From a27c0099ededac2d1fb1745a437a446450dfea10 Mon Sep 17 00:00:00 2001
From: bunnei <bunneidev@gmail.com>
Date: Mon, 23 Jul 2018 19:10:00 -0400
Subject: [PATCH] gl_rasterizer_cache: Implement RenderTargetFormat RG32_FLOAT.

---
 src/video_core/gpu.cpp                         |  1 +
 src/video_core/gpu.h                           |  1 +
 .../renderer_opengl/gl_rasterizer_cache.cpp    |  8 +++++---
 .../renderer_opengl/gl_rasterizer_cache.h      | 18 ++++++++++++++----
 src/video_core/textures/decoders.cpp           |  4 ++++
 5 files changed, 25 insertions(+), 7 deletions(-)

diff --git a/src/video_core/gpu.cpp b/src/video_core/gpu.cpp
index b094d48c3a..60c49d6723 100644
--- a/src/video_core/gpu.cpp
+++ b/src/video_core/gpu.cpp
@@ -35,6 +35,7 @@ u32 RenderTargetBytesPerPixel(RenderTargetFormat format) {
     case RenderTargetFormat::RGBA32_FLOAT:
         return 16;
     case RenderTargetFormat::RGBA16_FLOAT:
+    case RenderTargetFormat::RG32_FLOAT:
         return 8;
     case RenderTargetFormat::RGBA8_UNORM:
     case RenderTargetFormat::RGB10_A2_UNORM:
diff --git a/src/video_core/gpu.h b/src/video_core/gpu.h
index 9c74cfac38..58501ca8b2 100644
--- a/src/video_core/gpu.h
+++ b/src/video_core/gpu.h
@@ -18,6 +18,7 @@ enum class RenderTargetFormat : u32 {
     RGBA32_FLOAT = 0xC0,
     RGBA32_UINT = 0xC2,
     RGBA16_FLOAT = 0xCA,
+    RG32_FLOAT = 0xCB,
     BGRA8_UNORM = 0xCF,
     RGB10_A2_UNORM = 0xD1,
     RGBA8_UNORM = 0xD5,
diff --git a/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp b/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp
index 133a15a125..8f99864a04 100644
--- a/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp
+++ b/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp
@@ -109,6 +109,7 @@ static constexpr std::array<FormatTuple, SurfaceParams::MaxPixelFormat> tex_form
     {GL_RG8, GL_RG, GL_UNSIGNED_BYTE, ComponentType::UNorm, false},     // G8R8
     {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
 
     // DepthStencil formats
     {GL_DEPTH24_STENCIL8, GL_DEPTH_STENCIL, GL_UNSIGNED_INT_24_8, ComponentType::UNorm,
@@ -201,9 +202,9 @@ static constexpr std::array<void (*)(u32, u32, u32, u8*, Tegra::GPUVAddr),
         MortonCopy<true, PixelFormat::DXT45>,        MortonCopy<true, PixelFormat::DXN1>,
         MortonCopy<true, PixelFormat::BC7U>,         MortonCopy<true, PixelFormat::ASTC_2D_4X4>,
         MortonCopy<true, PixelFormat::G8R8>,         MortonCopy<true, PixelFormat::BGRA8>,
-        MortonCopy<true, PixelFormat::RGBA32F>,      MortonCopy<true, PixelFormat::Z24S8>,
-        MortonCopy<true, PixelFormat::S8Z24>,        MortonCopy<true, PixelFormat::Z32F>,
-        MortonCopy<true, PixelFormat::Z16>,
+        MortonCopy<true, PixelFormat::RGBA32F>,      MortonCopy<true, PixelFormat::RG32F>,
+        MortonCopy<true, PixelFormat::Z24S8>,        MortonCopy<true, PixelFormat::S8Z24>,
+        MortonCopy<true, PixelFormat::Z32F>,         MortonCopy<true, PixelFormat::Z16>,
 };
 
 static constexpr std::array<void (*)(u32, u32, u32, u8*, Tegra::GPUVAddr),
@@ -227,6 +228,7 @@ static constexpr std::array<void (*)(u32, u32, u32, u8*, Tegra::GPUVAddr),
         MortonCopy<false, PixelFormat::G8R8>,
         MortonCopy<false, PixelFormat::BGRA8>,
         MortonCopy<false, PixelFormat::RGBA32F>,
+        MortonCopy<false, PixelFormat::RG32F>,
         MortonCopy<false, PixelFormat::Z24S8>,
         MortonCopy<false, PixelFormat::S8Z24>,
         MortonCopy<false, PixelFormat::Z32F>,
diff --git a/src/video_core/renderer_opengl/gl_rasterizer_cache.h b/src/video_core/renderer_opengl/gl_rasterizer_cache.h
index 2feea3d4db..23efbe67c8 100644
--- a/src/video_core/renderer_opengl/gl_rasterizer_cache.h
+++ b/src/video_core/renderer_opengl/gl_rasterizer_cache.h
@@ -39,14 +39,15 @@ struct SurfaceParams {
         G8R8 = 14,
         BGRA8 = 15,
         RGBA32F = 16,
+        RG32F = 17,
 
         MaxColorFormat,
 
         // DepthStencil formats
-        Z24S8 = 17,
-        S8Z24 = 18,
-        Z32F = 19,
-        Z16 = 20,
+        Z24S8 = 18,
+        S8Z24 = 19,
+        Z32F = 20,
+        Z16 = 21,
 
         MaxDepthStencilFormat,
 
@@ -101,6 +102,7 @@ struct SurfaceParams {
             1, // G8R8
             1, // BGRA8
             1, // RGBA32F
+            1, // RG32F
             1, // Z24S8
             1, // S8Z24
             1, // Z32F
@@ -133,6 +135,7 @@ struct SurfaceParams {
             16,  // G8R8
             32,  // BGRA8
             128, // RGBA32F
+            64,  // RG32F
             32,  // Z24S8
             32,  // S8Z24
             32,  // Z32F
@@ -176,6 +179,8 @@ struct SurfaceParams {
             return PixelFormat::RGBA16F;
         case Tegra::RenderTargetFormat::RGBA32_FLOAT:
             return PixelFormat::RGBA32F;
+        case Tegra::RenderTargetFormat::RG32_FLOAT:
+            return PixelFormat::RG32F;
         case Tegra::RenderTargetFormat::R11G11B10_FLOAT:
             return PixelFormat::R11FG11FB10F;
         case Tegra::RenderTargetFormat::RGBA32_UINT:
@@ -216,6 +221,8 @@ struct SurfaceParams {
             LOG_CRITICAL(HW_GPU, "Unimplemented component_type={}",
                          static_cast<u32>(component_type));
             UNREACHABLE();
+        case Tegra::Texture::TextureFormat::R32_G32:
+            return PixelFormat::RG32F;
         case Tegra::Texture::TextureFormat::DXT1:
             return PixelFormat::DXT1;
         case Tegra::Texture::TextureFormat::DXT23:
@@ -274,6 +281,8 @@ struct SurfaceParams {
             return Tegra::Texture::TextureFormat::A8R8G8B8;
         case PixelFormat::RGBA32F:
             return Tegra::Texture::TextureFormat::R32_G32_B32_A32;
+        case PixelFormat::RG32F:
+            return Tegra::Texture::TextureFormat::R32_G32;
         default:
             LOG_CRITICAL(HW_GPU, "Unimplemented format={}", static_cast<u32>(format));
             UNREACHABLE();
@@ -320,6 +329,7 @@ struct SurfaceParams {
         case Tegra::RenderTargetFormat::RGBA16_FLOAT:
         case Tegra::RenderTargetFormat::R11G11B10_FLOAT:
         case Tegra::RenderTargetFormat::RGBA32_FLOAT:
+        case Tegra::RenderTargetFormat::RG32_FLOAT:
             return ComponentType::Float;
         case Tegra::RenderTargetFormat::RGBA32_UINT:
             return ComponentType::UInt;
diff --git a/src/video_core/textures/decoders.cpp b/src/video_core/textures/decoders.cpp
index a3e67d105f..e5e9e18981 100644
--- a/src/video_core/textures/decoders.cpp
+++ b/src/video_core/textures/decoders.cpp
@@ -72,6 +72,8 @@ u32 BytesPerPixel(TextureFormat format) {
         return 8;
     case TextureFormat::R32_G32_B32_A32:
         return 16;
+    case TextureFormat::R32_G32:
+        return 8;
     default:
         UNIMPLEMENTED_MSG("Format not implemented");
         break;
@@ -118,6 +120,7 @@ std::vector<u8> UnswizzleTexture(VAddr address, TextureFormat format, u32 width,
     case TextureFormat::G8R8:
     case TextureFormat::R16_G16_B16_A16:
     case TextureFormat::R32_G32_B32_A32:
+    case TextureFormat::R32_G32:
     case TextureFormat::BF10GF11RF11:
     case TextureFormat::ASTC_2D_4X4:
         CopySwizzledData(width, height, bytes_per_pixel, bytes_per_pixel, data,
@@ -174,6 +177,7 @@ std::vector<u8> DecodeTexture(const std::vector<u8>& texture_data, TextureFormat
     case TextureFormat::G8R8:
     case TextureFormat::BF10GF11RF11:
     case TextureFormat::R32_G32_B32_A32:
+    case TextureFormat::R32_G32:
         // TODO(Subv): For the time being just forward the same data without any decoding.
         rgba_data = texture_data;
         break;
-- 
GitLab