diff options
55 files changed, 306 insertions, 198 deletions
diff --git a/externals/dynarmic b/externals/dynarmic -Subproject af2d50288fc537201014c4230bb55ab9018a743 +Subproject 644172477eaf0d822178cb7e96c62b75caa9657 diff --git a/src/CMakeLists.txt b/src/CMakeLists.txt index 3ad2c0950..9182dbfd4 100644 --- a/src/CMakeLists.txt +++ b/src/CMakeLists.txt @@ -65,12 +65,14 @@ if (MSVC) /we4305 # 'context': truncation from 'type1' to 'type2' /we4388 # 'expression': signed/unsigned mismatch /we4389 # 'operator': signed/unsigned mismatch + /we4505 # 'function': unreferenced local function has been removed /we4547 # 'operator': operator before comma has no effect; expected operator with side-effect /we4549 # 'operator1': operator before comma has no effect; did you intend 'operator2'? /we4555 # Expression has no effect; expected expression with side-effect /we4715 # 'function': not all control paths return a value /we4834 # Discarding return value of function with 'nodiscard' attribute /we5038 # data member 'member1' will be initialized after data member 'member2' + /we5245 # 'function': unreferenced function with internal linkage has been removed ) if (ARCHITECTURE_x86_64) diff --git a/src/common/host_memory.cpp b/src/common/host_memory.cpp index e829af1ac..802943eb7 100644 --- a/src/common/host_memory.cpp +++ b/src/common/host_memory.cpp @@ -149,7 +149,7 @@ public: } void Unmap(size_t virtual_offset, size_t length) { - std::lock_guard lock{placeholder_mutex}; + std::scoped_lock lock{placeholder_mutex}; // Unmap until there are no more placeholders while (UnmapOnePlaceholder(virtual_offset, length)) { @@ -169,7 +169,7 @@ public: } const size_t virtual_end = virtual_offset + length; - std::lock_guard lock{placeholder_mutex}; + std::scoped_lock lock{placeholder_mutex}; auto [it, end] = placeholders.equal_range({virtual_offset, virtual_end}); while (it != end) { const size_t offset = std::max(it->lower(), virtual_offset); diff --git a/src/common/thread.h b/src/common/thread.h index a8c17c71a..626609372 100644 --- a/src/common/thread.h +++ b/src/common/thread.h @@ -17,7 +17,7 @@ namespace Common { class Event { public: void Set() { - std::lock_guard lk{mutex}; + std::scoped_lock lk{mutex}; if (!is_set) { is_set = true; condvar.notify_one(); diff --git a/src/common/threadsafe_queue.h b/src/common/threadsafe_queue.h index 2c8c2b90e..7272ac6e8 100644 --- a/src/common/threadsafe_queue.h +++ b/src/common/threadsafe_queue.h @@ -52,7 +52,7 @@ public: // line before cv.wait // TODO(bunnei): This can be replaced with C++20 waitable atomics when properly supported. // See discussion on https://github.com/yuzu-emu/yuzu/pull/3173 for details. - std::lock_guard lock{cv_mutex}; + std::scoped_lock lock{cv_mutex}; cv.notify_one(); } @@ -159,7 +159,7 @@ public: template <typename Arg> void Push(Arg&& t) { - std::lock_guard lock{write_lock}; + std::scoped_lock lock{write_lock}; spsc_queue.Push(t); } diff --git a/src/common/x64/native_clock.cpp b/src/common/x64/native_clock.cpp index 7a3f21dcf..7fd9d22f8 100644 --- a/src/common/x64/native_clock.cpp +++ b/src/common/x64/native_clock.cpp @@ -10,25 +10,49 @@ #include "common/uint128.h" #include "common/x64/native_clock.h" +#ifdef _MSC_VER +#include <intrin.h> +#endif + namespace Common { +#ifdef _MSC_VER +__forceinline static u64 FencedRDTSC() { + _mm_lfence(); + _ReadWriteBarrier(); + const u64 result = __rdtsc(); + _mm_lfence(); + _ReadWriteBarrier(); + return result; +} +#else +static u64 FencedRDTSC() { + u64 result; + asm volatile("lfence\n\t" + "rdtsc\n\t" + "shl $32, %%rdx\n\t" + "or %%rdx, %0\n\t" + "lfence" + : "=a"(result) + : + : "rdx", "memory", "cc"); + return result; +} +#endif + u64 EstimateRDTSCFrequency() { // Discard the first result measuring the rdtsc. - _mm_mfence(); - __rdtsc(); + FencedRDTSC(); std::this_thread::sleep_for(std::chrono::milliseconds{1}); - _mm_mfence(); - __rdtsc(); + FencedRDTSC(); // Get the current time. const auto start_time = std::chrono::steady_clock::now(); - _mm_mfence(); - const u64 tsc_start = __rdtsc(); + const u64 tsc_start = FencedRDTSC(); // Wait for 200 milliseconds. std::this_thread::sleep_for(std::chrono::milliseconds{200}); const auto end_time = std::chrono::steady_clock::now(); - _mm_mfence(); - const u64 tsc_end = __rdtsc(); + const u64 tsc_end = FencedRDTSC(); // Calculate differences. const u64 timer_diff = static_cast<u64>( std::chrono::duration_cast<std::chrono::nanoseconds>(end_time - start_time).count()); @@ -42,8 +66,7 @@ NativeClock::NativeClock(u64 emulated_cpu_frequency_, u64 emulated_clock_frequen u64 rtsc_frequency_) : WallClock(emulated_cpu_frequency_, emulated_clock_frequency_, true), rtsc_frequency{ rtsc_frequency_} { - _mm_mfence(); - time_point.inner.last_measure = __rdtsc(); + time_point.inner.last_measure = FencedRDTSC(); time_point.inner.accumulated_ticks = 0U; ns_rtsc_factor = GetFixedPoint64Factor(NS_RATIO, rtsc_frequency); us_rtsc_factor = GetFixedPoint64Factor(US_RATIO, rtsc_frequency); @@ -58,8 +81,7 @@ u64 NativeClock::GetRTSC() { current_time_point.pack = Common::AtomicLoad128(time_point.pack.data()); do { - _mm_mfence(); - const u64 current_measure = __rdtsc(); + const u64 current_measure = FencedRDTSC(); u64 diff = current_measure - current_time_point.inner.last_measure; diff = diff & ~static_cast<u64>(static_cast<s64>(diff) >> 63); // max(diff, 0) new_time_point.inner.last_measure = current_measure > current_time_point.inner.last_measure @@ -80,8 +102,7 @@ void NativeClock::Pause(bool is_paused) { current_time_point.pack = Common::AtomicLoad128(time_point.pack.data()); do { new_time_point.pack = current_time_point.pack; - _mm_mfence(); - new_time_point.inner.last_measure = __rdtsc(); + new_time_point.inner.last_measure = FencedRDTSC(); } while (!Common::AtomicCompareAndSwap(time_point.pack.data(), new_time_point.pack, current_time_point.pack, current_time_point.pack)); } diff --git a/src/core/arm/arm_interface.h b/src/core/arm/arm_interface.h index c60322442..dce2f4195 100644 --- a/src/core/arm/arm_interface.h +++ b/src/core/arm/arm_interface.h @@ -171,6 +171,9 @@ public: /// Prepare core for thread reschedule (if needed to correctly handle state) virtual void PrepareReschedule() = 0; + /// Signal an interrupt and ask the core to halt as soon as possible. + virtual void SignalInterrupt() = 0; + struct BacktraceEntry { std::string module; u64 address; diff --git a/src/core/arm/dynarmic/arm_dynarmic_32.cpp b/src/core/arm/dynarmic/arm_dynarmic_32.cpp index 054572445..ab3210d84 100644 --- a/src/core/arm/dynarmic/arm_dynarmic_32.cpp +++ b/src/core/arm/dynarmic/arm_dynarmic_32.cpp @@ -25,6 +25,9 @@ namespace Core { using namespace Common::Literals; +constexpr Dynarmic::HaltReason break_loop = Dynarmic::HaltReason::UserDefined2; +constexpr Dynarmic::HaltReason svc_call = Dynarmic::HaltReason::UserDefined3; + class DynarmicCallbacks32 : public Dynarmic::A32::UserCallbacks { public: explicit DynarmicCallbacks32(ARM_Dynarmic_32& parent_) @@ -84,15 +87,13 @@ public: } void CallSVC(u32 swi) override { - parent.svc_called = true; parent.svc_swi = swi; - parent.jit->HaltExecution(); + parent.jit->HaltExecution(svc_call); } void AddTicks(u64 ticks) override { - if (parent.uses_wall_clock) { - return; - } + ASSERT_MSG(!parent.uses_wall_clock, "This should never happen - dynarmic ticking disabled"); + // 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 @@ -108,12 +109,8 @@ public: } u64 GetTicksRemaining() override { - if (parent.uses_wall_clock) { - if (!parent.interrupt_handlers[parent.core_index].IsInterrupted()) { - return minimum_run_cycles; - } - return 0U; - } + ASSERT_MSG(!parent.uses_wall_clock, "This should never happen - dynarmic ticking disabled"); + return std::max<s64>(parent.system.CoreTiming().GetDowncount(), 0); } @@ -148,6 +145,7 @@ std::shared_ptr<Dynarmic::A32::Jit> ARM_Dynarmic_32::MakeJit(Common::PageTable* // Timing config.wall_clock_cntpct = uses_wall_clock; + config.enable_cycle_counting = !uses_wall_clock; // Code cache size config.code_cache_size = 512_MiB; @@ -230,13 +228,11 @@ std::shared_ptr<Dynarmic::A32::Jit> ARM_Dynarmic_32::MakeJit(Common::PageTable* void ARM_Dynarmic_32::Run() { while (true) { - jit->Run(); - if (!svc_called) { - break; + const auto hr = jit->Run(); + if (Has(hr, svc_call)) { + Kernel::Svc::Call(system, svc_swi); } - svc_called = false; - Kernel::Svc::Call(system, svc_swi); - if (shutdown) { + if (Has(hr, break_loop)) { break; } } @@ -322,8 +318,11 @@ void ARM_Dynarmic_32::LoadContext(const ThreadContext32& ctx) { } void ARM_Dynarmic_32::PrepareReschedule() { - jit->HaltExecution(); - shutdown = true; + jit->HaltExecution(break_loop); +} + +void ARM_Dynarmic_32::SignalInterrupt() { + jit->HaltExecution(break_loop); } void ARM_Dynarmic_32::ClearInstructionCache() { diff --git a/src/core/arm/dynarmic/arm_dynarmic_32.h b/src/core/arm/dynarmic/arm_dynarmic_32.h index 5d47b600d..3f68a4ff1 100644 --- a/src/core/arm/dynarmic/arm_dynarmic_32.h +++ b/src/core/arm/dynarmic/arm_dynarmic_32.h @@ -57,6 +57,7 @@ public: void LoadContext(const ThreadContext64& ctx) override {} void PrepareReschedule() override; + void SignalInterrupt() override; void ClearExclusiveState() override; void ClearInstructionCache() override; @@ -83,9 +84,6 @@ private: // SVC callback u32 svc_swi{}; - bool svc_called{}; - - bool shutdown{}; }; } // namespace Core diff --git a/src/core/arm/dynarmic/arm_dynarmic_64.cpp b/src/core/arm/dynarmic/arm_dynarmic_64.cpp index 7ff8f9495..68822a1fc 100644 --- a/src/core/arm/dynarmic/arm_dynarmic_64.cpp +++ b/src/core/arm/dynarmic/arm_dynarmic_64.cpp @@ -26,6 +26,9 @@ namespace Core { using Vector = Dynarmic::A64::Vector; using namespace Common::Literals; +constexpr Dynarmic::HaltReason break_loop = Dynarmic::HaltReason::UserDefined2; +constexpr Dynarmic::HaltReason svc_call = Dynarmic::HaltReason::UserDefined3; + class DynarmicCallbacks64 : public Dynarmic::A64::UserCallbacks { public: explicit DynarmicCallbacks64(ARM_Dynarmic_64& parent_) @@ -106,7 +109,7 @@ public: break; } - parent.jit->HaltExecution(); + parent.jit->HaltExecution(Dynarmic::HaltReason::CacheInvalidation); } void ExceptionRaised(u64 pc, Dynarmic::A64::Exception exception) override { @@ -126,15 +129,12 @@ public: } void CallSVC(u32 swi) override { - parent.svc_called = true; parent.svc_swi = swi; - parent.jit->HaltExecution(); + parent.jit->HaltExecution(svc_call); } void AddTicks(u64 ticks) override { - if (parent.uses_wall_clock) { - return; - } + ASSERT_MSG(!parent.uses_wall_clock, "This should never happen - dynarmic ticking disabled"); // 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 @@ -149,12 +149,8 @@ public: } u64 GetTicksRemaining() override { - if (parent.uses_wall_clock) { - if (!parent.interrupt_handlers[parent.core_index].IsInterrupted()) { - return minimum_run_cycles; - } - return 0U; - } + ASSERT_MSG(!parent.uses_wall_clock, "This should never happen - dynarmic ticking disabled"); + return std::max<s64>(parent.system.CoreTiming().GetDowncount(), 0); } @@ -210,6 +206,7 @@ std::shared_ptr<Dynarmic::A64::Jit> ARM_Dynarmic_64::MakeJit(Common::PageTable* // Timing config.wall_clock_cntpct = uses_wall_clock; + config.enable_cycle_counting = !uses_wall_clock; // Code cache size config.code_cache_size = 512_MiB; @@ -292,13 +289,11 @@ std::shared_ptr<Dynarmic::A64::Jit> ARM_Dynarmic_64::MakeJit(Common::PageTable* void ARM_Dynarmic_64::Run() { while (true) { - jit->Run(); - if (!svc_called) { - break; + const auto hr = jit->Run(); + if (Has(hr, svc_call)) { + Kernel::Svc::Call(system, svc_swi); } - svc_called = false; - Kernel::Svc::Call(system, svc_swi); - if (shutdown) { + if (Has(hr, break_loop)) { break; } } @@ -389,8 +384,11 @@ void ARM_Dynarmic_64::LoadContext(const ThreadContext64& ctx) { } void ARM_Dynarmic_64::PrepareReschedule() { - jit->HaltExecution(); - shutdown = true; + jit->HaltExecution(break_loop); +} + +void ARM_Dynarmic_64::SignalInterrupt() { + jit->HaltExecution(break_loop); } void ARM_Dynarmic_64::ClearInstructionCache() { diff --git a/src/core/arm/dynarmic/arm_dynarmic_64.h b/src/core/arm/dynarmic/arm_dynarmic_64.h index 0c4e46c64..58bc7fbec 100644 --- a/src/core/arm/dynarmic/arm_dynarmic_64.h +++ b/src/core/arm/dynarmic/arm_dynarmic_64.h @@ -51,6 +51,7 @@ public: void LoadContext(const ThreadContext64& ctx) override; void PrepareReschedule() override; + void SignalInterrupt() override; void ClearExclusiveState() override; void ClearInstructionCache() override; @@ -77,9 +78,6 @@ private: // SVC callback u32 svc_swi{}; - bool svc_called{}; - - bool shutdown{}; }; } // namespace Core diff --git a/src/core/hle/kernel/kernel.cpp b/src/core/hle/kernel/kernel.cpp index 6387d0c29..134a0b8e9 100644 --- a/src/core/hle/kernel/kernel.cpp +++ b/src/core/hle/kernel/kernel.cpp @@ -99,7 +99,7 @@ struct KernelCore::Impl { // Close all open server ports. std::unordered_set<KServerPort*> server_ports_; { - std::lock_guard lk(server_ports_lock); + std::scoped_lock lk{server_ports_lock}; server_ports_ = server_ports; server_ports.clear(); } @@ -157,7 +157,7 @@ struct KernelCore::Impl { // Close kernel objects that were not freed on shutdown { - std::lock_guard lk(registered_in_use_objects_lock); + std::scoped_lock lk{registered_in_use_objects_lock}; if (registered_in_use_objects.size()) { for (auto& object : registered_in_use_objects) { object->Close(); @@ -178,7 +178,7 @@ struct KernelCore::Impl { // Track kernel objects that were not freed on shutdown { - std::lock_guard lk(registered_objects_lock); + std::scoped_lock lk{registered_objects_lock}; if (registered_objects.size()) { LOG_DEBUG(Kernel, "{} kernel objects were dangling on shutdown!", registered_objects.size()); @@ -660,7 +660,7 @@ struct KernelCore::Impl { KClientPort* port = &search->second(system.ServiceManager(), system); { - std::lock_guard lk(server_ports_lock); + std::scoped_lock lk{server_ports_lock}; server_ports.insert(&port->GetParent()->GetServerPort()); } return port; @@ -929,22 +929,22 @@ KClientPort* KernelCore::CreateNamedServicePort(std::string name) { } void KernelCore::RegisterKernelObject(KAutoObject* object) { - std::lock_guard lk(impl->registered_objects_lock); + std::scoped_lock lk{impl->registered_objects_lock}; impl->registered_objects.insert(object); } void KernelCore::UnregisterKernelObject(KAutoObject* object) { - std::lock_guard lk(impl->registered_objects_lock); + std::scoped_lock lk{impl->registered_objects_lock}; impl->registered_objects.erase(object); } void KernelCore::RegisterInUseObject(KAutoObject* object) { - std::lock_guard lk(impl->registered_in_use_objects_lock); + std::scoped_lock lk{impl->registered_in_use_objects_lock}; impl->registered_in_use_objects.insert(object); } void KernelCore::UnregisterInUseObject(KAutoObject* object) { - std::lock_guard lk(impl->registered_in_use_objects_lock); + std::scoped_lock lk{impl->registered_in_use_objects_lock}; impl->registered_in_use_objects.erase(object); } diff --git a/src/core/hle/kernel/physical_core.cpp b/src/core/hle/kernel/physical_core.cpp index 7477668e4..18a5f40f8 100644 --- a/src/core/hle/kernel/physical_core.cpp +++ b/src/core/hle/kernel/physical_core.cpp @@ -58,6 +58,7 @@ bool PhysicalCore::IsInterrupted() const { void PhysicalCore::Interrupt() { guard->lock(); interrupts[core_index].SetInterrupt(true); + arm_interface->SignalInterrupt(); guard->unlock(); } diff --git a/src/core/hle/kernel/time_manager.cpp b/src/core/hle/kernel/time_manager.cpp index aa985d820..5b8fe8eae 100644 --- a/src/core/hle/kernel/time_manager.cpp +++ b/src/core/hle/kernel/time_manager.cpp @@ -24,7 +24,7 @@ TimeManager::TimeManager(Core::System& system_) : system{system_} { } void TimeManager::ScheduleTimeEvent(KThread* thread, s64 nanoseconds) { - std::lock_guard lock{mutex}; + std::scoped_lock lock{mutex}; if (nanoseconds > 0) { ASSERT(thread); ASSERT(thread->GetState() != ThreadState::Runnable); @@ -35,7 +35,7 @@ void TimeManager::ScheduleTimeEvent(KThread* thread, s64 nanoseconds) { } void TimeManager::UnscheduleTimeEvent(KThread* thread) { - std::lock_guard lock{mutex}; + std::scoped_lock lock{mutex}; system.CoreTiming().UnscheduleEvent(time_manager_event_type, reinterpret_cast<uintptr_t>(thread)); } diff --git a/src/core/hle/service/hid/controllers/npad.cpp b/src/core/hle/service/hid/controllers/npad.cpp index aa6cb34b7..4e17a952e 100644 --- a/src/core/hle/service/hid/controllers/npad.cpp +++ b/src/core/hle/service/hid/controllers/npad.cpp @@ -318,7 +318,7 @@ void Controller_NPad::OnRelease() { } void Controller_NPad::RequestPadStateUpdate(Core::HID::NpadIdType npad_id) { - std::lock_guard lock{mutex}; + std::scoped_lock lock{mutex}; auto& controller = GetControllerFromNpadIdType(npad_id); const auto controller_type = controller.device->GetNpadStyleIndex(); if (!controller.device->IsConnected()) { diff --git a/src/core/hle/service/nvflinger/buffer_item_consumer.cpp b/src/core/hle/service/nvflinger/buffer_item_consumer.cpp index 93fa1ec10..d7ee5362b 100644 --- a/src/core/hle/service/nvflinger/buffer_item_consumer.cpp +++ b/src/core/hle/service/nvflinger/buffer_item_consumer.cpp @@ -21,7 +21,7 @@ Status BufferItemConsumer::AcquireBuffer(BufferItem* item, std::chrono::nanoseco return Status::BadValue; } - std::scoped_lock lock(mutex); + std::scoped_lock lock{mutex}; if (const auto status = AcquireBufferLocked(item, present_when); status != Status::NoError) { if (status != Status::NoBufferAvailable) { @@ -40,7 +40,7 @@ Status BufferItemConsumer::AcquireBuffer(BufferItem* item, std::chrono::nanoseco } Status BufferItemConsumer::ReleaseBuffer(const BufferItem& item, Fence& release_fence) { - std::scoped_lock lock(mutex); + std::scoped_lock lock{mutex}; if (const auto status = AddReleaseFenceLocked(item.buf, item.graphic_buffer, release_fence); status != Status::NoError) { diff --git a/src/core/hle/service/nvflinger/buffer_queue_consumer.cpp b/src/core/hle/service/nvflinger/buffer_queue_consumer.cpp index c527c577e..3ab9a8c05 100644 --- a/src/core/hle/service/nvflinger/buffer_queue_consumer.cpp +++ b/src/core/hle/service/nvflinger/buffer_queue_consumer.cpp @@ -19,7 +19,7 @@ BufferQueueConsumer::~BufferQueueConsumer() = default; Status BufferQueueConsumer::AcquireBuffer(BufferItem* out_buffer, std::chrono::nanoseconds expected_present) { - std::scoped_lock lock(core->mutex); + std::scoped_lock lock{core->mutex}; // Check that the consumer doesn't currently have the maximum number of buffers acquired. const s32 num_acquired_buffers{ @@ -120,7 +120,7 @@ Status BufferQueueConsumer::ReleaseBuffer(s32 slot, u64 frame_number, const Fenc std::shared_ptr<IProducerListener> listener; { - std::scoped_lock lock(core->mutex); + std::scoped_lock lock{core->mutex}; // If the frame number has changed because the buffer has been reallocated, we can ignore // this ReleaseBuffer for the old buffer. @@ -180,7 +180,7 @@ Status BufferQueueConsumer::Connect(std::shared_ptr<IConsumerListener> consumer_ LOG_DEBUG(Service_NVFlinger, "controlled_by_app={}", controlled_by_app); - std::scoped_lock lock(core->mutex); + std::scoped_lock lock{core->mutex}; if (core->is_abandoned) { LOG_ERROR(Service_NVFlinger, "BufferQueue has been abandoned"); @@ -199,7 +199,7 @@ Status BufferQueueConsumer::GetReleasedBuffers(u64* out_slot_mask) { return Status::BadValue; } - std::scoped_lock lock(core->mutex); + std::scoped_lock lock{core->mutex}; if (core->is_abandoned) { LOG_ERROR(Service_NVFlinger, "BufferQueue has been abandoned"); diff --git a/src/core/hle/service/nvflinger/buffer_queue_core.cpp b/src/core/hle/service/nvflinger/buffer_queue_core.cpp index 3a0481786..ec5aabaeb 100644 --- a/src/core/hle/service/nvflinger/buffer_queue_core.cpp +++ b/src/core/hle/service/nvflinger/buffer_queue_core.cpp @@ -15,7 +15,7 @@ BufferQueueCore::BufferQueueCore() = default; BufferQueueCore::~BufferQueueCore() = default; void BufferQueueCore::NotifyShutdown() { - std::scoped_lock lock(mutex); + std::scoped_lock lock{mutex}; is_shutting_down = true; diff --git a/src/core/hle/service/nvflinger/buffer_queue_producer.cpp b/src/core/hle/service/nvflinger/buffer_queue_producer.cpp index 3d6e990c3..6f604a88e 100644 --- a/src/core/hle/service/nvflinger/buffer_queue_producer.cpp +++ b/src/core/hle/service/nvflinger/buffer_queue_producer.cpp @@ -38,7 +38,7 @@ BufferQueueProducer::~BufferQueueProducer() { Status BufferQueueProducer::RequestBuffer(s32 slot, std::shared_ptr<GraphicBuffer>* buf) { LOG_DEBUG(Service_NVFlinger, "slot {}", slot); - std::scoped_lock lock(core->mutex); + std::scoped_lock lock{core->mutex}; if (core->is_abandoned) { LOG_ERROR(Service_NVFlinger, "BufferQueue has been abandoned"); @@ -65,7 +65,7 @@ Status BufferQueueProducer::SetBufferCount(s32 buffer_count) { std::shared_ptr<IConsumerListener> listener; { - std::scoped_lock lock(core->mutex); + std::scoped_lock lock{core->mutex}; core->WaitWhileAllocatingLocked(); if (core->is_abandoned) { @@ -236,7 +236,7 @@ Status BufferQueueProducer::DequeueBuffer(s32* out_slot, Fence* out_fence, bool Status return_flags = Status::NoError; bool attached_by_consumer = false; { - std::scoped_lock lock(core->mutex); + std::scoped_lock lock{core->mutex}; core->WaitWhileAllocatingLocked(); if (format == PixelFormat::NoFormat) { @@ -295,7 +295,7 @@ Status BufferQueueProducer::DequeueBuffer(s32* out_slot, Fence* out_fence, bool } { - std::scoped_lock lock(core->mutex); + std::scoped_lock lock{core->mutex}; if (core->is_abandoned) { LOG_ERROR(Service_NVFlinger, "BufferQueue has been abandoned"); @@ -320,7 +320,7 @@ Status BufferQueueProducer::DequeueBuffer(s32* out_slot, Fence* out_fence, bool Status BufferQueueProducer::DetachBuffer(s32 slot) { LOG_DEBUG(Service_NVFlinger, "slot {}", slot); - std::scoped_lock lock(core->mutex); + std::scoped_lock lock{core->mutex}; if (core->is_abandoned) { LOG_ERROR(Service_NVFlinger, "BufferQueue has been abandoned"); @@ -356,7 +356,7 @@ Status BufferQueueProducer::DetachNextBuffer(std::shared_ptr<GraphicBuffer>* out return Status::BadValue; } - std::scoped_lock lock(core->mutex); + std::scoped_lock lock{core->mutex}; core->WaitWhileAllocatingLocked(); if (core->is_abandoned) { @@ -399,7 +399,7 @@ Status BufferQueueProducer::AttachBuffer(s32* out_slot, return Status::BadValue; } - std::scoped_lock lock(core->mutex); + std::scoped_lock lock{core->mutex}; core->WaitWhileAllocatingLocked(); Status return_flags = Status::NoError; @@ -460,7 +460,7 @@ Status BufferQueueProducer::QueueBuffer(s32 slot, const QueueBufferInput& input, BufferItem item; { - std::scoped_lock lock(core->mutex); + std::scoped_lock lock{core->mutex}; if (core->is_abandoned) { LOG_ERROR(Service_NVFlinger, "BufferQueue has been abandoned"); @@ -576,7 +576,7 @@ Status BufferQueueProducer::QueueBuffer(s32 slot, const QueueBufferInput& input, // Call back without the main BufferQueue lock held, but with the callback lock held so we can // ensure that callbacks occur in order { - std::scoped_lock lock(callback_mutex); + std::scoped_lock lock{callback_mutex}; while (callback_ticket != current_callback_ticket) { callback_condition.wait(callback_mutex); } @@ -597,7 +597,7 @@ Status BufferQueueProducer::QueueBuffer(s32 slot, const QueueBufferInput& input, void BufferQueueProducer::CancelBuffer(s32 slot, const Fence& fence) { LOG_DEBUG(Service_NVFlinger, "slot {}", slot); - std::scoped_lock lock(core->mutex); + std::scoped_lock lock{core->mutex}; if (core->is_abandoned) { LOG_ERROR(Service_NVFlinger, "BufferQueue has been abandoned"); @@ -623,7 +623,7 @@ void BufferQueueProducer::CancelBuffer(s32 slot, const Fence& fence) { } Status BufferQueueProducer::Query(NativeWindow what, s32* out_value) { - std::scoped_lock lock(core->mutex); + std::scoped_lock lock{core->mutex}; if (out_value == nullptr) { LOG_ERROR(Service_NVFlinger, "outValue was nullptr"); @@ -673,7 +673,7 @@ Status BufferQueueProducer::Query(NativeWindow what, s32* out_value) { Status BufferQueueProducer::Connect(const std::shared_ptr<IProducerListener>& listener, NativeWindowApi api, bool producer_controlled_by_app, QueueBufferOutput* output) { - std::scoped_lock lock(core->mutex); + std::scoped_lock lock{core->mutex}; LOG_DEBUG(Service_NVFlinger, "api = {} producer_controlled_by_app = {}", api, producer_controlled_by_app); @@ -730,7 +730,7 @@ Status BufferQueueProducer::Disconnect(NativeWindowApi api) { std::shared_ptr<IConsumerListener> listener; { - std::scoped_lock lock(core->mutex); + std::scoped_lock lock{core->mutex}; core->WaitWhileAllocatingLocked(); @@ -780,7 +780,7 @@ Status BufferQueueProducer::SetPreallocatedBuffer(s32 slot, return Status::BadValue; } - std::scoped_lock lock(core->mutex); + std::scoped_lock lock{core->mutex}; slots[slot] = {}; slots[slot].graphic_buffer = buffer; diff --git a/src/core/hle/service/nvflinger/consumer_base.cpp b/src/core/hle/service/nvflinger/consumer_base.cpp index c2c80832c..30fc21acc 100644 --- a/src/core/hle/service/nvflinger/consumer_base.cpp +++ b/src/core/hle/service/nvflinger/consumer_base.cpp @@ -18,7 +18,7 @@ ConsumerBase::ConsumerBase(std::unique_ptr<BufferQueueConsumer> consumer_) : consumer{std::move(consumer_)} {} ConsumerBase::~ConsumerBase() { - std::scoped_lock lock(mutex); + std::scoped_lock lock{mutex}; ASSERT_MSG(is_abandoned, "consumer is not abandoned!"); } @@ -44,7 +44,7 @@ void ConsumerBase::OnFrameReplaced(const BufferItem& item) { } void ConsumerBase::OnBuffersReleased() { - std::scoped_lock lock(mutex); + std::scoped_lock lock{mutex}; LOG_DEBUG(Service_NVFlinger, "called"); diff --git a/src/core/hle/service/nvflinger/graphic_buffer_producer.cpp b/src/core/hle/service/nvflinger/graphic_buffer_producer.cpp index d4da98ddb..04068827e 100644 --- a/src/core/hle/service/nvflinger/graphic_buffer_producer.cpp +++ b/src/core/hle/service/nvflinger/graphic_buffer_producer.cpp @@ -4,8 +4,6 @@ // Parts of this implementation were base on: // https://cs.android.com/android/platform/superproject/+/android-5.1.1_r38:frameworks/native/libs/gui/IGraphicBufferProducer.cpp -#pragma once - #include "core/hle/service/nvflinger/graphic_buffer_producer.h" #include "core/hle/service/nvflinger/parcel.h" diff --git a/src/core/hle/service/nvflinger/hos_binder_driver_server.cpp b/src/core/hle/service/nvflinger/hos_binder_driver_server.cpp index 0c937d682..094ba2542 100644 --- a/src/core/hle/service/nvflinger/hos_binder_driver_server.cpp +++ b/src/core/hle/service/nvflinger/hos_binder_driver_server.cpp @@ -14,7 +14,7 @@ HosBinderDriverServer::HosBinderDriverServer(Core::System& system_) HosBinderDriverServer::~HosBinderDriverServer() {} u64 HosBinderDriverServer::RegisterProducer(std::unique_ptr<android::IBinder>&& binder) { - std::lock_guard lk{lock}; + std::scoped_lock lk{lock}; last_id++; @@ -24,7 +24,7 @@ u64 HosBinderDriverServer::RegisterProducer(std::unique_ptr<android::IBinder>&& } android::IBinder* HosBinderDriverServer::TryGetProducer(u64 id) { - std::lock_guard lk{lock}; + std::scoped_lock lk{lock}; if (auto search = producers.find(id); search != producers.end()) { return search->second.get(); diff --git a/src/core/perf_stats.cpp b/src/core/perf_stats.cpp index 52c43c857..6ef459b7a 100644 --- a/src/core/perf_stats.cpp +++ b/src/core/perf_stats.cpp @@ -53,13 +53,13 @@ PerfStats::~PerfStats() { } void PerfStats::BeginSystemFrame() { - std::lock_guard lock{object_mutex}; + std::scoped_lock lock{object_mutex}; frame_begin = Clock::now(); } void PerfStats::EndSystemFrame() { - std::lock_guard lock{object_mutex}; + std::scoped_lock lock{object_mutex}; auto frame_end = Clock::now(); const auto frame_time = frame_end - frame_begin; @@ -79,7 +79,7 @@ void PerfStats::EndGameFrame() { } double PerfStats::GetMeanFrametime() const { - std::lock_guard lock{object_mutex}; + std::scoped_lock lock{object_mutex}; if (current_index <= IgnoreFrames) { return 0; @@ -91,7 +91,7 @@ double PerfStats::GetMeanFrametime() const { } PerfStatsResults PerfStats::GetAndResetStats(microseconds current_system_time_us) { - std::lock_guard lock{object_mutex}; + std::scoped_lock lock{object_mutex}; const auto now = Clock::now(); // Walltime elapsed since stats were reset @@ -120,7 +120,7 @@ PerfStatsResults PerfStats::GetAndResetStats(microseconds current_system_time_us } double PerfStats::GetLastFrameTimeScale() const { - std::lock_guard lock{object_mutex}; + std::scoped_lock lock{object_mutex}; constexpr double FRAME_LENGTH = 1.0 / 60; return duration_cast<DoubleSecs>(previous_frame_length).count() / FRAME_LENGTH; diff --git a/src/core/tools/freezer.cpp b/src/core/tools/freezer.cpp index 032c71aff..c81dc0e52 100644 --- a/src/core/tools/freezer.cpp +++ b/src/core/tools/freezer.cpp @@ -80,7 +80,7 @@ bool Freezer::IsActive() const { } void Freezer::Clear() { - std::lock_guard lock{entries_mutex}; + std::scoped_lock lock{entries_mutex}; LOG_DEBUG(Common_Memory, "Clearing all frozen memory values."); @@ -88,7 +88,7 @@ void Freezer::Clear() { } u64 Freezer::Freeze(VAddr address, u32 width) { - std::lock_guard lock{entries_mutex}; + std::scoped_lock lock{entries_mutex}; const auto current_value = MemoryReadWidth(memory, width, address); entries.push_back({address, width, current_value}); @@ -101,7 +101,7 @@ u64 Freezer::Freeze(VAddr address, u32 width) { } void Freezer::Unfreeze(VAddr address) { - std::lock_guard lock{entries_mutex}; + std::scoped_lock lock{entries_mutex}; LOG_DEBUG(Common_Memory, "Unfreezing memory for address={:016X}", address); @@ -109,13 +109,13 @@ void Freezer::Unfreeze(VAddr address) { } bool Freezer::IsFrozen(VAddr address) const { - std::lock_guard lock{entries_mutex}; + std::scoped_lock lock{entries_mutex}; return FindEntry(address) != entries.cend(); } void Freezer::SetFrozenValue(VAddr address, u64 value) { - std::lock_guard lock{entries_mutex}; + std::scoped_lock lock{entries_mutex}; const auto iter = FindEntry(address); @@ -132,7 +132,7 @@ void Freezer::SetFrozenValue(VAddr address, u64 value) { } std::optional<Freezer::Entry> Freezer::GetEntry(VAddr address) const { - std::lock_guard lock{entries_mutex}; + std::scoped_lock lock{entries_mutex}; const auto iter = FindEntry(address); @@ -144,7 +144,7 @@ std::optional<Freezer::Entry> Freezer::GetEntry(VAddr address) const { } std::vector<Freezer::Entry> Freezer::GetEntries() const { - std::lock_guard lock{entries_mutex}; + std::scoped_lock lock{entries_mutex}; return entries; } @@ -165,7 +165,7 @@ void Freezer::FrameCallback(std::uintptr_t, std::chrono::nanoseconds ns_late) { return; } - std::lock_guard lock{entries_mutex}; + std::scoped_lock lock{entries_mutex}; for (const auto& entry : entries) { LOG_DEBUG(Common_Memory, @@ -178,7 +178,7 @@ void Freezer::FrameCallback(std::uintptr_t, std::chrono::nanoseconds ns_late) { } void Freezer::FillEntryReads() { - std::lock_guard lock{entries_mutex}; + std::scoped_lock lock{entries_mutex}; LOG_DEBUG(Common_Memory, "Updating memory freeze entries to current values."); diff --git a/src/input_common/drivers/sdl_driver.cpp b/src/input_common/drivers/sdl_driver.cpp index c17ea305e..b3e4c3f64 100644 --- a/src/input_common/drivers/sdl_driver.cpp +++ b/src/input_common/drivers/sdl_driver.cpp @@ -62,7 +62,7 @@ public: bool UpdateMotion(SDL_ControllerSensorEvent event) { constexpr float gravity_constant = 9.80665f; - std::lock_guard lock{mutex}; + std::scoped_lock lock{mutex}; const u64 time_difference = event.timestamp - last_motion_update; last_motion_update = event.timestamp; switch (event.sensor) { @@ -241,7 +241,7 @@ private: }; std::shared_ptr<SDLJoystick> SDLDriver::GetSDLJoystickByGUID(const std::string& guid, int port) { - std::lock_guard lock{joystick_map_mutex}; + std::scoped_lock lock{joystick_map_mutex}; const auto it = joystick_map.find(guid); if (it != joystick_map.end()) { @@ -263,7 +263,7 @@ std::shared_ptr<SDLJoystick> SDLDriver::GetSDLJoystickBySDLID(SDL_JoystickID sdl auto sdl_joystick = SDL_JoystickFromInstanceID(sdl_id); const std::string guid = GetGUID(sdl_joystick); - std::lock_guard lock{joystick_map_mutex}; + std::scoped_lock lock{joystick_map_mutex}; const auto map_it = joystick_map.find(guid); if (map_it == joystick_map.end()) { @@ -297,7 +297,7 @@ void SDLDriver::InitJoystick(int joystick_index) { const std::string guid = GetGUID(sdl_joystick); - std::lock_guard lock{joystick_map_mutex}; + std::scoped_lock lock{joystick_map_mutex}; if (joystick_map.find(guid) == joystick_map.end()) { auto joystick = std::make_shared<SDLJoystick>(guid, 0, sdl_joystick, sdl_gamecontroller); PreSetController(joystick->GetPadIdentifier()); @@ -326,7 +326,7 @@ void SDLDriver::InitJoystick(int joystick_index) { void SDLDriver::CloseJoystick(SDL_Joystick* sdl_joystick) { const std::string guid = GetGUID(sdl_joystick); - std::lock_guard lock{joystick_map_mutex}; + std::scoped_lock lock{joystick_map_mutex}; // This call to guid is safe since the joystick is guaranteed to be in the map const auto& joystick_guid_list = joystick_map[guid]; const auto joystick_it = std::find_if(joystick_guid_list.begin(), joystick_guid_list.end(), @@ -392,7 +392,7 @@ void SDLDriver::HandleGameControllerEvent(const SDL_Event& event) { } void SDLDriver::CloseJoysticks() { - std::lock_guard lock{joystick_map_mutex}; + std::scoped_lock lock{joystick_map_mutex}; joystick_map.clear(); } diff --git a/src/input_common/input_engine.cpp b/src/input_common/input_engine.cpp index 738022ece..a16cca33d 100644 --- a/src/input_common/input_engine.cpp +++ b/src/input_common/input_engine.cpp @@ -8,37 +8,37 @@ namespace InputCommon { void InputEngine::PreSetController(const PadIdentifier& identifier) { - std::lock_guard lock{mutex}; + std::scoped_lock lock{mutex}; controller_list.try_emplace(identifier); } void InputEngine::PreSetButton(const PadIdentifier& identifier, int button) { - std::lock_guard lock{mutex}; + std::scoped_lock lock{mutex}; ControllerData& controller = controller_list.at(identifier); controller.buttons.try_emplace(button, false); } void InputEngine::PreSetHatButton(const PadIdentifier& identifier, int button) { - std::lock_guard lock{mutex}; + std::scoped_lock lock{mutex}; ControllerData& controller = controller_list.at(identifier); controller.hat_buttons.try_emplace(button, u8{0}); } void InputEngine::PreSetAxis(const PadIdentifier& identifier, int axis) { - std::lock_guard lock{mutex}; + std::scoped_lock lock{mutex}; ControllerData& controller = controller_list.at(identifier); controller.axes.try_emplace(axis, 0.0f); } void InputEngine::PreSetMotion(const PadIdentifier& identifier, int motion) { - std::lock_guard lock{mutex}; + std::scoped_lock lock{mutex}; ControllerData& controller = controller_list.at(identifier); controller.motions.try_emplace(motion); } void InputEngine::SetButton(const PadIdentifier& identifier, int button, bool value) { { - std::lock_guard lock{mutex}; + std::scoped_lock lock{mutex}; ControllerData& controller = controller_list.at(identifier); if (!configuring) { controller.buttons.insert_or_assign(button, value); @@ -49,7 +49,7 @@ void InputEngine::SetButton(const PadIdentifier& identifier, int button, bool va void InputEngine::SetHatButton(const PadIdentifier& identifier, int button, u8 value) { { - std::lock_guard lock{mutex}; + std::scoped_lock lock{mutex}; ControllerData& controller = controller_list.at(identifier); if (!configuring) { controller.hat_buttons.insert_or_assign(button, value); @@ -60,7 +60,7 @@ void InputEngine::SetHatButton(const PadIdentifier& identifier, int button, u8 v void InputEngine::SetAxis(const PadIdentifier& identifier, int axis, f32 value) { { - std::lock_guard lock{mutex}; + std::scoped_lock lock{mutex}; ControllerData& controller = controller_list.at(identifier); if (!configuring) { controller.axes.insert_or_assign(axis, value); @@ -71,7 +71,7 @@ void InputEngine::SetAxis(const PadIdentifier& identifier, int axis, f32 value) void InputEngine::SetBattery(const PadIdentifier& identifier, Common::Input::BatteryLevel value) { { - std::lock_guard lock{mutex}; + std::scoped_lock lock{mutex}; ControllerData& controller = controller_list.at(identifier); if (!configuring) { controller.battery = value; @@ -82,7 +82,7 @@ void InputEngine::SetBattery(const PadIdentifier& identifier, Common::Input::Bat void InputEngine::SetMotion(const PadIdentifier& identifier, int motion, const BasicMotion& value) { { - std::lock_guard lock{mutex}; + std::scoped_lock lock{mutex}; ControllerData& controller = controller_list.at(identifier); if (!configuring) { controller.motions.insert_or_assign(motion, value); @@ -92,7 +92,7 @@ void InputEngine::SetMotion(const PadIdentifier& identifier, int motion, const B } bool InputEngine::GetButton(const PadIdentifier& identifier, int button) const { - std::lock_guard lock{mutex}; + std::scoped_lock lock{mutex}; const auto controller_iter = controller_list.find(identifier); if (controller_iter == controller_list.cend()) { LOG_ERROR(Input, "Invalid identifier guid={}, pad={}, port={}", identifier.guid.RawString(), @@ -109,7 +109,7 @@ bool InputEngine::GetButton(const PadIdentifier& identifier, int button) const { } bool InputEngine::GetHatButton(const PadIdentifier& identifier, int button, u8 direction) const { - std::lock_guard lock{mutex}; + std::scoped_lock lock{mutex}; const auto controller_iter = controller_list.find(identifier); if (controller_iter == controller_list.cend()) { LOG_ERROR(Input, "Invalid identifier guid={}, pad={}, port={}", identifier.guid.RawString(), @@ -126,7 +126,7 @@ bool InputEngine::GetHatButton(const PadIdentifier& identifier, int button, u8 d } f32 InputEngine::GetAxis(const PadIdentifier& identifier, int axis) const { - std::lock_guard lock{mutex}; + std::scoped_lock lock{mutex}; const auto controller_iter = controller_list.find(identifier); if (controller_iter == controller_list.cend()) { LOG_ERROR(Input, "Invalid identifier guid={}, pad={}, port={}", identifier.guid.RawString(), @@ -143,7 +143,7 @@ f32 InputEngine::GetAxis(const PadIdentifier& identifier, int axis) const { } Common::Input::BatteryLevel InputEngine::GetBattery(const PadIdentifier& identifier) const { - std::lock_guard lock{mutex}; + std::scoped_lock lock{mutex}; const auto controller_iter = controller_list.find(identifier); if (controller_iter == controller_list.cend()) { LOG_ERROR(Input, "Invalid identifier guid={}, pad={}, port={}", identifier.guid.RawString(), @@ -155,7 +155,7 @@ Common::Input::BatteryLevel InputEngine::GetBattery(const PadIdentifier& identif } BasicMotion InputEngine::GetMotion(const PadIdentifier& identifier, int motion) const { - std::lock_guard lock{mutex}; + std::scoped_lock lock{mutex}; const auto controller_iter = controller_list.find(identifier); if (controller_iter == controller_list.cend()) { LOG_ERROR(Input, "Invalid identifier guid={}, pad={}, port={}", identifier.guid.RawString(), @@ -186,7 +186,7 @@ void InputEngine::ResetAnalogState() { } void InputEngine::TriggerOnButtonChange(const PadIdentifier& identifier, int button, bool value) { - std::lock_guard lock{mutex_callback}; + std::scoped_lock lock{mutex_callback}; for (const auto& poller_pair : callback_list) { const InputIdentifier& poller = poller_pair.second; if (!IsInputIdentifierEqual(poller, identifier, EngineInputType::Button, button)) { @@ -214,7 +214,7 @@ void InputEngine::TriggerOnButtonChange(const PadIdentifier& identifier, int but } void InputEngine::TriggerOnHatButtonChange(const PadIdentifier& identifier, int button, u8 value) { - std::lock_guard lock{mutex_callback}; + std::scoped_lock lock{mutex_callback}; for (const auto& poller_pair : callback_list) { const InputIdentifier& poller = poller_pair.second; if (!IsInputIdentifierEqual(poller, identifier, EngineInputType::HatButton, button)) { @@ -243,7 +243,7 @@ void InputEngine::TriggerOnHatButtonChange(const PadIdentifier& identifier, int } void InputEngine::TriggerOnAxisChange(const PadIdentifier& identifier, int axis, f32 value) { - std::lock_guard lock{mutex_callback}; + std::scoped_lock lock{mutex_callback}; for (const auto& poller_pair : callback_list) { const InputIdentifier& poller = poller_pair.second; if (!IsInputIdentifierEqual(poller, identifier, EngineInputType::Analog, axis)) { @@ -270,7 +270,7 @@ void InputEngine::TriggerOnAxisChange(const PadIdentifier& identifier, int axis, void InputEngine::TriggerOnBatteryChange(const PadIdentifier& identifier, [[maybe_unused]] Common::Input::BatteryLevel value) { - std::lock_guard lock{mutex_callback}; + std::scoped_lock lock{mutex_callback}; for (const auto& poller_pair : callback_list) { const InputIdentifier& poller = poller_pair.second; if (!IsInputIdentifierEqual(poller, identifier, EngineInputType::Battery, 0)) { @@ -284,7 +284,7 @@ void InputEngine::TriggerOnBatteryChange(const PadIdentifier& identifier, void InputEngine::TriggerOnMotionChange(const PadIdentifier& identifier, int motion, const BasicMotion& value) { - std::lock_guard lock{mutex_callback}; + std::scoped_lock lock{mutex_callback}; for (const auto& poller_pair : callback_list) { const InputIdentifier& poller = poller_pair.second; if (!IsInputIdentifierEqual(poller, identifier, EngineInputType::Motion, motion)) { @@ -346,18 +346,18 @@ const std::string& InputEngine::GetEngineName() const { } int InputEngine::SetCallback(InputIdentifier input_identifier) { - std::lock_guard lock{mutex_callback}; + std::scoped_lock lock{mutex_callback}; callback_list.insert_or_assign(last_callback_key, std::move(input_identifier)); return last_callback_key++; } void InputEngine::SetMappingCallback(MappingCallback callback) { - std::lock_guard lock{mutex_callback}; + std::scoped_lock lock{mutex_callback}; mapping_callback = std::move(callback); } void InputEngine::DeleteCallback(int key) { - std::lock_guard lock{mutex_callback}; + std::scoped_lock lock{mutex_callback}; const auto& iterator = callback_list.find(key); if (iterator == callback_list.end()) { LOG_ERROR(Input, "Tried to delete non-existent callback {}", key); diff --git a/src/shader_recompiler/backend/glsl/emit_glsl_context_get_set.cpp b/src/shader_recompiler/backend/glsl/emit_glsl_context_get_set.cpp index 0c1fbc7b1..282668b36 100644 --- a/src/shader_recompiler/backend/glsl/emit_glsl_context_get_set.cpp +++ b/src/shader_recompiler/backend/glsl/emit_glsl_context_get_set.cpp @@ -35,6 +35,15 @@ std::string_view OutputVertexIndex(EmitContext& ctx) { return ctx.stage == Stage::TessellationControl ? "[gl_InvocationID]" : ""; } +std::string ChooseCbuf(EmitContext& ctx, const IR::Value& binding, std::string_view index) { + if (binding.IsImmediate()) { + return fmt::format("{}_cbuf{}[{}]", ctx.stage_name, binding.U32(), index); + } else { + const auto binding_var{ctx.var_alloc.Consume(binding)}; + return fmt::format("GetCbufIndirect({},{})", binding_var, index); + } +} + void GetCbuf(EmitContext& ctx, std::string_view ret, const IR::Value& binding, const IR::Value& offset, u32 num_bits, std::string_view cast = {}, std::string_view bit_offset = {}) { @@ -55,8 +64,8 @@ void GetCbuf(EmitContext& ctx, std::string_view ret, const IR::Value& binding, const auto swizzle{is_immediate ? fmt::format(".{}", OffsetSwizzle(offset.U32())) : fmt::format("[({}>>2)%4]", offset_var)}; - const auto cbuf{fmt::format("{}_cbuf{}", ctx.stage_name, binding.U32())}; - const auto cbuf_cast{fmt::format("{}({}[{}]{{}})", cast, cbuf, index)}; + const auto cbuf{ChooseCbuf(ctx, binding, index)}; + const auto cbuf_cast{fmt::format("{}({}{{}})", cast, cbuf)}; const auto extraction{num_bits == 32 ? cbuf_cast : fmt::format("bitfieldExtract({},int({}),{})", cbuf_cast, bit_offset, num_bits)}; @@ -140,9 +149,9 @@ void EmitGetCbufF32(EmitContext& ctx, IR::Inst& inst, const IR::Value& binding, void EmitGetCbufU32x2(EmitContext& ctx, IR::Inst& inst, const IR::Value& binding, const IR::Value& offset) { - const auto cbuf{fmt::format("{}_cbuf{}", ctx.stage_name, binding.U32())}; const auto cast{ctx.profile.has_gl_cbuf_ftou_bug ? "" : "ftou"}; if (offset.IsImmediate()) { + const auto cbuf{fmt::format("{}_cbuf{}", ctx.stage_name, binding.U32())}; static constexpr u32 cbuf_size{0x10000}; const u32 u32_offset{offset.U32()}; const s32 signed_offset{static_cast<s32>(offset.U32())}; @@ -162,17 +171,17 @@ void EmitGetCbufU32x2(EmitContext& ctx, IR::Inst& inst, const IR::Value& binding return; } const auto offset_var{ctx.var_alloc.Consume(offset)}; + const auto cbuf{ChooseCbuf(ctx, binding, fmt::format("{}>>4", offset_var))}; if (!ctx.profile.has_gl_component_indexing_bug) { - ctx.AddU32x2("{}=uvec2({}({}[{}>>4][({}>>2)%4]),{}({}[({}+4)>>4][(({}+4)>>2)%4]));", inst, - cast, cbuf, offset_var, offset_var, cast, cbuf, offset_var, offset_var); + ctx.AddU32x2("{}=uvec2({}({}[({}>>2)%4]),{}({}[(({}+4)>>2)%4]));", inst, cast, cbuf, + offset_var, cast, cbuf, offset_var); return; } const auto ret{ctx.var_alloc.Define(inst, GlslVarType::U32x2)}; const auto cbuf_offset{fmt::format("{}>>2", offset_var)}; for (u32 swizzle = 0; swizzle < 4; ++swizzle) { - ctx.Add("if(({}&3)=={}){}=uvec2({}({}[{}>>4].{}),{}({}[({}+4)>>4].{}));", cbuf_offset, - swizzle, ret, cast, cbuf, offset_var, "xyzw"[swizzle], cast, cbuf, offset_var, - "xyzw"[(swizzle + 1) % 4]); + ctx.Add("if(({}&3)=={}){}=uvec2({}({}.{}),{}({}.{}));", cbuf_offset, swizzle, ret, cast, + cbuf, "xyzw"[swizzle], cast, cbuf, "xyzw"[(swizzle + 1) % 4]); } } diff --git a/src/shader_recompiler/backend/glsl/glsl_emit_context.cpp b/src/shader_recompiler/backend/glsl/glsl_emit_context.cpp index e816a93ec..17266f40d 100644 --- a/src/shader_recompiler/backend/glsl/glsl_emit_context.cpp +++ b/src/shader_recompiler/backend/glsl/glsl_emit_context.cpp @@ -359,6 +359,7 @@ EmitContext::EmitContext(IR::Program& program, Bindings& bindings, const Profile header += "layout(location=0) uniform vec4 scaling;"; } DefineConstantBuffers(bindings); + DefineConstantBufferIndirect(); DefineStorageBuffers(bindings); SetupImages(bindings); SetupTextures(bindings); @@ -436,6 +437,24 @@ void EmitContext::DefineConstantBuffers(Bindings& bindings) { } } +void EmitContext::DefineConstantBufferIndirect() { + if (!info.uses_cbuf_indirect) { + return; + } + + header += profile.has_gl_cbuf_ftou_bug ? "uvec4 " : "vec4 "; + header += "GetCbufIndirect(uint binding, uint offset){" + "switch(binding){" + "default:"; + + for (const auto& desc : info.constant_buffer_descriptors) { + header += + fmt::format("case {}:return {}_cbuf{}[offset];", desc.index, stage_name, desc.index); + } + + header += "}}"; +} + void EmitContext::DefineStorageBuffers(Bindings& bindings) { if (info.storage_buffers_descriptors.empty()) { return; diff --git a/src/shader_recompiler/backend/glsl/glsl_emit_context.h b/src/shader_recompiler/backend/glsl/glsl_emit_context.h index d9b639d29..2b13db6e6 100644 --- a/src/shader_recompiler/backend/glsl/glsl_emit_context.h +++ b/src/shader_recompiler/backend/glsl/glsl_emit_context.h @@ -162,6 +162,7 @@ public: private: void SetupExtensions(); void DefineConstantBuffers(Bindings& bindings); + void DefineConstantBufferIndirect(); void DefineStorageBuffers(Bindings& bindings); void DefineGenericOutput(size_t index, u32 invocations); void DefineHelperFunctions(); diff --git a/src/shader_recompiler/backend/spirv/spirv_emit_context.cpp b/src/shader_recompiler/backend/spirv/spirv_emit_context.cpp index 28f6a6184..9c83cd2e4 100644 --- a/src/shader_recompiler/backend/spirv/spirv_emit_context.cpp +++ b/src/shader_recompiler/backend/spirv/spirv_emit_context.cpp @@ -1043,15 +1043,15 @@ void EmitContext::DefineConstantBufferIndirectFunctions(const Info& info) { const Id merge_label{OpLabel()}; const Id uniform_type{uniform_types.*member_ptr}; - std::array<Id, Info::MAX_CBUFS> buf_labels; - std::array<Sirit::Literal, Info::MAX_CBUFS> buf_literals; - for (u32 i = 0; i < Info::MAX_CBUFS; i++) { + std::array<Id, Info::MAX_INDIRECT_CBUFS> buf_labels; + std::array<Sirit::Literal, Info::MAX_INDIRECT_CBUFS> buf_literals; + for (u32 i = 0; i < Info::MAX_INDIRECT_CBUFS; i++) { buf_labels[i] = OpLabel(); buf_literals[i] = Sirit::Literal{i}; } OpSelectionMerge(merge_label, spv::SelectionControlMask::MaskNone); OpSwitch(binding, buf_labels[0], buf_literals, buf_labels); - for (u32 i = 0; i < Info::MAX_CBUFS; i++) { + for (u32 i = 0; i < Info::MAX_INDIRECT_CBUFS; i++) { AddLabel(buf_labels[i]); const Id cbuf{cbufs[i].*member_ptr}; const Id access_chain{OpAccessChain(uniform_type, cbuf, u32_zero_value, offset)}; @@ -1064,22 +1064,23 @@ void EmitContext::DefineConstantBufferIndirectFunctions(const Info& info) { return func; }}; IR::Type types{info.used_indirect_cbuf_types}; - if (True(types & IR::Type::U8)) { + bool supports_aliasing = profile.support_descriptor_aliasing; + if (supports_aliasing && True(types & IR::Type::U8)) { load_const_func_u8 = make_accessor(U8, &UniformDefinitions::U8); } - if (True(types & IR::Type::U16)) { + if (supports_aliasing && True(types & IR::Type::U16)) { load_const_func_u16 = make_accessor(U16, &UniformDefinitions::U16); } - if (True(types & IR::Type::F32)) { + if (supports_aliasing && True(types & IR::Type::F32)) { load_const_func_f32 = make_accessor(F32[1], &UniformDefinitions::F32); } - if (True(types & IR::Type::U32)) { + if (supports_aliasing && True(types & IR::Type::U32)) { load_const_func_u32 = make_accessor(U32[1], &UniformDefinitions::U32); } - if (True(types & IR::Type::U32x2)) { + if (supports_aliasing && True(types & IR::Type::U32x2)) { load_const_func_u32x2 = make_accessor(U32[2], &UniformDefinitions::U32x2); } - if (True(types & IR::Type::U32x4)) { + if (!supports_aliasing || True(types & IR::Type::U32x4)) { load_const_func_u32x4 = make_accessor(U32[4], &UniformDefinitions::U32x4); } } diff --git a/src/shader_recompiler/ir_opt/collect_shader_info_pass.cpp b/src/shader_recompiler/ir_opt/collect_shader_info_pass.cpp index 0b2c60842..16278faab 100644 --- a/src/shader_recompiler/ir_opt/collect_shader_info_pass.cpp +++ b/src/shader_recompiler/ir_opt/collect_shader_info_pass.cpp @@ -32,13 +32,8 @@ void AddConstantBufferDescriptor(Info& info, u32 index, u32 count) { void AddRegisterIndexedLdc(Info& info) { info.uses_cbuf_indirect = true; - // The shader can use any possible constant buffer - info.constant_buffer_mask = (1 << Info::MAX_CBUFS) - 1; - - auto& cbufs{info.constant_buffer_descriptors}; - cbufs.clear(); - for (u32 i = 0; i < Info::MAX_CBUFS; i++) { - cbufs.push_back(ConstantBufferDescriptor{.index = i, .count = 1}); + for (u32 i = 0; i < Info::MAX_INDIRECT_CBUFS; i++) { + AddConstantBufferDescriptor(info, i, 1); // The shader can use any possible access size info.constant_buffer_used_sizes[i] = 0x10'000; diff --git a/src/shader_recompiler/shader_info.h b/src/shader_recompiler/shader_info.h index 9d36bd9eb..a3a09c71c 100644 --- a/src/shader_recompiler/shader_info.h +++ b/src/shader_recompiler/shader_info.h @@ -105,6 +105,7 @@ struct ImageDescriptor { using ImageDescriptors = boost::container::small_vector<ImageDescriptor, 4>; struct Info { + static constexpr size_t MAX_INDIRECT_CBUFS{14}; static constexpr size_t MAX_CBUFS{18}; static constexpr size_t MAX_SSBOS{32}; diff --git a/src/video_core/gpu.cpp b/src/video_core/gpu.cpp index ba9ba082f..789af452d 100644 --- a/src/video_core/gpu.cpp +++ b/src/video_core/gpu.cpp @@ -230,7 +230,7 @@ struct GPU::Impl { void IncrementSyncPoint(u32 syncpoint_id) { auto& syncpoint = syncpoints.at(syncpoint_id); syncpoint++; - std::lock_guard lock{sync_mutex}; + std::scoped_lock lock{sync_mutex}; sync_cv.notify_all(); auto& interrupt = syncpt_interrupts.at(syncpoint_id); if (!interrupt.empty()) { @@ -252,7 +252,7 @@ struct GPU::Impl { } void RegisterSyncptInterrupt(u32 syncpoint_id, u32 value) { - std::lock_guard lock{sync_mutex}; + std::scoped_lock lock{sync_mutex}; auto& interrupt = syncpt_interrupts.at(syncpoint_id); bool contains = std::any_of(interrupt.begin(), interrupt.end(), [value](u32 in_value) { return in_value == value; }); @@ -263,7 +263,7 @@ struct GPU::Impl { } [[nodiscard]] bool CancelSyncptInterrupt(u32 syncpoint_id, u32 value) { - std::lock_guard lock{sync_mutex}; + std::scoped_lock lock{sync_mutex}; auto& interrupt = syncpt_interrupts.at(syncpoint_id); const auto iter = std::find_if(interrupt.begin(), interrupt.end(), diff --git a/src/video_core/gpu_thread.cpp b/src/video_core/gpu_thread.cpp index 9547f277a..4e8999915 100644 --- a/src/video_core/gpu_thread.cpp +++ b/src/video_core/gpu_thread.cpp @@ -56,7 +56,7 @@ static void RunThread(std::stop_token stop_token, Core::System& system, if (next.block) { // We have to lock the write_lock to ensure that the condition_variable wait not get a // race between the check and the lock itself. - std::lock_guard lk(state.write_lock); + std::scoped_lock lk{state.write_lock}; state.cv.notify_all(); } } diff --git a/src/video_core/host_shaders/CMakeLists.txt b/src/video_core/host_shaders/CMakeLists.txt index af05d47d1..190fc6aea 100644 --- a/src/video_core/host_shaders/CMakeLists.txt +++ b/src/video_core/host_shaders/CMakeLists.txt @@ -18,6 +18,7 @@ set(SHADER_FILES full_screen_triangle.vert fxaa.frag fxaa.vert + opengl_convert_s8d24.comp opengl_copy_bc4.comp opengl_present.frag opengl_present.vert diff --git a/src/video_core/host_shaders/opengl_convert_s8d24.comp b/src/video_core/host_shaders/opengl_convert_s8d24.comp new file mode 100644 index 000000000..83e1ab176 --- /dev/null +++ b/src/video_core/host_shaders/opengl_convert_s8d24.comp @@ -0,0 +1,18 @@ +// Copyright 2022 yuzu Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#version 430 core + +layout(local_size_x = 16, local_size_y = 8) in; + +layout(binding = 0, rgba8ui) restrict uniform uimage2D destination; +layout(location = 0) uniform uvec3 size; + +void main() { + if (any(greaterThanEqual(gl_GlobalInvocationID, size))) { + return; + } + uvec4 components = imageLoad(destination, ivec2(gl_GlobalInvocationID.xy)); + imageStore(destination, ivec2(gl_GlobalInvocationID.xy), components.wxyz); +} diff --git a/src/video_core/renderer_opengl/gl_graphics_pipeline.cpp b/src/video_core/renderer_opengl/gl_graphics_pipeline.cpp index 9e6732abd..fd40966d5 100644 --- a/src/video_core/renderer_opengl/gl_graphics_pipeline.cpp +++ b/src/video_core/renderer_opengl/gl_graphics_pipeline.cpp @@ -253,7 +253,7 @@ GraphicsPipeline::GraphicsPipeline( } } if (in_parallel) { - std::lock_guard lock{built_mutex}; + std::scoped_lock lock{built_mutex}; built_fence.Create(); // Flush this context to ensure compilation commands and fence are in the GPU pipe. glFlush(); diff --git a/src/video_core/renderer_opengl/gl_rasterizer.cpp b/src/video_core/renderer_opengl/gl_rasterizer.cpp index e6f9ece8b..7ab7f0c0a 100644 --- a/src/video_core/renderer_opengl/gl_rasterizer.cpp +++ b/src/video_core/renderer_opengl/gl_rasterizer.cpp @@ -520,6 +520,8 @@ bool RasterizerOpenGL::AccelerateDisplay(const Tegra::FramebufferConfig& config, // ASSERT_MSG(image_view->size.width == config.width, "Framebuffer width is different"); // ASSERT_MSG(image_view->size.height == config.height, "Framebuffer height is different"); + screen_info.texture.width = image_view->size.width; + screen_info.texture.height = image_view->size.height; screen_info.display_texture = image_view->Handle(Shader::TextureType::Color2D); screen_info.display_srgb = VideoCore::Surface::IsPixelFormatSRGB(image_view->format); return true; diff --git a/src/video_core/renderer_opengl/gl_shader_cache.cpp b/src/video_core/renderer_opengl/gl_shader_cache.cpp index 6423992c3..05c5e702c 100644 --- a/src/video_core/renderer_opengl/gl_shader_cache.cpp +++ b/src/video_core/renderer_opengl/gl_shader_cache.cpp @@ -258,7 +258,7 @@ void ShaderCache::LoadDiskResources(u64 title_id, std::stop_token stop_loading, [this, key, env = std::move(env), &state, &callback](Context* ctx) mutable { ctx->pools.ReleaseContents(); auto pipeline{CreateComputePipeline(ctx->pools, key, env)}; - std::lock_guard lock{state.mutex}; + std::scoped_lock lock{state.mutex}; if (pipeline) { compute_cache.emplace(key, std::move(pipeline)); } @@ -280,7 +280,7 @@ void ShaderCache::LoadDiskResources(u64 title_id, std::stop_token stop_loading, } ctx->pools.ReleaseContents(); auto pipeline{CreateGraphicsPipeline(ctx->pools, key, MakeSpan(env_ptrs), false)}; - std::lock_guard lock{state.mutex}; + std::scoped_lock lock{state.mutex}; if (pipeline) { graphics_cache.emplace(key, std::move(pipeline)); } diff --git a/src/video_core/renderer_opengl/gl_texture_cache.cpp b/src/video_core/renderer_opengl/gl_texture_cache.cpp index 8f9a65beb..d12076358 100644 --- a/src/video_core/renderer_opengl/gl_texture_cache.cpp +++ b/src/video_core/renderer_opengl/gl_texture_cache.cpp @@ -409,8 +409,8 @@ ImageBufferMap::~ImageBufferMap() { TextureCacheRuntime::TextureCacheRuntime(const Device& device_, ProgramManager& program_manager, StateTracker& state_tracker_) - : device{device_}, state_tracker{state_tracker_}, - util_shaders(program_manager), resolution{Settings::values.resolution_info} { + : device{device_}, state_tracker{state_tracker_}, util_shaders(program_manager), + format_conversion_pass{util_shaders}, resolution{Settings::values.resolution_info} { static constexpr std::array TARGETS{GL_TEXTURE_1D_ARRAY, GL_TEXTURE_2D_ARRAY, GL_TEXTURE_3D}; for (size_t i = 0; i < TARGETS.size(); ++i) { const GLenum target = TARGETS[i]; @@ -1325,6 +1325,9 @@ Framebuffer::Framebuffer(TextureCacheRuntime& runtime, std::span<ImageView*, NUM Framebuffer::~Framebuffer() = default; +FormatConversionPass::FormatConversionPass(UtilShaders& util_shaders_) + : util_shaders{util_shaders_} {} + void FormatConversionPass::ConvertImage(Image& dst_image, Image& src_image, std::span<const VideoCommon::ImageCopy> copies) { const GLenum dst_target = ImageTarget(dst_image.info); @@ -1357,6 +1360,12 @@ void FormatConversionPass::ConvertImage(Image& dst_image, Image& src_image, dst_origin.z, region.width, region.height, region.depth, dst_image.GlFormat(), dst_image.GlType(), nullptr); } + + // Swap component order of S8D24 to ABGR8 reinterprets + if (src_image.info.format == PixelFormat::D24_UNORM_S8_UINT && + dst_image.info.format == PixelFormat::A8B8G8R8_UNORM) { + util_shaders.ConvertS8D24(dst_image, copies); + } } } // namespace OpenGL diff --git a/src/video_core/renderer_opengl/gl_texture_cache.h b/src/video_core/renderer_opengl/gl_texture_cache.h index 53088b66e..672fa8dde 100644 --- a/src/video_core/renderer_opengl/gl_texture_cache.h +++ b/src/video_core/renderer_opengl/gl_texture_cache.h @@ -55,13 +55,14 @@ struct FormatProperties { class FormatConversionPass { public: - FormatConversionPass() = default; + explicit FormatConversionPass(UtilShaders& util_shaders); ~FormatConversionPass() = default; void ConvertImage(Image& dst_image, Image& src_image, std::span<const VideoCommon::ImageCopy> copies); private: + UtilShaders& util_shaders; OGLBuffer intermediate_pbo; size_t pbo_size{}; }; diff --git a/src/video_core/renderer_opengl/renderer_opengl.cpp b/src/video_core/renderer_opengl/renderer_opengl.cpp index f8f29013a..3a3c213bb 100644 --- a/src/video_core/renderer_opengl/renderer_opengl.cpp +++ b/src/video_core/renderer_opengl/renderer_opengl.cpp @@ -208,6 +208,8 @@ void RendererOpenGL::LoadFBToScreenInfo(const Tegra::FramebufferConfig& framebuf // Framebuffer orientation handling framebuffer_transform_flags = framebuffer.transform_flags; framebuffer_crop_rect = framebuffer.crop_rect; + framebuffer_width = framebuffer.width; + framebuffer_height = framebuffer.height; const VAddr framebuffer_addr{framebuffer.address + framebuffer.offset}; screen_info.was_accelerated = @@ -480,9 +482,12 @@ void RendererOpenGL::DrawScreen(const Layout::FramebufferLayout& layout) { ASSERT_MSG(framebuffer_crop_rect.top == 0, "Unimplemented"); ASSERT_MSG(framebuffer_crop_rect.left == 0, "Unimplemented"); + f32 scale_u = static_cast<f32>(framebuffer_width) / static_cast<f32>(screen_info.texture.width); + f32 scale_v = + static_cast<f32>(framebuffer_height) / static_cast<f32>(screen_info.texture.height); + // Scale the output by the crop width/height. This is commonly used with 1280x720 rendering // (e.g. handheld mode) on a 1920x1080 framebuffer. - f32 scale_u = 1.f, scale_v = 1.f; if (framebuffer_crop_rect.GetWidth() > 0) { scale_u = static_cast<f32>(framebuffer_crop_rect.GetWidth()) / static_cast<f32>(screen_info.texture.width); diff --git a/src/video_core/renderer_opengl/renderer_opengl.h b/src/video_core/renderer_opengl/renderer_opengl.h index aa206878b..ae9558a33 100644 --- a/src/video_core/renderer_opengl/renderer_opengl.h +++ b/src/video_core/renderer_opengl/renderer_opengl.h @@ -137,6 +137,8 @@ private: /// Used for transforming the framebuffer orientation Service::android::BufferTransformFlags framebuffer_transform_flags{}; Common::Rectangle<int> framebuffer_crop_rect; + u32 framebuffer_width; + u32 framebuffer_height; }; } // namespace OpenGL diff --git a/src/video_core/renderer_opengl/util_shaders.cpp b/src/video_core/renderer_opengl/util_shaders.cpp index 897c380b3..04c482a09 100644 --- a/src/video_core/renderer_opengl/util_shaders.cpp +++ b/src/video_core/renderer_opengl/util_shaders.cpp @@ -13,6 +13,7 @@ #include "video_core/host_shaders/astc_decoder_comp.h" #include "video_core/host_shaders/block_linear_unswizzle_2d_comp.h" #include "video_core/host_shaders/block_linear_unswizzle_3d_comp.h" +#include "video_core/host_shaders/opengl_convert_s8d24_comp.h" #include "video_core/host_shaders/opengl_copy_bc4_comp.h" #include "video_core/host_shaders/pitch_unswizzle_comp.h" #include "video_core/renderer_opengl/gl_shader_manager.h" @@ -50,7 +51,8 @@ UtilShaders::UtilShaders(ProgramManager& program_manager_) block_linear_unswizzle_2d_program(MakeProgram(BLOCK_LINEAR_UNSWIZZLE_2D_COMP)), block_linear_unswizzle_3d_program(MakeProgram(BLOCK_LINEAR_UNSWIZZLE_3D_COMP)), pitch_unswizzle_program(MakeProgram(PITCH_UNSWIZZLE_COMP)), - copy_bc4_program(MakeProgram(OPENGL_COPY_BC4_COMP)) { + copy_bc4_program(MakeProgram(OPENGL_COPY_BC4_COMP)), + convert_s8d24_program(MakeProgram(OPENGL_CONVERT_S8D24_COMP)) { const auto swizzle_table = Tegra::Texture::MakeSwizzleTable(); swizzle_table_buffer.Create(); glNamedBufferStorage(swizzle_table_buffer.handle, sizeof(swizzle_table), &swizzle_table, 0); @@ -248,6 +250,26 @@ void UtilShaders::CopyBC4(Image& dst_image, Image& src_image, std::span<const Im program_manager.RestoreGuestCompute(); } +void UtilShaders::ConvertS8D24(Image& dst_image, std::span<const ImageCopy> copies) { + static constexpr GLuint BINDING_DESTINATION = 0; + static constexpr GLuint LOC_SIZE = 0; + + program_manager.BindComputeProgram(convert_s8d24_program.handle); + for (const ImageCopy& copy : copies) { + ASSERT(copy.src_subresource.base_layer == 0); + ASSERT(copy.src_subresource.num_layers == 1); + ASSERT(copy.dst_subresource.base_layer == 0); + ASSERT(copy.dst_subresource.num_layers == 1); + + glUniform3ui(LOC_SIZE, copy.extent.width, copy.extent.height, copy.extent.depth); + glBindImageTexture(BINDING_DESTINATION, dst_image.StorageHandle(), + copy.dst_subresource.base_level, GL_TRUE, 0, GL_READ_WRITE, GL_RGBA8UI); + glDispatchCompute(Common::DivCeil(copy.extent.width, 16u), + Common::DivCeil(copy.extent.height, 8u), copy.extent.depth); + } + program_manager.RestoreGuestCompute(); +} + GLenum StoreFormat(u32 bytes_per_block) { switch (bytes_per_block) { case 1: diff --git a/src/video_core/renderer_opengl/util_shaders.h b/src/video_core/renderer_opengl/util_shaders.h index 5de95ea7a..5c132e67f 100644 --- a/src/video_core/renderer_opengl/util_shaders.h +++ b/src/video_core/renderer_opengl/util_shaders.h @@ -39,6 +39,8 @@ public: void CopyBC4(Image& dst_image, Image& src_image, std::span<const VideoCommon::ImageCopy> copies); + void ConvertS8D24(Image& dst_image, std::span<const VideoCommon::ImageCopy> copies); + private: ProgramManager& program_manager; @@ -49,6 +51,7 @@ private: OGLProgram block_linear_unswizzle_3d_program; OGLProgram pitch_unswizzle_program; OGLProgram copy_bc4_program; + OGLProgram convert_s8d24_program; }; GLenum StoreFormat(u32 bytes_per_block); diff --git a/src/video_core/renderer_vulkan/pipeline_statistics.cpp b/src/video_core/renderer_vulkan/pipeline_statistics.cpp index bfec931a6..7ccadf084 100644 --- a/src/video_core/renderer_vulkan/pipeline_statistics.cpp +++ b/src/video_core/renderer_vulkan/pipeline_statistics.cpp @@ -57,7 +57,7 @@ void PipelineStatistics::Collect(VkPipeline pipeline) { stage_stats.basic_block_count = GetUint64(statistic); } } - std::lock_guard lock{mutex}; + std::scoped_lock lock{mutex}; collected_stats.push_back(stage_stats); } } @@ -66,7 +66,7 @@ void PipelineStatistics::Report() const { double num{}; Stats total; { - std::lock_guard lock{mutex}; + std::scoped_lock lock{mutex}; for (const Stats& stats : collected_stats) { total.code_size += stats.code_size; total.register_count += stats.register_count; diff --git a/src/video_core/renderer_vulkan/vk_blit_screen.cpp b/src/video_core/renderer_vulkan/vk_blit_screen.cpp index d893c1952..b866e9103 100644 --- a/src/video_core/renderer_vulkan/vk_blit_screen.cpp +++ b/src/video_core/renderer_vulkan/vk_blit_screen.cpp @@ -1406,8 +1406,9 @@ void VKBlitScreen::SetVertexData(BufferData& data, const Tegra::FramebufferConfi UNIMPLEMENTED_IF(framebuffer_crop_rect.top != 0); UNIMPLEMENTED_IF(framebuffer_crop_rect.left != 0); - f32 scale_u = 1.0f; - f32 scale_v = 1.0f; + f32 scale_u = static_cast<f32>(framebuffer.width) / static_cast<f32>(screen_info.width); + f32 scale_v = static_cast<f32>(framebuffer.height) / static_cast<f32>(screen_info.height); + // Scale the output by the crop width/height. This is commonly used with 1280x720 rendering // (e.g. handheld mode) on a 1920x1080 framebuffer. if (!fsr) { diff --git a/src/video_core/renderer_vulkan/vk_compute_pipeline.cpp b/src/video_core/renderer_vulkan/vk_compute_pipeline.cpp index de36bcdb7..97b3594c2 100644 --- a/src/video_core/renderer_vulkan/vk_compute_pipeline.cpp +++ b/src/video_core/renderer_vulkan/vk_compute_pipeline.cpp @@ -77,7 +77,7 @@ ComputePipeline::ComputePipeline(const Device& device_, DescriptorPool& descript if (pipeline_statistics) { pipeline_statistics->Collect(*pipeline); } - std::lock_guard lock{build_mutex}; + std::scoped_lock lock{build_mutex}; is_built = true; build_condvar.notify_one(); if (shader_notify) { diff --git a/src/video_core/renderer_vulkan/vk_graphics_pipeline.cpp b/src/video_core/renderer_vulkan/vk_graphics_pipeline.cpp index d514b71d0..8959d6059 100644 --- a/src/video_core/renderer_vulkan/vk_graphics_pipeline.cpp +++ b/src/video_core/renderer_vulkan/vk_graphics_pipeline.cpp @@ -258,7 +258,7 @@ GraphicsPipeline::GraphicsPipeline( pipeline_statistics->Collect(*pipeline); } - std::lock_guard lock{build_mutex}; + std::scoped_lock lock{build_mutex}; is_built = true; build_condvar.notify_one(); if (shader_notify) { diff --git a/src/video_core/renderer_vulkan/vk_pipeline_cache.cpp b/src/video_core/renderer_vulkan/vk_pipeline_cache.cpp index 27e59df73..336d1e9dc 100644 --- a/src/video_core/renderer_vulkan/vk_pipeline_cache.cpp +++ b/src/video_core/renderer_vulkan/vk_pipeline_cache.cpp @@ -404,7 +404,7 @@ void PipelineCache::LoadDiskResources(u64 title_id, std::stop_token stop_loading workers.QueueWork([this, key, env = std::move(env), &state, &callback]() mutable { ShaderPools pools; auto pipeline{CreateComputePipeline(pools, key, env, state.statistics.get(), false)}; - std::lock_guard lock{state.mutex}; + std::scoped_lock lock{state.mutex}; if (pipeline) { compute_cache.emplace(key, std::move(pipeline)); } @@ -434,7 +434,7 @@ void PipelineCache::LoadDiskResources(u64 title_id, std::stop_token stop_loading auto pipeline{CreateGraphicsPipeline(pools, key, MakeSpan(env_ptrs), state.statistics.get(), false)}; - std::lock_guard lock{state.mutex}; + std::scoped_lock lock{state.mutex}; graphics_cache.emplace(key, std::move(pipeline)); ++state.built; if (state.has_loaded) { diff --git a/src/video_core/renderer_vulkan/vk_render_pass_cache.cpp b/src/video_core/renderer_vulkan/vk_render_pass_cache.cpp index 451ffe019..d22bb6694 100644 --- a/src/video_core/renderer_vulkan/vk_render_pass_cache.cpp +++ b/src/video_core/renderer_vulkan/vk_render_pass_cache.cpp @@ -36,7 +36,7 @@ VkAttachmentDescription AttachmentDescription(const Device& device, PixelFormat RenderPassCache::RenderPassCache(const Device& device_) : device{&device_} {} VkRenderPass RenderPassCache::Get(const RenderPassKey& key) { - std::lock_guard lock{mutex}; + std::scoped_lock lock{mutex}; const auto [pair, is_new] = cache.try_emplace(key); if (!is_new) { return *pair->second; diff --git a/src/video_core/renderer_vulkan/vk_scheduler.cpp b/src/video_core/renderer_vulkan/vk_scheduler.cpp index ad320991b..6a9416457 100644 --- a/src/video_core/renderer_vulkan/vk_scheduler.cpp +++ b/src/video_core/renderer_vulkan/vk_scheduler.cpp @@ -73,7 +73,7 @@ void VKScheduler::DispatchWork() { return; } { - std::lock_guard lock{work_mutex}; + std::scoped_lock lock{work_mutex}; work_queue.push(std::move(chunk)); } work_cv.notify_one(); @@ -157,7 +157,7 @@ void VKScheduler::WorkerThread(std::stop_token stop_token) { if (has_submit) { AllocateWorkerCommandBuffer(); } - std::lock_guard reserve_lock{reserve_mutex}; + std::scoped_lock reserve_lock{reserve_mutex}; chunk_reserve.push_back(std::move(work)); } while (!stop_token.stop_requested()); } @@ -282,7 +282,7 @@ void VKScheduler::EndRenderPass() { } void VKScheduler::AcquireNewChunk() { - std::lock_guard lock{reserve_mutex}; + std::scoped_lock lock{reserve_mutex}; if (chunk_reserve.empty()) { chunk = std::make_unique<CommandChunk>(); return; diff --git a/src/video_core/shader_cache.cpp b/src/video_core/shader_cache.cpp index 87636857d..75031767a 100644 --- a/src/video_core/shader_cache.cpp +++ b/src/video_core/shader_cache.cpp @@ -25,7 +25,7 @@ void ShaderCache::InvalidateRegion(VAddr addr, size_t size) { } void ShaderCache::OnCPUWrite(VAddr addr, size_t size) { - std::lock_guard lock{invalidation_mutex}; + std::scoped_lock lock{invalidation_mutex}; InvalidatePagesInRegion(addr, size); } diff --git a/src/web_service/web_backend.cpp b/src/web_service/web_backend.cpp index 4208bd044..58b0c2f10 100644 --- a/src/web_service/web_backend.cpp +++ b/src/web_service/web_backend.cpp @@ -32,7 +32,7 @@ constexpr std::size_t TIMEOUT_SECONDS = 30; struct Client::Impl { Impl(std::string host, std::string username, std::string token) : host{std::move(host)}, username{std::move(username)}, token{std::move(token)} { - std::lock_guard lock{jwt_cache.mutex}; + std::scoped_lock lock{jwt_cache.mutex}; if (this->username == jwt_cache.username && this->token == jwt_cache.token) { jwt = jwt_cache.jwt; } @@ -147,7 +147,7 @@ struct Client::Impl { if (result.result_code != WebResult::Code::Success) { LOG_ERROR(WebService, "UpdateJWT failed"); } else { - std::lock_guard lock{jwt_cache.mutex}; + std::scoped_lock lock{jwt_cache.mutex}; jwt_cache.username = username; jwt_cache.token = token; jwt_cache.jwt = jwt = result.returned_data; diff --git a/src/yuzu/util/controller_navigation.cpp b/src/yuzu/util/controller_navigation.cpp index c2b13123d..9a1868cae 100644 --- a/src/yuzu/util/controller_navigation.cpp +++ b/src/yuzu/util/controller_navigation.cpp @@ -39,7 +39,7 @@ void ControllerNavigation::TriggerButton(Settings::NativeButton::Values native_b } void ControllerNavigation::ControllerUpdateEvent(Core::HID::ControllerTriggerType type) { - std::lock_guard lock{mutex}; + std::scoped_lock lock{mutex}; if (!Settings::values.controller_navigation) { return; } |