summaryrefslogtreecommitdiffstats
path: root/src
diff options
context:
space:
mode:
Diffstat (limited to 'src')
-rw-r--r--src/core/hle/service/sockets/bsd.cpp48
-rw-r--r--src/core/hle/service/sockets/bsd.h4
-rw-r--r--src/video_core/engines/shader_bytecode.h30
-rw-r--r--src/video_core/renderer_opengl/gl_shader_decompiler.cpp6
-rw-r--r--src/video_core/renderer_opengl/gl_texture_cache.cpp40
-rw-r--r--src/video_core/renderer_vulkan/vk_shader_decompiler.cpp66
-rw-r--r--src/video_core/shader/decode/memory.cpp22
-rw-r--r--src/video_core/shader/node.h2
-rw-r--r--src/video_core/texture_cache/surface_base.cpp2
9 files changed, 168 insertions, 52 deletions
diff --git a/src/core/hle/service/sockets/bsd.cpp b/src/core/hle/service/sockets/bsd.cpp
index 884ad173b..f67fab2f9 100644
--- a/src/core/hle/service/sockets/bsd.cpp
+++ b/src/core/hle/service/sockets/bsd.cpp
@@ -42,6 +42,26 @@ void BSD::Socket(Kernel::HLERequestContext& ctx) {
rb.Push<u32>(0); // bsd errno
}
+void BSD::Select(Kernel::HLERequestContext& ctx) {
+ LOG_WARNING(Service, "(STUBBED) called");
+
+ IPC::ResponseBuilder rb{ctx, 4};
+
+ rb.Push(RESULT_SUCCESS);
+ rb.Push<u32>(0); // ret
+ rb.Push<u32>(0); // bsd errno
+}
+
+void BSD::Bind(Kernel::HLERequestContext& ctx) {
+ LOG_WARNING(Service, "(STUBBED) called");
+
+ IPC::ResponseBuilder rb{ctx, 4};
+
+ rb.Push(RESULT_SUCCESS);
+ rb.Push<u32>(0); // ret
+ rb.Push<u32>(0); // bsd errno
+}
+
void BSD::Connect(Kernel::HLERequestContext& ctx) {
LOG_WARNING(Service, "(STUBBED) called");
@@ -52,6 +72,26 @@ void BSD::Connect(Kernel::HLERequestContext& ctx) {
rb.Push<u32>(0); // bsd errno
}
+void BSD::Listen(Kernel::HLERequestContext& ctx) {
+ LOG_WARNING(Service, "(STUBBED) called");
+
+ IPC::ResponseBuilder rb{ctx, 4};
+
+ rb.Push(RESULT_SUCCESS);
+ rb.Push<u32>(0); // ret
+ rb.Push<u32>(0); // bsd errno
+}
+
+void BSD::SetSockOpt(Kernel::HLERequestContext& ctx) {
+ LOG_WARNING(Service, "(STUBBED) called");
+
+ IPC::ResponseBuilder rb{ctx, 4};
+
+ rb.Push(RESULT_SUCCESS);
+ rb.Push<u32>(0); // ret
+ rb.Push<u32>(0); // bsd errno
+}
+
void BSD::SendTo(Kernel::HLERequestContext& ctx) {
LOG_WARNING(Service, "(STUBBED) called");
@@ -80,7 +120,7 @@ BSD::BSD(const char* name) : ServiceFramework(name) {
{2, &BSD::Socket, "Socket"},
{3, nullptr, "SocketExempt"},
{4, nullptr, "Open"},
- {5, nullptr, "Select"},
+ {5, &BSD::Select, "Select"},
{6, nullptr, "Poll"},
{7, nullptr, "Sysctl"},
{8, nullptr, "Recv"},
@@ -88,15 +128,15 @@ BSD::BSD(const char* name) : ServiceFramework(name) {
{10, nullptr, "Send"},
{11, &BSD::SendTo, "SendTo"},
{12, nullptr, "Accept"},
- {13, nullptr, "Bind"},
+ {13, &BSD::Bind, "Bind"},
{14, &BSD::Connect, "Connect"},
{15, nullptr, "GetPeerName"},
{16, nullptr, "GetSockName"},
{17, nullptr, "GetSockOpt"},
- {18, nullptr, "Listen"},
+ {18, &BSD::Listen, "Listen"},
{19, nullptr, "Ioctl"},
{20, nullptr, "Fcntl"},
- {21, nullptr, "SetSockOpt"},
+ {21, &BSD::SetSockOpt, "SetSockOpt"},
{22, nullptr, "Shutdown"},
{23, nullptr, "ShutdownAllSockets"},
{24, nullptr, "Write"},
diff --git a/src/core/hle/service/sockets/bsd.h b/src/core/hle/service/sockets/bsd.h
index 0fe0e65c6..3098e3baf 100644
--- a/src/core/hle/service/sockets/bsd.h
+++ b/src/core/hle/service/sockets/bsd.h
@@ -18,7 +18,11 @@ private:
void RegisterClient(Kernel::HLERequestContext& ctx);
void StartMonitoring(Kernel::HLERequestContext& ctx);
void Socket(Kernel::HLERequestContext& ctx);
+ void Select(Kernel::HLERequestContext& ctx);
+ void Bind(Kernel::HLERequestContext& ctx);
void Connect(Kernel::HLERequestContext& ctx);
+ void Listen(Kernel::HLERequestContext& ctx);
+ void SetSockOpt(Kernel::HLERequestContext& ctx);
void SendTo(Kernel::HLERequestContext& ctx);
void Close(Kernel::HLERequestContext& ctx);
diff --git a/src/video_core/engines/shader_bytecode.h b/src/video_core/engines/shader_bytecode.h
index 6f98bd827..f443ec0fe 100644
--- a/src/video_core/engines/shader_bytecode.h
+++ b/src/video_core/engines/shader_bytecode.h
@@ -227,6 +227,28 @@ enum class AtomicOp : u64 {
Exch = 8,
};
+enum class GlobalAtomicOp : u64 {
+ Add = 0,
+ Min = 1,
+ Max = 2,
+ Inc = 3,
+ Dec = 4,
+ And = 5,
+ Or = 6,
+ Xor = 7,
+ Exch = 8,
+ SafeAdd = 10,
+};
+
+enum class GlobalAtomicType : u64 {
+ U32 = 0,
+ S32 = 1,
+ U64 = 2,
+ F32_FTZ_RN = 3,
+ F16x2_FTZ_RN = 4,
+ S64 = 5,
+};
+
enum class UniformType : u64 {
UnsignedByte = 0,
SignedByte = 1,
@@ -958,6 +980,12 @@ union Instruction {
} stg;
union {
+ BitField<52, 4, GlobalAtomicOp> operation;
+ BitField<49, 3, GlobalAtomicType> type;
+ BitField<28, 20, s64> offset;
+ } atom;
+
+ union {
BitField<52, 4, AtomicOp> operation;
BitField<28, 2, AtomicType> type;
BitField<30, 22, s64> offset;
@@ -1690,6 +1718,7 @@ public:
ST_S,
ST, // Store in generic memory
STG, // Store in global memory
+ ATOM, // Atomic operation on global memory
ATOMS, // Atomic operation on shared memory
AL2P, // Transforms attribute memory into physical memory
TEX,
@@ -1994,6 +2023,7 @@ private:
INST("1110111101010---", Id::ST_L, Type::Memory, "ST_L"),
INST("101-------------", Id::ST, Type::Memory, "ST"),
INST("1110111011011---", Id::STG, Type::Memory, "STG"),
+ INST("11101101--------", Id::ATOM, Type::Memory, "ATOM"),
INST("11101100--------", Id::ATOMS, Type::Memory, "ATOMS"),
INST("1110111110100---", Id::AL2P, Type::Memory, "AL2P"),
INST("110000----111---", Id::TEX, Type::Texture, "TEX"),
diff --git a/src/video_core/renderer_opengl/gl_shader_decompiler.cpp b/src/video_core/renderer_opengl/gl_shader_decompiler.cpp
index 2996aaf08..a1ac3d7a9 100644
--- a/src/video_core/renderer_opengl/gl_shader_decompiler.cpp
+++ b/src/video_core/renderer_opengl/gl_shader_decompiler.cpp
@@ -1019,7 +1019,6 @@ private:
}
return {{"gl_ViewportIndex", Type::Int}};
case 3:
- UNIMPLEMENTED_MSG("Requires some state changes for gl_PointSize to work in shader");
return {{"gl_PointSize", Type::Float}};
}
return {};
@@ -1858,10 +1857,7 @@ private:
template <const std::string_view& opname, Type type>
Expression Atomic(Operation operation) {
- ASSERT(stage == ShaderType::Compute);
- auto& smem = std::get<SmemNode>(*operation[0]);
-
- return {fmt::format("atomic{}(smem[{} >> 2], {})", opname, Visit(smem.GetAddress()).AsInt(),
+ return {fmt::format("atomic{}({}, {})", opname, Visit(operation[0]).GetCode(),
Visit(operation[1]).As(type)),
type};
}
diff --git a/src/video_core/renderer_opengl/gl_texture_cache.cpp b/src/video_core/renderer_opengl/gl_texture_cache.cpp
index e95eb069e..d4b81cd87 100644
--- a/src/video_core/renderer_opengl/gl_texture_cache.cpp
+++ b/src/video_core/renderer_opengl/gl_texture_cache.cpp
@@ -176,6 +176,19 @@ GLint GetSwizzleSource(SwizzleSource source) {
return GL_NONE;
}
+GLenum GetComponent(PixelFormat format, bool is_first) {
+ switch (format) {
+ case PixelFormat::Z24S8:
+ case PixelFormat::Z32FS8:
+ return is_first ? GL_DEPTH_COMPONENT : GL_STENCIL_INDEX;
+ case PixelFormat::S8Z24:
+ return is_first ? GL_STENCIL_INDEX : GL_DEPTH_COMPONENT;
+ default:
+ UNREACHABLE();
+ return GL_DEPTH_COMPONENT;
+ }
+}
+
void ApplyTextureDefaults(const SurfaceParams& params, GLuint texture) {
if (params.IsBuffer()) {
return;
@@ -184,7 +197,7 @@ void ApplyTextureDefaults(const SurfaceParams& params, GLuint texture) {
glTextureParameteri(texture, GL_TEXTURE_MAG_FILTER, GL_LINEAR);
glTextureParameteri(texture, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE);
glTextureParameteri(texture, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE);
- glTextureParameteri(texture, GL_TEXTURE_MAX_LEVEL, params.num_levels - 1);
+ glTextureParameteri(texture, GL_TEXTURE_MAX_LEVEL, static_cast<GLint>(params.num_levels - 1));
if (params.num_levels == 1) {
glTextureParameterf(texture, GL_TEXTURE_LOD_BIAS, 1000.0f);
}
@@ -416,11 +429,21 @@ void CachedSurfaceView::ApplySwizzle(SwizzleSource x_source, SwizzleSource y_sou
if (new_swizzle == swizzle)
return;
swizzle = new_swizzle;
- const std::array<GLint, 4> gl_swizzle = {GetSwizzleSource(x_source), GetSwizzleSource(y_source),
- GetSwizzleSource(z_source),
- GetSwizzleSource(w_source)};
+ const std::array gl_swizzle = {GetSwizzleSource(x_source), GetSwizzleSource(y_source),
+ GetSwizzleSource(z_source), GetSwizzleSource(w_source)};
const GLuint handle = GetTexture();
- glTextureParameteriv(handle, GL_TEXTURE_SWIZZLE_RGBA, gl_swizzle.data());
+ const PixelFormat format = surface.GetSurfaceParams().pixel_format;
+ switch (format) {
+ case PixelFormat::Z24S8:
+ case PixelFormat::Z32FS8:
+ case PixelFormat::S8Z24:
+ glTextureParameteri(handle, GL_DEPTH_STENCIL_TEXTURE_MODE,
+ GetComponent(format, x_source == SwizzleSource::R));
+ break;
+ default:
+ glTextureParameteriv(handle, GL_TEXTURE_SWIZZLE_RGBA, gl_swizzle.data());
+ break;
+ }
}
OGLTextureView CachedSurfaceView::CreateTextureView() const {
@@ -529,8 +552,11 @@ void TextureCacheOpenGL::ImageBlit(View& src_view, View& dst_view,
const Common::Rectangle<u32>& dst_rect = copy_config.dst_rect;
const bool is_linear = copy_config.filter == Tegra::Engines::Fermi2D::Filter::Linear;
- glBlitFramebuffer(src_rect.left, src_rect.top, src_rect.right, src_rect.bottom, dst_rect.left,
- dst_rect.top, dst_rect.right, dst_rect.bottom, buffers,
+ glBlitFramebuffer(static_cast<GLint>(src_rect.left), static_cast<GLint>(src_rect.top),
+ static_cast<GLint>(src_rect.right), static_cast<GLint>(src_rect.bottom),
+ static_cast<GLint>(dst_rect.left), static_cast<GLint>(dst_rect.top),
+ static_cast<GLint>(dst_rect.right), static_cast<GLint>(dst_rect.bottom),
+ buffers,
is_linear && (buffers == GL_COLOR_BUFFER_BIT) ? GL_LINEAR : GL_NEAREST);
}
diff --git a/src/video_core/renderer_vulkan/vk_shader_decompiler.cpp b/src/video_core/renderer_vulkan/vk_shader_decompiler.cpp
index b53078721..1ab22251e 100644
--- a/src/video_core/renderer_vulkan/vk_shader_decompiler.cpp
+++ b/src/video_core/renderer_vulkan/vk_shader_decompiler.cpp
@@ -1123,15 +1123,7 @@ private:
}
if (const auto gmem = std::get_if<GmemNode>(&*node)) {
- const Id gmem_buffer = global_buffers.at(gmem->GetDescriptor());
- const Id real = AsUint(Visit(gmem->GetRealAddress()));
- const Id base = AsUint(Visit(gmem->GetBaseAddress()));
-
- Id offset = OpISub(t_uint, real, base);
- offset = OpUDiv(t_uint, offset, Constant(t_uint, 4U));
- return {OpLoad(t_float,
- OpAccessChain(t_gmem_float, gmem_buffer, Constant(t_uint, 0U), offset)),
- Type::Float};
+ return {OpLoad(t_uint, GetGlobalMemoryPointer(*gmem)), Type::Uint};
}
if (const auto lmem = std::get_if<LmemNode>(&*node)) {
@@ -1142,10 +1134,7 @@ private:
}
if (const auto smem = std::get_if<SmemNode>(&*node)) {
- Id address = AsUint(Visit(smem->GetAddress()));
- address = OpShiftRightLogical(t_uint, address, Constant(t_uint, 2U));
- const Id pointer = OpAccessChain(t_smem_uint, shared_memory, address);
- return {OpLoad(t_uint, pointer), Type::Uint};
+ return {OpLoad(t_uint, GetSharedMemoryPointer(*smem)), Type::Uint};
}
if (const auto internal_flag = std::get_if<InternalFlagNode>(&*node)) {
@@ -1339,20 +1328,10 @@ private:
target = {OpAccessChain(t_prv_float, local_memory, address), Type::Float};
} else if (const auto smem = std::get_if<SmemNode>(&*dest)) {
- ASSERT(stage == ShaderType::Compute);
- Id address = AsUint(Visit(smem->GetAddress()));
- address = OpShiftRightLogical(t_uint, address, Constant(t_uint, 2U));
- target = {OpAccessChain(t_smem_uint, shared_memory, address), Type::Uint};
+ target = {GetSharedMemoryPointer(*smem), Type::Uint};
} else if (const auto gmem = std::get_if<GmemNode>(&*dest)) {
- const Id real = AsUint(Visit(gmem->GetRealAddress()));
- const Id base = AsUint(Visit(gmem->GetBaseAddress()));
- const Id diff = OpISub(t_uint, real, base);
- const Id offset = OpShiftRightLogical(t_uint, diff, Constant(t_uint, 2));
-
- const Id gmem_buffer = global_buffers.at(gmem->GetDescriptor());
- target = {OpAccessChain(t_gmem_float, gmem_buffer, Constant(t_uint, 0), offset),
- Type::Float};
+ target = {GetGlobalMemoryPointer(*gmem), Type::Uint};
} else {
UNIMPLEMENTED();
@@ -1804,11 +1783,16 @@ private:
return {};
}
- Expression UAtomicAdd(Operation operation) {
- const auto& smem = std::get<SmemNode>(*operation[0]);
- Id address = AsUint(Visit(smem.GetAddress()));
- address = OpShiftRightLogical(t_uint, address, Constant(t_uint, 2U));
- const Id pointer = OpAccessChain(t_smem_uint, shared_memory, address);
+ Expression AtomicAdd(Operation operation) {
+ Id pointer;
+ if (const auto smem = std::get_if<SmemNode>(&*operation[0])) {
+ pointer = GetSharedMemoryPointer(*smem);
+ } else if (const auto gmem = std::get_if<GmemNode>(&*operation[0])) {
+ pointer = GetGlobalMemoryPointer(*gmem);
+ } else {
+ UNREACHABLE();
+ return {Constant(t_uint, 0), Type::Uint};
+ }
const Id scope = Constant(t_uint, static_cast<u32>(spv::Scope::Device));
const Id semantics = Constant(t_uint, 0U);
@@ -2243,6 +2227,22 @@ private:
return {};
}
+ Id GetGlobalMemoryPointer(const GmemNode& gmem) {
+ const Id real = AsUint(Visit(gmem.GetRealAddress()));
+ const Id base = AsUint(Visit(gmem.GetBaseAddress()));
+ const Id diff = OpISub(t_uint, real, base);
+ const Id offset = OpShiftRightLogical(t_uint, diff, Constant(t_uint, 2));
+ const Id buffer = global_buffers.at(gmem.GetDescriptor());
+ return OpAccessChain(t_gmem_uint, buffer, Constant(t_uint, 0), offset);
+ }
+
+ Id GetSharedMemoryPointer(const SmemNode& smem) {
+ ASSERT(stage == ShaderType::Compute);
+ Id address = AsUint(Visit(smem.GetAddress()));
+ address = OpShiftRightLogical(t_uint, address, Constant(t_uint, 2U));
+ return OpAccessChain(t_smem_uint, shared_memory, address);
+ }
+
static constexpr std::array operation_decompilers = {
&SPIRVDecompiler::Assign,
@@ -2389,7 +2389,7 @@ private:
&SPIRVDecompiler::AtomicImageXor,
&SPIRVDecompiler::AtomicImageExchange,
- &SPIRVDecompiler::UAtomicAdd,
+ &SPIRVDecompiler::AtomicAdd,
&SPIRVDecompiler::Branch,
&SPIRVDecompiler::BranchIndirect,
@@ -2485,9 +2485,9 @@ private:
Id t_smem_uint{};
- const Id t_gmem_float = TypePointer(spv::StorageClass::StorageBuffer, t_float);
+ const Id t_gmem_uint = TypePointer(spv::StorageClass::StorageBuffer, t_uint);
const Id t_gmem_array =
- Name(Decorate(TypeRuntimeArray(t_float), spv::Decoration::ArrayStride, 4U), "GmemArray");
+ Name(Decorate(TypeRuntimeArray(t_uint), spv::Decoration::ArrayStride, 4U), "GmemArray");
const Id t_gmem_struct = MemberDecorate(
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);
diff --git a/src/video_core/shader/decode/memory.cpp b/src/video_core/shader/decode/memory.cpp
index 58744d29a..b5fbc4d58 100644
--- a/src/video_core/shader/decode/memory.cpp
+++ b/src/video_core/shader/decode/memory.cpp
@@ -19,6 +19,8 @@ namespace VideoCommon::Shader {
using Tegra::Shader::AtomicOp;
using Tegra::Shader::AtomicType;
using Tegra::Shader::Attribute;
+using Tegra::Shader::GlobalAtomicOp;
+using Tegra::Shader::GlobalAtomicType;
using Tegra::Shader::Instruction;
using Tegra::Shader::OpCode;
using Tegra::Shader::Register;
@@ -360,6 +362,24 @@ u32 ShaderIR::DecodeMemory(NodeBlock& bb, u32 pc) {
}
break;
}
+ case OpCode::Id::ATOM: {
+ UNIMPLEMENTED_IF_MSG(instr.atom.operation != GlobalAtomicOp::Add, "operation={}",
+ static_cast<int>(instr.atom.operation.Value()));
+ UNIMPLEMENTED_IF_MSG(instr.atom.type != GlobalAtomicType::S32, "type={}",
+ static_cast<int>(instr.atom.type.Value()));
+
+ const auto [real_address, base_address, descriptor] =
+ TrackGlobalMemory(bb, instr, true, true);
+ if (!real_address || !base_address) {
+ // Tracking failed, skip atomic.
+ break;
+ }
+
+ Node gmem = MakeNode<GmemNode>(real_address, base_address, descriptor);
+ Node value = Operation(OperationCode::AtomicAdd, std::move(gmem), GetRegister(instr.gpr20));
+ SetRegister(bb, instr.gpr0, std::move(value));
+ break;
+ }
case OpCode::Id::ATOMS: {
UNIMPLEMENTED_IF_MSG(instr.atoms.operation != AtomicOp::Add, "operation={}",
static_cast<int>(instr.atoms.operation.Value()));
@@ -373,7 +393,7 @@ u32 ShaderIR::DecodeMemory(NodeBlock& bb, u32 pc) {
Node memory = GetSharedMemory(std::move(address));
Node data = GetRegister(instr.gpr20);
- Node value = Operation(OperationCode::UAtomicAdd, std::move(memory), std::move(data));
+ Node value = Operation(OperationCode::AtomicAdd, std::move(memory), std::move(data));
SetRegister(bb, instr.gpr0, std::move(value));
break;
}
diff --git a/src/video_core/shader/node.h b/src/video_core/shader/node.h
index 075c7d07c..9af1f0228 100644
--- a/src/video_core/shader/node.h
+++ b/src/video_core/shader/node.h
@@ -162,7 +162,7 @@ enum class OperationCode {
AtomicImageXor, /// (MetaImage, int[N] coords) -> void
AtomicImageExchange, /// (MetaImage, int[N] coords) -> void
- UAtomicAdd, /// (smem, uint) -> uint
+ AtomicAdd, /// (memory, {u}int) -> {u}int
Branch, /// (uint branch_target) -> void
BranchIndirect, /// (uint branch_target) -> void
diff --git a/src/video_core/texture_cache/surface_base.cpp b/src/video_core/texture_cache/surface_base.cpp
index 829268b4c..84469b7ba 100644
--- a/src/video_core/texture_cache/surface_base.cpp
+++ b/src/video_core/texture_cache/surface_base.cpp
@@ -135,7 +135,7 @@ std::vector<CopyParams> SurfaceBaseImpl::BreakDownLayered(const SurfaceParams& i
for (u32 level = 0; level < mipmaps; level++) {
const u32 width = SurfaceParams::IntersectWidth(params, in_params, level, level);
const u32 height = SurfaceParams::IntersectHeight(params, in_params, level, level);
- result.emplace_back(width, height, layer, level);
+ result.emplace_back(0, 0, layer, 0, 0, layer, level, level, width, height, 1);
}
}
return result;
generated by cgit v1.2.3 (git 2.25.1) at 2024-08-19 13:25:59 +0200