aboutsummaryrefslogtreecommitdiff
path: root/src/video_core
diff options
context:
space:
mode:
Diffstat (limited to 'src/video_core')
-rw-r--r--src/video_core/renderer_opengl/gl_rasterizer_cache.cpp14
-rw-r--r--src/video_core/renderer_opengl/gl_rasterizer_cache.h71
-rw-r--r--src/video_core/renderer_opengl/gl_shader_decompiler.cpp35
3 files changed, 73 insertions, 47 deletions
diff --git a/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp b/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp
index 5d58ebd4f..05f153599 100644
--- a/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp
+++ b/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp
@@ -119,7 +119,8 @@ static constexpr std::array<FormatTuple, SurfaceParams::MaxPixelFormat> tex_form
{GL_COMPRESSED_RGBA_BPTC_UNORM_ARB, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, ComponentType::UNorm,
true}, // BC7U
{GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_4X4
- {GL_RG8, GL_RG, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // G8R8
+ {GL_RG8, GL_RG, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // G8R8U
+ {GL_RG8, GL_RG, GL_BYTE, ComponentType::SNorm, false}, // G8R8S
{GL_RGBA8, GL_BGRA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // BGRA8
{GL_RGBA32F, GL_RGBA, GL_FLOAT, ComponentType::Float, false}, // RGBA32F
{GL_RG32F, GL_RG, GL_FLOAT, ComponentType::Float, false}, // RG32F
@@ -260,7 +261,8 @@ static constexpr std::array<void (*)(u32, u32, u32, std::vector<u8>&, Tegra::GPU
MortonCopy<true, PixelFormat::DXN2SNORM>,
MortonCopy<true, PixelFormat::BC7U>,
MortonCopy<true, PixelFormat::ASTC_2D_4X4>,
- MortonCopy<true, PixelFormat::G8R8>,
+ MortonCopy<true, PixelFormat::G8R8U>,
+ MortonCopy<true, PixelFormat::G8R8S>,
MortonCopy<true, PixelFormat::BGRA8>,
MortonCopy<true, PixelFormat::RGBA32F>,
MortonCopy<true, PixelFormat::RG32F>,
@@ -315,7 +317,8 @@ static constexpr std::array<void (*)(u32, u32, u32, std::vector<u8>&, Tegra::GPU
nullptr,
nullptr,
nullptr,
- MortonCopy<false, PixelFormat::G8R8>,
+ MortonCopy<false, PixelFormat::G8R8U>,
+ MortonCopy<false, PixelFormat::G8R8S>,
MortonCopy<false, PixelFormat::BGRA8>,
MortonCopy<false, PixelFormat::RGBA32F>,
MortonCopy<false, PixelFormat::RG32F>,
@@ -461,7 +464,7 @@ static void ConvertS8Z24ToZ24S8(std::vector<u8>& data, u32 width, u32 height) {
}
static void ConvertG8R8ToR8G8(std::vector<u8>& data, u32 width, u32 height) {
- const auto bpp{CachedSurface::GetGLBytesPerPixel(PixelFormat::G8R8)};
+ const auto bpp{CachedSurface::GetGLBytesPerPixel(PixelFormat::G8R8U)};
for (size_t y = 0; y < height; ++y) {
for (size_t x = 0; x < width; ++x) {
const size_t offset{bpp * (y * width + x)};
@@ -493,7 +496,8 @@ static void ConvertFormatAsNeeded_LoadGLBuffer(std::vector<u8>& data, PixelForma
ConvertS8Z24ToZ24S8(data, width, height);
break;
- case PixelFormat::G8R8:
+ case PixelFormat::G8R8U:
+ case PixelFormat::G8R8S:
// Convert the G8R8 color format to R8G8, as OpenGL does not support G8R8.
ConvertG8R8ToR8G8(data, width, height);
break;
diff --git a/src/video_core/renderer_opengl/gl_rasterizer_cache.h b/src/video_core/renderer_opengl/gl_rasterizer_cache.h
index 36a41522b..37eef5ad0 100644
--- a/src/video_core/renderer_opengl/gl_rasterizer_cache.h
+++ b/src/video_core/renderer_opengl/gl_rasterizer_cache.h
@@ -43,36 +43,37 @@ struct SurfaceParams {
DXN2SNORM = 17,
BC7U = 18,
ASTC_2D_4X4 = 19,
- G8R8 = 20,
- BGRA8 = 21,
- RGBA32F = 22,
- RG32F = 23,
- R32F = 24,
- R16F = 25,
- R16UNORM = 26,
- R16S = 27,
- R16UI = 28,
- R16I = 29,
- RG16 = 30,
- RG16F = 31,
- RG16UI = 32,
- RG16I = 33,
- RG16S = 34,
- RGB32F = 35,
- SRGBA8 = 36,
- RG8U = 37,
- RG8S = 38,
- RG32UI = 39,
- R32UI = 40,
+ G8R8U = 20,
+ G8R8S = 21,
+ BGRA8 = 22,
+ RGBA32F = 23,
+ RG32F = 24,
+ R32F = 25,
+ R16F = 26,
+ R16UNORM = 27,
+ R16S = 28,
+ R16UI = 29,
+ R16I = 30,
+ RG16 = 31,
+ RG16F = 32,
+ RG16UI = 33,
+ RG16I = 34,
+ RG16S = 35,
+ RGB32F = 36,
+ SRGBA8 = 37,
+ RG8U = 38,
+ RG8S = 39,
+ RG32UI = 40,
+ R32UI = 41,
MaxColorFormat,
// DepthStencil formats
- Z24S8 = 41,
- S8Z24 = 42,
- Z32F = 43,
- Z16 = 44,
- Z32FS8 = 45,
+ Z24S8 = 42,
+ S8Z24 = 43,
+ Z32F = 44,
+ Z16 = 45,
+ Z32FS8 = 46,
MaxDepthStencilFormat,
@@ -130,7 +131,8 @@ struct SurfaceParams {
4, // DXN2SNORM
4, // BC7U
4, // ASTC_2D_4X4
- 1, // G8R8
+ 1, // G8R8U
+ 1, // G8R8S
1, // BGRA8
1, // RGBA32F
1, // RG32F
@@ -187,7 +189,8 @@ struct SurfaceParams {
128, // DXN2SNORM
128, // BC7U
32, // ASTC_2D_4X4
- 16, // G8R8
+ 16, // G8R8U
+ 16, // G8R8S
32, // BGRA8
128, // RGBA32F
64, // RG32F
@@ -341,7 +344,15 @@ struct SurfaceParams {
static_cast<u32>(component_type));
UNREACHABLE();
case Tegra::Texture::TextureFormat::G8R8:
- return PixelFormat::G8R8;
+ switch (component_type) {
+ case Tegra::Texture::ComponentType::UNORM:
+ return PixelFormat::G8R8U;
+ case Tegra::Texture::ComponentType::SNORM:
+ return PixelFormat::G8R8S;
+ }
+ LOG_CRITICAL(HW_GPU, "Unimplemented component_type={}",
+ static_cast<u32>(component_type));
+ UNREACHABLE();
case Tegra::Texture::TextureFormat::R16_G16_B16_A16:
return PixelFormat::RGBA16F;
case Tegra::Texture::TextureFormat::BF10GF11RF11:
@@ -396,6 +407,8 @@ struct SurfaceParams {
UNREACHABLE();
case Tegra::Texture::TextureFormat::ZF32:
return PixelFormat::Z32F;
+ case Tegra::Texture::TextureFormat::Z16:
+ return PixelFormat::Z16;
case Tegra::Texture::TextureFormat::Z24S8:
return PixelFormat::Z24S8;
case Tegra::Texture::TextureFormat::DXT1:
diff --git a/src/video_core/renderer_opengl/gl_shader_decompiler.cpp b/src/video_core/renderer_opengl/gl_shader_decompiler.cpp
index 6834d7085..dabf98b74 100644
--- a/src/video_core/renderer_opengl/gl_shader_decompiler.cpp
+++ b/src/video_core/renderer_opengl/gl_shader_decompiler.cpp
@@ -383,15 +383,13 @@ public:
}
}
- std::string GetUniformIndirect(u64 index, s64 offset, const Register& index_reg,
+ std::string GetUniformIndirect(u64 cbuf_index, s64 offset, const std::string& index_str,
GLSLRegister::Type type) {
- declr_const_buffers[index].MarkAsUsedIndirect(index, stage);
+ declr_const_buffers[cbuf_index].MarkAsUsedIndirect(cbuf_index, stage);
- std::string final_offset = "((floatBitsToInt(" + GetRegister(index_reg, 0) + ") + " +
- std::to_string(offset) + ") / 4)";
-
- std::string value =
- 'c' + std::to_string(index) + '[' + final_offset + " / 4][" + final_offset + " % 4]";
+ std::string final_offset = fmt::format("({} + {})", index_str, offset / 4);
+ std::string value = 'c' + std::to_string(cbuf_index) + '[' + final_offset + " / 4][" +
+ final_offset + " % 4]";
if (type == GLSLRegister::Type::Float) {
return value;
@@ -1355,11 +1353,16 @@ private:
case OpCode::Id::LD_C: {
ASSERT_MSG(instr.ld_c.unknown == 0, "Unimplemented");
+ // Add an extra scope and declare the index register inside to prevent
+ // overwriting it in case it is used as an output of the LD instruction.
+ shader.AddLine("{");
+ ++shader.scope;
+
+ shader.AddLine("uint index = (" + regs.GetRegisterAsInteger(instr.gpr8, 0, false) +
+ " / 4) & (MAX_CONSTBUFFER_ELEMENTS - 1);");
+
std::string op_a =
- regs.GetUniformIndirect(instr.cbuf36.index, instr.cbuf36.offset + 0, instr.gpr8,
- GLSLRegister::Type::Float);
- std::string op_b =
- regs.GetUniformIndirect(instr.cbuf36.index, instr.cbuf36.offset + 4, instr.gpr8,
+ regs.GetUniformIndirect(instr.cbuf36.index, instr.cbuf36.offset + 0, "index",
GLSLRegister::Type::Float);
switch (instr.ld_c.type.Value()) {
@@ -1367,16 +1370,22 @@ private:
regs.SetRegisterToFloat(instr.gpr0, 0, op_a, 1, 1);
break;
- case Tegra::Shader::UniformType::Double:
+ case Tegra::Shader::UniformType::Double: {
+ std::string op_b =
+ regs.GetUniformIndirect(instr.cbuf36.index, instr.cbuf36.offset + 4,
+ "index", GLSLRegister::Type::Float);
regs.SetRegisterToFloat(instr.gpr0, 0, op_a, 1, 1);
regs.SetRegisterToFloat(instr.gpr0.Value() + 1, 0, op_b, 1, 1);
break;
-
+ }
default:
LOG_CRITICAL(HW_GPU, "Unhandled type: {}",
static_cast<unsigned>(instr.ld_c.type.Value()));
UNREACHABLE();
}
+
+ --shader.scope;
+ shader.AddLine("}");
break;
}
case OpCode::Id::ST_A: {
generated by cgit v1.2.3 (git 2.25.1) at 2025-12-29 20:49:07 +0000