nvflinger: Call MicroProfileFlip on NVFlinger::Compose.

Misc gpu improvements

src/common/bit_field.h

@@ -192,11 +192,6 @@ private:
     static_assert(position < 8 * sizeof(T), "Invalid position");
     static_assert(bits <= 8 * sizeof(T), "Invalid number of bits");
     static_assert(bits > 0, "Invalid number of bits");
-    static_assert(std::is_pod<T>::value, "Invalid base type");
+    static_assert(std::is_trivially_copyable_v<T>, "T must be trivially copyable in a BitField");
 };
 #pragma pack()
-
-#if (__GNUC__ >= 5) || defined(__clang__) || defined(_MSC_VER)
-static_assert(std::is_trivially_copyable<BitField<0, 1, unsigned>>::value,
-              "BitField must be trivially copyable");
-#endif

src/core/CMakeLists.txt

@@ -12,6 +12,8 @@ add_library(core STATIC
     file_sys/errors.h
     file_sys/filesystem.cpp
     file_sys/filesystem.h
+    file_sys/partition_filesystem.cpp
+    file_sys/partition_filesystem.h
     file_sys/path_parser.cpp
     file_sys/path_parser.h
     file_sys/program_metadata.cpp

src/core/file_sys/partition_filesystem.cpp

@@ -0,0 +1,125 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <cinttypes>
+#include <utility>
+#include "common/file_util.h"
+#include "common/logging/log.h"
+#include "core/file_sys/partition_filesystem.h"
+#include "core/loader/loader.h"
+
+namespace FileSys {
+
+Loader::ResultStatus PartitionFilesystem::Load(const std::string& file_path, size_t offset) {
+    FileUtil::IOFile file(file_path, "rb");
+    if (!file.IsOpen())
+        return Loader::ResultStatus::Error;
+
+    // At least be as large as the header
+    if (file.GetSize() < sizeof(Header))
+        return Loader::ResultStatus::Error;
+
+    // For cartridges, HFSs can get very large, so we need to calculate the size up to
+    // the actual content itself instead of just blindly reading in the entire file.
+    Header pfs_header;
+    if (!file.ReadBytes(&pfs_header, sizeof(Header)))
+        return Loader::ResultStatus::Error;
+
+    bool is_hfs = (memcmp(pfs_header.magic.data(), "HFS", 3) == 0);
+    size_t entry_size = is_hfs ? sizeof(HFSEntry) : sizeof(PFSEntry);
+    size_t metadata_size =
+        sizeof(Header) + (pfs_header.num_entries * entry_size) + pfs_header.strtab_size;
+
+    // Actually read in now...
+    file.Seek(offset, SEEK_SET);
+    std::vector<u8> file_data(metadata_size);
+
+    if (!file.ReadBytes(file_data.data(), metadata_size))
+        return Loader::ResultStatus::Error;
+
+    Loader::ResultStatus result = Load(file_data);
+    if (result != Loader::ResultStatus::Success)
+        LOG_ERROR(Service_FS, "Failed to load PFS from file %s!", file_path.c_str());
+
+    return result;
+}
+
+Loader::ResultStatus PartitionFilesystem::Load(const std::vector<u8>& file_data, size_t offset) {
+    size_t total_size = file_data.size() - offset;
+    if (total_size < sizeof(Header))
+        return Loader::ResultStatus::Error;
+
+    memcpy(&pfs_header, &file_data[offset], sizeof(Header));
+    is_hfs = (memcmp(pfs_header.magic.data(), "HFS", 3) == 0);
+
+    size_t entries_offset = offset + sizeof(Header);
+    size_t entry_size = is_hfs ? sizeof(HFSEntry) : sizeof(PFSEntry);
+    size_t strtab_offset = entries_offset + (pfs_header.num_entries * entry_size);
+    for (u16 i = 0; i < pfs_header.num_entries; i++) {
+        FileEntry entry;
+
+        memcpy(&entry.fs_entry, &file_data[entries_offset + (i * entry_size)], sizeof(FSEntry));
+        entry.name = std::string(reinterpret_cast<const char*>(
+            &file_data[strtab_offset + entry.fs_entry.strtab_offset]));
+        pfs_entries.push_back(std::move(entry));
+    }
+
+    content_offset = strtab_offset + pfs_header.strtab_size;
+
+    return Loader::ResultStatus::Success;
+}
+
+u32 PartitionFilesystem::GetNumEntries() const {
+    return pfs_header.num_entries;
+}
+
+u64 PartitionFilesystem::GetEntryOffset(int index) const {
+    if (index > GetNumEntries())
+        return 0;
+
+    return content_offset + pfs_entries[index].fs_entry.offset;
+}
+
+u64 PartitionFilesystem::GetEntrySize(int index) const {
+    if (index > GetNumEntries())
+        return 0;
+
+    return pfs_entries[index].fs_entry.size;
+}
+
+std::string PartitionFilesystem::GetEntryName(int index) const {
+    if (index > GetNumEntries())
+        return "";
+
+    return pfs_entries[index].name;
+}
+
+u64 PartitionFilesystem::GetFileOffset(const std::string& name) const {
+    for (u32 i = 0; i < pfs_header.num_entries; i++) {
+        if (pfs_entries[i].name == name)
+            return content_offset + pfs_entries[i].fs_entry.offset;
+    }
+
+    return 0;
+}
+
+u64 PartitionFilesystem::GetFileSize(const std::string& name) const {
+    for (u32 i = 0; i < pfs_header.num_entries; i++) {
+        if (pfs_entries[i].name == name)
+            return pfs_entries[i].fs_entry.size;
+    }
+
+    return 0;
+}
+
+void PartitionFilesystem::Print() const {
+    NGLOG_DEBUG(Service_FS, "Magic:                  {:.4}", pfs_header.magic.data());
+    NGLOG_DEBUG(Service_FS, "Files:                  {}", pfs_header.num_entries);
+    for (u32 i = 0; i < pfs_header.num_entries; i++) {
+        NGLOG_DEBUG(Service_FS, " > File {}:              {} (0x{:X} bytes, at 0x{:X})", i,
+                    pfs_entries[i].name.c_str(), pfs_entries[i].fs_entry.size,
+                    GetFileOffset(pfs_entries[i].name));
+    }
+}
+} // namespace FileSys

src/core/file_sys/partition_filesystem.h

@@ -0,0 +1,87 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <array>
+#include <string>
+#include <vector>
+#include "common/common_funcs.h"
+#include "common/common_types.h"
+#include "common/swap.h"
+
+namespace Loader {
+enum class ResultStatus;
+}
+
+namespace FileSys {
+
+/**
+ * Helper which implements an interface to parse PFS/HFS filesystems.
+ * Data can either be loaded from a file path or data with an offset into it.
+ */
+class PartitionFilesystem {
+public:
+    Loader::ResultStatus Load(const std::string& file_path, size_t offset = 0);
+    Loader::ResultStatus Load(const std::vector<u8>& file_data, size_t offset = 0);
+
+    u32 GetNumEntries() const;
+    u64 GetEntryOffset(int index) const;
+    u64 GetEntrySize(int index) const;
+    std::string GetEntryName(int index) const;
+    u64 GetFileOffset(const std::string& name) const;
+    u64 GetFileSize(const std::string& name) const;
+
+    void Print() const;
+
+private:
+    struct Header {
+        std::array<char, 4> magic;
+        u32_le num_entries;
+        u32_le strtab_size;
+        INSERT_PADDING_BYTES(0x4);
+    };
+
+    static_assert(sizeof(Header) == 0x10, "PFS/HFS header structure size is wrong");
+
+#pragma pack(push, 1)
+    struct FSEntry {
+        u64_le offset;
+        u64_le size;
+        u32_le strtab_offset;
+    };
+
+    static_assert(sizeof(FSEntry) == 0x14, "FS entry structure size is wrong");
+
+    struct PFSEntry {
+        FSEntry fs_entry;
+        INSERT_PADDING_BYTES(0x4);
+    };
+
+    static_assert(sizeof(PFSEntry) == 0x18, "PFS entry structure size is wrong");
+
+    struct HFSEntry {
+        FSEntry fs_entry;
+        u32_le hash_region_size;
+        INSERT_PADDING_BYTES(0x8);
+        std::array<char, 0x20> hash;
+    };
+
+    static_assert(sizeof(HFSEntry) == 0x40, "HFS entry structure size is wrong");
+
+#pragma pack(pop)
+
+    struct FileEntry {
+        FSEntry fs_entry;
+        std::string name;
+    };
+
+    Header pfs_header;
+    bool is_hfs;
+    size_t content_offset;
+
+    std::vector<FileEntry> pfs_entries;
+};
+
+} // namespace FileSys

src/core/hle/service/nvflinger/nvflinger.cpp

@@ -5,6 +5,7 @@
 #include <algorithm>
 
 #include "common/alignment.h"
+#include "common/microprofile.h"
 #include "common/scope_exit.h"
 #include "core/core.h"
 #include "core/core_timing.h"
@@ -128,6 +129,8 @@ void NVFlinger::Compose() {
         // Search for a queued buffer and acquire it
         auto buffer = buffer_queue->AcquireBuffer();
 
+        MicroProfileFlip();
+
         if (buffer == boost::none) {
             // There was no queued buffer to draw, render previous frame
             Core::System::GetInstance().perf_stats.EndGameFrame();

src/core/hle/service/vi/vi.cpp

@@ -587,7 +587,7 @@ public:
             {2203, nullptr, "SetLayerSize"},
             {2204, nullptr, "GetLayerZ"},
             {2205, &ISystemDisplayService::SetLayerZ, "SetLayerZ"},
-            {2207, &ISystemDisplayService::SetLayerVisibility, "SetLayerVisibility"},
+            {2207, nullptr, "SetLayerVisibility"},
             {2209, nullptr, "SetLayerAlpha"},
             {2312, nullptr, "CreateStrayLayer"},
             {2400, nullptr, "OpenIndirectLayer"},
@@ -637,16 +637,6 @@ private:
         IPC::ResponseBuilder rb = rp.MakeBuilder(2, 0, 0);
         rb.Push(RESULT_SUCCESS);
     }
-
-    void SetLayerVisibility(Kernel::HLERequestContext& ctx) {
-        IPC::RequestParser rp{ctx};
-        u64 layer_id = rp.Pop<u64>();
-        bool visibility = rp.Pop<bool>();
-        IPC::ResponseBuilder rb = rp.MakeBuilder(2, 0, 0);
-        rb.Push(RESULT_SUCCESS);
-        LOG_WARNING(Service_VI, "(STUBBED) called, layer_id=0x%x, visibility=%u", layer_id,
-                    visibility);
-    }
 };
 
 class IManagerDisplayService final : public ServiceFramework<IManagerDisplayService> {
@@ -678,7 +668,7 @@ public:
             {4206, nullptr, "SetDefaultDisplay"},
             {6000, &IManagerDisplayService::AddToLayerStack, "AddToLayerStack"},
             {6001, nullptr, "RemoveFromLayerStack"},
-            {6002, &IManagerDisplayService::SetLayerVisibility, "SetLayerVisibility"},
+            {6002, nullptr, "SetLayerVisibility"},
             {6003, nullptr, "SetLayerConfig"},
             {6004, nullptr, "AttachLayerPresentationTracer"},
             {6005, nullptr, "DetachLayerPresentationTracer"},
@@ -760,16 +750,6 @@ private:
         rb.Push(RESULT_SUCCESS);
     }
 
-    void SetLayerVisibility(Kernel::HLERequestContext& ctx) {
-        IPC::RequestParser rp{ctx};
-        u64 layer_id = rp.Pop<u64>();
-        bool visibility = rp.Pop<bool>();
-        IPC::ResponseBuilder rb = rp.MakeBuilder(2, 0, 0);
-        rb.Push(RESULT_SUCCESS);
-        LOG_WARNING(Service_VI, "(STUBBED) called, layer_id=0x%x, visibility=%u", layer_id,
-                    visibility);
-    }
-
     std::shared_ptr<NVFlinger::NVFlinger> nv_flinger;
 };
 
@@ -840,15 +820,15 @@ private:
         IPC::RequestParser rp{ctx};
         u64 display_id = rp.Pop<u64>();
 
-        IPC::ResponseBuilder rb = rp.MakeBuilder(6, 0, 0);
+        IPC::ResponseBuilder rb = rp.MakeBuilder(4, 0, 0);
         rb.Push(RESULT_SUCCESS);
 
         if (Settings::values.use_docked_mode) {
-            rb.Push(static_cast<u64>(DisplayResolution::DockedWidth));
-            rb.Push(static_cast<u64>(DisplayResolution::DockedHeight));
+            rb.Push(static_cast<u32>(DisplayResolution::DockedWidth));
+            rb.Push(static_cast<u32>(DisplayResolution::DockedHeight));
         } else {
-            rb.Push(static_cast<u64>(DisplayResolution::UndockedWidth));
-            rb.Push(static_cast<u64>(DisplayResolution::UndockedHeight));
+            rb.Push(static_cast<u32>(DisplayResolution::UndockedWidth));
+            rb.Push(static_cast<u32>(DisplayResolution::UndockedHeight));
         }
     }
 

src/video_core/engines/maxwell_3d.cpp

@@ -74,8 +74,6 @@ void Maxwell3D::WriteReg(u32 method, u32 value, u32 remaining_params) {
 
     regs.reg_array[method] = value;
 
-#define MAXWELL3D_REG_INDEX(field_name) (offsetof(Regs, field_name) / sizeof(u32))
-
     switch (method) {
     case MAXWELL3D_REG_INDEX(code_address.code_address_high):
     case MAXWELL3D_REG_INDEX(code_address.code_address_low): {
@@ -136,7 +134,7 @@ void Maxwell3D::WriteReg(u32 method, u32 value, u32 remaining_params) {
         break;
     }
 
-#undef MAXWELL3D_REG_INDEX
+    VideoCore::g_renderer->Rasterizer()->NotifyMaxwellRegisterChanged(method);
 
     if (debug_context) {
         debug_context->OnEvent(Tegra::DebugContext::Event::MaxwellCommandProcessed, nullptr);
@@ -223,7 +221,8 @@ Texture::TICEntry Maxwell3D::GetTICEntry(u32 tic_index) const {
     ASSERT_MSG(tic_entry.header_version == Texture::TICHeaderVersion::BlockLinear,
                "TIC versions other than BlockLinear are unimplemented");
 
-    ASSERT_MSG(tic_entry.texture_type == Texture::TextureType::Texture2D,
+    ASSERT_MSG((tic_entry.texture_type == Texture::TextureType::Texture2D) ||
+                   (tic_entry.texture_type == Texture::TextureType::Texture2DNoMipmap),
                "Texture types other than Texture2D are unimplemented");
 
     auto r_type = tic_entry.r_type.Value();

src/video_core/engines/maxwell_3d.h

@@ -20,6 +20,9 @@
 namespace Tegra {
 namespace Engines {
 
+#define MAXWELL3D_REG_INDEX(field_name)                                                            \
+    (offsetof(Tegra::Engines::Maxwell3D::Regs, field_name) / sizeof(u32))
+
 class Maxwell3D final {
 public:
     explicit Maxwell3D(MemoryManager& memory_manager);
@@ -254,6 +257,46 @@ public:
             UnsignedInt = 0x2,
         };
 
+        struct Blend {
+            enum class Equation : u32 {
+                Add = 1,
+                Subtract = 2,
+                ReverseSubtract = 3,
+                Min = 4,
+                Max = 5,
+            };
+
+            enum class Factor : u32 {
+                Zero = 0x1,
+                One = 0x2,
+                SourceColor = 0x3,
+                OneMinusSourceColor = 0x4,
+                SourceAlpha = 0x5,
+                OneMinusSourceAlpha = 0x6,
+                DestAlpha = 0x7,
+                OneMinusDestAlpha = 0x8,
+                DestColor = 0x9,
+                OneMinusDestColor = 0xa,
+                SourceAlphaSaturate = 0xb,
+                Source1Color = 0x10,
+                OneMinusSource1Color = 0x11,
+                Source1Alpha = 0x12,
+                OneMinusSource1Alpha = 0x13,
+                ConstantColor = 0x61,
+                OneMinusConstantColor = 0x62,
+                ConstantAlpha = 0x63,
+                OneMinusConstantAlpha = 0x64,
+            };
+
+            u32 separate_alpha;
+            Equation equation_rgb;
+            Factor factor_source_rgb;
+            Factor factor_dest_rgb;
+            Equation equation_a;
+            Factor factor_source_a;
+            Factor factor_dest_a;
+        };
+
         union {
             struct {
                 INSERT_PADDING_WORDS(0x200);
@@ -451,7 +494,9 @@ public:
                     }
                 } vertex_array[NumVertexArrays];
 
-                INSERT_PADDING_WORDS(0x40);
+                Blend blend;
+
+                INSERT_PADDING_WORDS(0x39);
 
                 struct {
                     u32 limit_high;
@@ -616,6 +661,7 @@ ASSERT_REG_POSITION(draw, 0x585);
 ASSERT_REG_POSITION(index_array, 0x5F2);
 ASSERT_REG_POSITION(query, 0x6C0);
 ASSERT_REG_POSITION(vertex_array[0], 0x700);
+ASSERT_REG_POSITION(blend, 0x780);
 ASSERT_REG_POSITION(vertex_array_limit[0], 0x7C0);
 ASSERT_REG_POSITION(shader_config[0], 0x800);
 ASSERT_REG_POSITION(const_buffer, 0x8E0);

src/video_core/engines/shader_bytecode.h

@@ -4,6 +4,7 @@
 
 #pragma once
 
+#include <cstring>
 #include <map>
 #include <string>
 #include "common/bit_field.h"
@@ -12,14 +13,10 @@ namespace Tegra {
 namespace Shader {
 
 struct Register {
-    Register() = default;
+    constexpr Register() = default;
 
     constexpr Register(u64 value) : value(value) {}
 
-    constexpr u64 GetIndex() const {
-        return value;
-    }
-
     constexpr operator u64() const {
         return value;
     }
@@ -43,13 +40,13 @@ struct Register {
     }
 
 private:
-    u64 value;
+    u64 value{};
 };
 
 union Attribute {
     Attribute() = default;
 
-    constexpr Attribute(u64 value) : value(value) {}
+    constexpr explicit Attribute(u64 value) : value(value) {}
 
     enum class Index : u64 {
         Position = 7,
@@ -68,7 +65,20 @@ union Attribute {
     } fmt28;
 
     BitField<39, 8, u64> reg;
-    u64 value;
+    u64 value{};
+};
+
+union Sampler {
+    Sampler() = default;
+
+    constexpr explicit Sampler(u64 value) : value(value) {}
+
+    enum class Index : u64 {
+        Sampler_0 = 8,
+    };
+
+    BitField<36, 13, Index> index;
+    u64 value{};
 };
 
 union Uniform {
@@ -238,7 +248,7 @@ union OpCode {
     BitField<55, 9, Id> op3;
     BitField<52, 12, Id> op4;
     BitField<51, 13, Id> op5;
-    u64 value;
+    u64 value{};
 };
 static_assert(sizeof(OpCode) == 0x8, "Incorrect structure size");
 
@@ -280,6 +290,7 @@ enum class SubOp : u64 {
     Lg2 = 0x3,
     Rcp = 0x4,
     Rsq = 0x5,
+    Min = 0x8,
 };
 
 union Instruction {
@@ -295,15 +306,25 @@ union Instruction {
     BitField<20, 8, Register> gpr20;
     BitField<20, 7, SubOp> sub_op;
     BitField<28, 8, Register> gpr28;
-    BitField<36, 13, u64> imm36;
     BitField<39, 8, Register> gpr39;
 
     union {
+        BitField<20, 19, u64> imm20;
         BitField<45, 1, u64> negate_b;
         BitField<46, 1, u64> abs_a;
         BitField<48, 1, u64> negate_a;
         BitField<49, 1, u64> abs_b;
         BitField<50, 1, u64> abs_d;
+        BitField<56, 1, u64> negate_imm;
+
+        float GetImm20() const {
+            float result{};
+            u32 imm{static_cast<u32>(imm20)};
+            imm <<= 12;
+            imm |= negate_imm ? 0x80000000 : 0;
+            std::memcpy(&result, &imm, sizeof(imm));
+            return result;
+        }
     } alu;
 
     union {
@@ -311,11 +332,13 @@ union Instruction {
         BitField<49, 1, u64> negate_c;
     } ffma;
 
+    BitField<61, 1, u64> is_b_imm;
     BitField<60, 1, u64> is_b_gpr;
     BitField<59, 1, u64> is_c_gpr;
 
     Attribute attribute;
     Uniform uniform;
+    Sampler sampler;
 
     u64 hex;
 };

src/video_core/gpu.h

@@ -15,7 +15,10 @@ namespace Tegra {
 
 enum class RenderTargetFormat : u32 {
     NONE = 0x0,
+    RGBA16_FLOAT = 0xCA,
+    RGB10_A2_UNORM = 0xD1,
     RGBA8_UNORM = 0xD5,
+    RGBA8_SRGB = 0xD6,
 };
 
 class DebugContext;

src/video_core/rasterizer_interface.h

@@ -19,7 +19,7 @@ public:
     virtual void DrawArrays() = 0;
 
     /// Notify rasterizer that the specified Maxwell register has been changed
-    virtual void NotifyMaxwellRegisterChanged(u32 id) = 0;
+    virtual void NotifyMaxwellRegisterChanged(u32 method) = 0;
 
     /// Notify rasterizer that all caches should be flushed to Switch memory
     virtual void FlushAll() = 0;

src/video_core/renderer_opengl/gl_rasterizer.cpp

@@ -446,7 +446,32 @@ void RasterizerOpenGL::BindTextures() {
     }
 }
 
-void RasterizerOpenGL::NotifyMaxwellRegisterChanged(u32 id) {}
+void RasterizerOpenGL::NotifyMaxwellRegisterChanged(u32 method) {
+    const auto& regs = Core::System().GetInstance().GPU().Maxwell3D().regs;
+    switch (method) {
+    case MAXWELL3D_REG_INDEX(blend.separate_alpha):
+        ASSERT_MSG(false, "unimplemented");
+        break;
+    case MAXWELL3D_REG_INDEX(blend.equation_rgb):
+        state.blend.rgb_equation = MaxwellToGL::BlendEquation(regs.blend.equation_rgb);
+        break;
+    case MAXWELL3D_REG_INDEX(blend.factor_source_rgb):
+        state.blend.src_rgb_func = MaxwellToGL::BlendFunc(regs.blend.factor_source_rgb);
+        break;
+    case MAXWELL3D_REG_INDEX(blend.factor_dest_rgb):
+        state.blend.dst_rgb_func = MaxwellToGL::BlendFunc(regs.blend.factor_dest_rgb);
+        break;
+    case MAXWELL3D_REG_INDEX(blend.equation_a):
+        state.blend.a_equation = MaxwellToGL::BlendEquation(regs.blend.equation_a);
+        break;
+    case MAXWELL3D_REG_INDEX(blend.factor_source_a):
+        state.blend.src_a_func = MaxwellToGL::BlendFunc(regs.blend.factor_source_a);
+        break;
+    case MAXWELL3D_REG_INDEX(blend.factor_dest_a):
+        state.blend.dst_a_func = MaxwellToGL::BlendFunc(regs.blend.factor_dest_a);
+        break;
+    }
+}
 
 void RasterizerOpenGL::FlushAll() {
     MICROPROFILE_SCOPE(OpenGL_CacheManagement);

src/video_core/renderer_opengl/gl_rasterizer.h

@@ -32,7 +32,7 @@ public:
     ~RasterizerOpenGL() override;
 
     void DrawArrays() override;
-    void NotifyMaxwellRegisterChanged(u32 id) override;
+    void NotifyMaxwellRegisterChanged(u32 method) override;
     void FlushAll() override;
     void FlushRegion(VAddr addr, u64 size) override;
     void InvalidateRegion(VAddr addr, u64 size) override;

src/video_core/renderer_opengl/gl_rasterizer_cache.h

@@ -87,6 +87,7 @@ struct SurfaceParams {
         case Tegra::RenderTargetFormat::RGBA8_UNORM:
             return PixelFormat::RGBA8;
         default:
+            NGLOG_CRITICAL(HW_GPU, "Unimplemented format={}", static_cast<u32>(format));
             UNREACHABLE();
         }
     }
@@ -96,6 +97,7 @@ struct SurfaceParams {
         case Tegra::FramebufferConfig::PixelFormat::ABGR8:
             return PixelFormat::RGBA8;
         default:
+            NGLOG_CRITICAL(HW_GPU, "Unimplemented format={}", static_cast<u32>(format));
             UNREACHABLE();
         }
     }
@@ -108,6 +110,7 @@ struct SurfaceParams {
         case Tegra::Texture::TextureFormat::DXT1:
             return PixelFormat::DXT1;
         default:
+            NGLOG_CRITICAL(HW_GPU, "Unimplemented format={}", static_cast<u32>(format));
             UNREACHABLE();
         }
     }

src/video_core/renderer_opengl/gl_shader_decompiler.cpp

@@ -17,6 +17,7 @@ using Tegra::Shader::Attribute;
 using Tegra::Shader::Instruction;
 using Tegra::Shader::OpCode;
 using Tegra::Shader::Register;
+using Tegra::Shader::Sampler;
 using Tegra::Shader::SubOp;
 using Tegra::Shader::Uniform;
 
@@ -155,23 +156,27 @@ private:
 
     /// Generates code representing an input attribute register.
     std::string GetInputAttribute(Attribute::Index attribute) {
-        declr_input_attribute.insert(attribute);
+        switch (attribute) {
+        case Attribute::Index::Position:
+            return "position";
+        default:
+            const u32 index{static_cast<u32>(attribute) -
+                            static_cast<u32>(Attribute::Index::Attribute_0)};
+            if (attribute >= Attribute::Index::Attribute_0) {
+                declr_input_attribute.insert(attribute);
+                return "input_attribute_" + std::to_string(index);
+            }
 
-        const u32 index{static_cast<u32>(attribute) -
-                        static_cast<u32>(Attribute::Index::Attribute_0)};
-        if (attribute >= Attribute::Index::Attribute_0) {
-            return "input_attribute_" + std::to_string(index);
+            NGLOG_CRITICAL(HW_GPU, "Unhandled input attribute: {}", index);
+            UNREACHABLE();
         }
-
-        LOG_CRITICAL(HW_GPU, "Unhandled input attribute: 0x%02x", index);
-        UNREACHABLE();
     }
 
     /// Generates code representing an output attribute register.
     std::string GetOutputAttribute(Attribute::Index attribute) {
         switch (attribute) {
         case Attribute::Index::Position:
-            return "gl_Position";
+            return "position";
         default:
             const u32 index{static_cast<u32>(attribute) -
                             static_cast<u32>(Attribute::Index::Attribute_0)};
@@ -180,22 +185,42 @@ private:
                 return "output_attribute_" + std::to_string(index);
             }
 
-            LOG_CRITICAL(HW_GPU, "Unhandled output attribute: 0x%02x", index);
+            NGLOG_CRITICAL(HW_GPU, "Unhandled output attribute: {}", index);
             UNREACHABLE();
         }
     }
 
+    /// Generates code representing an immediate value
+    static std::string GetImmediate(const Instruction& instr) {
+        return std::to_string(instr.alu.GetImm20());
+    }
+
     /// Generates code representing a temporary (GPR) register.
-    std::string GetRegister(const Register& reg) {
-        return *declr_register.insert("register_" + std::to_string(reg)).first;
+    std::string GetRegister(const Register& reg, unsigned elem = 0) {
+        if (stage == Maxwell3D::Regs::ShaderStage::Fragment && reg < 4) {
+            // GPRs 0-3 are output color for the fragment shader
+            return std::string{"color."} + "rgba"[(reg + elem) & 3];
+        }
+
+        return *declr_register.insert("register_" + std::to_string(reg + elem)).first;
     }
 
     /// Generates code representing a uniform (C buffer) register.
     std::string GetUniform(const Uniform& reg) {
-        declr_const_buffers[reg.index].MarkAsUsed(reg.index, reg.offset, stage);
+        declr_const_buffers[reg.index].MarkAsUsed(static_cast<unsigned>(reg.index),
+                                                  static_cast<unsigned>(reg.offset), stage);
         return 'c' + std::to_string(reg.index) + '[' + std::to_string(reg.offset) + ']';
     }
 
+    /// Generates code representing a texture sampler.
+    std::string GetSampler(const Sampler& sampler) const {
+        // TODO(Subv): Support more than just texture sampler 0
+        ASSERT_MSG(sampler.index == Sampler::Index::Sampler_0, "unsupported");
+        const unsigned index{static_cast<unsigned>(sampler.index.Value()) -
+                             static_cast<unsigned>(Sampler::Index::Sampler_0)};
+        return "tex[" + std::to_string(index) + "]";
+    }
+
     /**
      * Adds code that calls a subroutine.
      * @param subroutine the subroutine to call.
@@ -217,12 +242,13 @@ private:
      * @param value the code representing the value to assign.
      */
     void SetDest(u64 elem, const std::string& reg, const std::string& value,
-                 u64 dest_num_components, u64 value_num_components) {
+                 u64 dest_num_components, u64 value_num_components, bool is_abs = false) {
         std::string swizzle = ".";
         swizzle += "xyzw"[elem];
 
         std::string dest = reg + (dest_num_components != 1 ? swizzle : "");
         std::string src = "(" + value + ")" + (value_num_components != 1 ? swizzle : "");
+        src = is_abs ? "abs(" + src + ")" : src;
 
         shader.AddLine(dest + " = " + src + ";");
     }
@@ -240,8 +266,6 @@ private:
 
         switch (OpCode::GetInfo(instr.opcode).type) {
         case OpCode::Type::Arithmetic: {
-            ASSERT(!instr.alu.abs_d);
-
             std::string dest = GetRegister(instr.gpr0);
             std::string op_a = instr.alu.negate_a ? "-" : "";
             op_a += GetRegister(instr.gpr8);
@@ -250,63 +274,109 @@ private:
             }
 
             std::string op_b = instr.alu.negate_b ? "-" : "";
-            if (instr.is_b_gpr) {
-                op_b += GetRegister(instr.gpr20);
+
+            if (instr.is_b_imm) {
+                op_b += GetImmediate(instr);
             } else {
-                op_b += GetUniform(instr.uniform);
+                if (instr.is_b_gpr) {
+                    op_b += GetRegister(instr.gpr20);
+                } else {
+                    op_b += GetUniform(instr.uniform);
+                }
             }
+
             if (instr.alu.abs_b) {
                 op_b = "abs(" + op_b + ")";
             }
 
             switch (instr.opcode.EffectiveOpCode()) {
             case OpCode::Id::FMUL_C:
-            case OpCode::Id::FMUL_R: {
-                SetDest(0, dest, op_a + " * " + op_b, 1, 1);
+            case OpCode::Id::FMUL_R:
+            case OpCode::Id::FMUL_IMM: {
+                SetDest(0, dest, op_a + " * " + op_b, 1, 1, instr.alu.abs_d);
                 break;
             }
             case OpCode::Id::FADD_C:
-            case OpCode::Id::FADD_R: {
-                SetDest(0, dest, op_a + " + " + op_b, 1, 1);
+            case OpCode::Id::FADD_R:
+            case OpCode::Id::FADD_IMM: {
+                SetDest(0, dest, op_a + " + " + op_b, 1, 1, instr.alu.abs_d);
+                break;
+            }
+            case OpCode::Id::MUFU: {
+                switch (instr.sub_op) {
+                case SubOp::Cos:
+                    SetDest(0, dest, "cos(" + op_a + ")", 1, 1, instr.alu.abs_d);
+                    break;
+                case SubOp::Sin:
+                    SetDest(0, dest, "sin(" + op_a + ")", 1, 1, instr.alu.abs_d);
+                    break;
+                case SubOp::Ex2:
+                    SetDest(0, dest, "exp2(" + op_a + ")", 1, 1, instr.alu.abs_d);
+                    break;
+                case SubOp::Lg2:
+                    SetDest(0, dest, "log2(" + op_a + ")", 1, 1, instr.alu.abs_d);
+                    break;
+                case SubOp::Rcp:
+                    SetDest(0, dest, "1.0 / " + op_a, 1, 1, instr.alu.abs_d);
+                    break;
+                case SubOp::Rsq:
+                    SetDest(0, dest, "inversesqrt(" + op_a + ")", 1, 1, instr.alu.abs_d);
+                    break;
+                case SubOp::Min:
+                    SetDest(0, dest, "min(" + op_a + "," + op_b + ")", 1, 1, instr.alu.abs_d);
+                    break;
+                default:
+                    NGLOG_CRITICAL(HW_GPU, "Unhandled MUFU sub op: {}",
+                                   static_cast<unsigned>(instr.sub_op.Value()));
+                    UNREACHABLE();
+                }
                 break;
             }
             default: {
-                LOG_CRITICAL(HW_GPU, "Unhandled arithmetic instruction: 0x%02x (%s): 0x%08x",
-                             static_cast<unsigned>(instr.opcode.EffectiveOpCode()),
-                             OpCode::GetInfo(instr.opcode).name.c_str(), instr.hex);
-                throw DecompileFail("Unhandled instruction");
-                break;
+                NGLOG_CRITICAL(HW_GPU, "Unhandled arithmetic instruction: {} ({}): {}",
+                               static_cast<unsigned>(instr.opcode.EffectiveOpCode()),
+                               OpCode::GetInfo(instr.opcode).name, instr.hex);
+                UNREACHABLE();
             }
             }
             break;
         }
         case OpCode::Type::Ffma: {
-            ASSERT_MSG(!instr.ffma.negate_b, "untested");
-            ASSERT_MSG(!instr.ffma.negate_c, "untested");
-
             std::string dest = GetRegister(instr.gpr0);
             std::string op_a = GetRegister(instr.gpr8);
-
             std::string op_b = instr.ffma.negate_b ? "-" : "";
-            op_b += GetUniform(instr.uniform);
-
             std::string op_c = instr.ffma.negate_c ? "-" : "";
-            op_c += GetRegister(instr.gpr39);
 
             switch (instr.opcode.EffectiveOpCode()) {
             case OpCode::Id::FFMA_CR: {
-                SetDest(0, dest, op_a + " * " + op_b + " + " + op_c, 1, 1);
+                op_b += GetUniform(instr.uniform);
+                op_c += GetRegister(instr.gpr39);
                 break;
             }
+            case OpCode::Id::FFMA_RR: {
+                op_b += GetRegister(instr.gpr20);
+                op_c += GetRegister(instr.gpr39);
+                break;
+            }
+            case OpCode::Id::FFMA_RC: {
+                op_b += GetRegister(instr.gpr39);
+                op_c += GetUniform(instr.uniform);
+                break;
+            }
+            case OpCode::Id::FFMA_IMM: {
+                op_b += GetImmediate(instr);
+                op_c += GetRegister(instr.gpr39);
+                break;
+            }
+            default: {
+                NGLOG_CRITICAL(HW_GPU, "Unhandled FFMA instruction: {} ({}): {}",
+                               static_cast<unsigned>(instr.opcode.EffectiveOpCode()),
+                               OpCode::GetInfo(instr.opcode).name, instr.hex);
+                UNREACHABLE();
+            }
+            }
 
-            default: {
-                LOG_CRITICAL(HW_GPU, "Unhandled arithmetic FFMA instruction: 0x%02x (%s): 0x%08x",
-                             static_cast<unsigned>(instr.opcode.EffectiveOpCode()),
-                             OpCode::GetInfo(instr.opcode).name.c_str(), instr.hex);
-                throw DecompileFail("Unhandled instruction");
-                break;
-            }
-            }
+            SetDest(0, dest, op_a + " * " + op_b + " + " + op_c, 1, 1);
             break;
         }
         case OpCode::Type::Memory: {
@@ -315,22 +385,33 @@ private:
 
             switch (instr.opcode.EffectiveOpCode()) {
             case OpCode::Id::LD_A: {
-                ASSERT(instr.attribute.fmt20.size == 0);
+                ASSERT_MSG(instr.attribute.fmt20.size == 0, "untested");
                 SetDest(instr.attribute.fmt20.element, gpr0, GetInputAttribute(attribute), 1, 4);
                 break;
             }
             case OpCode::Id::ST_A: {
-                ASSERT(instr.attribute.fmt20.size == 0);
+                ASSERT_MSG(instr.attribute.fmt20.size == 0, "untested");
                 SetDest(instr.attribute.fmt20.element, GetOutputAttribute(attribute), gpr0, 4, 1);
                 break;
             }
-            default: {
-                LOG_CRITICAL(HW_GPU, "Unhandled memory instruction: 0x%02x (%s): 0x%08x",
-                             static_cast<unsigned>(instr.opcode.EffectiveOpCode()),
-                             OpCode::GetInfo(instr.opcode).name.c_str(), instr.hex);
-                throw DecompileFail("Unhandled instruction");
+            case OpCode::Id::TEXS: {
+                ASSERT_MSG(instr.attribute.fmt20.size == 4, "untested");
+                const std::string op_a = GetRegister(instr.gpr8);
+                const std::string op_b = GetRegister(instr.gpr20);
+                const std::string sampler = GetSampler(instr.sampler);
+                const std::string coord = "vec2(" + op_a + ", " + op_b + ")";
+                const std::string texture = "texture(" + sampler + ", " + coord + ")";
+                for (unsigned elem = 0; elem < instr.attribute.fmt20.size; ++elem) {
+                    SetDest(elem, GetRegister(instr.gpr0, elem), texture, 1, 4);
+                }
                 break;
             }
+            default: {
+                NGLOG_CRITICAL(HW_GPU, "Unhandled memory instruction: {} ({}): {}",
+                               static_cast<unsigned>(instr.opcode.EffectiveOpCode()),
+                               OpCode::GetInfo(instr.opcode).name, instr.hex);
+                UNREACHABLE();
+            }
             }
             break;
         }
@@ -342,14 +423,18 @@ private:
                 offset = PROGRAM_END - 1;
                 break;
             }
-
-            default: {
-                LOG_CRITICAL(HW_GPU, "Unhandled instruction: 0x%02x (%s): 0x%08x",
-                             static_cast<unsigned>(instr.opcode.EffectiveOpCode()),
-                             OpCode::GetInfo(instr.opcode).name.c_str(), instr.hex);
-                throw DecompileFail("Unhandled instruction");
+            case OpCode::Id::IPA: {
+                const auto& attribute = instr.attribute.fmt28;
+                std::string dest = GetRegister(instr.gpr0);
+                SetDest(attribute.element, dest, GetInputAttribute(attribute.index), 1, 4);
                 break;
             }
+            default: {
+                NGLOG_CRITICAL(HW_GPU, "Unhandled instruction: {} ({}): {}",
+                               static_cast<unsigned>(instr.opcode.EffectiveOpCode()),
+                               OpCode::GetInfo(instr.opcode).name, instr.hex);
+                UNREACHABLE();
+            }
             }
 
             break;
@@ -514,7 +599,7 @@ boost::optional<ProgramResult> DecompileProgram(const ProgramCode& program_code,
         GLSLGenerator generator(subroutines, program_code, main_offset, stage);
         return ProgramResult{generator.GetShaderCode(), generator.GetEntries()};
     } catch (const DecompileFail& exception) {
-        LOG_ERROR(HW_GPU, "Shader decompilation failed: %s", exception.what());
+        NGLOG_ERROR(HW_GPU, "Shader decompilation failed: {}", exception.what());
     }
     return boost::none;
 }

src/video_core/renderer_opengl/gl_shader_gen.cpp

@@ -27,10 +27,13 @@ out gl_PerVertex {
     vec4 gl_Position;
 };
 
+out vec4 position;
+
 void main() {
     exec_shader();
-}
 
+    gl_Position = position;
+}
 )";
     out += program.first;
     return {out, program.second};
@@ -46,6 +49,7 @@ ProgramResult GenerateFragmentShader(const ShaderSetup& setup, const MaxwellFSCo
                                 .get_value_or({});
     out += R"(
 
+in vec4 position;
 out vec4 color;
 
 uniform sampler2D tex[32];

src/video_core/renderer_opengl/maxwell_to_gl.h

@@ -102,4 +102,68 @@ inline GLenum WrapMode(Tegra::Texture::WrapMode wrap_mode) {
     return {};
 }
 
+inline GLenum BlendEquation(Maxwell::Blend::Equation equation) {
+    switch (equation) {
+    case Maxwell::Blend::Equation::Add:
+        return GL_FUNC_ADD;
+    case Maxwell::Blend::Equation::Subtract:
+        return GL_FUNC_SUBTRACT;
+    case Maxwell::Blend::Equation::ReverseSubtract:
+        return GL_FUNC_REVERSE_SUBTRACT;
+    case Maxwell::Blend::Equation::Min:
+        return GL_MIN;
+    case Maxwell::Blend::Equation::Max:
+        return GL_MAX;
+    }
+    NGLOG_CRITICAL(Render_OpenGL, "Unimplemented blend equation={}", static_cast<u32>(equation));
+    UNREACHABLE();
+    return {};
+}
+
+inline GLenum BlendFunc(Maxwell::Blend::Factor factor) {
+    switch (factor) {
+    case Maxwell::Blend::Factor::Zero:
+        return GL_ZERO;
+    case Maxwell::Blend::Factor::One:
+        return GL_ONE;
+    case Maxwell::Blend::Factor::SourceColor:
+        return GL_SRC_COLOR;
+    case Maxwell::Blend::Factor::OneMinusSourceColor:
+        return GL_ONE_MINUS_SRC_COLOR;
+    case Maxwell::Blend::Factor::SourceAlpha:
+        return GL_SRC_ALPHA;
+    case Maxwell::Blend::Factor::OneMinusSourceAlpha:
+        return GL_ONE_MINUS_SRC_ALPHA;
+    case Maxwell::Blend::Factor::DestAlpha:
+        return GL_DST_ALPHA;
+    case Maxwell::Blend::Factor::OneMinusDestAlpha:
+        return GL_ONE_MINUS_DST_ALPHA;
+    case Maxwell::Blend::Factor::DestColor:
+        return GL_DST_COLOR;
+    case Maxwell::Blend::Factor::OneMinusDestColor:
+        return GL_ONE_MINUS_DST_COLOR;
+    case Maxwell::Blend::Factor::SourceAlphaSaturate:
+        return GL_SRC_ALPHA_SATURATE;
+    case Maxwell::Blend::Factor::Source1Color:
+        return GL_SRC1_COLOR;
+    case Maxwell::Blend::Factor::OneMinusSource1Color:
+        return GL_ONE_MINUS_SRC1_COLOR;
+    case Maxwell::Blend::Factor::Source1Alpha:
+        return GL_SRC1_ALPHA;
+    case Maxwell::Blend::Factor::OneMinusSource1Alpha:
+        return GL_ONE_MINUS_SRC1_ALPHA;
+    case Maxwell::Blend::Factor::ConstantColor:
+        return GL_CONSTANT_COLOR;
+    case Maxwell::Blend::Factor::OneMinusConstantColor:
+        return GL_ONE_MINUS_CONSTANT_COLOR;
+    case Maxwell::Blend::Factor::ConstantAlpha:
+        return GL_CONSTANT_ALPHA;
+    case Maxwell::Blend::Factor::OneMinusConstantAlpha:
+        return GL_ONE_MINUS_CONSTANT_ALPHA;
+    }
+    NGLOG_CRITICAL(Render_OpenGL, "Unimplemented blend factor={}", static_cast<u32>(factor));
+    UNREACHABLE();
+    return {};
+}
+
 } // namespace MaxwellToGL

src/video_core/textures/texture.h

@@ -13,8 +13,10 @@ namespace Tegra {
 namespace Texture {
 
 enum class TextureFormat : u32 {
-    A8R8G8B8 = 8,
+    A8R8G8B8 = 0x8,
     DXT1 = 0x24,
+    DXT23 = 0x25,
+    DXT45 = 0x26,
 };
 
 enum class TextureType : u32 {