Merge 88897b67aa into ac959799e4

gl_rasterizer: Implement multiple color attachments.

src/audio_core/audio_renderer.cpp

@@ -17,10 +17,10 @@ AudioRenderer::AudioRenderer(AudioRendererParameter params,
                              Kernel::SharedPtr<Kernel::Event> buffer_event)
     : worker_params{params}, buffer_event{buffer_event}, voices(params.voice_count) {
 
-    audio_core = std::make_unique<AudioCore::AudioOut>();
-    stream = audio_core->OpenStream(STREAM_SAMPLE_RATE, STREAM_NUM_CHANNELS, "AudioRenderer",
-                                    [=]() { buffer_event->Signal(); });
-    audio_core->StartStream(stream);
+    audio_out = std::make_unique<AudioCore::AudioOut>();
+    stream = audio_out->OpenStream(STREAM_SAMPLE_RATE, STREAM_NUM_CHANNELS, "AudioRenderer",
+                                   [=]() { buffer_event->Signal(); });
+    audio_out->StartStream(stream);
 
     QueueMixedBuffer(0);
     QueueMixedBuffer(1);
@@ -236,11 +236,11 @@ void AudioRenderer::QueueMixedBuffer(Buffer::Tag tag) {
             }
         }
     }
-    audio_core->QueueBuffer(stream, tag, std::move(buffer));
+    audio_out->QueueBuffer(stream, tag, std::move(buffer));
 }
 
 void AudioRenderer::ReleaseAndQueueBuffers() {
-    const auto released_buffers{audio_core->GetTagsAndReleaseBuffers(stream, 2)};
+    const auto released_buffers{audio_out->GetTagsAndReleaseBuffers(stream, 2)};
     for (const auto& tag : released_buffers) {
         QueueMixedBuffer(tag);
     }

src/audio_core/audio_renderer.h

@@ -204,7 +204,7 @@ private:
     AudioRendererParameter worker_params;
     Kernel::SharedPtr<Kernel::Event> buffer_event;
     std::vector<VoiceState> voices;
-    std::unique_ptr<AudioCore::AudioOut> audio_core;
+    std::unique_ptr<AudioCore::AudioOut> audio_out;
     AudioCore::StreamPtr stream;
 };
 

src/common/CMakeLists.txt

@@ -1,13 +1,16 @@
 # Generate cpp with Git revision from template
-# Also if this is a CI build, add the build name (ie: Nightly, Bleeding Edge) to the scm_rev file as well
+# Also if this is a CI build, add the build name (ie: Nightly, Canary) to the scm_rev file as well
 set(REPO_NAME "")
+set(BUILD_VERSION "0")
 if ($ENV{CI})
   if ($ENV{TRAVIS})
     set(BUILD_REPOSITORY $ENV{TRAVIS_REPO_SLUG})
+    set(BUILD_TAG $ENV{TRAVIS_TAG})
   elseif($ENV{APPVEYOR})
     set(BUILD_REPOSITORY $ENV{APPVEYOR_REPO_NAME})
+    set(BUILD_TAG $ENV{APPVEYOR_REPO_TAG_NAME})
   endif()
-  # regex capture the string nightly or bleeding-edge into CMAKE_MATCH_1
+  # regex capture the string nightly or canary into CMAKE_MATCH_1
   string(REGEX MATCH "yuzu-emu/yuzu-?(.*)" OUTVAR ${BUILD_REPOSITORY})
   if (${CMAKE_MATCH_COUNT} GREATER 0)
     # capitalize the first letter of each word in the repo name.
@@ -16,10 +19,21 @@ if ($ENV{CI})
       string(SUBSTRING ${WORD} 0 1 FIRST_LETTER)
       string(SUBSTRING ${WORD} 1 -1 REMAINDER)
       string(TOUPPER ${FIRST_LETTER} FIRST_LETTER)
-      # this leaves a trailing space on the last word, but we actually want that
-      # because of how it's styled in the title bar.
-      set(REPO_NAME "${REPO_NAME}${FIRST_LETTER}${REMAINDER} ")
+      set(REPO_NAME "${REPO_NAME}${FIRST_LETTER}${REMAINDER}")
     endforeach()
+    if (BUILD_TAG)
+      string(REGEX MATCH "${CMAKE_MATCH_1}-([0-9]+)" OUTVAR ${BUILD_TAG})
+      if (${CMAKE_MATCH_COUNT} GREATER 0)
+        set(BUILD_VERSION ${CMAKE_MATCH_1})
+      endif()
+      if (BUILD_VERSION)
+        # This leaves a trailing space on the last word, but we actually want that
+        # because of how it's styled in the title bar.
+        set(BUILD_FULLNAME "${REPO_NAME} #${BUILD_VERSION} ")
+      else()
+        set(BUILD_FULLNAME "")
+      endif()
+    endif()
   endif()
 endif()
 configure_file("${CMAKE_CURRENT_SOURCE_DIR}/scm_rev.cpp.in" "${CMAKE_CURRENT_SOURCE_DIR}/scm_rev.cpp" @ONLY)

src/common/scm_rev.cpp.in

@@ -9,6 +9,8 @@
 #define GIT_DESC     "@GIT_DESC@"
 #define BUILD_NAME   "@REPO_NAME@"
 #define BUILD_DATE   "@BUILD_DATE@"
+#define BUILD_FULLNAME "@BUILD_FULLNAME@"
+#define BUILD_VERSION "@BUILD_VERSION@"
 
 namespace Common {
 
@@ -17,6 +19,8 @@ const char g_scm_branch[]   = GIT_BRANCH;
 const char g_scm_desc[]     = GIT_DESC;
 const char g_build_name[]   = BUILD_NAME;
 const char g_build_date[]   = BUILD_DATE;
+const char g_build_fullname[] = BUILD_FULLNAME;
+const char g_build_version[]  = BUILD_VERSION;
 
 } // namespace
 

src/common/scm_rev.h

@@ -11,5 +11,7 @@ extern const char g_scm_branch[];
 extern const char g_scm_desc[];
 extern const char g_build_name[];
 extern const char g_build_date[];
+extern const char g_build_fullname[];
+extern const char g_build_version[];
 
 } // namespace Common

src/core/core.cpp

@@ -24,7 +24,6 @@
 #include "core/hle/kernel/scheduler.h"
 #include "core/hle/kernel/thread.h"
 #include "core/hle/service/service.h"
-#include "core/hle/service/sm/controller.h"
 #include "core/hle/service/sm/sm.h"
 #include "core/loader/loader.h"
 #include "core/perf_stats.h"
@@ -137,7 +136,7 @@ struct System::Impl {
         if (virtual_filesystem == nullptr)
             virtual_filesystem = std::make_shared<FileSys::RealVfsFilesystem>();
 
-        current_process = Kernel::Process::Create(kernel, "main");
+        kernel.MakeCurrentProcess(Kernel::Process::Create(kernel, "main"));
 
         cpu_barrier = std::make_shared<CpuBarrier>();
         cpu_exclusive_monitor = Cpu::MakeExclusiveMonitor(cpu_cores.size());
@@ -203,7 +202,7 @@ struct System::Impl {
             return init_result;
         }
 
-        const Loader::ResultStatus load_result{app_loader->Load(current_process)};
+        const Loader::ResultStatus load_result{app_loader->Load(kernel.CurrentProcess())};
         if (load_result != Loader::ResultStatus::Success) {
             LOG_CRITICAL(Core, "Failed to load ROM (Error {})!", static_cast<int>(load_result));
             Shutdown();
@@ -282,7 +281,6 @@ struct System::Impl {
     std::unique_ptr<VideoCore::RendererBase> renderer;
     std::unique_ptr<Tegra::GPU> gpu_core;
     std::shared_ptr<Tegra::DebugContext> debug_context;
-    Kernel::SharedPtr<Kernel::Process> current_process;
     std::shared_ptr<ExclusiveMonitor> cpu_exclusive_monitor;
     std::shared_ptr<CpuBarrier> cpu_barrier;
     std::array<std::shared_ptr<Cpu>, NUM_CPU_CORES> cpu_cores;
@@ -364,7 +362,11 @@ const std::shared_ptr<Kernel::Scheduler>& System::Scheduler(size_t core_index) {
 }
 
 Kernel::SharedPtr<Kernel::Process>& System::CurrentProcess() {
-    return impl->current_process;
+    return impl->kernel.CurrentProcess();
+}
+
+const Kernel::SharedPtr<Kernel::Process>& System::CurrentProcess() const {
+    return impl->kernel.CurrentProcess();
 }
 
 ARM_Interface& System::ArmInterface(size_t core_index) {

src/core/core.h

@@ -174,9 +174,12 @@ public:
     /// Gets the scheduler for the CPU core with the specified index
     const std::shared_ptr<Kernel::Scheduler>& Scheduler(size_t core_index);
 
-    /// Gets the current process
+    /// Provides a reference to the current process
     Kernel::SharedPtr<Kernel::Process>& CurrentProcess();
 
+    /// Provides a constant reference to the current process.
+    const Kernel::SharedPtr<Kernel::Process>& CurrentProcess() const;
+
     /// Provides a reference to the kernel instance.
     Kernel::KernelCore& Kernel();
 

src/core/file_sys/nca_patch.cpp

@@ -2,6 +2,10 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include <algorithm>
+#include <cstddef>
+#include <cstring>
+
 #include "common/assert.h"
 #include "core/crypto/aes_util.h"
 #include "core/file_sys/nca_patch.h"
@@ -13,9 +17,9 @@ BKTR::BKTR(VirtualFile base_romfs_, VirtualFile bktr_romfs_, RelocationBlock rel
            std::vector<SubsectionBucket> subsection_buckets_, bool is_encrypted_,
            Core::Crypto::Key128 key_, u64 base_offset_, u64 ivfc_offset_,
            std::array<u8, 8> section_ctr_)
-    : base_romfs(std::move(base_romfs_)), bktr_romfs(std::move(bktr_romfs_)),
-      relocation(relocation_), relocation_buckets(std::move(relocation_buckets_)),
+    : relocation(relocation_), relocation_buckets(std::move(relocation_buckets_)),
       subsection(subsection_), subsection_buckets(std::move(subsection_buckets_)),
+      base_romfs(std::move(base_romfs_)), bktr_romfs(std::move(bktr_romfs_)),
       encrypted(is_encrypted_), key(key_), base_offset(base_offset_), ivfc_offset(ivfc_offset_),
       section_ctr(section_ctr_) {
     for (size_t i = 0; i < relocation.number_buckets - 1; ++i) {

src/core/file_sys/nca_patch.h

@@ -5,10 +5,13 @@
 #pragma once
 
 #include <array>
+#include <memory>
 #include <vector>
-#include <common/common_funcs.h>
+
+#include "common/common_funcs.h"
+#include "common/common_types.h"
+#include "common/swap.h"
 #include "core/crypto/key_manager.h"
-#include "core/file_sys/romfs.h"
 
 namespace FileSys {
 

src/core/file_sys/patch_manager.cpp

@@ -2,6 +2,10 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include <array>
+#include <cstddef>
+
+#include "common/logging/log.h"
 #include "core/file_sys/content_archive.h"
 #include "core/file_sys/control_metadata.h"
 #include "core/file_sys/patch_manager.h"

src/core/file_sys/patch_manager.h

@@ -5,6 +5,7 @@
 #pragma once
 
 #include <map>
+#include <memory>
 #include <string>
 #include "common/common_types.h"
 #include "core/file_sys/nca_metadata.h"

src/core/hle/kernel/kernel.cpp

@@ -116,6 +116,7 @@ struct KernelCore::Impl {
         next_thread_id = 1;
 
         process_list.clear();
+        current_process.reset();
 
         handle_table.Clear();
         resource_limits.fill(nullptr);
@@ -206,6 +207,7 @@ struct KernelCore::Impl {
 
     // Lists all processes that exist in the current session.
     std::vector<SharedPtr<Process>> process_list;
+    SharedPtr<Process> current_process;
 
     Kernel::HandleTable handle_table;
     std::array<SharedPtr<ResourceLimit>, 4> resource_limits;
@@ -264,6 +266,18 @@ void KernelCore::AppendNewProcess(SharedPtr<Process> process) {
     impl->process_list.push_back(std::move(process));
 }
 
+void KernelCore::MakeCurrentProcess(SharedPtr<Process> process) {
+    impl->current_process = std::move(process);
+}
+
+SharedPtr<Process>& KernelCore::CurrentProcess() {
+    return impl->current_process;
+}
+
+const SharedPtr<Process>& KernelCore::CurrentProcess() const {
+    return impl->current_process;
+}
+
 void KernelCore::AddNamedPort(std::string name, SharedPtr<ClientPort> port) {
     impl->named_ports.emplace(std::move(name), std::move(port));
 }

src/core/hle/kernel/kernel.h

@@ -65,6 +65,15 @@ public:
     /// Adds the given shared pointer to an internal list of active processes.
     void AppendNewProcess(SharedPtr<Process> process);
 
+    /// Makes the given process the new current process.
+    void MakeCurrentProcess(SharedPtr<Process> process);
+
+    /// Retrieves a reference to the current process.
+    SharedPtr<Process>& CurrentProcess();
+
+    /// Retrieves a const reference to the current process.
+    const SharedPtr<Process>& CurrentProcess() const;
+
     /// Adds a port to the named port table
     void AddNamedPort(std::string name, SharedPtr<ClientPort> port);
 

src/core/hle/service/hid/hid.cpp

@@ -2,7 +2,6 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
-#include <atomic>
 #include "common/logging/log.h"
 #include "core/core.h"
 #include "core/core_timing.h"
@@ -78,7 +77,7 @@ private:
         SharedMemory mem{};
         std::memcpy(&mem, shared_mem->GetPointer(), sizeof(SharedMemory));
 
-        if (is_device_reload_pending.exchange(false))
+        if (Settings::values.is_device_reload_pending.exchange(false))
             LoadInputDevices();
 
         // Set up controllers as neon red+blue Joy-Con attached to console
@@ -267,7 +266,6 @@ private:
     CoreTiming::EventType* pad_update_event;
 
     // Stored input state info
-    std::atomic<bool> is_device_reload_pending{true};
     std::array<std::unique_ptr<Input::ButtonDevice>, Settings::NativeButton::NUM_BUTTONS_HID>
         buttons;
     std::array<std::unique_ptr<Input::AnalogDevice>, Settings::NativeAnalog::NUM_STICKS_HID> sticks;
@@ -797,7 +795,9 @@ public:
     }
 };
 
-void ReloadInputDevices() {}
+void ReloadInputDevices() {
+    Settings::values.is_device_reload_pending.store(true);
+}
 
 void InstallInterfaces(SM::ServiceManager& service_manager) {
     std::make_shared<Hid>()->InstallAsService(service_manager);

src/core/hle/service/service.cpp

@@ -74,8 +74,6 @@ using Kernel::SharedPtr;
 
 namespace Service {
 
-std::unordered_map<std::string, SharedPtr<ClientPort>> g_kernel_named_ports;
-
 /**
  * Creates a function string for logging, complete with the name (or header code, depending
  * on what's passed in) the port name, and all the cmd_buff arguments.

src/core/hle/service/sm/controller.cpp

@@ -57,4 +57,6 @@ Controller::Controller() : ServiceFramework("IpcController") {
     RegisterHandlers(functions);
 }
 
+Controller::~Controller() = default;
+
 } // namespace Service::SM

src/core/hle/service/sm/controller.h

@@ -11,7 +11,7 @@ namespace Service::SM {
 class Controller final : public ServiceFramework<Controller> {
 public:
     Controller();
-    ~Controller() = default;
+    ~Controller() override;
 
 private:
     void ConvertSessionToDomain(Kernel::HLERequestContext& ctx);

src/core/hle/service/sm/sm.cpp

@@ -15,6 +15,7 @@
 
 namespace Service::SM {
 
+ServiceManager::ServiceManager() = default;
 ServiceManager::~ServiceManager() = default;
 
 void ServiceManager::InvokeControlRequest(Kernel::HLERequestContext& context) {

src/core/hle/service/sm/sm.h

@@ -46,6 +46,7 @@ class ServiceManager {
 public:
     static void InstallInterfaces(std::shared_ptr<ServiceManager> self);
 
+    ServiceManager();
     ~ServiceManager();
 
     ResultVal<Kernel::SharedPtr<Kernel::ServerPort>> RegisterService(std::string name,

src/core/settings.h

@@ -5,6 +5,7 @@
 #pragma once
 
 #include <array>
+#include <atomic>
 #include <string>
 #include "common/common_types.h"
 
@@ -120,6 +121,7 @@ struct Values {
     std::array<std::string, NativeAnalog::NumAnalogs> analogs;
     std::string motion_device;
     std::string touch_device;
+    std::atomic_bool is_device_reload_pending{true};
 
     // Core
     bool use_cpu_jit;

src/video_core/CMakeLists.txt

@@ -23,6 +23,7 @@ add_library(video_core STATIC
     renderer_base.cpp
     renderer_base.h
     renderer_opengl/gl_buffer_cache.cpp
+    renderer_opengl/gl_buffer_cache.h
     renderer_opengl/gl_rasterizer.cpp
     renderer_opengl/gl_rasterizer.h
     renderer_opengl/gl_rasterizer_cache.cpp

src/video_core/engines/maxwell_3d.cpp

@@ -293,10 +293,6 @@ Texture::TICEntry Maxwell3D::GetTICEntry(u32 tic_index) const {
                    tic_entry.header_version == Texture::TICHeaderVersion::Pitch,
                "TIC versions other than BlockLinear or Pitch are unimplemented");
 
-    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();
     auto g_type = tic_entry.g_type.Value();
     auto b_type = tic_entry.b_type.Value();

src/video_core/engines/maxwell_3d.h

@@ -533,7 +533,11 @@ public:
                 u32 stencil_back_mask;
                 u32 stencil_back_func_mask;
 
-                INSERT_PADDING_WORDS(0x20);
+                INSERT_PADDING_WORDS(0x13);
+
+                u32 rt_separate_frag_data;
+
+                INSERT_PADDING_WORDS(0xC);
 
                 struct {
                     u32 address_high;
@@ -557,7 +561,22 @@ public:
                 struct {
                     union {
                         BitField<0, 4, u32> count;
+                        BitField<4, 3, u32> map_0;
+                        BitField<7, 3, u32> map_1;
+                        BitField<10, 3, u32> map_2;
+                        BitField<13, 3, u32> map_3;
+                        BitField<16, 3, u32> map_4;
+                        BitField<19, 3, u32> map_5;
+                        BitField<22, 3, u32> map_6;
+                        BitField<25, 3, u32> map_7;
                     };
+
+                    u32 GetMap(size_t index) const {
+                        const std::array<u32, NumRenderTargets> maps{map_0, map_1, map_2, map_3,
+                                                                     map_4, map_5, map_6, map_7};
+                        ASSERT(index < maps.size());
+                        return maps[index];
+                    }
                 } rt_control;
 
                 INSERT_PADDING_WORDS(0x2);
@@ -968,6 +987,7 @@ ASSERT_REG_POSITION(clear_stencil, 0x368);
 ASSERT_REG_POSITION(stencil_back_func_ref, 0x3D5);
 ASSERT_REG_POSITION(stencil_back_mask, 0x3D6);
 ASSERT_REG_POSITION(stencil_back_func_mask, 0x3D7);
+ASSERT_REG_POSITION(rt_separate_frag_data, 0x3EB);
 ASSERT_REG_POSITION(zeta, 0x3F8);
 ASSERT_REG_POSITION(vertex_attrib_format, 0x458);
 ASSERT_REG_POSITION(rt_control, 0x487);

src/video_core/engines/maxwell_dma.cpp

@@ -41,7 +41,6 @@ void MaxwellDMA::HandleCopy() {
 
     // TODO(Subv): Perform more research and implement all features of this engine.
     ASSERT(regs.exec.enable_swizzle == 0);
-    ASSERT(regs.exec.enable_2d == 1);
     ASSERT(regs.exec.query_mode == Regs::QueryMode::None);
     ASSERT(regs.exec.query_intr == Regs::QueryIntr::None);
     ASSERT(regs.exec.copy_mode == Regs::CopyMode::Unk2);
@@ -51,10 +50,19 @@ void MaxwellDMA::HandleCopy() {
     ASSERT(regs.dst_params.pos_y == 0);
 
     if (regs.exec.is_dst_linear == regs.exec.is_src_linear) {
-        Memory::CopyBlock(dest_cpu, source_cpu, regs.x_count * regs.y_count);
+        size_t copy_size = regs.x_count;
+
+        // When the enable_2d bit is disabled, the copy is performed as if we were copying a 1D
+        // buffer of length `x_count`, otherwise we copy a 2D buffer of size (x_count, y_count).
+        if (regs.exec.enable_2d) {
+            copy_size = copy_size * regs.y_count;
+        }
+
+        Memory::CopyBlock(dest_cpu, source_cpu, copy_size);
         return;
     }
 
+    ASSERT(regs.exec.enable_2d == 1);
     u8* src_buffer = Memory::GetPointer(source_cpu);
     u8* dst_buffer = Memory::GetPointer(dest_cpu);
 

src/video_core/engines/shader_bytecode.h

@@ -244,6 +244,16 @@ enum class TextureType : u64 {
     TextureCube = 3,
 };
 
+enum class TextureQueryType : u64 {
+    Dimension = 1,
+    TextureType = 2,
+    SamplePosition = 5,
+    Filter = 16,
+    LevelOfDetail = 18,
+    Wrap = 20,
+    BorderColor = 22,
+};
+
 enum class IpaInterpMode : u64 { Linear = 0, Perspective = 1, Flat = 2, Sc = 3 };
 enum class IpaSampleMode : u64 { Default = 0, Centroid = 1, Offset = 2 };
 
@@ -518,6 +528,21 @@ union Instruction {
         }
     } tex;
 
+    union {
+        BitField<22, 6, TextureQueryType> query_type;
+        BitField<31, 4, u64> component_mask;
+    } txq;
+
+    union {
+        BitField<28, 1, u64> array;
+        BitField<29, 2, TextureType> texture_type;
+        BitField<31, 4, u64> component_mask;
+
+        bool IsComponentEnabled(size_t component) const {
+            return ((1ull << component) & component_mask) != 0;
+        }
+    } tmml;
+
     union {
         BitField<28, 1, u64> array;
         BitField<29, 2, TextureType> texture_type;
@@ -670,11 +695,13 @@ public:
         LDG, // Load from global memory
         STG, // Store in global memory
         TEX,
-        TEXQ,  // Texture Query
-        TEXS,  // Texture Fetch with scalar/non-vec4 source/destinations
-        TLDS,  // Texture Load with scalar/non-vec4 source/destinations
-        TLD4,  // Texture Load 4
-        TLD4S, // Texture Load 4 with scalar / non - vec4 source / destinations
+        TXQ,    // Texture Query
+        TEXS,   // Texture Fetch with scalar/non-vec4 source/destinations
+        TLDS,   // Texture Load with scalar/non-vec4 source/destinations
+        TLD4,   // Texture Load 4
+        TLD4S,  // Texture Load 4 with scalar / non - vec4 source / destinations
+        TMML_B, // Texture Mip Map Level
+        TMML,   // Texture Mip Map Level
         EXIT,
         IPA,
         FFMA_IMM, // Fused Multiply and Add
@@ -894,11 +921,13 @@ private:
             INST("1110111011010---", Id::LDG, Type::Memory, "LDG"),
             INST("1110111011011---", Id::STG, Type::Memory, "STG"),
             INST("110000----111---", Id::TEX, Type::Memory, "TEX"),
-            INST("1101111101001---", Id::TEXQ, Type::Memory, "TEXQ"),
+            INST("1101111101001---", Id::TXQ, Type::Memory, "TXQ"),
             INST("1101100---------", Id::TEXS, Type::Memory, "TEXS"),
             INST("1101101---------", Id::TLDS, Type::Memory, "TLDS"),
             INST("110010----111---", Id::TLD4, Type::Memory, "TLD4"),
             INST("1101111100------", Id::TLD4S, Type::Memory, "TLD4S"),
+            INST("110111110110----", Id::TMML_B, Type::Memory, "TMML_B"),
+            INST("1101111101011---", Id::TMML, Type::Memory, "TMML"),
             INST("111000110000----", Id::EXIT, Type::Trivial, "EXIT"),
             INST("11100000--------", Id::IPA, Type::Trivial, "IPA"),
             INST("0011001-1-------", Id::FFMA_IMM, Type::Ffma, "FFMA_IMM"),

src/video_core/renderer_opengl/gl_rasterizer.cpp

@@ -151,24 +151,17 @@ void RasterizerOpenGL::SetupVertexArrays() {
         Tegra::GPUVAddr start = vertex_array.StartAddress();
         const Tegra::GPUVAddr end = regs.vertex_array_limit[index].LimitAddress();
 
-        if (regs.instanced_arrays.IsInstancingEnabled(index) && vertex_array.divisor != 0) {
-            start += vertex_array.stride * (gpu.state.current_instance / vertex_array.divisor);
-        }
-
         ASSERT(end > start);
-        u64 size = end - start + 1;
-
-        GLintptr vertex_buffer_offset = buffer_cache.UploadMemory(start, size);
+        const u64 size = end - start + 1;
+        const GLintptr vertex_buffer_offset = buffer_cache.UploadMemory(start, size);
 
         // Bind the vertex array to the buffer at the current offset.
         glBindVertexBuffer(index, buffer_cache.GetHandle(), vertex_buffer_offset,
                            vertex_array.stride);
 
         if (regs.instanced_arrays.IsInstancingEnabled(index) && vertex_array.divisor != 0) {
-            // Tell OpenGL that this is an instanced vertex buffer to prevent accessing different
-            // indexes on each vertex. We do the instance indexing manually by incrementing the
-            // start address of the vertex buffer.
-            glVertexBindingDivisor(index, 1);
+            // Enable vertex buffer instancing with the specified divisor.
+            glVertexBindingDivisor(index, vertex_array.divisor);
         } else {
             // Disable the vertex buffer instancing.
             glVertexBindingDivisor(index, 0);
@@ -178,7 +171,7 @@ void RasterizerOpenGL::SetupVertexArrays() {
 
 void RasterizerOpenGL::SetupShaders() {
     MICROPROFILE_SCOPE(OpenGL_Shader);
-    auto& gpu = Core::System::GetInstance().GPU().Maxwell3D();
+    const auto& gpu = Core::System::GetInstance().GPU().Maxwell3D();
 
     // Next available bindpoints to use when uploading the const buffers and textures to the GLSL
     // shaders. The constbuffer bindpoint starts after the shader stage configuration bind points.
@@ -186,7 +179,7 @@ void RasterizerOpenGL::SetupShaders() {
     u32 current_texture_bindpoint = 0;
 
     for (size_t index = 0; index < Maxwell::MaxShaderProgram; ++index) {
-        auto& shader_config = gpu.regs.shader_config[index];
+        const auto& shader_config = gpu.regs.shader_config[index];
         const Maxwell::ShaderProgram program{static_cast<Maxwell::ShaderProgram>(index)};
 
         // Skip stages that are not enabled
@@ -198,7 +191,7 @@ void RasterizerOpenGL::SetupShaders() {
 
         GLShader::MaxwellUniformData ubo{};
         ubo.SetFromRegs(gpu.state.shader_stages[stage]);
-        GLintptr offset = buffer_cache.UploadHostMemory(
+        const GLintptr offset = buffer_cache.UploadHostMemory(
             &ubo, sizeof(ubo), static_cast<size_t>(uniform_buffer_alignment));
 
         // Bind the buffer
@@ -237,6 +230,8 @@ void RasterizerOpenGL::SetupShaders() {
         }
     }
 
+    state.Apply();
+
     shader_program_manager->UseTrivialGeometryShader();
 }
 
@@ -299,17 +294,10 @@ void RasterizerOpenGL::UpdatePagesCachedCount(VAddr addr, u64 size, int delta) {
         cached_pages.add({pages_interval, delta});
 }
 
-std::pair<Surface, Surface> RasterizerOpenGL::ConfigureFramebuffers(bool using_color_fb,
-                                                                    bool using_depth_fb,
-                                                                    bool preserve_contents) {
+void RasterizerOpenGL::ConfigureFramebuffers(bool using_depth_fb, bool preserve_contents) {
     MICROPROFILE_SCOPE(OpenGL_Framebuffer);
     const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
 
-    if (regs.rt[0].format == Tegra::RenderTargetFormat::NONE) {
-        LOG_ERROR(HW_GPU, "RenderTargetFormat is not configured");
-        using_color_fb = false;
-    }
-
     const bool has_stencil = regs.stencil_enable;
     const bool write_color_fb =
         state.color_mask.red_enabled == GL_TRUE || state.color_mask.green_enabled == GL_TRUE ||
@@ -319,41 +307,52 @@ std::pair<Surface, Surface> RasterizerOpenGL::ConfigureFramebuffers(bool using_c
         (state.depth.test_enabled && state.depth.write_mask == GL_TRUE) ||
         (has_stencil && (state.stencil.front.write_mask || state.stencil.back.write_mask));
 
-    Surface color_surface;
     Surface depth_surface;
-    MathUtil::Rectangle<u32> surfaces_rect;
-    std::tie(color_surface, depth_surface, surfaces_rect) =
-        res_cache.GetFramebufferSurfaces(using_color_fb, using_depth_fb, preserve_contents);
+    if (using_depth_fb) {
+        depth_surface = res_cache.GetDepthBufferSurface(preserve_contents);
+    }
 
-    const MathUtil::Rectangle<s32> viewport_rect{regs.viewport_transform[0].GetRect()};
-    const MathUtil::Rectangle<u32> draw_rect{
-        static_cast<u32>(std::clamp<s32>(static_cast<s32>(surfaces_rect.left) + viewport_rect.left,
-                                         surfaces_rect.left, surfaces_rect.right)), // Left
-        static_cast<u32>(std::clamp<s32>(static_cast<s32>(surfaces_rect.bottom) + viewport_rect.top,
-                                         surfaces_rect.bottom, surfaces_rect.top)), // Top
-        static_cast<u32>(std::clamp<s32>(static_cast<s32>(surfaces_rect.left) + viewport_rect.right,
-                                         surfaces_rect.left, surfaces_rect.right)), // Right
-        static_cast<u32>(
-            std::clamp<s32>(static_cast<s32>(surfaces_rect.bottom) + viewport_rect.bottom,
-                            surfaces_rect.bottom, surfaces_rect.top))}; // Bottom
+    // TODO(bunnei): Figure out how the below register works. According to envytools, this should be
+    // used to enable multiple render targets. However, it is left unset on all games that I have
+    // tested.
+    ASSERT_MSG(regs.rt_separate_frag_data == 0, "Unimplemented");
 
     // Bind the framebuffer surfaces
-    BindFramebufferSurfaces(color_surface, depth_surface, has_stencil);
-
-    SyncViewport(surfaces_rect);
-
-    // Viewport can have negative offsets or larger dimensions than our framebuffer sub-rect. Enable
-    // scissor test to prevent drawing outside of the framebuffer region
-    state.scissor.enabled = true;
-    state.scissor.x = draw_rect.left;
-    state.scissor.y = draw_rect.bottom;
-    state.scissor.width = draw_rect.GetWidth();
-    state.scissor.height = draw_rect.GetHeight();
+    state.draw.draw_framebuffer = framebuffer.handle;
     state.Apply();
 
-    // Only return the surface to be marked as dirty if writing to it is enabled.
-    return std::make_pair(write_color_fb ? color_surface : nullptr,
-                          write_depth_fb ? depth_surface : nullptr);
+    std::array<GLenum, Maxwell::NumRenderTargets> buffers;
+    for (size_t index = 0; index < Maxwell::NumRenderTargets; ++index) {
+        Surface color_surface = res_cache.GetColorBufferSurface(index, preserve_contents);
+        buffers[index] = GL_COLOR_ATTACHMENT0 + regs.rt_control.GetMap(index);
+        glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER,
+                               GL_COLOR_ATTACHMENT0 + static_cast<GLenum>(index), GL_TEXTURE_2D,
+                               color_surface != nullptr ? color_surface->Texture().handle : 0, 0);
+    }
+
+    glDrawBuffers(regs.rt_control.count, buffers.data());
+
+    if (depth_surface) {
+        if (has_stencil) {
+            // Attach both depth and stencil
+            glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, GL_TEXTURE_2D,
+                                   depth_surface->Texture().handle, 0);
+        } else {
+            // Attach depth
+            glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_DEPTH_ATTACHMENT, GL_TEXTURE_2D,
+                                   depth_surface->Texture().handle, 0);
+            // Clear stencil attachment
+            glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_STENCIL_ATTACHMENT, GL_TEXTURE_2D, 0, 0);
+        }
+    } else {
+        // Clear both depth and stencil attachment
+        glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, GL_TEXTURE_2D, 0,
+                               0);
+    }
+
+    SyncViewport();
+
+    state.Apply();
 }
 
 void RasterizerOpenGL::Clear() {
@@ -412,8 +411,7 @@ void RasterizerOpenGL::Clear() {
 
     ScopeAcquireGLContext acquire_context{emu_window};
 
-    auto [dirty_color_surface, dirty_depth_surface] =
-        ConfigureFramebuffers(use_color_fb, use_depth_fb, false);
+    ConfigureFramebuffers(use_depth_fb, false);
 
     clear_state.Apply();
 
@@ -430,12 +428,12 @@ void RasterizerOpenGL::DrawArrays() {
         return;
 
     MICROPROFILE_SCOPE(OpenGL_Drawing);
-    const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
+    const auto& gpu = Core::System::GetInstance().GPU().Maxwell3D();
+    const auto& regs = gpu.regs;
 
     ScopeAcquireGLContext acquire_context{emu_window};
 
-    auto [dirty_color_surface, dirty_depth_surface] =
-        ConfigureFramebuffers(true, regs.zeta.Address() != 0 && regs.zeta_enable != 0, true);
+    ConfigureFramebuffers(true, true);
 
     SyncDepthTestState();
     SyncStencilTestState();
@@ -448,7 +446,8 @@ void RasterizerOpenGL::DrawArrays() {
 
     // Draw the vertex batch
     const bool is_indexed = accelerate_draw == AccelDraw::Indexed;
-    const u64 index_buffer_size{regs.index_array.count * regs.index_array.FormatSizeInBytes()};
+    const u64 index_buffer_size{static_cast<u64>(regs.index_array.count) *
+                                static_cast<u64>(regs.index_array.FormatSizeInBytes())};
 
     state.draw.vertex_buffer = buffer_cache.GetHandle();
     state.Apply();
@@ -491,13 +490,29 @@ void RasterizerOpenGL::DrawArrays() {
         const GLint base_vertex{static_cast<GLint>(regs.vb_element_base)};
 
         // Adjust the index buffer offset so it points to the first desired index.
-        index_buffer_offset += regs.index_array.first * regs.index_array.FormatSizeInBytes();
+        index_buffer_offset += static_cast<GLintptr>(regs.index_array.first) *
+                               static_cast<GLintptr>(regs.index_array.FormatSizeInBytes());
 
-        glDrawElementsBaseVertex(primitive_mode, regs.index_array.count,
-                                 MaxwellToGL::IndexFormat(regs.index_array.format),
-                                 reinterpret_cast<const void*>(index_buffer_offset), base_vertex);
+        if (gpu.state.current_instance > 0) {
+            glDrawElementsInstancedBaseVertexBaseInstance(
+                primitive_mode, regs.index_array.count,
+                MaxwellToGL::IndexFormat(regs.index_array.format),
+                reinterpret_cast<const void*>(index_buffer_offset), 1, base_vertex,
+                gpu.state.current_instance);
+        } else {
+            glDrawElementsBaseVertex(primitive_mode, regs.index_array.count,
+                                     MaxwellToGL::IndexFormat(regs.index_array.format),
+                                     reinterpret_cast<const void*>(index_buffer_offset),
+                                     base_vertex);
+        }
     } else {
-        glDrawArrays(primitive_mode, regs.vertex_buffer.first, regs.vertex_buffer.count);
+        if (gpu.state.current_instance > 0) {
+            glDrawArraysInstancedBaseInstance(primitive_mode, regs.vertex_buffer.first,
+                                              regs.vertex_buffer.count, 1,
+                                              gpu.state.current_instance);
+        } else {
+            glDrawArrays(primitive_mode, regs.vertex_buffer.first, regs.vertex_buffer.count);
+        }
     }
 
     // Disable scissor test
@@ -514,13 +529,9 @@ void RasterizerOpenGL::DrawArrays() {
 
 void RasterizerOpenGL::NotifyMaxwellRegisterChanged(u32 method) {}
 
-void RasterizerOpenGL::FlushAll() {
-    MICROPROFILE_SCOPE(OpenGL_CacheManagement);
-}
+void RasterizerOpenGL::FlushAll() {}
 
-void RasterizerOpenGL::FlushRegion(VAddr addr, u64 size) {
-    MICROPROFILE_SCOPE(OpenGL_CacheManagement);
-}
+void RasterizerOpenGL::FlushRegion(VAddr addr, u64 size) {}
 
 void RasterizerOpenGL::InvalidateRegion(VAddr addr, u64 size) {
     MICROPROFILE_SCOPE(OpenGL_CacheManagement);
@@ -530,7 +541,6 @@ void RasterizerOpenGL::InvalidateRegion(VAddr addr, u64 size) {
 }
 
 void RasterizerOpenGL::FlushAndInvalidateRegion(VAddr addr, u64 size) {
-    MICROPROFILE_SCOPE(OpenGL_CacheManagement);
     InvalidateRegion(addr, size);
 }
 
@@ -578,7 +588,7 @@ bool RasterizerOpenGL::AccelerateDisplay(const Tegra::FramebufferConfig& config,
 void RasterizerOpenGL::SamplerInfo::Create() {
     sampler.Create();
     mag_filter = min_filter = Tegra::Texture::TextureFilter::Linear;
-    wrap_u = wrap_v = Tegra::Texture::WrapMode::Wrap;
+    wrap_u = wrap_v = wrap_p = Tegra::Texture::WrapMode::Wrap;
 
     // default is GL_LINEAR_MIPMAP_LINEAR
     glSamplerParameteri(sampler.handle, GL_TEXTURE_MIN_FILTER, GL_LINEAR);
@@ -586,7 +596,7 @@ void RasterizerOpenGL::SamplerInfo::Create() {
 }
 
 void RasterizerOpenGL::SamplerInfo::SyncWithConfig(const Tegra::Texture::TSCEntry& config) {
-    GLuint s = sampler.handle;
+    const GLuint s = sampler.handle;
 
     if (mag_filter != config.mag_filter) {
         mag_filter = config.mag_filter;
@@ -605,8 +615,13 @@ void RasterizerOpenGL::SamplerInfo::SyncWithConfig(const Tegra::Texture::TSCEntr
         wrap_v = config.wrap_v;
         glSamplerParameteri(s, GL_TEXTURE_WRAP_T, MaxwellToGL::WrapMode(wrap_v));
     }
+    if (wrap_p != config.wrap_p) {
+        wrap_p = config.wrap_p;
+        glSamplerParameteri(s, GL_TEXTURE_WRAP_R, MaxwellToGL::WrapMode(wrap_p));
+    }
 
-    if (wrap_u == Tegra::Texture::WrapMode::Border || wrap_v == Tegra::Texture::WrapMode::Border) {
+    if (wrap_u == Tegra::Texture::WrapMode::Border || wrap_v == Tegra::Texture::WrapMode::Border ||
+        wrap_p == Tegra::Texture::WrapMode::Border) {
         const GLvec4 new_border_color = {{config.border_color_r, config.border_color_g,
                                           config.border_color_b, config.border_color_a}};
         if (border_color != new_border_color) {
@@ -666,8 +681,6 @@ u32 RasterizerOpenGL::SetupConstBuffers(Maxwell::ShaderStage stage, Shader& shad
                               current_bindpoint + bindpoint);
     }
 
-    state.Apply();
-
     return current_bindpoint + static_cast<u32>(entries.size());
 }
 
@@ -682,7 +695,7 @@ u32 RasterizerOpenGL::SetupTextures(Maxwell::ShaderStage stage, Shader& shader,
 
     for (u32 bindpoint = 0; bindpoint < entries.size(); ++bindpoint) {
         const auto& entry = entries[bindpoint];
-        u32 current_bindpoint = current_unit + bindpoint;
+        const u32 current_bindpoint = current_unit + bindpoint;
 
         // Bind the uniform to the sampler.
 
@@ -692,14 +705,15 @@ u32 RasterizerOpenGL::SetupTextures(Maxwell::ShaderStage stage, Shader& shader,
         const auto texture = maxwell3d.GetStageTexture(entry.GetStage(), entry.GetOffset());
 
         if (!texture.enabled) {
-            state.texture_units[current_bindpoint].texture_2d = 0;
+            state.texture_units[current_bindpoint].texture = 0;
             continue;
         }
 
         texture_samplers[current_bindpoint].SyncWithConfig(texture.tsc);
         Surface surface = res_cache.GetTextureSurface(texture);
         if (surface != nullptr) {
-            state.texture_units[current_bindpoint].texture_2d = surface->Texture().handle;
+            state.texture_units[current_bindpoint].texture = surface->Texture().handle;
+            state.texture_units[current_bindpoint].target = surface->Target();
             state.texture_units[current_bindpoint].swizzle.r =
                 MaxwellToGL::SwizzleSource(texture.tic.x_source);
             state.texture_units[current_bindpoint].swizzle.g =
@@ -710,47 +724,19 @@ u32 RasterizerOpenGL::SetupTextures(Maxwell::ShaderStage stage, Shader& shader,
                 MaxwellToGL::SwizzleSource(texture.tic.w_source);
         } else {
             // Can occur when texture addr is null or its memory is unmapped/invalid
-            state.texture_units[current_bindpoint].texture_2d = 0;
+            state.texture_units[current_bindpoint].texture = 0;
         }
     }
 
-    state.Apply();
-
     return current_unit + static_cast<u32>(entries.size());
 }
 
-void RasterizerOpenGL::BindFramebufferSurfaces(const Surface& color_surface,
-                                               const Surface& depth_surface, bool has_stencil) {
-    state.draw.draw_framebuffer = framebuffer.handle;
-    state.Apply();
-
-    glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D,
-                           color_surface != nullptr ? color_surface->Texture().handle : 0, 0);
-    if (depth_surface != nullptr) {
-        if (has_stencil) {
-            // attach both depth and stencil
-            glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, GL_TEXTURE_2D,
-                                   depth_surface->Texture().handle, 0);
-        } else {
-            // attach depth
-            glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_DEPTH_ATTACHMENT, GL_TEXTURE_2D,
-                                   depth_surface->Texture().handle, 0);
-            // clear stencil attachment
-            glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_STENCIL_ATTACHMENT, GL_TEXTURE_2D, 0, 0);
-        }
-    } else {
-        // clear both depth and stencil attachment
-        glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, GL_TEXTURE_2D, 0,
-                               0);
-    }
-}
-
-void RasterizerOpenGL::SyncViewport(const MathUtil::Rectangle<u32>& surfaces_rect) {
+void RasterizerOpenGL::SyncViewport() {
     const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
     const MathUtil::Rectangle<s32> viewport_rect{regs.viewport_transform[0].GetRect()};
 
-    state.viewport.x = static_cast<GLint>(surfaces_rect.left) + viewport_rect.left;
-    state.viewport.y = static_cast<GLint>(surfaces_rect.bottom) + viewport_rect.bottom;
+    state.viewport.x = viewport_rect.left;
+    state.viewport.y = viewport_rect.bottom;
     state.viewport.width = static_cast<GLsizei>(viewport_rect.GetWidth());
     state.viewport.height = static_cast<GLsizei>(viewport_rect.GetHeight());
 }

src/video_core/renderer_opengl/gl_rasterizer.h

@@ -93,17 +93,12 @@ private:
         Tegra::Texture::TextureFilter min_filter;
         Tegra::Texture::WrapMode wrap_u;
         Tegra::Texture::WrapMode wrap_v;
+        Tegra::Texture::WrapMode wrap_p;
         GLvec4 border_color;
     };
 
-    /// Configures the color and depth framebuffer states and returns the dirty <Color, Depth>
-    /// surfaces if writing was enabled.
-    std::pair<Surface, Surface> ConfigureFramebuffers(bool using_color_fb, bool using_depth_fb,
-                                                      bool preserve_contents);
-
-    /// Binds the framebuffer color and depth surface
-    void BindFramebufferSurfaces(const Surface& color_surface, const Surface& depth_surface,
-                                 bool has_stencil);
+    /// Configures the color and depth framebuffer states
+    void ConfigureFramebuffers(bool using_depth_fb, bool preserve_contents);
 
     /*
      * Configures the current constbuffers to use for the draw command.
@@ -126,7 +121,7 @@ private:
                       u32 current_unit);
 
     /// Syncs the viewport to match the guest state
-    void SyncViewport(const MathUtil::Rectangle<u32>& surfaces_rect);
+    void SyncViewport();
 
     /// Syncs the clip enabled status to match the guest state
     void SyncClipEnabled();

src/video_core/renderer_opengl/gl_rasterizer_cache.cpp

@@ -7,6 +7,7 @@
 
 #include "common/alignment.h"
 #include "common/assert.h"
+#include "common/logging/log.h"
 #include "common/microprofile.h"
 #include "common/scope_exit.h"
 #include "core/core.h"
@@ -51,15 +52,17 @@ static VAddr TryGetCpuAddr(Tegra::GPUVAddr gpu_addr) {
     params.type = GetFormatType(params.pixel_format);
     params.width = Common::AlignUp(config.tic.Width(), GetCompressionFactor(params.pixel_format));
     params.height = Common::AlignUp(config.tic.Height(), GetCompressionFactor(params.pixel_format));
+    params.depth = config.tic.Depth();
     params.unaligned_height = config.tic.Height();
     params.size_in_bytes = params.SizeInBytes();
-    params.cache_width = Common::AlignUp(params.width, 16);
-    params.cache_height = Common::AlignUp(params.height, 16);
+    params.cache_width = Common::AlignUp(params.width, 8);
+    params.cache_height = Common::AlignUp(params.height, 8);
+    params.target = SurfaceTargetFromTextureType(config.tic.texture_type);
     return params;
 }
 
-/*static*/ SurfaceParams SurfaceParams::CreateForFramebuffer(
-    const Tegra::Engines::Maxwell3D::Regs::RenderTargetConfig& config) {
+/*static*/ SurfaceParams SurfaceParams::CreateForFramebuffer(size_t index) {
+    const auto& config{Core::System::GetInstance().GPU().Maxwell3D().regs.rt[index]};
     SurfaceParams params{};
     params.addr = TryGetCpuAddr(config.Address());
     params.is_tiled = true;
@@ -69,10 +72,12 @@ static VAddr TryGetCpuAddr(Tegra::GPUVAddr gpu_addr) {
     params.type = GetFormatType(params.pixel_format);
     params.width = config.width;
     params.height = config.height;
+    params.depth = 1;
     params.unaligned_height = config.height;
     params.size_in_bytes = params.SizeInBytes();
-    params.cache_width = Common::AlignUp(params.width, 16);
-    params.cache_height = Common::AlignUp(params.height, 16);
+    params.cache_width = Common::AlignUp(params.width, 8);
+    params.cache_height = Common::AlignUp(params.height, 8);
+    params.target = SurfaceTarget::Texture2D;
     return params;
 }
 
@@ -86,13 +91,14 @@ static VAddr TryGetCpuAddr(Tegra::GPUVAddr gpu_addr) {
     params.pixel_format = PixelFormatFromDepthFormat(format);
     params.component_type = ComponentTypeFromDepthFormat(format);
     params.type = GetFormatType(params.pixel_format);
-    params.size_in_bytes = params.SizeInBytes();
     params.width = zeta_width;
     params.height = zeta_height;
+    params.depth = 1;
     params.unaligned_height = zeta_height;
     params.size_in_bytes = params.SizeInBytes();
-    params.cache_width = Common::AlignUp(params.width, 16);
-    params.cache_height = Common::AlignUp(params.height, 16);
+    params.cache_width = Common::AlignUp(params.width, 8);
+    params.cache_height = Common::AlignUp(params.height, 8);
+    params.target = SurfaceTarget::Texture2D;
     return params;
 }
 
@@ -166,6 +172,26 @@ static constexpr std::array<FormatTuple, SurfaceParams::MaxPixelFormat> tex_form
      ComponentType::Float, false}, // Z32FS8
 }};
 
+static GLenum SurfaceTargetToGL(SurfaceParams::SurfaceTarget target) {
+    switch (target) {
+    case SurfaceParams::SurfaceTarget::Texture1D:
+        return GL_TEXTURE_1D;
+    case SurfaceParams::SurfaceTarget::Texture2D:
+        return GL_TEXTURE_2D;
+    case SurfaceParams::SurfaceTarget::Texture3D:
+        return GL_TEXTURE_3D;
+    case SurfaceParams::SurfaceTarget::Texture1DArray:
+        return GL_TEXTURE_1D_ARRAY;
+    case SurfaceParams::SurfaceTarget::Texture2DArray:
+        return GL_TEXTURE_2D_ARRAY;
+    case SurfaceParams::SurfaceTarget::TextureCubemap:
+        return GL_TEXTURE_CUBE_MAP;
+    }
+    LOG_CRITICAL(Render_OpenGL, "Unimplemented texture target={}", static_cast<u32>(target));
+    UNREACHABLE();
+    return {};
+}
+
 static const FormatTuple& GetFormatTuple(PixelFormat pixel_format, ComponentType component_type) {
     ASSERT(static_cast<size_t>(pixel_format) < tex_format_tuples.size());
     auto& format = tex_format_tuples[static_cast<unsigned int>(pixel_format)];
@@ -220,7 +246,8 @@ 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, VAddr addr) {
+void MortonCopy(u32 stride, u32 block_height, u32 height, u8* gl_buffer, size_t gl_buffer_size,
+                VAddr addr) {
     constexpr u32 bytes_per_pixel = SurfaceParams::GetFormatBpp(format) / CHAR_BIT;
     constexpr u32 gl_bytes_per_pixel = CachedSurface::GetGLBytesPerPixel(format);
 
@@ -230,18 +257,18 @@ void MortonCopy(u32 stride, u32 block_height, u32 height, std::vector<u8>& gl_bu
         const u32 tile_size{IsFormatBCn(format) ? 4U : 1U};
         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);
+        const size_t size_to_copy{std::min(gl_buffer_size, data.size())};
+        memcpy(gl_buffer, data.data(), 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(addr), gl_buffer.data(), morton_to_gl);
+                                       Memory::GetPointer(addr), gl_buffer, morton_to_gl);
     }
 }
 
-static constexpr std::array<void (*)(u32, u32, u32, std::vector<u8>&, VAddr),
+static constexpr std::array<void (*)(u32, u32, u32, u8*, size_t, VAddr),
                             SurfaceParams::MaxPixelFormat>
     morton_to_gl_fns = {
         // clang-format off
@@ -298,7 +325,7 @@ static constexpr std::array<void (*)(u32, u32, u32, std::vector<u8>&, VAddr),
         // clang-format on
 };
 
-static constexpr std::array<void (*)(u32, u32, u32, std::vector<u8>&, VAddr),
+static constexpr std::array<void (*)(u32, u32, u32, u8*, size_t, VAddr),
                             SurfaceParams::MaxPixelFormat>
     gl_to_morton_fns = {
         // clang-format off
@@ -357,33 +384,6 @@ static constexpr std::array<void (*)(u32, u32, u32, std::vector<u8>&, VAddr),
         // clang-format on
 };
 
-// Allocate an uninitialized texture of appropriate size and format for the surface
-static void AllocateSurfaceTexture(GLuint texture, const FormatTuple& format_tuple, u32 width,
-                                   u32 height) {
-    OpenGLState cur_state = OpenGLState::GetCurState();
-
-    // Keep track of previous texture bindings
-    GLuint old_tex = cur_state.texture_units[0].texture_2d;
-    cur_state.texture_units[0].texture_2d = texture;
-    cur_state.Apply();
-    glActiveTexture(GL_TEXTURE0);
-
-    if (!format_tuple.compressed) {
-        // Only pre-create the texture for non-compressed textures.
-        glTexImage2D(GL_TEXTURE_2D, 0, format_tuple.internal_format, width, height, 0,
-                     format_tuple.format, format_tuple.type, nullptr);
-    }
-
-    glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAX_LEVEL, 0);
-    glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_LINEAR);
-    glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE);
-    glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE);
-
-    // Restore previous texture bindings
-    cur_state.texture_units[0].texture_2d = old_tex;
-    cur_state.Apply();
-}
-
 static bool BlitTextures(GLuint src_tex, const MathUtil::Rectangle<u32>& src_rect, GLuint dst_tex,
                          const MathUtil::Rectangle<u32>& dst_rect, SurfaceType type,
                          GLuint read_fb_handle, GLuint draw_fb_handle) {
@@ -438,12 +438,56 @@ static bool BlitTextures(GLuint src_tex, const MathUtil::Rectangle<u32>& src_rec
     return true;
 }
 
-CachedSurface::CachedSurface(const SurfaceParams& params) : params(params) {
+CachedSurface::CachedSurface(const SurfaceParams& params)
+    : params(params), gl_target(SurfaceTargetToGL(params.target)) {
     texture.Create();
     const auto& rect{params.GetRect()};
-    AllocateSurfaceTexture(texture.handle,
-                           GetFormatTuple(params.pixel_format, params.component_type),
-                           rect.GetWidth(), rect.GetHeight());
+
+    // Keep track of previous texture bindings
+    OpenGLState cur_state = OpenGLState::GetCurState();
+    const auto& old_tex = cur_state.texture_units[0];
+    SCOPE_EXIT({
+        cur_state.texture_units[0] = old_tex;
+        cur_state.Apply();
+    });
+
+    cur_state.texture_units[0].texture = texture.handle;
+    cur_state.texture_units[0].target = SurfaceTargetToGL(params.target);
+    cur_state.Apply();
+    glActiveTexture(GL_TEXTURE0);
+
+    const auto& format_tuple = GetFormatTuple(params.pixel_format, params.component_type);
+    if (!format_tuple.compressed) {
+        // Only pre-create the texture for non-compressed textures.
+        switch (params.target) {
+        case SurfaceParams::SurfaceTarget::Texture1D:
+            glTexImage1D(SurfaceTargetToGL(params.target), 0, format_tuple.internal_format,
+                         rect.GetWidth(), 0, format_tuple.format, format_tuple.type, nullptr);
+            break;
+        case SurfaceParams::SurfaceTarget::Texture2D:
+            glTexImage2D(SurfaceTargetToGL(params.target), 0, format_tuple.internal_format,
+                         rect.GetWidth(), rect.GetHeight(), 0, format_tuple.format,
+                         format_tuple.type, nullptr);
+            break;
+        case SurfaceParams::SurfaceTarget::Texture3D:
+        case SurfaceParams::SurfaceTarget::Texture2DArray:
+            glTexImage3D(SurfaceTargetToGL(params.target), 0, format_tuple.internal_format,
+                         rect.GetWidth(), rect.GetHeight(), params.depth, 0, format_tuple.format,
+                         format_tuple.type, nullptr);
+            break;
+        default:
+            LOG_CRITICAL(Render_OpenGL, "Unimplemented surface target={}",
+                         static_cast<u32>(params.target));
+            UNREACHABLE();
+            glTexImage2D(GL_TEXTURE_2D, 0, format_tuple.internal_format, rect.GetWidth(),
+                         rect.GetHeight(), 0, format_tuple.format, format_tuple.type, nullptr);
+        }
+    }
+
+    glTexParameteri(SurfaceTargetToGL(params.target), GL_TEXTURE_MAX_LEVEL, 0);
+    glTexParameteri(SurfaceTargetToGL(params.target), GL_TEXTURE_MIN_FILTER, GL_LINEAR);
+    glTexParameteri(SurfaceTargetToGL(params.target), GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE);
+    glTexParameteri(SurfaceTargetToGL(params.target), GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE);
 }
 
 static void ConvertS8Z24ToZ24S8(std::vector<u8>& data, u32 width, u32 height) {
@@ -461,7 +505,7 @@ static void ConvertS8Z24ToZ24S8(std::vector<u8>& data, u32 width, u32 height) {
 
     S8Z24 input_pixel{};
     Z24S8 output_pixel{};
-    const auto bpp{CachedSurface::GetGLBytesPerPixel(PixelFormat::S8Z24)};
+    constexpr auto bpp{CachedSurface::GetGLBytesPerPixel(PixelFormat::S8Z24)};
     for (size_t y = 0; y < height; ++y) {
         for (size_t x = 0; x < width; ++x) {
             const size_t offset{bpp * (y * width + x)};
@@ -474,7 +518,7 @@ 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::G8R8U)};
+    constexpr 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)};
@@ -514,23 +558,6 @@ static void ConvertFormatAsNeeded_LoadGLBuffer(std::vector<u8>& data, PixelForma
     }
 }
 
-/**
- * Helper function to perform software conversion (as needed) when flushing a buffer to Switch
- * memory. This is for Maxwell pixel formats that cannot be represented as-is in OpenGL or with
- * typical desktop GPUs.
- */
-static void ConvertFormatAsNeeded_FlushGLBuffer(std::vector<u8>& /*data*/, PixelFormat pixel_format,
-                                                u32 /*width*/, u32 /*height*/) {
-    switch (pixel_format) {
-    case PixelFormat::ASTC_2D_4X4:
-    case PixelFormat::S8Z24:
-        LOG_CRITICAL(Render_OpenGL, "Unimplemented pixel_format={}",
-                     static_cast<u32>(pixel_format));
-        UNREACHABLE();
-        break;
-    }
-}
-
 MICROPROFILE_DEFINE(OpenGL_SurfaceLoad, "OpenGL", "Surface Load", MP_RGB(128, 64, 192));
 void CachedSurface::LoadGLBuffer() {
     ASSERT(params.type != SurfaceType::Fill);
@@ -545,13 +572,25 @@ void CachedSurface::LoadGLBuffer() {
     MICROPROFILE_SCOPE(OpenGL_SurfaceLoad);
 
     if (params.is_tiled) {
-        gl_buffer.resize(copy_size);
+        // TODO(bunnei): This only unswizzles and copies a 2D texture - we do not yet know how to do
+        // this for 3D textures, etc.
+        switch (params.target) {
+        case SurfaceParams::SurfaceTarget::Texture2D:
+            // Pass impl. to the fallback code below
+            break;
+        default:
+            LOG_CRITICAL(HW_GPU, "Unimplemented tiled load for target={}",
+                         static_cast<u32>(params.target));
+            UNREACHABLE();
+        }
 
+        gl_buffer.resize(static_cast<size_t>(params.depth) * 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.data(), copy_size,
+            params.addr);
     } else {
-        const u8* const texture_src_data_end = texture_src_data + copy_size;
-
+        const u8* const texture_src_data_end{texture_src_data +
+                                             (static_cast<size_t>(params.depth) * copy_size)};
         gl_buffer.assign(texture_src_data, texture_src_data_end);
     }
 
@@ -560,23 +599,7 @@ 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.addr);
-
-    ASSERT(dst_buffer);
-    ASSERT(gl_buffer.size() ==
-           params.width * params.height * GetGLBytesPerPixel(params.pixel_format));
-
-    MICROPROFILE_SCOPE(OpenGL_SurfaceFlush);
-
-    ConvertFormatAsNeeded_FlushGLBuffer(gl_buffer, params.pixel_format, params.width,
-                                        params.height);
-
-    if (!params.is_tiled) {
-        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);
-    }
+    ASSERT_MSG(false, "Unimplemented");
 }
 
 MICROPROFILE_DEFINE(OpenGL_TextureUL, "OpenGL", "Texture Upload", MP_RGB(128, 64, 192));
@@ -586,22 +609,29 @@ void CachedSurface::UploadGLTexture(GLuint read_fb_handle, GLuint draw_fb_handle
 
     MICROPROFILE_SCOPE(OpenGL_TextureUL);
 
-    ASSERT(gl_buffer.size() ==
-           params.width * params.height * GetGLBytesPerPixel(params.pixel_format));
+    ASSERT(gl_buffer.size() == static_cast<size_t>(params.width) * params.height *
+                                   GetGLBytesPerPixel(params.pixel_format) * params.depth);
 
     const auto& rect{params.GetRect()};
 
     // Load data from memory to the surface
-    GLint x0 = static_cast<GLint>(rect.left);
-    GLint y0 = static_cast<GLint>(rect.bottom);
-    size_t buffer_offset = (y0 * params.width + x0) * GetGLBytesPerPixel(params.pixel_format);
+    const GLint x0 = static_cast<GLint>(rect.left);
+    const GLint y0 = static_cast<GLint>(rect.bottom);
+    const size_t buffer_offset =
+        static_cast<size_t>(static_cast<size_t>(y0) * params.width + static_cast<size_t>(x0)) *
+        GetGLBytesPerPixel(params.pixel_format);
 
     const FormatTuple& tuple = GetFormatTuple(params.pixel_format, params.component_type);
-    GLuint target_tex = texture.handle;
+    const GLuint target_tex = texture.handle;
     OpenGLState cur_state = OpenGLState::GetCurState();
 
-    GLuint old_tex = cur_state.texture_units[0].texture_2d;
-    cur_state.texture_units[0].texture_2d = target_tex;
+    const auto& old_tex = cur_state.texture_units[0];
+    SCOPE_EXIT({
+        cur_state.texture_units[0] = old_tex;
+        cur_state.Apply();
+    });
+    cur_state.texture_units[0].texture = target_tex;
+    cur_state.texture_units[0].target = SurfaceTargetToGL(params.target);
     cur_state.Apply();
 
     // Ensure no bad interactions with GL_UNPACK_ALIGNMENT
@@ -610,136 +640,102 @@ void CachedSurface::UploadGLTexture(GLuint read_fb_handle, GLuint draw_fb_handle
 
     glActiveTexture(GL_TEXTURE0);
     if (tuple.compressed) {
-        glCompressedTexImage2D(
-            GL_TEXTURE_2D, 0, tuple.internal_format, static_cast<GLsizei>(params.width),
-            static_cast<GLsizei>(params.height), 0, static_cast<GLsizei>(params.size_in_bytes),
-            &gl_buffer[buffer_offset]);
+        switch (params.target) {
+        case SurfaceParams::SurfaceTarget::Texture2D:
+            glCompressedTexImage2D(
+                SurfaceTargetToGL(params.target), 0, tuple.internal_format,
+                static_cast<GLsizei>(params.width), static_cast<GLsizei>(params.height), 0,
+                static_cast<GLsizei>(params.size_in_bytes), &gl_buffer[buffer_offset]);
+            break;
+        case SurfaceParams::SurfaceTarget::Texture3D:
+        case SurfaceParams::SurfaceTarget::Texture2DArray:
+            glCompressedTexImage3D(
+                SurfaceTargetToGL(params.target), 0, tuple.internal_format,
+                static_cast<GLsizei>(params.width), static_cast<GLsizei>(params.height),
+                static_cast<GLsizei>(params.depth), 0, static_cast<GLsizei>(params.size_in_bytes),
+                &gl_buffer[buffer_offset]);
+            break;
+        default:
+            LOG_CRITICAL(Render_OpenGL, "Unimplemented surface target={}",
+                         static_cast<u32>(params.target));
+            UNREACHABLE();
+            glCompressedTexImage2D(
+                GL_TEXTURE_2D, 0, tuple.internal_format, static_cast<GLsizei>(params.width),
+                static_cast<GLsizei>(params.height), 0, static_cast<GLsizei>(params.size_in_bytes),
+                &gl_buffer[buffer_offset]);
+        }
     } else {
-        glTexSubImage2D(GL_TEXTURE_2D, 0, x0, y0, static_cast<GLsizei>(rect.GetWidth()),
-                        static_cast<GLsizei>(rect.GetHeight()), tuple.format, tuple.type,
-                        &gl_buffer[buffer_offset]);
+
+        switch (params.target) {
+        case SurfaceParams::SurfaceTarget::Texture1D:
+            glTexSubImage1D(SurfaceTargetToGL(params.target), 0, x0,
+                            static_cast<GLsizei>(rect.GetWidth()), tuple.format, tuple.type,
+                            &gl_buffer[buffer_offset]);
+            break;
+        case SurfaceParams::SurfaceTarget::Texture2D:
+            glTexSubImage2D(SurfaceTargetToGL(params.target), 0, x0, y0,
+                            static_cast<GLsizei>(rect.GetWidth()),
+                            static_cast<GLsizei>(rect.GetHeight()), tuple.format, tuple.type,
+                            &gl_buffer[buffer_offset]);
+            break;
+        case SurfaceParams::SurfaceTarget::Texture3D:
+        case SurfaceParams::SurfaceTarget::Texture2DArray:
+            glTexSubImage3D(SurfaceTargetToGL(params.target), 0, x0, y0, 0,
+                            static_cast<GLsizei>(rect.GetWidth()),
+                            static_cast<GLsizei>(rect.GetHeight()), params.depth, tuple.format,
+                            tuple.type, &gl_buffer[buffer_offset]);
+            break;
+        default:
+            LOG_CRITICAL(Render_OpenGL, "Unimplemented surface target={}",
+                         static_cast<u32>(params.target));
+            UNREACHABLE();
+            glTexSubImage2D(GL_TEXTURE_2D, 0, x0, y0, static_cast<GLsizei>(rect.GetWidth()),
+                            static_cast<GLsizei>(rect.GetHeight()), tuple.format, tuple.type,
+                            &gl_buffer[buffer_offset]);
+        }
     }
 
     glPixelStorei(GL_UNPACK_ROW_LENGTH, 0);
-
-    cur_state.texture_units[0].texture_2d = old_tex;
-    cur_state.Apply();
-}
-
-MICROPROFILE_DEFINE(OpenGL_TextureDL, "OpenGL", "Texture Download", MP_RGB(128, 192, 64));
-void CachedSurface::DownloadGLTexture(GLuint read_fb_handle, GLuint draw_fb_handle) {
-    if (params.type == SurfaceType::Fill)
-        return;
-
-    MICROPROFILE_SCOPE(OpenGL_TextureDL);
-
-    gl_buffer.resize(params.width * params.height * GetGLBytesPerPixel(params.pixel_format));
-
-    OpenGLState state = OpenGLState::GetCurState();
-    OpenGLState prev_state = state;
-    SCOPE_EXIT({ prev_state.Apply(); });
-
-    const FormatTuple& tuple = GetFormatTuple(params.pixel_format, params.component_type);
-
-    // Ensure no bad interactions with GL_PACK_ALIGNMENT
-    ASSERT(params.width * GetGLBytesPerPixel(params.pixel_format) % 4 == 0);
-    glPixelStorei(GL_PACK_ROW_LENGTH, static_cast<GLint>(params.width));
-
-    const auto& rect{params.GetRect()};
-    size_t buffer_offset =
-        (rect.bottom * params.width + rect.left) * GetGLBytesPerPixel(params.pixel_format);
-
-    state.UnbindTexture(texture.handle);
-    state.draw.read_framebuffer = read_fb_handle;
-    state.Apply();
-
-    if (params.type == SurfaceType::ColorTexture) {
-        glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D,
-                               texture.handle, 0);
-        glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, GL_TEXTURE_2D, 0,
-                               0);
-    } else if (params.type == SurfaceType::Depth) {
-        glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, 0, 0);
-        glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_DEPTH_ATTACHMENT, GL_TEXTURE_2D,
-                               texture.handle, 0);
-        glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_STENCIL_ATTACHMENT, GL_TEXTURE_2D, 0, 0);
-    } else {
-        glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, 0, 0);
-        glFramebufferTexture2D(GL_READ_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, GL_TEXTURE_2D,
-                               texture.handle, 0);
-    }
-    glReadPixels(static_cast<GLint>(rect.left), static_cast<GLint>(rect.bottom),
-                 static_cast<GLsizei>(rect.GetWidth()), static_cast<GLsizei>(rect.GetHeight()),
-                 tuple.format, tuple.type, &gl_buffer[buffer_offset]);
-
-    glPixelStorei(GL_PACK_ROW_LENGTH, 0);
 }
 
 RasterizerCacheOpenGL::RasterizerCacheOpenGL() {
     read_framebuffer.Create();
     draw_framebuffer.Create();
+    copy_pbo.Create();
 }
 
 Surface RasterizerCacheOpenGL::GetTextureSurface(const Tegra::Texture::FullTextureInfo& config) {
     return GetSurface(SurfaceParams::CreateForTexture(config));
 }
 
-SurfaceSurfaceRect_Tuple RasterizerCacheOpenGL::GetFramebufferSurfaces(bool using_color_fb,
-                                                                       bool using_depth_fb,
-                                                                       bool preserve_contents) {
-    const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
-
-    // TODO(bunnei): This is hard corded to use just the first render buffer
-    LOG_TRACE(Render_OpenGL, "hard-coded for render target 0!");
-
-    // get color and depth surfaces
-    SurfaceParams color_params{};
-    SurfaceParams depth_params{};
-
-    if (using_color_fb) {
-        color_params = SurfaceParams::CreateForFramebuffer(regs.rt[0]);
+Surface RasterizerCacheOpenGL::GetDepthBufferSurface(bool preserve_contents) {
+    const auto& regs{Core::System::GetInstance().GPU().Maxwell3D().regs};
+    if (!regs.zeta.Address() || !regs.zeta_enable) {
+        return {};
     }
 
-    if (using_depth_fb) {
-        depth_params = SurfaceParams::CreateForDepthBuffer(regs.zeta_width, regs.zeta_height,
-                                                           regs.zeta.Address(), regs.zeta.format);
+    SurfaceParams depth_params{SurfaceParams::CreateForDepthBuffer(
+        regs.zeta_width, regs.zeta_height, regs.zeta.Address(), regs.zeta.format)};
+
+    return GetSurface(depth_params, preserve_contents);
+}
+
+Surface RasterizerCacheOpenGL::GetColorBufferSurface(size_t index, bool preserve_contents) {
+    const auto& regs{Core::System::GetInstance().GPU().Maxwell3D().regs};
+
+    ASSERT(index < Tegra::Engines::Maxwell3D::Regs::NumRenderTargets);
+
+    if (index >= regs.rt_control.count) {
+        return {};
     }
 
-    MathUtil::Rectangle<u32> color_rect{};
-    Surface color_surface;
-    if (using_color_fb) {
-        color_surface = GetSurface(color_params, preserve_contents);
-        if (color_surface) {
-            color_rect = color_surface->GetSurfaceParams().GetRect();
-        }
+    if (regs.rt[index].Address() == 0 || regs.rt[index].format == Tegra::RenderTargetFormat::NONE) {
+        return {};
     }
 
-    MathUtil::Rectangle<u32> depth_rect{};
-    Surface depth_surface;
-    if (using_depth_fb) {
-        depth_surface = GetSurface(depth_params, preserve_contents);
-        if (depth_surface) {
-            depth_rect = depth_surface->GetSurfaceParams().GetRect();
-        }
-    }
+    const SurfaceParams color_params{SurfaceParams::CreateForFramebuffer(index)};
 
-    MathUtil::Rectangle<u32> fb_rect{};
-    if (color_surface && depth_surface) {
-        fb_rect = color_rect;
-        // Color and Depth surfaces must have the same dimensions and offsets
-        if (color_rect.bottom != depth_rect.bottom || color_rect.top != depth_rect.top ||
-            color_rect.left != depth_rect.left || color_rect.right != depth_rect.right) {
-            color_surface = GetSurface(color_params);
-            depth_surface = GetSurface(depth_params);
-            fb_rect = color_surface->GetSurfaceParams().GetRect();
-        }
-    } else if (color_surface) {
-        fb_rect = color_rect;
-    } else if (depth_surface) {
-        fb_rect = depth_rect;
-    }
-
-    return std::make_tuple(color_surface, depth_surface, fb_rect);
+    return GetSurface(color_params, preserve_contents);
 }
 
 void RasterizerCacheOpenGL::LoadSurface(const Surface& surface) {
@@ -748,7 +744,6 @@ void RasterizerCacheOpenGL::LoadSurface(const Surface& surface) {
 }
 
 void RasterizerCacheOpenGL::FlushSurface(const Surface& surface) {
-    surface->DownloadGLTexture(read_framebuffer.handle, draw_framebuffer.handle);
     surface->FlushGLBuffer();
 }
 
@@ -809,8 +804,8 @@ Surface RasterizerCacheOpenGL::RecreateSurface(const Surface& surface,
     // If format is unchanged, we can do a faster blit without reinterpreting pixel data
     if (params.pixel_format == new_params.pixel_format) {
         BlitTextures(surface->Texture().handle, params.GetRect(), new_surface->Texture().handle,
-                     new_surface->GetSurfaceParams().GetRect(), params.type,
-                     read_framebuffer.handle, draw_framebuffer.handle);
+                     params.GetRect(), params.type, read_framebuffer.handle,
+                     draw_framebuffer.handle);
         return new_surface;
     }
 
@@ -821,12 +816,7 @@ Surface RasterizerCacheOpenGL::RecreateSurface(const Surface& surface,
 
         size_t buffer_size = std::max(params.SizeInBytes(), new_params.SizeInBytes());
 
-        // Use a Pixel Buffer Object to download the previous texture and then upload it to the new
-        // one using the new format.
-        OGLBuffer pbo;
-        pbo.Create();
-
-        glBindBuffer(GL_PIXEL_PACK_BUFFER, pbo.handle);
+        glBindBuffer(GL_PIXEL_PACK_BUFFER, copy_pbo.handle);
         glBufferData(GL_PIXEL_PACK_BUFFER, buffer_size, nullptr, GL_STREAM_DRAW_ARB);
         if (source_format.compressed) {
             glGetCompressedTextureImage(surface->Texture().handle, 0,
@@ -845,8 +835,8 @@ Surface RasterizerCacheOpenGL::RecreateSurface(const Surface& surface,
                 // of the data in this case. Games like Super Mario Odyssey seem to hit this case
                 // when drawing, it re-uses the memory of a previous texture as a bigger framebuffer
                 // but it doesn't clear it beforehand, the texture is already full of zeros.
-                LOG_CRITICAL(HW_GPU, "Trying to upload extra texture data from the CPU during "
-                                     "reinterpretation but the texture is tiled.");
+                LOG_DEBUG(HW_GPU, "Trying to upload extra texture data from the CPU during "
+                                  "reinterpretation but the texture is tiled.");
             }
             size_t remaining_size = new_params.SizeInBytes() - params.SizeInBytes();
             std::vector<u8> data(remaining_size);
@@ -859,21 +849,38 @@ Surface RasterizerCacheOpenGL::RecreateSurface(const Surface& surface,
 
         const auto& dest_rect{new_params.GetRect()};
 
-        glBindBuffer(GL_PIXEL_UNPACK_BUFFER, pbo.handle);
+        glBindBuffer(GL_PIXEL_UNPACK_BUFFER, copy_pbo.handle);
         if (dest_format.compressed) {
-            glCompressedTexSubImage2D(
-                GL_TEXTURE_2D, 0, 0, 0, static_cast<GLsizei>(dest_rect.GetWidth()),
-                static_cast<GLsizei>(dest_rect.GetHeight()), dest_format.format,
-                static_cast<GLsizei>(new_params.SizeInBytes()), nullptr);
+            LOG_CRITICAL(HW_GPU, "Compressed copy is unimplemented!");
+            UNREACHABLE();
         } 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);
+            switch (new_params.target) {
+            case SurfaceParams::SurfaceTarget::Texture1D:
+                glTextureSubImage1D(new_surface->Texture().handle, 0, 0,
+                                    static_cast<GLsizei>(dest_rect.GetWidth()), dest_format.format,
+                                    dest_format.type, nullptr);
+                break;
+            case SurfaceParams::SurfaceTarget::Texture2D:
+                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);
+                break;
+            case SurfaceParams::SurfaceTarget::Texture3D:
+            case SurfaceParams::SurfaceTarget::Texture2DArray:
+                glTextureSubImage3D(new_surface->Texture().handle, 0, 0, 0, 0,
+                                    static_cast<GLsizei>(dest_rect.GetWidth()),
+                                    static_cast<GLsizei>(dest_rect.GetHeight()),
+                                    static_cast<GLsizei>(new_params.depth), dest_format.format,
+                                    dest_format.type, nullptr);
+                break;
+            default:
+                LOG_CRITICAL(Render_OpenGL, "Unimplemented surface target={}",
+                             static_cast<u32>(params.target));
+                UNREACHABLE();
+            }
         }
         glBindBuffer(GL_PIXEL_UNPACK_BUFFER, 0);
-
-        pbo.Release();
     }
 
     return new_surface;

src/video_core/renderer_opengl/gl_rasterizer_cache.h

@@ -109,6 +109,33 @@ struct SurfaceParams {
         Invalid = 4,
     };
 
+    enum class SurfaceTarget {
+        Texture1D,
+        Texture2D,
+        Texture3D,
+        Texture1DArray,
+        Texture2DArray,
+        TextureCubemap,
+    };
+
+    static SurfaceTarget SurfaceTargetFromTextureType(Tegra::Texture::TextureType texture_type) {
+        switch (texture_type) {
+        case Tegra::Texture::TextureType::Texture1D:
+            return SurfaceTarget::Texture1D;
+        case Tegra::Texture::TextureType::Texture2D:
+        case Tegra::Texture::TextureType::Texture2DNoMipmap:
+            return SurfaceTarget::Texture2D;
+        case Tegra::Texture::TextureType::Texture1DArray:
+            return SurfaceTarget::Texture1DArray;
+        case Tegra::Texture::TextureType::Texture2DArray:
+            return SurfaceTarget::Texture2DArray;
+        default:
+            LOG_CRITICAL(HW_GPU, "Unimplemented texture_type={}", static_cast<u32>(texture_type));
+            UNREACHABLE();
+            return SurfaceTarget::Texture2D;
+        }
+    }
+
     /**
      * Gets the compression factor for the specified PixelFormat. This applies to just the
      * "compressed width" and "compressed height", not the overall compression factor of a
@@ -635,15 +662,14 @@ struct SurfaceParams {
         ASSERT(width % compression_factor == 0);
         ASSERT(height % compression_factor == 0);
         return (width / compression_factor) * (height / compression_factor) *
-               GetFormatBpp(pixel_format) / CHAR_BIT;
+               GetFormatBpp(pixel_format) * depth / CHAR_BIT;
     }
 
     /// Creates SurfaceParams from a texture configuration
     static SurfaceParams CreateForTexture(const Tegra::Texture::FullTextureInfo& config);
 
     /// Creates SurfaceParams from a framebuffer configuration
-    static SurfaceParams CreateForFramebuffer(
-        const Tegra::Engines::Maxwell3D::Regs::RenderTargetConfig& config);
+    static SurfaceParams CreateForFramebuffer(size_t index);
 
     /// Creates SurfaceParams for a depth buffer configuration
     static SurfaceParams CreateForDepthBuffer(u32 zeta_width, u32 zeta_height,
@@ -664,8 +690,10 @@ struct SurfaceParams {
     SurfaceType type;
     u32 width;
     u32 height;
+    u32 depth;
     u32 unaligned_height;
     size_t size_in_bytes;
+    SurfaceTarget target;
 
     // Parameters used for caching only
     u32 cache_width;
@@ -709,6 +737,10 @@ public:
         return texture;
     }
 
+    GLenum Target() const {
+        return gl_target;
+    }
+
     static constexpr unsigned int GetGLBytesPerPixel(SurfaceParams::PixelFormat format) {
         if (format == SurfaceParams::PixelFormat::Invalid)
             return 0;
@@ -724,14 +756,14 @@ public:
     void LoadGLBuffer();
     void FlushGLBuffer();
 
-    // Upload/Download data in gl_buffer in/to this surface's texture
+    // Upload data in gl_buffer to this surface's texture
     void UploadGLTexture(GLuint read_fb_handle, GLuint draw_fb_handle);
-    void DownloadGLTexture(GLuint read_fb_handle, GLuint draw_fb_handle);
 
 private:
     OGLTexture texture;
     std::vector<u8> gl_buffer;
     SurfaceParams params;
+    GLenum gl_target;
 };
 
 class RasterizerCacheOpenGL final : public RasterizerCache<Surface> {
@@ -741,9 +773,11 @@ public:
     /// Get a surface based on the texture configuration
     Surface GetTextureSurface(const Tegra::Texture::FullTextureInfo& config);
 
-    /// Get the color and depth surfaces based on the framebuffer configuration
-    SurfaceSurfaceRect_Tuple GetFramebufferSurfaces(bool using_color_fb, bool using_depth_fb,
-                                                    bool preserve_contents);
+    /// Get the depth surface based on the framebuffer configuration
+    Surface GetDepthBufferSurface(bool preserve_contents);
+
+    /// Get the color surface based on the framebuffer configuration and the specified render target
+    Surface GetColorBufferSurface(size_t index, bool preserve_contents);
 
     /// Flushes the surface to Switch memory
     void FlushSurface(const Surface& surface);
@@ -774,6 +808,10 @@ private:
 
     OGLFramebuffer read_framebuffer;
     OGLFramebuffer draw_framebuffer;
+
+    /// Use a Pixel Buffer Object to download the previous texture and then upload it to the new one
+    /// using the new format.
+    OGLBuffer copy_pbo;
 };
 
 } // namespace OpenGL

src/video_core/renderer_opengl/gl_shader_cache.cpp

@@ -13,8 +13,8 @@ namespace OpenGL {
 
 /// Gets the address for the specified shader stage program
 static VAddr GetShaderAddress(Maxwell::ShaderProgram program) {
-    auto& gpu = Core::System::GetInstance().GPU().Maxwell3D();
-    auto& shader_config = gpu.regs.shader_config[static_cast<size_t>(program)];
+    const auto& gpu = Core::System::GetInstance().GPU().Maxwell3D();
+    const auto& shader_config = gpu.regs.shader_config[static_cast<size_t>(program)];
     return *gpu.memory_manager.GpuToCpuAddress(gpu.regs.code_address.CodeAddress() +
                                                shader_config.offset);
 }
@@ -86,7 +86,7 @@ CachedShader::CachedShader(VAddr addr, Maxwell::ShaderProgram program_type)
 }
 
 GLuint CachedShader::GetProgramResourceIndex(const GLShader::ConstBufferEntry& buffer) {
-    auto search{resource_cache.find(buffer.GetHash())};
+    const auto search{resource_cache.find(buffer.GetHash())};
     if (search == resource_cache.end()) {
         const GLuint index{
             glGetProgramResourceIndex(program.handle, GL_UNIFORM_BLOCK, buffer.GetName().c_str())};
@@ -98,7 +98,7 @@ GLuint CachedShader::GetProgramResourceIndex(const GLShader::ConstBufferEntry& b
 }
 
 GLint CachedShader::GetUniformLocation(const GLShader::SamplerEntry& sampler) {
-    auto search{uniform_cache.find(sampler.GetHash())};
+    const auto search{uniform_cache.find(sampler.GetHash())};
     if (search == uniform_cache.end()) {
         const GLint index{glGetUniformLocation(program.handle, sampler.GetName().c_str())};
         uniform_cache[sampler.GetHash()] = index;

src/video_core/renderer_opengl/gl_shader_decompiler.cpp

@@ -113,7 +113,7 @@ private:
 
     /// Scans a range of code for labels and determines the exit method.
     ExitMethod Scan(u32 begin, u32 end, std::set<u32>& labels) {
-        auto [iter, inserted] =
+        const auto [iter, inserted] =
             exit_method_map.emplace(std::make_pair(begin, end), ExitMethod::Undetermined);
         ExitMethod& exit_method = iter->second;
         if (!inserted)
@@ -131,22 +131,22 @@ private:
                     if (instr.pred.pred_index == static_cast<u64>(Pred::UnusedIndex)) {
                         return exit_method = ExitMethod::AlwaysEnd;
                     } else {
-                        ExitMethod not_met = Scan(offset + 1, end, labels);
+                        const ExitMethod not_met = Scan(offset + 1, end, labels);
                         return exit_method = ParallelExit(ExitMethod::AlwaysEnd, not_met);
                     }
                 }
                 case OpCode::Id::BRA: {
-                    u32 target = offset + instr.bra.GetBranchTarget();
+                    const u32 target = offset + instr.bra.GetBranchTarget();
                     labels.insert(target);
-                    ExitMethod no_jmp = Scan(offset + 1, end, labels);
-                    ExitMethod jmp = Scan(target, end, labels);
+                    const ExitMethod no_jmp = Scan(offset + 1, end, labels);
+                    const ExitMethod jmp = Scan(target, end, labels);
                     return exit_method = ParallelExit(no_jmp, jmp);
                 }
                 case OpCode::Id::SSY: {
                     // The SSY instruction uses a similar encoding as the BRA instruction.
                     ASSERT_MSG(instr.bra.constant_buffer == 0,
                                "Constant buffer SSY is not supported");
-                    u32 target = offset + instr.bra.GetBranchTarget();
+                    const u32 target = offset + instr.bra.GetBranchTarget();
                     labels.insert(target);
                     // Continue scanning for an exit method.
                     break;
@@ -346,8 +346,8 @@ public:
      */
     void SetRegisterToInputAttibute(const Register& reg, u64 elem, Attribute::Index attribute,
                                     const Tegra::Shader::IpaMode& input_mode) {
-        std::string dest = GetRegisterAsFloat(reg);
-        std::string src = GetInputAttribute(attribute, input_mode) + GetSwizzle(elem);
+        const std::string dest = GetRegisterAsFloat(reg);
+        const std::string src = GetInputAttribute(attribute, input_mode) + GetSwizzle(elem);
         shader.AddLine(dest + " = " + src + ';');
     }
 
@@ -359,8 +359,8 @@ public:
      * @param reg The register to use as the source value.
      */
     void SetOutputAttributeToRegister(Attribute::Index attribute, u64 elem, const Register& reg) {
-        std::string dest = GetOutputAttribute(attribute);
-        std::string src = GetRegisterAsFloat(reg);
+        const std::string dest = GetOutputAttribute(attribute);
+        const std::string src = GetRegisterAsFloat(reg);
 
         if (!dest.empty()) {
             // Can happen with unknown/unimplemented output attributes, in which case we ignore the
@@ -393,9 +393,9 @@ public:
                                    GLSLRegister::Type type) {
         declr_const_buffers[cbuf_index].MarkAsUsedIndirect(cbuf_index, stage);
 
-        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]";
+        const std::string final_offset = fmt::format("({} + {})", index_str, offset / 4);
+        const std::string value = 'c' + std::to_string(cbuf_index) + '[' + final_offset + " / 4][" +
+                                  final_offset + " % 4]";
 
         if (type == GLSLRegister::Type::Float) {
             return value;
@@ -443,13 +443,12 @@ public:
         }
         declarations.AddNewLine();
 
-        // Append the sampler2D array for the used textures.
-        const size_t num_samplers = used_samplers.size();
-        if (num_samplers > 0) {
-            declarations.AddLine("uniform sampler2D " + SamplerEntry::GetArrayName(stage) + '[' +
-                                 std::to_string(num_samplers) + "];");
-            declarations.AddNewLine();
+        const auto& samplers = GetSamplers();
+        for (const auto& sampler : samplers) {
+            declarations.AddLine("uniform " + sampler.GetTypeString() + ' ' + sampler.GetName() +
+                                 ';');
         }
+        declarations.AddNewLine();
     }
 
     /// Returns a list of constant buffer declarations
@@ -461,27 +460,29 @@ public:
     }
 
     /// Returns a list of samplers used in the shader
-    std::vector<SamplerEntry> GetSamplers() const {
+    const std::vector<SamplerEntry>& GetSamplers() const {
         return used_samplers;
     }
 
     /// Returns the GLSL sampler used for the input shader sampler, and creates a new one if
     /// necessary.
-    std::string AccessSampler(const Sampler& sampler) {
-        size_t offset = static_cast<size_t>(sampler.index.Value());
+    std::string AccessSampler(const Sampler& sampler, Tegra::Shader::TextureType type,
+                              bool is_array) {
+        const size_t offset = static_cast<size_t>(sampler.index.Value());
 
         // If this sampler has already been used, return the existing mapping.
-        auto itr =
+        const auto itr =
             std::find_if(used_samplers.begin(), used_samplers.end(),
                          [&](const SamplerEntry& entry) { return entry.GetOffset() == offset; });
 
         if (itr != used_samplers.end()) {
+            ASSERT(itr->GetType() == type && itr->IsArray() == is_array);
             return itr->GetName();
         }
 
         // Otherwise create a new mapping for this sampler
-        size_t next_index = used_samplers.size();
-        SamplerEntry entry{stage, offset, next_index};
+        const size_t next_index = used_samplers.size();
+        const SamplerEntry entry{stage, offset, next_index, type, is_array};
         used_samplers.emplace_back(entry);
         return entry.GetName();
     }
@@ -698,7 +699,7 @@ private:
         };
 
         bool IsColorComponentOutputEnabled(u32 render_target, u32 component) const {
-            u32 bit = render_target * 4 + component;
+            const u32 bit = render_target * 4 + component;
             return enabled_color_outputs & (1 << bit);
         }
     };
@@ -706,7 +707,7 @@ private:
 
     /// Gets the Subroutine object corresponding to the specified address.
     const Subroutine& GetSubroutine(u32 begin, u32 end) const {
-        auto iter = subroutines.find(Subroutine{begin, end, suffix});
+        const auto iter = subroutines.find(Subroutine{begin, end, suffix});
         ASSERT(iter != subroutines.end());
         return *iter;
     }
@@ -722,8 +723,8 @@ private:
     }
 
     /// Generates code representing a texture sampler.
-    std::string GetSampler(const Sampler& sampler) {
-        return regs.AccessSampler(sampler);
+    std::string GetSampler(const Sampler& sampler, Tegra::Shader::TextureType type, bool is_array) {
+        return regs.AccessSampler(sampler, type, is_array);
     }
 
     /**
@@ -751,7 +752,7 @@ private:
         // Can't assign to the constant predicate.
         ASSERT(pred != static_cast<u64>(Pred::UnusedIndex));
 
-        std::string variable = 'p' + std::to_string(pred) + '_' + suffix;
+        const std::string variable = 'p' + std::to_string(pred) + '_' + suffix;
         shader.AddLine(variable + " = " + value + ';');
         declr_predicates.insert(std::move(variable));
     }
@@ -1032,7 +1033,11 @@ private:
         if (header.writes_depth) {
             // The depth output is always 2 registers after the last color output, and current_reg
             // already contains one past the last color register.
-            shader.AddLine("gl_FragDepth = " + regs.GetRegisterAsFloat(current_reg + 1) + ';');
+
+            shader.AddLine(
+                "gl_FragDepth = " +
+                regs.GetRegisterAsFloat(static_cast<Tegra::Shader::Register>(current_reg) + 1) +
+                ';');
         }
     }
 
@@ -1434,7 +1439,7 @@ private:
                 if (instr.alu_integer.negate_b)
                     op_b = "-(" + op_b + ')';
 
-                std::string shift = std::to_string(instr.alu_integer.shift_amount.Value());
+                const std::string shift = std::to_string(instr.alu_integer.shift_amount.Value());
 
                 regs.SetRegisterToInteger(instr.gpr0, true, 0,
                                           "((" + op_a + " << " + shift + ") + " + op_b + ')', 1, 1);
@@ -1452,7 +1457,7 @@ private:
             case OpCode::Id::SEL_C:
             case OpCode::Id::SEL_R:
             case OpCode::Id::SEL_IMM: {
-                std::string condition =
+                const std::string condition =
                     GetPredicateCondition(instr.sel.pred, instr.sel.neg_pred != 0);
                 regs.SetRegisterToInteger(instr.gpr0, true, 0,
                                           '(' + condition + ") ? " + op_a + " : " + op_b, 1, 1);
@@ -1474,8 +1479,9 @@ private:
             case OpCode::Id::LOP3_C:
             case OpCode::Id::LOP3_R:
             case OpCode::Id::LOP3_IMM: {
-                std::string op_c = regs.GetRegisterAsInteger(instr.gpr39);
+                const 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 {
@@ -1490,9 +1496,9 @@ private:
             case OpCode::Id::IMNMX_IMM: {
                 ASSERT_MSG(instr.imnmx.exchange == Tegra::Shader::IMinMaxExchange::None,
                            "Unimplemented");
-                std::string condition =
+                const std::string condition =
                     GetPredicateCondition(instr.imnmx.pred, instr.imnmx.negate_pred != 0);
-                std::string parameters = op_a + ',' + op_b;
+                const std::string parameters = op_a + ',' + op_b;
                 regs.SetRegisterToInteger(instr.gpr0, instr.imnmx.is_signed, 0,
                                           '(' + condition + ") ? min(" + parameters + ") : max(" +
                                               parameters + ')',
@@ -1509,7 +1515,7 @@ private:
             break;
         }
         case OpCode::Type::Ffma: {
-            std::string op_a = regs.GetRegisterAsFloat(instr.gpr8);
+            const std::string op_a = regs.GetRegisterAsFloat(instr.gpr8);
             std::string op_b = instr.ffma.negate_b ? "-" : "";
             std::string op_c = instr.ffma.negate_c ? "-" : "";
 
@@ -1719,7 +1725,7 @@ private:
                 shader.AddLine("uint index = (" + regs.GetRegisterAsInteger(instr.gpr8, 0, false) +
                                " / 4) & (MAX_CONSTBUFFER_ELEMENTS - 1);");
 
-                std::string op_a =
+                const std::string op_a =
                     regs.GetUniformIndirect(instr.cbuf36.index, instr.cbuf36.offset + 0, "index",
                                             GLSLRegister::Type::Float);
 
@@ -1729,7 +1735,7 @@ private:
                     break;
 
                 case Tegra::Shader::UniformType::Double: {
-                    std::string op_b =
+                    const 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);
@@ -1753,10 +1759,35 @@ private:
                 break;
             }
             case OpCode::Id::TEX: {
-                const std::string op_a = regs.GetRegisterAsFloat(instr.gpr8);
-                const std::string op_b = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
-                const std::string sampler = GetSampler(instr.sampler);
-                const std::string coord = "vec2 coords = vec2(" + op_a + ", " + op_b + ");";
+                ASSERT_MSG(instr.tex.array == 0, "TEX arrays unimplemented");
+                Tegra::Shader::TextureType texture_type{instr.tex.texture_type};
+                std::string coord;
+
+                switch (texture_type) {
+                case Tegra::Shader::TextureType::Texture1D: {
+                    const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
+                    coord = "float coords = " + x + ';';
+                    break;
+                }
+                case Tegra::Shader::TextureType::Texture2D: {
+                    const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
+                    const std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
+                    coord = "vec2 coords = vec2(" + x + ", " + y + ");";
+                    break;
+                }
+                default:
+                    LOG_CRITICAL(HW_GPU, "Unhandled texture type {}",
+                                 static_cast<u32>(texture_type));
+                    UNREACHABLE();
+
+                    // Fallback to interpreting as a 2D texture for now
+                    const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
+                    const std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
+                    coord = "vec2 coords = vec2(" + x + ", " + y + ");";
+                    texture_type = Tegra::Shader::TextureType::Texture2D;
+                }
+
+                const std::string sampler = GetSampler(instr.sampler, texture_type, false);
                 // Add an extra scope and declare the texture coords inside to prevent
                 // overwriting them in case they are used as outputs of the texs instruction.
                 shader.AddLine("{");
@@ -1778,20 +1809,65 @@ private:
                 break;
             }
             case OpCode::Id::TEXS: {
-                const std::string op_a = regs.GetRegisterAsFloat(instr.gpr8);
-                const std::string op_b = regs.GetRegisterAsFloat(instr.gpr20);
-                const std::string sampler = GetSampler(instr.sampler);
-                const std::string coord = "vec2 coords = vec2(" + op_a + ", " + op_b + ");";
+                std::string coord;
+                Tegra::Shader::TextureType texture_type{instr.texs.GetTextureType()};
+                bool is_array{instr.texs.IsArrayTexture()};
 
+                switch (texture_type) {
+                case Tegra::Shader::TextureType::Texture2D: {
+                    if (is_array) {
+                        const std::string index = regs.GetRegisterAsInteger(instr.gpr8);
+                        const std::string x = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
+                        const std::string y = regs.GetRegisterAsFloat(instr.gpr20);
+                        coord = "vec3 coords = vec3(" + x + ", " + y + ", " + index + ");";
+                    } else {
+                        const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
+                        const std::string y = regs.GetRegisterAsFloat(instr.gpr20);
+                        coord = "vec2 coords = vec2(" + x + ", " + y + ");";
+                    }
+                    break;
+                }
+                default:
+                    LOG_CRITICAL(HW_GPU, "Unhandled texture type {}",
+                                 static_cast<u32>(texture_type));
+                    UNREACHABLE();
+
+                    // Fallback to interpreting as a 2D texture for now
+                    const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
+                    const std::string y = regs.GetRegisterAsFloat(instr.gpr20);
+                    coord = "vec2 coords = vec2(" + x + ", " + y + ");";
+                    texture_type = Tegra::Shader::TextureType::Texture2D;
+                    is_array = false;
+                }
+                const std::string sampler = GetSampler(instr.sampler, texture_type, is_array);
                 const std::string texture = "texture(" + sampler + ", coords)";
                 WriteTexsInstruction(instr, coord, texture);
                 break;
             }
             case OpCode::Id::TLDS: {
-                const std::string op_a = regs.GetRegisterAsInteger(instr.gpr8);
-                const std::string op_b = regs.GetRegisterAsInteger(instr.gpr20);
-                const std::string sampler = GetSampler(instr.sampler);
-                const std::string coord = "ivec2 coords = ivec2(" + op_a + ", " + op_b + ");";
+                ASSERT(instr.tlds.GetTextureType() == Tegra::Shader::TextureType::Texture2D);
+                ASSERT(instr.tlds.IsArrayTexture() == false);
+                std::string coord;
+
+                switch (instr.tlds.GetTextureType()) {
+                case Tegra::Shader::TextureType::Texture2D: {
+                    if (instr.tlds.IsArrayTexture()) {
+                        LOG_CRITICAL(HW_GPU, "Unhandled 2d array texture");
+                        UNREACHABLE();
+                    } else {
+                        const std::string x = regs.GetRegisterAsInteger(instr.gpr8);
+                        const std::string y = regs.GetRegisterAsInteger(instr.gpr20);
+                        coord = "ivec2 coords = ivec2(" + x + ", " + y + ");";
+                    }
+                    break;
+                }
+                default:
+                    LOG_CRITICAL(HW_GPU, "Unhandled texture type {}",
+                                 static_cast<u32>(instr.tlds.GetTextureType()));
+                    UNREACHABLE();
+                }
+                const std::string sampler = GetSampler(instr.sampler, instr.tlds.GetTextureType(),
+                                                       instr.tlds.IsArrayTexture());
                 const std::string texture = "texelFetch(" + sampler + ", coords, 0)";
                 WriteTexsInstruction(instr, coord, texture);
                 break;
@@ -1799,12 +1875,12 @@ private:
             case OpCode::Id::TLD4: {
                 ASSERT(instr.tld4.texture_type == Tegra::Shader::TextureType::Texture2D);
                 ASSERT(instr.tld4.array == 0);
-                std::string coord{};
+                std::string coord;
 
                 switch (instr.tld4.texture_type) {
                 case Tegra::Shader::TextureType::Texture2D: {
-                    std::string x = regs.GetRegisterAsFloat(instr.gpr8);
-                    std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
+                    const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
+                    const std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
                     coord = "vec2 coords = vec2(" + x + ", " + y + ");";
                     break;
                 }
@@ -1814,7 +1890,8 @@ private:
                     UNREACHABLE();
                 }
 
-                const std::string sampler = GetSampler(instr.sampler);
+                const std::string sampler =
+                    GetSampler(instr.sampler, instr.tld4.texture_type, false);
                 // Add an extra scope and declare the texture coords inside to prevent
                 // overwriting them in case they are used as outputs of the texs instruction.
                 shader.AddLine("{");
@@ -1840,13 +1917,82 @@ private:
                 const std::string op_a = regs.GetRegisterAsFloat(instr.gpr8);
                 const std::string op_b = regs.GetRegisterAsFloat(instr.gpr20);
                 // TODO(Subv): Figure out how the sampler type is encoded in the TLD4S instruction.
-                const std::string sampler = GetSampler(instr.sampler);
+                const std::string sampler =
+                    GetSampler(instr.sampler, Tegra::Shader::TextureType::Texture2D, false);
                 const std::string coord = "vec2 coords = vec2(" + op_a + ", " + op_b + ");";
                 const std::string texture = "textureGather(" + sampler + ", coords, " +
                                             std::to_string(instr.tld4s.component) + ')';
                 WriteTexsInstruction(instr, coord, texture);
                 break;
             }
+            case OpCode::Id::TXQ: {
+                // TODO: the new commits on the texture refactor, change the way samplers work.
+                // Sadly, not all texture instructions specify the type of texture their sampler
+                // uses. This must be fixed at a later instance.
+                const std::string sampler =
+                    GetSampler(instr.sampler, Tegra::Shader::TextureType::Texture2D, false);
+                switch (instr.txq.query_type) {
+                case Tegra::Shader::TextureQueryType::Dimension: {
+                    const std::string texture = "textureQueryLevels(" + sampler + ')';
+                    regs.SetRegisterToInteger(instr.gpr0, true, 0, texture, 1, 1);
+                    break;
+                }
+                default: {
+                    LOG_CRITICAL(HW_GPU, "Unhandled texture query type: {}",
+                                 static_cast<u32>(instr.txq.query_type.Value()));
+                    UNREACHABLE();
+                }
+                }
+                break;
+            }
+            case OpCode::Id::TMML: {
+                const std::string op_a = regs.GetRegisterAsFloat(instr.gpr8);
+                const std::string op_b = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
+                const bool is_array = instr.tmml.array != 0;
+                auto texture_type = instr.tmml.texture_type.Value();
+                const std::string sampler = GetSampler(instr.sampler, texture_type, is_array);
+
+                // TODO: add coordinates for different samplers once other texture types are
+                // implemented.
+                std::string coord;
+                switch (texture_type) {
+                case Tegra::Shader::TextureType::Texture1D: {
+                    std::string x = regs.GetRegisterAsFloat(instr.gpr8);
+                    coord = "float coords = " + x + ';';
+                    break;
+                }
+                case Tegra::Shader::TextureType::Texture2D: {
+                    std::string x = regs.GetRegisterAsFloat(instr.gpr8);
+                    std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
+                    coord = "vec2 coords = vec2(" + x + ", " + y + ");";
+                    break;
+                }
+                default:
+                    LOG_CRITICAL(HW_GPU, "Unhandled texture type {}",
+                                 static_cast<u32>(texture_type));
+                    UNREACHABLE();
+
+                    // Fallback to interpreting as a 2D texture for now
+                    std::string x = regs.GetRegisterAsFloat(instr.gpr8);
+                    std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
+                    coord = "vec2 coords = vec2(" + x + ", " + y + ");";
+                    texture_type = Tegra::Shader::TextureType::Texture2D;
+                }
+                // Add an extra scope and declare the texture coords inside to prevent
+                // overwriting them in case they are used as outputs of the texs instruction.
+                shader.AddLine('{');
+                ++shader.scope;
+                shader.AddLine(coord);
+                const std::string texture = "textureQueryLod(" + sampler + ", coords)";
+                const std::string tmp = "vec2 tmp = " + texture + "*vec2(256.0, 256.0);";
+                shader.AddLine(tmp);
+
+                regs.SetRegisterToInteger(instr.gpr0, true, 0, "int(tmp.y)", 1, 1);
+                regs.SetRegisterToInteger(instr.gpr0.Value() + 1, false, 0, "uint(tmp.x)", 1, 1);
+                --shader.scope;
+                shader.AddLine('}');
+                break;
+            }
             default: {
                 LOG_CRITICAL(HW_GPU, "Unhandled memory instruction: {}", opcode->GetName());
                 UNREACHABLE();
@@ -1886,12 +2032,12 @@ private:
             // We can't use the constant predicate as destination.
             ASSERT(instr.fsetp.pred3 != static_cast<u64>(Pred::UnusedIndex));
 
-            std::string second_pred =
+            const std::string second_pred =
                 GetPredicateCondition(instr.fsetp.pred39, instr.fsetp.neg_pred != 0);
 
-            std::string combiner = GetPredicateCombiner(instr.fsetp.op);
+            const std::string combiner = GetPredicateCombiner(instr.fsetp.op);
 
-            std::string predicate = GetPredicateComparison(instr.fsetp.cond, op_a, op_b);
+            const std::string predicate = GetPredicateComparison(instr.fsetp.cond, op_a, op_b);
             // Set the primary predicate to the result of Predicate OP SecondPredicate
             SetPredicate(instr.fsetp.pred3,
                          '(' + predicate + ") " + combiner + " (" + second_pred + ')');
@@ -1905,7 +2051,8 @@ private:
             break;
         }
         case OpCode::Type::IntegerSetPredicate: {
-            std::string op_a = regs.GetRegisterAsInteger(instr.gpr8, 0, instr.isetp.is_signed);
+            const std::string op_a =
+                regs.GetRegisterAsInteger(instr.gpr8, 0, instr.isetp.is_signed);
             std::string op_b;
 
             if (instr.is_b_imm) {
@@ -1922,12 +2069,12 @@ private:
             // We can't use the constant predicate as destination.
             ASSERT(instr.isetp.pred3 != static_cast<u64>(Pred::UnusedIndex));
 
-            std::string second_pred =
+            const std::string second_pred =
                 GetPredicateCondition(instr.isetp.pred39, instr.isetp.neg_pred != 0);
 
-            std::string combiner = GetPredicateCombiner(instr.isetp.op);
+            const std::string combiner = GetPredicateCombiner(instr.isetp.op);
 
-            std::string predicate = GetPredicateComparison(instr.isetp.cond, op_a, op_b);
+            const std::string predicate = GetPredicateComparison(instr.isetp.cond, op_a, op_b);
             // Set the primary predicate to the result of Predicate OP SecondPredicate
             SetPredicate(instr.isetp.pred3,
                          '(' + predicate + ") " + combiner + " (" + second_pred + ')');
@@ -1941,20 +2088,20 @@ private:
             break;
         }
         case OpCode::Type::PredicateSetPredicate: {
-            std::string op_a =
+            const std::string op_a =
                 GetPredicateCondition(instr.psetp.pred12, instr.psetp.neg_pred12 != 0);
-            std::string op_b =
+            const std::string op_b =
                 GetPredicateCondition(instr.psetp.pred29, instr.psetp.neg_pred29 != 0);
 
             // We can't use the constant predicate as destination.
             ASSERT(instr.psetp.pred3 != static_cast<u64>(Pred::UnusedIndex));
 
-            std::string second_pred =
+            const std::string second_pred =
                 GetPredicateCondition(instr.psetp.pred39, instr.psetp.neg_pred39 != 0);
 
-            std::string combiner = GetPredicateCombiner(instr.psetp.op);
+            const std::string combiner = GetPredicateCombiner(instr.psetp.op);
 
-            std::string predicate =
+            const std::string predicate =
                 '(' + op_a + ") " + GetPredicateCombiner(instr.psetp.cond) + " (" + op_b + ')';
 
             // Set the primary predicate to the result of Predicate OP SecondPredicate
@@ -1980,7 +2127,7 @@ private:
             std::string op_b = instr.fset.neg_b ? "-" : "";
 
             if (instr.is_b_imm) {
-                std::string imm = GetImmediate19(instr);
+                const std::string imm = GetImmediate19(instr);
                 if (instr.fset.neg_imm)
                     op_b += "(-" + imm + ')';
                 else
@@ -2000,13 +2147,14 @@ private:
 
             // The fset instruction sets a register to 1.0 or -1 (depending on the bf bit) if the
             // condition is true, and to 0 otherwise.
-            std::string second_pred =
+            const std::string second_pred =
                 GetPredicateCondition(instr.fset.pred39, instr.fset.neg_pred != 0);
 
-            std::string combiner = GetPredicateCombiner(instr.fset.op);
+            const std::string combiner = GetPredicateCombiner(instr.fset.op);
 
-            std::string predicate = "((" + GetPredicateComparison(instr.fset.cond, op_a, op_b) +
-                                    ") " + combiner + " (" + second_pred + "))";
+            const std::string predicate = "((" +
+                                          GetPredicateComparison(instr.fset.cond, op_a, op_b) +
+                                          ") " + combiner + " (" + second_pred + "))";
 
             if (instr.fset.bf) {
                 regs.SetRegisterToFloat(instr.gpr0, 0, predicate + " ? 1.0 : 0.0", 1, 1);
@@ -2017,7 +2165,7 @@ private:
             break;
         }
         case OpCode::Type::IntegerSet: {
-            std::string op_a = regs.GetRegisterAsInteger(instr.gpr8, 0, instr.iset.is_signed);
+            const std::string op_a = regs.GetRegisterAsInteger(instr.gpr8, 0, instr.iset.is_signed);
 
             std::string op_b;
 
@@ -2034,13 +2182,14 @@ private:
 
             // The iset instruction sets a register to 1.0 or -1 (depending on the bf bit) if the
             // condition is true, and to 0 otherwise.
-            std::string second_pred =
+            const std::string second_pred =
                 GetPredicateCondition(instr.iset.pred39, instr.iset.neg_pred != 0);
 
-            std::string combiner = GetPredicateCombiner(instr.iset.op);
+            const std::string combiner = GetPredicateCombiner(instr.iset.op);
 
-            std::string predicate = "((" + GetPredicateComparison(instr.iset.cond, op_a, op_b) +
-                                    ") " + combiner + " (" + second_pred + "))";
+            const std::string predicate = "((" +
+                                          GetPredicateComparison(instr.iset.cond, op_a, op_b) +
+                                          ") " + combiner + " (" + second_pred + "))";
 
             if (instr.iset.bf) {
                 regs.SetRegisterToFloat(instr.gpr0, 0, predicate + " ? 1.0 : 0.0", 1, 1);
@@ -2190,18 +2339,22 @@ private:
             case OpCode::Id::BRA: {
                 ASSERT_MSG(instr.bra.constant_buffer == 0,
                            "BRA with constant buffers are not implemented");
-                u32 target = offset + instr.bra.GetBranchTarget();
+                const u32 target = offset + instr.bra.GetBranchTarget();
                 shader.AddLine("{ jmp_to = " + std::to_string(target) + "u; break; }");
                 break;
             }
             case OpCode::Id::IPA: {
                 const auto& attribute = instr.attribute.fmt28;
                 const auto& reg = instr.gpr0;
-                ASSERT_MSG(instr.ipa.saturate == 0, "IPA saturate not implemented");
+
                 Tegra::Shader::IpaMode input_mode{instr.ipa.interp_mode.Value(),
                                                   instr.ipa.sample_mode.Value()};
                 regs.SetRegisterToInputAttibute(reg, attribute.element, attribute.index,
                                                 input_mode);
+
+                if (instr.ipa.saturate) {
+                    regs.SetRegisterToFloat(reg, 0, regs.GetRegisterAsFloat(reg), 1, 1, true);
+                }
                 break;
             }
             case OpCode::Id::SSY: {
@@ -2210,7 +2363,7 @@ private:
                 // has a similar structure to the BRA opcode.
                 ASSERT_MSG(instr.bra.constant_buffer == 0, "Constant buffer SSY is not supported");
 
-                u32 target = offset + instr.bra.GetBranchTarget();
+                const u32 target = offset + instr.bra.GetBranchTarget();
                 EmitPushToSSYStack(target);
                 break;
             }
@@ -2304,10 +2457,10 @@ private:
                     shader.AddLine("case " + std::to_string(label) + "u: {");
                     ++shader.scope;
 
-                    auto next_it = labels.lower_bound(label + 1);
-                    u32 next_label = next_it == labels.end() ? subroutine.end : *next_it;
+                    const auto next_it = labels.lower_bound(label + 1);
+                    const u32 next_label = next_it == labels.end() ? subroutine.end : *next_it;
 
-                    u32 compile_end = CompileRange(label, next_label);
+                    const u32 compile_end = CompileRange(label, next_label);
                     if (compile_end > next_label && compile_end != PROGRAM_END) {
                         // This happens only when there is a label inside a IF/LOOP block
                         shader.AddLine(" jmp_to = " + std::to_string(compile_end) + "u; break; }");
@@ -2370,7 +2523,8 @@ boost::optional<ProgramResult> DecompileProgram(const ProgramCode& program_code,
                                                 Maxwell3D::Regs::ShaderStage stage,
                                                 const std::string& suffix) {
     try {
-        auto subroutines = ControlFlowAnalyzer(program_code, main_offset, suffix).GetSubroutines();
+        const auto subroutines =
+            ControlFlowAnalyzer(program_code, main_offset, suffix).GetSubroutines();
         GLSLGenerator generator(subroutines, program_code, main_offset, stage, suffix);
         return ProgramResult{generator.GetShaderCode(), generator.GetEntries()};
     } catch (const DecompileFail& exception) {

src/video_core/renderer_opengl/gl_shader_gen.h

@@ -9,6 +9,7 @@
 #include <vector>
 
 #include "common/common_types.h"
+#include "video_core/engines/shader_bytecode.h"
 
 namespace OpenGL::GLShader {
 
@@ -73,8 +74,9 @@ class SamplerEntry {
     using Maxwell = Tegra::Engines::Maxwell3D::Regs;
 
 public:
-    SamplerEntry(Maxwell::ShaderStage stage, size_t offset, size_t index)
-        : offset(offset), stage(stage), sampler_index(index) {}
+    SamplerEntry(Maxwell::ShaderStage stage, size_t offset, size_t index,
+                 Tegra::Shader::TextureType type, bool is_array)
+        : offset(offset), stage(stage), sampler_index(index), type(type), is_array(is_array) {}
 
     size_t GetOffset() const {
         return offset;
@@ -89,8 +91,41 @@ public:
     }
 
     std::string GetName() const {
-        return std::string(TextureSamplerNames[static_cast<size_t>(stage)]) + '[' +
-               std::to_string(sampler_index) + ']';
+        return std::string(TextureSamplerNames[static_cast<size_t>(stage)]) + '_' +
+               std::to_string(sampler_index);
+    }
+
+    std::string GetTypeString() const {
+        using Tegra::Shader::TextureType;
+        std::string glsl_type;
+
+        switch (type) {
+        case TextureType::Texture1D:
+            glsl_type = "sampler1D";
+            break;
+        case TextureType::Texture2D:
+            glsl_type = "sampler2D";
+            break;
+        case TextureType::Texture3D:
+            glsl_type = "sampler3D";
+            break;
+        case TextureType::TextureCube:
+            glsl_type = "samplerCube";
+            break;
+        default:
+            UNIMPLEMENTED();
+        }
+        if (is_array)
+            glsl_type += "Array";
+        return glsl_type;
+    }
+
+    Tegra::Shader::TextureType GetType() const {
+        return type;
+    }
+
+    bool IsArray() const {
+        return is_array;
     }
 
     u32 GetHash() const {
@@ -105,11 +140,14 @@ private:
     static constexpr std::array<const char*, Maxwell::MaxShaderStage> TextureSamplerNames = {
         "tex_vs", "tex_tessc", "tex_tesse", "tex_gs", "tex_fs",
     };
+
     /// Offset in TSC memory from which to read the sampler object, as specified by the sampling
     /// instruction.
     size_t offset;
-    Maxwell::ShaderStage stage; ///< Shader stage where this sampler was used.
-    size_t sampler_index;       ///< Value used to index into the generated GLSL sampler array.
+    Maxwell::ShaderStage stage;      ///< Shader stage where this sampler was used.
+    size_t sampler_index;            ///< Value used to index into the generated GLSL sampler array.
+    Tegra::Shader::TextureType type; ///< The type used to sample this texture (Texture2D, etc)
+    bool is_array; ///< Whether the texture is being sampled as an array texture or not.
 };
 
 struct ShaderEntries {

src/video_core/renderer_opengl/gl_shader_util.cpp

@@ -25,7 +25,7 @@ GLuint LoadShader(const char* source, GLenum type) {
     default:
         UNREACHABLE();
     }
-    GLuint shader_id = glCreateShader(type);
+    const GLuint shader_id = glCreateShader(type);
     glShaderSource(shader_id, 1, &source, nullptr);
     LOG_DEBUG(Render_OpenGL, "Compiling {} shader...", debug_type);
     glCompileShader(shader_id);

src/video_core/renderer_opengl/gl_state.cpp

@@ -200,9 +200,9 @@ void OpenGLState::Apply() const {
         const auto& texture_unit = texture_units[i];
         const auto& cur_state_texture_unit = cur_state.texture_units[i];
 
-        if (texture_unit.texture_2d != cur_state_texture_unit.texture_2d) {
+        if (texture_unit.texture != cur_state_texture_unit.texture) {
             glActiveTexture(TextureUnits::MaxwellTexture(static_cast<int>(i)).Enum());
-            glBindTexture(GL_TEXTURE_2D, texture_unit.texture_2d);
+            glBindTexture(texture_unit.target, texture_unit.texture);
         }
         if (texture_unit.sampler != cur_state_texture_unit.sampler) {
             glBindSampler(static_cast<GLuint>(i), texture_unit.sampler);
@@ -214,7 +214,7 @@ void OpenGLState::Apply() const {
             texture_unit.swizzle.a != cur_state_texture_unit.swizzle.a) {
             std::array<GLint, 4> mask = {texture_unit.swizzle.r, texture_unit.swizzle.g,
                                          texture_unit.swizzle.b, texture_unit.swizzle.a};
-            glTexParameteriv(GL_TEXTURE_2D, GL_TEXTURE_SWIZZLE_RGBA, mask.data());
+            glTexParameteriv(texture_unit.target, GL_TEXTURE_SWIZZLE_RGBA, mask.data());
         }
     }
 
@@ -287,7 +287,7 @@ void OpenGLState::Apply() const {
 
 OpenGLState& OpenGLState::UnbindTexture(GLuint handle) {
     for (auto& unit : texture_units) {
-        if (unit.texture_2d == handle) {
+        if (unit.texture == handle) {
             unit.Unbind();
         }
     }

src/video_core/renderer_opengl/gl_state.h

@@ -94,8 +94,9 @@ public:
 
     // 3 texture units - one for each that is used in PICA fragment shader emulation
     struct TextureUnit {
-        GLuint texture_2d; // GL_TEXTURE_BINDING_2D
-        GLuint sampler;    // GL_SAMPLER_BINDING
+        GLuint texture; // GL_TEXTURE_BINDING_2D
+        GLuint sampler; // GL_SAMPLER_BINDING
+        GLenum target;
         struct {
             GLint r; // GL_TEXTURE_SWIZZLE_R
             GLint g; // GL_TEXTURE_SWIZZLE_G
@@ -104,7 +105,7 @@ public:
         } swizzle;
 
         void Unbind() {
-            texture_2d = 0;
+            texture = 0;
             swizzle.r = GL_RED;
             swizzle.g = GL_GREEN;
             swizzle.b = GL_BLUE;
@@ -114,6 +115,7 @@ public:
         void Reset() {
             Unbind();
             sampler = 0;
+            target = GL_TEXTURE_2D;
         }
     };
     std::array<TextureUnit, 32> texture_units;

src/video_core/renderer_opengl/gl_stream_buffer.cpp

@@ -29,7 +29,7 @@ OGLStreamBuffer::OGLStreamBuffer(GLenum target, GLsizeiptr size, bool prefer_coh
     if (GLAD_GL_ARB_buffer_storage) {
         persistent = true;
         coherent = prefer_coherent;
-        GLbitfield flags =
+        const GLbitfield flags =
             GL_MAP_WRITE_BIT | GL_MAP_PERSISTENT_BIT | (coherent ? GL_MAP_COHERENT_BIT : 0);
         glBufferStorage(gl_target, allocate_size, nullptr, flags);
         mapped_ptr = static_cast<u8*>(glMapBufferRange(

src/video_core/renderer_opengl/renderer_opengl.cpp

@@ -177,7 +177,7 @@ void RendererOpenGL::LoadFBToScreenInfo(const Tegra::FramebufferConfig& framebuf
                                        Memory::GetPointer(framebuffer_addr),
                                        gl_framebuffer_data.data(), true);
 
-        state.texture_units[0].texture_2d = screen_info.texture.resource.handle;
+        state.texture_units[0].texture = screen_info.texture.resource.handle;
         state.Apply();
 
         glActiveTexture(GL_TEXTURE0);
@@ -194,7 +194,7 @@ void RendererOpenGL::LoadFBToScreenInfo(const Tegra::FramebufferConfig& framebuf
 
         glPixelStorei(GL_UNPACK_ROW_LENGTH, 0);
 
-        state.texture_units[0].texture_2d = 0;
+        state.texture_units[0].texture = 0;
         state.Apply();
     }
 }
@@ -205,7 +205,7 @@ void RendererOpenGL::LoadFBToScreenInfo(const Tegra::FramebufferConfig& framebuf
  */
 void RendererOpenGL::LoadColorToActiveGLTexture(u8 color_r, u8 color_g, u8 color_b, u8 color_a,
                                                 const TextureInfo& texture) {
-    state.texture_units[0].texture_2d = texture.resource.handle;
+    state.texture_units[0].texture = texture.resource.handle;
     state.Apply();
 
     glActiveTexture(GL_TEXTURE0);
@@ -214,7 +214,7 @@ void RendererOpenGL::LoadColorToActiveGLTexture(u8 color_r, u8 color_g, u8 color
     // Update existing texture
     glTexImage2D(GL_TEXTURE_2D, 0, GL_RGBA, 1, 1, 0, GL_RGBA, GL_UNSIGNED_BYTE, framebuffer_data);
 
-    state.texture_units[0].texture_2d = 0;
+    state.texture_units[0].texture = 0;
     state.Apply();
 }
 
@@ -260,7 +260,7 @@ void RendererOpenGL::InitOpenGLObjects() {
     // Allocation of storage is deferred until the first frame, when we
     // know the framebuffer size.
 
-    state.texture_units[0].texture_2d = screen_info.texture.resource.handle;
+    state.texture_units[0].texture = screen_info.texture.resource.handle;
     state.Apply();
 
     glActiveTexture(GL_TEXTURE0);
@@ -272,7 +272,7 @@ void RendererOpenGL::InitOpenGLObjects() {
 
     screen_info.display_texture = screen_info.texture.resource.handle;
 
-    state.texture_units[0].texture_2d = 0;
+    state.texture_units[0].texture = 0;
     state.Apply();
 
     // Clear screen to black
@@ -305,14 +305,14 @@ void RendererOpenGL::ConfigureFramebufferTexture(TextureInfo& texture,
         UNREACHABLE();
     }
 
-    state.texture_units[0].texture_2d = texture.resource.handle;
+    state.texture_units[0].texture = texture.resource.handle;
     state.Apply();
 
     glActiveTexture(GL_TEXTURE0);
     glTexImage2D(GL_TEXTURE_2D, 0, internal_format, texture.width, texture.height, 0,
                  texture.gl_format, texture.gl_type, nullptr);
 
-    state.texture_units[0].texture_2d = 0;
+    state.texture_units[0].texture = 0;
     state.Apply();
 }
 
@@ -354,14 +354,14 @@ void RendererOpenGL::DrawScreenTriangles(const ScreenInfo& screen_info, float x,
         ScreenRectVertex(x + w, y + h, texcoords.bottom * scale_u, right * scale_v),
     }};
 
-    state.texture_units[0].texture_2d = screen_info.display_texture;
+    state.texture_units[0].texture = screen_info.display_texture;
     state.texture_units[0].swizzle = {GL_RED, GL_GREEN, GL_BLUE, GL_ALPHA};
     state.Apply();
 
     glBufferSubData(GL_ARRAY_BUFFER, 0, sizeof(vertices), vertices.data());
     glDrawArrays(GL_TRIANGLE_STRIP, 0, 4);
 
-    state.texture_units[0].texture_2d = 0;
+    state.texture_units[0].texture = 0;
     state.Apply();
 }
 

src/video_core/textures/texture.h

@@ -170,8 +170,12 @@ struct TICEntry {
         BitField<0, 16, u32> width_minus_1;
         BitField<23, 4, TextureType> texture_type;
     };
-    u16 height_minus_1;
-    INSERT_PADDING_BYTES(10);
+    union {
+        BitField<0, 16, u32> height_minus_1;
+        BitField<16, 15, u32> depth_minus_1;
+    };
+
+    INSERT_PADDING_BYTES(8);
 
     GPUVAddr Address() const {
         return static_cast<GPUVAddr>((static_cast<GPUVAddr>(address_high) << 32) | address_low);
@@ -192,6 +196,10 @@ struct TICEntry {
         return height_minus_1 + 1;
     }
 
+    u32 Depth() const {
+        return depth_minus_1 + 1;
+    }
+
     u32 BlockHeight() const {
         ASSERT(header_version == TICHeaderVersion::BlockLinear ||
                header_version == TICHeaderVersion::BlockLinearColorKey);

src/yuzu/CMakeLists.txt

@@ -9,6 +9,8 @@ add_executable(yuzu
     about_dialog.h
     bootmanager.cpp
     bootmanager.h
+    compatibility_list.cpp
+    compatibility_list.h
     configuration/config.cpp
     configuration/config.h
     configuration/configure_audio.cpp
@@ -43,6 +45,8 @@ add_executable(yuzu
     game_list.cpp
     game_list.h
     game_list_p.h
+    game_list_worker.cpp
+    game_list_worker.h
     hotkeys.cpp
     hotkeys.h
     main.cpp

src/yuzu/about_dialog.cpp

@@ -11,7 +11,7 @@ AboutDialog::AboutDialog(QWidget* parent) : QDialog(parent), ui(new Ui::AboutDia
     ui->setupUi(this);
     ui->labelLogo->setPixmap(QIcon::fromTheme("yuzu").pixmap(200));
     ui->labelBuildInfo->setText(
-        ui->labelBuildInfo->text().arg(Common::g_build_name, Common::g_scm_branch,
+        ui->labelBuildInfo->text().arg(Common::g_build_fullname, Common::g_scm_branch,
                                        Common::g_scm_desc, QString(Common::g_build_date).left(10)));
 }
 

src/yuzu/bootmanager.cpp

@@ -256,6 +256,7 @@ void GRenderWindow::InitRenderTarget() {
     QGLFormat fmt;
     fmt.setVersion(3, 3);
     fmt.setProfile(QGLFormat::CoreProfile);
+    fmt.setSwapInterval(false);
 
     // Requests a forward-compatible context, which is required to get a 3.2+ context on OS X
     fmt.setOption(QGL::NoDeprecatedFunctions);

src/yuzu/compatibility_list.cpp

@@ -0,0 +1,18 @@
+// Copyright 2018 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <algorithm>
+
+#include <fmt/format.h>
+
+#include "yuzu/compatibility_list.h"
+
+CompatibilityList::const_iterator FindMatchingCompatibilityEntry(
+    const CompatibilityList& compatibility_list, u64 program_id) {
+    return std::find_if(compatibility_list.begin(), compatibility_list.end(),
+                        [program_id](const auto& element) {
+                            std::string pid = fmt::format("{:016X}", program_id);
+                            return element.first == pid;
+                        });
+}

src/yuzu/compatibility_list.h

@@ -0,0 +1,17 @@
+// Copyright 2018 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <string>
+#include <unordered_map>
+
+#include <QString>
+
+#include "common/common_types.h"
+
+using CompatibilityList = std::unordered_map<std::string, std::pair<QString, QString>>;
+
+CompatibilityList::const_iterator FindMatchingCompatibilityEntry(
+    const CompatibilityList& compatibility_list, u64 program_id);

src/yuzu/configuration/configure_general.ui

@@ -61,7 +61,7 @@
           <item>
            <widget class="QCheckBox" name="use_multi_core">
             <property name="text">
-             <string>Enable multi-core</string>
+             <string>Enable multi-core (Currently broken.)</string>
             </property>
            </widget>
           </item>

src/yuzu/game_list.cpp

@@ -18,17 +18,11 @@
 #include "common/common_types.h"
 #include "common/file_util.h"
 #include "common/logging/log.h"
-#include "core/file_sys/content_archive.h"
-#include "core/file_sys/control_metadata.h"
-#include "core/file_sys/nca_metadata.h"
 #include "core/file_sys/patch_manager.h"
-#include "core/file_sys/registered_cache.h"
-#include "core/file_sys/romfs.h"
-#include "core/file_sys/vfs_real.h"
-#include "core/hle/service/filesystem/filesystem.h"
-#include "core/loader/loader.h"
+#include "yuzu/compatibility_list.h"
 #include "yuzu/game_list.h"
 #include "yuzu/game_list_p.h"
+#include "yuzu/game_list_worker.h"
 #include "yuzu/main.h"
 #include "yuzu/ui_settings.h"
 
@@ -436,45 +430,6 @@ void GameList::LoadInterfaceLayout() {
 
 const QStringList GameList::supported_file_extensions = {"nso", "nro", "nca", "xci", "nsp"};
 
-static bool HasSupportedFileExtension(const std::string& file_name) {
-    const QFileInfo file = QFileInfo(QString::fromStdString(file_name));
-    return GameList::supported_file_extensions.contains(file.suffix(), Qt::CaseInsensitive);
-}
-
-static bool IsExtractedNCAMain(const std::string& file_name) {
-    return QFileInfo(QString::fromStdString(file_name)).fileName() == "main";
-}
-
-static QString FormatGameName(const std::string& physical_name) {
-    const QString physical_name_as_qstring = QString::fromStdString(physical_name);
-    const QFileInfo file_info(physical_name_as_qstring);
-
-    if (IsExtractedNCAMain(physical_name)) {
-        return file_info.dir().path();
-    }
-
-    return physical_name_as_qstring;
-}
-
-static QString FormatPatchNameVersions(const FileSys::PatchManager& patch_manager,
-                                       bool updatable = true) {
-    QString out;
-    for (const auto& kv : patch_manager.GetPatchVersionNames()) {
-        if (!updatable && kv.first == FileSys::PatchType::Update)
-            continue;
-
-        if (kv.second.empty()) {
-            out.append(fmt::format("{}\n", FileSys::FormatPatchTypeName(kv.first)).c_str());
-        } else {
-            out.append(fmt::format("{} ({})\n", FileSys::FormatPatchTypeName(kv.first), kv.second)
-                           .c_str());
-        }
-    }
-
-    out.chop(1);
-    return out;
-}
-
 void GameList::RefreshGameDirectory() {
     if (!UISettings::values.gamedir.isEmpty() && current_worker != nullptr) {
         LOG_INFO(Frontend, "Change detected in the games directory. Reloading game list.");
@@ -482,176 +437,3 @@ void GameList::RefreshGameDirectory() {
         PopulateAsync(UISettings::values.gamedir, UISettings::values.gamedir_deepscan);
     }
 }
-
-static void GetMetadataFromControlNCA(const FileSys::PatchManager& patch_manager,
-                                      const std::shared_ptr<FileSys::NCA>& nca,
-                                      std::vector<u8>& icon, std::string& name) {
-    auto [nacp, icon_file] = patch_manager.ParseControlNCA(nca);
-    if (icon_file != nullptr)
-        icon = icon_file->ReadAllBytes();
-    if (nacp != nullptr)
-        name = nacp->GetApplicationName();
-}
-
-GameListWorker::GameListWorker(
-    FileSys::VirtualFilesystem vfs, QString dir_path, bool deep_scan,
-    const std::unordered_map<std::string, std::pair<QString, QString>>& compatibility_list)
-    : vfs(std::move(vfs)), dir_path(std::move(dir_path)), deep_scan(deep_scan),
-      compatibility_list(compatibility_list) {}
-
-GameListWorker::~GameListWorker() = default;
-
-void GameListWorker::AddInstalledTitlesToGameList() {
-    const auto cache = Service::FileSystem::GetUnionContents();
-    const auto installed_games = cache->ListEntriesFilter(FileSys::TitleType::Application,
-                                                          FileSys::ContentRecordType::Program);
-
-    for (const auto& game : installed_games) {
-        const auto& file = cache->GetEntryUnparsed(game);
-        std::unique_ptr<Loader::AppLoader> loader = Loader::GetLoader(file);
-        if (!loader)
-            continue;
-
-        std::vector<u8> icon;
-        std::string name;
-        u64 program_id = 0;
-        loader->ReadProgramId(program_id);
-
-        const FileSys::PatchManager patch{program_id};
-        const auto& control = cache->GetEntry(game.title_id, FileSys::ContentRecordType::Control);
-        if (control != nullptr)
-            GetMetadataFromControlNCA(patch, control, icon, name);
-
-        auto it = FindMatchingCompatibilityEntry(compatibility_list, program_id);
-
-        // The game list uses this as compatibility number for untested games
-        QString compatibility("99");
-        if (it != compatibility_list.end())
-            compatibility = it->second.first;
-
-        emit EntryReady({
-            new GameListItemPath(
-                FormatGameName(file->GetFullPath()), icon, QString::fromStdString(name),
-                QString::fromStdString(Loader::GetFileTypeString(loader->GetFileType())),
-                program_id),
-            new GameListItemCompat(compatibility),
-            new GameListItem(FormatPatchNameVersions(patch)),
-            new GameListItem(
-                QString::fromStdString(Loader::GetFileTypeString(loader->GetFileType()))),
-            new GameListItemSize(file->GetSize()),
-        });
-    }
-
-    const auto control_data = cache->ListEntriesFilter(FileSys::TitleType::Application,
-                                                       FileSys::ContentRecordType::Control);
-
-    for (const auto& entry : control_data) {
-        const auto nca = cache->GetEntry(entry);
-        if (nca != nullptr)
-            nca_control_map.insert_or_assign(entry.title_id, nca);
-    }
-}
-
-void GameListWorker::FillControlMap(const std::string& dir_path) {
-    const auto nca_control_callback = [this](u64* num_entries_out, const std::string& directory,
-                                             const std::string& virtual_name) -> bool {
-        std::string physical_name = directory + DIR_SEP + virtual_name;
-
-        if (stop_processing)
-            return false; // Breaks the callback loop.
-
-        bool is_dir = FileUtil::IsDirectory(physical_name);
-        QFileInfo file_info(physical_name.c_str());
-        if (!is_dir && file_info.suffix().toStdString() == "nca") {
-            auto nca =
-                std::make_shared<FileSys::NCA>(vfs->OpenFile(physical_name, FileSys::Mode::Read));
-            if (nca->GetType() == FileSys::NCAContentType::Control)
-                nca_control_map.insert_or_assign(nca->GetTitleId(), nca);
-        }
-        return true;
-    };
-
-    FileUtil::ForeachDirectoryEntry(nullptr, dir_path, nca_control_callback);
-}
-
-void GameListWorker::AddFstEntriesToGameList(const std::string& dir_path, unsigned int recursion) {
-    const auto callback = [this, recursion](u64* num_entries_out, const std::string& directory,
-                                            const std::string& virtual_name) -> bool {
-        std::string physical_name = directory + DIR_SEP + virtual_name;
-
-        if (stop_processing)
-            return false; // Breaks the callback loop.
-
-        bool is_dir = FileUtil::IsDirectory(physical_name);
-        if (!is_dir &&
-            (HasSupportedFileExtension(physical_name) || IsExtractedNCAMain(physical_name))) {
-            std::unique_ptr<Loader::AppLoader> loader =
-                Loader::GetLoader(vfs->OpenFile(physical_name, FileSys::Mode::Read));
-            if (!loader || ((loader->GetFileType() == Loader::FileType::Unknown ||
-                             loader->GetFileType() == Loader::FileType::Error) &&
-                            !UISettings::values.show_unknown))
-                return true;
-
-            std::vector<u8> icon;
-            const auto res1 = loader->ReadIcon(icon);
-
-            u64 program_id = 0;
-            const auto res2 = loader->ReadProgramId(program_id);
-
-            std::string name = " ";
-            const auto res3 = loader->ReadTitle(name);
-
-            const FileSys::PatchManager patch{program_id};
-
-            if (res1 != Loader::ResultStatus::Success && res3 != Loader::ResultStatus::Success &&
-                res2 == Loader::ResultStatus::Success) {
-                // Use from metadata pool.
-                if (nca_control_map.find(program_id) != nca_control_map.end()) {
-                    const auto nca = nca_control_map[program_id];
-                    GetMetadataFromControlNCA(patch, nca, icon, name);
-                }
-            }
-
-            auto it = FindMatchingCompatibilityEntry(compatibility_list, program_id);
-
-            // The game list uses this as compatibility number for untested games
-            QString compatibility("99");
-            if (it != compatibility_list.end())
-                compatibility = it->second.first;
-
-            emit EntryReady({
-                new GameListItemPath(
-                    FormatGameName(physical_name), icon, QString::fromStdString(name),
-                    QString::fromStdString(Loader::GetFileTypeString(loader->GetFileType())),
-                    program_id),
-                new GameListItemCompat(compatibility),
-                new GameListItem(FormatPatchNameVersions(patch, loader->IsRomFSUpdatable())),
-                new GameListItem(
-                    QString::fromStdString(Loader::GetFileTypeString(loader->GetFileType()))),
-                new GameListItemSize(FileUtil::GetSize(physical_name)),
-            });
-        } else if (is_dir && recursion > 0) {
-            watch_list.append(QString::fromStdString(physical_name));
-            AddFstEntriesToGameList(physical_name, recursion - 1);
-        }
-
-        return true;
-    };
-
-    FileUtil::ForeachDirectoryEntry(nullptr, dir_path, callback);
-}
-
-void GameListWorker::run() {
-    stop_processing = false;
-    watch_list.append(dir_path);
-    FillControlMap(dir_path.toStdString());
-    AddInstalledTitlesToGameList();
-    AddFstEntriesToGameList(dir_path.toStdString(), deep_scan ? 256 : 0);
-    nca_control_map.clear();
-    emit Finished(watch_list);
-}
-
-void GameListWorker::Cancel() {
-    this->disconnect();
-    stop_processing = true;
-}

src/yuzu/game_list.h

@@ -4,8 +4,6 @@
 
 #pragma once
 
-#include <unordered_map>
-
 #include <QFileSystemWatcher>
 #include <QHBoxLayout>
 #include <QLabel>
@@ -21,6 +19,7 @@
 #include <QWidget>
 
 #include "common/common_types.h"
+#include "yuzu/compatibility_list.h"
 
 class GameListWorker;
 class GMainWindow;
@@ -90,9 +89,8 @@ signals:
     void GameChosen(QString game_path);
     void ShouldCancelWorker();
     void OpenFolderRequested(u64 program_id, GameListOpenTarget target);
-    void NavigateToGamedbEntryRequested(
-        u64 program_id,
-        std::unordered_map<std::string, std::pair<QString, QString>>& compatibility_list);
+    void NavigateToGamedbEntryRequested(u64 program_id,
+                                        const CompatibilityList& compatibility_list);
 
 private slots:
     void onTextChanged(const QString& newText);
@@ -114,7 +112,7 @@ private:
     QStandardItemModel* item_model = nullptr;
     GameListWorker* current_worker = nullptr;
     QFileSystemWatcher* watcher = nullptr;
-    std::unordered_map<std::string, std::pair<QString, QString>> compatibility_list;
+    CompatibilityList compatibility_list;
 };
 
 Q_DECLARE_METATYPE(GameListOpenTarget);

src/yuzu/game_list_p.h

@@ -6,9 +6,7 @@
 
 #include <algorithm>
 #include <array>
-#include <atomic>
 #include <map>
-#include <memory>
 #include <string>
 #include <unordered_map>
 #include <utility>
@@ -16,7 +14,6 @@
 #include <QCoreApplication>
 #include <QImage>
 #include <QObject>
-#include <QRunnable>
 #include <QStandardItem>
 #include <QString>
 
@@ -26,12 +23,6 @@
 #include "yuzu/ui_settings.h"
 #include "yuzu/util/util.h"
 
-namespace FileSys {
-class NCA;
-class RegisteredCache;
-class VfsFilesystem;
-} // namespace FileSys
-
 /**
  * Gets the default icon (for games without valid SMDH)
  * @param large If true, returns large icon (48x48), otherwise returns small icon (24x24)
@@ -43,17 +34,6 @@ static QPixmap GetDefaultIcon(u32 size) {
     return icon;
 }
 
-static auto FindMatchingCompatibilityEntry(
-    const std::unordered_map<std::string, std::pair<QString, QString>>& compatibility_list,
-    u64 program_id) {
-    return std::find_if(
-        compatibility_list.begin(), compatibility_list.end(),
-        [program_id](const std::pair<std::string, std::pair<QString, QString>>& element) {
-            std::string pid = fmt::format("{:016X}", program_id);
-            return element.first == pid;
-        });
-}
-
 class GameListItem : public QStandardItem {
 
 public:
@@ -196,50 +176,3 @@ public:
         return data(SizeRole).toULongLong() < other.data(SizeRole).toULongLong();
     }
 };
-
-/**
- * Asynchronous worker object for populating the game list.
- * Communicates with other threads through Qt's signal/slot system.
- */
-class GameListWorker : public QObject, public QRunnable {
-    Q_OBJECT
-
-public:
-    GameListWorker(
-        std::shared_ptr<FileSys::VfsFilesystem> vfs, QString dir_path, bool deep_scan,
-        const std::unordered_map<std::string, std::pair<QString, QString>>& compatibility_list);
-    ~GameListWorker() override;
-
-public slots:
-    /// Starts the processing of directory tree information.
-    void run() override;
-    /// Tells the worker that it should no longer continue processing. Thread-safe.
-    void Cancel();
-
-signals:
-    /**
-     * The `EntryReady` signal is emitted once an entry has been prepared and is ready
-     * to be added to the game list.
-     * @param entry_items a list with `QStandardItem`s that make up the columns of the new entry.
-     */
-    void EntryReady(QList<QStandardItem*> entry_items);
-
-    /**
-     * After the worker has traversed the game directory looking for entries, this signal is emmited
-     * with a list of folders that should be watched for changes as well.
-     */
-    void Finished(QStringList watch_list);
-
-private:
-    std::shared_ptr<FileSys::VfsFilesystem> vfs;
-    std::map<u64, std::shared_ptr<FileSys::NCA>> nca_control_map;
-    QStringList watch_list;
-    QString dir_path;
-    bool deep_scan;
-    const std::unordered_map<std::string, std::pair<QString, QString>>& compatibility_list;
-    std::atomic_bool stop_processing;
-
-    void AddInstalledTitlesToGameList();
-    void FillControlMap(const std::string& dir_path);
-    void AddFstEntriesToGameList(const std::string& dir_path, unsigned int recursion = 0);
-};

src/yuzu/game_list_worker.cpp

@@ -0,0 +1,239 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <memory>
+#include <string>
+#include <utility>
+#include <vector>
+
+#include <QDir>
+#include <QFileInfo>
+
+#include "common/common_paths.h"
+#include "common/file_util.h"
+#include "core/file_sys/content_archive.h"
+#include "core/file_sys/control_metadata.h"
+#include "core/file_sys/mode.h"
+#include "core/file_sys/nca_metadata.h"
+#include "core/file_sys/patch_manager.h"
+#include "core/file_sys/registered_cache.h"
+#include "core/hle/service/filesystem/filesystem.h"
+#include "core/loader/loader.h"
+#include "yuzu/compatibility_list.h"
+#include "yuzu/game_list.h"
+#include "yuzu/game_list_p.h"
+#include "yuzu/game_list_worker.h"
+#include "yuzu/ui_settings.h"
+
+namespace {
+void GetMetadataFromControlNCA(const FileSys::PatchManager& patch_manager,
+                               const std::shared_ptr<FileSys::NCA>& nca, std::vector<u8>& icon,
+                               std::string& name) {
+    auto [nacp, icon_file] = patch_manager.ParseControlNCA(nca);
+    if (icon_file != nullptr)
+        icon = icon_file->ReadAllBytes();
+    if (nacp != nullptr)
+        name = nacp->GetApplicationName();
+}
+
+bool HasSupportedFileExtension(const std::string& file_name) {
+    const QFileInfo file = QFileInfo(QString::fromStdString(file_name));
+    return GameList::supported_file_extensions.contains(file.suffix(), Qt::CaseInsensitive);
+}
+
+bool IsExtractedNCAMain(const std::string& file_name) {
+    return QFileInfo(QString::fromStdString(file_name)).fileName() == "main";
+}
+
+QString FormatGameName(const std::string& physical_name) {
+    const QString physical_name_as_qstring = QString::fromStdString(physical_name);
+    const QFileInfo file_info(physical_name_as_qstring);
+
+    if (IsExtractedNCAMain(physical_name)) {
+        return file_info.dir().path();
+    }
+
+    return physical_name_as_qstring;
+}
+
+QString FormatPatchNameVersions(const FileSys::PatchManager& patch_manager, bool updatable = true) {
+    QString out;
+    for (const auto& kv : patch_manager.GetPatchVersionNames()) {
+        if (!updatable && kv.first == FileSys::PatchType::Update)
+            continue;
+
+        if (kv.second.empty()) {
+            out.append(fmt::format("{}\n", FileSys::FormatPatchTypeName(kv.first)).c_str());
+        } else {
+            out.append(fmt::format("{} ({})\n", FileSys::FormatPatchTypeName(kv.first), kv.second)
+                           .c_str());
+        }
+    }
+
+    out.chop(1);
+    return out;
+}
+} // Anonymous namespace
+
+GameListWorker::GameListWorker(FileSys::VirtualFilesystem vfs, QString dir_path, bool deep_scan,
+                               const CompatibilityList& compatibility_list)
+    : vfs(std::move(vfs)), dir_path(std::move(dir_path)), deep_scan(deep_scan),
+      compatibility_list(compatibility_list) {}
+
+GameListWorker::~GameListWorker() = default;
+
+void GameListWorker::AddInstalledTitlesToGameList() {
+    const auto cache = Service::FileSystem::GetUnionContents();
+    const auto installed_games = cache->ListEntriesFilter(FileSys::TitleType::Application,
+                                                          FileSys::ContentRecordType::Program);
+
+    for (const auto& game : installed_games) {
+        const auto& file = cache->GetEntryUnparsed(game);
+        std::unique_ptr<Loader::AppLoader> loader = Loader::GetLoader(file);
+        if (!loader)
+            continue;
+
+        std::vector<u8> icon;
+        std::string name;
+        u64 program_id = 0;
+        loader->ReadProgramId(program_id);
+
+        const FileSys::PatchManager patch{program_id};
+        const auto& control = cache->GetEntry(game.title_id, FileSys::ContentRecordType::Control);
+        if (control != nullptr)
+            GetMetadataFromControlNCA(patch, control, icon, name);
+
+        auto it = FindMatchingCompatibilityEntry(compatibility_list, program_id);
+
+        // The game list uses this as compatibility number for untested games
+        QString compatibility("99");
+        if (it != compatibility_list.end())
+            compatibility = it->second.first;
+
+        emit EntryReady({
+            new GameListItemPath(
+                FormatGameName(file->GetFullPath()), icon, QString::fromStdString(name),
+                QString::fromStdString(Loader::GetFileTypeString(loader->GetFileType())),
+                program_id),
+            new GameListItemCompat(compatibility),
+            new GameListItem(FormatPatchNameVersions(patch)),
+            new GameListItem(
+                QString::fromStdString(Loader::GetFileTypeString(loader->GetFileType()))),
+            new GameListItemSize(file->GetSize()),
+        });
+    }
+
+    const auto control_data = cache->ListEntriesFilter(FileSys::TitleType::Application,
+                                                       FileSys::ContentRecordType::Control);
+
+    for (const auto& entry : control_data) {
+        const auto nca = cache->GetEntry(entry);
+        if (nca != nullptr)
+            nca_control_map.insert_or_assign(entry.title_id, nca);
+    }
+}
+
+void GameListWorker::FillControlMap(const std::string& dir_path) {
+    const auto nca_control_callback = [this](u64* num_entries_out, const std::string& directory,
+                                             const std::string& virtual_name) -> bool {
+        std::string physical_name = directory + DIR_SEP + virtual_name;
+
+        if (stop_processing)
+            return false; // Breaks the callback loop.
+
+        bool is_dir = FileUtil::IsDirectory(physical_name);
+        QFileInfo file_info(physical_name.c_str());
+        if (!is_dir && file_info.suffix().toStdString() == "nca") {
+            auto nca =
+                std::make_shared<FileSys::NCA>(vfs->OpenFile(physical_name, FileSys::Mode::Read));
+            if (nca->GetType() == FileSys::NCAContentType::Control)
+                nca_control_map.insert_or_assign(nca->GetTitleId(), nca);
+        }
+        return true;
+    };
+
+    FileUtil::ForeachDirectoryEntry(nullptr, dir_path, nca_control_callback);
+}
+
+void GameListWorker::AddFstEntriesToGameList(const std::string& dir_path, unsigned int recursion) {
+    const auto callback = [this, recursion](u64* num_entries_out, const std::string& directory,
+                                            const std::string& virtual_name) -> bool {
+        std::string physical_name = directory + DIR_SEP + virtual_name;
+
+        if (stop_processing)
+            return false; // Breaks the callback loop.
+
+        bool is_dir = FileUtil::IsDirectory(physical_name);
+        if (!is_dir &&
+            (HasSupportedFileExtension(physical_name) || IsExtractedNCAMain(physical_name))) {
+            std::unique_ptr<Loader::AppLoader> loader =
+                Loader::GetLoader(vfs->OpenFile(physical_name, FileSys::Mode::Read));
+            if (!loader || ((loader->GetFileType() == Loader::FileType::Unknown ||
+                             loader->GetFileType() == Loader::FileType::Error) &&
+                            !UISettings::values.show_unknown))
+                return true;
+
+            std::vector<u8> icon;
+            const auto res1 = loader->ReadIcon(icon);
+
+            u64 program_id = 0;
+            const auto res2 = loader->ReadProgramId(program_id);
+
+            std::string name = " ";
+            const auto res3 = loader->ReadTitle(name);
+
+            const FileSys::PatchManager patch{program_id};
+
+            if (res1 != Loader::ResultStatus::Success && res3 != Loader::ResultStatus::Success &&
+                res2 == Loader::ResultStatus::Success) {
+                // Use from metadata pool.
+                if (nca_control_map.find(program_id) != nca_control_map.end()) {
+                    const auto nca = nca_control_map[program_id];
+                    GetMetadataFromControlNCA(patch, nca, icon, name);
+                }
+            }
+
+            auto it = FindMatchingCompatibilityEntry(compatibility_list, program_id);
+
+            // The game list uses this as compatibility number for untested games
+            QString compatibility("99");
+            if (it != compatibility_list.end())
+                compatibility = it->second.first;
+
+            emit EntryReady({
+                new GameListItemPath(
+                    FormatGameName(physical_name), icon, QString::fromStdString(name),
+                    QString::fromStdString(Loader::GetFileTypeString(loader->GetFileType())),
+                    program_id),
+                new GameListItemCompat(compatibility),
+                new GameListItem(FormatPatchNameVersions(patch, loader->IsRomFSUpdatable())),
+                new GameListItem(
+                    QString::fromStdString(Loader::GetFileTypeString(loader->GetFileType()))),
+                new GameListItemSize(FileUtil::GetSize(physical_name)),
+            });
+        } else if (is_dir && recursion > 0) {
+            watch_list.append(QString::fromStdString(physical_name));
+            AddFstEntriesToGameList(physical_name, recursion - 1);
+        }
+
+        return true;
+    };
+
+    FileUtil::ForeachDirectoryEntry(nullptr, dir_path, callback);
+}
+
+void GameListWorker::run() {
+    stop_processing = false;
+    watch_list.append(dir_path);
+    FillControlMap(dir_path.toStdString());
+    AddInstalledTitlesToGameList();
+    AddFstEntriesToGameList(dir_path.toStdString(), deep_scan ? 256 : 0);
+    nca_control_map.clear();
+    emit Finished(watch_list);
+}
+
+void GameListWorker::Cancel() {
+    this->disconnect();
+    stop_processing = true;
+}

src/yuzu/game_list_worker.h

@@ -0,0 +1,72 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <atomic>
+#include <map>
+#include <memory>
+#include <string>
+#include <unordered_map>
+
+#include <QList>
+#include <QObject>
+#include <QRunnable>
+#include <QString>
+
+#include "common/common_types.h"
+#include "yuzu/compatibility_list.h"
+
+class QStandardItem;
+
+namespace FileSys {
+class NCA;
+class VfsFilesystem;
+} // namespace FileSys
+
+/**
+ * Asynchronous worker object for populating the game list.
+ * Communicates with other threads through Qt's signal/slot system.
+ */
+class GameListWorker : public QObject, public QRunnable {
+    Q_OBJECT
+
+public:
+    GameListWorker(std::shared_ptr<FileSys::VfsFilesystem> vfs, QString dir_path, bool deep_scan,
+                   const CompatibilityList& compatibility_list);
+    ~GameListWorker() override;
+
+    /// Starts the processing of directory tree information.
+    void run() override;
+
+    /// Tells the worker that it should no longer continue processing. Thread-safe.
+    void Cancel();
+
+signals:
+    /**
+     * The `EntryReady` signal is emitted once an entry has been prepared and is ready
+     * to be added to the game list.
+     * @param entry_items a list with `QStandardItem`s that make up the columns of the new entry.
+     */
+    void EntryReady(QList<QStandardItem*> entry_items);
+
+    /**
+     * After the worker has traversed the game directory looking for entries, this signal is emitted
+     * with a list of folders that should be watched for changes as well.
+     */
+    void Finished(QStringList watch_list);
+
+private:
+    void AddInstalledTitlesToGameList();
+    void FillControlMap(const std::string& dir_path);
+    void AddFstEntriesToGameList(const std::string& dir_path, unsigned int recursion = 0);
+
+    std::shared_ptr<FileSys::VfsFilesystem> vfs;
+    std::map<u64, std::shared_ptr<FileSys::NCA>> nca_control_map;
+    QStringList watch_list;
+    QString dir_path;
+    bool deep_scan;
+    const CompatibilityList& compatibility_list;
+    std::atomic_bool stop_processing;
+};

src/yuzu/main.cpp

@@ -47,6 +47,7 @@
 #include "video_core/debug_utils/debug_utils.h"
 #include "yuzu/about_dialog.h"
 #include "yuzu/bootmanager.h"
+#include "yuzu/compatibility_list.h"
 #include "yuzu/configuration/config.h"
 #include "yuzu/configuration/configure_dialog.h"
 #include "yuzu/debugger/console.h"
@@ -136,11 +137,11 @@ GMainWindow::GMainWindow()
 
     ConnectMenuEvents();
     ConnectWidgetEvents();
-    LOG_INFO(Frontend, "yuzu Version: {} | {}-{}", Common::g_build_name, Common::g_scm_branch,
+    LOG_INFO(Frontend, "yuzu Version: {} | {}-{}", Common::g_build_fullname, Common::g_scm_branch,
              Common::g_scm_desc);
 
     setWindowTitle(QString("yuzu %1| %2-%3")
-                       .arg(Common::g_build_name, Common::g_scm_branch, Common::g_scm_desc));
+                       .arg(Common::g_build_fullname, Common::g_scm_branch, Common::g_scm_desc));
     show();
 
     // Necessary to load titles from nand in gamelist.
@@ -444,6 +445,8 @@ QStringList GMainWindow::GetUnsupportedGLExtensions() {
         unsupported_ext.append("ARB_vertex_type_10f_11f_11f_rev");
     if (!GLAD_GL_ARB_texture_mirror_clamp_to_edge)
         unsupported_ext.append("ARB_texture_mirror_clamp_to_edge");
+    if (!GLAD_GL_ARB_base_instance)
+        unsupported_ext.append("ARB_base_instance");
 
     // Extensions required to support some texture formats.
     if (!GLAD_GL_EXT_texture_compression_s3tc)
@@ -606,7 +609,7 @@ void GMainWindow::BootGame(const QString& filename) {
     }
 
     setWindowTitle(QString("yuzu %1| %4 | %2-%3")
-                       .arg(Common::g_build_name, Common::g_scm_branch, Common::g_scm_desc,
+                       .arg(Common::g_build_fullname, Common::g_scm_branch, Common::g_scm_desc,
                             QString::fromStdString(title_name)));
 
     render_window->show();
@@ -641,7 +644,7 @@ void GMainWindow::ShutdownGame() {
     game_list->show();
     game_list->setFilterFocus();
     setWindowTitle(QString("yuzu %1| %2-%3")
-                       .arg(Common::g_build_name, Common::g_scm_branch, Common::g_scm_desc));
+                       .arg(Common::g_build_fullname, Common::g_scm_branch, Common::g_scm_desc));
 
     // Disable status bar updates
     status_bar_update_timer.stop();
@@ -723,14 +726,11 @@ void GMainWindow::OnGameListOpenFolder(u64 program_id, GameListOpenTarget target
     QDesktopServices::openUrl(QUrl::fromLocalFile(qpath));
 }
 
-void GMainWindow::OnGameListNavigateToGamedbEntry(
-    u64 program_id,
-    std::unordered_map<std::string, std::pair<QString, QString>>& compatibility_list) {
-
-    auto it = FindMatchingCompatibilityEntry(compatibility_list, program_id);
+void GMainWindow::OnGameListNavigateToGamedbEntry(u64 program_id,
+                                                  const CompatibilityList& compatibility_list) {
+    const auto it = FindMatchingCompatibilityEntry(compatibility_list, program_id);
 
     QString directory;
-
     if (it != compatibility_list.end())
         directory = it->second.second;
 

src/yuzu/main.h

@@ -13,6 +13,7 @@
 #include "common/common_types.h"
 #include "core/core.h"
 #include "ui_main.h"
+#include "yuzu/compatibility_list.h"
 #include "yuzu/hotkeys.h"
 
 class Config;
@@ -137,9 +138,8 @@ private slots:
     /// Called whenever a user selects a game in the game list widget.
     void OnGameListLoadFile(QString game_path);
     void OnGameListOpenFolder(u64 program_id, GameListOpenTarget target);
-    void OnGameListNavigateToGamedbEntry(
-        u64 program_id,
-        std::unordered_map<std::string, std::pair<QString, QString>>& compatibility_list);
+    void OnGameListNavigateToGamedbEntry(u64 program_id,
+                                         const CompatibilityList& compatibility_list);
     void OnMenuLoadFile();
     void OnMenuLoadFolder();
     void OnMenuInstallToNAND();

src/yuzu_cmd/emu_window/emu_window_sdl2.cpp

@@ -91,6 +91,8 @@ bool EmuWindow_SDL2::SupportsRequiredGLExtensions() {
         unsupported_ext.push_back("ARB_vertex_type_10f_11f_11f_rev");
     if (!GLAD_GL_ARB_texture_mirror_clamp_to_edge)
         unsupported_ext.push_back("ARB_texture_mirror_clamp_to_edge");
+    if (!GLAD_GL_ARB_base_instance)
+        unsupported_ext.push_back("ARB_base_instance");
 
     // Extensions required to support some texture formats.
     if (!GLAD_GL_EXT_texture_compression_s3tc)
@@ -128,7 +130,7 @@ EmuWindow_SDL2::EmuWindow_SDL2(bool fullscreen) {
     SDL_GL_SetAttribute(SDL_GL_BLUE_SIZE, 8);
     SDL_GL_SetAttribute(SDL_GL_ALPHA_SIZE, 0);
 
-    std::string window_title = fmt::format("yuzu {} | {}-{}", Common::g_build_name,
+    std::string window_title = fmt::format("yuzu {} | {}-{}", Common::g_build_fullname,
                                            Common::g_scm_branch, Common::g_scm_desc);
     render_window =
         SDL_CreateWindow(window_title.c_str(),
@@ -166,7 +168,8 @@ EmuWindow_SDL2::EmuWindow_SDL2(bool fullscreen) {
     OnResize();
     OnMinimalClientAreaChangeRequest(GetActiveConfig().min_client_area_size);
     SDL_PumpEvents();
-    LOG_INFO(Frontend, "yuzu Version: {} | {}-{}", Common::g_build_name, Common::g_scm_branch,
+    SDL_GL_SetSwapInterval(false);
+    LOG_INFO(Frontend, "yuzu Version: {} | {}-{}", Common::g_build_fullname, Common::g_scm_branch,
              Common::g_scm_desc);
 
     DoneCurrent();

src/yuzu_cmd/yuzu.cpp

@@ -82,6 +82,9 @@ int main(int argc, char** argv) {
     int option_index = 0;
     bool use_gdbstub = Settings::values.use_gdbstub;
     u32 gdb_port = static_cast<u32>(Settings::values.gdbstub_port);
+
+    InitializeLogging();
+
     char* endarg;
 #ifdef _WIN32
     int argc_w;
@@ -144,8 +147,6 @@ int main(int argc, char** argv) {
     LocalFree(argv_w);
 #endif
 
-    InitializeLogging();
-
     MicroProfileOnThreadCreate("EmuThread");
     SCOPE_EXIT({ MicroProfileShutdown(); });