Merge cda72cde54 into c7f2fb2151

gl_shader_gen: Use a std::vector to represent program code instead of std::array

src/core/hle/service/nvdrv/devices/nvhost_gpu.h

@@ -10,6 +10,7 @@
 #include "common/common_types.h"
 #include "common/swap.h"
 #include "core/hle/service/nvdrv/devices/nvdevice.h"
+#include "video_core/memory_manager.h"
 
 namespace Service::Nvidia::Devices {
 

src/core/memory.cpp

@@ -251,8 +251,8 @@ std::string ReadCString(VAddr vaddr, std::size_t max_length) {
     return string;
 }
 
-void RasterizerMarkRegionCached(Tegra::GPUVAddr gpu_addr, u64 size, bool cached) {
-    if (gpu_addr == 0) {
+void RasterizerMarkRegionCached(VAddr vaddr, u64 size, bool cached) {
+    if (vaddr == 0) {
         return;
     }
 
@@ -261,19 +261,8 @@ void RasterizerMarkRegionCached(Tegra::GPUVAddr gpu_addr, u64 size, bool cached)
     // CPU pages, hence why we iterate on a CPU page basis (note: GPU page size is different). This
     // assumes the specified GPU address region is contiguous as well.
 
-    u64 num_pages = ((gpu_addr + size - 1) >> PAGE_BITS) - (gpu_addr >> PAGE_BITS) + 1;
-    for (unsigned i = 0; i < num_pages; ++i, gpu_addr += PAGE_SIZE) {
-        boost::optional<VAddr> maybe_vaddr =
-            Core::System::GetInstance().GPU().memory_manager->GpuToCpuAddress(gpu_addr);
-        // The GPU <-> CPU virtual memory mapping is not 1:1
-        if (!maybe_vaddr) {
-            LOG_ERROR(HW_Memory,
-                      "Trying to flush a cached region to an invalid physical address {:016X}",
-                      gpu_addr);
-            continue;
-        }
-        VAddr vaddr = *maybe_vaddr;
-
+    u64 num_pages = ((vaddr + size - 1) >> PAGE_BITS) - (vaddr >> PAGE_BITS) + 1;
+    for (unsigned i = 0; i < num_pages; ++i, vaddr += PAGE_SIZE) {
         PageType& page_type = current_page_table->attributes[vaddr >> PAGE_BITS];
 
         if (cached) {
@@ -344,29 +333,19 @@ void RasterizerFlushVirtualRegion(VAddr start, u64 size, FlushMode mode) {
 
         const VAddr overlap_start = std::max(start, region_start);
         const VAddr overlap_end = std::min(end, region_end);
-
-        const std::vector<Tegra::GPUVAddr> gpu_addresses =
-            system_instance.GPU().memory_manager->CpuToGpuAddress(overlap_start);
-
-        if (gpu_addresses.empty()) {
-            return;
-        }
-
         const u64 overlap_size = overlap_end - overlap_start;
 
-        for (const auto& gpu_address : gpu_addresses) {
-            auto& rasterizer = system_instance.Renderer().Rasterizer();
-            switch (mode) {
-            case FlushMode::Flush:
-                rasterizer.FlushRegion(gpu_address, overlap_size);
-                break;
-            case FlushMode::Invalidate:
-                rasterizer.InvalidateRegion(gpu_address, overlap_size);
-                break;
-            case FlushMode::FlushAndInvalidate:
-                rasterizer.FlushAndInvalidateRegion(gpu_address, overlap_size);
-                break;
-            }
+        auto& rasterizer = system_instance.Renderer().Rasterizer();
+        switch (mode) {
+        case FlushMode::Flush:
+            rasterizer.FlushRegion(overlap_start, overlap_size);
+            break;
+        case FlushMode::Invalidate:
+            rasterizer.InvalidateRegion(overlap_start, overlap_size);
+            break;
+        case FlushMode::FlushAndInvalidate:
+            rasterizer.FlushAndInvalidateRegion(overlap_start, overlap_size);
+            break;
         }
     };
 

src/core/memory.h

@@ -11,7 +11,6 @@
 #include <boost/icl/interval_map.hpp>
 #include "common/common_types.h"
 #include "core/memory_hook.h"
-#include "video_core/memory_manager.h"
 
 namespace Kernel {
 class Process;
@@ -179,7 +178,7 @@ enum class FlushMode {
 /**
  * Mark each page touching the region as cached.
  */
-void RasterizerMarkRegionCached(Tegra::GPUVAddr gpu_addr, u64 size, bool cached);
+void RasterizerMarkRegionCached(VAddr vaddr, u64 size, bool cached);
 
 /**
  * Flushes and invalidates any externally cached rasterizer resources touching the given virtual

src/video_core/engines/shader_bytecode.h

@@ -280,6 +280,19 @@ union Instruction {
             BitField<56, 1, u64> invert_b;
         } lop32i;
 
+        union {
+            BitField<28, 8, u64> imm_lut28;
+            BitField<48, 8, u64> imm_lut48;
+
+            u32 GetImmLut28() const {
+                return static_cast<u32>(imm_lut28);
+            }
+
+            u32 GetImmLut48() const {
+                return static_cast<u32>(imm_lut48);
+            }
+        } lop3;
+
         u32 GetImm20_19() const {
             u32 imm{static_cast<u32>(imm20_19)};
             imm <<= 12;
@@ -650,6 +663,9 @@ public:
         LOP_R,
         LOP_IMM,
         LOP32I,
+        LOP3_C,
+        LOP3_R,
+        LOP3_IMM,
         MOV_C,
         MOV_R,
         MOV_IMM,
@@ -872,6 +888,9 @@ private:
             INST("0101110001000---", Id::LOP_R, Type::ArithmeticInteger, "LOP_R"),
             INST("0011100001000---", Id::LOP_IMM, Type::ArithmeticInteger, "LOP_IMM"),
             INST("000001----------", Id::LOP32I, Type::ArithmeticIntegerImmediate, "LOP32I"),
+            INST("0000001---------", Id::LOP3_C, Type::ArithmeticInteger, "LOP3_C"),
+            INST("0101101111100---", Id::LOP3_R, Type::ArithmeticInteger, "LOP3_R"),
+            INST("0011110---------", Id::LOP3_IMM, Type::ArithmeticInteger, "LOP3_IMM"),
             INST("0100110001001---", Id::SHL_C, Type::Shift, "SHL_C"),
             INST("0101110001001---", Id::SHL_R, Type::Shift, "SHL_R"),
             INST("0011100-01001---", Id::SHL_IMM, Type::Shift, "SHL_IMM"),

src/video_core/rasterizer_interface.h

@@ -27,14 +27,14 @@ public:
     virtual void FlushAll() = 0;
 
     /// Notify rasterizer that any caches of the specified region should be flushed to Switch memory
-    virtual void FlushRegion(Tegra::GPUVAddr addr, u64 size) = 0;
+    virtual void FlushRegion(VAddr addr, u64 size) = 0;
 
     /// Notify rasterizer that any caches of the specified region should be invalidated
-    virtual void InvalidateRegion(Tegra::GPUVAddr addr, u64 size) = 0;
+    virtual void InvalidateRegion(VAddr addr, u64 size) = 0;
 
     /// Notify rasterizer that any caches of the specified region should be flushed to Switch memory
     /// and invalidated
-    virtual void FlushAndInvalidateRegion(Tegra::GPUVAddr addr, u64 size) = 0;
+    virtual void FlushAndInvalidateRegion(VAddr addr, u64 size) = 0;
 
     /// Attempt to use a faster method to perform a display transfer with is_texture_copy = 0
     virtual bool AccelerateDisplayTransfer(const void* config) {

src/video_core/renderer_opengl/gl_rasterizer.cpp

@@ -25,6 +25,8 @@
 #include "video_core/renderer_opengl/renderer_opengl.h"
 #include "video_core/video_core.h"
 
+namespace OpenGL {
+
 using Maxwell = Tegra::Engines::Maxwell3D::Regs;
 using PixelFormat = SurfaceParams::PixelFormat;
 using SurfaceType = SurfaceParams::SurfaceType;
@@ -179,7 +181,7 @@ static GLShader::ProgramCode GetShaderProgramCode(Maxwell::ShaderProgram program
     auto& gpu = Core::System::GetInstance().GPU().Maxwell3D();
 
     // Fetch program code from memory
-    GLShader::ProgramCode program_code;
+    GLShader::ProgramCode program_code(GLShader::MAX_PROGRAM_CODE_LENGTH);
     auto& shader_config = gpu.regs.shader_config[static_cast<size_t>(program)];
     const u64 gpu_address{gpu.regs.code_address.CodeAddress() + shader_config.offset};
     const boost::optional<VAddr> cpu_address{gpu.memory_manager.GpuToCpuAddress(gpu_address)};
@@ -543,17 +545,17 @@ void RasterizerOpenGL::FlushAll() {
     res_cache.FlushRegion(0, Kernel::VMManager::MAX_ADDRESS);
 }
 
-void RasterizerOpenGL::FlushRegion(Tegra::GPUVAddr addr, u64 size) {
+void RasterizerOpenGL::FlushRegion(VAddr addr, u64 size) {
     MICROPROFILE_SCOPE(OpenGL_CacheManagement);
     res_cache.FlushRegion(addr, size);
 }
 
-void RasterizerOpenGL::InvalidateRegion(Tegra::GPUVAddr addr, u64 size) {
+void RasterizerOpenGL::InvalidateRegion(VAddr addr, u64 size) {
     MICROPROFILE_SCOPE(OpenGL_CacheManagement);
     res_cache.InvalidateRegion(addr, size);
 }
 
-void RasterizerOpenGL::FlushAndInvalidateRegion(Tegra::GPUVAddr addr, u64 size) {
+void RasterizerOpenGL::FlushAndInvalidateRegion(VAddr addr, u64 size) {
     MICROPROFILE_SCOPE(OpenGL_CacheManagement);
     res_cache.FlushRegion(addr, size);
     res_cache.InvalidateRegion(addr, size);
@@ -874,3 +876,5 @@ void RasterizerOpenGL::SyncLogicOpState() {
 
     state.logic_op.operation = MaxwellToGL::LogicOp(regs.logic_op.operation);
 }
+
+} // namespace OpenGL

src/video_core/renderer_opengl/gl_rasterizer.h

@@ -22,12 +22,14 @@
 #include "video_core/renderer_opengl/gl_state.h"
 #include "video_core/renderer_opengl/gl_stream_buffer.h"
 
-struct ScreenInfo;
-
 namespace Core::Frontend {
 class EmuWindow;
 }
 
+namespace OpenGL {
+
+struct ScreenInfo;
+
 class RasterizerOpenGL : public VideoCore::RasterizerInterface {
 public:
     explicit RasterizerOpenGL(Core::Frontend::EmuWindow& renderer, ScreenInfo& info);
@@ -37,9 +39,9 @@ public:
     void Clear() override;
     void NotifyMaxwellRegisterChanged(u32 method) override;
     void FlushAll() override;
-    void FlushRegion(Tegra::GPUVAddr addr, u64 size) override;
-    void InvalidateRegion(Tegra::GPUVAddr addr, u64 size) override;
-    void FlushAndInvalidateRegion(Tegra::GPUVAddr addr, u64 size) override;
+    void FlushRegion(VAddr addr, u64 size) override;
+    void InvalidateRegion(VAddr addr, u64 size) override;
+    void FlushAndInvalidateRegion(VAddr addr, u64 size) override;
     bool AccelerateDisplayTransfer(const void* config) override;
     bool AccelerateTextureCopy(const void* config) override;
     bool AccelerateFill(const void* config) override;
@@ -184,3 +186,5 @@ private:
     enum class AccelDraw { Disabled, Arrays, Indexed };
     AccelDraw accelerate_draw = AccelDraw::Disabled;
 };
+
+} // namespace OpenGL

src/video_core/renderer_opengl/gl_rasterizer_cache.cpp

@@ -19,6 +19,8 @@
 #include "video_core/textures/decoders.h"
 #include "video_core/utils.h"
 
+namespace OpenGL {
+
 using SurfaceType = SurfaceParams::SurfaceType;
 using PixelFormat = SurfaceParams::PixelFormat;
 using ComponentType = SurfaceParams::ComponentType;
@@ -33,9 +35,9 @@ struct FormatTuple {
 
 /*static*/ SurfaceParams SurfaceParams::CreateForTexture(
     const Tegra::Texture::FullTextureInfo& config) {
-
+    const auto& gpu = Core::System::GetInstance().GPU();
     SurfaceParams params{};
-    params.addr = config.tic.Address();
+    params.cpu_addr = *gpu.memory_manager->GpuToCpuAddress(config.tic.Address());
     params.is_tiled = config.tic.IsTiled();
     params.block_height = params.is_tiled ? config.tic.BlockHeight() : 0,
     params.pixel_format =
@@ -53,9 +55,9 @@ struct FormatTuple {
 
 /*static*/ SurfaceParams SurfaceParams::CreateForFramebuffer(
     const Tegra::Engines::Maxwell3D::Regs::RenderTargetConfig& config) {
-
+    const auto& gpu = Core::System::GetInstance().GPU();
     SurfaceParams params{};
-    params.addr = config.Address();
+    params.cpu_addr = *gpu.memory_manager->GpuToCpuAddress(config.Address());
     params.is_tiled = true;
     params.block_height = Tegra::Texture::TICEntry::DefaultBlockHeight;
     params.pixel_format = PixelFormatFromRenderTargetFormat(config.format);
@@ -73,9 +75,9 @@ struct FormatTuple {
 /*static*/ SurfaceParams SurfaceParams::CreateForDepthBuffer(u32 zeta_width, u32 zeta_height,
                                                              Tegra::GPUVAddr zeta_address,
                                                              Tegra::DepthFormat format) {
-
+    const auto& gpu = Core::System::GetInstance().GPU();
     SurfaceParams params{};
-    params.addr = zeta_address;
+    params.cpu_addr = *gpu.memory_manager->GpuToCpuAddress(zeta_address);
     params.is_tiled = true;
     params.block_height = Tegra::Texture::TICEntry::DefaultBlockHeight;
     params.pixel_format = PixelFormatFromDepthFormat(format);
@@ -165,11 +167,6 @@ static const FormatTuple& GetFormatTuple(PixelFormat pixel_format, ComponentType
     return format;
 }
 
-VAddr SurfaceParams::GetCpuAddr() const {
-    const auto& gpu = Core::System::GetInstance().GPU();
-    return *gpu.memory_manager->GpuToCpuAddress(addr);
-}
-
 static bool IsPixelFormatASTC(PixelFormat format) {
     switch (format) {
     case PixelFormat::ASTC_2D_4X4:
@@ -214,33 +211,28 @@ static bool IsFormatBCn(PixelFormat format) {
 }
 
 template <bool morton_to_gl, PixelFormat format>
-void MortonCopy(u32 stride, u32 block_height, u32 height, std::vector<u8>& gl_buffer,
-                Tegra::GPUVAddr addr) {
+void MortonCopy(u32 stride, u32 block_height, u32 height, std::vector<u8>& gl_buffer, VAddr addr) {
     constexpr u32 bytes_per_pixel = SurfaceParams::GetFormatBpp(format) / CHAR_BIT;
     constexpr u32 gl_bytes_per_pixel = CachedSurface::GetGLBytesPerPixel(format);
-    const auto& gpu = Core::System::GetInstance().GPU();
 
     if (morton_to_gl) {
         // With the BCn formats (DXT and DXN), each 4x4 tile is swizzled instead of just individual
         // pixel values.
         const u32 tile_size{IsFormatBCn(format) ? 4U : 1U};
-        const std::vector<u8> data =
-            Tegra::Texture::UnswizzleTexture(*gpu.memory_manager->GpuToCpuAddress(addr), tile_size,
-                                             bytes_per_pixel, stride, height, block_height);
+        const std::vector<u8> data = Tegra::Texture::UnswizzleTexture(
+            addr, tile_size, bytes_per_pixel, stride, height, block_height);
         const size_t size_to_copy{std::min(gl_buffer.size(), data.size())};
         gl_buffer.assign(data.begin(), data.begin() + size_to_copy);
     } else {
         // TODO(bunnei): Assumes the default rendering GOB size of 16 (128 lines). We should
         // check the configuration for this and perform more generic un/swizzle
         LOG_WARNING(Render_OpenGL, "need to use correct swizzle/GOB parameters!");
-        VideoCore::MortonCopyPixels128(
-            stride, height, bytes_per_pixel, gl_bytes_per_pixel,
-            Memory::GetPointer(*gpu.memory_manager->GpuToCpuAddress(addr)), gl_buffer.data(),
-            morton_to_gl);
+        VideoCore::MortonCopyPixels128(stride, height, bytes_per_pixel, gl_bytes_per_pixel,
+                                       Memory::GetPointer(addr), gl_buffer.data(), morton_to_gl);
     }
 }
 
-static constexpr std::array<void (*)(u32, u32, u32, std::vector<u8>&, Tegra::GPUVAddr),
+static constexpr std::array<void (*)(u32, u32, u32, std::vector<u8>&, VAddr),
                             SurfaceParams::MaxPixelFormat>
     morton_to_gl_fns = {
         // clang-format off
@@ -295,7 +287,7 @@ static constexpr std::array<void (*)(u32, u32, u32, std::vector<u8>&, Tegra::GPU
         // clang-format on
 };
 
-static constexpr std::array<void (*)(u32, u32, u32, std::vector<u8>&, Tegra::GPUVAddr),
+static constexpr std::array<void (*)(u32, u32, u32, std::vector<u8>&, VAddr),
                             SurfaceParams::MaxPixelFormat>
     gl_to_morton_fns = {
         // clang-format off
@@ -530,7 +522,7 @@ MICROPROFILE_DEFINE(OpenGL_SurfaceLoad, "OpenGL", "Surface Load", MP_RGB(128, 64
 void CachedSurface::LoadGLBuffer() {
     ASSERT(params.type != SurfaceType::Fill);
 
-    const u8* const texture_src_data = Memory::GetPointer(params.GetCpuAddr());
+    const u8* const texture_src_data = Memory::GetPointer(params.cpu_addr);
 
     ASSERT(texture_src_data);
 
@@ -543,7 +535,7 @@ void CachedSurface::LoadGLBuffer() {
         gl_buffer.resize(copy_size);
 
         morton_to_gl_fns[static_cast<size_t>(params.pixel_format)](
-            params.width, params.block_height, params.height, gl_buffer, params.addr);
+            params.width, params.block_height, params.height, gl_buffer, params.cpu_addr);
     } else {
         const u8* const texture_src_data_end = texture_src_data + copy_size;
 
@@ -555,7 +547,7 @@ void CachedSurface::LoadGLBuffer() {
 
 MICROPROFILE_DEFINE(OpenGL_SurfaceFlush, "OpenGL", "Surface Flush", MP_RGB(128, 192, 64));
 void CachedSurface::FlushGLBuffer() {
-    u8* const dst_buffer = Memory::GetPointer(params.GetCpuAddr());
+    u8* const dst_buffer = Memory::GetPointer(params.cpu_addr);
 
     ASSERT(dst_buffer);
     ASSERT(gl_buffer.size() ==
@@ -570,7 +562,7 @@ void CachedSurface::FlushGLBuffer() {
         std::memcpy(dst_buffer, gl_buffer.data(), params.size_in_bytes);
     } else {
         gl_to_morton_fns[static_cast<size_t>(params.pixel_format)](
-            params.width, params.block_height, params.height, gl_buffer, params.addr);
+            params.width, params.block_height, params.height, gl_buffer, params.cpu_addr);
     }
 }
 
@@ -754,17 +746,12 @@ void RasterizerCacheOpenGL::FlushSurface(const Surface& surface) {
 }
 
 Surface RasterizerCacheOpenGL::GetSurface(const SurfaceParams& params, bool preserve_contents) {
-    if (params.addr == 0 || params.height * params.width == 0) {
+    if (params.cpu_addr == 0 || params.height * params.width == 0) {
         return {};
     }
 
-    const auto& gpu = Core::System::GetInstance().GPU();
-    // Don't try to create any entries in the cache if the address of the texture is invalid.
-    if (gpu.memory_manager->GpuToCpuAddress(params.addr) == boost::none)
-        return {};
-
     // Look up surface in the cache based on address
-    const auto& search{surface_cache.find(params.addr)};
+    const auto& search{surface_cache.find(params.cpu_addr)};
     Surface surface;
     if (search != surface_cache.end()) {
         surface = search->second;
@@ -797,9 +784,6 @@ Surface RasterizerCacheOpenGL::RecreateSurface(const Surface& surface,
                                                const SurfaceParams& new_params) {
     // Verify surface is compatible for blitting
     const auto& params{surface->GetSurfaceParams()};
-    ASSERT(params.type == new_params.type);
-    ASSERT_MSG(params.GetCompressionFactor(params.pixel_format) == 1,
-               "Compressed texture reinterpretation is not supported");
 
     // Create a new surface with the new parameters, and blit the previous surface to it
     Surface new_surface{std::make_shared<CachedSurface>(new_params)};
@@ -816,9 +800,12 @@ Surface RasterizerCacheOpenGL::RecreateSurface(const Surface& surface,
 
     glBindBuffer(GL_PIXEL_PACK_BUFFER, pbo.handle);
     glBufferData(GL_PIXEL_PACK_BUFFER, buffer_size, nullptr, GL_STREAM_DRAW_ARB);
-    glGetTextureImage(surface->Texture().handle, 0, source_format.format, source_format.type,
-                      params.SizeInBytes(), nullptr);
-
+    if (source_format.compressed) {
+        glGetCompressedTextureImage(surface->Texture().handle, 0, params.SizeInBytes(), nullptr);
+    } else {
+        glGetTextureImage(surface->Texture().handle, 0, source_format.format, source_format.type,
+                          params.SizeInBytes(), nullptr);
+    }
     // If the new texture is bigger than the previous one, we need to fill in the rest with data
     // from the CPU.
     if (params.SizeInBytes() < new_params.SizeInBytes()) {
@@ -832,10 +819,8 @@ Surface RasterizerCacheOpenGL::RecreateSurface(const Surface& surface,
                                  "reinterpretation but the texture is tiled.");
         }
         size_t remaining_size = new_params.SizeInBytes() - params.SizeInBytes();
-        auto address = Core::System::GetInstance().GPU().memory_manager->GpuToCpuAddress(
-            new_params.addr + params.SizeInBytes());
         std::vector<u8> data(remaining_size);
-        Memory::ReadBlock(*address, data.data(), data.size());
+        Memory::ReadBlock(new_params.cpu_addr + params.SizeInBytes(), data.data(), data.size());
         glBufferSubData(GL_PIXEL_PACK_BUFFER, params.SizeInBytes(), remaining_size, data.data());
     }
 
@@ -844,9 +829,32 @@ Surface RasterizerCacheOpenGL::RecreateSurface(const Surface& surface,
     const auto& dest_rect{new_params.GetRect()};
 
     glBindBuffer(GL_PIXEL_UNPACK_BUFFER, pbo.handle);
-    glTextureSubImage2D(
-        new_surface->Texture().handle, 0, 0, 0, static_cast<GLsizei>(dest_rect.GetWidth()),
-        static_cast<GLsizei>(dest_rect.GetHeight()), dest_format.format, dest_format.type, nullptr);
+    if (dest_format.compressed) {
+        OpenGLState cur_state = OpenGLState::GetCurState();
+
+        GLuint old_tex = cur_state.texture_units[0].texture_2d;
+        cur_state.texture_units[0].texture_2d = new_surface->Texture().handle;
+        cur_state.Apply();
+
+        // Ensure no bad interactions with GL_UNPACK_ALIGNMENT
+        ASSERT(new_params.width * CachedSurface::GetGLBytesPerPixel(new_params.pixel_format) % 4 ==
+               0);
+        glPixelStorei(GL_UNPACK_ROW_LENGTH, static_cast<GLint>(params.width));
+        glActiveTexture(GL_TEXTURE0);
+        glCompressedTexImage2D(GL_TEXTURE_2D, 0, dest_format.internal_format,
+                               static_cast<GLsizei>(dest_rect.GetWidth()),
+                               static_cast<GLsizei>(dest_rect.GetHeight()), 0,
+                               static_cast<GLsizei>(new_params.SizeInBytes()), nullptr);
+        glPixelStorei(GL_UNPACK_ROW_LENGTH, 0);
+
+        cur_state.texture_units[0].texture_2d = old_tex;
+        cur_state.Apply();
+    } else {
+        glTextureSubImage2D(new_surface->Texture().handle, 0, 0, 0,
+                            static_cast<GLsizei>(dest_rect.GetWidth()),
+                            static_cast<GLsizei>(dest_rect.GetHeight()), dest_format.format,
+                            dest_format.type, nullptr);
+    }
     glBindBuffer(GL_PIXEL_UNPACK_BUFFER, 0);
 
     pbo.Release();
@@ -868,9 +876,8 @@ Surface RasterizerCacheOpenGL::TryFindFramebufferSurface(VAddr cpu_addr) const {
     std::vector<Surface> surfaces;
     for (const auto& surface : surface_cache) {
         const auto& params = surface.second->GetSurfaceParams();
-        const VAddr surface_cpu_addr = params.GetCpuAddr();
-        if (cpu_addr >= surface_cpu_addr && cpu_addr < (surface_cpu_addr + params.size_in_bytes)) {
-            ASSERT_MSG(cpu_addr == surface_cpu_addr, "overlapping surfaces are unsupported");
+        if (cpu_addr >= params.cpu_addr && cpu_addr < (params.cpu_addr + params.size_in_bytes)) {
+            ASSERT_MSG(cpu_addr == params.cpu_addr, "overlapping surfaces are unsupported");
             surfaces.push_back(surface.second);
         }
     }
@@ -884,13 +891,13 @@ Surface RasterizerCacheOpenGL::TryFindFramebufferSurface(VAddr cpu_addr) const {
     return surfaces[0];
 }
 
-void RasterizerCacheOpenGL::FlushRegion(Tegra::GPUVAddr /*addr*/, size_t /*size*/) {
+void RasterizerCacheOpenGL::FlushRegion(VAddr /*addr*/, size_t /*size*/) {
     // TODO(bunnei): This is unused in the current implementation of the rasterizer cache. We should
     // probably implement this in the future, but for now, the `use_accurate_framebufers` setting
     // can be used to always flush.
 }
 
-void RasterizerCacheOpenGL::InvalidateRegion(Tegra::GPUVAddr addr, size_t size) {
+void RasterizerCacheOpenGL::InvalidateRegion(VAddr addr, size_t size) {
     for (auto iter = surface_cache.cbegin(); iter != surface_cache.cend();) {
         const auto& surface{iter->second};
         const auto& params{surface->GetSurfaceParams()};
@@ -905,27 +912,27 @@ void RasterizerCacheOpenGL::InvalidateRegion(Tegra::GPUVAddr addr, size_t size)
 
 void RasterizerCacheOpenGL::RegisterSurface(const Surface& surface) {
     const auto& params{surface->GetSurfaceParams()};
-    const auto& search{surface_cache.find(params.addr)};
+    const auto& search{surface_cache.find(params.cpu_addr)};
 
     if (search != surface_cache.end()) {
         // Registered already
         return;
     }
 
-    surface_cache[params.addr] = surface;
-    UpdatePagesCachedCount(params.addr, params.size_in_bytes, 1);
+    surface_cache[params.cpu_addr] = surface;
+    UpdatePagesCachedCount(params.cpu_addr, params.size_in_bytes, 1);
 }
 
 void RasterizerCacheOpenGL::UnregisterSurface(const Surface& surface) {
     const auto& params{surface->GetSurfaceParams()};
-    const auto& search{surface_cache.find(params.addr)};
+    const auto& search{surface_cache.find(params.cpu_addr)};
 
     if (search == surface_cache.end()) {
         // Unregistered already
         return;
     }
 
-    UpdatePagesCachedCount(params.addr, params.size_in_bytes, -1);
+    UpdatePagesCachedCount(params.cpu_addr, params.size_in_bytes, -1);
     surface_cache.erase(search);
 }
 
@@ -934,10 +941,10 @@ constexpr auto RangeFromInterval(Map& map, const Interval& interval) {
     return boost::make_iterator_range(map.equal_range(interval));
 }
 
-void RasterizerCacheOpenGL::UpdatePagesCachedCount(Tegra::GPUVAddr addr, u64 size, int delta) {
-    const u64 num_pages = ((addr + size - 1) >> Tegra::MemoryManager::PAGE_BITS) -
-                          (addr >> Tegra::MemoryManager::PAGE_BITS) + 1;
-    const u64 page_start = addr >> Tegra::MemoryManager::PAGE_BITS;
+void RasterizerCacheOpenGL::UpdatePagesCachedCount(VAddr addr, u64 size, int delta) {
+    const u64 num_pages =
+        ((addr + size - 1) >> Memory::PAGE_BITS) - (addr >> Memory::PAGE_BITS) + 1;
+    const u64 page_start = addr >> Memory::PAGE_BITS;
     const u64 page_end = page_start + num_pages;
 
     // Interval maps will erase segments if count reaches 0, so if delta is negative we have to
@@ -950,10 +957,8 @@ void RasterizerCacheOpenGL::UpdatePagesCachedCount(Tegra::GPUVAddr addr, u64 siz
         const auto interval = pair.first & pages_interval;
         const int count = pair.second;
 
-        const Tegra::GPUVAddr interval_start_addr = boost::icl::first(interval)
-                                                    << Tegra::MemoryManager::PAGE_BITS;
-        const Tegra::GPUVAddr interval_end_addr = boost::icl::last_next(interval)
-                                                  << Tegra::MemoryManager::PAGE_BITS;
+        const VAddr interval_start_addr = boost::icl::first(interval) << Memory::PAGE_BITS;
+        const VAddr interval_end_addr = boost::icl::last_next(interval) << Memory::PAGE_BITS;
         const u64 interval_size = interval_end_addr - interval_start_addr;
 
         if (delta > 0 && count == delta)
@@ -967,3 +972,5 @@ void RasterizerCacheOpenGL::UpdatePagesCachedCount(Tegra::GPUVAddr addr, u64 siz
     if (delta < 0)
         cached_pages.add({pages_interval, delta});
 }
+
+} // namespace OpenGL

src/video_core/renderer_opengl/gl_rasterizer_cache.h

@@ -16,6 +16,8 @@
 #include "video_core/renderer_opengl/gl_resource_manager.h"
 #include "video_core/textures/texture.h"
 
+namespace OpenGL {
+
 class CachedSurface;
 using Surface = std::shared_ptr<CachedSurface>;
 using SurfaceSurfaceRect_Tuple = std::tuple<Surface, Surface, MathUtil::Rectangle<u32>>;
@@ -626,12 +628,9 @@ struct SurfaceParams {
                GetFormatBpp(pixel_format) / CHAR_BIT;
     }
 
-    /// Returns the CPU virtual address for this surface
-    VAddr GetCpuAddr() const;
-
     /// Returns true if the specified region overlaps with this surface's region in Switch memory
-    bool IsOverlappingRegion(Tegra::GPUVAddr region_addr, size_t region_size) const {
-        return addr <= (region_addr + region_size) && region_addr <= (addr + size_in_bytes);
+    bool IsOverlappingRegion(VAddr region_addr, size_t region_size) const {
+        return cpu_addr <= (region_addr + region_size) && region_addr <= (cpu_addr + size_in_bytes);
     }
 
     /// Creates SurfaceParams from a texture configuration
@@ -647,9 +646,9 @@ struct SurfaceParams {
                                               Tegra::DepthFormat format);
 
     bool operator==(const SurfaceParams& other) const {
-        return std::tie(addr, is_tiled, block_height, pixel_format, component_type, type, width,
+        return std::tie(cpu_addr, is_tiled, block_height, pixel_format, component_type, type, width,
                         height, unaligned_height, size_in_bytes) ==
-               std::tie(other.addr, other.is_tiled, other.block_height, other.pixel_format,
+               std::tie(other.cpu_addr, other.is_tiled, other.block_height, other.pixel_format,
                         other.component_type, other.type, other.width, other.height,
                         other.unaligned_height, other.size_in_bytes);
     }
@@ -664,7 +663,7 @@ struct SurfaceParams {
                std::tie(other.pixel_format, other.type, other.cache_width, other.cache_height);
     }
 
-    Tegra::GPUVAddr addr;
+    VAddr cpu_addr;
     bool is_tiled;
     u32 block_height;
     PixelFormat pixel_format;
@@ -732,10 +731,10 @@ public:
     Surface TryFindFramebufferSurface(VAddr cpu_addr) const;
 
     /// Write any cached resources overlapping the region back to memory (if dirty)
-    void FlushRegion(Tegra::GPUVAddr addr, size_t size);
+    void FlushRegion(VAddr addr, size_t size);
 
     /// Mark the specified region as being invalidated
-    void InvalidateRegion(Tegra::GPUVAddr addr, size_t size);
+    void InvalidateRegion(VAddr addr, size_t size);
 
 private:
     void LoadSurface(const Surface& surface);
@@ -751,11 +750,13 @@ private:
     void UnregisterSurface(const Surface& surface);
 
     /// Increase/decrease the number of surface in pages touching the specified region
-    void UpdatePagesCachedCount(Tegra::GPUVAddr addr, u64 size, int delta);
+    void UpdatePagesCachedCount(VAddr addr, u64 size, int delta);
 
-    std::unordered_map<Tegra::GPUVAddr, Surface> surface_cache;
+    std::unordered_map<VAddr, Surface> surface_cache;
     PageMap cached_pages;
 
     OGLFramebuffer read_framebuffer;
     OGLFramebuffer draw_framebuffer;
 };
+
+} // namespace OpenGL

src/video_core/renderer_opengl/gl_resource_manager.h

@@ -10,6 +10,8 @@
 #include "video_core/renderer_opengl/gl_shader_util.h"
 #include "video_core/renderer_opengl/gl_state.h"
 
+namespace OpenGL {
+
 class OGLTexture : private NonCopyable {
 public:
     OGLTexture() = default;
@@ -331,3 +333,5 @@ public:
 
     GLuint handle = 0;
 };
+
+} // namespace OpenGL

src/video_core/renderer_opengl/gl_shader_decompiler.cpp

@@ -15,7 +15,7 @@
 #include "video_core/renderer_opengl/gl_rasterizer.h"
 #include "video_core/renderer_opengl/gl_shader_decompiler.h"
 
-namespace GLShader::Decompiler {
+namespace OpenGL::GLShader::Decompiler {
 
 using Tegra::Shader::Attribute;
 using Tegra::Shader::Instruction;
@@ -849,6 +849,33 @@ private:
         }
     }
 
+    void WriteLop3Instruction(Register dest, const std::string& op_a, const std::string& op_b,
+                              const std::string& op_c, const std::string& imm_lut) {
+        if (dest == Tegra::Shader::Register::ZeroIndex) {
+            return;
+        }
+
+        static constexpr std::array<const char*, 32> shift_amounts = {
+            "0",  "1",  "2",  "3",  "4",  "5",  "6",  "7",  "8",  "9",  "10",
+            "11", "12", "13", "14", "15", "16", "17", "18", "19", "20", "21",
+            "22", "23", "24", "25", "26", "27", "28", "29", "30", "31"};
+
+        std::string result;
+        result += '(';
+
+        for (size_t i = 0; i < shift_amounts.size(); ++i) {
+            if (i)
+                result += '|';
+            result += "(((" + imm_lut + " >> (((" + op_c + " >> " + shift_amounts[i] +
+                      ") & 1) | ((" + op_b + " >> " + shift_amounts[i] + ") & 1) << 1 | ((" + op_a +
+                      " >> " + shift_amounts[i] + ") & 1) << 2)) & 1) << " + shift_amounts[i] + ")";
+        }
+
+        result += ')';
+
+        regs.SetRegisterToInteger(dest, true, 0, result, 1, 1);
+    }
+
     void WriteTexsInstruction(const Instruction& instr, const std::string& coord,
                               const std::string& texture) {
         // Add an extra scope and declare the texture coords inside to prevent
@@ -1297,6 +1324,20 @@ private:
                                     instr.alu.lop.pred_result_mode, instr.alu.lop.pred48);
                 break;
             }
+            case OpCode::Id::LOP3_C:
+            case OpCode::Id::LOP3_R:
+            case OpCode::Id::LOP3_IMM: {
+                std::string op_c = regs.GetRegisterAsInteger(instr.gpr39);
+                std::string lut;
+                if (opcode->GetId() == OpCode::Id::LOP3_R) {
+                    lut = '(' + std::to_string(instr.alu.lop3.GetImmLut28()) + ')';
+                } else {
+                    lut = '(' + std::to_string(instr.alu.lop3.GetImmLut48()) + ')';
+                }
+
+                WriteLop3Instruction(instr.gpr0, op_a, op_b, op_c, lut);
+                break;
+            }
             case OpCode::Id::IMNMX_C:
             case OpCode::Id::IMNMX_R:
             case OpCode::Id::IMNMX_IMM: {
@@ -2176,4 +2217,4 @@ boost::optional<ProgramResult> DecompileProgram(const ProgramCode& program_code,
     return boost::none;
 }
 
-} // namespace GLShader::Decompiler
+} // namespace OpenGL::GLShader::Decompiler

src/video_core/renderer_opengl/gl_shader_decompiler.h

@@ -12,7 +12,7 @@
 #include "video_core/engines/maxwell_3d.h"
 #include "video_core/renderer_opengl/gl_shader_gen.h"
 
-namespace GLShader::Decompiler {
+namespace OpenGL::GLShader::Decompiler {
 
 using Tegra::Engines::Maxwell3D;
 
@@ -22,4 +22,4 @@ boost::optional<ProgramResult> DecompileProgram(const ProgramCode& program_code,
                                                 Maxwell3D::Regs::ShaderStage stage,
                                                 const std::string& suffix);
 
-} // namespace GLShader::Decompiler
+} // namespace OpenGL::GLShader::Decompiler

src/video_core/renderer_opengl/gl_shader_gen.cpp

@@ -7,7 +7,7 @@
 #include "video_core/renderer_opengl/gl_shader_decompiler.h"
 #include "video_core/renderer_opengl/gl_shader_gen.h"
 
-namespace GLShader {
+namespace OpenGL::GLShader {
 
 using Tegra::Engines::Maxwell3D;
 
@@ -103,4 +103,4 @@ void main() {
     return {out, program.second};
 }
 
-} // namespace GLShader
+} // namespace OpenGL::GLShader

src/video_core/renderer_opengl/gl_shader_gen.h

@@ -9,14 +9,14 @@
 #include <type_traits>
 #include <utility>
 #include <vector>
+#include <boost/functional/hash.hpp>
 #include "common/common_types.h"
 #include "common/hash.h"
 
-namespace GLShader {
+namespace OpenGL::GLShader {
 
 constexpr size_t MAX_PROGRAM_CODE_LENGTH{0x1000};
-
-using ProgramCode = std::array<u64, MAX_PROGRAM_CODE_LENGTH>;
+using ProgramCode = std::vector<u64>;
 
 class ConstBufferEntry {
     using Maxwell = Tegra::Engines::Maxwell3D::Regs;
@@ -115,8 +115,8 @@ struct ShaderEntries {
 using ProgramResult = std::pair<std::string, ShaderEntries>;
 
 struct ShaderSetup {
-    ShaderSetup(const ProgramCode& program_code) {
-        program.code = program_code;
+    explicit ShaderSetup(ProgramCode program_code) {
+        program.code = std::move(program_code);
     }
 
     struct {
@@ -135,8 +135,8 @@ struct ShaderSetup {
     }
 
     /// Used in scenarios where we have a dual vertex shaders
-    void SetProgramB(const ProgramCode& program_b) {
-        program.code_b = program_b;
+    void SetProgramB(ProgramCode program_b) {
+        program.code_b = std::move(program_b);
         has_program_b = true;
     }
 
@@ -146,13 +146,18 @@ struct ShaderSetup {
 
 private:
     u64 GetNewHash() const {
+        size_t hash = 0;
+
+        const u64 hash_a = Common::ComputeHash64(program.code.data(), program.code.size());
+        boost::hash_combine(hash, hash_a);
+
         if (has_program_b) {
             // Compute hash over dual shader programs
-            return Common::ComputeHash64(&program, sizeof(program));
-        } else {
-            // Compute hash over a single shader program
-            return Common::ComputeHash64(&program.code, program.code.size());
+            const u64 hash_b = Common::ComputeHash64(program.code_b.data(), program.code_b.size());
+            boost::hash_combine(hash, hash_b);
         }
+
+        return hash;
     }
 
     u64 program_code_hash{};
@@ -191,20 +196,20 @@ ProgramResult GenerateVertexShader(const ShaderSetup& setup, const MaxwellVSConf
  */
 ProgramResult GenerateFragmentShader(const ShaderSetup& setup, const MaxwellFSConfig& config);
 
-} // namespace GLShader
+} // namespace OpenGL::GLShader
 
 namespace std {
 
 template <>
-struct hash<GLShader::MaxwellVSConfig> {
-    size_t operator()(const GLShader::MaxwellVSConfig& k) const {
+struct hash<OpenGL::GLShader::MaxwellVSConfig> {
+    size_t operator()(const OpenGL::GLShader::MaxwellVSConfig& k) const {
         return k.Hash();
     }
 };
 
 template <>
-struct hash<GLShader::MaxwellFSConfig> {
-    size_t operator()(const GLShader::MaxwellFSConfig& k) const {
+struct hash<OpenGL::GLShader::MaxwellFSConfig> {
+    size_t operator()(const OpenGL::GLShader::MaxwellFSConfig& k) const {
         return k.Hash();
     }
 };

src/video_core/renderer_opengl/gl_shader_manager.cpp

@@ -7,7 +7,7 @@
 #include "video_core/engines/maxwell_3d.h"
 #include "video_core/renderer_opengl/gl_shader_manager.h"
 
-namespace GLShader {
+namespace OpenGL::GLShader {
 
 namespace Impl {
 static void SetShaderUniformBlockBinding(GLuint shader, const char* name,
@@ -49,4 +49,4 @@ void MaxwellUniformData::SetFromRegs(const Maxwell3D::State::ShaderStageInfo& sh
     instance_id[0] = state.current_instance;
 }
 
-} // namespace GLShader
+} // namespace OpenGL::GLShader

src/video_core/renderer_opengl/gl_shader_manager.h

@@ -12,7 +12,7 @@
 #include "video_core/renderer_opengl/gl_shader_gen.h"
 #include "video_core/renderer_opengl/maxwell_to_gl.h"
 
-namespace GLShader {
+namespace OpenGL::GLShader {
 
 /// Number of OpenGL texture samplers that can be used in the fragment shader
 static constexpr size_t NumTextureSamplers = 32;
@@ -171,4 +171,4 @@ private:
     OGLPipeline pipeline;
 };
 
-} // namespace GLShader
+} // namespace OpenGL::GLShader

src/video_core/renderer_opengl/gl_shader_util.cpp

@@ -8,7 +8,7 @@
 #include "common/logging/log.h"
 #include "video_core/renderer_opengl/gl_shader_util.h"
 
-namespace GLShader {
+namespace OpenGL::GLShader {
 
 GLuint LoadShader(const char* source, GLenum type) {
     const char* debug_type;
@@ -47,4 +47,4 @@ GLuint LoadShader(const char* source, GLenum type) {
     return shader_id;
 }
 
-} // namespace GLShader
+} // namespace OpenGL::GLShader

src/video_core/renderer_opengl/gl_shader_util.h

@@ -10,7 +10,7 @@
 #include "common/assert.h"
 #include "common/logging/log.h"
 
-namespace GLShader {
+namespace OpenGL::GLShader {
 
 /**
  * Utility function to log the source code of a list of shaders.
@@ -89,4 +89,4 @@ GLuint LoadProgram(bool separable_program, T... shaders) {
     return program_id;
 }
 
-} // namespace GLShader
+} // namespace OpenGL::GLShader

src/video_core/renderer_opengl/gl_state.cpp

@@ -7,6 +7,8 @@
 #include "common/logging/log.h"
 #include "video_core/renderer_opengl/gl_state.h"
 
+namespace OpenGL {
+
 OpenGLState OpenGLState::cur_state;
 
 OpenGLState::OpenGLState() {
@@ -338,3 +340,5 @@ OpenGLState& OpenGLState::ResetFramebuffer(GLuint handle) {
     }
     return *this;
 }
+
+} // namespace OpenGL

src/video_core/renderer_opengl/gl_state.h

@@ -9,6 +9,8 @@
 
 #include "video_core/engines/maxwell_3d.h"
 
+namespace OpenGL {
+
 using Regs = Tegra::Engines::Maxwell3D::Regs;
 
 namespace TextureUnits {
@@ -163,3 +165,5 @@ public:
 private:
     static OpenGLState cur_state;
 };
+
+} // namespace OpenGL

src/video_core/renderer_opengl/gl_stream_buffer.cpp

@@ -9,6 +9,8 @@
 #include "video_core/renderer_opengl/gl_state.h"
 #include "video_core/renderer_opengl/gl_stream_buffer.h"
 
+namespace OpenGL {
+
 OGLStreamBuffer::OGLStreamBuffer(GLenum target, GLsizeiptr size, bool prefer_coherent)
     : gl_target(target), buffer_size(size) {
     gl_buffer.Create();
@@ -97,3 +99,5 @@ void OGLStreamBuffer::Unmap(GLsizeiptr size) {
 
     buffer_pos += size;
 }
+
+} // namespace OpenGL

src/video_core/renderer_opengl/gl_stream_buffer.h

@@ -9,6 +9,8 @@
 #include "common/common_types.h"
 #include "video_core/renderer_opengl/gl_resource_manager.h"
 
+namespace OpenGL {
+
 class OGLStreamBuffer : private NonCopyable {
 public:
     explicit OGLStreamBuffer(GLenum target, GLsizeiptr size, bool prefer_coherent = false);
@@ -42,3 +44,5 @@ private:
     GLsizeiptr mapped_size = 0;
     u8* mapped_ptr = nullptr;
 };
+
+} // namespace OpenGL

src/video_core/renderer_opengl/maxwell_to_gl.h

@@ -10,6 +10,8 @@
 #include "common/logging/log.h"
 #include "video_core/engines/maxwell_3d.h"
 
+namespace OpenGL {
+
 using GLvec2 = std::array<GLfloat, 2>;
 using GLvec3 = std::array<GLfloat, 3>;
 using GLvec4 = std::array<GLfloat, 4>;
@@ -360,3 +362,4 @@ inline GLenum LogicOp(Maxwell::LogicOperation operation) {
 }
 
 } // namespace MaxwellToGL
+} // namespace OpenGL

src/video_core/renderer_opengl/renderer_opengl.cpp

@@ -20,6 +20,8 @@
 #include "video_core/renderer_opengl/renderer_opengl.h"
 #include "video_core/utils.h"
 
+namespace OpenGL {
+
 static const char vertex_shader[] = R"(
 #version 150 core
 
@@ -476,3 +478,5 @@ bool RendererOpenGL::Init() {
 
 /// Shutdown the renderer
 void RendererOpenGL::ShutDown() {}
+
+} // namespace OpenGL

src/video_core/renderer_opengl/renderer_opengl.h

@@ -16,6 +16,8 @@ namespace Core::Frontend {
 class EmuWindow;
 }
 
+namespace OpenGL {
+
 /// Structure used for storing information about the textures for the Switch screen
 struct TextureInfo {
     OGLTexture resource;
@@ -98,3 +100,5 @@ private:
     Tegra::FramebufferConfig::TransformFlags framebuffer_transform_flags;
     MathUtil::Rectangle<int> framebuffer_crop_rect;
 };
+
+} // namespace OpenGL

src/video_core/video_core.cpp

@@ -10,7 +10,7 @@
 namespace VideoCore {
 
 std::unique_ptr<RendererBase> CreateRenderer(Core::Frontend::EmuWindow& emu_window) {
-    return std::make_unique<RendererOpenGL>(emu_window);
+    return std::make_unique<OpenGL::RendererOpenGL>(emu_window);
 }
 
 } // namespace VideoCore

src/yuzu/configuration/configure_input.ui

@@ -29,24 +29,6 @@
        </property>
        <layout class="QGridLayout" name="gridLayout_6">
         <item row="0" column="0">
-         <layout class="QVBoxLayout" name="buttonMiscPlusVerticalLayout">
-          <item>
-           <widget class="QLabel" name="labelPlus">
-            <property name="text">
-             <string>Plus:</string>
-            </property>
-           </widget>
-          </item>
-          <item>
-           <widget class="QPushButton" name="buttonPlus">
-            <property name="text">
-             <string/>
-            </property>
-           </widget>
-          </item>
-         </layout>
-        </item>
-        <item row="0" column="1">
          <layout class="QVBoxLayout" name="buttonMiscMinusVerticalLayout">
           <item>
            <widget class="QLabel" name="labelMinus">
@@ -64,6 +46,24 @@
           </item>
          </layout>
         </item>
+        <item row="0" column="1">
+         <layout class="QVBoxLayout" name="buttonMiscPlusVerticalLayout">
+          <item>
+           <widget class="QLabel" name="labelPlus">
+            <property name="text">
+             <string>Plus:</string>
+            </property>
+           </widget>
+          </item>
+          <item>
+           <widget class="QPushButton" name="buttonPlus">
+            <property name="text">
+             <string/>
+            </property>
+           </widget>
+          </item>
+         </layout>
+        </item>
         <item row="1" column="0">
          <layout class="QVBoxLayout" name="buttonMiscHomeVerticalLayout">
           <item>