Compare commits

...

25 Commits

Author SHA1 Message Date
Subv
636ad34707 MaxwellToGL: Implemented tex wrap mode 1 (Wrap, GL_REPEAT). 2018-04-17 10:17:18 -05:00
Subv
7fc516cc1a MaxwellToGL: Added a TODO and partial implementation of maxwell wrap mode 4 (Clamp, GL_CLAMP).
This clamp mode was removed from OpenGL as of 3.1, we can emulate it by using GL_CLAMP_TO_BORDER to get the border color of the texture, and then manually sampling the edge to mix them in the fragment shader.
2018-04-17 10:16:50 -05:00
bunnei
44e09ba807 Merge pull request #338 from bunnei/unrequire-shared-font
pl_u: Use empty shared font if none is available.
2018-04-15 16:54:36 -04:00
bunnei
d6d7d0989c Merge pull request #337 from Subv/used_buffers
GPU: Don't use explicit binding points when uploading the constbuffers to opengl
2018-04-15 16:30:57 -04:00
bunnei
ac628f139d pl_u: Use empty shared font if none is available.
- Makes games work in lieu of shared_font.bin.
2018-04-15 16:15:34 -04:00
Subv
477aab5960 GPU: Use the same buffer names in the generated GLSL and the buffer uploading code. 2018-04-15 15:02:50 -05:00
Subv
14ac40436e GPU: Don't use explicit binding points when uploading the constbuffers to opengl.
The bindpoints will now be dynamically calculated based on the number of buffers used by the previous shader stage.
2018-04-15 14:14:57 -05:00
bunnei
b8825fbf10 Merge pull request #335 from bunnei/delete-file
fsp_srv: Implement DeleteFile.
2018-04-15 15:13:02 -04:00
bunnei
b60834ac41 Merge pull request #334 from Subv/used_buffers
GPU: Use the buffer hints from the shader decompiler to upload only the necessary const buffers for each shader stage
2018-04-15 13:17:30 -04:00
bunnei
bddad50dd4 fsp_srv: Implement DeleteFile.
- Used by Binding of Isaac.
2018-04-15 13:15:18 -04:00
Subv
e128e90350 GPU: Don't use GetPointer when uploading the constbuffer data to the GPU. 2018-04-15 11:18:09 -05:00
Subv
7da47da66e GPU: Use the buffer hints from the shader decompiler to upload only the necessary const buffers for each shader stage. 2018-04-15 11:15:54 -05:00
bunnei
34264879b3 Merge pull request #333 from bunnei/const-buff-hints
shaders: Expose hints about used const buffers.
2018-04-15 12:12:17 -04:00
bunnei
73d9c494ea shaders: Expose hints about used const buffers. 2018-04-15 11:50:10 -04:00
bunnei
bb0c3fc828 Merge pull request #328 from Subv/constbuffers
GPU: Upload the shader Constant Buffers as SSBOs to the GPU
2018-04-15 03:19:57 -04:00
Subv
c9b511da08 GPU: Upload the entirety of each constbuffer for each shader stage as SSBOs.
We're going to need the shader generator to give us a mapping of the actual used const buffers to properly bind them to the shader.
2018-04-14 23:02:05 -05:00
Subv
1957640ea2 GPU: Allow configuring ssbos in the opengl state manager. 2018-04-14 22:54:23 -05:00
Subv
ae58e46036 GPU: Added a function to determine whether a shader stage is enabled or not. 2018-04-14 22:54:23 -05:00
bunnei
2b9a6b3281 Merge pull request #332 from bunnei/fix-total-mem-usage
vm_manager: Increase GetTotalMemoryUsage value.
2018-04-14 22:29:19 -04:00
bunnei
43f0f163e1 vm_manager: Increase GetTotalMemoryUsage value.
- Gets Binding of Isaac running.
2018-04-14 22:04:10 -04:00
bunnei
9f66cae865 Merge pull request #327 from adityaruplaha/fullscreen-fix
Fix the stuck in fullscreen bug
2018-04-14 21:24:36 -04:00
bunnei
778be45103 Merge pull request #331 from bunnei/fsp-flush
fsp_srv: Implement IFile::Flush.
2018-04-14 21:21:34 -04:00
bunnei
fdca7b5f7a Merge pull request #329 from bunnei/shader-gen-part-1
OpenGL shader generation part 1
2018-04-14 20:40:39 -04:00
bunnei
9cab6809f2 fsp_srv: Implement IFile::Flush. 2018-04-14 19:46:09 -04:00
adityaruplaha
958c98bdae Fix the stuck in fullscreen bug (Original PR: citra-emu/citra#3611) 2018-04-14 16:41:56 +05:30
22 changed files with 370 additions and 77 deletions

View File

@@ -57,10 +57,14 @@ ResultVal<std::unique_ptr<StorageBackend>> Disk_FileSystem::OpenFile(const std::
std::make_unique<Disk_Storage>(std::move(file)));
}
ResultCode Disk_FileSystem::DeleteFile(const Path& path) const {
LOG_WARNING(Service_FS, "(STUBBED) called");
// TODO(bunnei): Use correct error code
return ResultCode(-1);
ResultCode Disk_FileSystem::DeleteFile(const std::string& path) const {
if (!FileUtil::Exists(path)) {
return ERROR_PATH_NOT_FOUND;
}
FileUtil::Delete(path);
return RESULT_SUCCESS;
}
ResultCode Disk_FileSystem::RenameFile(const Path& src_path, const Path& dest_path) const {

View File

@@ -25,7 +25,7 @@ public:
ResultVal<std::unique_ptr<StorageBackend>> OpenFile(const std::string& path,
Mode mode) const override;
ResultCode DeleteFile(const Path& path) const override;
ResultCode DeleteFile(const std::string& path) const override;
ResultCode RenameFile(const Path& src_path, const Path& dest_path) const override;
ResultCode DeleteDirectory(const Path& path) const override;
ResultCode DeleteDirectoryRecursively(const Path& path) const override;

View File

@@ -97,7 +97,7 @@ public:
* @param path Path relative to the archive
* @return Result of the operation
*/
virtual ResultCode DeleteFile(const Path& path) const = 0;
virtual ResultCode DeleteFile(const std::string& path) const = 0;
/**
* Create a directory specified by its path

View File

@@ -20,7 +20,7 @@ ResultVal<std::unique_ptr<StorageBackend>> RomFS_FileSystem::OpenFile(const std:
std::make_unique<RomFS_Storage>(romfs_file, data_offset, data_size));
}
ResultCode RomFS_FileSystem::DeleteFile(const Path& path) const {
ResultCode RomFS_FileSystem::DeleteFile(const std::string& path) const {
LOG_CRITICAL(Service_FS, "Attempted to delete a file from an ROMFS archive (%s).",
GetName().c_str());
// TODO(bunnei): Use correct error code

View File

@@ -31,7 +31,7 @@ public:
ResultVal<std::unique_ptr<StorageBackend>> OpenFile(const std::string& path,
Mode mode) const override;
ResultCode DeleteFile(const Path& path) const override;
ResultCode DeleteFile(const std::string& path) const override;
ResultCode RenameFile(const Path& src_path, const Path& dest_path) const override;
ResultCode DeleteDirectory(const Path& path) const override;
ResultCode DeleteDirectoryRecursively(const Path& path) const override;

View File

@@ -380,7 +380,7 @@ void VMManager::UpdatePageTableForVMA(const VirtualMemoryArea& vma) {
u64 VMManager::GetTotalMemoryUsage() {
LOG_WARNING(Kernel, "(STUBBED) called");
return 0xBE000000;
return 0xF8000000;
}
u64 VMManager::GetTotalHeapUsage() {

View File

@@ -73,7 +73,7 @@ public:
: ServiceFramework("IFile"), backend(std::move(backend)) {
static const FunctionInfo functions[] = {
{0, &IFile::Read, "Read"}, {1, &IFile::Write, "Write"},
{2, nullptr, "Flush"}, {3, &IFile::SetSize, "SetSize"},
{2, &IFile::Flush, "Flush"}, {3, &IFile::SetSize, "SetSize"},
{4, &IFile::GetSize, "GetSize"}, {5, nullptr, "OperateRange"},
};
RegisterHandlers(functions);
@@ -152,6 +152,14 @@ private:
rb.Push(RESULT_SUCCESS);
}
void Flush(Kernel::HLERequestContext& ctx) {
LOG_DEBUG(Service_FS, "called");
backend->Flush();
IPC::ResponseBuilder rb{ctx, 2};
rb.Push(RESULT_SUCCESS);
}
void SetSize(Kernel::HLERequestContext& ctx) {
IPC::RequestParser rp{ctx};
const u64 size = rp.Pop<u64>();
@@ -228,7 +236,7 @@ public:
: ServiceFramework("IFileSystem"), backend(std::move(backend)) {
static const FunctionInfo functions[] = {
{0, &IFileSystem::CreateFile, "CreateFile"},
{1, nullptr, "DeleteFile"},
{1, &IFileSystem::DeleteFile, "DeleteFile"},
{2, &IFileSystem::CreateDirectory, "CreateDirectory"},
{3, nullptr, "DeleteDirectory"},
{4, nullptr, "DeleteDirectoryRecursively"},
@@ -265,6 +273,20 @@ public:
rb.Push(backend->CreateFile(name, size));
}
void DeleteFile(Kernel::HLERequestContext& ctx) {
IPC::RequestParser rp{ctx};
auto file_buffer = ctx.ReadBuffer();
auto end = std::find(file_buffer.begin(), file_buffer.end(), '\0');
std::string name(file_buffer.begin(), end);
LOG_DEBUG(Service_FS, "called file %s", name.c_str());
IPC::ResponseBuilder rb{ctx, 2};
rb.Push(backend->DeleteFile(name));
}
void CreateDirectory(Kernel::HLERequestContext& ctx) {
IPC::RequestParser rp{ctx};

View File

@@ -47,10 +47,10 @@ PL_U::PL_U() : ServiceFramework("pl:u") {
FileUtil::CreateFullPath(filepath); // Create path if not already created
FileUtil::IOFile file(filepath, "rb");
shared_font = std::make_shared<std::vector<u8>>(SHARED_FONT_MEM_SIZE);
if (file.IsOpen()) {
// Read shared font data
ASSERT(file.GetSize() == SHARED_FONT_MEM_SIZE);
shared_font = std::make_shared<std::vector<u8>>(static_cast<size_t>(file.GetSize()));
file.ReadBytes(shared_font->data(), shared_font->size());
} else {
LOG_WARNING(Service_NS, "Unable to load shared font: %s", filepath.c_str());
@@ -97,22 +97,19 @@ void PL_U::GetSharedMemoryAddressOffset(Kernel::HLERequestContext& ctx) {
}
void PL_U::GetSharedMemoryNativeHandle(Kernel::HLERequestContext& ctx) {
if (shared_font != nullptr) {
// TODO(bunnei): This is a less-than-ideal solution to load a RAM dump of the Switch shared
// font data. This (likely) relies on exact address, size, and offsets from the original
// dump. In the future, we need to replace this with a more robust solution.
// TODO(bunnei): This is a less-than-ideal solution to load a RAM dump of the Switch shared
// font data. This (likely) relies on exact address, size, and offsets from the original
// dump. In the future, we need to replace this with a more robust solution.
// Map backing memory for the font data
Core::CurrentProcess()->vm_manager.MapMemoryBlock(SHARED_FONT_MEM_VADDR, shared_font, 0,
SHARED_FONT_MEM_SIZE,
Kernel::MemoryState::Shared);
// Map backing memory for the font data
Core::CurrentProcess()->vm_manager.MapMemoryBlock(
SHARED_FONT_MEM_VADDR, shared_font, 0, SHARED_FONT_MEM_SIZE, Kernel::MemoryState::Shared);
// Create shared font memory object
shared_font_mem = Kernel::SharedMemory::Create(
Core::CurrentProcess(), SHARED_FONT_MEM_SIZE, Kernel::MemoryPermission::ReadWrite,
Kernel::MemoryPermission::Read, SHARED_FONT_MEM_VADDR, Kernel::MemoryRegion::BASE,
"PL_U:shared_font_mem");
}
// Create shared font memory object
shared_font_mem = Kernel::SharedMemory::Create(
Core::CurrentProcess(), SHARED_FONT_MEM_SIZE, Kernel::MemoryPermission::ReadWrite,
Kernel::MemoryPermission::Read, SHARED_FONT_MEM_VADDR, Kernel::MemoryRegion::BASE,
"PL_U:shared_font_mem");
LOG_DEBUG(Service_NS, "called");
IPC::ResponseBuilder rb{ctx, 2, 1};

View File

@@ -301,5 +301,26 @@ u32 Maxwell3D::GetRegisterValue(u32 method) const {
return regs.reg_array[method];
}
bool Maxwell3D::IsShaderStageEnabled(Regs::ShaderStage stage) const {
// The Vertex stage is always enabled.
if (stage == Regs::ShaderStage::Vertex)
return true;
switch (stage) {
case Regs::ShaderStage::TesselationControl:
return regs.shader_config[static_cast<size_t>(Regs::ShaderProgram::TesselationControl)]
.enable != 0;
case Regs::ShaderStage::TesselationEval:
return regs.shader_config[static_cast<size_t>(Regs::ShaderProgram::TesselationEval)]
.enable != 0;
case Regs::ShaderStage::Geometry:
return regs.shader_config[static_cast<size_t>(Regs::ShaderProgram::Geometry)].enable != 0;
case Regs::ShaderStage::Fragment:
return regs.shader_config[static_cast<size_t>(Regs::ShaderProgram::Fragment)].enable != 0;
}
UNREACHABLE();
}
} // namespace Engines
} // namespace Tegra

View File

@@ -518,6 +518,9 @@ public:
/// Returns a list of enabled textures for the specified shader stage.
std::vector<Texture::FullTextureInfo> GetStageTextures(Regs::ShaderStage stage) const;
/// Returns whether the specified shader stage is enabled or not.
bool IsShaderStageEnabled(Regs::ShaderStage stage) const;
private:
std::unordered_map<u32, std::vector<u32>> uploaded_macros;

View File

@@ -46,6 +46,14 @@ RasterizerOpenGL::RasterizerOpenGL() {
state.texture_units[i].sampler = texture_samplers[i].sampler.handle;
}
// Create SSBOs
for (size_t stage = 0; stage < ssbos.size(); ++stage) {
for (size_t buffer = 0; buffer < ssbos[stage].size(); ++buffer) {
ssbos[stage][buffer].Create();
state.draw.const_buffers[stage][buffer].ssbo = ssbos[stage][buffer].handle;
}
}
GLint ext_num;
glGetIntegerv(GL_NUM_EXTENSIONS, &ext_num);
for (GLint i = 0; i < ext_num; i++) {
@@ -185,14 +193,18 @@ void RasterizerOpenGL::SetupShaders(u8* buffer_ptr, GLintptr buffer_offset, size
auto& gpu = Core::System().GetInstance().GPU().Maxwell3D();
ASSERT_MSG(!gpu.regs.shader_config[0].enable, "VertexA is unsupported!");
// Next available bindpoint to use when uploading the const buffers to the GLSL shaders.
u32 current_constbuffer_bindpoint = 0;
for (unsigned index = 1; index < Maxwell::MaxShaderProgram; ++index) {
ptr_pos += sizeof(GLShader::MaxwellUniformData);
auto& shader_config = gpu.regs.shader_config[index];
const Maxwell::ShaderProgram program{static_cast<Maxwell::ShaderProgram>(index)};
// VertexB program is always enabled, despite bit setting
const bool is_enabled{shader_config.enable || program == Maxwell::ShaderProgram::VertexB};
const auto& stage = index - 1; // Stage indices are 0 - 5
const bool is_enabled = gpu.IsShaderStageEnabled(static_cast<Maxwell::ShaderStage>(stage));
// Skip stages that are not enabled
if (!is_enabled) {
@@ -200,7 +212,6 @@ void RasterizerOpenGL::SetupShaders(u8* buffer_ptr, GLintptr buffer_offset, size
}
// Upload uniform data as one UBO per stage
const auto& stage = index - 1; // Stage indices are 0 - 5
const GLintptr ubo_offset = buffer_offset + static_cast<GLintptr>(ptr_pos);
copy_buffer(uniform_buffers[stage].handle, ubo_offset,
sizeof(GLShader::MaxwellUniformData));
@@ -215,15 +226,19 @@ void RasterizerOpenGL::SetupShaders(u8* buffer_ptr, GLintptr buffer_offset, size
Memory::ReadBlock(cpu_address, program_code.data(), program_code.size() * sizeof(u64));
GLShader::ShaderSetup setup{std::move(program_code)};
GLShader::ShaderEntries shader_resources;
switch (program) {
case Maxwell::ShaderProgram::VertexB: {
GLShader::MaxwellVSConfig vs_config{setup};
shader_program_manager->UseProgrammableVertexShader(vs_config, setup);
shader_resources =
shader_program_manager->UseProgrammableVertexShader(vs_config, setup);
break;
}
case Maxwell::ShaderProgram::Fragment: {
GLShader::MaxwellFSConfig fs_config{setup};
shader_program_manager->UseProgrammableFragmentShader(fs_config, setup);
shader_resources =
shader_program_manager->UseProgrammableFragmentShader(fs_config, setup);
break;
}
default:
@@ -231,6 +246,14 @@ void RasterizerOpenGL::SetupShaders(u8* buffer_ptr, GLintptr buffer_offset, size
shader_config.enable.Value(), shader_config.offset);
UNREACHABLE();
}
GLuint gl_stage_program = shader_program_manager->GetCurrentProgramStage(
static_cast<Maxwell::ShaderStage>(stage));
// Configure the const buffers for this shader stage.
current_constbuffer_bindpoint =
SetupConstBuffers(static_cast<Maxwell::ShaderStage>(stage), gl_stage_program,
current_constbuffer_bindpoint, shader_resources.const_buffer_entries);
}
shader_program_manager->UseTrivialGeometryShader();
@@ -380,7 +403,7 @@ void RasterizerOpenGL::DrawArrays() {
void RasterizerOpenGL::BindTextures() {
using Regs = Tegra::Engines::Maxwell3D::Regs;
auto maxwell3d = Core::System::GetInstance().GPU().Get3DEngine();
auto& maxwell3d = Core::System::GetInstance().GPU().Get3DEngine();
// Each Maxwell shader stage can have an arbitrary number of textures, but we're limited to a
// certain number in OpenGL. We try to only use the minimum amount of host textures by not
@@ -527,6 +550,55 @@ void RasterizerOpenGL::SamplerInfo::SyncWithConfig(const Tegra::Texture::TSCEntr
}
}
u32 RasterizerOpenGL::SetupConstBuffers(Maxwell::ShaderStage stage, GLuint program,
u32 current_bindpoint,
const std::vector<GLShader::ConstBufferEntry>& entries) {
auto& gpu = Core::System::GetInstance().GPU();
auto& maxwell3d = gpu.Get3DEngine();
ASSERT_MSG(maxwell3d.IsShaderStageEnabled(stage),
"Attempted to upload constbuffer of disabled shader stage");
// Reset all buffer draw state for this stage.
for (auto& buffer : state.draw.const_buffers[static_cast<size_t>(stage)]) {
buffer.bindpoint = 0;
buffer.enabled = false;
}
// Upload only the enabled buffers from the 16 constbuffers of each shader stage
auto& shader_stage = maxwell3d.state.shader_stages[static_cast<size_t>(stage)];
for (u32 bindpoint = 0; bindpoint < entries.size(); ++bindpoint) {
const auto& used_buffer = entries[bindpoint];
const auto& buffer = shader_stage.const_buffers[used_buffer.GetIndex()];
auto& buffer_draw_state =
state.draw.const_buffers[static_cast<size_t>(stage)][used_buffer.GetIndex()];
ASSERT_MSG(buffer.enabled, "Attempted to upload disabled constbuffer");
buffer_draw_state.enabled = true;
buffer_draw_state.bindpoint = current_bindpoint + bindpoint;
VAddr addr = gpu.memory_manager->PhysicalToVirtualAddress(buffer.address);
std::vector<u8> data(used_buffer.GetSize() * sizeof(float));
Memory::ReadBlock(addr, data.data(), data.size());
glBindBuffer(GL_SHADER_STORAGE_BUFFER, buffer_draw_state.ssbo);
glBufferData(GL_SHADER_STORAGE_BUFFER, data.size(), data.data(), GL_DYNAMIC_DRAW);
glBindBuffer(GL_SHADER_STORAGE_BUFFER, 0);
// Now configure the bindpoint of the buffer inside the shader
std::string buffer_name = used_buffer.GetName();
GLuint index =
glGetProgramResourceIndex(program, GL_SHADER_STORAGE_BLOCK, buffer_name.c_str());
if (index != -1)
glShaderStorageBlockBinding(program, index, buffer_draw_state.bindpoint);
}
state.Apply();
return current_bindpoint + entries.size();
}
void RasterizerOpenGL::BindFramebufferSurfaces(const Surface& color_surface,
const Surface& depth_surface, bool has_stencil) {
state.draw.draw_framebuffer = framebuffer.handle;

View File

@@ -87,6 +87,18 @@ private:
/// Binds the required textures to OpenGL before drawing a batch.
void BindTextures();
/*
* Configures the current constbuffers to use for the draw command.
* @param stage The shader stage to configure buffers for.
* @param program The OpenGL program object that contains the specified stage.
* @param current_bindpoint The offset at which to start counting new buffer bindpoints.
* @param entries Vector describing the buffers that are actually used in the guest shader.
* @returns The next available bindpoint for use in the next shader stage.
*/
u32 SetupConstBuffers(Tegra::Engines::Maxwell3D::Regs::ShaderStage stage, GLuint program,
u32 current_bindpoint,
const std::vector<GLShader::ConstBufferEntry>& entries);
/// Syncs the viewport to match the guest state
void SyncViewport(const MathUtil::Rectangle<u32>& surfaces_rect, u16 res_scale);
@@ -129,6 +141,10 @@ private:
std::array<bool, 16> hw_vao_enabled_attributes;
std::array<SamplerInfo, GLShader::NumTextureSamplers> texture_samplers;
std::array<std::array<OGLBuffer, Tegra::Engines::Maxwell3D::Regs::MaxConstBuffers>,
Tegra::Engines::Maxwell3D::Regs::MaxShaderStage>
ssbos;
static constexpr size_t VERTEX_BUFFER_SIZE = 128 * 1024 * 1024;
std::unique_ptr<OGLStreamBuffer> vertex_buffer;
OGLBuffer uniform_buffer;

View File

@@ -140,6 +140,11 @@ public:
return declarations.GetResult() + shader.GetResult();
}
/// Returns entries in the shader that are useful for external functions
ShaderEntries GetEntries() const {
return {GetConstBuffersDeclarations()};
}
private:
/// Gets the Subroutine object corresponding to the specified address.
const Subroutine& GetSubroutine(u32 begin, u32 end) const {
@@ -186,10 +191,9 @@ private:
}
/// Generates code representing a uniform (C buffer) register.
std::string GetUniform(const Uniform& reg) const {
std::string index = std::to_string(reg.index);
return "uniform_" + index + "[" + std::to_string(reg.offset >> 2) + "][" +
std::to_string(reg.offset & 3) + "]";
std::string GetUniform(const Uniform& reg) {
declr_const_buffers[reg.index].MarkAsUsed(reg.index, reg.offset, stage);
return 'c' + std::to_string(reg.index) + '[' + std::to_string(reg.offset) + ']';
}
/**
@@ -439,6 +443,14 @@ private:
GenerateDeclarations();
}
/// Returns a list of constant buffer declarations
std::vector<ConstBufferEntry> GetConstBuffersDeclarations() const {
std::vector<ConstBufferEntry> result;
std::copy_if(declr_const_buffers.begin(), declr_const_buffers.end(),
std::back_inserter(result), [](const auto& entry) { return entry.IsUsed(); });
return result;
}
/// Add declarations for registers
void GenerateDeclarations() {
for (const auto& reg : declr_register) {
@@ -463,6 +475,16 @@ private:
") out vec4 " + GetOutputAttribute(index) + ";");
}
declarations.AddLine("");
unsigned const_buffer_layout = 0;
for (const auto& entry : GetConstBuffersDeclarations()) {
declarations.AddLine("layout(std430) buffer " + entry.GetName());
declarations.AddLine("{");
declarations.AddLine(" float c" + std::to_string(entry.GetIndex()) + "[];");
declarations.AddLine("};");
declarations.AddLine("");
++const_buffer_layout;
}
}
private:
@@ -478,18 +500,19 @@ private:
std::set<std::string> declr_register;
std::set<Attribute::Index> declr_input_attribute;
std::set<Attribute::Index> declr_output_attribute;
}; // namespace Decompiler
std::array<ConstBufferEntry, Maxwell3D::Regs::MaxConstBuffers> declr_const_buffers;
};
std::string GetCommonDeclarations() {
return "bool exec_shader();";
}
boost::optional<std::string> DecompileProgram(const ProgramCode& program_code, u32 main_offset,
Maxwell3D::Regs::ShaderStage stage) {
boost::optional<ProgramResult> DecompileProgram(const ProgramCode& program_code, u32 main_offset,
Maxwell3D::Regs::ShaderStage stage) {
try {
auto subroutines = ControlFlowAnalyzer(program_code, main_offset).GetSubroutines();
GLSLGenerator generator(subroutines, program_code, main_offset, stage);
return generator.GetShaderCode();
return ProgramResult{generator.GetShaderCode(), generator.GetEntries()};
} catch (const DecompileFail& exception) {
LOG_ERROR(HW_GPU, "Shader decompilation failed: %s", exception.what());
}

View File

@@ -17,8 +17,8 @@ using Tegra::Engines::Maxwell3D;
std::string GetCommonDeclarations();
boost::optional<std::string> DecompileProgram(const ProgramCode& program_code, u32 main_offset,
Maxwell3D::Regs::ShaderStage stage);
boost::optional<ProgramResult> DecompileProgram(const ProgramCode& program_code, u32 main_offset,
Maxwell3D::Regs::ShaderStage stage);
} // namespace Decompiler
} // namespace GLShader

View File

@@ -3,18 +3,60 @@
// Refer to the license.txt file included.
#include "common/assert.h"
#include "video_core/engines/maxwell_3d.h"
#include "video_core/renderer_opengl/gl_shader_decompiler.h"
#include "video_core/renderer_opengl/gl_shader_gen.h"
namespace GLShader {
std::string GenerateVertexShader(const ShaderSetup& setup, const MaxwellVSConfig& config) {
UNREACHABLE();
return {};
using Tegra::Engines::Maxwell3D;
static constexpr u32 PROGRAM_OFFSET{10};
ProgramResult GenerateVertexShader(const ShaderSetup& setup, const MaxwellVSConfig& config) {
std::string out = "#version 430 core\n";
out += "#extension GL_ARB_separate_shader_objects : enable\n\n";
out += Decompiler::GetCommonDeclarations();
ProgramResult program = Decompiler::DecompileProgram(setup.program_code, PROGRAM_OFFSET,
Maxwell3D::Regs::ShaderStage::Vertex)
.get_value_or({});
out += R"(
out gl_PerVertex {
vec4 gl_Position;
};
void main() {
exec_shader();
}
std::string GenerateFragmentShader(const ShaderSetup& setup, const MaxwellFSConfig& config) {
UNREACHABLE();
return {};
)";
out += program.first;
return {out, program.second};
}
ProgramResult GenerateFragmentShader(const ShaderSetup& setup, const MaxwellFSConfig& config) {
std::string out = "#version 430 core\n";
out += "#extension GL_ARB_separate_shader_objects : enable\n\n";
out += Decompiler::GetCommonDeclarations();
ProgramResult program = Decompiler::DecompileProgram(setup.program_code, PROGRAM_OFFSET,
Maxwell3D::Regs::ShaderStage::Fragment)
.get_value_or({});
out += R"(
out vec4 color;
uniform sampler2D tex[32];
void main() {
exec_shader();
}
)";
out += program.first;
return {out, program.second};
}
} // namespace GLShader

View File

@@ -7,6 +7,8 @@
#include <array>
#include <string>
#include <type_traits>
#include <utility>
#include <vector>
#include "common/common_types.h"
#include "common/hash.h"
@@ -16,6 +18,50 @@ constexpr size_t MAX_PROGRAM_CODE_LENGTH{0x1000};
using ProgramCode = std::array<u64, MAX_PROGRAM_CODE_LENGTH>;
class ConstBufferEntry {
using Maxwell = Tegra::Engines::Maxwell3D::Regs;
public:
void MarkAsUsed(unsigned index, unsigned offset, Maxwell::ShaderStage stage) {
is_used = true;
this->index = index;
this->stage = stage;
max_offset = std::max(max_offset, offset);
}
bool IsUsed() const {
return is_used;
}
unsigned GetIndex() const {
return index;
}
unsigned GetSize() const {
return max_offset + 1;
}
std::string GetName() const {
return BufferBaseNames[static_cast<size_t>(stage)] + std::to_string(index);
}
private:
static constexpr std::array<const char*, Maxwell::MaxShaderStage> BufferBaseNames = {
"buffer_vs_c", "buffer_tessc_c", "buffer_tesse_c", "buffer_gs_c", "buffer_fs_c",
};
bool is_used{};
unsigned index{};
unsigned max_offset{};
Maxwell::ShaderStage stage;
};
struct ShaderEntries {
std::vector<ConstBufferEntry> const_buffer_entries;
};
using ProgramResult = std::pair<std::string, ShaderEntries>;
struct ShaderSetup {
ShaderSetup(ProgramCode&& program_code) : program_code(std::move(program_code)) {}
@@ -58,13 +104,13 @@ struct MaxwellFSConfig : Common::HashableStruct<MaxwellShaderConfigCommon> {
* Generates the GLSL vertex shader program source code for the given VS program
* @returns String of the shader source code
*/
std::string GenerateVertexShader(const ShaderSetup& setup, const MaxwellVSConfig& config);
ProgramResult GenerateVertexShader(const ShaderSetup& setup, const MaxwellVSConfig& config);
/**
* Generates the GLSL fragment shader program source code for the given FS program
* @returns String of the shader source code
*/
std::string GenerateFragmentShader(const ShaderSetup& setup, const MaxwellFSConfig& config);
ProgramResult GenerateFragmentShader(const ShaderSetup& setup, const MaxwellFSConfig& config);
} // namespace GLShader

View File

@@ -53,13 +53,6 @@ void SetShaderSamplerBindings(GLuint shader) {
} // namespace Impl
void MaxwellUniformData::SetFromRegs(const Maxwell3D::State::ShaderStageInfo& shader_stage) {
const auto& memory_manager = Core::System().GetInstance().GPU().memory_manager;
for (unsigned index = 0; index < shader_stage.const_buffers.size(); ++index) {
const auto& const_buffer = shader_stage.const_buffers[index];
const VAddr vaddr = memory_manager->PhysicalToVirtualAddress(const_buffer.address);
Memory::ReadBlock(vaddr, const_buffers[index].data(), sizeof(ConstBuffer));
}
}
void MaxwellUniformData::SetFromRegs(const Maxwell3D::State::ShaderStageInfo& shader_stage) {}
} // namespace GLShader

View File

@@ -30,11 +30,10 @@ void SetShaderSamplerBindings(GLuint shader);
// Not following that rule will cause problems on some AMD drivers.
struct MaxwellUniformData {
void SetFromRegs(const Maxwell3D::State::ShaderStageInfo& shader_stage);
using ConstBuffer = std::array<GLvec4, 4>;
alignas(16) std::array<ConstBuffer, Maxwell3D::Regs::MaxConstBuffers> const_buffers;
// TODO(Subv): Use this for something.
};
static_assert(sizeof(MaxwellUniformData) == 1024, "MaxwellUniformData structure size is incorrect");
// static_assert(sizeof(MaxwellUniformData) == 1024, "MaxwellUniformData structure size is
// incorrect");
static_assert(sizeof(MaxwellUniformData) < 16384,
"MaxwellUniformData structure must be less than 16kb as per the OpenGL spec");
@@ -42,19 +41,25 @@ class OGLShaderStage {
public:
OGLShaderStage() = default;
void Create(const char* source, GLenum type) {
void Create(const ProgramResult& program_result, GLenum type) {
OGLShader shader;
shader.Create(source, type);
shader.Create(program_result.first.c_str(), type);
program.Create(true, shader.handle);
Impl::SetShaderUniformBlockBindings(program.handle);
Impl::SetShaderSamplerBindings(program.handle);
entries = program_result.second;
}
GLuint GetHandle() const {
return program.handle;
}
ShaderEntries GetEntries() const {
return entries;
}
private:
OGLProgram program;
ShaderEntries entries;
};
// TODO(wwylele): beautify this doc
@@ -62,25 +67,28 @@ private:
// The double cache is needed because diffent KeyConfigType, which includes a hash of the code
// region (including its leftover unused code) can generate the same GLSL code.
template <typename KeyConfigType,
std::string (*CodeGenerator)(const ShaderSetup&, const KeyConfigType&), GLenum ShaderType>
ProgramResult (*CodeGenerator)(const ShaderSetup&, const KeyConfigType&),
GLenum ShaderType>
class ShaderCache {
public:
ShaderCache() = default;
GLuint Get(const KeyConfigType& key, const ShaderSetup& setup) {
using Result = std::pair<GLuint, ShaderEntries>;
Result Get(const KeyConfigType& key, const ShaderSetup& setup) {
auto map_it = shader_map.find(key);
if (map_it == shader_map.end()) {
std::string program = CodeGenerator(setup, key);
ProgramResult program = CodeGenerator(setup, key);
auto [iter, new_shader] = shader_cache.emplace(program, OGLShaderStage{});
auto [iter, new_shader] = shader_cache.emplace(program.first, OGLShaderStage{});
OGLShaderStage& cached_shader = iter->second;
if (new_shader) {
cached_shader.Create(program.c_str(), ShaderType);
cached_shader.Create(program, ShaderType);
}
shader_map[key] = &cached_shader;
return cached_shader.GetHandle();
return {cached_shader.GetHandle(), program.second};
} else {
return map_it->second->GetHandle();
return {map_it->second->GetHandle(), map_it->second->GetEntries()};
}
}
@@ -99,12 +107,29 @@ public:
pipeline.Create();
}
void UseProgrammableVertexShader(const MaxwellVSConfig& config, const ShaderSetup setup) {
current.vs = vertex_shaders.Get(config, setup);
ShaderEntries UseProgrammableVertexShader(const MaxwellVSConfig& config,
const ShaderSetup setup) {
ShaderEntries result;
std::tie(current.vs, result) = vertex_shaders.Get(config, setup);
return result;
}
void UseProgrammableFragmentShader(const MaxwellFSConfig& config, const ShaderSetup setup) {
current.fs = fragment_shaders.Get(config, setup);
ShaderEntries UseProgrammableFragmentShader(const MaxwellFSConfig& config,
const ShaderSetup setup) {
ShaderEntries result;
std::tie(current.fs, result) = fragment_shaders.Get(config, setup);
return result;
}
GLuint GetCurrentProgramStage(Maxwell3D::Regs::ShaderStage stage) {
switch (stage) {
case Maxwell3D::Regs::ShaderStage::Vertex:
return current.vs;
case Maxwell3D::Regs::ShaderStage::Fragment:
return current.fs;
}
UNREACHABLE();
}
void UseTrivialGeometryShader() {

View File

@@ -202,6 +202,20 @@ void OpenGLState::Apply() const {
}
}
// Constbuffers
for (u32 stage = 0; stage < draw.const_buffers.size(); ++stage) {
for (u32 buffer_id = 0; buffer_id < draw.const_buffers[stage].size(); ++buffer_id) {
auto& current = cur_state.draw.const_buffers[stage][buffer_id];
auto& new_state = draw.const_buffers[stage][buffer_id];
if (current.enabled != new_state.enabled || current.bindpoint != new_state.bindpoint ||
current.ssbo != new_state.ssbo) {
if (new_state.enabled) {
glBindBufferBase(GL_SHADER_STORAGE_BUFFER, new_state.bindpoint, new_state.ssbo);
}
}
}
}
// Lighting LUTs
if (lighting_lut.texture_buffer != cur_state.lighting_lut.texture_buffer) {
glActiveTexture(TextureUnits::LightingLUT.Enum());

View File

@@ -123,6 +123,12 @@ public:
GLuint uniform_buffer; // GL_UNIFORM_BUFFER_BINDING
GLuint shader_program; // GL_CURRENT_PROGRAM
GLuint program_pipeline; // GL_PROGRAM_PIPELINE_BINDING
struct ConstBufferConfig {
bool enabled = false;
GLuint bindpoint;
GLuint ssbo;
};
std::array<std::array<ConstBufferConfig, 16>, 5> const_buffers{};
} draw;
struct {

View File

@@ -72,8 +72,15 @@ inline GLenum TextureFilterMode(Tegra::Texture::TextureFilter filter_mode) {
inline GLenum WrapMode(Tegra::Texture::WrapMode wrap_mode) {
switch (wrap_mode) {
case Tegra::Texture::WrapMode::Wrap:
return GL_REPEAT;
case Tegra::Texture::WrapMode::ClampToEdge:
return GL_CLAMP_TO_EDGE;
case Tegra::Texture::WrapMode::ClampOGL:
// TODO(Subv): GL_CLAMP was removed as of OpenGL 3.1, to implement GL_CLAMP, we can use
// GL_CLAMP_TO_BORDER to get the border color of the texture, and then sample the edge to
// manually mix them. However the shader part of this is not yet implemented.
return GL_CLAMP_TO_BORDER;
}
LOG_CRITICAL(Render_OpenGL, "Unimplemented texture wrap mode=%u", static_cast<u32>(wrap_mode));
UNREACHABLE();

View File

@@ -776,9 +776,11 @@ void GMainWindow::closeEvent(QCloseEvent* event) {
return;
}
UISettings::values.geometry = saveGeometry();
if (ui.action_Fullscreen->isChecked()) {
UISettings::values.geometry = saveGeometry();
UISettings::values.renderwindow_geometry = render_window->saveGeometry();
}
UISettings::values.state = saveState();
UISettings::values.renderwindow_geometry = render_window->saveGeometry();
#if MICROPROFILE_ENABLED
UISettings::values.microprofile_geometry = microProfileDialog->saveGeometry();
UISettings::values.microprofile_visible = microProfileDialog->isVisible();