back().image.imageLayout; } -private: - using Variant = std::variant<VkDescriptorImageInfo, VkDescriptorBufferInfo, VkBufferView>; + const VkImageLayout* LastImageLayout() const { + return &payload.back().image.imageLayout; + } +private: const VKDevice& device; VKScheduler& scheduler; - boost::container::static_vector<Variant, 0x400> entries; + const DescriptorUpdateEntry* upload_start = nullptr; boost::container::static_vector<DescriptorUpdateEntry, 0x10000> payload; }; diff --git a/src/video_core/renderer_vulkan/wrapper.cpp b/src/video_core/renderer_vulkan/wrapper.cpp index 2ce9b0626..0d485a662 100644 --- a/src/video_core/renderer_vulkan/wrapper.cpp +++ b/src/video_core/renderer_vulkan/wrapper.cpp @@ -153,7 +153,8 @@ void Load(VkDevice device, DeviceDispatch& dld) noexcept { bool Load(InstanceDispatch& dld) noexcept { #define X(name) Proc(dld.name, dld, #name) - return X(vkCreateInstance) && X(vkEnumerateInstanceExtensionProperties); + return X(vkCreateInstance) && X(vkEnumerateInstanceExtensionProperties) && + X(vkEnumerateInstanceLayerProperties); #undef X } @@ -725,8 +726,7 @@ bool PhysicalDevice::GetSurfaceSupportKHR(u32 queue_family_index, VkSurfaceKHR s return supported == VK_TRUE; } -VkSurfaceCapabilitiesKHR PhysicalDevice::GetSurfaceCapabilitiesKHR(VkSurfaceKHR surface) const - noexcept { +VkSurfaceCapabilitiesKHR PhysicalDevice::GetSurfaceCapabilitiesKHR(VkSurfaceKHR surface) const { VkSurfaceCapabilitiesKHR capabilities; Check(dld->vkGetPhysicalDeviceSurfaceCapabilitiesKHR(physical_device, surface, &capabilities)); return capabilities; @@ -771,4 +771,17 @@ std::optional<std::vector<VkExtensionProperties>> EnumerateInstanceExtensionProp return properties; } +std::optional<std::vector<VkLayerProperties>> EnumerateInstanceLayerProperties( + const InstanceDispatch& dld) { + u32 num; + if (dld.vkEnumerateInstanceLayerProperties(&num, nullptr) != VK_SUCCESS) { + return std::nullopt; + } + std::vector<VkLayerProperties> properties(num); + if (dld.vkEnumerateInstanceLayerProperties(&num, properties.data()) != VK_SUCCESS) { + return std::nullopt; + } + return properties; +} + } // namespace Vulkan::vk diff --git a/src/video_core/renderer_vulkan/wrapper.h b/src/video_core/renderer_vulkan/wrapper.h index 98937a77a..d56fdb3f9 100644 --- a/src/video_core/renderer_vulkan/wrapper.h +++ b/src/video_core/renderer_vulkan/wrapper.h @@ -141,6 +141,7 @@ struct InstanceDispatch { PFN_vkCreateInstance vkCreateInstance; PFN_vkDestroyInstance vkDestroyInstance; PFN_vkEnumerateInstanceExtensionProperties vkEnumerateInstanceExtensionProperties; + PFN_vkEnumerateInstanceLayerProperties vkEnumerateInstanceLayerProperties; PFN_vkCreateDebugUtilsMessengerEXT vkCreateDebugUtilsMessengerEXT; PFN_vkCreateDevice vkCreateDevice; @@ -779,7 +780,7 @@ public: bool GetSurfaceSupportKHR(u32 queue_family_index, VkSurfaceKHR) const; - VkSurfaceCapabilitiesKHR GetSurfaceCapabilitiesKHR(VkSurfaceKHR) const noexcept; + VkSurfaceCapabilitiesKHR GetSurfaceCapabilitiesKHR(VkSurfaceKHR) const; std::vector<VkSurfaceFormatKHR> GetSurfaceFormatsKHR(VkSurfaceKHR) const; @@ -996,4 +997,7 @@ private: std::optional<std::vector<VkExtensionProperties>> EnumerateInstanceExtensionProperties( const InstanceDispatch& dld); +std::optional<std::vector<VkLayerProperties>> EnumerateInstanceLayerProperties( + const InstanceDispatch& dld); + } // namespace Vulkan::vk diff --git a/src/video_core/shader/decode.cpp b/src/video_core/shader/decode.cpp index a75a5cc63..eeac328a6 100644 --- a/src/video_core/shader/decode.cpp +++ b/src/video_core/shader/decode.cpp @@ -255,7 +255,7 @@ void ShaderIR::InsertControlFlow(NodeBlock& bb, const ShaderBlock& block) { Node n = Operation(OperationCode::Branch, Immediate(branch_case.address)); Node op_b = Immediate(branch_case.cmp_value); Node condition = - GetPredicateComparisonInteger(Tegra::Shader::PredCondition::Equal, false, op_a, op_b); + GetPredicateComparisonInteger(Tegra::Shader::PredCondition::EQ, false, op_a, op_b); auto result = Conditional(condition, {n}); bb.push_back(result); global_code.push_back(result); diff --git a/src/video_core/shader/decode/half_set.cpp b/src/video_core/shader/decode/half_set.cpp index 848e46874..b2e88fa20 100644 --- a/src/video_core/shader/decode/half_set.cpp +++ b/src/video_core/shader/decode/half_set.cpp @@ -13,55 +13,101 @@ namespace VideoCommon::Shader { +using std::move; using Tegra::Shader::Instruction; using Tegra::Shader::OpCode; +using Tegra::Shader::PredCondition; u32 ShaderIR::DecodeHalfSet(NodeBlock& bb, u32 pc) { const Instruction instr = {program_code[pc]}; const auto opcode = OpCode::Decode(instr); - if (instr.hset2.ftz == 0) { - LOG_DEBUG(HW_GPU, "{} without FTZ is not implemented", opcode->get().GetName()); + PredCondition cond; + bool bf; + bool ftz; + bool neg_a; + bool abs_a; + bool neg_b; + bool abs_b; + switch (opcode->get().GetId()) { + case OpCode::Id::HSET2_C: + case OpCode::Id::HSET2_IMM: + cond = instr.hsetp2.cbuf_and_imm.cond; + bf = instr.Bit(53); + ftz = instr.Bit(54); + neg_a = instr.Bit(43); + abs_a = instr.Bit(44); + neg_b = instr.Bit(56); + abs_b = instr.Bit(54); + break; + case OpCode::Id::HSET2_R: + cond = instr.hsetp2.reg.cond; + bf = instr.Bit(49); + ftz = instr.Bit(50); + neg_a = instr.Bit(43); + abs_a = instr.Bit(44); + neg_b = instr.Bit(31); + abs_b = instr.Bit(30); + break; + default: + UNREACHABLE(); } - Node op_a = UnpackHalfFloat(GetRegister(instr.gpr8), instr.hset2.type_a); - op_a = GetOperandAbsNegHalf(op_a, instr.hset2.abs_a, instr.hset2.negate_a); - - Node op_b = [&]() { + Node op_b = [this, instr, opcode] { switch (opcode->get().GetId()) { + case OpCode::Id::HSET2_C: + // Inform as unimplemented as this is not tested. + UNIMPLEMENTED_MSG("HSET2_C is not implemented"); + return GetConstBuffer(instr.cbuf34.index, instr.cbuf34.GetOffset()); case OpCode::Id::HSET2_R: return GetRegister(instr.gpr20); + case OpCode::Id::HSET2_IMM: + return UnpackHalfImmediate(instr, true); default: UNREACHABLE(); - return Immediate(0); + return Node{}; } }(); - op_b = UnpackHalfFloat(op_b, instr.hset2.type_b); - op_b = GetOperandAbsNegHalf(op_b, instr.hset2.abs_b, instr.hset2.negate_b); - const Node second_pred = GetPredicate(instr.hset2.pred39, instr.hset2.neg_pred); + if (!ftz) { + LOG_DEBUG(HW_GPU, "{} without FTZ is not implemented", opcode->get().GetName()); + } + + Node op_a = UnpackHalfFloat(GetRegister(instr.gpr8), instr.hset2.type_a); + op_a = GetOperandAbsNegHalf(op_a, abs_a, neg_a); + + switch (opcode->get().GetId()) { + case OpCode::Id::HSET2_R: + op_b = GetOperandAbsNegHalf(move(op_b), abs_b, neg_b); + [[fallthrough]]; + case OpCode::Id::HSET2_C: + op_b = UnpackHalfFloat(move(op_b), instr.hset2.type_b); + break; + default: + break; + } - const Node comparison_pair = GetPredicateComparisonHalf(instr.hset2.cond, op_a, op_b); + Node second_pred = GetPredicate(instr.hset2.pred39, instr.hset2.neg_pred); + + Node comparison_pair = GetPredicateComparisonHalf(cond, op_a, op_b); const OperationCode combiner = GetPredicateCombiner(instr.hset2.op); // HSET2 operates on each half float in the pack. std::array<Node, 2> values; for (u32 i = 0; i < 2; ++i) { - const u32 raw_value = instr.hset2.bf ? 0x3c00 : 0xffff; - const Node true_value = Immediate(raw_value << (i * 16)); - const Node false_value = Immediate(0); - - const Node comparison = - Operation(OperationCode::LogicalPick2, comparison_pair, Immediate(i)); - const Node predicate = Operation(combiner, comparison, second_pred); + const u32 raw_value = bf ? 0x3c00 : 0xffff; + Node true_value = Immediate(raw_value << (i * 16)); + Node false_value = Immediate(0); + Node comparison = Operation(OperationCode::LogicalPick2, comparison_pair, Immediate(i)); + Node predicate = Operation(combiner, comparison, second_pred); values[i] = - Operation(OperationCode::Select, NO_PRECISE, predicate, true_value, false_value); + Operation(OperationCode::Select, predicate, move(true_value), move(false_value)); } - const Node value = Operation(OperationCode::UBitwiseOr, NO_PRECISE, values[0], values[1]); - SetRegister(bb, instr.gpr0, value); + Node value = Operation(OperationCode::UBitwiseOr, values[0], values[1]); + SetRegister(bb, instr.gpr0, move(value)); return pc; } diff --git a/src/video_core/shader/decode/image.cpp b/src/video_core/shader/decode/image.cpp index 60b6ad72a..07778dc3e 100644 --- a/src/video_core/shader/decode/image.cpp +++ b/src/video_core/shader/decode/image.cpp @@ -97,6 +97,7 @@ ComponentType GetComponentType(Tegra::Engines::SamplerDescriptor descriptor, break; case TextureFormat::B5G6R5: case TextureFormat::B6G5R5: + case TextureFormat::BF10GF11RF11: if (component == 0) { return descriptor.b_type; } @@ -119,7 +120,7 @@ ComponentType GetComponentType(Tegra::Engines::SamplerDescriptor descriptor, } break; } - UNIMPLEMENTED_MSG("texture format not implement={}", format); + UNIMPLEMENTED_MSG("Texture format not implemented={}", format); return ComponentType::FLOAT; } @@ -191,6 +192,14 @@ u32 GetComponentSize(TextureFormat format, std::size_t component) { return 6; } return 0; + case TextureFormat::BF10GF11RF11: + if (component == 1 || component == 2) { + return 11; + } + if (component == 0) { + return 10; + } + return 0; case TextureFormat::G8R24: if (component == 0) { return 8; @@ -211,10 +220,9 @@ u32 GetComponentSize(TextureFormat format, std::size_t component) { return (component == 0 || component == 1) ? 8 : 0; case TextureFormat::G4R4: return (component == 0 || component == 1) ? 4 : 0; - default: - UNIMPLEMENTED_MSG("texture format not implement={}", format); - return 0; } + UNIMPLEMENTED_MSG("Texture format not implemented={}", format); + return 0; } std::size_t GetImageComponentMask(TextureFormat format) { @@ -235,6 +243,7 @@ std::size_t GetImageComponentMask(TextureFormat format) { case TextureFormat::R32_B24G8: case TextureFormat::B5G6R5: case TextureFormat::B6G5R5: + case TextureFormat::BF10GF11RF11: return std::size_t{R | G | B}; case TextureFormat::R32_G32: case TextureFormat::R16_G16: @@ -248,10 +257,9 @@ std::size_t GetImageComponentMask(TextureFormat format) { case TextureFormat::R8: case TextureFormat::R1: return std::size_t{R}; - default: - UNIMPLEMENTED_MSG("texture format not implement={}", format); - return std::size_t{R | G | B | A}; } + UNIMPLEMENTED_MSG("Texture format not implemented={}", format); + return std::size_t{R | G | B | A}; } std::size_t GetImageTypeNumCoordinates(Tegra::Shader::ImageType image_type) { @@ -299,7 +307,7 @@ std::pair<Node, bool> ShaderIR::GetComponentValue(ComponentType component_type, return {std::move(original_value), true}; } default: - UNIMPLEMENTED_MSG("Unimplement component type={}", component_type); + UNIMPLEMENTED_MSG("Unimplemented component type={}", component_type); return {std::move(original_value), true}; } } @@ -459,7 +467,7 @@ u32 ShaderIR::DecodeImage(NodeBlock& bb, u32 pc) { default: break; } - UNIMPLEMENTED_MSG("Unimplemented operation={} type={}", + UNIMPLEMENTED_MSG("Unimplemented operation={}, type={}", static_cast<u64>(instr.suatom_d.operation.Value()), static_cast<u64>(instr.suatom_d.operation_type.Value())); return OperationCode::AtomicImageAdd; diff --git a/src/video_core/shader/decode/memory.cpp b/src/video_core/shader/decode/memory.cpp index 9392f065b..63adbc4a3 100644 --- a/src/video_core/shader/decode/memory.cpp +++ b/src/video_core/shader/decode/memory.cpp @@ -387,7 +387,6 @@ u32 ShaderIR::DecodeMemory(NodeBlock& bb, u32 pc) { } case OpCode::Id::RED: { UNIMPLEMENTED_IF_MSG(instr.red.type != GlobalAtomicType::U32); - UNIMPLEMENTED_IF_MSG(instr.red.operation != AtomicOp::Add); const auto [real_address, base_address, descriptor] = TrackGlobalMemory(bb, instr, true, true); if (!real_address || !base_address) { @@ -396,7 +395,7 @@ u32 ShaderIR::DecodeMemory(NodeBlock& bb, u32 pc) { } Node gmem = MakeNode<GmemNode>(real_address, base_address, descriptor); Node value = GetRegister(instr.gpr0); - bb.push_back(Operation(OperationCode::ReduceIAdd, move(gmem), move(value))); + bb.push_back(Operation(GetAtomOperation(instr.red.operation), move(gmem), move(value))); break; } case OpCode::Id::ATOM: { diff --git a/src/video_core/shader/decode/other.cpp b/src/video_core/shader/decode/other.cpp index d4f95b18c..c0a8f233f 100644 --- a/src/video_core/shader/decode/other.cpp +++ b/src/video_core/shader/decode/other.cpp @@ -83,7 +83,7 @@ u32 ShaderIR::DecodeOther(NodeBlock& bb, u32 pc) { return Operation(OperationCode::YNegate); case SystemVariable::InvocationInfo: LOG_WARNING(HW_GPU, "S2R instruction with InvocationInfo is incomplete"); - return Immediate(0U); + return Immediate(0x00ff'0000U); case SystemVariable::WscaleFactorXY: UNIMPLEMENTED_MSG("S2R WscaleFactorXY is not implemented"); return Immediate(0U); @@ -109,6 +109,27 @@ u32 ShaderIR::DecodeOther(NodeBlock& bb, u32 pc) { return Operation(OperationCode::WorkGroupIdY); case SystemVariable::CtaIdZ: return Operation(OperationCode::WorkGroupIdZ); + case SystemVariable::EqMask: + case SystemVariable::LtMask: + case SystemVariable::LeMask: + case SystemVariable::GtMask: + case SystemVariable::GeMask: + uses_warps = true; + switch (instr.sys20) { + case SystemVariable::EqMask: + return Operation(OperationCode::ThreadEqMask); + case SystemVariable::LtMask: + return Operation(OperationCode::ThreadLtMask); + case SystemVariable::LeMask: + return Operation(OperationCode::ThreadLeMask); + case SystemVariable::GtMask: + return Operation(OperationCode::ThreadGtMask); + case SystemVariable::GeMask: + return Operation(OperationCode::ThreadGeMask); + default: + UNREACHABLE(); + return Immediate(0u); + } default: UNIMPLEMENTED_MSG("Unhandled system move: {}", static_cast<u32>(instr.sys20.Value())); @@ -272,10 +293,25 @@ u32 ShaderIR::DecodeOther(NodeBlock& bb, u32 pc) { SetRegister(bb, instr.gpr0, GetRegister(instr.gpr8)); break; } + case OpCode::Id::BAR: { + UNIMPLEMENTED_IF_MSG(instr.value != 0xF0A81B8000070000ULL, "BAR is not BAR.SYNC 0x0"); + bb.push_back(Operation(OperationCode::Barrier)); + break; + } case OpCode::Id::MEMBAR: { - UNIMPLEMENTED_IF(instr.membar.type != Tegra::Shader::MembarType::GL); UNIMPLEMENTED_IF(instr.membar.unknown != Tegra::Shader::MembarUnknown::Default); - bb.push_back(Operation(OperationCode::MemoryBarrierGL)); + const OperationCode type = [instr] { + switch (instr.membar.type) { + case Tegra::Shader::MembarType::CTA: + return OperationCode::MemoryBarrierGroup; + case Tegra::Shader::MembarType::GL: + return OperationCode::MemoryBarrierGlobal; + default: + UNIMPLEMENTED_MSG("MEMBAR type={}", static_cast<int>(instr.membar.type.Value())); + return OperationCode::MemoryBarrierGlobal; + } + }(); + bb.push_back(Operation(type)); break; } case OpCode::Id::DEPBAR: { diff --git a/src/video_core/shader/decode/texture.cpp b/src/video_core/shader/decode/texture.cpp index 8f0bb996e..29ebf65ba 100644 --- a/src/video_core/shader/decode/texture.cpp +++ b/src/video_core/shader/decode/texture.cpp @@ -357,13 +357,11 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) { return pc; } -ShaderIR::SamplerInfo ShaderIR::GetSamplerInfo(SamplerInfo info, u32 offset, - std::optional<u32> buffer) { +ShaderIR::SamplerInfo ShaderIR::GetSamplerInfo( + SamplerInfo info, std::optional<Tegra::Engines::SamplerDescriptor> sampler) { if (info.IsComplete()) { return info; } - const auto sampler = buffer ? registry.ObtainBindlessSampler(*buffer, offset) - : registry.ObtainBoundSampler(offset); if (!sampler) { LOG_WARNING(HW_GPU, "Unknown sampler info"); info.type = info.type.value_or(Tegra::Shader::TextureType::Texture2D); @@ -381,8 +379,8 @@ ShaderIR::SamplerInfo ShaderIR::GetSamplerInfo(SamplerInfo info, u32 offset, std::optional<Sampler> ShaderIR::GetSampler(Tegra::Shader::Sampler sampler, SamplerInfo sampler_info) { - const auto offset = static_cast<u32>(sampler.index.Value()); - const auto info = GetSamplerInfo(sampler_info, offset); + const u32 offset = static_cast<u32>(sampler.index.Value()); + const auto info = GetSamplerInfo(sampler_info, registry.ObtainBoundSampler(offset)); // If this sampler has already been used, return the existing mapping. const auto it = std::find_if(used_samplers.begin(), used_samplers.end(), @@ -404,20 +402,19 @@ std::optional<Sampler> ShaderIR::GetBindlessSampler(Tegra::Shader::Register reg, const Node sampler_register = GetRegister(reg); const auto [base_node, tracked_sampler_info] = TrackBindlessSampler(sampler_register, global_code, static_cast<s64>(global_code.size())); - ASSERT(base_node != nullptr); - if (base_node == nullptr) { + if (!base_node) { + UNREACHABLE(); return std::nullopt; } - if (const auto bindless_sampler_info = - std::get_if<BindlessSamplerNode>(&*tracked_sampler_info)) { - const u32 buffer = bindless_sampler_info->GetIndex(); - const u32 offset = bindless_sampler_info->GetOffset(); - info = GetSamplerInfo(info, offset, buffer); + if (const auto sampler_info = std::get_if<BindlessSamplerNode>(&*tracked_sampler_info)) { + const u32 buffer = sampler_info->index; + const u32 offset = sampler_info->offset; + info = GetSamplerInfo(info, registry.ObtainBindlessSampler(buffer, offset)); // If this sampler has already been used, return the existing mapping. const auto it = std::find_if(used_samplers.begin(), used_samplers.end(), - [buffer = buffer, offset = offset](const Sampler& entry) { + [buffer, offset](const Sampler& entry) { return entry.buffer == buffer && entry.offset == offset; }); if (it != used_samplers.end()) { @@ -431,10 +428,32 @@ std::optional<Sampler> ShaderIR::GetBindlessSampler(Tegra::Shader::Register reg, return used_samplers.emplace_back(next_index, offset, buffer, *info.type, *info.is_array, *info.is_shadow, *info.is_buffer, false); } - if (const auto array_sampler_info = std::get_if<ArraySamplerNode>(&*tracked_sampler_info)) { - const u32 base_offset = array_sampler_info->GetBaseOffset() / 4; - index_var = GetCustomVariable(array_sampler_info->GetIndexVar()); - info = GetSamplerInfo(info, base_offset); + if (const auto sampler_info = std::get_if<SeparateSamplerNode>(&*tracked_sampler_info)) { + const std::pair indices = sampler_info->indices; + const std::pair offsets = sampler_info->offsets; + info = GetSamplerInfo(info, registry.ObtainSeparateSampler(indices, offsets)); + + // Try to use an already created sampler if it exists + const auto it = std::find_if( + used_samplers.begin(), used_samplers.end(), [indices, offsets](const Sampler& entry) { + return offsets == std::pair{entry.offset, entry.secondary_offset} && + indices == std::pair{entry.buffer, entry.secondary_buffer}; + }); + if (it != used_samplers.end()) { + ASSERT(it->is_separated && it->type == info.type && it->is_array == info.is_array && + it->is_shadow == info.is_shadow && it->is_buffer == info.is_buffer); + return *it; + } + + // Otherwise create a new mapping for this sampler + const u32 next_index = static_cast<u32>(used_samplers.size()); + return used_samplers.emplace_back(next_index, offsets, indices, *info.type, *info.is_array, + *info.is_shadow, *info.is_buffer); + } + if (const auto sampler_info = std::get_if<ArraySamplerNode>(&*tracked_sampler_info)) { + const u32 base_offset = sampler_info->base_offset / 4; + index_var = GetCustomVariable(sampler_info->bindless_var); + info = GetSamplerInfo(info, registry.ObtainBoundSampler(base_offset)); // If this sampler has already been used, return the existing mapping. const auto it = std::find_if( diff --git a/src/video_core/shader/decode/xmad.cpp b/src/video_core/shader/decode/xmad.cpp index 6191ffba1..c83dc6615 100644 --- a/src/video_core/shader/decode/xmad.cpp +++ b/src/video_core/shader/decode/xmad.cpp @@ -97,19 +97,19 @@ u32 ShaderIR::DecodeXmad(NodeBlock& bb, u32 pc) { return SignedOperation(OperationCode::IAdd, is_signed_c, original_c, shifted_b); } case Tegra::Shader::XmadMode::CSfu: { - const Node comp_a = GetPredicateComparisonInteger(PredCondition::Equal, is_signed_a, - op_a, Immediate(0)); - const Node comp_b = GetPredicateComparisonInteger(PredCondition::Equal, is_signed_b, - op_b, Immediate(0)); + const Node comp_a = + GetPredicateComparisonInteger(PredCondition::EQ, is_signed_a, op_a, Immediate(0)); + const Node comp_b = + GetPredicateComparisonInteger(PredCondition::EQ, is_signed_b, op_b, Immediate(0)); const Node comp = Operation(OperationCode::LogicalOr, comp_a, comp_b); const Node comp_minus_a = GetPredicateComparisonInteger( - PredCondition::NotEqual, is_signed_a, + PredCondition::NE, is_signed_a, SignedOperation(OperationCode::IBitwiseAnd, is_signed_a, op_a, Immediate(0x80000000)), Immediate(0)); const Node comp_minus_b = GetPredicateComparisonInteger( - PredCondition::NotEqual, is_signed_b, + PredCondition::NE, is_signed_b, SignedOperation(OperationCode::IBitwiseAnd, is_signed_b, op_b, Immediate(0x80000000)), Immediate(0)); diff --git a/src/video_core/shader/memory_util.cpp b/src/video_core/shader/memory_util.cpp index 074f21691..5071c83ca 100644 --- a/src/video_core/shader/memory_util.cpp +++ b/src/video_core/shader/memory_util.cpp @@ -66,12 +66,12 @@ ProgramCode GetShaderCode(Tegra::MemoryManager& memory_manager, GPUVAddr gpu_add u64 GetUniqueIdentifier(Tegra::Engines::ShaderType shader_type, bool is_a, const ProgramCode& code, const ProgramCode& code_b) { - u64 unique_identifier = boost::hash_value(code); + size_t unique_identifier = boost::hash_value(code); if (is_a) { // VertexA programs include two programs boost::hash_combine(unique_identifier, boost::hash_value(code_b)); } - return unique_identifier; + return static_cast<u64>(unique_identifier); } } // namespace VideoCommon::Shader diff --git a/src/video_core/shader/node.h b/src/video_core/shader/node.h index 601c822d2..8f230d57a 100644 --- a/src/video_core/shader/node.h +++ b/src/video_core/shader/node.h @@ -110,13 +110,20 @@ enum class OperationCode { LogicalPick2, /// (bool2 pair, uint index) -> bool LogicalAnd2, /// (bool2 a) -> bool - LogicalFLessThan, /// (float a, float b) -> bool - LogicalFEqual, /// (float a, float b) -> bool - LogicalFLessEqual, /// (float a, float b) -> bool - LogicalFGreaterThan, /// (float a, float b) -> bool - LogicalFNotEqual, /// (float a, float b) -> bool - LogicalFGreaterEqual, /// (float a, float b) -> bool - LogicalFIsNan, /// (float a) -> bool + LogicalFOrdLessThan, /// (float a, float b) -> bool + LogicalFOrdEqual, /// (float a, float b) -> bool + LogicalFOrdLessEqual, /// (float a, float b) -> bool + LogicalFOrdGreaterThan, /// (float a, float b) -> bool + LogicalFOrdNotEqual, /// (float a, float b) -> bool + LogicalFOrdGreaterEqual, /// (float a, float b) -> bool + LogicalFOrdered, /// (float a, float b) -> bool + LogicalFUnordered, /// (float a, float b) -> bool + LogicalFUnordLessThan, /// (float a, float b) -> bool + LogicalFUnordEqual, /// (float a, float b) -> bool + LogicalFUnordLessEqual, /// (float a, float b) -> bool + LogicalFUnordGreaterThan, /// (float a, float b) -> bool + LogicalFUnordNotEqual, /// (float a, float b) -> bool + LogicalFUnordGreaterEqual, /// (float a, float b) -> bool LogicalILessThan, /// (int a, int b) -> bool LogicalIEqual, /// (int a, int b) -> bool @@ -219,9 +226,16 @@ enum class OperationCode { VoteEqual, /// (bool) -> bool ThreadId, /// () -> uint + ThreadEqMask, /// () -> uint + ThreadGeMask, /// () -> uint + ThreadGtMask, /// () -> uint + ThreadLeMask, /// () -> uint + ThreadLtMask, /// () -> uint ShuffleIndexed, /// (uint value, uint index) -> uint - MemoryBarrierGL, /// () -> void + Barrier, /// () -> void + MemoryBarrierGroup, /// () -> void + MemoryBarrierGlobal, /// () -> void Amount, }; @@ -261,10 +275,11 @@ using Node = std::shared_ptr<NodeData>; using Node4 = std::array<Node, 4>; using NodeBlock = std::vector<Node>; -class BindlessSamplerNode; -class ArraySamplerNode; +struct ArraySamplerNode; +struct BindlessSamplerNode; +struct SeparateSamplerNode; -using TrackSamplerData = std::variant<BindlessSamplerNode, ArraySamplerNode>; +using TrackSamplerData = std::variant<BindlessSamplerNode, SeparateSamplerNode, ArraySamplerNode>; using TrackSampler = std::shared_ptr<TrackSamplerData>; struct Sampler { @@ -274,63 +289,51 @@ struct Sampler { : index{index}, offset{offset}, type{type}, is_array{is_array}, is_shadow{is_shadow}, is_buffer{is_buffer}, is_indexed{is_indexed} {} + /// Separate sampler constructor + constexpr explicit Sampler(u32 index, std::pair<u32, u32> offsets, std::pair<u32, u32> buffers, + Tegra::Shader::TextureType type, bool is_array, bool is_shadow, + bool is_buffer) + : index{index}, offset{offsets.first}, secondary_offset{offsets.second}, + buffer{buffers.first}, secondary_buffer{buffers.second}, type{type}, is_array{is_array}, + is_shadow{is_shadow}, is_buffer{is_buffer}, is_separated{true} {} + /// Bindless samplers constructor constexpr explicit Sampler(u32 index, u32 offset, u32 buffer, Tegra::Shader::TextureType type, bool is_array, bool is_shadow, bool is_buffer, bool is_indexed) : index{index}, offset{offset}, buffer{buffer}, type{type}, is_array{is_array}, is_shadow{is_shadow}, is_buffer{is_buffer}, is_bindless{true}, is_indexed{is_indexed} {} - u32 index = 0; ///< Emulated index given for the this sampler. - u32 offset = 0; ///< Offset in the const buffer from where the sampler is being read. - u32 buffer = 0; ///< Buffer where the bindless sampler is being read (unused on bound samplers). - u32 size = 1; ///< Size of the sampler. + u32 index = 0; ///< Emulated index given for the this sampler. + u32 offset = 0; ///< Offset in the const buffer from where the sampler is being read. + u32 secondary_offset = 0; ///< Secondary offset in the const buffer. + u32 buffer = 0; ///< Buffer where the bindless sampler is read. + u32 secondary_buffer = 0; ///< Secondary buffer where the bindless sampler is read. + u32 size = 1; ///< Size of the sampler. Tegra::Shader::TextureType type{}; ///< The type used to sample this texture (Texture2D, etc) - bool is_array = false; ///< Whether the texture is being sampled as an array texture or not. - bool is_shadow = false; ///< Whether the texture is being sampled as a depth texture or not. - bool is_buffer = false; ///< Whether the texture is a texture buffer without sampler. - bool is_bindless = false; ///< Whether this sampler belongs to a bindless texture or not. - bool is_indexed = false; ///< Whether this sampler is an indexed array of textures. + bool is_array = false; ///< Whether the texture is being sampled as an array texture or not. + bool is_shadow = false; ///< Whether the texture is being sampled as a depth texture or not. + bool is_buffer = false; ///< Whether the texture is a texture buffer without sampler. + bool is_bindless = false; ///< Whether this sampler belongs to a bindless texture or not. + bool is_indexed = false; ///< Whether this sampler is an indexed array of textures. + bool is_separated = false; ///< Whether the image and sampler is separated or not. }; /// Represents a tracked bindless sampler into a direct const buffer -class ArraySamplerNode final { -public: - explicit ArraySamplerNode(u32 index, u32 base_offset, u32 bindless_var) - : index{index}, base_offset{base_offset}, bindless_var{bindless_var} {} - - constexpr u32 GetIndex() const { - return index; - } - - constexpr u32 GetBaseOffset() const { - return base_offset; - } - - constexpr u32 GetIndexVar() const { - return bindless_var; - } - -private: +struct ArraySamplerNode { u32 index; u32 base_offset; u32 bindless_var; }; -/// Represents a tracked bindless sampler into a direct const buffer -class BindlessSamplerNode final { -public: - explicit BindlessSamplerNode(u32 index, u32 offset) : index{index}, offset{offset} {} - - constexpr u32 GetIndex() const { - return index; - } - - constexpr u32 GetOffset() const { - return offset; - } +/// Represents a tracked separate sampler image pair that was folded statically +struct SeparateSamplerNode { + std::pair<u32, u32> indices; + std::pair<u32, u32> offsets; +}; -private: +/// Represents a tracked bindless sampler into a direct const buffer +struct BindlessSamplerNode { u32 index; u32 offset; }; diff --git a/src/video_core/shader/node_helper.h b/src/video_core/shader/node_helper.h index 11231bbea..1e0886185 100644 --- a/src/video_core/shader/node_helper.h +++ b/src/video_core/shader/node_helper.h @@ -48,7 +48,7 @@ Node MakeNode(Args&&... args) { template <typename T, typename... Args> TrackSampler MakeTrackSampler(Args&&... args) { static_assert(std::is_convertible_v<T, TrackSamplerData>); - return std::make_shared<TrackSamplerData>(T(std::forward<Args>(args)...)); + return std::make_shared<TrackSamplerData>(T{std::forward<Args>(args)...}); } template <typename... Args> diff --git a/src/video_core/shader/registry.cpp b/src/video_core/shader/registry.cpp index af70b3f35..cdf274e54 100644 --- a/src/video_core/shader/registry.cpp +++ b/src/video_core/shader/registry.cpp @@ -93,6 +93,26 @@ std::optional<SamplerDescriptor> Registry::ObtainBoundSampler(u32 offset) { return value; } +std::optional<Tegra::Engines::SamplerDescriptor> Registry::ObtainSeparateSampler( + std::pair<u32, u32> buffers, std::pair<u32, u32> offsets) { + SeparateSamplerKey key; + key.buffers = buffers; + key.offsets = offsets; + const auto iter = separate_samplers.find(key); + if (iter != separate_samplers.end()) { + return iter->second; + } + if (!engine) { + return std::nullopt; + } + + const u32 handle_1 = engine->AccessConstBuffer32(stage, key.buffers.first, key.offsets.first); + const u32 handle_2 = engine->AccessConstBuffer32(stage, key.buffers.second, key.offsets.second); + const SamplerDescriptor value = engine->AccessSampler(handle_1 | handle_2); + separate_samplers.emplace(key, value); + return value; +} + std::optional<Tegra::Engines::SamplerDescriptor> Registry::ObtainBindlessSampler(u32 buffer, u32 offset) { const std::pair key = {buffer, offset}; diff --git a/src/video_core/shader/registry.h b/src/video_core/shader/registry.h index 0c80d35fd..231206765 100644 --- a/src/video_core/shader/registry.h +++ b/src/video_core/shader/registry.h @@ -19,8 +19,39 @@ namespace VideoCommon::Shader { +struct SeparateSamplerKey { + std::pair<u32, u32> buffers; + std::pair<u32, u32> offsets; +}; + +} // namespace VideoCommon::Shader + +namespace std { + +template <> +struct hash<VideoCommon::Shader::SeparateSamplerKey> { + std::size_t operator()(const VideoCommon::Shader::SeparateSamplerKey& key) const noexcept { + return std::hash<u32>{}(key.buffers.first ^ key.buffers.second ^ key.offsets.first ^ + key.offsets.second); + } +}; + +template <> +struct equal_to<VideoCommon::Shader::SeparateSamplerKey> { + bool operator()(const VideoCommon::Shader::SeparateSamplerKey& lhs, + const VideoCommon::Shader::SeparateSamplerKey& rhs) const noexcept { + return lhs.buffers == rhs.buffers && lhs.offsets == rhs.offsets; + } +}; + +} // namespace std + +namespace VideoCommon::Shader { + using KeyMap = std::unordered_map<std::pair<u32, u32>, u32, Common::PairHash>; using BoundSamplerMap = std::unordered_map<u32, Tegra::Engines::SamplerDescriptor>; +using SeparateSamplerMap = + std::unordered_map<SeparateSamplerKey, Tegra::Engines::SamplerDescriptor>; using BindlessSamplerMap = std::unordered_map<std::pair<u32, u32>, Tegra::Engines::SamplerDescriptor, Common::PairHash>; @@ -73,6 +104,9 @@ public: std::optional<Tegra::Engines::SamplerDescriptor> ObtainBoundSampler(u32 offset); + std::optional<Tegra::Engines::SamplerDescriptor> ObtainSeparateSampler( + std::pair<u32, u32> buffers, std::pair<u32, u32> offsets); + std::optional<Tegra::Engines::SamplerDescriptor> ObtainBindlessSampler(u32 buffer, u32 offset); /// Inserts a key. @@ -128,6 +162,7 @@ private: Tegra::Engines::ConstBufferEngineInterface* engine = nullptr; KeyMap keys; BoundSamplerMap bound_samplers; + SeparateSamplerMap separate_samplers; BindlessSamplerMap bindless_samplers; u32 bound_buffer; GraphicsInfo graphics_info; diff --git a/src/video_core/shader/shader_ir.cpp b/src/video_core/shader/shader_ir.cpp index 822674926..e322c3402 100644 --- a/src/video_core/shader/shader_ir.cpp +++ b/src/video_core/shader/shader_ir.cpp @@ -10,6 +10,7 @@ #include "common/common_types.h" #include "common/logging/log.h" #include "video_core/engines/shader_bytecode.h" +#include "video_core/shader/node.h" #include "video_core/shader/node_helper.h" #include "video_core/shader/registry.h" #include "video_core/shader/shader_ir.h" @@ -243,56 +244,44 @@ Node ShaderIR::GetSaturatedHalfFloat(Node value, bool saturate) { } Node ShaderIR::GetPredicateComparisonFloat(PredCondition condition, Node op_a, Node op_b) { + if (condition == PredCondition::T) { + return GetPredicate(true); + } else if (condition == PredCondition::F) { + return GetPredicate(false); + } + static constexpr std::array comparison_table{ - std::pair{PredCondition::LessThan, OperationCode::LogicalFLessThan}, - std::pair{PredCondition::Equal, OperationCode::LogicalFEqual}, - std::pair{PredCondition::LessEqual, OperationCode::LogicalFLessEqual}, - std::pair{PredCondition::GreaterThan, OperationCode::LogicalFGreaterThan}, - std::pair{PredCondition::NotEqual, OperationCode::LogicalFNotEqual}, - std::pair{PredCondition::GreaterEqual, OperationCode::LogicalFGreaterEqual}, - std::pair{PredCondition::LessThanWithNan, OperationCode::LogicalFLessThan}, - std::pair{PredCondition::NotEqualWithNan, OperationCode::LogicalFNotEqual}, - std::pair{PredCondition::LessEqualWithNan, OperationCode::LogicalFLessEqual}, - std::pair{PredCondition::GreaterThanWithNan, OperationCode::LogicalFGreaterThan}, - std::pair{PredCondition::GreaterEqualWithNan, OperationCode::LogicalFGreaterEqual}, + OperationCode(0), + OperationCode::LogicalFOrdLessThan, // LT + OperationCode::LogicalFOrdEqual, // EQ + OperationCode::LogicalFOrdLessEqual, // LE + OperationCode::LogicalFOrdGreaterThan, // GT + OperationCode::LogicalFOrdNotEqual, // NE + OperationCode::LogicalFOrdGreaterEqual, // GE + OperationCode::LogicalFOrdered, // NUM + OperationCode::LogicalFUnordered, // NAN + OperationCode::LogicalFUnordLessThan, // LTU + OperationCode::LogicalFUnordEqual, // EQU + OperationCode::LogicalFUnordLessEqual, // LEU + OperationCode::LogicalFUnordGreaterThan, // GTU + OperationCode::LogicalFUnordNotEqual, // NEU + OperationCode::LogicalFUnordGreaterEqual, // GEU }; + const std::size_t index = static_cast<std::size_t>(condition); + ASSERT_MSG(index < std::size(comparison_table), "Invalid condition={}", index); - const auto comparison = - std::find_if(comparison_table.cbegin(), comparison_table.cend(), - [condition](const auto entry) { return condition == entry.first; }); - UNIMPLEMENTED_IF_MSG(comparison == comparison_table.cend(), - "Unknown predicate comparison operation"); - - Node predicate = Operation(comparison->second, NO_PRECISE, op_a, op_b); - - if (condition == PredCondition::LessThanWithNan || - condition == PredCondition::NotEqualWithNan || - condition == PredCondition::LessEqualWithNan || - condition == PredCondition::GreaterThanWithNan || - condition == PredCondition::GreaterEqualWithNan) { - predicate = Operation(OperationCode::LogicalOr, predicate, - Operation(OperationCode::LogicalFIsNan, op_a)); - predicate = Operation(OperationCode::LogicalOr, predicate, - Operation(OperationCode::LogicalFIsNan, op_b)); - } - - return predicate; + return Operation(comparison_table[index], op_a, op_b); } Node ShaderIR::GetPredicateComparisonInteger(PredCondition condition, bool is_signed, Node op_a, Node op_b) { static constexpr std::array comparison_table{ - std::pair{PredCondition::LessThan, OperationCode::LogicalILessThan}, - std::pair{PredCondition::Equal, OperationCode::LogicalIEqual}, - std::pair{PredCondition::LessEqual, OperationCode::LogicalILessEqual}, - std::pair{PredCondition::GreaterThan, OperationCode::LogicalIGreaterThan}, - std::pair{PredCondition::NotEqual, OperationCode::LogicalINotEqual}, - std::pair{PredCondition::GreaterEqual, OperationCode::LogicalIGreaterEqual}, - std::pair{PredCondition::LessThanWithNan, OperationCode::LogicalILessThan}, - std::pair{PredCondition::NotEqualWithNan, OperationCode::LogicalINotEqual}, - std::pair{PredCondition::LessEqualWithNan, OperationCode::LogicalILessEqual}, - std::pair{PredCondition::GreaterThanWithNan, OperationCode::LogicalIGreaterThan}, - std::pair{PredCondition::GreaterEqualWithNan, OperationCode::LogicalIGreaterEqual}, + std::pair{PredCondition::LT, OperationCode::LogicalILessThan}, + std::pair{PredCondition::EQ, OperationCode::LogicalIEqual}, + std::pair{PredCondition::LE, OperationCode::LogicalILessEqual}, + std::pair{PredCondition::GT, OperationCode::LogicalIGreaterThan}, + std::pair{PredCondition::NE, OperationCode::LogicalINotEqual}, + std::pair{PredCondition::GE, OperationCode::LogicalIGreaterEqual}, }; const auto comparison = @@ -301,32 +290,24 @@ Node ShaderIR::GetPredicateComparisonInteger(PredCondition condition, bool is_si UNIMPLEMENTED_IF_MSG(comparison == comparison_table.cend(), "Unknown predicate comparison operation"); - Node predicate = SignedOperation(comparison->second, is_signed, NO_PRECISE, std::move(op_a), - std::move(op_b)); - - UNIMPLEMENTED_IF_MSG(condition == PredCondition::LessThanWithNan || - condition == PredCondition::NotEqualWithNan || - condition == PredCondition::LessEqualWithNan || - condition == PredCondition::GreaterThanWithNan || - condition == PredCondition::GreaterEqualWithNan, - "NaN comparisons for integers are not implemented"); - return predicate; + return SignedOperation(comparison->second, is_signed, NO_PRECISE, std::move(op_a), + std::move(op_b)); } Node ShaderIR::GetPredicateComparisonHalf(Tegra::Shader::PredCondition condition, Node op_a, Node op_b) { static constexpr std::array comparison_table{ - std::pair{PredCondition::LessThan, OperationCode::Logical2HLessThan}, - std::pair{PredCondition::Equal, OperationCode::Logical2HEqual}, - std::pair{PredCondition::LessEqual, OperationCode::Logical2HLessEqual}, - std::pair{PredCondition::GreaterThan, OperationCode::Logical2HGreaterThan}, - std::pair{PredCondition::NotEqual, OperationCode::Logical2HNotEqual}, - std::pair{PredCondition::GreaterEqual, OperationCode::Logical2HGreaterEqual}, - std::pair{PredCondition::LessThanWithNan, OperationCode::Logical2HLessThanWithNan}, - std::pair{PredCondition::NotEqualWithNan, OperationCode::Logical2HNotEqualWithNan}, - std::pair{PredCondition::LessEqualWithNan, OperationCode::Logical2HLessEqualWithNan}, - std::pair{PredCondition::GreaterThanWithNan, OperationCode::Logical2HGreaterThanWithNan}, - std::pair{PredCondition::GreaterEqualWithNan, OperationCode::Logical2HGreaterEqualWithNan}, + std::pair{PredCondition::LT, OperationCode::Logical2HLessThan}, + std::pair{PredCondition::EQ, OperationCode::Logical2HEqual}, + std::pair{PredCondition::LE, OperationCode::Logical2HLessEqual}, + std::pair{PredCondition::GT, OperationCode::Logical2HGreaterThan}, + std::pair{PredCondition::NE, OperationCode::Logical2HNotEqual}, + std::pair{PredCondition::GE, OperationCode::Logical2HGreaterEqual}, + std::pair{PredCondition::LTU, OperationCode::Logical2HLessThanWithNan}, + std::pair{PredCondition::LEU, OperationCode::Logical2HLessEqualWithNan}, + std::pair{PredCondition::GTU, OperationCode::Logical2HGreaterThanWithNan}, + std::pair{PredCondition::NEU, OperationCode::Logical2HNotEqualWithNan}, + std::pair{PredCondition::GEU, OperationCode::Logical2HGreaterEqualWithNan}, }; const auto comparison = @@ -397,7 +378,7 @@ void ShaderIR::SetInternalFlagsFromFloat(NodeBlock& bb, Node value, bool sets_cc if (!sets_cc) { return; } - Node zerop = Operation(OperationCode::LogicalFEqual, std::move(value), Immediate(0.0f)); + Node zerop = Operation(OperationCode::LogicalFOrdEqual, std::move(value), Immediate(0.0f)); SetInternalFlag(bb, InternalFlag::Zero, std::move(zerop)); LOG_WARNING(HW_GPU, "Condition codes implementation is incomplete"); } diff --git a/src/video_core/shader/shader_ir.h b/src/video_core/shader/shader_ir.h index 15ae152f2..3a98b2104 100644 --- a/src/video_core/shader/shader_ir.h +++ b/src/video_core/shader/shader_ir.h @@ -330,8 +330,8 @@ private: OperationCode GetPredicateCombiner(Tegra::Shader::PredOperation operation); /// Queries the missing sampler info from the execution context. - SamplerInfo GetSamplerInfo(SamplerInfo info, u32 offset, - std::optional<u32> buffer = std::nullopt); + SamplerInfo GetSamplerInfo(SamplerInfo info, + std::optional<Tegra::Engines::SamplerDescriptor> sampler); /// Accesses a texture sampler. std::optional<Sampler> GetSampler(Tegra::Shader::Sampler sampler, SamplerInfo info); @@ -409,8 +409,14 @@ private: std::tuple<Node, u32, u32> TrackCbuf(Node tracked, const NodeBlock& code, s64 cursor) const; - std::tuple<Node, TrackSampler> TrackBindlessSampler(Node tracked, const NodeBlock& code, - s64 cursor); + std::pair<Node, TrackSampler> TrackBindlessSampler(Node tracked, const NodeBlock& code, + s64 cursor); + + std::pair<Node, TrackSampler> HandleBindlessIndirectRead(const CbufNode& cbuf, + const OperationNode& operation, + Node gpr, Node base_offset, + Node tracked, const NodeBlock& code, + s64 cursor); std::optional<u32> TrackImmediate(Node tracked, const NodeBlock& code, s64 cursor) const; diff --git a/src/video_core/shader/track.cpp b/src/video_core/shader/track.cpp index eb97bfd41..d5ed81442 100644 --- a/src/video_core/shader/track.cpp +++ b/src/video_core/shader/track.cpp @@ -14,6 +14,7 @@ namespace VideoCommon::Shader { namespace { + std::pair<Node, s64> FindOperation(const NodeBlock& code, s64 cursor, OperationCode operation_code) { for (; cursor >= 0; --cursor) { @@ -63,7 +64,8 @@ bool AmendNodeCv(std::size_t amend_index, Node node) { if (const auto operation = std::get_if<OperationNode>(&*node)) { operation->SetAmendIndex(amend_index); return true; - } else if (const auto conditional = std::get_if<ConditionalNode>(&*node)) { + } + if (const auto conditional = std::get_if<ConditionalNode>(&*node)) { conditional->SetAmendIndex(amend_index); return true; } @@ -72,40 +74,27 @@ bool AmendNodeCv(std::size_t amend_index, Node node) { } // Anonymous namespace -std::tuple<Node, TrackSampler> ShaderIR::TrackBindlessSampler(Node tracked, const NodeBlock& code, - s64 cursor) { +std::pair<Node, TrackSampler> ShaderIR::TrackBindlessSampler(Node tracked, const NodeBlock& code, + s64 cursor) { if (const auto cbuf = std::get_if<CbufNode>(&*tracked)) { + const u32 cbuf_index = cbuf->GetIndex(); + // Constant buffer found, test if it's an immediate const auto& offset = cbuf->GetOffset(); if (const auto immediate = std::get_if<ImmediateNode>(&*offset)) { - auto track = - MakeTrackSampler<BindlessSamplerNode>(cbuf->GetIndex(), immediate->GetValue()); + auto track = MakeTrackSampler<BindlessSamplerNode>(cbuf_index, immediate->GetValue()); return {tracked, track}; } if (const auto operation = std::get_if<OperationNode>(&*offset)) { const u32 bound_buffer = registry.GetBoundBuffer(); - if (bound_buffer != cbuf->GetIndex()) { + if (bound_buffer != cbuf_index) { return {}; } - const auto pair = DecoupleIndirectRead(*operation); - if (!pair) { - return {}; + if (const std::optional pair = DecoupleIndirectRead(*operation)) { + auto [gpr, base_offset] = *pair; + return HandleBindlessIndirectRead(*cbuf, *operation, gpr, base_offset, tracked, + code, cursor); } - auto [gpr, base_offset] = *pair; - const auto offset_inm = std::get_if<ImmediateNode>(&*base_offset); - const auto& gpu_driver = registry.AccessGuestDriverProfile(); - const u32 bindless_cv = NewCustomVariable(); - Node op = - Operation(OperationCode::UDiv, gpr, Immediate(gpu_driver.GetTextureHandlerSize())); - - const Node cv_node = GetCustomVariable(bindless_cv); - Node amend_op = Operation(OperationCode::Assign, cv_node, std::move(op)); - const std::size_t amend_index = DeclareAmend(std::move(amend_op)); - AmendNodeCv(amend_index, code[cursor]); - // TODO Implement Bindless Index custom variable - auto track = MakeTrackSampler<ArraySamplerNode>(cbuf->GetIndex(), - offset_inm->GetValue(), bindless_cv); - return {tracked, track}; } return {}; } @@ -122,10 +111,23 @@ std::tuple<Node, TrackSampler> ShaderIR::TrackBindlessSampler(Node tracked, cons return TrackBindlessSampler(source, code, new_cursor); } if (const auto operation = std::get_if<OperationNode>(&*tracked)) { - for (std::size_t i = operation->GetOperandsCount(); i > 0; --i) { - if (auto found = TrackBindlessSampler((*operation)[i - 1], code, cursor); - std::get<0>(found)) { - // Cbuf found in operand. + const OperationNode& op = *operation; + + const OperationCode opcode = operation->GetCode(); + if (opcode == OperationCode::IBitwiseOr || opcode == OperationCode::UBitwiseOr) { + ASSERT(op.GetOperandsCount() == 2); + auto [node_a, index_a, offset_a] = TrackCbuf(op[0], code, cursor); + auto [node_b, index_b, offset_b] = TrackCbuf(op[1], code, cursor); + if (node_a && node_b) { + auto track = MakeTrackSampler<SeparateSamplerNode>(std::pair{index_a, index_b}, + std::pair{offset_a, offset_b}); + return {tracked, std::move(track)}; + } + } + std::size_t i = op.GetOperandsCount(); + while (i--) { + if (auto found = TrackBindlessSampler(op[i - 1], code, cursor); std::get<0>(found)) { + // Constant buffer found in operand. return found; } } @@ -139,6 +141,26 @@ std::tuple<Node, TrackSampler> ShaderIR::TrackBindlessSampler(Node tracked, cons return {}; } +std::pair<Node, TrackSampler> ShaderIR::HandleBindlessIndirectRead( + const CbufNode& cbuf, const OperationNode& operation, Node gpr, Node base_offset, Node tracked, + const NodeBlock& code, s64 cursor) { + const auto offset_imm = std::get<ImmediateNode>(*base_offset); + const auto& gpu_driver = registry.AccessGuestDriverProfile(); + const u32 bindless_cv = NewCustomVariable(); + const u32 texture_handler_size = gpu_driver.GetTextureHandlerSize(); + Node op = Operation(OperationCode::UDiv, gpr, Immediate(texture_handler_size)); + + Node cv_node = GetCustomVariable(bindless_cv); + Node amend_op = Operation(OperationCode::Assign, std::move(cv_node), std::move(op)); + const std::size_t amend_index = DeclareAmend(std::move(amend_op)); + AmendNodeCv(amend_index, code[cursor]); + + // TODO: Implement bindless index custom variable + auto track = + MakeTrackSampler<ArraySamplerNode>(cbuf.GetIndex(), offset_imm.GetValue(), bindless_cv); + return {tracked, track}; +} + std::tuple<Node, u32, u32> ShaderIR::TrackCbuf(Node tracked, const NodeBlock& code, s64 cursor) const { if (const auto cbuf = std::get_if<CbufNode>(&*tracked)) { diff --git a/src/video_core/shader_cache.h b/src/video_core/shader_cache.h new file mode 100644 index 000000000..2dd270e99 --- /dev/null +++ b/src/video_core/shader_cache.h @@ -0,0 +1,228 @@ +// Copyright 2020 yuzu Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include <algorithm> +#include <memory> +#include <mutex> +#include <unordered_map> +#include <utility> +#include <vector> + +#include "common/assert.h" +#include "common/common_types.h" +#include "video_core/rasterizer_interface.h" + +namespace VideoCommon { + +template <class T> +class ShaderCache { + static constexpr u64 PAGE_BITS = 14; + + struct Entry { + VAddr addr_start; + VAddr addr_end; + T* data; + + bool is_memory_marked = true; + + constexpr bool Overlaps(VAddr start, VAddr end) const noexcept { + return start < addr_end && addr_start < end; + } + }; + +public: + virtual ~ShaderCache() = default; + + /// @brief Removes shaders inside a given region + /// @note Checks for ranges + /// @param addr Start address of the invalidation + /// @param size Number of bytes of the invalidation + void InvalidateRegion(VAddr addr, std::size_t size) { + std::scoped_lock lock{invalidation_mutex}; + InvalidatePagesInRegion(addr, size); + RemovePendingShaders(); + } + + /// @brief Unmarks a memory region as cached and marks it for removal + /// @param addr Start address of the CPU write operation + /// @param size Number of bytes of the CPU write operation + void OnCPUWrite(VAddr addr, std::size_t size) { + std::lock_guard lock{invalidation_mutex}; + InvalidatePagesInRegion(addr, size); + } + + /// @brief Flushes delayed removal operations + void SyncGuestHost() { + std::scoped_lock lock{invalidation_mutex}; + RemovePendingShaders(); + } + + /// @brief Tries to obtain a cached shader starting in a given address + /// @note Doesn't check for ranges, the given address has to be the start of the shader + /// @param addr Start address of the shader, this doesn't cache for region + /// @return Pointer to a valid shader, nullptr when nothing is found + T* TryGet(VAddr addr) const { + std::scoped_lock lock{lookup_mutex}; + + const auto it = lookup_cache.find(addr); + if (it == lookup_cache.end()) { + return nullptr; + } + return it->second->data; + } + +protected: + explicit ShaderCache(VideoCore::RasterizerInterface& rasterizer_) : rasterizer{rasterizer_} {} + + /// @brief Register in the cache a given entry + /// @param data Shader to store in the cache + /// @param addr Start address of the shader that will be registered + /// @param size Size in bytes of the shader + void Register(std::unique_ptr<T> data, VAddr addr, std::size_t size) { + std::scoped_lock lock{invalidation_mutex, lookup_mutex}; + + const VAddr addr_end = addr + size; + Entry* const entry = NewEntry(addr, addr_end, data.get()); + + const u64 page_end = addr_end >> PAGE_BITS; + for (u64 page = addr >> PAGE_BITS; page <= page_end; ++page) { + invalidation_cache[page].push_back(entry); + } + + storage.push_back(std::move(data)); + + rasterizer.UpdatePagesCachedCount(addr, size, 1); + } + + /// @brief Called when a shader is going to be removed + /// @param shader Shader that will be removed + /// @pre invalidation_cache is locked + /// @pre lookup_mutex is locked + virtual void OnShaderRemoval([[maybe_unused]] T* shader) {} + +private: + /// @brief Invalidate pages in a given region + /// @pre invalidation_mutex is locked + void InvalidatePagesInRegion(VAddr addr, std::size_t size) { + const VAddr addr_end = addr + size; + const u64 page_end = addr_end >> PAGE_BITS; + for (u64 page = addr >> PAGE_BITS; page <= page_end; ++page) { + const auto it = invalidation_cache.find(page); + if (it == invalidation_cache.end()) { + continue; + } + + std::vector<Entry*>& entries = it->second; + InvalidatePageEntries(entries, addr, addr_end); + + // If there's nothing else in this page, remove it to avoid overpopulating the hash map. + if (entries.empty()) { + invalidation_cache.erase(it); + } + } + } + + /// @brief Remove shaders marked for deletion + /// @pre invalidation_mutex is locked + void RemovePendingShaders() { + if (marked_for_removal.empty()) { + return; + } + std::scoped_lock lock{lookup_mutex}; + + std::vector<T*> removed_shaders; + removed_shaders.reserve(marked_for_removal.size()); + + for (Entry* const entry : marked_for_removal) { + if (lookup_cache.erase(entry->addr_start) > 0) { + removed_shaders.push_back(entry->data); + } + } + marked_for_removal.clear(); + + if (!removed_shaders.empty()) { + RemoveShadersFromStorage(std::move(removed_shaders)); + } + } + + /// @brief Invalidates entries in a given range for the passed page + /// @param entries Vector of entries in the page, it will be modified on overlaps + /// @param addr Start address of the invalidation + /// @param addr_end Non-inclusive end address of the invalidation + /// @pre invalidation_mutex is locked + void InvalidatePageEntries(std::vector<Entry*>& entries, VAddr addr, VAddr addr_end) { + auto it = entries.begin(); + while (it != entries.end()) { + Entry* const entry = *it; + if (!entry->Overlaps(addr, addr_end)) { + ++it; + continue; + } + UnmarkMemory(entry); + marked_for_removal.push_back(entry); + + it = entries.erase(it); + } + } + + /// @brief Unmarks an entry from the rasterizer cache + /// @param entry Entry to unmark from memory + void UnmarkMemory(Entry* entry) { + if (!entry->is_memory_marked) { + return; + } + entry->is_memory_marked = false; + + const VAddr addr = entry->addr_start; + const std::size_t size = entry->addr_end - addr; + rasterizer.UpdatePagesCachedCount(addr, size, -1); + } + + /// @brief Removes a vector of shaders from a list + /// @param removed_shaders Shaders to be removed from the storage, it can contain duplicates + /// @pre invalidation_mutex is locked + /// @pre lookup_mutex is locked + void RemoveShadersFromStorage(std::vector<T*> removed_shaders) { + // Remove duplicates + std::sort(removed_shaders.begin(), removed_shaders.end()); + removed_shaders.erase(std::unique(removed_shaders.begin(), removed_shaders.end()), + removed_shaders.end()); + + // Now that there are no duplicates, we can notify removals + for (T* const shader : removed_shaders) { + OnShaderRemoval(shader); + } + + // Remove them from the cache + const auto is_removed = [&removed_shaders](std::unique_ptr<T>& shader) { + return std::find(removed_shaders.begin(), removed_shaders.end(), shader.get()) != + removed_shaders.end(); + }; + storage.erase(std::remove_if(storage.begin(), storage.end(), is_removed), storage.end()); + } + + /// @brief Creates a new entry in the lookup cache and returns its pointer + /// @pre lookup_mutex is locked + Entry* NewEntry(VAddr addr, VAddr addr_end, T* data) { + auto entry = std::make_unique<Entry>(Entry{addr, addr_end, data}); + Entry* const entry_pointer = entry.get(); + + lookup_cache.emplace(addr, std::move(entry)); + return entry_pointer; + } + + VideoCore::RasterizerInterface& rasterizer; + + mutable std::mutex lookup_mutex; + std::mutex invalidation_mutex; + + std::unordered_map<u64, std::unique_ptr<Entry>> lookup_cache; + std::unordered_map<u64, std::vector<Entry*>> invalidation_cache; + std::vector<std::unique_ptr<T>> storage; + std::vector<Entry*> marked_for_removal; +}; + +} // namespace VideoCommon diff --git a/src/video_core/surface.cpp b/src/video_core/surface.cpp index cc7181229..bbe93903c 100644 --- a/src/video_core/surface.cpp +++ b/src/video_core/surface.cpp @@ -145,6 +145,8 @@ PixelFormat PixelFormatFromRenderTargetFormat(Tegra::RenderTargetFormat format) return PixelFormat::RG8U; case Tegra::RenderTargetFormat::RG8_SNORM: return PixelFormat::RG8S; + case Tegra::RenderTargetFormat::RG8_UINT: + return PixelFormat::RG8UI; case Tegra::RenderTargetFormat::R16_FLOAT: return PixelFormat::R16F; case Tegra::RenderTargetFormat::R16_UNORM: diff --git a/src/video_core/surface.h b/src/video_core/surface.h index e0acd44d3..6da6a1b97 100644 --- a/src/video_core/surface.h +++ b/src/video_core/surface.h @@ -57,51 +57,52 @@ enum class PixelFormat { RGBA8_SRGB = 39, RG8U = 40, RG8S = 41, - RG32UI = 42, - RGBX16F = 43, - R32UI = 44, - R32I = 45, - ASTC_2D_8X8 = 46, - ASTC_2D_8X5 = 47, - ASTC_2D_5X4 = 48, - BGRA8_SRGB = 49, - DXT1_SRGB = 50, - DXT23_SRGB = 51, - DXT45_SRGB = 52, - BC7U_SRGB = 53, - R4G4B4A4U = 54, - ASTC_2D_4X4_SRGB = 55, - ASTC_2D_8X8_SRGB = 56, - ASTC_2D_8X5_SRGB = 57, - ASTC_2D_5X4_SRGB = 58, - ASTC_2D_5X5 = 59, - ASTC_2D_5X5_SRGB = 60, - ASTC_2D_10X8 = 61, - ASTC_2D_10X8_SRGB = 62, - ASTC_2D_6X6 = 63, - ASTC_2D_6X6_SRGB = 64, - ASTC_2D_10X10 = 65, - ASTC_2D_10X10_SRGB = 66, - ASTC_2D_12X12 = 67, - ASTC_2D_12X12_SRGB = 68, - ASTC_2D_8X6 = 69, - ASTC_2D_8X6_SRGB = 70, - ASTC_2D_6X5 = 71, - ASTC_2D_6X5_SRGB = 72, - E5B9G9R9F = 73, + RG8UI = 42, + RG32UI = 43, + RGBX16F = 44, + R32UI = 45, + R32I = 46, + ASTC_2D_8X8 = 47, + ASTC_2D_8X5 = 48, + ASTC_2D_5X4 = 49, + BGRA8_SRGB = 50, + DXT1_SRGB = 51, + DXT23_SRGB = 52, + DXT45_SRGB = 53, + BC7U_SRGB = 54, + R4G4B4A4U = 55, + ASTC_2D_4X4_SRGB = 56, + ASTC_2D_8X8_SRGB = 57, + ASTC_2D_8X5_SRGB = 58, + ASTC_2D_5X4_SRGB = 59, + ASTC_2D_5X5 = 60, + ASTC_2D_5X5_SRGB = 61, + ASTC_2D_10X8 = 62, + ASTC_2D_10X8_SRGB = 63, + ASTC_2D_6X6 = 64, + ASTC_2D_6X6_SRGB = 65, + ASTC_2D_10X10 = 66, + ASTC_2D_10X10_SRGB = 67, + ASTC_2D_12X12 = 68, + ASTC_2D_12X12_SRGB = 69, + ASTC_2D_8X6 = 70, + ASTC_2D_8X6_SRGB = 71, + ASTC_2D_6X5 = 72, + ASTC_2D_6X5_SRGB = 73, + E5B9G9R9F = 74, MaxColorFormat, // Depth formats - Z32F = 74, - Z16 = 75, + Z32F = 75, + Z16 = 76, MaxDepthFormat, // DepthStencil formats - Z24S8 = 76, - S8Z24 = 77, - Z32FS8 = 78, + Z24S8 = 77, + S8Z24 = 78, + Z32FS8 = 79, MaxDepthStencilFormat, @@ -171,6 +172,7 @@ constexpr std::array<u32, MaxPixelFormat> compression_factor_shift_table = {{ 0, // RGBA8_SRGB 0, // RG8U 0, // RG8S + 0, // RG8UI 0, // RG32UI 0, // RGBX16F 0, // R32UI @@ -269,6 +271,7 @@ constexpr std::array<u32, MaxPixelFormat> block_width_table = {{ 1, // RGBA8_SRGB 1, // RG8U 1, // RG8S + 1, // RG8UI 1, // RG32UI 1, // RGBX16F 1, // R32UI @@ -359,6 +362,7 @@ constexpr std::array<u32, MaxPixelFormat> block_height_table = {{ 1, // RGBA8_SRGB 1, // RG8U 1, // RG8S + 1, // RG8UI 1, // RG32UI 1, // RGBX16F 1, // R32UI @@ -449,6 +453,7 @@ constexpr std::array<u32, MaxPixelFormat> bpp_table = {{ 32, // RGBA8_SRGB 16, // RG8U 16, // RG8S + 16, // RG8UI 64, // RG32UI 64, // RGBX16F 32, // R32UI diff --git a/src/video_core/texture_cache/format_lookup_table.cpp b/src/video_core/texture_cache/format_lookup_table.cpp index 25d2ee2e8..f476f03b0 100644 --- a/src/video_core/texture_cache/format_lookup_table.cpp +++ b/src/video_core/texture_cache/format_lookup_table.cpp @@ -41,7 +41,7 @@ struct Table { ComponentType alpha_component; bool is_srgb; }; -constexpr std::array<Table, 76> DefinitionTable = {{ +constexpr std::array<Table, 78> DefinitionTable = {{ {TextureFormat::A8R8G8B8, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::ABGR8U}, {TextureFormat::A8R8G8B8, C, SNORM, SNORM, SNORM, SNORM, PixelFormat::ABGR8S}, {TextureFormat::A8R8G8B8, C, UINT, UINT, UINT, UINT, PixelFormat::ABGR8UI}, @@ -60,6 +60,7 @@ constexpr std::array<Table, 76> DefinitionTable = {{ {TextureFormat::G8R8, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::RG8U}, {TextureFormat::G8R8, C, SNORM, SNORM, SNORM, SNORM, PixelFormat::RG8S}, + {TextureFormat::G8R8, C, UINT, UINT, UINT, UINT, PixelFormat::RG8UI}, {TextureFormat::R16_G16_B16_A16, C, SNORM, SNORM, SNORM, SNORM, PixelFormat::RGBA16S}, {TextureFormat::R16_G16_B16_A16, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::RGBA16U}, @@ -97,6 +98,7 @@ constexpr std::array<Table, 76> DefinitionTable = {{ {TextureFormat::ZF32, C, FLOAT, FLOAT, FLOAT, FLOAT, PixelFormat::Z32F}, {TextureFormat::Z16, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::Z16}, {TextureFormat::S8Z24, C, UINT, UNORM, UNORM, UNORM, PixelFormat::S8Z24}, + {TextureFormat::G24R8, C, UINT, UNORM, UNORM, UNORM, PixelFormat::S8Z24}, {TextureFormat::ZF32_X24S8, C, FLOAT, UINT, UNORM, UNORM, PixelFormat::Z32FS8}, {TextureFormat::DXT1, C, UNORM, UNORM, UNORM, UNORM, PixelFormat::DXT1}, diff --git a/src/video_core/texture_cache/surface_base.cpp b/src/video_core/texture_cache/surface_base.cpp index 715f39d0d..0caf3b4f0 100644 --- a/src/video_core/texture_cache/surface_base.cpp +++ b/src/video_core/texture_cache/surface_base.cpp @@ -120,6 +120,9 @@ std::optional<std::pair<u32, u32>> SurfaceBaseImpl::GetLayerMipmap( } const auto relative_address{static_cast<GPUVAddr>(candidate_gpu_addr - gpu_addr)}; const auto layer{static_cast<u32>(relative_address / layer_size)}; + if (layer >= params.depth) { + return {}; + } const GPUVAddr mipmap_address = relative_address - layer_size * layer; const auto mipmap_it = Common::BinaryFind(mipmap_offsets.begin(), mipmap_offsets.end(), mipmap_address); @@ -248,12 +251,11 @@ void SurfaceBaseImpl::FlushBuffer(Tegra::MemoryManager& memory_manager, // Use an extra temporal buffer auto& tmp_buffer = staging_cache.GetBuffer(1); - // Special case for 3D Texture Segments - const bool must_read_current_data = - params.block_depth > 0 && params.target == VideoCore::Surface::SurfaceTarget::Texture2D; tmp_buffer.resize(guest_memory_size); host_ptr = tmp_buffer.data(); - if (must_read_current_data) { + + if (params.target == SurfaceTarget::Texture3D) { + // Special case for 3D texture segments memory_manager.ReadBlockUnsafe(gpu_addr, host_ptr, guest_memory_size); } diff --git a/src/video_core/texture_cache/surface_base.h b/src/video_core/texture_cache/surface_base.h index 79e10ffbb..173f2edba 100644 --- a/src/video_core/texture_cache/surface_base.h +++ b/src/video_core/texture_cache/surface_base.h @@ -217,8 +217,8 @@ public: } bool IsProtected() const { - // Only 3D Slices are to be protected - return is_target && params.block_depth > 0; + // Only 3D slices are to be protected + return is_target && params.target == SurfaceTarget::Texture3D; } bool IsRenderTarget() const { @@ -250,6 +250,11 @@ public: return GetView(ViewParams(overview_params.target, 0, num_layers, 0, params.num_levels)); } + TView Emplace3DView(u32 slice, u32 depth, u32 base_level, u32 num_levels) { + return GetView(ViewParams(VideoCore::Surface::SurfaceTarget::Texture3D, slice, depth, + base_level, num_levels)); + } + std::optional<TView> EmplaceIrregularView(const SurfaceParams& view_params, const GPUVAddr view_addr, const std::size_t candidate_size, const u32 mipmap, @@ -272,8 +277,8 @@ public: std::optional<TView> EmplaceView(const SurfaceParams& view_params, const GPUVAddr view_addr, const std::size_t candidate_size) { if (params.target == SurfaceTarget::Texture3D || - (params.num_levels == 1 && !params.is_layered) || - view_params.target == SurfaceTarget::Texture3D) { + view_params.target == SurfaceTarget::Texture3D || + (params.num_levels == 1 && !params.is_layered)) { return {}; } const auto layer_mipmap{GetLayerMipmap(view_addr)}; diff --git a/src/video_core/texture_cache/surface_params.cpp b/src/video_core/texture_cache/surface_params.cpp index 884fabffe..0b2b2b8c4 100644 --- a/src/video_core/texture_cache/surface_params.cpp +++ b/src/video_core/texture_cache/surface_params.cpp @@ -215,10 +215,19 @@ SurfaceParams SurfaceParams::CreateForFramebuffer(Core::System& system, std::siz params.num_levels = 1; params.emulated_levels = 1; - const bool is_layered = config.layers > 1 && params.block_depth == 0; - params.is_layered = is_layered; - params.depth = is_layered ? config.layers.Value() : 1; - params.target = is_layered ? SurfaceTarget::Texture2DArray : SurfaceTarget::Texture2D; + if (config.memory_layout.is_3d != 0) { + params.depth = config.layers.Value(); + params.is_layered = false; + params.target = SurfaceTarget::Texture3D; + } else if (config.layers > 1) { + params.depth = config.layers.Value(); + params.is_layered = true; + params.target = SurfaceTarget::Texture2DArray; + } else { + params.depth = 1; + params.is_layered = false; + params.target = SurfaceTarget::Texture2D; + } return params; } @@ -237,7 +246,7 @@ SurfaceParams SurfaceParams::CreateForFermiCopySurface( params.width = config.width; params.height = config.height; params.pitch = config.pitch; - // TODO(Rodrigo): Try to guess the surface target from depth and layer parameters + // TODO(Rodrigo): Try to guess texture arrays from parameters params.target = SurfaceTarget::Texture2D; params.depth = 1; params.num_levels = 1; diff --git a/src/video_core/texture_cache/texture_cache.h b/src/video_core/texture_cache/texture_cache.h index d6efc34b2..6207d8dfe 100644 --- a/src/video_core/texture_cache/texture_cache.h +++ b/src/video_core/texture_cache/texture_cache.h @@ -14,6 +14,7 @@ #include <unordered_map> #include <vector> +#include <boost/container/small_vector.hpp> #include <boost/icl/interval_map.hpp> #include <boost/range/iterator_range.hpp> @@ -23,6 +24,7 @@ #include "core/core.h" #include "core/memory.h" #include "core/settings.h" +#include "video_core/compatible_formats.h" #include "video_core/dirty_flags.h" #include "video_core/engines/fermi_2d.h" #include "video_core/engines/maxwell_3d.h" @@ -46,13 +48,14 @@ class RasterizerInterface; namespace VideoCommon { +using VideoCore::Surface::FormatCompatibility; using VideoCore::Surface::PixelFormat; - using VideoCore::Surface::SurfaceTarget; using RenderTargetConfig = Tegra::Engines::Maxwell3D::Regs::RenderTargetConfig; template <typename TSurface, typename TView> class TextureCache { + using VectorSurface = boost::container::small_vector<TSurface, 1>; public: void InvalidateRegion(VAddr addr, std::size_t size) { @@ -296,30 +299,30 @@ public: const GPUVAddr src_gpu_addr = src_config.Address(); const GPUVAddr dst_gpu_addr = dst_config.Address(); DeduceBestBlit(src_params, dst_params, src_gpu_addr, dst_gpu_addr); - const std::optional<VAddr> dst_cpu_addr = - system.GPU().MemoryManager().GpuToCpuAddress(dst_gpu_addr); - const std::optional<VAddr> src_cpu_addr = - system.GPU().MemoryManager().GpuToCpuAddress(src_gpu_addr); - std::pair<TSurface, TView> dst_surface = - GetSurface(dst_gpu_addr, *dst_cpu_addr, dst_params, true, false); - std::pair<TSurface, TView> src_surface = - GetSurface(src_gpu_addr, *src_cpu_addr, src_params, true, false); - ImageBlit(src_surface.second, dst_surface.second, copy_config); + + const auto& memory_manager = system.GPU().MemoryManager(); + const std::optional<VAddr> dst_cpu_addr = memory_manager.GpuToCpuAddress(dst_gpu_addr); + const std::optional<VAddr> src_cpu_addr = memory_manager.GpuToCpuAddress(src_gpu_addr); + std::pair dst_surface = GetSurface(dst_gpu_addr, *dst_cpu_addr, dst_params, true, false); + TView src_surface = GetSurface(src_gpu_addr, *src_cpu_addr, src_params, true, false).second; + ImageBlit(src_surface, dst_surface.second, copy_config); dst_surface.first->MarkAsModified(true, Tick()); } - TSurface TryFindFramebufferSurface(VAddr addr) { + TSurface TryFindFramebufferSurface(VAddr addr) const { if (!addr) { return nullptr; } const VAddr page = addr >> registry_page_bits; - std::vector<TSurface>& list = registry[page]; - for (auto& surface : list) { - if (surface->GetCpuAddr() == addr) { - return surface; - } + const auto it = registry.find(page); + if (it == registry.end()) { + return nullptr; } - return nullptr; + const auto& list = it->second; + const auto found = std::find_if(list.begin(), list.end(), [addr](const auto& surface) { + return surface->GetCpuAddr() == addr; + }); + return found != list.end() ? *found : nullptr; } u64 Tick() { @@ -498,18 +501,18 @@ private: * @param untopological Indicates to the recycler that the texture has no way * to match the overlaps due to topological reasons. **/ - RecycleStrategy PickStrategy(std::vector<TSurface>& overlaps, const SurfaceParams& params, + RecycleStrategy PickStrategy(VectorSurface& overlaps, const SurfaceParams& params, const GPUVAddr gpu_addr, const MatchTopologyResult untopological) { if (Settings::IsGPULevelExtreme()) { return RecycleStrategy::Flush; } // 3D Textures decision - if (params.block_depth > 1 || params.target == SurfaceTarget::Texture3D) { + if (params.target == SurfaceTarget::Texture3D) { return RecycleStrategy::Flush; } for (const auto& s : overlaps) { const auto& s_params = s->GetSurfaceParams(); - if (s_params.block_depth > 1 || s_params.target == SurfaceTarget::Texture3D) { + if (s_params.target == SurfaceTarget::Texture3D) { return RecycleStrategy::Flush; } } @@ -538,9 +541,8 @@ private: * @param untopological Indicates to the recycler that the texture has no way to match the * overlaps due to topological reasons. **/ - std::pair<TSurface, TView> RecycleSurface(std::vector<TSurface>& overlaps, - const SurfaceParams& params, const GPUVAddr gpu_addr, - const bool preserve_contents, + std::pair<TSurface, TView> RecycleSurface(VectorSurface& overlaps, const SurfaceParams& params, + const GPUVAddr gpu_addr, const bool preserve_contents, const MatchTopologyResult untopological) { const bool do_load = preserve_contents && Settings::IsGPULevelExtreme(); for (auto& surface : overlaps) { @@ -594,7 +596,7 @@ private: } else { new_surface = GetUncachedSurface(gpu_addr, params); } - const auto& final_params = new_surface->GetSurfaceParams(); + const SurfaceParams& final_params = new_surface->GetSurfaceParams(); if (cr_params.type != final_params.type) { if (Settings::IsGPULevelExtreme()) { BufferCopy(current_surface, new_surface); @@ -602,7 +604,7 @@ private: } else { std::vector<CopyParams> bricks = current_surface->BreakDown(final_params); for (auto& brick : bricks) { - ImageCopy(current_surface, new_surface, brick); + TryCopyImage(current_surface, new_surface, brick); } } Unregister(current_surface); @@ -650,47 +652,65 @@ private: * @param params The parameters on the new surface. * @param gpu_addr The starting address of the new surface. **/ - std::optional<std::pair<TSurface, TView>> TryReconstructSurface(std::vector<TSurface>& overlaps, + std::optional<std::pair<TSurface, TView>> TryReconstructSurface(VectorSurface& overlaps, const SurfaceParams& params, - const GPUVAddr gpu_addr) { + GPUVAddr gpu_addr) { if (params.target == SurfaceTarget::Texture3D) { - return {}; + return std::nullopt; } - bool modified = false; + const auto test_modified = [](TSurface& surface) { return surface->IsModified(); }; TSurface new_surface = GetUncachedSurface(gpu_addr, params); - u32 passed_tests = 0; + + if (std::none_of(overlaps.begin(), overlaps.end(), test_modified)) { + LoadSurface(new_surface); + for (const auto& surface : overlaps) { + Unregister(surface); + } + Register(new_surface); + return {{new_surface, new_surface->GetMainView()}}; + } + + std::size_t passed_tests = 0; for (auto& surface : overlaps) { const SurfaceParams& src_params = surface->GetSurfaceParams(); - if (src_params.is_layered || src_params.num_levels > 1) { - // We send this cases to recycle as they are more complex to handle - return {}; - } - const std::size_t candidate_size = surface->GetSizeInBytes(); - auto mipmap_layer{new_surface->GetLayerMipmap(surface->GetGpuAddr())}; + const auto mipmap_layer{new_surface->GetLayerMipmap(surface->GetGpuAddr())}; if (!mipmap_layer) { continue; } - const auto [layer, mipmap] = *mipmap_layer; - if (new_surface->GetMipmapSize(mipmap) != candidate_size) { + const auto [base_layer, base_mipmap] = *mipmap_layer; + if (new_surface->GetMipmapSize(base_mipmap) != surface->GetMipmapSize(0)) { continue; } - modified |= surface->IsModified(); - // Now we got all the data set up - const u32 width = SurfaceParams::IntersectWidth(src_params, params, 0, mipmap); - const u32 height = SurfaceParams::IntersectHeight(src_params, params, 0, mipmap); - const CopyParams copy_params(0, 0, 0, 0, 0, layer, 0, mipmap, width, height, 1); - passed_tests++; - ImageCopy(surface, new_surface, copy_params); + ++passed_tests; + + // Copy all mipmaps and layers + const u32 block_width = params.GetDefaultBlockWidth(); + const u32 block_height = params.GetDefaultBlockHeight(); + for (u32 mipmap = base_mipmap; mipmap < base_mipmap + src_params.num_levels; ++mipmap) { + const u32 width = SurfaceParams::IntersectWidth(src_params, params, 0, mipmap); + const u32 height = SurfaceParams::IntersectHeight(src_params, params, 0, mipmap); + if (width < block_width || height < block_height) { + // Current APIs forbid copying small compressed textures, avoid errors + break; + } + const CopyParams copy_params(0, 0, 0, 0, 0, base_layer, 0, mipmap, width, height, + src_params.depth); + TryCopyImage(surface, new_surface, copy_params); + } } if (passed_tests == 0) { - return {}; + return std::nullopt; + } + if (Settings::IsGPULevelExtreme() && passed_tests != overlaps.size()) { // In Accurate GPU all tests should pass, else we recycle - } else if (Settings::IsGPULevelExtreme() && passed_tests != overlaps.size()) { - return {}; + return std::nullopt; } + + const bool modified = std::any_of(overlaps.begin(), overlaps.end(), test_modified); for (const auto& surface : overlaps) { Unregister(surface); } + new_surface->MarkAsModified(modified, Tick()); Register(new_surface); return {{new_surface, new_surface->GetMainView()}}; @@ -708,53 +728,11 @@ private: * @param preserve_contents Indicates that the new surface should be loaded from memory or * left blank. */ - std::optional<std::pair<TSurface, TView>> Manage3DSurfaces(std::vector<TSurface>& overlaps, + std::optional<std::pair<TSurface, TView>> Manage3DSurfaces(VectorSurface& overlaps, const SurfaceParams& params, - const GPUVAddr gpu_addr, - const VAddr cpu_addr, + GPUVAddr gpu_addr, VAddr cpu_addr, bool preserve_contents) { - if (params.target == SurfaceTarget::Texture3D) { - bool failed = false; - if (params.num_levels > 1) { - // We can't handle mipmaps in 3D textures yet, better fallback to LLE approach - return std::nullopt; - } - TSurface new_surface = GetUncachedSurface(gpu_addr, params); - bool modified = false; - for (auto& surface : overlaps) { - const SurfaceParams& src_params = surface->GetSurfaceParams(); - if (src_params.target != SurfaceTarget::Texture2D) { - failed = true; - break; - } - if (src_params.height != params.height) { - failed = true; - break; - } - if (src_params.block_depth != params.block_depth || - src_params.block_height != params.block_height) { - failed = true; - break; - } - const u32 offset = static_cast<u32>(surface->GetCpuAddr() - cpu_addr); - const auto offsets = params.GetBlockOffsetXYZ(offset); - const auto z = std::get<2>(offsets); - modified |= surface->IsModified(); - const CopyParams copy_params(0, 0, 0, 0, 0, z, 0, 0, params.width, params.height, - 1); - ImageCopy(surface, new_surface, copy_params); - } - if (failed) { - return std::nullopt; - } - for (const auto& surface : overlaps) { - Unregister(surface); - } - new_surface->MarkAsModified(modified, Tick()); - Register(new_surface); - auto view = new_surface->GetMainView(); - return {{std::move(new_surface), view}}; - } else { + if (params.target != SurfaceTarget::Texture3D) { for (const auto& surface : overlaps) { if (!surface->MatchTarget(params.target)) { if (overlaps.size() == 1 && surface->GetCpuAddr() == cpu_addr) { @@ -770,11 +748,60 @@ private: continue; } if (surface->MatchesStructure(params) == MatchStructureResult::FullMatch) { - return {{surface, surface->GetMainView()}}; + return std::make_pair(surface, surface->GetMainView()); } } return InitializeSurface(gpu_addr, params, preserve_contents); } + + if (params.num_levels > 1) { + // We can't handle mipmaps in 3D textures yet, better fallback to LLE approach + return std::nullopt; + } + + if (overlaps.size() == 1) { + const auto& surface = overlaps[0]; + const SurfaceParams& overlap_params = surface->GetSurfaceParams(); + // Don't attempt to render to textures with more than one level for now + // The texture has to be to the right or the sample address if we want to render to it + if (overlap_params.num_levels == 1 && cpu_addr >= surface->GetCpuAddr()) { + const u32 offset = static_cast<u32>(cpu_addr - surface->GetCpuAddr()); + const u32 slice = std::get<2>(params.GetBlockOffsetXYZ(offset)); + if (slice < overlap_params.depth) { + auto view = surface->Emplace3DView(slice, params.depth, 0, 1); + return std::make_pair(std::move(surface), std::move(view)); + } + } + } + + TSurface new_surface = GetUncachedSurface(gpu_addr, params); + bool modified = false; + + for (auto& surface : overlaps) { + const SurfaceParams& src_params = surface->GetSurfaceParams(); + if (src_params.target != SurfaceTarget::Texture2D || + src_params.height != params.height || + src_params.block_depth != params.block_depth || + src_params.block_height != params.block_height) { + return std::nullopt; + } + modified |= surface->IsModified(); + + const u32 offset = static_cast<u32>(surface->GetCpuAddr() - cpu_addr); + const u32 slice = std::get<2>(params.GetBlockOffsetXYZ(offset)); + const u32 width = params.width; + const u32 height = params.height; + const CopyParams copy_params(0, 0, 0, 0, 0, slice, 0, 0, width, height, 1); + TryCopyImage(surface, new_surface, copy_params); + } + for (const auto& surface : overlaps) { + Unregister(surface); + } + new_surface->MarkAsModified(modified, Tick()); + Register(new_surface); + + TView view = new_surface->GetMainView(); + return std::make_pair(std::move(new_surface), std::move(view)); } /** @@ -810,7 +837,7 @@ private: TSurface& current_surface = iter->second; const auto topological_result = current_surface->MatchesTopology(params); if (topological_result != MatchTopologyResult::FullMatch) { - std::vector<TSurface> overlaps{current_surface}; + VectorSurface overlaps{current_surface}; return RecycleSurface(overlaps, params, gpu_addr, preserve_contents, topological_result); } @@ -852,7 +879,7 @@ private: } } - // Check if it's a 3D texture + // Manage 3D textures if (params.block_depth > 0) { auto surface = Manage3DSurfaces(overlaps, params, gpu_addr, cpu_addr, preserve_contents); @@ -868,12 +895,9 @@ private: // two things either the candidate surface is a supertexture of the overlap // or they don't match in any known way. if (!current_surface->IsInside(gpu_addr, gpu_addr + candidate_size)) { - if (current_surface->GetGpuAddr() == gpu_addr) { - std::optional<std::pair<TSurface, TView>> view = - TryReconstructSurface(overlaps, params, gpu_addr); - if (view) { - return *view; - } + const std::optional view = TryReconstructSurface(overlaps, params, gpu_addr); + if (view) { + return *view; } return RecycleSurface(overlaps, params, gpu_addr, preserve_contents, MatchTopologyResult::FullMatch); @@ -991,7 +1015,9 @@ private: params.target = target; params.is_tiled = false; params.srgb_conversion = false; - params.is_layered = false; + params.is_layered = + target == SurfaceTarget::Texture1DArray || target == SurfaceTarget::Texture2DArray || + target == SurfaceTarget::TextureCubemap || target == SurfaceTarget::TextureCubeArray; params.block_width = 0; params.block_height = 0; params.block_depth = 0; @@ -1028,7 +1054,7 @@ private: void DeduceBestBlit(SurfaceParams& src_params, SurfaceParams& dst_params, const GPUVAddr src_gpu_addr, const GPUVAddr dst_gpu_addr) { auto deduced_src = DeduceSurface(src_gpu_addr, src_params); - auto deduced_dst = DeduceSurface(src_gpu_addr, src_params); + auto deduced_dst = DeduceSurface(dst_gpu_addr, dst_params); if (deduced_src.Failed() || deduced_dst.Failed()) { return; } @@ -1124,23 +1150,25 @@ private: } } - std::vector<TSurface> GetSurfacesInRegion(const VAddr cpu_addr, const std::size_t size) { + VectorSurface GetSurfacesInRegion(const VAddr cpu_addr, const std::size_t size) { if (size == 0) { return {}; } const VAddr cpu_addr_end = cpu_addr + size; - VAddr start = cpu_addr >> registry_page_bits; const VAddr end = (cpu_addr_end - 1) >> registry_page_bits; - std::vector<TSurface> surfaces; - while (start <= end) { - std::vector<TSurface>& list = registry[start]; - for (auto& surface : list) { - if (!surface->IsPicked() && surface->Overlaps(cpu_addr, cpu_addr_end)) { - surface->MarkAsPicked(true); - surfaces.push_back(surface); + VectorSurface surfaces; + for (VAddr start = cpu_addr >> registry_page_bits; start <= end; ++start) { + const auto it = registry.find(start); + if (it == registry.end()) { + continue; + } + for (auto& surface : it->second) { + if (surface->IsPicked() || !surface->Overlaps(cpu_addr, cpu_addr_end)) { + continue; } + surface->MarkAsPicked(true); + surfaces.push_back(surface); } - start++; } for (auto& surface : surfaces) { surface->MarkAsPicked(false); @@ -1165,6 +1193,19 @@ private: return {}; } + /// Try to do an image copy logging when formats are incompatible. + void TryCopyImage(TSurface& src, TSurface& dst, const CopyParams& copy) { + const SurfaceParams& src_params = src->GetSurfaceParams(); + const SurfaceParams& dst_params = dst->GetSurfaceParams(); + if (!format_compatibility.TestCopy(src_params.pixel_format, dst_params.pixel_format)) { + LOG_ERROR(HW_GPU, "Illegal copy between formats={{{}, {}}}", + static_cast<int>(dst_params.pixel_format), + static_cast<int>(src_params.pixel_format)); + return; + } + ImageCopy(src, dst, copy); + } + constexpr PixelFormat GetSiblingFormat(PixelFormat format) const { return siblings_table[static_cast<std::size_t>(format)]; } @@ -1214,6 +1255,7 @@ private: VideoCore::RasterizerInterface& rasterizer; FormatLookupTable format_lookup_table; + FormatCompatibility format_compatibility; u64 ticks{}; diff --git a/src/web_service/CMakeLists.txt b/src/web_service/CMakeLists.txt index 0c9bb0d55..06ab7c59d 100644 --- a/src/web_service/CMakeLists.txt +++ b/src/web_service/CMakeLists.txt @@ -8,4 +8,4 @@ add_library(web_service STATIC ) create_target_directory_groups(web_service) -target_link_libraries(web_service PRIVATE common json-headers httplib lurlparser) +target_link_libraries(web_service PRIVATE common nlohmann_json::nlohmann_json httplib lurlparser) diff --git a/src/web_service/telemetry_json.cpp b/src/web_service/telemetry_json.cpp index 7538389bf..7a480e33c 100644 --- a/src/web_service/telemetry_json.cpp +++ b/src/web_service/telemetry_json.cpp @@ -2,7 +2,7 @@ // Licensed under GPLv2 or any later version // Refer to the license.txt file included. -#include <json.hpp> +#include <nlohmann/json.hpp> #include "common/detached_tasks.h" #include "common/web_result.h" #include "web_service/telemetry_json.h" diff --git a/src/web_service/verify_login.cpp b/src/web_service/verify_login.cpp index ca4b43b93..bfaa5b70a 100644 --- a/src/web_service/verify_login.cpp +++ b/src/web_service/verify_login.cpp @@ -2,7 +2,7 @@ // Licensed under GPLv2 or any later version // Refer to the license.txt file included. -#include <json.hpp> +#include <nlohmann/json.hpp> #include "common/web_result.h" #include "web_service/verify_login.h" #include "web_service/web_backend.h" diff --git a/src/yuzu/CMakeLists.txt b/src/yuzu/CMakeLists.txt index 8b9404718..75c27e39e 100644 --- a/src/yuzu/CMakeLists.txt +++ b/src/yuzu/CMakeLists.txt @@ -208,6 +208,10 @@ if (MSVC) copy_yuzu_unicorn_deps(yuzu) endif() +if (NOT APPLE) + target_compile_definitions(yuzu PRIVATE HAS_OPENGL) +endif() + if (ENABLE_VULKAN) target_include_directories(yuzu PRIVATE ../../externals/Vulkan-Headers/include) target_compile_definitions(yuzu PRIVATE HAS_VULKAN) diff --git a/src/yuzu/bootmanager.cpp b/src/yuzu/bootmanager.cpp index 3d759f77b..4bfce48a4 100644 --- a/src/yuzu/bootmanager.cpp +++ b/src/yuzu/bootmanager.cpp @@ -8,13 +8,16 @@ #include <QHBoxLayout> #include <QKeyEvent> #include <QMessageBox> -#include <QOffscreenSurface> -#include <QOpenGLContext> #include <QPainter> #include <QScreen> #include <QStringList> #include <QWindow> +#ifdef HAS_OPENGL +#include <QOffscreenSurface> +#include <QOpenGLContext> +#endif + #if !defined(WIN32) && HAS_VULKAN #include <qpa/qplatformnativeinterface.h> #endif @@ -41,49 +44,65 @@ EmuThread::EmuThread() = default; EmuThread::~EmuThread() = default; void EmuThread::run() { - MicroProfileOnThreadCreate("EmuThread"); + std::string name = "yuzu:EmuControlThread"; + MicroProfileOnThreadCreate(name.c_str()); + Common::SetCurrentThreadName(name.c_str()); + + auto& system = Core::System::GetInstance(); + + system.RegisterHostThread(); + + auto& gpu = system.GPU(); // Main process has been loaded. Make the context current to this thread and begin GPU and CPU // execution. - Core::System::GetInstance().GPU().Start(); + gpu.Start(); + + gpu.ObtainContext(); emit LoadProgress(VideoCore::LoadCallbackStage::Prepare, 0, 0); - Core::System::GetInstance().Renderer().Rasterizer().LoadDiskResources( + system.Renderer().Rasterizer().LoadDiskResources( stop_run, [this](VideoCore::LoadCallbackStage stage, std::size_t value, std::size_t total) { emit LoadProgress(stage, value, total); }); emit LoadProgress(VideoCore::LoadCallbackStage::Complete, 0, 0); + gpu.ReleaseContext(); + // Holds whether the cpu was running during the last iteration, // so that the DebugModeLeft signal can be emitted before the // next execution step bool was_active = false; while (!stop_run) { if (running) { - if (!was_active) + if (was_active) { emit DebugModeLeft(); + } - Core::System::ResultStatus result = Core::System::GetInstance().RunLoop(); + running_guard = true; + Core::System::ResultStatus result = system.Run(); + if (result != Core::System::ResultStatus::Success) { + running_guard = false; + this->SetRunning(false); + emit ErrorThrown(result, system.GetStatusDetails()); + } + running_wait.Wait(); + result = system.Pause(); if (result != Core::System::ResultStatus::Success) { + running_guard = false; this->SetRunning(false); - emit ErrorThrown(result, Core::System::GetInstance().GetStatusDetails()); + emit ErrorThrown(result, system.GetStatusDetails()); } + running_guard = false; - was_active = running || exec_step; - if (!was_active && !stop_run) + if (!stop_run) { + was_active = true; emit DebugModeEntered(); + } } else if (exec_step) { - if (!was_active) - emit DebugModeLeft(); - - exec_step = false; - Core::System::GetInstance().SingleStep(); - emit DebugModeEntered(); - yieldCurrentThread(); - - was_active = false; + UNIMPLEMENTED(); } else { std::unique_lock lock{running_mutex}; running_cv.wait(lock, [this] { return IsRunning() || exec_step || stop_run; }); @@ -91,13 +110,14 @@ void EmuThread::run() { } // Shutdown the core emulation - Core::System::GetInstance().Shutdown(); + system.Shutdown(); #if MICROPROFILE_ENABLED MicroProfileOnThreadExit(); #endif } +#ifdef HAS_OPENGL class OpenGLSharedContext : public Core::Frontend::GraphicsContext { public: /// Create the original context that should be shared from @@ -106,6 +126,9 @@ public: format.setVersion(4, 3); format.setProfile(QSurfaceFormat::CompatibilityProfile); format.setOption(QSurfaceFormat::FormatOption::DeprecatedFunctions); + if (Settings::values.renderer_debug) { + format.setOption(QSurfaceFormat::FormatOption::DebugContext); + } // TODO: expose a setting for buffer value (ie default/single/double/triple) format.setSwapBehavior(QSurfaceFormat::DefaultSwapBehavior); format.setSwapInterval(0); @@ -150,18 +173,19 @@ public: } void MakeCurrent() override { - if (is_current) { - return; + // We can't track the current state of the underlying context in this wrapper class because + // Qt may make the underlying context not current for one reason or another. In particular, + // the WebBrowser uses GL, so it seems to conflict if we aren't careful. + // Instead of always just making the context current (which does not have any caching to + // check if the underlying context is already current) we can check for the current context + // in the thread local data by calling `currentContext()` and checking if its ours. + if (QOpenGLContext::currentContext() != context.get()) { + context->makeCurrent(surface); } - is_current = context->makeCurrent(surface); } void DoneCurrent() override { - if (!is_current) { - return; - } context->doneCurrent(); - is_current = false; } QOpenGLContext* GetShareContext() { @@ -178,8 +202,8 @@ private: std::unique_ptr<QOpenGLContext> context; std::unique_ptr<QOffscreenSurface> offscreen_surface{}; QSurface* surface; - bool is_current = false; }; +#endif class DummyContext : public Core::Frontend::GraphicsContext {}; @@ -352,7 +376,7 @@ QByteArray GRenderWindow::saveGeometry() { } qreal GRenderWindow::windowPixelRatio() const { - return devicePixelRatio(); + return devicePixelRatioF(); } std::pair<u32, u32> GRenderWindow::ScaleTouch(const QPointF& pos) const { @@ -470,6 +494,7 @@ void GRenderWindow::resizeEvent(QResizeEvent* event) { } std::unique_ptr<Core::Frontend::GraphicsContext> GRenderWindow::CreateSharedContext() const { +#ifdef HAS_OPENGL if (Settings::values.renderer_backend == Settings::RendererBackend::OpenGL) { auto c = static_cast<OpenGLSharedContext*>(main_context.get()); // Bind the shared contexts to the main surface in case the backend wants to take over @@ -477,6 +502,7 @@ std::unique_ptr<Core::Frontend::GraphicsContext> GRenderWindow::CreateSharedCont return std::make_unique<OpenGLSharedContext>(c->GetShareContext(), child_widget->windowHandle()); } +#endif return std::make_unique<DummyContext>(); } @@ -557,6 +583,7 @@ void GRenderWindow::OnMinimalClientAreaChangeRequest(std::pair<u32, u32> minimal } bool GRenderWindow::InitializeOpenGL() { +#ifdef HAS_OPENGL // TODO: One of these flags might be interesting: WA_OpaquePaintEvent, WA_NoBackground, // WA_DontShowOnScreen, WA_DeleteOnClose auto child = new OpenGLRenderWidget(this); @@ -568,6 +595,11 @@ bool GRenderWindow::InitializeOpenGL() { std::make_unique<OpenGLSharedContext>(context->GetShareContext(), child->windowHandle())); return true; +#else + QMessageBox::warning(this, tr("OpenGL not available!"), + tr("yuzu has not been compiled with OpenGL support.")); + return false; +#endif } bool GRenderWindow::InitializeVulkan() { diff --git a/src/yuzu/bootmanager.h b/src/yuzu/bootmanager.h index 3626604ca..6c59b4d5c 100644 --- a/src/yuzu/bootmanager.h +++ b/src/yuzu/bootmanager.h @@ -59,6 +59,12 @@ public: this->running = running; lock.unlock(); running_cv.notify_all(); + if (!running) { + running_wait.Set(); + /// Wait until effectively paused + while (running_guard) + ; + } } /** @@ -84,6 +90,8 @@ private: std::atomic_bool stop_run{false}; std::mutex running_mutex; std::condition_variable running_cv; + Common::Event running_wait{}; + std::atomic_bool running_guard{false}; signals: /** diff --git a/src/yuzu/configuration/config.cpp b/src/yuzu/configuration/config.cpp index 75c6cf20b..bbbd96113 100644 --- a/src/yuzu/configuration/config.cpp +++ b/src/yuzu/configuration/config.cpp @@ -211,7 +211,7 @@ const std::array<int, Settings::NativeKeyboard::NumKeyboardMods> Config::default // This must be in alphabetical order according to action name as it must have the same order as // UISetting::values.shortcuts, which is alphabetically ordered. // clang-format off -const std::array<UISettings::Shortcut, 15> Config::default_hotkeys{{ +const std::array<UISettings::Shortcut, 16> Config::default_hotkeys{{ {QStringLiteral("Capture Screenshot"), QStringLiteral("Main Window"), {QStringLiteral("Ctrl+P"), Qt::ApplicationShortcut}}, {QStringLiteral("Change Docked Mode"), QStringLiteral("Main Window"), {QStringLiteral("F10"), Qt::ApplicationShortcut}}, {QStringLiteral("Continue/Pause Emulation"), QStringLiteral("Main Window"), {QStringLiteral("F4"), Qt::WindowShortcut}}, @@ -222,6 +222,7 @@ const std::array<UISettings::Shortcut, 15> Config::default_hotkeys{{ {QStringLiteral("Increase Speed Limit"), QStringLiteral("Main Window"), {QStringLiteral("+"), Qt::ApplicationShortcut}}, {QStringLiteral("Load Amiibo"), QStringLiteral("Main Window"), {QStringLiteral("F2"), Qt::ApplicationShortcut}}, {QStringLiteral("Load File"), QStringLiteral("Main Window"), {QStringLiteral("Ctrl+O"), Qt::WindowShortcut}}, + {QStringLiteral("Mute Audio"), QStringLiteral("Main Window"), {QStringLiteral("Ctrl+M"), Qt::WindowShortcut}}, {QStringLiteral("Restart Emulation"), QStringLiteral("Main Window"), {QStringLiteral("F6"), Qt::WindowShortcut}}, {QStringLiteral("Stop Emulation"), QStringLiteral("Main Window"), {QStringLiteral("F5"), Qt::WindowShortcut}}, {QStringLiteral("Toggle Filter Bar"), QStringLiteral("Main Window"), {QStringLiteral("Ctrl+F"), Qt::WindowShortcut}}, @@ -533,6 +534,8 @@ void Config::ReadDebuggingValues() { Settings::values.quest_flag = ReadSetting(QStringLiteral("quest_flag"), false).toBool(); Settings::values.disable_cpu_opt = ReadSetting(QStringLiteral("disable_cpu_opt"), false).toBool(); + Settings::values.disable_macro_jit = + ReadSetting(QStringLiteral("disable_macro_jit"), false).toBool(); qt_config->endGroup(); } @@ -629,13 +632,11 @@ void Config::ReadRendererValues() { static_cast<Settings::RendererBackend>(ReadSetting(QStringLiteral("backend"), 0).toInt()); Settings::values.renderer_debug = ReadSetting(QStringLiteral("debug"), false).toBool(); Settings::values.vulkan_device = ReadSetting(QStringLiteral("vulkan_device"), 0).toInt(); - Settings::values.resolution_factor = - ReadSetting(QStringLiteral("resolution_factor"), 1.0).toFloat(); Settings::values.aspect_ratio = ReadSetting(QStringLiteral("aspect_ratio"), 0).toInt(); Settings::values.max_anisotropy = ReadSetting(QStringLiteral("max_anisotropy"), 0).toInt(); Settings::values.use_frame_limit = ReadSetting(QStringLiteral("use_frame_limit"), true).toBool(); - Settings::values.frame_limit = ReadSetting(QStringLiteral("frame_limit"), 100).toInt(); + Settings::values.frame_limit = ReadSetting(QStringLiteral("frame_limit"), 100).toUInt(); Settings::values.use_disk_shader_cache = ReadSetting(QStringLiteral("use_disk_shader_cache"), true).toBool(); const int gpu_accuracy_level = ReadSetting(QStringLiteral("gpu_accuracy"), 0).toInt(); @@ -643,6 +644,8 @@ void Config::ReadRendererValues() { Settings::values.use_asynchronous_gpu_emulation = ReadSetting(QStringLiteral("use_asynchronous_gpu_emulation"), false).toBool(); Settings::values.use_vsync = ReadSetting(QStringLiteral("use_vsync"), true).toBool(); + Settings::values.use_assembly_shaders = + ReadSetting(QStringLiteral("use_assembly_shaders"), false).toBool(); Settings::values.use_fast_gpu_time = ReadSetting(QStringLiteral("use_fast_gpu_time"), true).toBool(); Settings::values.force_30fps_mode = @@ -687,6 +690,8 @@ void Config::ReadSystemValues() { Settings::values.region_index = ReadSetting(QStringLiteral("region_index"), 1).toInt(); + Settings::values.time_zone_index = ReadSetting(QStringLiteral("time_zone_index"), 0).toInt(); + const auto rng_seed_enabled = ReadSetting(QStringLiteral("rng_seed_enabled"), false).toBool(); if (rng_seed_enabled) { Settings::values.rng_seed = ReadSetting(QStringLiteral("rng_seed"), 0).toULongLong(); @@ -716,8 +721,6 @@ void Config::ReadUIValues() { .toString(); UISettings::values.enable_discord_presence = ReadSetting(QStringLiteral("enable_discord_presence"), true).toBool(); - UISettings::values.screenshot_resolution_factor = - static_cast<u16>(ReadSetting(QStringLiteral("screenshot_resolution_factor"), 0).toUInt()); UISettings::values.select_user_on_boot = ReadSetting(QStringLiteral("select_user_on_boot"), false).toBool(); @@ -1007,6 +1010,7 @@ void Config::SaveDebuggingValues() { WriteSetting(QStringLiteral("dump_nso"), Settings::values.dump_nso, false); WriteSetting(QStringLiteral("quest_flag"), Settings::values.quest_flag, false); WriteSetting(QStringLiteral("disable_cpu_opt"), Settings::values.disable_cpu_opt, false); + WriteSetting(QStringLiteral("disable_macro_jit"), Settings::values.disable_macro_jit, false); qt_config->endGroup(); } @@ -1075,8 +1079,6 @@ void Config::SaveRendererValues() { WriteSetting(QStringLiteral("backend"), static_cast<int>(Settings::values.renderer_backend), 0); WriteSetting(QStringLiteral("debug"), Settings::values.renderer_debug, false); WriteSetting(QStringLiteral("vulkan_device"), Settings::values.vulkan_device, 0); - WriteSetting(QStringLiteral("resolution_factor"), - static_cast<double>(Settings::values.resolution_factor), 1.0); WriteSetting(QStringLiteral("aspect_ratio"), Settings::values.aspect_ratio, 0); WriteSetting(QStringLiteral("max_anisotropy"), Settings::values.max_anisotropy, 0); WriteSetting(QStringLiteral("use_frame_limit"), Settings::values.use_frame_limit, true); @@ -1088,6 +1090,8 @@ void Config::SaveRendererValues() { WriteSetting(QStringLiteral("use_asynchronous_gpu_emulation"), Settings::values.use_asynchronous_gpu_emulation, false); WriteSetting(QStringLiteral("use_vsync"), Settings::values.use_vsync, true); + WriteSetting(QStringLiteral("use_assembly_shaders"), Settings::values.use_assembly_shaders, + false); WriteSetting(QStringLiteral("use_fast_gpu_time"), Settings::values.use_fast_gpu_time, true); WriteSetting(QStringLiteral("force_30fps_mode"), Settings::values.force_30fps_mode, false); @@ -1126,6 +1130,7 @@ void Config::SaveSystemValues() { WriteSetting(QStringLiteral("current_user"), Settings::values.current_user, 0); WriteSetting(QStringLiteral("language_index"), Settings::values.language_index, 1); WriteSetting(QStringLiteral("region_index"), Settings::values.region_index, 1); + WriteSetting(QStringLiteral("time_zone_index"), Settings::values.time_zone_index, 0); WriteSetting(QStringLiteral("rng_seed_enabled"), Settings::values.rng_seed.has_value(), false); WriteSetting(QStringLiteral("rng_seed"), Settings::values.rng_seed.value_or(0), 0); @@ -1149,8 +1154,6 @@ void Config::SaveUIValues() { QString::fromUtf8(UISettings::themes[0].second)); WriteSetting(QStringLiteral("enable_discord_presence"), UISettings::values.enable_discord_presence, true); - WriteSetting(QStringLiteral("screenshot_resolution_factor"), - UISettings::values.screenshot_resolution_factor, 0); WriteSetting(QStringLiteral("select_user_on_boot"), UISettings::values.select_user_on_boot, false); diff --git a/src/yuzu/configuration/config.h b/src/yuzu/configuration/config.h index 5cd2a5feb..09316382c 100644 --- a/src/yuzu/configuration/config.h +++ b/src/yuzu/configuration/config.h @@ -27,7 +27,7 @@ public: default_mouse_buttons; static const std::array<int, Settings::NativeKeyboard::NumKeyboardKeys> default_keyboard_keys; static const std::array<int, Settings::NativeKeyboard::NumKeyboardMods> default_keyboard_mods; - static const std::array<UISettings::Shortcut, 15> default_hotkeys; + static const std::array<UISettings::Shortcut, 16> default_hotkeys; private: void ReadValues(); diff --git a/src/yuzu/configuration/configure_debug.cpp b/src/yuzu/configuration/configure_debug.cpp index c2026763e..2c77441fd 100644 --- a/src/yuzu/configuration/configure_debug.cpp +++ b/src/yuzu/configuration/configure_debug.cpp @@ -39,6 +39,8 @@ void ConfigureDebug::SetConfiguration() { ui->disable_cpu_opt->setChecked(Settings::values.disable_cpu_opt); ui->enable_graphics_debugging->setEnabled(!Core::System::GetInstance().IsPoweredOn()); ui->enable_graphics_debugging->setChecked(Settings::values.renderer_debug); + ui->disable_macro_jit->setEnabled(!Core::System::GetInstance().IsPoweredOn()); + ui->disable_macro_jit->setChecked(Settings::values.disable_macro_jit); } void ConfigureDebug::ApplyConfiguration() { @@ -51,6 +53,7 @@ void ConfigureDebug::ApplyConfiguration() { Settings::values.quest_flag = ui->quest_flag->isChecked(); Settings::values.disable_cpu_opt = ui->disable_cpu_opt->isChecked(); Settings::values.renderer_debug = ui->enable_graphics_debugging->isChecked(); + Settings::values.disable_macro_jit = ui->disable_macro_jit->isChecked(); Debugger::ToggleConsole(); Log::Filter filter; filter.ParseFilterString(Settings::values.log_filter); diff --git a/src/yuzu/configuration/configure_debug.ui b/src/yuzu/configuration/configure_debug.ui index e0d4c4a44..46f0208c6 100644 --- a/src/yuzu/configuration/configure_debug.ui +++ b/src/yuzu/configuration/configure_debug.ui @@ -148,6 +148,19 @@ </property> </widget> </item> + <item> + <widget class="QCheckBox" name="disable_macro_jit"> + <property name="enabled"> + <bool>true</bool> + </property> + <property name="whatsThis"> + <string>When checked, it disables the macro Just In Time compiler. Enabled this makes games run slower</string> + </property> + <property name="text"> + <string>Disable Macro JIT</string> + </property> + </widget> + </item> </layout> </widget> </item> diff --git a/src/yuzu/configuration/configure_general.cpp b/src/yuzu/configuration/configure_general.cpp index cb95423e0..74b2ad537 100644 --- a/src/yuzu/configuration/configure_general.cpp +++ b/src/yuzu/configuration/configure_general.cpp @@ -23,6 +23,11 @@ ConfigureGeneral::ConfigureGeneral(QWidget* parent) ConfigureGeneral::~ConfigureGeneral() = default; void ConfigureGeneral::SetConfiguration() { + const bool runtime_lock = !Core::System::GetInstance().IsPoweredOn(); + + ui->use_multi_core->setEnabled(runtime_lock); + ui->use_multi_core->setChecked(Settings::values.use_multi_core); + ui->toggle_check_exit->setChecked(UISettings::values.confirm_before_closing); ui->toggle_user_on_boot->setChecked(UISettings::values.select_user_on_boot); ui->toggle_background_pause->setChecked(UISettings::values.pause_when_in_background); @@ -41,6 +46,7 @@ void ConfigureGeneral::ApplyConfiguration() { Settings::values.use_frame_limit = ui->toggle_frame_limit->isChecked(); Settings::values.frame_limit = ui->frame_limit->value(); + Settings::values.use_multi_core = ui->use_multi_core->isChecked(); } void ConfigureGeneral::changeEvent(QEvent* event) { diff --git a/src/yuzu/configuration/configure_general.ui b/src/yuzu/configuration/configure_general.ui index fc3b7e65a..2711116a2 100644 --- a/src/yuzu/configuration/configure_general.ui +++ b/src/yuzu/configuration/configure_general.ui @@ -52,6 +52,13 @@ </layout> </item> <item> + <widget class="QCheckBox" name="use_multi_core"> + <property name="text"> + <string>Multicore CPU Emulation</string> + </property> + </widget> + </item> + <item> <widget class="QCheckBox" name="toggle_check_exit"> <property name="text"> <string>Confirm exit while emulation is running</string> diff --git a/src/yuzu/configuration/configure_graphics.cpp b/src/yuzu/configuration/configure_graphics.cpp index ea667caef..304625cd7 100644 --- a/src/yuzu/configuration/configure_graphics.cpp +++ b/src/yuzu/configuration/configure_graphics.cpp @@ -19,47 +19,6 @@ #include "video_core/renderer_vulkan/renderer_vulkan.h" #endif -namespace { -enum class Resolution : int { - Auto, - Scale1x, - Scale2x, - Scale3x, - Scale4x, -}; - -float ToResolutionFactor(Resolution option) { - switch (option) { - case Resolution::Auto: - return 0.f; - case Resolution::Scale1x: - return 1.f; - case Resolution::Scale2x: - return 2.f; - case Resolution::Scale3x: - return 3.f; - case Resolution::Scale4x: - return 4.f; - } - return 0.f; -} - -Resolution FromResolutionFactor(float factor) { - if (factor == 0.f) { - return Resolution::Auto; - } else if (factor == 1.f) { - return Resolution::Scale1x; - } else if (factor == 2.f) { - return Resolution::Scale2x; - } else if (factor == 3.f) { - return Resolution::Scale3x; - } else if (factor == 4.f) { - return Resolution::Scale4x; - } - return Resolution::Auto; -} -} // Anonymous namespace - ConfigureGraphics::ConfigureGraphics(QWidget* parent) : QWidget(parent), ui(new Ui::ConfigureGraphics) { vulkan_device = Settings::values.vulkan_device; @@ -99,8 +58,6 @@ void ConfigureGraphics::SetConfiguration() { ui->api->setEnabled(runtime_lock); ui->api->setCurrentIndex(static_cast<int>(Settings::values.renderer_backend)); - ui->resolution_factor_combobox->setCurrentIndex( - static_cast<int>(FromResolutionFactor(Settings::values.resolution_factor))); ui->aspect_ratio_combobox->setCurrentIndex(Settings::values.aspect_ratio); ui->use_disk_shader_cache->setEnabled(runtime_lock); ui->use_disk_shader_cache->setChecked(Settings::values.use_disk_shader_cache); @@ -114,8 +71,6 @@ void ConfigureGraphics::SetConfiguration() { void ConfigureGraphics::ApplyConfiguration() { Settings::values.renderer_backend = GetCurrentGraphicsBackend(); Settings::values.vulkan_device = vulkan_device; - Settings::values.resolution_factor = - ToResolutionFactor(static_cast<Resolution>(ui->resolution_factor_combobox->currentIndex())); Settings::values.aspect_ratio = ui->aspect_ratio_combobox->currentIndex(); Settings::values.use_disk_shader_cache = ui->use_disk_shader_cache->isChecked(); Settings::values.use_asynchronous_gpu_emulation = diff --git a/src/yuzu/configuration/configure_graphics.ui b/src/yuzu/configuration/configure_graphics.ui index c816d6108..6e75447a5 100644 --- a/src/yuzu/configuration/configure_graphics.ui +++ b/src/yuzu/configuration/configure_graphics.ui @@ -85,46 +85,6 @@ </widget> </item> <item> - <layout class="QHBoxLayout" name="horizontalLayout_2"> - <item> - <widget class="QLabel" name="label"> - <property name="text"> - <string>Internal Resolution:</string> - </property> - </widget> - </item> - <item> - <widget class="QComboBox" name="resolution_factor_combobox"> - <item> - <property name="text"> - <string>Auto (Window Size)</string> - </property> - </item> - <item> - <property name="text"> - <string>Native (1280x720)</string> - </property> - </item> - <item> - <property name="text"> - <string>2x Native (2560x1440)</string> - </property> - </item> - <item> - <property name="text"> - <string>3x Native (3840x2160)</string> - </property> - </item> - <item> - <property name="text"> - <string>4x Native (5120x2880)</string> - </property> - </item> - </widget> - </item> - </layout> - </item> - <item> <layout class="QHBoxLayout" name="horizontalLayout_6"> <item> <widget class="QLabel" name="ar_label"> diff --git a/src/yuzu/configuration/configure_graphics_advanced.cpp b/src/yuzu/configuration/configure_graphics_advanced.cpp index 5bb2ae555..be5006ad3 100644 --- a/src/yuzu/configuration/configure_graphics_advanced.cpp +++ b/src/yuzu/configuration/configure_graphics_advanced.cpp @@ -22,6 +22,8 @@ void ConfigureGraphicsAdvanced::SetConfiguration() { ui->gpu_accuracy->setCurrentIndex(static_cast<int>(Settings::values.gpu_accuracy)); ui->use_vsync->setEnabled(runtime_lock); ui->use_vsync->setChecked(Settings::values.use_vsync); + ui->use_assembly_shaders->setEnabled(runtime_lock); + ui->use_assembly_shaders->setChecked(Settings::values.use_assembly_shaders); ui->use_fast_gpu_time->setChecked(Settings::values.use_fast_gpu_time); ui->force_30fps_mode->setEnabled(runtime_lock); ui->force_30fps_mode->setChecked(Settings::values.force_30fps_mode); @@ -33,6 +35,7 @@ void ConfigureGraphicsAdvanced::ApplyConfiguration() { auto gpu_accuracy = static_cast<Settings::GPUAccuracy>(ui->gpu_accuracy->currentIndex()); Settings::values.gpu_accuracy = gpu_accuracy; Settings::values.use_vsync = ui->use_vsync->isChecked(); + Settings::values.use_assembly_shaders = ui->use_assembly_shaders->isChecked(); Settings::values.use_fast_gpu_time = ui->use_fast_gpu_time->isChecked(); Settings::values.force_30fps_mode = ui->force_30fps_mode->isChecked(); Settings::values.max_anisotropy = ui->anisotropic_filtering_combobox->currentIndex(); diff --git a/src/yuzu/configuration/configure_graphics_advanced.ui b/src/yuzu/configuration/configure_graphics_advanced.ui index 770b80c50..0021607ac 100644 --- a/src/yuzu/configuration/configure_graphics_advanced.ui +++ b/src/yuzu/configuration/configure_graphics_advanced.ui @@ -63,6 +63,16 @@ </widget> </item> <item> + <widget class="QCheckBox" name="use_assembly_shaders"> + <property name="toolTip"> + <string>Enabling this reduces shader stutter. Enables OpenGL assembly shaders on supported Nvidia devices (NV_gpu_program5 is required). This feature is experimental.</string> + </property> + <property name="text"> + <string>Use assembly shaders (experimental, Nvidia OpenGL only)</string> + </property> + </widget> + </item> + <item> <widget class="QCheckBox" name="force_30fps_mode"> <property name="text"> <string>Force 30 FPS mode</string> diff --git a/src/yuzu/configuration/configure_input_player.cpp b/src/yuzu/configuration/configure_input_player.cpp index e4eb5594b..a05fa64ba 100644 --- a/src/yuzu/configuration/configure_input_player.cpp +++ b/src/yuzu/configuration/configure_input_player.cpp @@ -480,7 +480,9 @@ void ConfigureInputPlayer::RestoreDefaults() { SetAnalogButton(params, analogs_param[analog_id], analog_sub_buttons[sub_button_id]); } } + UpdateButtonLabels(); + ApplyConfiguration(); } void ConfigureInputPlayer::ClearAll() { @@ -505,6 +507,7 @@ void ConfigureInputPlayer::ClearAll() { } UpdateButtonLabels(); + ApplyConfiguration(); } void ConfigureInputPlayer::UpdateButtonLabels() { diff --git a/src/yuzu/configuration/configure_service.cpp b/src/yuzu/configuration/configure_service.cpp index 06566e981..0de7a4f0b 100644 --- a/src/yuzu/configuration/configure_service.cpp +++ b/src/yuzu/configuration/configure_service.cpp @@ -68,6 +68,7 @@ void ConfigureService::SetConfiguration() { } std::pair<QString, QString> ConfigureService::BCATDownloadEvents() { +#ifdef YUZU_ENABLE_BOXCAT std::optional<std::string> global; std::map<std::string, Service::BCAT::EventStatus> map; const auto res = Service::BCAT::Boxcat::GetStatus(global, map); @@ -105,7 +106,10 @@ std::pair<QString, QString> ConfigureService::BCATDownloadEvents() { .arg(QString::fromStdString(key)) .arg(FormatEventStatusString(value)); } - return {QStringLiteral("Current Boxcat Events"), std::move(out)}; + return {tr("Current Boxcat Events"), std::move(out)}; +#else + return {tr("Current Boxcat Events"), tr("There are currently no events on boxcat.")}; +#endif } void ConfigureService::OnBCATImplChanged() { diff --git a/src/yuzu/configuration/configure_system.cpp b/src/yuzu/configuration/configure_system.cpp index f49cd4c8f..10315e7a6 100644 --- a/src/yuzu/configuration/configure_system.cpp +++ b/src/yuzu/configuration/configure_system.cpp @@ -57,6 +57,7 @@ void ConfigureSystem::SetConfiguration() { ui->combo_language->setCurrentIndex(Settings::values.language_index); ui->combo_region->setCurrentIndex(Settings::values.region_index); + ui->combo_time_zone->setCurrentIndex(Settings::values.time_zone_index); ui->combo_sound->setCurrentIndex(Settings::values.sound_index); ui->rng_seed_checkbox->setChecked(Settings::values.rng_seed.has_value()); @@ -84,6 +85,7 @@ void ConfigureSystem::ApplyConfiguration() { Settings::values.language_index = ui->combo_language->currentIndex(); Settings::values.region_index = ui->combo_region->currentIndex(); + Settings::values.time_zone_index = ui->combo_time_zone->currentIndex(); Settings::values.sound_index = ui->combo_sound->currentIndex(); if (ui->rng_seed_checkbox->isChecked()) { diff --git a/src/yuzu/configuration/configure_system.h b/src/yuzu/configuration/configure_system.h index d8fa2d2cc..26d42d5c5 100644 --- a/src/yuzu/configuration/configure_system.h +++ b/src/yuzu/configuration/configure_system.h @@ -37,5 +37,6 @@ private: int language_index = 0; int region_index = 0; + int time_zone_index = 0; int sound_index = 0; }; diff --git a/src/yuzu/configuration/configure_system.ui b/src/yuzu/configuration/configure_system.ui index 4e2c7e76e..9c8cca6dc 100644 --- a/src/yuzu/configuration/configure_system.ui +++ b/src/yuzu/configuration/configure_system.ui @@ -22,14 +22,14 @@ <string>System Settings</string> </property> <layout class="QGridLayout" name="gridLayout"> - <item row="2" column="0"> + <item row="3" column="0"> <widget class="QLabel" name="label_sound"> <property name="text"> <string>Sound output mode</string> </property> </widget> </item> - <item row="3" column="0"> + <item row="4" column="0"> <widget class="QLabel" name="label_console_id"> <property name="text"> <string>Console ID:</string> @@ -174,14 +174,255 @@ </item> </widget> </item> - <item row="5" column="0"> + <item row="2" column="0"> + <widget class="QLabel" name="label_timezone"> + <property name="text"> + <string>Time Zone:</string> + </property> + </widget> + </item> + <item row="2" column="1"> + <widget class="QComboBox" name="combo_time_zone"> + <item> + <property name="text"> + <string>Auto</string> + </property> + </item> + <item> + <property name="text"> + <string>Default</string> + </property> + </item> + <item> + <property name="text"> + <string>CET</string> + </property> + </item> + <item> + <property name="text"> + <string>CST6CDT</string> + </property> + </item> + <item> + <property name="text"> + <string>Cuba</string> + </property> + </item> + <item> + <property name="text"> + <string>EET</string> + </property> + </item> + <item> + <property name="text"> + <string>Egypt</string> + </property> + </item> + <item> + <property name="text"> + <string>Eire</string> + </property> + </item> + <item> + <property name="text"> + <string>EST</string> + </property> + </item> + <item> + <property name="text"> + <string>EST5EDT</string> + </property> + </item> + <item> + <property name="text"> + <string>GB</string> + </property> + </item> + <item> + <property name="text"> + <string>GB-Eire</string> + </property> + </item> + <item> + <property name="text"> + <string>GMT</string> + </property> + </item> + <item> + <property name="text"> + <string>GMT+0</string> + </property> + </item> + <item> + <property name="text"> + <string>GMT-0</string> + </property> + </item> + <item> + <property name="text"> + <string>GMT0</string> + </property> + </item> + <item> + <property name="text"> + <string>Greenwich</string> + </property> + </item> + <item> + <property name="text"> + <string>Hongkong</string> + </property> + </item> + <item> + <property name="text"> + <string>HST</string> + </property> + </item> + <item> + <property name="text"> + <string>Iceland</string> + </property> + </item> + <item> + <property name="text"> + <string>Iran</string> + </property> + </item> + <item> + <property name="text"> + <string>Israel</string> + </property> + </item> + <item> + <property name="text"> + <string>Jamaica</string> + </property> + </item> + <item> + <property name="text"> + <string>Japan</string> + </property> + </item> + <item> + <property name="text"> + <string>Kwajalein</string> + </property> + </item> + <item> + <property name="text"> + <string>Libya</string> + </property> + </item> + <item> + <property name="text"> + <string>MET</string> + </property> + </item> + <item> + <property name="text"> + <string>MST</string> + </property> + </item> + <item> + <property name="text"> + <string>MST7MDT</string> + </property> + </item> + <item> + <property name="text"> + <string>Navajo</string> + </property> + </item> + <item> + <property name="text"> + <string>NZ</string> + </property> + </item> + <item> + <property name="text"> + <string>NZ-CHAT</string> + </property> + </item> + <item> + <property name="text"> + <string>Poland</string> + </property> + </item> + <item> + <property name="text"> + <string>Portugal</string> + </property> + </item> + <item> + <property name="text"> + <string>PRC</string> + </property> + </item> + <item> + <property name="text"> + <string>PST8PDT</string> + </property> + </item> + <item> + <property name="text"> + <string>ROC</string> + </property> + </item> + <item> + <property name="text"> + <string>ROK</string> + </property> + </item> + <item> + <property name="text"> + <string>Singapore</string> + </property> + </item> + <item> + <property name="text"> + <string>Turkey</string> + </property> + </item> + <item> + <property name="text"> + <string>UCT</string> + </property> + </item> + <item> + <property name="text"> + <string>Universal</string> + </property> + </item> + <item> + <property name="text"> + <string>UTC</string> + </property> + </item> + <item> + <property name="text"> + <string>W-SU</string> + </property> + </item> + <item> + <property name="text"> + <string>WET</string> + </property> + </item> + <item> + <property name="text"> + <string>Zulu</string> + </property> + </item> + </widget> + </item> + <item row="6" column="0"> <widget class="QCheckBox" name="rng_seed_checkbox"> <property name="text"> <string>RNG Seed</string> </property> </widget> </item> - <item row="2" column="1"> + <item row="3" column="1"> <widget class="QComboBox" name="combo_sound"> <item> <property name="text"> @@ -207,7 +448,7 @@ </property> </widget> </item> - <item row="3" column="1"> + <item row="4" column="1"> <widget class="QPushButton" name="button_regenerate_console_id"> <property name="sizePolicy"> <sizepolicy hsizetype="Fixed" vsizetype="Fixed"> @@ -223,14 +464,14 @@ </property> </widget> </item> - <item row="4" column="0"> + <item row="5" column="0"> <widget class="QCheckBox" name="custom_rtc_checkbox"> <property name="text"> <string>Custom RTC</string> </property> </widget> </item> - <item row="4" column="1"> + <item row="5" column="1"> <widget class="QDateTimeEdit" name="custom_rtc_edit"> <property name="minimumDate"> <date> @@ -244,7 +485,7 @@ </property> </widget> </item> - <item row="5" column="1"> + <item row="6" column="1"> <widget class="QLineEdit" name="rng_seed_edit"> <property name="sizePolicy"> <sizepolicy hsizetype="Minimum" vsizetype="Fixed"> diff --git a/src/yuzu/debugger/wait_tree.cpp b/src/yuzu/debugger/wait_tree.cpp index c1ea25fb8..9bb0a0109 100644 --- a/src/yuzu/debugger/wait_tree.cpp +++ b/src/yuzu/debugger/wait_tree.cpp @@ -2,10 +2,13 @@ // Licensed under GPLv2 or any later version // Refer to the license.txt file included. +#include <fmt/format.h> + #include "yuzu/debugger/wait_tree.h" #include "yuzu/util/util.h" #include "common/assert.h" +#include "core/arm/arm_interface.h" #include "core/core.h" #include "core/hle/kernel/handle_table.h" #include "core/hle/kernel/mutex.h" @@ -59,8 +62,10 @@ std::vector<std::unique_ptr<WaitTreeThread>> WaitTreeItem::MakeThreadItemList() std::size_t row = 0; auto add_threads = [&](const std::vector<std::shared_ptr<Kernel::Thread>>& threads) { for (std::size_t i = 0; i < threads.size(); ++i) { - item_list.push_back(std::make_unique<WaitTreeThread>(*threads[i])); - item_list.back()->row = row; + if (!threads[i]->IsHLEThread()) { + item_list.push_back(std::make_unique<WaitTreeThread>(*threads[i])); + item_list.back()->row = row; + } ++row; } }; @@ -114,20 +119,21 @@ QString WaitTreeCallstack::GetText() const { std::vector<std::unique_ptr<WaitTreeItem>> WaitTreeCallstack::GetChildren() const { std::vector<std::unique_ptr<WaitTreeItem>> list; - constexpr std::size_t BaseRegister = 29; - auto& memory = Core::System::GetInstance().Memory(); - u64 base_pointer = thread.GetContext64().cpu_registers[BaseRegister]; + if (thread.IsHLEThread()) { + return list; + } - while (base_pointer != 0) { - const u64 lr = memory.Read64(base_pointer + sizeof(u64)); - if (lr == 0) { - break; - } + if (thread.GetOwnerProcess() == nullptr || !thread.GetOwnerProcess()->Is64BitProcess()) { + return list; + } - list.push_back(std::make_unique<WaitTreeText>( - tr("0x%1").arg(lr - sizeof(u32), 16, 16, QLatin1Char{'0'}))); + auto backtrace = Core::ARM_Interface::GetBacktraceFromContext(Core::System::GetInstance(), + thread.GetContext64()); - base_pointer = memory.Read64(base_pointer); + for (auto& entry : backtrace) { + std::string s = fmt::format("{:20}{:016X} {:016X} {:016X} {}", entry.module, entry.address, + entry.original_address, entry.offset, entry.name); + list.push_back(std::make_unique<WaitTreeText>(QString::fromStdString(s))); } return list; @@ -206,7 +212,15 @@ QString WaitTreeThread::GetText() const { status = tr("running"); break; case Kernel::ThreadStatus::Ready: - status = tr("ready"); + if (!thread.IsPaused()) { + if (thread.WasRunning()) { + status = tr("running"); + } else { + status = tr("ready"); + } + } else { + status = tr("paused"); + } break; case Kernel::ThreadStatus::Paused: status = tr("paused"); @@ -254,7 +268,15 @@ QColor WaitTreeThread::GetColor() const { case Kernel::ThreadStatus::Running: return QColor(Qt::GlobalColor::darkGreen); case Kernel::ThreadStatus::Ready: - return QColor(Qt::GlobalColor::darkBlue); + if (!thread.IsPaused()) { + if (thread.WasRunning()) { + return QColor(Qt::GlobalColor::darkGreen); + } else { + return QColor(Qt::GlobalColor::darkBlue); + } + } else { + return QColor(Qt::GlobalColor::lightGray); + } case Kernel::ThreadStatus::Paused: return QColor(Qt::GlobalColor::lightGray); case Kernel::ThreadStatus::WaitHLEEvent: @@ -319,7 +341,7 @@ std::vector<std::unique_ptr<WaitTreeItem>> WaitTreeThread::GetChildren() const { if (thread.GetStatus() == Kernel::ThreadStatus::WaitSynch) { list.push_back(std::make_unique<WaitTreeObjectList>(thread.GetSynchronizationObjects(), - thread.IsSleepingOnWait())); + thread.IsWaitingSync())); } list.push_back(std::make_unique<WaitTreeCallstack>(thread)); diff --git a/src/yuzu/discord_impl.cpp b/src/yuzu/discord_impl.cpp index ea0079353..a93733b26 100644 --- a/src/yuzu/discord_impl.cpp +++ b/src/yuzu/discord_impl.cpp @@ -18,7 +18,7 @@ DiscordImpl::DiscordImpl() { // The number is the client ID for yuzu, it's used for images and the // application name - Discord_Initialize("471872241299226636", &handlers, 1, nullptr); + Discord_Initialize("712465656758665259", &handlers, 1, nullptr); } DiscordImpl::~DiscordImpl() { diff --git a/src/yuzu/game_list.cpp b/src/yuzu/game_list.cpp index dccbabcbf..bfb600df0 100644 --- a/src/yuzu/game_list.cpp +++ b/src/yuzu/game_list.cpp @@ -488,11 +488,11 @@ void GameList::AddGamePopup(QMenu& context_menu, u64 program_id, std::string pat auto it = FindMatchingCompatibilityEntry(compatibility_list, program_id); navigate_to_gamedb_entry->setVisible(it != compatibility_list.end() && program_id != 0); - connect(open_save_location, &QAction::triggered, [this, program_id]() { - emit OpenFolderRequested(program_id, GameListOpenTarget::SaveData); + connect(open_save_location, &QAction::triggered, [this, program_id, path]() { + emit OpenFolderRequested(GameListOpenTarget::SaveData, path); }); - connect(open_lfs_location, &QAction::triggered, [this, program_id]() { - emit OpenFolderRequested(program_id, GameListOpenTarget::ModData); + connect(open_lfs_location, &QAction::triggered, [this, program_id, path]() { + emit OpenFolderRequested(GameListOpenTarget::ModData, path); }); connect(open_transferable_shader_cache, &QAction::triggered, [this, program_id]() { emit OpenTransferableShaderCacheRequested(program_id); }); diff --git a/src/yuzu/game_list.h b/src/yuzu/game_list.h index 878d94413..a38cb2fc3 100644 --- a/src/yuzu/game_list.h +++ b/src/yuzu/game_list.h @@ -73,7 +73,7 @@ public: signals: void GameChosen(QString game_path); void ShouldCancelWorker(); - void OpenFolderRequested(u64 program_id, GameListOpenTarget target); + void OpenFolderRequested(GameListOpenTarget target, const std::string& game_path); void OpenTransferableShaderCacheRequested(u64 program_id); void DumpRomFSRequested(u64 program_id, const std::string& game_path); void CopyTIDRequested(u64 program_id); diff --git a/src/yuzu/loading_screen.cpp b/src/yuzu/loading_screen.cpp index 2a6483370..ae842306c 100644 --- a/src/yuzu/loading_screen.cpp +++ b/src/yuzu/loading_screen.cpp @@ -19,6 +19,7 @@ #include <QTime> #include <QtConcurrent/QtConcurrentRun> #include "common/logging/log.h" +#include "core/frontend/framebuffer_layout.h" #include "core/loader/loader.h" #include "ui_loading_screen.h" #include "video_core/rasterizer_interface.h" @@ -61,7 +62,7 @@ LoadingScreen::LoadingScreen(QWidget* parent) : QWidget(parent), ui(std::make_unique<Ui::LoadingScreen>()), previous_stage(VideoCore::LoadCallbackStage::Complete) { ui->setupUi(this); - setMinimumSize(1280, 720); + setMinimumSize(Layout::MinimumSize::Width, Layout::MinimumSize::Height); // Create a fade out effect to hide this loading screen widget. // When fading opacity, it will fade to the parent widgets background color, which is why we diff --git a/src/yuzu/main.cpp b/src/yuzu/main.cpp index 62f9d2ab2..f586950e7 100644 --- a/src/yuzu/main.cpp +++ b/src/yuzu/main.cpp @@ -56,6 +56,7 @@ static FileSys::VirtualFile VfsDirectoryCreateFileWrapper(const FileSys::Virtual #include <QShortcut> #include <QStatusBar> #include <QSysInfo> +#include <QUrl> #include <QtConcurrent/QtConcurrent> #include <fmt/format.h> @@ -65,6 +66,7 @@ static FileSys::VirtualFile VfsDirectoryCreateFileWrapper(const FileSys::Virtual #include "common/logging/backend.h" #include "common/logging/filter.h" #include "common/logging/log.h" +#include "common/memory_detect.h" #include "common/microprofile.h" #include "common/scm_rev.h" #include "common/scope_exit.h" @@ -216,9 +218,26 @@ GMainWindow::GMainWindow() LOG_INFO(Frontend, "yuzu Version: {} | {}-{}", yuzu_build_version, Common::g_scm_branch, Common::g_scm_desc); #ifdef ARCHITECTURE_x86_64 - LOG_INFO(Frontend, "Host CPU: {}", Common::GetCPUCaps().cpu_string); + const auto& caps = Common::GetCPUCaps(); + std::string cpu_string = caps.cpu_string; + if (caps.avx || caps.avx2 || caps.avx512) { + cpu_string += " | AVX"; + if (caps.avx512) { + cpu_string += "512"; + } else if (caps.avx2) { + cpu_string += '2'; + } + if (caps.fma || caps.fma4) { + cpu_string += " | FMA"; + } + } + LOG_INFO(Frontend, "Host CPU: {}", cpu_string); #endif LOG_INFO(Frontend, "Host OS: {}", QSysInfo::prettyProductName().toStdString()); + LOG_INFO(Frontend, "Host RAM: {:.2f} GB", + Common::GetMemInfo().TotalPhysicalMemory / 1024.0f / 1024 / 1024); + LOG_INFO(Frontend, "Host Swap: {:.2f} GB", + Common::GetMemInfo().TotalSwapMemory / 1024.0f / 1024 / 1024); UpdateWindowTitle(); show(); @@ -515,14 +534,36 @@ void GMainWindow::InitializeWidgets() { if (emulation_running) { return; } - Settings::values.use_asynchronous_gpu_emulation = - !Settings::values.use_asynchronous_gpu_emulation; + bool is_async = + !Settings::values.use_asynchronous_gpu_emulation || Settings::values.use_multi_core; + Settings::values.use_asynchronous_gpu_emulation = is_async; async_status_button->setChecked(Settings::values.use_asynchronous_gpu_emulation); Settings::Apply(); }); async_status_button->setText(tr("ASYNC")); async_status_button->setCheckable(true); async_status_button->setChecked(Settings::values.use_asynchronous_gpu_emulation); + + // Setup Multicore button + multicore_status_button = new QPushButton(); + multicore_status_button->setObjectName(QStringLiteral("TogglableStatusBarButton")); + multicore_status_button->setFocusPolicy(Qt::NoFocus); + connect(multicore_status_button, &QPushButton::clicked, [&] { + if (emulation_running) { + return; + } + Settings::values.use_multi_core = !Settings::values.use_multi_core; + bool is_async = + Settings::values.use_asynchronous_gpu_emulation || Settings::values.use_multi_core; + Settings::values.use_asynchronous_gpu_emulation = is_async; + async_status_button->setChecked(Settings::values.use_asynchronous_gpu_emulation); + multicore_status_button->setChecked(Settings::values.use_multi_core); + Settings::Apply(); + }); + multicore_status_button->setText(tr("MULTICORE")); + multicore_status_button->setCheckable(true); + multicore_status_button->setChecked(Settings::values.use_multi_core); + statusBar()->insertPermanentWidget(0, multicore_status_button); statusBar()->insertPermanentWidget(0, async_status_button); // Setup Renderer API button @@ -648,6 +689,11 @@ void GMainWindow::InitializeHotkeys() { ui.action_Capture_Screenshot->setShortcutContext( hotkey_registry.GetShortcutContext(main_window, capture_screenshot)); + ui.action_Fullscreen->setShortcut( + hotkey_registry.GetHotkey(main_window, fullscreen, this)->key()); + ui.action_Fullscreen->setShortcutContext( + hotkey_registry.GetShortcutContext(main_window, fullscreen)); + connect(hotkey_registry.GetHotkey(main_window, QStringLiteral("Load File"), this), &QShortcut::activated, this, &GMainWindow::OnMenuLoadFile); connect( @@ -684,10 +730,7 @@ void GMainWindow::InitializeHotkeys() { Settings::values.use_frame_limit = !Settings::values.use_frame_limit; UpdateStatusBar(); }); - // TODO: Remove this comment/static whenever the next major release of - // MSVC occurs and we make it a requirement (see: - // https://developercommunity.visualstudio.com/content/problem/93922/constexprs-are-trying-to-be-captured-in-lambda-fun.html) - static constexpr u16 SPEED_LIMIT_STEP = 5; + constexpr u16 SPEED_LIMIT_STEP = 5; connect(hotkey_registry.GetHotkey(main_window, QStringLiteral("Increase Speed Limit"), this), &QShortcut::activated, this, [&] { if (Settings::values.frame_limit < 9999 - SPEED_LIMIT_STEP) { @@ -721,16 +764,19 @@ void GMainWindow::InitializeHotkeys() { Settings::values.use_docked_mode); dock_status_button->setChecked(Settings::values.use_docked_mode); }); + connect(hotkey_registry.GetHotkey(main_window, QStringLiteral("Mute Audio"), this), + &QShortcut::activated, this, + [] { Settings::values.audio_muted = !Settings::values.audio_muted; }); } void GMainWindow::SetDefaultUIGeometry() { - // geometry: 55% of the window contents are in the upper screen half, 45% in the lower half + // geometry: 53% of the window contents are in the upper screen half, 47% in the lower half const QRect screenRect = QApplication::desktop()->screenGeometry(this); const int w = screenRect.width() * 2 / 3; - const int h = screenRect.height() / 2; + const int h = screenRect.height() * 2 / 3; const int x = (screenRect.x() + screenRect.width()) / 2 - w / 2; - const int y = (screenRect.y() + screenRect.height()) / 2 - h * 55 / 100; + const int y = (screenRect.y() + screenRect.height()) / 2 - h * 53 / 100; setGeometry(x, y, w, h); } @@ -821,6 +867,10 @@ void GMainWindow::ConnectMenuEvents() { connect(ui.action_Stop, &QAction::triggered, this, &GMainWindow::OnStopGame); connect(ui.action_Report_Compatibility, &QAction::triggered, this, &GMainWindow::OnMenuReportCompatibility); + connect(ui.action_Open_Mods_Page, &QAction::triggered, this, &GMainWindow::OnOpenModsPage); + connect(ui.action_Open_Quickstart_Guide, &QAction::triggered, this, + &GMainWindow::OnOpenQuickstartGuide); + connect(ui.action_Open_FAQ, &QAction::triggered, this, &GMainWindow::OnOpenFAQ); connect(ui.action_Restart, &QAction::triggered, this, [this] { BootGame(QString(game_path)); }); connect(ui.action_Configure, &QAction::triggered, this, &GMainWindow::OnConfigure); @@ -831,12 +881,9 @@ void GMainWindow::ConnectMenuEvents() { &GMainWindow::OnDisplayTitleBars); connect(ui.action_Show_Filter_Bar, &QAction::triggered, this, &GMainWindow::OnToggleFilterBar); connect(ui.action_Show_Status_Bar, &QAction::triggered, statusBar(), &QStatusBar::setVisible); + connect(ui.action_Reset_Window_Size, &QAction::triggered, this, &GMainWindow::ResetWindowSize); // Fullscreen - ui.action_Fullscreen->setShortcut( - hotkey_registry - .GetHotkey(QStringLiteral("Main Window"), QStringLiteral("Fullscreen"), this) - ->key()); connect(ui.action_Fullscreen, &QAction::triggered, this, &GMainWindow::ToggleFullscreen); // Movie @@ -904,6 +951,8 @@ bool GMainWindow::LoadROM(const QString& filename) { nullptr, // E-Commerce }); + system.RegisterHostThread(); + const Core::System::ResultStatus result{system.Load(*render_window, filename.toStdString())}; const auto drd_callout = @@ -1020,29 +1069,31 @@ void GMainWindow::BootGame(const QString& filename) { } status_bar_update_timer.start(2000); async_status_button->setDisabled(true); + multicore_status_button->setDisabled(true); renderer_status_button->setDisabled(true); if (UISettings::values.hide_mouse) { mouse_hide_timer.start(); setMouseTracking(true); ui.centralwidget->setMouseTracking(true); - ui.menubar->setMouseTracking(true); } const u64 title_id = Core::System::GetInstance().CurrentProcess()->GetTitleID(); std::string title_name; + std::string title_version; const auto res = Core::System::GetInstance().GetGameName(title_name); - if (res != Loader::ResultStatus::Success) { - const auto metadata = FileSys::PatchManager(title_id).GetControlMetadata(); - if (metadata.first != nullptr) - title_name = metadata.first->GetApplicationName(); - if (title_name.empty()) - title_name = FileUtil::GetFilename(filename.toStdString()); + const auto metadata = FileSys::PatchManager(title_id).GetControlMetadata(); + if (metadata.first != nullptr) { + title_version = metadata.first->GetVersionString(); + title_name = metadata.first->GetApplicationName(); } - LOG_INFO(Frontend, "Booting game: {:016X} | {}", title_id, title_name); - UpdateWindowTitle(QString::fromStdString(title_name)); + if (res != Loader::ResultStatus::Success || title_name.empty()) { + title_name = FileUtil::GetFilename(filename.toStdString()); + } + LOG_INFO(Frontend, "Booting game: {:016X} | {} | {}", title_id, title_name, title_version); + UpdateWindowTitle(title_name, title_version); loading_screen->Prepare(Core::System::GetInstance().GetAppLoader()); loading_screen->show(); @@ -1099,7 +1150,6 @@ void GMainWindow::ShutdownGame() { setMouseTracking(false); ui.centralwidget->setMouseTracking(false); - ui.menubar->setMouseTracking(false); UpdateWindowTitle(); @@ -1109,6 +1159,7 @@ void GMainWindow::ShutdownGame() { game_fps_label->setVisible(false); emu_frametime_label->setVisible(false); async_status_button->setEnabled(true); + multicore_status_button->setEnabled(true); #ifdef HAS_VULKAN renderer_status_button->setEnabled(true); #endif @@ -1156,39 +1207,61 @@ void GMainWindow::OnGameListLoadFile(QString game_path) { BootGame(game_path); } -void GMainWindow::OnGameListOpenFolder(u64 program_id, GameListOpenTarget target) { +void GMainWindow::OnGameListOpenFolder(GameListOpenTarget target, const std::string& game_path) { std::string path; QString open_target; + + const auto v_file = Core::GetGameFileFromPath(vfs, game_path); + const auto loader = Loader::GetLoader(v_file); + FileSys::NACP control{}; + u64 program_id{}; + + loader->ReadControlData(control); + loader->ReadProgramId(program_id); + + const bool has_user_save{control.GetDefaultNormalSaveSize() > 0}; + const bool has_device_save{control.GetDeviceSaveDataSize() > 0}; + + ASSERT_MSG(has_user_save != has_device_save, "Game uses both user and device savedata?"); + switch (target) { case GameListOpenTarget::SaveData: { open_target = tr("Save Data"); const std::string nand_dir = FileUtil::GetUserPath(FileUtil::UserPath::NANDDir); ASSERT(program_id != 0); - const auto select_profile = [this] { - QtProfileSelectionDialog dialog(this); - dialog.setWindowFlags(Qt::Dialog | Qt::CustomizeWindowHint | Qt::WindowTitleHint | - Qt::WindowSystemMenuHint | Qt::WindowCloseButtonHint); - dialog.setWindowModality(Qt::WindowModal); + if (has_user_save) { + // User save data + const auto select_profile = [this] { + QtProfileSelectionDialog dialog(this); + dialog.setWindowFlags(Qt::Dialog | Qt::CustomizeWindowHint | Qt::WindowTitleHint | + Qt::WindowSystemMenuHint | Qt::WindowCloseButtonHint); + dialog.setWindowModality(Qt::WindowModal); - if (dialog.exec() == QDialog::Rejected) { - return -1; - } + if (dialog.exec() == QDialog::Rejected) { + return -1; + } - return dialog.GetIndex(); - }; + return dialog.GetIndex(); + }; - const auto index = select_profile(); - if (index == -1) { - return; - } + const auto index = select_profile(); + if (index == -1) { + return; + } - Service::Account::ProfileManager manager; - const auto user_id = manager.GetUser(static_cast<std::size_t>(index)); - ASSERT(user_id); - path = nand_dir + FileSys::SaveDataFactory::GetFullPath(FileSys::SaveDataSpaceId::NandUser, - FileSys::SaveDataType::SaveData, - program_id, user_id->uuid, 0); + Service::Account::ProfileManager manager; + const auto user_id = manager.GetUser(static_cast<std::size_t>(index)); + ASSERT(user_id); + path = nand_dir + FileSys::SaveDataFactory::GetFullPath( + FileSys::SaveDataSpaceId::NandUser, + FileSys::SaveDataType::SaveData, program_id, user_id->uuid, 0); + } else { + // Device save data + path = nand_dir + FileSys::SaveDataFactory::GetFullPath( + FileSys::SaveDataSpaceId::NandUser, + FileSys::SaveDataType::SaveData, program_id, {}, 0); + } if (!FileUtil::Exists(path)) { FileUtil::CreateFullPath(path); @@ -1771,6 +1844,26 @@ void GMainWindow::OnMenuReportCompatibility() { } } +void GMainWindow::OpenURL(const QUrl& url) { + const bool open = QDesktopServices::openUrl(url); + if (!open) { + QMessageBox::warning(this, tr("Error opening URL"), + tr("Unable to open the URL \"%1\".").arg(url.toString())); + } +} + +void GMainWindow::OnOpenModsPage() { + OpenURL(QUrl(QStringLiteral("https://github.com/yuzu-emu/yuzu/wiki/Switch-Mods"))); +} + +void GMainWindow::OnOpenQuickstartGuide() { + OpenURL(QUrl(QStringLiteral("https://yuzu-emu.org/help/quickstart/"))); +} + +void GMainWindow::OnOpenFAQ() { + OpenURL(QUrl(QStringLiteral("https://yuzu-emu.org/wiki/faq/"))); +} + void GMainWindow::ToggleFullscreen() { if (!emulation_running) { return; @@ -1831,6 +1924,20 @@ void GMainWindow::ToggleWindowMode() { } } +void GMainWindow::ResetWindowSize() { + const auto aspect_ratio = Layout::EmulationAspectRatio( + static_cast<Layout::AspectRatio>(Settings::values.aspect_ratio), + static_cast<float>(Layout::ScreenUndocked::Height) / Layout::ScreenUndocked::Width); + if (!ui.action_Single_Window_Mode->isChecked()) { + render_window->resize(Layout::ScreenUndocked::Height / aspect_ratio, + Layout::ScreenUndocked::Height); + } else { + resize(Layout::ScreenUndocked::Height / aspect_ratio, + Layout::ScreenUndocked::Height + menuBar()->height() + + (ui.action_Show_Status_Bar->isChecked() ? statusBar()->height() : 0)); + } +} + void GMainWindow::OnConfigure() { const auto old_theme = UISettings::values.theme; const bool old_discord_presence = UISettings::values.enable_discord_presence; @@ -1861,16 +1968,18 @@ void GMainWindow::OnConfigure() { if (UISettings::values.hide_mouse && emulation_running) { setMouseTracking(true); ui.centralwidget->setMouseTracking(true); - ui.menubar->setMouseTracking(true); mouse_hide_timer.start(); } else { setMouseTracking(false); ui.centralwidget->setMouseTracking(false); - ui.menubar->setMouseTracking(false); } dock_status_button->setChecked(Settings::values.use_docked_mode); + multicore_status_button->setChecked(Settings::values.use_multi_core); + Settings::values.use_asynchronous_gpu_emulation = + Settings::values.use_asynchronous_gpu_emulation || Settings::values.use_multi_core; async_status_button->setChecked(Settings::values.use_asynchronous_gpu_emulation); + #ifdef HAS_VULKAN renderer_status_button->setChecked(Settings::values.renderer_backend == Settings::RendererBackend::Vulkan); @@ -1957,7 +2066,8 @@ void GMainWindow::OnCaptureScreenshot() { OnStartGame(); } -void GMainWindow::UpdateWindowTitle(const QString& title_name) { +void GMainWindow::UpdateWindowTitle(const std::string& title_name, + const std::string& title_version) { const auto full_name = std::string(Common::g_build_fullname); const auto branch_name = std::string(Common::g_scm_branch); const auto description = std::string(Common::g_scm_desc); @@ -1966,7 +2076,7 @@ void GMainWindow::UpdateWindowTitle(const QString& title_name) { const auto date = QDateTime::currentDateTime().toString(QStringLiteral("yyyy-MM-dd")).toStdString(); - if (title_name.isEmpty()) { + if (title_name.empty()) { const auto fmt = std::string(Common::g_title_bar_format_idle); setWindowTitle(QString::fromStdString(fmt::format(fmt.empty() ? "yuzu {0}| {1}-{2}" : fmt, full_name, branch_name, description, @@ -1974,8 +2084,8 @@ void GMainWindow::UpdateWindowTitle(const QString& title_name) { } else { const auto fmt = std::string(Common::g_title_bar_format_running); setWindowTitle(QString::fromStdString( - fmt::format(fmt.empty() ? "yuzu {0}| {3} | {1}-{2}" : fmt, full_name, branch_name, - description, title_name.toStdString(), date, build_id))); + fmt::format(fmt.empty() ? "yuzu {0}| {3} | {6} | {1}-{2}" : fmt, full_name, branch_name, + description, title_name, date, build_id, title_version))); } } @@ -1997,7 +2107,7 @@ void GMainWindow::UpdateStatusBar() { game_fps_label->setText(tr("Game: %1 FPS").arg(results.game_fps, 0, 'f', 0)); emu_frametime_label->setText(tr("Frame: %1 ms").arg(results.frametime * 1000.0, 0, 'f', 2)); - emu_speed_label->setVisible(true); + emu_speed_label->setVisible(!Settings::values.use_multi_core); game_fps_label->setVisible(true); emu_frametime_label->setVisible(true); } diff --git a/src/yuzu/main.h b/src/yuzu/main.h index 60b17c54a..66c84e5c0 100644 --- a/src/yuzu/main.h +++ b/src/yuzu/main.h @@ -181,9 +181,12 @@ private slots: void OnPauseGame(); void OnStopGame(); void OnMenuReportCompatibility(); + void OnOpenModsPage(); + void OnOpenQuickstartGuide(); + void OnOpenFAQ(); /// Called whenever a user selects a game in the game list widget. void OnGameListLoadFile(QString game_path); - void OnGameListOpenFolder(u64 program_id, GameListOpenTarget target); + void OnGameListOpenFolder(GameListOpenTarget target, const std::string& game_path); void OnTransferableShaderCacheOpenFile(u64 program_id); void OnGameListDumpRomFS(u64 program_id, const std::string& game_path); void OnGameListCopyTID(u64 program_id); @@ -208,16 +211,19 @@ private slots: void ShowFullscreen(); void HideFullscreen(); void ToggleWindowMode(); + void ResetWindowSize(); void OnCaptureScreenshot(); void OnCoreError(Core::System::ResultStatus, std::string); void OnReinitializeKeys(ReinitializeKeyBehavior behavior); private: std::optional<u64> SelectRomFSDumpTarget(const FileSys::ContentProvider&, u64 program_id); - void UpdateWindowTitle(const QString& title_name = {}); + void UpdateWindowTitle(const std::string& title_name = {}, + const std::string& title_version = {}); void UpdateStatusBar(); void HideMouseCursor(); void ShowMouseCursor(); + void OpenURL(const QUrl& url); Ui::MainWindow ui; @@ -233,6 +239,7 @@ private: QLabel* game_fps_label = nullptr; QLabel* emu_frametime_label = nullptr; QPushButton* async_status_button = nullptr; + QPushButton* multicore_status_button = nullptr; QPushButton* renderer_status_button = nullptr; QPushButton* dock_status_button = nullptr; QTimer status_bar_update_timer; diff --git a/src/yuzu/main.ui b/src/yuzu/main.ui index ae414241e..bee6e107e 100644 --- a/src/yuzu/main.ui +++ b/src/yuzu/main.ui @@ -6,8 +6,8 @@ <rect> <x>0</x> <y>0</y> - <width>1081</width> - <height>730</height> + <width>1280</width> + <height>720</height> </rect> </property> <property name="windowTitle"> @@ -44,7 +44,7 @@ <rect> <x>0</x> <y>0</y> - <width>1081</width> + <width>1280</width> <height>21</height> </rect> </property> @@ -96,6 +96,7 @@ <addaction name="action_Display_Dock_Widget_Headers"/> <addaction name="action_Show_Filter_Bar"/> <addaction name="action_Show_Status_Bar"/> + <addaction name="action_Reset_Window_Size"/> <addaction name="separator"/> <addaction name="menu_View_Debugging"/> </widget> @@ -112,6 +113,9 @@ <string>&Help</string> </property> <addaction name="action_Report_Compatibility"/> + <addaction name="action_Open_Mods_Page"/> + <addaction name="action_Open_Quickstart_Guide"/> + <addaction name="action_Open_FAQ"/> <addaction name="separator"/> <addaction name="action_About"/> </widget> @@ -215,6 +219,11 @@ <string>Show Status Bar</string> </property> </action> + <action name="action_Reset_Window_Size"> + <property name="text"> + <string>Reset Window Size</string> + </property> + </action> <action name="action_Fullscreen"> <property name="checkable"> <bool>true</bool> @@ -250,6 +259,21 @@ <bool>false</bool> </property> </action> + <action name="action_Open_Mods_Page"> + <property name="text"> + <string>Open Mods Page</string> + </property> + </action> + <action name="action_Open_Quickstart_Guide"> + <property name="text"> + <string>Open Quickstart Guide</string> + </property> + </action> + <action name="action_Open_FAQ"> + <property name="text"> + <string>FAQ</string> + </property> + </action> <action name="action_Open_yuzu_Folder"> <property name="text"> <string>Open yuzu Folder</string> diff --git a/src/yuzu/yuzu.rc b/src/yuzu/yuzu.rc index 1b253653f..4a3645a71 100644 --- a/src/yuzu/yuzu.rc +++ b/src/yuzu/yuzu.rc @@ -16,4 +16,4 @@ IDI_ICON1 ICON "../../dist/yuzu.ico" // RT_MANIFEST // -1 RT_MANIFEST "../../dist/yuzu.manifest" +0 RT_MANIFEST "../../dist/yuzu.manifest" diff --git a/src/yuzu_cmd/config.cpp b/src/yuzu_cmd/config.cpp index 8476a5a16..659b9f701 100644 --- a/src/yuzu_cmd/config.cpp +++ b/src/yuzu_cmd/config.cpp @@ -367,6 +367,9 @@ void Config::ReadValues() { Settings::values.custom_rtc = std::nullopt; } + Settings::values.language_index = sdl2_config->GetInteger("System", "language_index", 1); + Settings::values.time_zone_index = sdl2_config->GetInteger("System", "time_zone_index", 0); + // Core Settings::values.use_multi_core = sdl2_config->GetBoolean("Core", "use_multi_core", false); @@ -377,8 +380,6 @@ void Config::ReadValues() { Settings::values.renderer_debug = sdl2_config->GetBoolean("Renderer", "debug", false); Settings::values.vulkan_device = sdl2_config->GetInteger("Renderer", "vulkan_device", 0); - Settings::values.resolution_factor = - static_cast<float>(sdl2_config->GetReal("Renderer", "resolution_factor", 1.0)); Settings::values.aspect_ratio = static_cast<int>(sdl2_config->GetInteger("Renderer", "aspect_ratio", 0)); Settings::values.max_anisotropy = @@ -394,6 +395,8 @@ void Config::ReadValues() { sdl2_config->GetBoolean("Renderer", "use_asynchronous_gpu_emulation", false); Settings::values.use_vsync = static_cast<u16>(sdl2_config->GetInteger("Renderer", "use_vsync", 1)); + Settings::values.use_assembly_shaders = + sdl2_config->GetBoolean("Renderer", "use_assembly_shaders", false); Settings::values.use_fast_gpu_time = sdl2_config->GetBoolean("Renderer", "use_fast_gpu_time", true); @@ -409,8 +412,6 @@ void Config::ReadValues() { Settings::values.audio_device_id = sdl2_config->Get("Audio", "output_device", "auto"); Settings::values.volume = static_cast<float>(sdl2_config->GetReal("Audio", "volume", 1)); - Settings::values.language_index = sdl2_config->GetInteger("System", "language_index", 1); - // Miscellaneous Settings::values.log_filter = sdl2_config->Get("Miscellaneous", "log_filter", "*:Trace"); Settings::values.use_dev_keys = sdl2_config->GetBoolean("Miscellaneous", "use_dev_keys", false); @@ -429,6 +430,8 @@ void Config::ReadValues() { Settings::values.quest_flag = sdl2_config->GetBoolean("Debugging", "quest_flag", false); Settings::values.disable_cpu_opt = sdl2_config->GetBoolean("Debugging", "disable_cpu_opt", false); + Settings::values.disable_macro_jit = + sdl2_config->GetBoolean("Debugging", "disable_macro_jit", false); const auto title_list = sdl2_config->Get("AddOns", "title_ids", ""); std::stringstream ss(title_list); diff --git a/src/yuzu_cmd/default_ini.h b/src/yuzu_cmd/default_ini.h index 60b1a62fa..45c07ed5d 100644 --- a/src/yuzu_cmd/default_ini.h +++ b/src/yuzu_cmd/default_ini.h @@ -117,11 +117,6 @@ use_hw_renderer = # 0: Interpreter (slow), 1 (default): JIT (fast) use_shader_jit = -# Resolution scale factor -# 0: Auto (scales resolution to window size), 1: Native Switch screen resolution, Otherwise a scale -# factor for the Switch resolution -resolution_factor = - # Aspect ratio # 0: Default (16:9), 1: Force 4:3, 2: Force 21:9, 3: Stretch to Window aspect_ratio = @@ -134,6 +129,10 @@ max_anisotropy = # 0 (default): Off, 1: On use_vsync = +# Whether to use OpenGL assembly shaders or not. NV_gpu_program5 is required. +# 0 (default): Off, 1: On +use_assembly_shaders = + # Turns on the frame limiter, which will limit frames output to the target game speed # 0: Off, 1: On (default) use_frame_limit = @@ -262,6 +261,10 @@ language_index = # -1: Auto-select (default), 0: Japan, 1: USA, 2: Europe, 3: Australia, 4: China, 5: Korea, 6: Taiwan region_value = +# The system time zone that yuzu will use during emulation +# 0: Auto-select (default), 1: Default (system archive value), Others: Index for specified time zone +time_zone_index = + [Miscellaneous] # A filter which removes logs below a certain logging level. # Examples: *:Debug Kernel.SVC:Trace Service.*:Critical @@ -283,6 +286,8 @@ quest_flag = # Determines whether or not JIT CPU optimizations are enabled # false: Optimizations Enabled, true: Optimizations Disabled disable_cpu_opt = +# Enables/Disables the macro JIT compiler +disable_macro_jit=false [WebService] # Whether or not to enable telemetry diff --git a/src/yuzu_cmd/emu_window/emu_window_sdl2_gl.cpp b/src/yuzu_cmd/emu_window/emu_window_sdl2_gl.cpp index 411e7e647..09cc0a3b5 100644 --- a/src/yuzu_cmd/emu_window/emu_window_sdl2_gl.cpp +++ b/src/yuzu_cmd/emu_window/emu_window_sdl2_gl.cpp @@ -98,6 +98,9 @@ EmuWindow_SDL2_GL::EmuWindow_SDL2_GL(Core::System& system, bool fullscreen) SDL_GL_SetAttribute(SDL_GL_BLUE_SIZE, 8); SDL_GL_SetAttribute(SDL_GL_ALPHA_SIZE, 0); SDL_GL_SetAttribute(SDL_GL_SHARE_WITH_CURRENT_CONTEXT, 1); + if (Settings::values.renderer_debug) { + SDL_GL_SetAttribute(SDL_GL_CONTEXT_FLAGS, SDL_GL_CONTEXT_DEBUG_FLAG); + } SDL_GL_SetSwapInterval(0); std::string window_title = fmt::format("yuzu {} | {}-{}", Common::g_build_fullname, diff --git a/src/yuzu_cmd/yuzu.cpp b/src/yuzu_cmd/yuzu.cpp index 4d2ea7e9e..e6c6a839d 100644 --- a/src/yuzu_cmd/yuzu.cpp +++ b/src/yuzu_cmd/yuzu.cpp @@ -2,6 +2,7 @@ // Licensed under GPLv2 or any later version // Refer to the license.txt file included. +#include <chrono> #include <iostream> #include <memory> #include <string> @@ -236,9 +237,11 @@ int main(int argc, char** argv) { system.Renderer().Rasterizer().LoadDiskResources(); std::thread render_thread([&emu_window] { emu_window->Present(); }); + system.Run(); while (emu_window->IsOpen()) { - system.RunLoop(); + std::this_thread::sleep_for(std::chrono::milliseconds(1)); } + system.Pause(); render_thread.join(); system.Shutdown(); diff --git a/src/yuzu_cmd/yuzu.rc b/src/yuzu_cmd/yuzu.rc index 7de8ef3d9..0cde75e2f 100644 --- a/src/yuzu_cmd/yuzu.rc +++ b/src/yuzu_cmd/yuzu.rc @@ -14,4 +14,4 @@ YUZU_ICON ICON "../../dist/yuzu.ico" // RT_MANIFEST // -1 RT_MANIFEST "../../dist/yuzu.manifest" +0 RT_MANIFEST "../../dist/yuzu.manifest" diff --git a/src/yuzu_tester/config.cpp b/src/yuzu_tester/config.cpp index 3be58b15d..1566c2e3f 100644 --- a/src/yuzu_tester/config.cpp +++ b/src/yuzu_tester/config.cpp @@ -116,8 +116,6 @@ void Config::ReadValues() { Settings::values.use_multi_core = sdl2_config->GetBoolean("Core", "use_multi_core", false); // Renderer - Settings::values.resolution_factor = - static_cast<float>(sdl2_config->GetReal("Renderer", "resolution_factor", 1.0)); Settings::values.aspect_ratio = static_cast<int>(sdl2_config->GetInteger("Renderer", "aspect_ratio", 0)); Settings::values.max_anisotropy = diff --git a/src/yuzu_tester/default_ini.h b/src/yuzu_tester/default_ini.h index ca203b64d..41bbbbf60 100644 --- a/src/yuzu_tester/default_ini.h +++ b/src/yuzu_tester/default_ini.h @@ -21,11 +21,6 @@ use_hw_renderer = # 0: Interpreter (slow), 1 (default): JIT (fast) use_shader_jit = -# Resolution scale factor -# 0: Auto (scales resolution to window size), 1: Native Switch screen resolution, Otherwise a scale -# factor for the Switch resolution -resolution_factor = - # Aspect ratio # 0: Default (16:9), 1: Force 4:3, 2: Force 21:9, 3: Stretch to Window aspect_ratio = diff --git a/src/yuzu_tester/service/yuzutest.cpp b/src/yuzu_tester/service/yuzutest.cpp index 85d3f436b..2d3f6e3a7 100644 --- a/src/yuzu_tester/service/yuzutest.cpp +++ b/src/yuzu_tester/service/yuzutest.cpp @@ -53,7 +53,7 @@ private: IPC::ResponseBuilder rb{ctx, 3}; rb.Push(RESULT_SUCCESS); - rb.Push<u32>(write_size); + rb.Push<u32>(static_cast<u32>(write_size)); } void StartIndividual(Kernel::HLERequestContext& ctx) { diff --git a/src/yuzu_tester/yuzu.cpp b/src/yuzu_tester/yuzu.cpp index 676e70ebd..083667baf 100644 --- a/src/yuzu_tester/yuzu.cpp +++ b/src/yuzu_tester/yuzu.cpp @@ -2,6 +2,7 @@ // Licensed under GPLv2 or any later version // Refer to the license.txt file included. +#include <chrono> #include <iostream> #include <memory> #include <string> @@ -255,9 +256,11 @@ int main(int argc, char** argv) { system.GPU().Start(); system.Renderer().Rasterizer().LoadDiskResources(); + system.Run(); while (!finished) { - system.RunLoop(); + std::this_thread::sleep_for(std::chrono::milliseconds(1)); } + system.Pause(); detached_tasks.WaitForAllTasks(); return return_value; diff --git a/src/yuzu_tester/yuzu.rc b/src/yuzu_tester/yuzu.rc index 7de8ef3d9..0cde75e2f 100644 --- a/src/yuzu_tester/yuzu.rc +++ b/src/yuzu_tester/yuzu.rc @@ -14,4 +14,4 @@ YUZU_ICON ICON "../../dist/yuzu.ico" // RT_MANIFEST // -1 RT_MANIFEST "../../dist/yuzu.manifest" +0 RT_MANIFEST "../../dist/yuzu.manifest" |