renderer_vulkan: Integrate Nvidia Nsight Aftermath on Windows

Adds optional support for Nsight Aftermath. It is enabled through
ENABLE_NSIGHT_AFTERMATH in cmake. A path to the SDK has to be provided
by the environment variable NSIGHT_AFTERMATH_SDK.

Nsight Aftermath allows an application to generate "minidumps" of the
GPU state when a device loss happens. By analysing these on Nsight we
can know what a game was doing and why it triggered a device loss.

The dump is generated inside %APPDATA%\yuzu\log\gpucrash and this
directory is deleted every time a new instance is initialized with
Nsight enabled.

To enable it on yuzu there has a to be a driver and device capable of
running Nsight Aftermath on Vulkan. That means only Turing based GPUs
on the latest stable driver, beta drivers won't work for now.

It is manually enabled in Configuration>Debug>Enable Graphics Debugging
because when using all debugging capabilities there is a runtime cost.

.ci/templates/build-msvc.yml

@@ -4,7 +4,7 @@ parameters:
   version: ''
 
 steps:
-- script: mkdir build && cd build && cmake -G "Visual Studio 15 2017 Win64" --config Release -DYUZU_USE_BUNDLED_QT=1 -DYUZU_USE_BUNDLED_SDL2=1 -DYUZU_USE_BUNDLED_UNICORN=1 -DYUZU_USE_QT_WEB_ENGINE=ON -DENABLE_COMPATIBILITY_LIST_DOWNLOAD=ON -DYUZU_ENABLE_COMPATIBILITY_REPORTING=${COMPAT} -DUSE_DISCORD_PRESENCE=ON -DDISPLAY_VERSION=${{ parameters['version'] }} .. && cd ..
+- script: mkdir build && cd build && cmake -G "Visual Studio 16 2019" -A x64 --config Release -DYUZU_USE_BUNDLED_QT=1 -DYUZU_USE_BUNDLED_SDL2=1 -DYUZU_USE_BUNDLED_UNICORN=1 -DYUZU_USE_QT_WEB_ENGINE=ON -DENABLE_COMPATIBILITY_LIST_DOWNLOAD=ON -DYUZU_ENABLE_COMPATIBILITY_REPORTING=${COMPAT} -DUSE_DISCORD_PRESENCE=ON -DDISPLAY_VERSION=${{ parameters['version'] }} .. && cd ..
   displayName: 'Configure CMake'
 - task: MSBuild@1
   displayName: 'Build'

.ci/yuzu-mainline-step2.yml

@@ -45,7 +45,7 @@ stages:
   - job: build
     displayName: 'msvc'
     pool:
-      vmImage: vs2017-win2016
+      vmImage: windows-2019
     steps:
     - template: ./templates/sync-source.yml
       parameters:

.ci/yuzu-patreon-step2.yml

@@ -22,7 +22,7 @@ stages:
   - job: build
     displayName: 'windows-msvc'
     pool:
-      vmImage: vs2017-win2016
+      vmImage: windows-2019
     steps:
     - template: ./templates/sync-source.yml
       parameters:

CMakeModules/CopyYuzuQt5Deps.cmake

@@ -6,9 +6,9 @@ function(copy_yuzu_Qt5_deps target_dir)
     set(Qt5_STYLES_DIR "${Qt5_DIR}/../../../plugins/styles/")
     set(Qt5_IMAGEFORMATS_DIR "${Qt5_DIR}/../../../plugins/imageformats/")
     set(Qt5_RESOURCES_DIR "${Qt5_DIR}/../../../resources/")
-    set(PLATFORMS ${DLL_DEST}platforms/)
-    set(STYLES ${DLL_DEST}styles/)
-    set(IMAGEFORMATS ${DLL_DEST}imageformats/)
+    set(PLATFORMS ${DLL_DEST}plugins/platforms/)
+    set(STYLES ${DLL_DEST}plugins/styles/)
+    set(IMAGEFORMATS ${DLL_DEST}plugins/imageformats/)
     windows_copy_files(${target_dir} ${Qt5_DLL_DIR} ${DLL_DEST}
         icudt*.dll
         icuin*.dll
@@ -42,11 +42,15 @@ function(copy_yuzu_Qt5_deps target_dir)
             icudtl.dat
         )
     endif ()
-
     windows_copy_files(yuzu ${Qt5_PLATFORMS_DIR} ${PLATFORMS} qwindows$<$<CONFIG:Debug>:d>.*)
     windows_copy_files(yuzu ${Qt5_STYLES_DIR} ${STYLES} qwindowsvistastyle$<$<CONFIG:Debug>:d>.*)
     windows_copy_files(yuzu ${Qt5_IMAGEFORMATS_DIR} ${IMAGEFORMATS}
         qjpeg$<$<CONFIG:Debug>:d>.*
         qgif$<$<CONFIG:Debug>:d>.*
         )
+    # Create an empty qt.conf file. Qt will detect that this file exists, and use the folder that its in as the root folder.
+    # This way it'll look for plugins in the root/plugins/ folder
+    add_custom_command(TARGET yuzu POST_BUILD
+        COMMAND ${CMAKE_COMMAND} -E touch ${DLL_DEST}qt.conf
+    )
 endfunction(copy_yuzu_Qt5_deps)

CMakeModules/GenerateSCMRev.cmake

@@ -57,8 +57,6 @@ set(HASH_FILES
     "${VIDEO_CORE}/renderer_opengl/gl_shader_decompiler.h"
     "${VIDEO_CORE}/renderer_opengl/gl_shader_disk_cache.cpp"
     "${VIDEO_CORE}/renderer_opengl/gl_shader_disk_cache.h"
-    "${VIDEO_CORE}/renderer_opengl/gl_shader_gen.cpp"
-    "${VIDEO_CORE}/renderer_opengl/gl_shader_gen.h"
     "${VIDEO_CORE}/shader/decode/arithmetic.cpp"
     "${VIDEO_CORE}/shader/decode/arithmetic_half.cpp"
     "${VIDEO_CORE}/shader/decode/arithmetic_half_immediate.cpp"
@@ -91,8 +89,6 @@ set(HASH_FILES
     "${VIDEO_CORE}/shader/ast.h"
     "${VIDEO_CORE}/shader/compiler_settings.cpp"
     "${VIDEO_CORE}/shader/compiler_settings.h"
-    "${VIDEO_CORE}/shader/const_buffer_locker.cpp"
-    "${VIDEO_CORE}/shader/const_buffer_locker.h"
     "${VIDEO_CORE}/shader/control_flow.cpp"
     "${VIDEO_CORE}/shader/control_flow.h"
     "${VIDEO_CORE}/shader/decode.cpp"
@@ -101,9 +97,13 @@ set(HASH_FILES
     "${VIDEO_CORE}/shader/node.h"
     "${VIDEO_CORE}/shader/node_helper.cpp"
     "${VIDEO_CORE}/shader/node_helper.h"
+    "${VIDEO_CORE}/shader/registry.cpp"
+    "${VIDEO_CORE}/shader/registry.h"
     "${VIDEO_CORE}/shader/shader_ir.cpp"
     "${VIDEO_CORE}/shader/shader_ir.h"
     "${VIDEO_CORE}/shader/track.cpp"
+    "${VIDEO_CORE}/shader/transform_feedback.cpp"
+    "${VIDEO_CORE}/shader/transform_feedback.h"
 )
 set(COMBINED "")
 foreach (F IN LISTS HASH_FILES)

README.md

@@ -2,6 +2,7 @@ yuzu emulator
 =============
 [![Travis CI Build Status](https://travis-ci.com/yuzu-emu/yuzu.svg?branch=master)](https://travis-ci.com/yuzu-emu/yuzu)
 [![Azure Mainline CI Build Status](https://dev.azure.com/yuzu-emu/yuzu/_apis/build/status/yuzu%20mainline?branchName=master)](https://dev.azure.com/yuzu-emu/yuzu/)
+[![Discord](https://img.shields.io/discord/398318088170242053?color=%237289DA&label=yuzu&logo=discord&logoColor=white)](https://discord.gg/XQV6dn9)
 
 yuzu is an experimental open-source emulator for the Nintendo Switch from the creators of [Citra](https://citra-emu.org/).
 
@@ -21,7 +22,7 @@ For development discussion, please join us on [Discord](https://discord.gg/XQV6d
 
 Most of the development happens on GitHub. It's also where [our central repository](https://github.com/yuzu-emu/yuzu) is hosted.
 
-If you want to contribute please take a look at the [Contributor's Guide](CONTRIBUTING.md) and [Developer Information](https://github.com/yuzu-emu/yuzu/wiki/Developer-Information). You should also contact any of the developers on Discord in order to know about the current state of the emulator.
+If you want to contribute please take a look at the [Contributor's Guide](https://github.com/yuzu-emu/yuzu/wiki/Contributing) and [Developer Information](https://github.com/yuzu-emu/yuzu/wiki/Developer-Information). You should also contact any of the developers on Discord in order to know about the current state of the emulator.
 
 ### Building
 

dist/license.md

@@ -2,8 +2,8 @@ The icons in this folder and its subfolders have the following licenses:
 
 Icon Name | License | Origin/Author
 --- | --- | ---
-qt_themes/default/icons/16x16/checked.png | Free for non-commercial use
-qt_themes/default/icons/16x16/failed.png | Free for non-commercial use
+qt_themes/default/icons/16x16/checked.png | CC BY-ND 3.0 | https://icons8.com
+qt_themes/default/icons/16x16/failed.png | CC BY-ND 3.0 | https://icons8.com
 qt_themes/default/icons/16x16/lock.png | CC BY-ND 3.0 | https://icons8.com
 qt_themes/default/icons/256x256/plus_folder.png | CC BY-ND 3.0 | https://icons8.com
 qt_themes/default/icons/48x48/bad_folder.png | CC BY-ND 3.0 | https://icons8.com
@@ -11,8 +11,6 @@ qt_themes/default/icons/48x48/chip.png | CC BY-ND 3.0 | https://icons8.com
 qt_themes/default/icons/48x48/folder.png | CC BY-ND 3.0 | https://icons8.com
 qt_themes/default/icons/48x48/plus.png | CC0 1.0 | Designed by BreadFish64 from the Citra team
 qt_themes/default/icons/48x48/sd_card.png | CC BY-ND 3.0 | https://icons8.com
-qt_themes/qdarkstyle/icons/16x16/checked.png | Free for non-commercial use
-qt_themes/qdarkstyle/icons/16x16/failed.png | Free for non-commercial use
 qt_themes/qdarkstyle/icons/16x16/lock.png | CC BY-ND 3.0 | https://icons8.com
 qt_themes/qdarkstyle/icons/256x256/plus_folder.png | CC BY-ND 3.0 | https://icons8.com
 qt_themes/qdarkstyle/icons/48x48/bad_folder.png | CC BY-ND 3.0 | https://icons8.com

externals/microprofile/microprofile.h

@@ -243,6 +243,7 @@ typedef uint32_t ThreadIdType;
 #define MICROPROFILE_DEFINE_GPU(var, name, color) MicroProfileToken g_mp_##var = MicroProfileGetToken("GPU", name, color, MicroProfileTokenTypeGpu)
 #define MICROPROFILE_TOKEN_PASTE0(a, b) a ## b
 #define MICROPROFILE_TOKEN_PASTE(a, b)  MICROPROFILE_TOKEN_PASTE0(a,b)
+#define MICROPROFILE_TOKEN(var) g_mp_##var
 #define MICROPROFILE_SCOPE(var) MicroProfileScopeHandler MICROPROFILE_TOKEN_PASTE(foo, __LINE__)(g_mp_##var)
 #define MICROPROFILE_SCOPE_TOKEN(token) MicroProfileScopeHandler MICROPROFILE_TOKEN_PASTE(foo, __LINE__)(token)
 #define MICROPROFILE_SCOPEI(group, name, color) static MicroProfileToken MICROPROFILE_TOKEN_PASTE(g_mp,__LINE__) = MicroProfileGetToken(group, name, color, MicroProfileTokenTypeCpu); MicroProfileScopeHandler MICROPROFILE_TOKEN_PASTE(foo,__LINE__)( MICROPROFILE_TOKEN_PASTE(g_mp,__LINE__))
@@ -827,7 +828,7 @@ inline MicroProfileLogEntry MicroProfileMakeLogIndex(uint64_t nBegin, MicroProfi
     MicroProfileLogEntry Entry =  (nBegin<<62) | ((0x3fff&nToken)<<48) | (MP_LOG_TICK_MASK&nTick);
     int t = MicroProfileLogType(Entry);
     uint64_t nTimerIndex = MicroProfileLogTimerIndex(Entry);
-    MP_ASSERT(t == nBegin);
+    MP_ASSERT((uint64_t)t == nBegin);
     MP_ASSERT(nTimerIndex == (nToken&0x3fff));
     return Entry;
 
@@ -1555,10 +1556,10 @@ void MicroProfileFlip()
 
         pFramePut->nFrameStartCpu = MP_TICK();
         pFramePut->nFrameStartGpu = (uint32_t)MicroProfileGpuInsertTimeStamp();
-        if(pFrameNext->nFrameStartGpu != (uint64_t)-1)
+        if(pFrameNext->nFrameStartGpu != -1)
             pFrameNext->nFrameStartGpu = MicroProfileGpuGetTimeStamp((uint32_t)pFrameNext->nFrameStartGpu);
 
-        if(pFrameCurrent->nFrameStartGpu == (uint64_t)-1)
+        if(pFrameCurrent->nFrameStartGpu == -1)
             pFrameCurrent->nFrameStartGpu = pFrameNext->nFrameStartGpu + 1;
 
         uint64_t nFrameStartCpu = pFrameCurrent->nFrameStartCpu;

license.txt

@@ -343,8 +343,8 @@ The icons used in this project have the following licenses:
 
 Icon Name                                   | License       | Origin/Author
 ---                                         | ---           | ---
-checked.png                                 | Free for non-commercial use
-failed.png                                  | Free for non-commercial use
+checked.png                                 | CC BY-ND 3.0  | https://icons8.com
+failed.png                                  | CC BY-ND 3.0  | https://icons8.com
 lock.png                                    | CC BY-ND 3.0  | https://icons8.com
 plus_folder.png (Default, Dark)             | CC BY-ND 3.0  | https://icons8.com
 bad_folder.png (Default, Dark)              | CC BY-ND 3.0  | https://icons8.com

src/audio_core/audio_renderer.h

@@ -189,7 +189,7 @@ struct UpdateDataHeader {
     UpdateDataHeader() {}
 
     explicit UpdateDataHeader(const AudioRendererParameter& config) {
-        revision = Common::MakeMagic('R', 'E', 'V', '4'); // 5.1.0 Revision
+        revision = Common::MakeMagic('R', 'E', 'V', '8'); // 9.2.0 Revision
         behavior_size = 0xb0;
         memory_pools_size = (config.effect_count + (config.voice_count * 4)) * 0x10;
         voices_size = config.voice_count * 0x10;

src/audio_core/cubeb_sink.cpp

@@ -8,6 +8,7 @@
 #include "audio_core/cubeb_sink.h"
 #include "audio_core/stream.h"
 #include "audio_core/time_stretch.h"
+#include "common/assert.h"
 #include "common/logging/log.h"
 #include "common/ring_buffer.h"
 #include "core/settings.h"
@@ -65,12 +66,25 @@ public:
     void EnqueueSamples(u32 source_num_channels, const std::vector<s16>& samples) override {
         if (source_num_channels > num_channels) {
             // Downsample 6 channels to 2
+            ASSERT_MSG(source_num_channels == 6, "Channel count must be 6");
+
             std::vector<s16> buf;
             buf.reserve(samples.size() * num_channels / source_num_channels);
             for (std::size_t i = 0; i < samples.size(); i += source_num_channels) {
-                for (std::size_t ch = 0; ch < num_channels; ch++) {
-                    buf.push_back(samples[i + ch]);
-                }
+                // Downmixing implementation taken from the ATSC standard
+                const s16 left{samples[i + 0]};
+                const s16 right{samples[i + 1]};
+                const s16 center{samples[i + 2]};
+                const s16 surround_left{samples[i + 4]};
+                const s16 surround_right{samples[i + 5]};
+                // Not used in the ATSC reference implementation
+                [[maybe_unused]] const s16 low_frequency_effects { samples[i + 3] };
+
+                constexpr s32 clev{707}; // center mixing level coefficient
+                constexpr s32 slev{707}; // surround mixing level coefficient
+
+                buf.push_back(left + (clev * center / 1000) + (slev * surround_left / 1000));
+                buf.push_back(right + (clev * center / 1000) + (slev * surround_right / 1000));
             }
             queue.Push(buf);
             return;

src/common/CMakeLists.txt

@@ -38,8 +38,6 @@ add_custom_command(OUTPUT scm_rev.cpp
       "${VIDEO_CORE}/renderer_opengl/gl_shader_decompiler.h"
       "${VIDEO_CORE}/renderer_opengl/gl_shader_disk_cache.cpp"
       "${VIDEO_CORE}/renderer_opengl/gl_shader_disk_cache.h"
-      "${VIDEO_CORE}/renderer_opengl/gl_shader_gen.cpp"
-      "${VIDEO_CORE}/renderer_opengl/gl_shader_gen.h"
       "${VIDEO_CORE}/shader/decode/arithmetic.cpp"
       "${VIDEO_CORE}/shader/decode/arithmetic_half.cpp"
       "${VIDEO_CORE}/shader/decode/arithmetic_half_immediate.cpp"
@@ -72,8 +70,6 @@ add_custom_command(OUTPUT scm_rev.cpp
       "${VIDEO_CORE}/shader/ast.h"
       "${VIDEO_CORE}/shader/compiler_settings.cpp"
       "${VIDEO_CORE}/shader/compiler_settings.h"
-      "${VIDEO_CORE}/shader/const_buffer_locker.cpp"
-      "${VIDEO_CORE}/shader/const_buffer_locker.h"
       "${VIDEO_CORE}/shader/control_flow.cpp"
       "${VIDEO_CORE}/shader/control_flow.h"
       "${VIDEO_CORE}/shader/decode.cpp"
@@ -82,9 +78,13 @@ add_custom_command(OUTPUT scm_rev.cpp
       "${VIDEO_CORE}/shader/node.h"
       "${VIDEO_CORE}/shader/node_helper.cpp"
       "${VIDEO_CORE}/shader/node_helper.h"
+      "${VIDEO_CORE}/shader/registry.cpp"
+      "${VIDEO_CORE}/shader/registry.h"
       "${VIDEO_CORE}/shader/shader_ir.cpp"
       "${VIDEO_CORE}/shader/shader_ir.h"
       "${VIDEO_CORE}/shader/track.cpp"
+      "${VIDEO_CORE}/shader/transform_feedback.cpp"
+      "${VIDEO_CORE}/shader/transform_feedback.h"
       # and also check that the scm_rev files haven't changed
       "${CMAKE_CURRENT_SOURCE_DIR}/scm_rev.cpp.in"
       "${CMAKE_CURRENT_SOURCE_DIR}/scm_rev.h"
@@ -106,6 +106,8 @@ add_library(common STATIC
     common_funcs.h
     common_paths.h
     common_types.h
+    dynamic_library.cpp
+    dynamic_library.h
     file_util.cpp
     file_util.h
     hash.h

src/common/dynamic_library.cpp

@@ -0,0 +1,106 @@
+// Copyright 2019 Dolphin Emulator Project
+// Licensed under GPLv2+
+// Refer to the license.txt file included.
+
+#include <cstring>
+#include <string>
+#include <utility>
+
+#include <fmt/format.h>
+
+#include "common/dynamic_library.h"
+
+#ifdef _WIN32
+#include <windows.h>
+#else
+#include <dlfcn.h>
+#endif
+
+namespace Common {
+
+DynamicLibrary::DynamicLibrary() = default;
+
+DynamicLibrary::DynamicLibrary(const char* filename) {
+    Open(filename);
+}
+
+DynamicLibrary::DynamicLibrary(DynamicLibrary&& rhs) noexcept
+    : handle{std::exchange(rhs.handle, nullptr)} {}
+
+DynamicLibrary& DynamicLibrary::operator=(DynamicLibrary&& rhs) noexcept {
+    Close();
+    handle = std::exchange(rhs.handle, nullptr);
+    return *this;
+}
+
+DynamicLibrary::~DynamicLibrary() {
+    Close();
+}
+
+std::string DynamicLibrary::GetUnprefixedFilename(const char* filename) {
+#if defined(_WIN32)
+    return std::string(filename) + ".dll";
+#elif defined(__APPLE__)
+    return std::string(filename) + ".dylib";
+#else
+    return std::string(filename) + ".so";
+#endif
+}
+
+std::string DynamicLibrary::GetVersionedFilename(const char* libname, int major, int minor) {
+#if defined(_WIN32)
+    if (major >= 0 && minor >= 0)
+        return fmt::format("{}-{}-{}.dll", libname, major, minor);
+    else if (major >= 0)
+        return fmt::format("{}-{}.dll", libname, major);
+    else
+        return fmt::format("{}.dll", libname);
+#elif defined(__APPLE__)
+    const char* prefix = std::strncmp(libname, "lib", 3) ? "lib" : "";
+    if (major >= 0 && minor >= 0)
+        return fmt::format("{}{}.{}.{}.dylib", prefix, libname, major, minor);
+    else if (major >= 0)
+        return fmt::format("{}{}.{}.dylib", prefix, libname, major);
+    else
+        return fmt::format("{}{}.dylib", prefix, libname);
+#else
+    const char* prefix = std::strncmp(libname, "lib", 3) ? "lib" : "";
+    if (major >= 0 && minor >= 0)
+        return fmt::format("{}{}.so.{}.{}", prefix, libname, major, minor);
+    else if (major >= 0)
+        return fmt::format("{}{}.so.{}", prefix, libname, major);
+    else
+        return fmt::format("{}{}.so", prefix, libname);
+#endif
+}
+
+bool DynamicLibrary::Open(const char* filename) {
+#ifdef _WIN32
+    handle = reinterpret_cast<void*>(LoadLibraryA(filename));
+#else
+    handle = dlopen(filename, RTLD_NOW);
+#endif
+    return handle != nullptr;
+}
+
+void DynamicLibrary::Close() {
+    if (!IsOpen())
+        return;
+
+#ifdef _WIN32
+    FreeLibrary(reinterpret_cast<HMODULE>(handle));
+#else
+    dlclose(handle);
+#endif
+    handle = nullptr;
+}
+
+void* DynamicLibrary::GetSymbolAddress(const char* name) const {
+#ifdef _WIN32
+    return reinterpret_cast<void*>(GetProcAddress(reinterpret_cast<HMODULE>(handle), name));
+#else
+    return reinterpret_cast<void*>(dlsym(handle, name));
+#endif
+}
+
+} // namespace Common

src/common/dynamic_library.h

@@ -0,0 +1,75 @@
+// Copyright 2019 Dolphin Emulator Project
+// Licensed under GPLv2+
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <string>
+
+namespace Common {
+
+/**
+ * Provides a platform-independent interface for loading a dynamic library and retrieving symbols.
+ * The interface maintains an internal reference count to allow one handle to be shared between
+ * multiple users.
+ */
+class DynamicLibrary final {
+public:
+    /// Default constructor, does not load a library.
+    explicit DynamicLibrary();
+
+    /// Automatically loads the specified library. Call IsOpen() to check validity before use.
+    explicit DynamicLibrary(const char* filename);
+
+    /// Moves the library.
+    DynamicLibrary(DynamicLibrary&&) noexcept;
+    DynamicLibrary& operator=(DynamicLibrary&&) noexcept;
+
+    /// Delete copies, we can't copy a dynamic library.
+    DynamicLibrary(const DynamicLibrary&) = delete;
+    DynamicLibrary& operator=(const DynamicLibrary&) = delete;
+
+    /// Closes the library.
+    ~DynamicLibrary();
+
+    /// Returns the specified library name with the platform-specific suffix added.
+    static std::string GetUnprefixedFilename(const char* filename);
+
+    /// Returns the specified library name in platform-specific format.
+    /// Major/minor versions will not be included if set to -1.
+    /// If libname already contains the "lib" prefix, it will not be added again.
+    /// Windows: LIBNAME-MAJOR-MINOR.dll
+    /// Linux: libLIBNAME.so.MAJOR.MINOR
+    /// Mac: libLIBNAME.MAJOR.MINOR.dylib
+    static std::string GetVersionedFilename(const char* libname, int major = -1, int minor = -1);
+
+    /// Returns true if a module is loaded, otherwise false.
+    bool IsOpen() const {
+        return handle != nullptr;
+    }
+
+    /// Loads (or replaces) the handle with the specified library file name.
+    /// Returns true if the library was loaded and can be used.
+    bool Open(const char* filename);
+
+    /// Unloads the library, any function pointers from this library are no longer valid.
+    void Close();
+
+    /// Returns the address of the specified symbol (function or variable) as an untyped pointer.
+    /// If the specified symbol does not exist in this library, nullptr is returned.
+    void* GetSymbolAddress(const char* name) const;
+
+    /// Obtains the address of the specified symbol, automatically casting to the correct type.
+    /// Returns true if the symbol was found and assigned, otherwise false.
+    template <typename T>
+    bool GetSymbol(const char* name, T* ptr) const {
+        *ptr = reinterpret_cast<T>(GetSymbolAddress(name));
+        return *ptr != nullptr;
+    }
+
+private:
+    /// Platform-dependent data type representing a dynamic library handle.
+    void* handle = nullptr;
+};
+
+} // namespace Common

src/common/file_util.cpp

@@ -3,6 +3,7 @@
 // Refer to the license.txt file included.
 
 #include <array>
+#include <limits>
 #include <memory>
 #include <sstream>
 #include <unordered_map>
@@ -530,11 +531,11 @@ void CopyDir(const std::string& source_path, const std::string& dest_path) {
 std::optional<std::string> GetCurrentDir() {
 // Get the current working directory (getcwd uses malloc)
 #ifdef _WIN32
-    wchar_t* dir;
-    if (!(dir = _wgetcwd(nullptr, 0))) {
+    wchar_t* dir = _wgetcwd(nullptr, 0);
+    if (!dir) {
 #else
-    char* dir;
-    if (!(dir = getcwd(nullptr, 0))) {
+    char* dir = getcwd(nullptr, 0);
+    if (!dir) {
 #endif
         LOG_ERROR(Common_Filesystem, "GetCurrentDirectory failed: {}", GetLastErrorMsg());
         return {};
@@ -918,19 +919,22 @@ void IOFile::Swap(IOFile& other) noexcept {
 
 bool IOFile::Open(const std::string& filename, const char openmode[], int flags) {
     Close();
+    bool m_good;
 #ifdef _WIN32
     if (flags != 0) {
         m_file = _wfsopen(Common::UTF8ToUTF16W(filename).c_str(),
                           Common::UTF8ToUTF16W(openmode).c_str(), flags);
+        m_good = m_file != nullptr;
     } else {
-        _wfopen_s(&m_file, Common::UTF8ToUTF16W(filename).c_str(),
-                  Common::UTF8ToUTF16W(openmode).c_str());
+        m_good = _wfopen_s(&m_file, Common::UTF8ToUTF16W(filename).c_str(),
+                           Common::UTF8ToUTF16W(openmode).c_str()) == 0;
     }
 #else
-    m_file = fopen(filename.c_str(), openmode);
+    m_file = std::fopen(filename.c_str(), openmode);
+    m_good = m_file != nullptr;
 #endif
 
-    return IsOpen();
+    return m_good;
 }
 
 bool IOFile::Close() {
@@ -956,7 +960,7 @@ u64 IOFile::Tell() const {
     if (IsOpen())
         return ftello(m_file);
 
-    return -1;
+    return std::numeric_limits<u64>::max();
 }
 
 bool IOFile::Flush() {

src/common/math_util.h

@@ -24,17 +24,29 @@ struct Rectangle {
         : left(left), top(top), right(right), bottom(bottom) {}
 
     T GetWidth() const {
-        return std::abs(static_cast<std::make_signed_t<T>>(right - left));
+        if constexpr (std::is_floating_point_v<T>) {
+            return std::abs(right - left);
+        } else {
+            return std::abs(static_cast<std::make_signed_t<T>>(right - left));
+        }
     }
+
     T GetHeight() const {
-        return std::abs(static_cast<std::make_signed_t<T>>(bottom - top));
+        if constexpr (std::is_floating_point_v<T>) {
+            return std::abs(bottom - top);
+        } else {
+            return std::abs(static_cast<std::make_signed_t<T>>(bottom - top));
+        }
     }
+
     Rectangle<T> TranslateX(const T x) const {
         return Rectangle{left + x, top, right + x, bottom};
     }
+
     Rectangle<T> TranslateY(const T y) const {
         return Rectangle{left, top + y, right, bottom + y};
     }
+
     Rectangle<T> Scale(const float s) const {
         return Rectangle{left, top, static_cast<T>(left + GetWidth() * s),
                          static_cast<T>(top + GetHeight() * s)};

src/common/page_table.cpp

@@ -16,7 +16,6 @@ void PageTable::Resize(std::size_t address_space_width_in_bits) {
 
     pointers.resize(num_page_table_entries);
     attributes.resize(num_page_table_entries);
-    backing_addr.resize(num_page_table_entries);
 
     // The default is a 39-bit address space, which causes an initial 1GB allocation size. If the
     // vector size is subsequently decreased (via resize), the vector might not automatically
@@ -25,6 +24,17 @@ void PageTable::Resize(std::size_t address_space_width_in_bits) {
 
     pointers.shrink_to_fit();
     attributes.shrink_to_fit();
+}
+
+BackingPageTable::BackingPageTable(std::size_t page_size_in_bits) : PageTable{page_size_in_bits} {}
+
+BackingPageTable::~BackingPageTable() = default;
+
+void BackingPageTable::Resize(std::size_t address_space_width_in_bits) {
+    PageTable::Resize(address_space_width_in_bits);
+    const std::size_t num_page_table_entries = 1ULL
+                                               << (address_space_width_in_bits - page_size_in_bits);
+    backing_addr.resize(num_page_table_entries);
     backing_addr.shrink_to_fit();
 }
 

src/common/page_table.h

@@ -76,9 +76,20 @@ struct PageTable {
      */
     std::vector<PageType> attributes;
 
-    std::vector<u64> backing_addr;
-
     const std::size_t page_size_in_bits{};
 };
 
+/**
+ * A more advanced Page Table with the ability to save a backing address when using it
+ * depends on another MMU.
+ */
+struct BackingPageTable : PageTable {
+    explicit BackingPageTable(std::size_t page_size_in_bits);
+    ~BackingPageTable();
+
+    void Resize(std::size_t address_space_width_in_bits);
+
+    std::vector<u64> backing_addr;
+};
+
 } // namespace Common

src/common/thread.cpp

@@ -28,11 +28,8 @@ namespace Common {
 #ifdef _MSC_VER
 
 // Sets the debugger-visible name of the current thread.
-// Uses undocumented (actually, it is now documented) trick.
-// http://msdn.microsoft.com/library/default.asp?url=/library/en-us/vsdebug/html/vxtsksettingthreadname.asp
-
-// This is implemented much nicer in upcoming msvc++, see:
-// http://msdn.microsoft.com/en-us/library/xcb2z8hs(VS.100).aspx
+// Uses trick documented in:
+// https://docs.microsoft.com/en-us/visualstudio/debugger/how-to-set-a-thread-name-in-native-code
 void SetCurrentThreadName(const char* name) {
     static const DWORD MS_VC_EXCEPTION = 0x406D1388;
 
@@ -47,7 +44,7 @@ void SetCurrentThreadName(const char* name) {
 
     info.dwType = 0x1000;
     info.szName = name;
-    info.dwThreadID = -1; // dwThreadID;
+    info.dwThreadID = std::numeric_limits<DWORD>::max();
     info.dwFlags = 0;
 
     __try {

src/core/CMakeLists.txt

@@ -131,8 +131,6 @@ add_library(core STATIC
     frontend/framebuffer_layout.cpp
     frontend/framebuffer_layout.h
     frontend/input.h
-    frontend/scope_acquire_context.cpp
-    frontend/scope_acquire_context.h
     gdbstub/gdbstub.cpp
     gdbstub/gdbstub.h
     hardware_interrupt_manager.cpp
@@ -287,6 +285,18 @@ add_library(core STATIC
     hle/service/btm/btm.h
     hle/service/caps/caps.cpp
     hle/service/caps/caps.h
+    hle/service/caps/caps_a.cpp
+    hle/service/caps/caps_a.h
+    hle/service/caps/caps_c.cpp
+    hle/service/caps/caps_c.h
+    hle/service/caps/caps_u.cpp
+    hle/service/caps/caps_u.h
+    hle/service/caps/caps_sc.cpp
+    hle/service/caps/caps_sc.h
+    hle/service/caps/caps_ss.cpp
+    hle/service/caps/caps_ss.h
+    hle/service/caps/caps_su.cpp
+    hle/service/caps/caps_su.h
     hle/service/erpt/erpt.cpp
     hle/service/erpt/erpt.h
     hle/service/es/es.cpp

src/core/core.cpp

@@ -24,7 +24,6 @@
 #include "core/file_sys/sdmc_factory.h"
 #include "core/file_sys/vfs_concat.h"
 #include "core/file_sys/vfs_real.h"
-#include "core/frontend/scope_acquire_context.h"
 #include "core/gdbstub/gdbstub.h"
 #include "core/hardware_interrupt_manager.h"
 #include "core/hle/kernel/client_port.h"
@@ -166,14 +165,14 @@ struct System::Impl {
         service_manager = std::make_shared<Service::SM::ServiceManager>();
 
         Service::Init(service_manager, system);
-        GDBStub::Init();
+        GDBStub::DeferStart();
 
-        renderer = VideoCore::CreateRenderer(emu_window, system);
-        if (!renderer->Init()) {
+        interrupt_manager = std::make_unique<Core::Hardware::InterruptManager>(system);
+        gpu_core = VideoCore::CreateGPU(emu_window, system);
+        if (!gpu_core) {
             return ResultStatus::ErrorVideoCore;
         }
-        interrupt_manager = std::make_unique<Core::Hardware::InterruptManager>(system);
-        gpu_core = VideoCore::CreateGPU(system);
+        gpu_core->Renderer().Rasterizer().SetupDirtyFlags();
 
         is_powered_on = true;
         exit_lock = false;
@@ -185,8 +184,6 @@ struct System::Impl {
 
     ResultStatus Load(System& system, Frontend::EmuWindow& emu_window,
                       const std::string& filepath) {
-        Core::Frontend::ScopeAcquireContext acquire_context{emu_window};
-
         app_loader = Loader::GetLoader(GetGameFileFromPath(virtual_filesystem, filepath));
         if (!app_loader) {
             LOG_CRITICAL(Core, "Failed to obtain loader for {}!", filepath);
@@ -215,10 +212,6 @@ struct System::Impl {
         AddGlueRegistrationForProcess(*app_loader, *main_process);
         kernel.MakeCurrentProcess(main_process.get());
 
-        // Main process has been loaded and been made current.
-        // Begin GPU and CPU execution.
-        gpu_core->Start();
-
         // Initialize cheat engine
         if (cheat_engine) {
             cheat_engine->Initialize();
@@ -276,7 +269,6 @@ struct System::Impl {
         }
 
         // Shutdown emulation session
-        renderer.reset();
         GDBStub::Shutdown();
         Service::Shutdown();
         service_manager.reset();
@@ -352,7 +344,6 @@ struct System::Impl {
     Service::FileSystem::FileSystemController fs_controller;
     /// AppLoader used to load the current executing application
     std::unique_ptr<Loader::AppLoader> app_loader;
-    std::unique_ptr<VideoCore::RendererBase> renderer;
     std::unique_ptr<Tegra::GPU> gpu_core;
     std::unique_ptr<Hardware::InterruptManager> interrupt_manager;
     Memory::Memory memory;
@@ -535,11 +526,11 @@ const Core::Hardware::InterruptManager& System::InterruptManager() const {
 }
 
 VideoCore::RendererBase& System::Renderer() {
-    return *impl->renderer;
+    return impl->gpu_core->Renderer();
 }
 
 const VideoCore::RendererBase& System::Renderer() const {
-    return *impl->renderer;
+    return impl->gpu_core->Renderer();
 }
 
 Kernel::KernelCore& System::Kernel() {

src/core/file_sys/romfs.cpp

@@ -5,6 +5,7 @@
 #include <memory>
 
 #include "common/common_types.h"
+#include "common/string_util.h"
 #include "common/swap.h"
 #include "core/file_sys/fsmitm_romfsbuild.h"
 #include "core/file_sys/romfs.h"
@@ -126,7 +127,7 @@ VirtualDir ExtractRomFS(VirtualFile file, RomFSExtractionType type) {
         return out->GetSubdirectories().front();
 
     while (out->GetSubdirectories().size() == 1 && out->GetFiles().empty()) {
-        if (out->GetSubdirectories().front()->GetName() == "data" &&
+        if (Common::ToLower(out->GetSubdirectories().front()->GetName()) == "data" &&
             type == RomFSExtractionType::Truncated)
             break;
         out = out->GetSubdirectories().front();

src/core/frontend/emu_window.h

@@ -12,20 +12,49 @@
 
 namespace Core::Frontend {
 
+/// Information for the Graphics Backends signifying what type of screen pointer is in
+/// WindowInformation
+enum class WindowSystemType {
+    Headless,
+    Windows,
+    X11,
+    Wayland,
+};
+
 /**
- * Represents a graphics context that can be used for background computation or drawing. If the
- * graphics backend doesn't require the context, then the implementation of these methods can be
- * stubs
+ * Represents a drawing context that supports graphics operations.
  */
 class GraphicsContext {
 public:
     virtual ~GraphicsContext();
 
+    /// Inform the driver to swap the front/back buffers and present the current image
+    virtual void SwapBuffers() {}
+
     /// Makes the graphics context current for the caller thread
-    virtual void MakeCurrent() = 0;
+    virtual void MakeCurrent() {}
 
     /// Releases (dunno if this is the "right" word) the context from the caller thread
-    virtual void DoneCurrent() = 0;
+    virtual void DoneCurrent() {}
+
+    class Scoped {
+    public:
+        explicit Scoped(GraphicsContext& context_) : context(context_) {
+            context.MakeCurrent();
+        }
+        ~Scoped() {
+            context.DoneCurrent();
+        }
+
+    private:
+        GraphicsContext& context;
+    };
+
+    /// Calls MakeCurrent on the context and calls DoneCurrent when the scope for the returned value
+    /// ends
+    Scoped Acquire() {
+        return Scoped{*this};
+    }
 };
 
 /**
@@ -46,7 +75,7 @@ public:
  * - DO NOT TREAT THIS CLASS AS A GUI TOOLKIT ABSTRACTION LAYER. That's not what it is. Please
  *   re-read the upper points again and think about it if you don't see this.
  */
-class EmuWindow : public GraphicsContext {
+class EmuWindow {
 public:
     /// Data structure to store emuwindow configuration
     struct WindowConfig {
@@ -56,29 +85,34 @@ public:
         std::pair<unsigned, unsigned> min_client_area_size;
     };
 
+    /// Data describing host window system information
+    struct WindowSystemInfo {
+        // Window system type. Determines which GL context or Vulkan WSI is used.
+        WindowSystemType type = WindowSystemType::Headless;
+
+        // Connection to a display server. This is used on X11 and Wayland platforms.
+        void* display_connection = nullptr;
+
+        // Render surface. This is a pointer to the native window handle, which depends
+        // on the platform. e.g. HWND for Windows, Window for X11. If the surface is
+        // set to nullptr, the video backend will run in headless mode.
+        void* render_surface = nullptr;
+
+        // Scale of the render surface. For hidpi systems, this will be >1.
+        float render_surface_scale = 1.0f;
+    };
+
     /// Polls window events
     virtual void PollEvents() = 0;
 
     /**
-     * Returns a GraphicsContext that the frontend provides that is shared with the emu window. This
-     * context can be used from other threads for background graphics computation. If the frontend
-     * is using a graphics backend that doesn't need anything specific to run on a different thread,
-     * then it can use a stubbed implemenation for GraphicsContext.
-     *
-     * If the return value is null, then the core should assume that the frontend cannot provide a
-     * Shared Context
+     * Returns a GraphicsContext that the frontend provides to be used for rendering.
      */
-    virtual std::unique_ptr<GraphicsContext> CreateSharedContext() const {
-        return nullptr;
-    }
+    virtual std::unique_ptr<GraphicsContext> CreateSharedContext() const = 0;
 
     /// Returns if window is shown (not minimized)
     virtual bool IsShown() const = 0;
 
-    /// Retrieves Vulkan specific handlers from the window
-    virtual void RetrieveVulkanHandlers(void* get_instance_proc_addr, void* instance,
-                                        void* surface) const = 0;
-
     /**
      * Signal that a touch pressed event has occurred (e.g. mouse click pressed)
      * @param framebuffer_x Framebuffer x-coordinate that was pressed
@@ -115,6 +149,13 @@ public:
         config = val;
     }
 
+    /**
+     * Returns system information about the drawing area.
+     */
+    const WindowSystemInfo& GetWindowInfo() const {
+        return window_info;
+    }
+
     /**
      * Gets the framebuffer layout (width, height, and screen regions)
      * @note This method is thread-safe
@@ -130,7 +171,7 @@ public:
     void UpdateCurrentFramebufferLayout(unsigned width, unsigned height);
 
 protected:
-    EmuWindow();
+    explicit EmuWindow();
     virtual ~EmuWindow();
 
     /**
@@ -167,6 +208,8 @@ protected:
         client_area_height = size.second;
     }
 
+    WindowSystemInfo window_info;
+
 private:
     /**
      * Handler called when the minimal client area was requested to be changed via SetConfig.

src/core/frontend/framebuffer_layout.cpp

@@ -48,8 +48,8 @@ FramebufferLayout FrameLayoutFromResolutionScale(u32 res_scale) {
     u32 width, height;
 
     if (Settings::values.use_docked_mode) {
-        width = ScreenDocked::WidthDocked * res_scale;
-        height = ScreenDocked::HeightDocked * res_scale;
+        width = ScreenDocked::Width * res_scale;
+        height = ScreenDocked::Height * res_scale;
     } else {
         width = ScreenUndocked::Width * res_scale;
         height = ScreenUndocked::Height * res_scale;

src/core/frontend/framebuffer_layout.h

@@ -8,15 +8,15 @@
 
 namespace Layout {
 
-enum ScreenUndocked : u32 {
-    Width = 1280,
-    Height = 720,
-};
+namespace ScreenUndocked {
+constexpr u32 Width = 1280;
+constexpr u32 Height = 720;
+} // namespace ScreenUndocked
 
-enum ScreenDocked : u32 {
-    WidthDocked = 1920,
-    HeightDocked = 1080,
-};
+namespace ScreenDocked {
+constexpr u32 Width = 1920;
+constexpr u32 Height = 1080;
+} // namespace ScreenDocked
 
 enum class AspectRatio {
     Default,

src/core/frontend/scope_acquire_context.cpp

@@ -1,18 +0,0 @@
-// Copyright 2019 yuzu Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#include "core/frontend/emu_window.h"
-#include "core/frontend/scope_acquire_context.h"
-
-namespace Core::Frontend {
-
-ScopeAcquireContext::ScopeAcquireContext(Core::Frontend::GraphicsContext& context)
-    : context{context} {
-    context.MakeCurrent();
-}
-ScopeAcquireContext::~ScopeAcquireContext() {
-    context.DoneCurrent();
-}
-
-} // namespace Core::Frontend

src/core/frontend/scope_acquire_context.h

@@ -1,23 +0,0 @@
-// Copyright 2019 yuzu Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#pragma once
-
-#include "common/common_types.h"
-
-namespace Core::Frontend {
-
-class GraphicsContext;
-
-/// Helper class to acquire/release window context within a given scope
-class ScopeAcquireContext : NonCopyable {
-public:
-    explicit ScopeAcquireContext(Core::Frontend::GraphicsContext& context);
-    ~ScopeAcquireContext();
-
-private:
-    Core::Frontend::GraphicsContext& context;
-};
-
-} // namespace Core::Frontend

src/core/gdbstub/gdbstub.cpp

@@ -141,6 +141,7 @@ constexpr char target_xml[] =
 )";
 
 int gdbserver_socket = -1;
+bool defer_start = false;
 
 u8 command_buffer[GDB_BUFFER_SIZE];
 u32 command_length;
@@ -1166,6 +1167,9 @@ static void RemoveBreakpoint() {
 
 void HandlePacket() {
     if (!IsConnected()) {
+        if (defer_start) {
+            ToggleServer(true);
+        }
         return;
     }
 
@@ -1256,6 +1260,10 @@ void ToggleServer(bool status) {
     }
 }
 
+void DeferStart() {
+    defer_start = true;
+}
+
 static void Init(u16 port) {
     if (!server_enabled) {
         // Set the halt loop to false in case the user enabled the gdbstub mid-execution.
@@ -1341,6 +1349,7 @@ void Shutdown() {
     if (!server_enabled) {
         return;
     }
+    defer_start = false;
 
     LOG_INFO(Debug_GDBStub, "Stopping GDB ...");
     if (gdbserver_socket != -1) {

src/core/gdbstub/gdbstub.h

@@ -43,6 +43,13 @@ void ToggleServer(bool status);
 /// Start the gdbstub server.
 void Init();
 
+/**
+ * Defer initialization of the gdbstub to the first packet processing functions.
+ * This avoids a case where the gdbstub thread is frozen after initialization
+ * and fails to respond in time to packets.
+ */
+void DeferStart();
+
 /// Stop gdbstub server.
 void Shutdown();
 

src/core/hle/service/am/am.cpp

@@ -52,6 +52,11 @@ enum class LaunchParameterKind : u32 {
     AccountPreselectedUser = 2,
 };
 
+enum class VrMode : u8 {
+    Disabled = 0,
+    Enabled = 1,
+};
+
 constexpr u32 LAUNCH_PARAMETER_ACCOUNT_PRESELECTED_USER_MAGIC = 0xC79497CA;
 
 struct LaunchParameterAccountPreselectedUser {
@@ -605,11 +610,11 @@ ICommonStateGetter::ICommonStateGetter(Core::System& system,
         {30, nullptr, "GetHomeButtonReaderLockAccessor"},
         {31, nullptr, "GetReaderLockAccessorEx"},
         {40, nullptr, "GetCradleFwVersion"},
-        {50, nullptr, "IsVrModeEnabled"},
-        {51, nullptr, "SetVrModeEnabled"},
+        {50, &ICommonStateGetter::IsVrModeEnabled, "IsVrModeEnabled"},
+        {51, &ICommonStateGetter::SetVrModeEnabled, "SetVrModeEnabled"},
         {52, &ICommonStateGetter::SetLcdBacklighOffEnabled, "SetLcdBacklighOffEnabled"},
         {53, nullptr, "BeginVrModeEx"},
-        {54, nullptr, "EndVrModeEx"},
+        {54, &ICommonStateGetter::EndVrModeEx, "EndVrModeEx"},
         {55, nullptr, "IsInControllerFirmwareUpdateSection"},
         {60, &ICommonStateGetter::GetDefaultDisplayResolution, "GetDefaultDisplayResolution"},
         {61, &ICommonStateGetter::GetDefaultDisplayResolutionChangeEvent, "GetDefaultDisplayResolutionChangeEvent"},
@@ -672,6 +677,30 @@ void ICommonStateGetter::GetCurrentFocusState(Kernel::HLERequestContext& ctx) {
     rb.Push(static_cast<u8>(FocusState::InFocus));
 }
 
+void ICommonStateGetter::IsVrModeEnabled(Kernel::HLERequestContext& ctx) {
+    LOG_WARNING(Service_AM, "(STUBBED) called");
+
+    IPC::ResponseBuilder rb{ctx, 3};
+    rb.Push(RESULT_SUCCESS);
+    rb.PushEnum(VrMode::Disabled);
+}
+
+void ICommonStateGetter::SetVrModeEnabled(Kernel::HLERequestContext& ctx) {
+    IPC::RequestParser rp{ctx};
+    const auto is_vr_mode_enabled = rp.Pop<bool>();
+
+    LOG_WARNING(Service_AM, "(STUBBED) called. is_vr_mode_enabled={}", is_vr_mode_enabled);
+
+    IPC::ResponseBuilder rb{ctx, 2};
+    if (!is_vr_mode_enabled) {
+        rb.Push(RESULT_SUCCESS);
+    } else {
+        // TODO: Find better error code for this
+        UNIMPLEMENTED_MSG("is_vr_mode_enabled={}", is_vr_mode_enabled);
+        rb.Push(RESULT_UNKNOWN);
+    }
+}
+
 void ICommonStateGetter::SetLcdBacklighOffEnabled(Kernel::HLERequestContext& ctx) {
     IPC::RequestParser rp{ctx};
     const auto is_lcd_backlight_off_enabled = rp.Pop<bool>();
@@ -683,6 +712,13 @@ void ICommonStateGetter::SetLcdBacklighOffEnabled(Kernel::HLERequestContext& ctx
     rb.Push(RESULT_SUCCESS);
 }
 
+void ICommonStateGetter::EndVrModeEx(Kernel::HLERequestContext& ctx) {
+    LOG_WARNING(Service_AM, "(STUBBED) called");
+
+    IPC::ResponseBuilder rb{ctx, 2};
+    rb.Push(RESULT_SUCCESS);
+}
+
 void ICommonStateGetter::GetDefaultDisplayResolutionChangeEvent(Kernel::HLERequestContext& ctx) {
     LOG_DEBUG(Service_AM, "called");
 

src/core/hle/service/am/am.h

@@ -182,7 +182,10 @@ private:
     void GetOperationMode(Kernel::HLERequestContext& ctx);
     void GetPerformanceMode(Kernel::HLERequestContext& ctx);
     void GetBootMode(Kernel::HLERequestContext& ctx);
+    void IsVrModeEnabled(Kernel::HLERequestContext& ctx);
+    void SetVrModeEnabled(Kernel::HLERequestContext& ctx);
     void SetLcdBacklighOffEnabled(Kernel::HLERequestContext& ctx);
+    void EndVrModeEx(Kernel::HLERequestContext& ctx);
     void GetDefaultDisplayResolution(Kernel::HLERequestContext& ctx);
     void SetCpuBoostMode(Kernel::HLERequestContext& ctx);
 

src/core/hle/service/am/applets/web_browser.cpp

@@ -254,6 +254,12 @@ void WebBrowser::Execute() {
 
     if (status != RESULT_SUCCESS) {
         complete = true;
+
+        // This is a workaround in order not to softlock yuzu when an error happens during the
+        // webapplet init. In order to avoid an svcBreak, the status is set to RESULT_SUCCESS
+        Finalize();
+        status = RESULT_SUCCESS;
+
         return;
     }
 

src/core/hle/service/caps/caps.cpp

@@ -2,168 +2,24 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
-#include <memory>
-
 #include "core/hle/service/caps/caps.h"
+#include "core/hle/service/caps/caps_a.h"
+#include "core/hle/service/caps/caps_c.h"
+#include "core/hle/service/caps/caps_sc.h"
+#include "core/hle/service/caps/caps_ss.h"
+#include "core/hle/service/caps/caps_su.h"
+#include "core/hle/service/caps/caps_u.h"
 #include "core/hle/service/service.h"
-#include "core/hle/service/sm/sm.h"
 
 namespace Service::Capture {
 
-class CAPS_A final : public ServiceFramework<CAPS_A> {
-public:
-    explicit CAPS_A() : ServiceFramework{"caps:a"} {
-        // clang-format off
-        static const FunctionInfo functions[] = {
-            {0, nullptr, "GetAlbumFileCount"},
-            {1, nullptr, "GetAlbumFileList"},
-            {2, nullptr, "LoadAlbumFile"},
-            {3, nullptr, "DeleteAlbumFile"},
-            {4, nullptr, "StorageCopyAlbumFile"},
-            {5, nullptr, "IsAlbumMounted"},
-            {6, nullptr, "GetAlbumUsage"},
-            {7, nullptr, "GetAlbumFileSize"},
-            {8, nullptr, "LoadAlbumFileThumbnail"},
-            {9, nullptr, "LoadAlbumScreenShotImage"},
-            {10, nullptr, "LoadAlbumScreenShotThumbnailImage"},
-            {11, nullptr, "GetAlbumEntryFromApplicationAlbumEntry"},
-            {12, nullptr, "Unknown12"},
-            {13, nullptr, "Unknown13"},
-            {14, nullptr, "Unknown14"},
-            {15, nullptr, "Unknown15"},
-            {16, nullptr, "Unknown16"},
-            {17, nullptr, "Unknown17"},
-            {18, nullptr, "Unknown18"},
-            {202, nullptr, "SaveEditedScreenShot"},
-            {301, nullptr, "GetLastThumbnail"},
-            {401, nullptr, "GetAutoSavingStorage"},
-            {501, nullptr, "GetRequiredStorageSpaceSizeToCopyAll"},
-            {1001, nullptr, "Unknown1001"},
-            {1002, nullptr, "Unknown1002"},
-            {1003, nullptr, "Unknown1003"},
-            {8001, nullptr, "ForceAlbumUnmounted"},
-            {8002, nullptr, "ResetAlbumMountStatus"},
-            {8011, nullptr, "RefreshAlbumCache"},
-            {8012, nullptr, "GetAlbumCache"},
-            {8013, nullptr, "Unknown8013"},
-            {8021, nullptr, "GetAlbumEntryFromApplicationAlbumEntryAruid"},
-            {10011, nullptr, "SetInternalErrorConversionEnabled"},
-            {50000, nullptr, "Unknown50000"},
-            {60002, nullptr, "Unknown60002"},
-        };
-        // clang-format on
-
-        RegisterHandlers(functions);
-    }
-};
-
-class CAPS_C final : public ServiceFramework<CAPS_C> {
-public:
-    explicit CAPS_C() : ServiceFramework{"caps:c"} {
-        // clang-format off
-        static const FunctionInfo functions[] = {
-            {33, nullptr, "Unknown33"},
-            {2001, nullptr, "Unknown2001"},
-            {2002, nullptr, "Unknown2002"},
-            {2011, nullptr, "Unknown2011"},
-            {2012, nullptr, "Unknown2012"},
-            {2013, nullptr, "Unknown2013"},
-            {2014, nullptr, "Unknown2014"},
-            {2101, nullptr, "Unknown2101"},
-            {2102, nullptr, "Unknown2102"},
-            {2201, nullptr, "Unknown2201"},
-            {2301, nullptr, "Unknown2301"},
-        };
-        // clang-format on
-
-        RegisterHandlers(functions);
-    }
-};
-
-class CAPS_SC final : public ServiceFramework<CAPS_SC> {
-public:
-    explicit CAPS_SC() : ServiceFramework{"caps:sc"} {
-        // clang-format off
-        static const FunctionInfo functions[] = {
-            {1, nullptr, "Unknown1"},
-            {2, nullptr, "Unknown2"},
-            {1001, nullptr, "Unknown3"},
-            {1002, nullptr, "Unknown4"},
-            {1003, nullptr, "Unknown5"},
-            {1011, nullptr, "Unknown6"},
-            {1012, nullptr, "Unknown7"},
-            {1201, nullptr, "Unknown8"},
-            {1202, nullptr, "Unknown9"},
-            {1203, nullptr, "Unknown10"},
-        };
-        // clang-format on
-
-        RegisterHandlers(functions);
-    }
-};
-
-class CAPS_SS final : public ServiceFramework<CAPS_SS> {
-public:
-    explicit CAPS_SS() : ServiceFramework{"caps:ss"} {
-        // clang-format off
-        static const FunctionInfo functions[] = {
-            {201, nullptr, "Unknown1"},
-            {202, nullptr, "Unknown2"},
-            {203, nullptr, "Unknown3"},
-            {204, nullptr, "Unknown4"},
-        };
-        // clang-format on
-
-        RegisterHandlers(functions);
-    }
-};
-
-class CAPS_SU final : public ServiceFramework<CAPS_SU> {
-public:
-    explicit CAPS_SU() : ServiceFramework{"caps:su"} {
-        // clang-format off
-        static const FunctionInfo functions[] = {
-            {201, nullptr, "SaveScreenShot"},
-            {203, nullptr, "SaveScreenShotEx0"},
-        };
-        // clang-format on
-
-        RegisterHandlers(functions);
-    }
-};
-
-class CAPS_U final : public ServiceFramework<CAPS_U> {
-public:
-    explicit CAPS_U() : ServiceFramework{"caps:u"} {
-        // clang-format off
-        static const FunctionInfo functions[] = {
-            {32, nullptr, "SetShimLibraryVersion"},
-            {102, nullptr, "GetAlbumFileListByAruid"},
-            {103, nullptr, "DeleteAlbumFileByAruid"},
-            {104, nullptr, "GetAlbumFileSizeByAruid"},
-            {105, nullptr, "DeleteAlbumFileByAruidForDebug"},
-            {110, nullptr, "LoadAlbumScreenShotImageByAruid"},
-            {120, nullptr, "LoadAlbumScreenShotThumbnailImageByAruid"},
-            {130, nullptr, "PrecheckToCreateContentsByAruid"},
-            {140, nullptr, "GetAlbumFileList1AafeAruidDeprecated"},
-            {141, nullptr, "GetAlbumFileList2AafeUidAruidDeprecated"},
-            {142, nullptr, "GetAlbumFileList3AaeAruid"},
-            {143, nullptr, "GetAlbumFileList4AaeUidAruid"},
-            {60002, nullptr, "OpenAccessorSessionForApplication"},
-        };
-        // clang-format on
-
-        RegisterHandlers(functions);
-    }
-};
-
 void InstallInterfaces(SM::ServiceManager& sm) {
     std::make_shared<CAPS_A>()->InstallAsService(sm);
     std::make_shared<CAPS_C>()->InstallAsService(sm);
+    std::make_shared<CAPS_U>()->InstallAsService(sm);
     std::make_shared<CAPS_SC>()->InstallAsService(sm);
     std::make_shared<CAPS_SS>()->InstallAsService(sm);
     std::make_shared<CAPS_SU>()->InstallAsService(sm);
-    std::make_shared<CAPS_U>()->InstallAsService(sm);
 }
 
 } // namespace Service::Capture

src/core/hle/service/caps/caps.h

@@ -4,12 +4,83 @@
 
 #pragma once
 
+#include "core/hle/service/service.h"
+
 namespace Service::SM {
 class ServiceManager;
 }
 
 namespace Service::Capture {
 
+enum AlbumImageOrientation {
+    Orientation0 = 0,
+    Orientation1 = 1,
+    Orientation2 = 2,
+    Orientation3 = 3,
+};
+
+enum AlbumReportOption {
+    Disable = 0,
+    Enable = 1,
+};
+
+enum ContentType : u8 {
+    Screenshot = 0,
+    Movie = 1,
+    ExtraMovie = 3,
+};
+
+enum AlbumStorage : u8 {
+    NAND = 0,
+    SD = 1,
+};
+
+struct AlbumFileDateTime {
+    u16 year;
+    u8 month;
+    u8 day;
+    u8 hour;
+    u8 minute;
+    u8 second;
+    u8 uid;
+};
+
+struct AlbumEntry {
+    u64 size;
+    u64 application_id;
+    AlbumFileDateTime datetime;
+    AlbumStorage storage;
+    ContentType content;
+    u8 padding[6];
+};
+
+struct AlbumFileEntry {
+    u64 size;
+    u64 hash;
+    AlbumFileDateTime datetime;
+    AlbumStorage storage;
+    ContentType content;
+    u8 padding[5];
+    u8 unknown;
+};
+
+struct ApplicationAlbumEntry {
+    u64 size;
+    u64 hash;
+    AlbumFileDateTime datetime;
+    AlbumStorage storage;
+    ContentType content;
+    u8 padding[5];
+    u8 unknown;
+};
+
+struct ApplicationAlbumFileEntry {
+    ApplicationAlbumEntry entry;
+    AlbumFileDateTime datetime;
+    u64 unknown;
+};
+
+/// Registers all Capture services with the specified service manager.
 void InstallInterfaces(SM::ServiceManager& sm);
 
 } // namespace Service::Capture

src/core/hle/service/caps/caps_a.cpp

@@ -0,0 +1,78 @@
+// Copyright 2020 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "core/hle/service/caps/caps_a.h"
+
+namespace Service::Capture {
+
+class IAlbumAccessorSession final : public ServiceFramework<IAlbumAccessorSession> {
+public:
+    explicit IAlbumAccessorSession() : ServiceFramework{"IAlbumAccessorSession"} {
+        // clang-format off
+        static const FunctionInfo functions[] = {
+            {2001, nullptr, "OpenAlbumMovieReadStream"},
+            {2002, nullptr, "CloseAlbumMovieReadStream"},
+            {2003, nullptr, "GetAlbumMovieReadStreamMovieDataSize"},
+            {2004, nullptr, "ReadMovieDataFromAlbumMovieReadStream"},
+            {2005, nullptr, "GetAlbumMovieReadStreamBrokenReason"},
+            {2006, nullptr, "GetAlbumMovieReadStreamImageDataSize"},
+            {2007, nullptr, "ReadImageDataFromAlbumMovieReadStream"},
+            {2008, nullptr, "ReadFileAttributeFromAlbumMovieReadStream"},
+        };
+        // clang-format on
+
+        RegisterHandlers(functions);
+    }
+};
+
+CAPS_A::CAPS_A() : ServiceFramework("caps:a") {
+    // clang-format off
+    static const FunctionInfo functions[] = {
+        {0, nullptr, "GetAlbumFileCount"},
+        {1, nullptr, "GetAlbumFileList"},
+        {2, nullptr, "LoadAlbumFile"},
+        {3, nullptr, "DeleteAlbumFile"},
+        {4, nullptr, "StorageCopyAlbumFile"},
+        {5, nullptr, "IsAlbumMounted"},
+        {6, nullptr, "GetAlbumUsage"},
+        {7, nullptr, "GetAlbumFileSize"},
+        {8, nullptr, "LoadAlbumFileThumbnail"},
+        {9, nullptr, "LoadAlbumScreenShotImage"},
+        {10, nullptr, "LoadAlbumScreenShotThumbnailImage"},
+        {11, nullptr, "GetAlbumEntryFromApplicationAlbumEntry"},
+        {12, nullptr, "LoadAlbumScreenShotImageEx"},
+        {13, nullptr, "LoadAlbumScreenShotThumbnailImageEx"},
+        {14, nullptr, "LoadAlbumScreenShotImageEx0"},
+        {15, nullptr, "GetAlbumUsage3"},
+        {16, nullptr, "GetAlbumMountResult"},
+        {17, nullptr, "GetAlbumUsage16"},
+        {18, nullptr, "Unknown18"},
+        {100, nullptr, "GetAlbumFileCountEx0"},
+        {101, nullptr, "GetAlbumFileListEx0"},
+        {202, nullptr, "SaveEditedScreenShot"},
+        {301, nullptr, "GetLastThumbnail"},
+        {302, nullptr, "GetLastOverlayMovieThumbnail"},
+        {401, nullptr, "GetAutoSavingStorage"},
+        {501, nullptr, "GetRequiredStorageSpaceSizeToCopyAll"},
+        {1001, nullptr, "LoadAlbumScreenShotThumbnailImageEx0"},
+        {1002, nullptr, "LoadAlbumScreenShotImageEx1"},
+        {1003, nullptr, "LoadAlbumScreenShotThumbnailImageEx1"},
+        {8001, nullptr, "ForceAlbumUnmounted"},
+        {8002, nullptr, "ResetAlbumMountStatus"},
+        {8011, nullptr, "RefreshAlbumCache"},
+        {8012, nullptr, "GetAlbumCache"},
+        {8013, nullptr, "GetAlbumCacheEx"},
+        {8021, nullptr, "GetAlbumEntryFromApplicationAlbumEntryAruid"},
+        {10011, nullptr, "SetInternalErrorConversionEnabled"},
+        {50000, nullptr, "LoadMakerNoteInfoForDebug"},
+        {60002, nullptr, "OpenAccessorSession"},
+    };
+    // clang-format on
+
+    RegisterHandlers(functions);
+}
+
+CAPS_A::~CAPS_A() = default;
+
+} // namespace Service::Capture

src/core/hle/service/caps/caps_a.h

@@ -0,0 +1,21 @@
+// Copyright 2020 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "core/hle/service/service.h"
+
+namespace Kernel {
+class HLERequestContext;
+}
+
+namespace Service::Capture {
+
+class CAPS_A final : public ServiceFramework<CAPS_A> {
+public:
+    explicit CAPS_A();
+    ~CAPS_A() override;
+};
+
+} // namespace Service::Capture

src/core/hle/service/caps/caps_c.cpp

@@ -0,0 +1,75 @@
+// Copyright 2020 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "core/hle/service/caps/caps_c.h"
+
+namespace Service::Capture {
+
+class IAlbumControlSession final : public ServiceFramework<IAlbumControlSession> {
+public:
+    explicit IAlbumControlSession() : ServiceFramework{"IAlbumControlSession"} {
+        // clang-format off
+        static const FunctionInfo functions[] = {
+            {2001, nullptr, "OpenAlbumMovieReadStream"},
+            {2002, nullptr, "CloseAlbumMovieReadStream"},
+            {2003, nullptr, "GetAlbumMovieReadStreamMovieDataSize"},
+            {2004, nullptr, "ReadMovieDataFromAlbumMovieReadStream"},
+            {2005, nullptr, "GetAlbumMovieReadStreamBrokenReason"},
+            {2006, nullptr, "GetAlbumMovieReadStreamImageDataSize"},
+            {2007, nullptr, "ReadImageDataFromAlbumMovieReadStream"},
+            {2008, nullptr, "ReadFileAttributeFromAlbumMovieReadStream"},
+            {2401, nullptr, "OpenAlbumMovieWriteStream"},
+            {2402, nullptr, "FinishAlbumMovieWriteStream"},
+            {2403, nullptr, "CommitAlbumMovieWriteStream"},
+            {2404, nullptr, "DiscardAlbumMovieWriteStream"},
+            {2405, nullptr, "DiscardAlbumMovieWriteStreamNoDelete"},
+            {2406, nullptr, "CommitAlbumMovieWriteStreamEx"},
+            {2411, nullptr, "StartAlbumMovieWriteStreamDataSection"},
+            {2412, nullptr, "EndAlbumMovieWriteStreamDataSection"},
+            {2413, nullptr, "StartAlbumMovieWriteStreamMetaSection"},
+            {2414, nullptr, "EndAlbumMovieWriteStreamMetaSection"},
+            {2421, nullptr, "ReadDataFromAlbumMovieWriteStream"},
+            {2422, nullptr, "WriteDataToAlbumMovieWriteStream"},
+            {2424, nullptr, "WriteMetaToAlbumMovieWriteStream"},
+            {2431, nullptr, "GetAlbumMovieWriteStreamBrokenReason"},
+            {2433, nullptr, "GetAlbumMovieWriteStreamDataSize"},
+            {2434, nullptr, "SetAlbumMovieWriteStreamDataSize"},
+        };
+        // clang-format on
+
+        RegisterHandlers(functions);
+    }
+};
+
+CAPS_C::CAPS_C() : ServiceFramework("caps:c") {
+    // clang-format off
+    static const FunctionInfo functions[] = {
+        {1, nullptr, "CaptureRawImage"},
+        {2, nullptr, "CaptureRawImageWithTimeout"},
+        {33, nullptr, "Unknown33"},
+        {1001, nullptr, "RequestTakingScreenShot"},
+        {1002, nullptr, "RequestTakingScreenShotWithTimeout"},
+        {1011, nullptr, "NotifyTakingScreenShotRefused"},
+        {2001, nullptr, "NotifyAlbumStorageIsAvailable"},
+        {2002, nullptr, "NotifyAlbumStorageIsUnavailable"},
+        {2011, nullptr, "RegisterAppletResourceUserId"},
+        {2012, nullptr, "UnregisterAppletResourceUserId"},
+        {2013, nullptr, "GetApplicationIdFromAruid"},
+        {2014, nullptr, "CheckApplicationIdRegistered"},
+        {2101, nullptr, "GenerateCurrentAlbumFileId"},
+        {2102, nullptr, "GenerateApplicationAlbumEntry"},
+        {2201, nullptr, "SaveAlbumScreenShotFile"},
+        {2202, nullptr, "SaveAlbumScreenShotFileEx"},
+        {2301, nullptr, "SetOverlayScreenShotThumbnailData"},
+        {2302, nullptr, "SetOverlayMovieThumbnailData"},
+        {60001, nullptr, "OpenControlSession"},
+    };
+    // clang-format on
+
+    RegisterHandlers(functions);
+}
+
+CAPS_C::~CAPS_C() = default;
+
+} // namespace Service::Capture

src/core/hle/service/caps/caps_c.h

@@ -0,0 +1,21 @@
+// Copyright 2020 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "core/hle/service/service.h"
+
+namespace Kernel {
+class HLERequestContext;
+}
+
+namespace Service::Capture {
+
+class CAPS_C final : public ServiceFramework<CAPS_C> {
+public:
+    explicit CAPS_C();
+    ~CAPS_C() override;
+};
+
+} // namespace Service::Capture

src/core/hle/service/caps/caps_sc.cpp

@@ -0,0 +1,40 @@
+// Copyright 2020 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "core/hle/service/caps/caps_sc.h"
+
+namespace Service::Capture {
+
+CAPS_SC::CAPS_SC() : ServiceFramework("caps:sc") {
+    // clang-format off
+    static const FunctionInfo functions[] = {
+        {1, nullptr, "CaptureRawImage"},
+        {2, nullptr, "CaptureRawImageWithTimeout"},
+        {3, nullptr, "AttachSharedBuffer"},
+        {5, nullptr, "CaptureRawImageToAttachedSharedBuffer"},
+        {210, nullptr, "Unknown210"},
+        {1001, nullptr, "RequestTakingScreenShot"},
+        {1002, nullptr, "RequestTakingScreenShotWithTimeout"},
+        {1003, nullptr, "RequestTakingScreenShotEx"},
+        {1004, nullptr, "RequestTakingScreenShotEx1"},
+        {1009, nullptr, "CancelTakingScreenShot"},
+        {1010, nullptr, "SetTakingScreenShotCancelState"},
+        {1011, nullptr, "NotifyTakingScreenShotRefused"},
+        {1012, nullptr, "NotifyTakingScreenShotFailed"},
+        {1101, nullptr, "SetupOverlayMovieThumbnail"},
+        {1106, nullptr, "Unknown1106"},
+        {1107, nullptr, "Unknown1107"},
+        {1201, nullptr, "OpenRawScreenShotReadStream"},
+        {1202, nullptr, "CloseRawScreenShotReadStream"},
+        {1203, nullptr, "ReadRawScreenShotReadStream"},
+        {1204, nullptr, "Unknown1204"},
+    };
+    // clang-format on
+
+    RegisterHandlers(functions);
+}
+
+CAPS_SC::~CAPS_SC() = default;
+
+} // namespace Service::Capture

src/core/hle/service/caps/caps_sc.h

@@ -0,0 +1,21 @@
+// Copyright 2020 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "core/hle/service/service.h"
+
+namespace Kernel {
+class HLERequestContext;
+}
+
+namespace Service::Capture {
+
+class CAPS_SC final : public ServiceFramework<CAPS_SC> {
+public:
+    explicit CAPS_SC();
+    ~CAPS_SC() override;
+};
+
+} // namespace Service::Capture

src/core/hle/service/caps/caps_ss.cpp

@@ -0,0 +1,26 @@
+// Copyright 2020 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "core/hle/service/caps/caps_ss.h"
+
+namespace Service::Capture {
+
+CAPS_SS::CAPS_SS() : ServiceFramework("caps:ss") {
+    // clang-format off
+    static const FunctionInfo functions[] = {
+        {201, nullptr, "SaveScreenShot"},
+        {202, nullptr, "SaveEditedScreenShot"},
+        {203, nullptr, "SaveScreenShotEx0"},
+        {204, nullptr, "SaveEditedScreenShotEx0"},
+        {206, nullptr, "Unknown206"},
+        {208, nullptr, "SaveScreenShotOfMovieEx1"},
+    };
+    // clang-format on
+
+    RegisterHandlers(functions);
+}
+
+CAPS_SS::~CAPS_SS() = default;
+
+} // namespace Service::Capture

src/core/hle/service/caps/caps_ss.h

@@ -0,0 +1,21 @@
+// Copyright 2020 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "core/hle/service/service.h"
+
+namespace Kernel {
+class HLERequestContext;
+}
+
+namespace Service::Capture {
+
+class CAPS_SS final : public ServiceFramework<CAPS_SS> {
+public:
+    explicit CAPS_SS();
+    ~CAPS_SS() override;
+};
+
+} // namespace Service::Capture

src/core/hle/service/caps/caps_su.cpp

@@ -0,0 +1,22 @@
+// Copyright 2020 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "core/hle/service/caps/caps_su.h"
+
+namespace Service::Capture {
+
+CAPS_SU::CAPS_SU() : ServiceFramework("caps:su") {
+    // clang-format off
+    static const FunctionInfo functions[] = {
+        {201, nullptr, "SaveScreenShot"},
+        {203, nullptr, "SaveScreenShotEx0"},
+    };
+    // clang-format on
+
+    RegisterHandlers(functions);
+}
+
+CAPS_SU::~CAPS_SU() = default;
+
+} // namespace Service::Capture

src/core/hle/service/caps/caps_su.h

@@ -0,0 +1,21 @@
+// Copyright 2020 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "core/hle/service/service.h"
+
+namespace Kernel {
+class HLERequestContext;
+}
+
+namespace Service::Capture {
+
+class CAPS_SU final : public ServiceFramework<CAPS_SU> {
+public:
+    explicit CAPS_SU();
+    ~CAPS_SU() override;
+};
+
+} // namespace Service::Capture

src/core/hle/service/caps/caps_u.cpp

@@ -0,0 +1,76 @@
+// Copyright 2020 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "common/logging/log.h"
+#include "core/hle/ipc_helpers.h"
+#include "core/hle/service/caps/caps.h"
+#include "core/hle/service/caps/caps_u.h"
+
+namespace Service::Capture {
+
+class IAlbumAccessorApplicationSession final
+    : public ServiceFramework<IAlbumAccessorApplicationSession> {
+public:
+    explicit IAlbumAccessorApplicationSession()
+        : ServiceFramework{"IAlbumAccessorApplicationSession"} {
+        // clang-format off
+        static const FunctionInfo functions[] = {
+            {2001, nullptr, "OpenAlbumMovieReadStream"},
+            {2002, nullptr, "CloseAlbumMovieReadStream"},
+            {2003, nullptr, "GetAlbumMovieReadStreamMovieDataSize"},
+            {2004, nullptr, "ReadMovieDataFromAlbumMovieReadStream"},
+            {2005, nullptr, "GetAlbumMovieReadStreamBrokenReason"},
+        };
+        // clang-format on
+
+        RegisterHandlers(functions);
+    }
+};
+
+CAPS_U::CAPS_U() : ServiceFramework("caps:u") {
+    // clang-format off
+    static const FunctionInfo functions[] = {
+        {31, nullptr, "GetShimLibraryVersion"},
+        {32, nullptr, "SetShimLibraryVersion"},
+        {102, &CAPS_U::GetAlbumContentsFileListForApplication, "GetAlbumContentsFileListForApplication"},
+        {103, nullptr, "DeleteAlbumContentsFileForApplication"},
+        {104, nullptr, "GetAlbumContentsFileSizeForApplication"},
+        {105, nullptr, "DeleteAlbumFileByAruidForDebug"},
+        {110, nullptr, "LoadAlbumContentsFileScreenShotImageForApplication"},
+        {120, nullptr, "LoadAlbumContentsFileThumbnailImageForApplication"},
+        {130, nullptr, "PrecheckToCreateContentsForApplication"},
+        {140, nullptr, "GetAlbumFileList1AafeAruidDeprecated"},
+        {141, nullptr, "GetAlbumFileList2AafeUidAruidDeprecated"},
+        {142, nullptr, "GetAlbumFileList3AaeAruid"},
+        {143, nullptr, "GetAlbumFileList4AaeUidAruid"},
+        {60002, nullptr, "OpenAccessorSessionForApplication"},
+    };
+    // clang-format on
+
+    RegisterHandlers(functions);
+}
+
+CAPS_U::~CAPS_U() = default;
+
+void CAPS_U::GetAlbumContentsFileListForApplication(Kernel::HLERequestContext& ctx) {
+    // Takes a type-0x6 output buffer containing an array of ApplicationAlbumFileEntry, a PID, an
+    // u8 ContentType, two s64s, and an u64 AppletResourceUserId. Returns an output u64 for total
+    // output entries (which is copied to a s32 by official SW).
+    IPC::RequestParser rp{ctx};
+    [[maybe_unused]] const auto application_album_file_entries = rp.PopRaw<std::array<u8, 0x30>>();
+    const auto pid = rp.Pop<s32>();
+    const auto content_type = rp.PopRaw<ContentType>();
+    [[maybe_unused]] const auto start_datetime = rp.PopRaw<AlbumFileDateTime>();
+    [[maybe_unused]] const auto end_datetime = rp.PopRaw<AlbumFileDateTime>();
+    const auto applet_resource_user_id = rp.Pop<u64>();
+    LOG_WARNING(Service_Capture,
+                "(STUBBED) called. pid={}, content_type={}, applet_resource_user_id={}", pid,
+                content_type, applet_resource_user_id);
+
+    IPC::ResponseBuilder rb{ctx, 3};
+    rb.Push(RESULT_SUCCESS);
+    rb.Push<s32>(0);
+}
+
+} // namespace Service::Capture

src/core/hle/service/caps/caps_u.h

@@ -0,0 +1,24 @@
+// Copyright 2020 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "core/hle/service/service.h"
+
+namespace Kernel {
+class HLERequestContext;
+}
+
+namespace Service::Capture {
+
+class CAPS_U final : public ServiceFramework<CAPS_U> {
+public:
+    explicit CAPS_U();
+    ~CAPS_U() override;
+
+private:
+    void GetAlbumContentsFileListForApplication(Kernel::HLERequestContext& ctx);
+};
+
+} // namespace Service::Capture

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

@@ -235,7 +235,7 @@ Hid::Hid(Core::System& system) : ServiceFramework("hid"), system(system) {
         {303, nullptr, "ActivateSevenSixAxisSensor"},
         {304, nullptr, "StartSevenSixAxisSensor"},
         {305, nullptr, "StopSevenSixAxisSensor"},
-        {306, nullptr, "InitializeSevenSixAxisSensor"},
+        {306, &Hid::InitializeSevenSixAxisSensor, "InitializeSevenSixAxisSensor"},
         {307, nullptr, "FinalizeSevenSixAxisSensor"},
         {308, nullptr, "SetSevenSixAxisSensorFusionStrength"},
         {309, nullptr, "GetSevenSixAxisSensorFusionStrength"},
@@ -853,6 +853,13 @@ void Hid::SetPalmaBoostMode(Kernel::HLERequestContext& ctx) {
     rb.Push(RESULT_SUCCESS);
 }
 
+void Hid::InitializeSevenSixAxisSensor(Kernel::HLERequestContext& ctx) {
+    LOG_WARNING(Service_HID, "(STUBBED) called");
+
+    IPC::ResponseBuilder rb{ctx, 2};
+    rb.Push(RESULT_SUCCESS);
+}
+
 class HidDbg final : public ServiceFramework<HidDbg> {
 public:
     explicit HidDbg() : ServiceFramework{"hid:dbg"} {

src/core/hle/service/hid/hid.h

@@ -128,6 +128,7 @@ private:
     void StopSixAxisSensor(Kernel::HLERequestContext& ctx);
     void SetIsPalmaAllConnectable(Kernel::HLERequestContext& ctx);
     void SetPalmaBoostMode(Kernel::HLERequestContext& ctx);
+    void InitializeSevenSixAxisSensor(Kernel::HLERequestContext& ctx);
 
     std::shared_ptr<IAppletResource> applet_resource;
     Core::System& system;

src/core/hle/service/ldr/ldr.cpp

@@ -342,17 +342,27 @@ public:
             return;
         }
 
-        ASSERT(
-            vm_manager
-                .MirrorMemory(*map_address, nro_address, nro_size, Kernel::MemoryState::ModuleCode)
-                .IsSuccess());
+        // Mark text and read-only region as ModuleCode
+        ASSERT(vm_manager
+                   .MirrorMemory(*map_address, nro_address, header.text_size + header.ro_size,
+                                 Kernel::MemoryState::ModuleCode)
+                   .IsSuccess());
+        // Mark read/write region as ModuleCodeData, which is necessary if this region is used for
+        // TransferMemory (e.g. Final Fantasy VIII Remastered does this)
+        ASSERT(vm_manager
+                   .MirrorMemory(*map_address + header.rw_offset, nro_address + header.rw_offset,
+                                 header.rw_size, Kernel::MemoryState::ModuleCodeData)
+                   .IsSuccess());
+        // Revoke permissions from the old memory region
         ASSERT(vm_manager.ReprotectRange(nro_address, nro_size, Kernel::VMAPermission::None)
                    .IsSuccess());
 
         if (bss_size > 0) {
+            // Mark BSS region as ModuleCodeData, which is necessary if this region is used for
+            // TransferMemory (e.g. Final Fantasy VIII Remastered does this)
             ASSERT(vm_manager
                        .MirrorMemory(*map_address + nro_size, bss_address, bss_size,
-                                     Kernel::MemoryState::ModuleCode)
+                                     Kernel::MemoryState::ModuleCodeData)
                        .IsSuccess());
             ASSERT(vm_manager.ReprotectRange(bss_address, bss_size, Kernel::VMAPermission::None)
                        .IsSuccess());

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

@@ -28,6 +28,7 @@ void BufferQueue::SetPreallocatedBuffer(u32 slot, const IGBPBuffer& igbp_buffer)
     buffer.slot = slot;
     buffer.igbp_buffer = igbp_buffer;
     buffer.status = Buffer::Status::Free;
+    free_buffers.push_back(slot);
 
     queue.emplace_back(buffer);
     buffer_wait_event.writable->Signal();
@@ -35,16 +36,37 @@ void BufferQueue::SetPreallocatedBuffer(u32 slot, const IGBPBuffer& igbp_buffer)
 
 std::optional<std::pair<u32, Service::Nvidia::MultiFence*>> BufferQueue::DequeueBuffer(u32 width,
                                                                                        u32 height) {
-    auto itr = std::find_if(queue.begin(), queue.end(), [&](const Buffer& buffer) {
-        // Only consider free buffers. Buffers become free once again after they've been Acquired
-        // and Released by the compositor, see the NVFlinger::Compose method.
-        if (buffer.status != Buffer::Status::Free) {
-            return false;
-        }
 
-        // Make sure that the parameters match.
-        return buffer.igbp_buffer.width == width && buffer.igbp_buffer.height == height;
-    });
+    if (free_buffers.empty()) {
+        return {};
+    }
+
+    auto f_itr = free_buffers.begin();
+    auto itr = queue.end();
+
+    while (f_itr != free_buffers.end()) {
+        auto slot = *f_itr;
+        itr = std::find_if(queue.begin(), queue.end(), [&](const Buffer& buffer) {
+            // Only consider free buffers. Buffers become free once again after they've been
+            // Acquired and Released by the compositor, see the NVFlinger::Compose method.
+            if (buffer.status != Buffer::Status::Free) {
+                return false;
+            }
+
+            if (buffer.slot != slot) {
+                return false;
+            }
+
+            // Make sure that the parameters match.
+            return buffer.igbp_buffer.width == width && buffer.igbp_buffer.height == height;
+        });
+
+        if (itr != queue.end()) {
+            free_buffers.erase(f_itr);
+            break;
+        }
+        ++f_itr;
+    }
 
     if (itr == queue.end()) {
         return {};
@@ -99,10 +121,18 @@ void BufferQueue::ReleaseBuffer(u32 slot) {
     ASSERT(itr != queue.end());
     ASSERT(itr->status == Buffer::Status::Acquired);
     itr->status = Buffer::Status::Free;
+    free_buffers.push_back(slot);
 
     buffer_wait_event.writable->Signal();
 }
 
+void BufferQueue::Disconnect() {
+    queue.clear();
+    queue_sequence.clear();
+    id = 1;
+    layer_id = 1;
+}
+
 u32 BufferQueue::Query(QueryType type) {
     LOG_WARNING(Service, "(STUBBED) called type={}", static_cast<u32>(type));
 

src/core/hle/service/nvflinger/buffer_queue.h

@@ -87,6 +87,7 @@ public:
                      Service::Nvidia::MultiFence& multi_fence);
     std::optional<std::reference_wrapper<const Buffer>> AcquireBuffer();
     void ReleaseBuffer(u32 slot);
+    void Disconnect();
     u32 Query(QueryType type);
 
     u32 GetId() const {
@@ -101,6 +102,7 @@ private:
     u32 id;
     u64 layer_id;
 
+    std::list<u32> free_buffers;
     std::vector<Buffer> queue;
     std::list<u32> queue_sequence;
     Kernel::EventPair buffer_wait_event;

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

@@ -191,8 +191,6 @@ void NVFlinger::Compose() {
         // Search for a queued buffer and acquire it
         auto buffer = buffer_queue.AcquireBuffer();
 
-        MicroProfileFlip();
-
         if (!buffer) {
             continue;
         }
@@ -206,6 +204,8 @@ void NVFlinger::Compose() {
             gpu.WaitFence(fence.id, fence.value);
         }
 
+        MicroProfileFlip();
+
         // Now send the buffer to the GPU for drawing.
         // TODO(Subv): Support more than just disp0. The display device selection is probably based
         // on which display we're drawing (Default, Internal, External, etc)

src/core/hle/service/set/set.cpp

@@ -111,6 +111,14 @@ void SET::GetLanguageCode(Kernel::HLERequestContext& ctx) {
     rb.PushEnum(available_language_codes[Settings::values.language_index]);
 }
 
+void SET::GetRegionCode(Kernel::HLERequestContext& ctx) {
+    LOG_DEBUG(Service_SET, "called");
+
+    IPC::ResponseBuilder rb{ctx, 3};
+    rb.Push(RESULT_SUCCESS);
+    rb.Push(Settings::values.region_index);
+}
+
 SET::SET() : ServiceFramework("set") {
     // clang-format off
     static const FunctionInfo functions[] = {
@@ -118,7 +126,7 @@ SET::SET() : ServiceFramework("set") {
         {1, &SET::GetAvailableLanguageCodes, "GetAvailableLanguageCodes"},
         {2, &SET::MakeLanguageCode, "MakeLanguageCode"},
         {3, &SET::GetAvailableLanguageCodeCount, "GetAvailableLanguageCodeCount"},
-        {4, nullptr, "GetRegionCode"},
+        {4, &SET::GetRegionCode, "GetRegionCode"},
         {5, &SET::GetAvailableLanguageCodes2, "GetAvailableLanguageCodes2"},
         {6, &SET::GetAvailableLanguageCodeCount2, "GetAvailableLanguageCodeCount2"},
         {7, nullptr, "GetKeyCodeMap"},

src/core/hle/service/set/set.h

@@ -43,6 +43,7 @@ private:
     void GetAvailableLanguageCodeCount(Kernel::HLERequestContext& ctx);
     void GetAvailableLanguageCodeCount2(Kernel::HLERequestContext& ctx);
     void GetQuestFlag(Kernel::HLERequestContext& ctx);
+    void GetRegionCode(Kernel::HLERequestContext& ctx);
 };
 
 } // namespace Service::Set

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

@@ -44,7 +44,7 @@ void Controller::QueryPointerBufferSize(Kernel::HLERequestContext& ctx) {
 
     IPC::ResponseBuilder rb{ctx, 3};
     rb.Push(RESULT_SUCCESS);
-    rb.Push<u16>(0x500);
+    rb.Push<u16>(0x1000);
 }
 
 Controller::Controller() : ServiceFramework("IpcController") {

src/core/hle/service/time/interface.cpp

@@ -30,7 +30,7 @@ Time::Time(std::shared_ptr<Module> module, Core::System& system, const char* nam
         {400, &Time::GetClockSnapshot, "GetClockSnapshot"},
         {401, &Time::GetClockSnapshotFromSystemClockContext, "GetClockSnapshotFromSystemClockContext"},
         {500, nullptr, "CalculateStandardUserSystemClockDifferenceByUser"},
-        {501, nullptr, "CalculateSpanBetween"},
+        {501, &Time::CalculateSpanBetween, "CalculateSpanBetween"},
     };
     // clang-format on
 

src/core/hle/service/time/time.cpp

@@ -308,6 +308,35 @@ void Module::Interface::GetClockSnapshotFromSystemClockContext(Kernel::HLEReques
     ctx.WriteBuffer(&clock_snapshot, sizeof(Clock::ClockSnapshot));
 }
 
+void Module::Interface::CalculateSpanBetween(Kernel::HLERequestContext& ctx) {
+    LOG_DEBUG(Service_Time, "called");
+
+    IPC::RequestParser rp{ctx};
+    const auto snapshot_a = rp.PopRaw<Clock::ClockSnapshot>();
+    const auto snapshot_b = rp.PopRaw<Clock::ClockSnapshot>();
+
+    Clock::TimeSpanType time_span_type{};
+    s64 span{};
+    if (const ResultCode result{snapshot_a.steady_clock_time_point.GetSpanBetween(
+            snapshot_b.steady_clock_time_point, span)};
+        result != RESULT_SUCCESS) {
+        if (snapshot_a.network_time && snapshot_b.network_time) {
+            time_span_type =
+                Clock::TimeSpanType::FromSeconds(snapshot_b.network_time - snapshot_a.network_time);
+        } else {
+            IPC::ResponseBuilder rb{ctx, 2};
+            rb.Push(ERROR_TIME_NOT_FOUND);
+            return;
+        }
+    } else {
+        time_span_type = Clock::TimeSpanType::FromSeconds(span);
+    }
+
+    IPC::ResponseBuilder rb{ctx, (sizeof(s64) / 4) + 2};
+    rb.Push(RESULT_SUCCESS);
+    rb.PushRaw(time_span_type.nanoseconds);
+}
+
 void Module::Interface::GetSharedMemoryNativeHandle(Kernel::HLERequestContext& ctx) {
     LOG_DEBUG(Service_Time, "called");
     IPC::ResponseBuilder rb{ctx, 2, 1};

src/core/hle/service/time/time.h

@@ -32,6 +32,7 @@ public:
         void CalculateMonotonicSystemClockBaseTimePoint(Kernel::HLERequestContext& ctx);
         void GetClockSnapshot(Kernel::HLERequestContext& ctx);
         void GetClockSnapshotFromSystemClockContext(Kernel::HLERequestContext& ctx);
+        void CalculateSpanBetween(Kernel::HLERequestContext& ctx);
         void GetSharedMemoryNativeHandle(Kernel::HLERequestContext& ctx);
 
     private:

src/core/hle/service/time/time_zone_content_manager.cpp

@@ -53,7 +53,7 @@ static std::vector<std::string> BuildLocationNameCache(Core::System& system) {
         return {};
     }
 
-    std::vector<char> raw_data(binary_list->GetSize());
+    std::vector<char> raw_data(binary_list->GetSize() + 1);
     binary_list->ReadBytes<char>(raw_data.data(), binary_list->GetSize());
 
     std::stringstream data_stream{raw_data.data()};

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

@@ -513,7 +513,8 @@ private:
 
         auto& buffer_queue = nv_flinger->FindBufferQueue(id);
 
-        if (transaction == TransactionId::Connect) {
+        switch (transaction) {
+        case TransactionId::Connect: {
             IGBPConnectRequestParcel request{ctx.ReadBuffer()};
             IGBPConnectResponseParcel response{
                 static_cast<u32>(static_cast<u32>(DisplayResolution::UndockedWidth) *
@@ -521,14 +522,18 @@ private:
                 static_cast<u32>(static_cast<u32>(DisplayResolution::UndockedHeight) *
                                  Settings::values.resolution_factor)};
             ctx.WriteBuffer(response.Serialize());
-        } else if (transaction == TransactionId::SetPreallocatedBuffer) {
+            break;
+        }
+        case TransactionId::SetPreallocatedBuffer: {
             IGBPSetPreallocatedBufferRequestParcel request{ctx.ReadBuffer()};
 
             buffer_queue.SetPreallocatedBuffer(request.data.slot, request.buffer);
 
             IGBPSetPreallocatedBufferResponseParcel response{};
             ctx.WriteBuffer(response.Serialize());
-        } else if (transaction == TransactionId::DequeueBuffer) {
+            break;
+        }
+        case TransactionId::DequeueBuffer: {
             IGBPDequeueBufferRequestParcel request{ctx.ReadBuffer()};
             const u32 width{request.data.width};
             const u32 height{request.data.height};
@@ -556,14 +561,18 @@ private:
                     },
                     buffer_queue.GetWritableBufferWaitEvent());
             }
-        } else if (transaction == TransactionId::RequestBuffer) {
+            break;
+        }
+        case TransactionId::RequestBuffer: {
             IGBPRequestBufferRequestParcel request{ctx.ReadBuffer()};
 
             auto& buffer = buffer_queue.RequestBuffer(request.slot);
 
             IGBPRequestBufferResponseParcel response{buffer};
             ctx.WriteBuffer(response.Serialize());
-        } else if (transaction == TransactionId::QueueBuffer) {
+            break;
+        }
+        case TransactionId::QueueBuffer: {
             IGBPQueueBufferRequestParcel request{ctx.ReadBuffer()};
 
             buffer_queue.QueueBuffer(request.data.slot, request.data.transform,
@@ -572,7 +581,9 @@ private:
 
             IGBPQueueBufferResponseParcel response{1280, 720};
             ctx.WriteBuffer(response.Serialize());
-        } else if (transaction == TransactionId::Query) {
+            break;
+        }
+        case TransactionId::Query: {
             IGBPQueryRequestParcel request{ctx.ReadBuffer()};
 
             const u32 value =
@@ -580,15 +591,30 @@ private:
 
             IGBPQueryResponseParcel response{value};
             ctx.WriteBuffer(response.Serialize());
-        } else if (transaction == TransactionId::CancelBuffer) {
+            break;
+        }
+        case TransactionId::CancelBuffer: {
             LOG_CRITICAL(Service_VI, "(STUBBED) called, transaction=CancelBuffer");
-        } else if (transaction == TransactionId::Disconnect ||
-                   transaction == TransactionId::DetachBuffer) {
+            break;
+        }
+        case TransactionId::Disconnect: {
+            LOG_WARNING(Service_VI, "(STUBBED) called, transaction=Disconnect");
+            const auto buffer = ctx.ReadBuffer();
+
+            buffer_queue.Disconnect();
+
+            IGBPEmptyResponseParcel response{};
+            ctx.WriteBuffer(response.Serialize());
+            break;
+        }
+        case TransactionId::DetachBuffer: {
             const auto buffer = ctx.ReadBuffer();
 
             IGBPEmptyResponseParcel response{};
             ctx.WriteBuffer(response.Serialize());
-        } else {
+            break;
+        }
+        default:
             ASSERT_MSG(false, "Unimplemented");
         }
 

src/core/memory.cpp

@@ -242,7 +242,52 @@ struct Memory::Impl {
             }
             case Common::PageType::RasterizerCachedMemory: {
                 const u8* const host_ptr = GetPointerFromVMA(process, current_vaddr);
-                system.GPU().FlushRegion(ToCacheAddr(host_ptr), copy_amount);
+                system.GPU().FlushRegion(current_vaddr, copy_amount);
+                std::memcpy(dest_buffer, host_ptr, copy_amount);
+                break;
+            }
+            default:
+                UNREACHABLE();
+            }
+
+            page_index++;
+            page_offset = 0;
+            dest_buffer = static_cast<u8*>(dest_buffer) + copy_amount;
+            remaining_size -= copy_amount;
+        }
+    }
+
+    void ReadBlockUnsafe(const Kernel::Process& process, const VAddr src_addr, void* dest_buffer,
+                         const std::size_t size) {
+        const auto& page_table = process.VMManager().page_table;
+
+        std::size_t remaining_size = size;
+        std::size_t page_index = src_addr >> PAGE_BITS;
+        std::size_t page_offset = src_addr & PAGE_MASK;
+
+        while (remaining_size > 0) {
+            const std::size_t copy_amount =
+                std::min(static_cast<std::size_t>(PAGE_SIZE) - page_offset, remaining_size);
+            const auto current_vaddr = static_cast<VAddr>((page_index << PAGE_BITS) + page_offset);
+
+            switch (page_table.attributes[page_index]) {
+            case Common::PageType::Unmapped: {
+                LOG_ERROR(HW_Memory,
+                          "Unmapped ReadBlock @ 0x{:016X} (start address = 0x{:016X}, size = {})",
+                          current_vaddr, src_addr, size);
+                std::memset(dest_buffer, 0, copy_amount);
+                break;
+            }
+            case Common::PageType::Memory: {
+                DEBUG_ASSERT(page_table.pointers[page_index]);
+
+                const u8* const src_ptr =
+                    page_table.pointers[page_index] + page_offset + (page_index << PAGE_BITS);
+                std::memcpy(dest_buffer, src_ptr, copy_amount);
+                break;
+            }
+            case Common::PageType::RasterizerCachedMemory: {
+                const u8* const host_ptr = GetPointerFromVMA(process, current_vaddr);
                 std::memcpy(dest_buffer, host_ptr, copy_amount);
                 break;
             }
@@ -261,6 +306,10 @@ struct Memory::Impl {
         ReadBlock(*system.CurrentProcess(), src_addr, dest_buffer, size);
     }
 
+    void ReadBlockUnsafe(const VAddr src_addr, void* dest_buffer, const std::size_t size) {
+        ReadBlockUnsafe(*system.CurrentProcess(), src_addr, dest_buffer, size);
+    }
+
     void WriteBlock(const Kernel::Process& process, const VAddr dest_addr, const void* src_buffer,
                     const std::size_t size) {
         const auto& page_table = process.VMManager().page_table;
@@ -290,7 +339,50 @@ struct Memory::Impl {
             }
             case Common::PageType::RasterizerCachedMemory: {
                 u8* const host_ptr = GetPointerFromVMA(process, current_vaddr);
-                system.GPU().InvalidateRegion(ToCacheAddr(host_ptr), copy_amount);
+                system.GPU().InvalidateRegion(current_vaddr, copy_amount);
+                std::memcpy(host_ptr, src_buffer, copy_amount);
+                break;
+            }
+            default:
+                UNREACHABLE();
+            }
+
+            page_index++;
+            page_offset = 0;
+            src_buffer = static_cast<const u8*>(src_buffer) + copy_amount;
+            remaining_size -= copy_amount;
+        }
+    }
+
+    void WriteBlockUnsafe(const Kernel::Process& process, const VAddr dest_addr,
+                          const void* src_buffer, const std::size_t size) {
+        const auto& page_table = process.VMManager().page_table;
+        std::size_t remaining_size = size;
+        std::size_t page_index = dest_addr >> PAGE_BITS;
+        std::size_t page_offset = dest_addr & PAGE_MASK;
+
+        while (remaining_size > 0) {
+            const std::size_t copy_amount =
+                std::min(static_cast<std::size_t>(PAGE_SIZE) - page_offset, remaining_size);
+            const auto current_vaddr = static_cast<VAddr>((page_index << PAGE_BITS) + page_offset);
+
+            switch (page_table.attributes[page_index]) {
+            case Common::PageType::Unmapped: {
+                LOG_ERROR(HW_Memory,
+                          "Unmapped WriteBlock @ 0x{:016X} (start address = 0x{:016X}, size = {})",
+                          current_vaddr, dest_addr, size);
+                break;
+            }
+            case Common::PageType::Memory: {
+                DEBUG_ASSERT(page_table.pointers[page_index]);
+
+                u8* const dest_ptr =
+                    page_table.pointers[page_index] + page_offset + (page_index << PAGE_BITS);
+                std::memcpy(dest_ptr, src_buffer, copy_amount);
+                break;
+            }
+            case Common::PageType::RasterizerCachedMemory: {
+                u8* const host_ptr = GetPointerFromVMA(process, current_vaddr);
                 std::memcpy(host_ptr, src_buffer, copy_amount);
                 break;
             }
@@ -309,6 +401,10 @@ struct Memory::Impl {
         WriteBlock(*system.CurrentProcess(), dest_addr, src_buffer, size);
     }
 
+    void WriteBlockUnsafe(const VAddr dest_addr, const void* src_buffer, const std::size_t size) {
+        WriteBlockUnsafe(*system.CurrentProcess(), dest_addr, src_buffer, size);
+    }
+
     void ZeroBlock(const Kernel::Process& process, const VAddr dest_addr, const std::size_t size) {
         const auto& page_table = process.VMManager().page_table;
         std::size_t remaining_size = size;
@@ -337,7 +433,7 @@ struct Memory::Impl {
             }
             case Common::PageType::RasterizerCachedMemory: {
                 u8* const host_ptr = GetPointerFromVMA(process, current_vaddr);
-                system.GPU().InvalidateRegion(ToCacheAddr(host_ptr), copy_amount);
+                system.GPU().InvalidateRegion(current_vaddr, copy_amount);
                 std::memset(host_ptr, 0, copy_amount);
                 break;
             }
@@ -384,7 +480,7 @@ struct Memory::Impl {
             }
             case Common::PageType::RasterizerCachedMemory: {
                 const u8* const host_ptr = GetPointerFromVMA(process, current_vaddr);
-                system.GPU().FlushRegion(ToCacheAddr(host_ptr), copy_amount);
+                system.GPU().FlushRegion(current_vaddr, copy_amount);
                 WriteBlock(process, dest_addr, host_ptr, copy_amount);
                 break;
             }
@@ -545,7 +641,7 @@ struct Memory::Impl {
             break;
         case Common::PageType::RasterizerCachedMemory: {
             const u8* const host_ptr = GetPointerFromVMA(vaddr);
-            system.GPU().FlushRegion(ToCacheAddr(host_ptr), sizeof(T));
+            system.GPU().FlushRegion(vaddr, sizeof(T));
             T value;
             std::memcpy(&value, host_ptr, sizeof(T));
             return value;
@@ -587,7 +683,7 @@ struct Memory::Impl {
             break;
         case Common::PageType::RasterizerCachedMemory: {
             u8* const host_ptr{GetPointerFromVMA(vaddr)};
-            system.GPU().InvalidateRegion(ToCacheAddr(host_ptr), sizeof(T));
+            system.GPU().InvalidateRegion(vaddr, sizeof(T));
             std::memcpy(host_ptr, &data, sizeof(T));
             break;
         }
@@ -696,6 +792,15 @@ void Memory::ReadBlock(const VAddr src_addr, void* dest_buffer, const std::size_
     impl->ReadBlock(src_addr, dest_buffer, size);
 }
 
+void Memory::ReadBlockUnsafe(const Kernel::Process& process, const VAddr src_addr,
+                             void* dest_buffer, const std::size_t size) {
+    impl->ReadBlockUnsafe(process, src_addr, dest_buffer, size);
+}
+
+void Memory::ReadBlockUnsafe(const VAddr src_addr, void* dest_buffer, const std::size_t size) {
+    impl->ReadBlockUnsafe(src_addr, dest_buffer, size);
+}
+
 void Memory::WriteBlock(const Kernel::Process& process, VAddr dest_addr, const void* src_buffer,
                         std::size_t size) {
     impl->WriteBlock(process, dest_addr, src_buffer, size);
@@ -705,6 +810,16 @@ void Memory::WriteBlock(const VAddr dest_addr, const void* src_buffer, const std
     impl->WriteBlock(dest_addr, src_buffer, size);
 }
 
+void Memory::WriteBlockUnsafe(const Kernel::Process& process, VAddr dest_addr,
+                              const void* src_buffer, std::size_t size) {
+    impl->WriteBlockUnsafe(process, dest_addr, src_buffer, size);
+}
+
+void Memory::WriteBlockUnsafe(const VAddr dest_addr, const void* src_buffer,
+                              const std::size_t size) {
+    impl->WriteBlockUnsafe(dest_addr, src_buffer, size);
+}
+
 void Memory::ZeroBlock(const Kernel::Process& process, VAddr dest_addr, std::size_t size) {
     impl->ZeroBlock(process, dest_addr, size);
 }

src/core/memory.h

@@ -294,6 +294,27 @@ public:
     void ReadBlock(const Kernel::Process& process, VAddr src_addr, void* dest_buffer,
                    std::size_t size);
 
+    /**
+     * Reads a contiguous block of bytes from a specified process' address space.
+     * This unsafe version does not trigger GPU flushing.
+     *
+     * @param process     The process to read the data from.
+     * @param src_addr    The virtual address to begin reading from.
+     * @param dest_buffer The buffer to place the read bytes into.
+     * @param size        The amount of data to read, in bytes.
+     *
+     * @note If a size of 0 is specified, then this function reads nothing and
+     *       no attempts to access memory are made at all.
+     *
+     * @pre dest_buffer must be at least size bytes in length, otherwise a
+     *      buffer overrun will occur.
+     *
+     * @post The range [dest_buffer, size) contains the read bytes from the
+     *       process' address space.
+     */
+    void ReadBlockUnsafe(const Kernel::Process& process, VAddr src_addr, void* dest_buffer,
+                         std::size_t size);
+
     /**
      * Reads a contiguous block of bytes from the current process' address space.
      *
@@ -312,6 +333,25 @@ public:
      */
     void ReadBlock(VAddr src_addr, void* dest_buffer, std::size_t size);
 
+    /**
+     * Reads a contiguous block of bytes from the current process' address space.
+     * This unsafe version does not trigger GPU flushing.
+     *
+     * @param src_addr    The virtual address to begin reading from.
+     * @param dest_buffer The buffer to place the read bytes into.
+     * @param size        The amount of data to read, in bytes.
+     *
+     * @note If a size of 0 is specified, then this function reads nothing and
+     *       no attempts to access memory are made at all.
+     *
+     * @pre dest_buffer must be at least size bytes in length, otherwise a
+     *      buffer overrun will occur.
+     *
+     * @post The range [dest_buffer, size) contains the read bytes from the
+     *       current process' address space.
+     */
+    void ReadBlockUnsafe(VAddr src_addr, void* dest_buffer, std::size_t size);
+
     /**
      * Writes a range of bytes into a given process' address space at the specified
      * virtual address.
@@ -335,6 +375,26 @@ public:
     void WriteBlock(const Kernel::Process& process, VAddr dest_addr, const void* src_buffer,
                     std::size_t size);
 
+    /**
+     * Writes a range of bytes into a given process' address space at the specified
+     * virtual address.
+     * This unsafe version does not invalidate GPU Memory.
+     *
+     * @param process    The process to write data into the address space of.
+     * @param dest_addr  The destination virtual address to begin writing the data at.
+     * @param src_buffer The data to write into the process' address space.
+     * @param size       The size of the data to write, in bytes.
+     *
+     * @post The address range [dest_addr, size) in the process' address space
+     *       contains the data that was within src_buffer.
+     *
+     * @post If an attempt is made to write into an unmapped region of memory, the writes
+     *       will be ignored and an error will be logged.
+     *
+     */
+    void WriteBlockUnsafe(const Kernel::Process& process, VAddr dest_addr, const void* src_buffer,
+                          std::size_t size);
+
     /**
      * Writes a range of bytes into the current process' address space at the specified
      * virtual address.
@@ -356,6 +416,24 @@ public:
      */
     void WriteBlock(VAddr dest_addr, const void* src_buffer, std::size_t size);
 
+    /**
+     * Writes a range of bytes into the current process' address space at the specified
+     * virtual address.
+     * This unsafe version does not invalidate GPU Memory.
+     *
+     * @param dest_addr  The destination virtual address to begin writing the data at.
+     * @param src_buffer The data to write into the current process' address space.
+     * @param size       The size of the data to write, in bytes.
+     *
+     * @post The address range [dest_addr, size) in the current process' address space
+     *       contains the data that was within src_buffer.
+     *
+     * @post If an attempt is made to write into an unmapped region of memory, the writes
+     *       will be ignored and an error will be logged.
+     *
+     */
+    void WriteBlockUnsafe(VAddr dest_addr, const void* src_buffer, std::size_t size);
+
     /**
      * Fills the specified address range within a process' address space with zeroes.
      *

src/core/settings.cpp

@@ -86,6 +86,7 @@ void LogSettings() {
     LogSetting("System_RngSeed", Settings::values.rng_seed.value_or(0));
     LogSetting("System_CurrentUser", Settings::values.current_user);
     LogSetting("System_LanguageIndex", Settings::values.language_index);
+    LogSetting("System_RegionIndex", Settings::values.region_index);
     LogSetting("Core_UseMultiCore", Settings::values.use_multi_core);
     LogSetting("Renderer_UseResolutionFactor", Settings::values.resolution_factor);
     LogSetting("Renderer_UseFrameLimit", Settings::values.use_frame_limit);

src/core/settings.h

@@ -387,6 +387,8 @@ struct Values {
 
     s32 current_user;
     s32 language_index;
+    s32 region_index;
+    s32 sound_index;
 
     // Controls
     std::array<PlayerInput, 10> players;
@@ -430,6 +432,7 @@ struct Values {
 
     float resolution_factor;
     int aspect_ratio;
+    int max_anisotropy;
     bool use_frame_limit;
     u16 frame_limit;
     bool use_disk_shader_cache;

src/input_common/udp/client.cpp

@@ -32,8 +32,16 @@ public:
                     SocketCallback callback)
         : callback(std::move(callback)), timer(io_service),
           socket(io_service, udp::endpoint(udp::v4(), 0)), client_id(client_id),
-          pad_index(pad_index),
-          send_endpoint(udp::endpoint(boost::asio::ip::make_address_v4(host), port)) {}
+          pad_index(pad_index) {
+        boost::system::error_code ec{};
+        auto ipv4 = boost::asio::ip::make_address_v4(host, ec);
+        if (ec.value() != boost::system::errc::success) {
+            LOG_ERROR(Input, "Invalid IPv4 address \"{}\" provided to socket", host);
+            ipv4 = boost::asio::ip::address_v4{};
+        }
+
+        send_endpoint = {udp::endpoint(ipv4, port)};
+    }
 
     void Stop() {
         io_service.stop();
@@ -85,17 +93,18 @@ private:
     }
 
     void HandleSend(const boost::system::error_code& error) {
+        boost::system::error_code _ignored{};
         // Send a request for getting port info for the pad
         Request::PortInfo port_info{1, {pad_index, 0, 0, 0}};
         const auto port_message = Request::Create(port_info, client_id);
         std::memcpy(&send_buffer1, &port_message, PORT_INFO_SIZE);
-        socket.send_to(boost::asio::buffer(send_buffer1), send_endpoint);
+        socket.send_to(boost::asio::buffer(send_buffer1), send_endpoint, {}, _ignored);
 
         // Send a request for getting pad data for the pad
         Request::PadData pad_data{Request::PadData::Flags::Id, pad_index, EMPTY_MAC_ADDRESS};
         const auto pad_message = Request::Create(pad_data, client_id);
         std::memcpy(send_buffer2.data(), &pad_message, PAD_DATA_SIZE);
-        socket.send_to(boost::asio::buffer(send_buffer2), send_endpoint);
+        socket.send_to(boost::asio::buffer(send_buffer2), send_endpoint, {}, _ignored);
         StartSend(timer.expiry());
     }
 

src/input_common/udp/protocol.cpp

@@ -31,7 +31,6 @@ namespace Response {
  */
 std::optional<Type> Validate(u8* data, std::size_t size) {
     if (size < sizeof(Header)) {
-        LOG_DEBUG(Input, "Invalid UDP packet received");
         return std::nullopt;
     }
     Header header{};

src/input_common/udp/udp.cpp

@@ -3,6 +3,7 @@
 // Refer to the license.txt file included.
 
 #include <mutex>
+#include <optional>
 #include <tuple>
 
 #include "common/param_package.h"
@@ -44,7 +45,7 @@ public:
     std::unique_ptr<Input::TouchDevice> Create(const Common::ParamPackage& params) override {
         {
             std::lock_guard guard(status->update_mutex);
-            status->touch_calibration.emplace();
+            status->touch_calibration = DeviceStatus::CalibrationData{};
             // These default values work well for DS4 but probably not other touch inputs
             status->touch_calibration->min_x = params.Get("min_x", 100);
             status->touch_calibration->min_y = params.Get("min_y", 50);

src/video_core/CMakeLists.txt

@@ -2,6 +2,8 @@ add_library(video_core STATIC
     buffer_cache/buffer_block.h
     buffer_cache/buffer_cache.h
     buffer_cache/map_interval.h
+    dirty_flags.cpp
+    dirty_flags.h
     dma_pusher.cpp
     dma_pusher.h
     engines/const_buffer_engine_interface.h
@@ -63,14 +65,12 @@ add_library(video_core STATIC
     renderer_opengl/gl_shader_decompiler.h
     renderer_opengl/gl_shader_disk_cache.cpp
     renderer_opengl/gl_shader_disk_cache.h
-    renderer_opengl/gl_shader_gen.cpp
-    renderer_opengl/gl_shader_gen.h
     renderer_opengl/gl_shader_manager.cpp
     renderer_opengl/gl_shader_manager.h
     renderer_opengl/gl_shader_util.cpp
     renderer_opengl/gl_shader_util.h
-    renderer_opengl/gl_state.cpp
-    renderer_opengl/gl_state.h
+    renderer_opengl/gl_state_tracker.cpp
+    renderer_opengl/gl_state_tracker.h
     renderer_opengl/gl_stream_buffer.cpp
     renderer_opengl/gl_stream_buffer.h
     renderer_opengl/gl_texture_cache.cpp
@@ -116,8 +116,6 @@ add_library(video_core STATIC
     shader/ast.h
     shader/compiler_settings.cpp
     shader/compiler_settings.h
-    shader/const_buffer_locker.cpp
-    shader/const_buffer_locker.h
     shader/control_flow.cpp
     shader/control_flow.h
     shader/decode.cpp
@@ -126,9 +124,13 @@ add_library(video_core STATIC
     shader/node_helper.cpp
     shader/node_helper.h
     shader/node.h
+    shader/registry.cpp
+    shader/registry.h
     shader/shader_ir.cpp
     shader/shader_ir.h
     shader/track.cpp
+    shader/transform_feedback.cpp
+    shader/transform_feedback.h
     surface.cpp
     surface.h
     texture_cache/format_lookup_table.cpp
@@ -146,6 +148,7 @@ add_library(video_core STATIC
     textures/convert.h
     textures/decoders.cpp
     textures/decoders.h
+    textures/texture.cpp
     textures/texture.h
     video_core.cpp
     video_core.h
@@ -153,11 +156,12 @@ add_library(video_core STATIC
 
 if (ENABLE_VULKAN)
     target_sources(video_core PRIVATE
-        renderer_vulkan/declarations.h
         renderer_vulkan/fixed_pipeline_state.cpp
         renderer_vulkan/fixed_pipeline_state.h
         renderer_vulkan/maxwell_to_vk.cpp
         renderer_vulkan/maxwell_to_vk.h
+        renderer_vulkan/nsight_aftermath_tracker.cpp
+        renderer_vulkan/nsight_aftermath_tracker.h
         renderer_vulkan/renderer_vulkan.h
         renderer_vulkan/renderer_vulkan.cpp
         renderer_vulkan/vk_blit_screen.cpp
@@ -198,6 +202,8 @@ if (ENABLE_VULKAN)
         renderer_vulkan/vk_shader_util.h
         renderer_vulkan/vk_staging_buffer_pool.cpp
         renderer_vulkan/vk_staging_buffer_pool.h
+        renderer_vulkan/vk_state_tracker.cpp
+        renderer_vulkan/vk_state_tracker.h
         renderer_vulkan/vk_stream_buffer.cpp
         renderer_vulkan/vk_stream_buffer.h
         renderer_vulkan/vk_swapchain.cpp
@@ -206,20 +212,33 @@ if (ENABLE_VULKAN)
         renderer_vulkan/vk_texture_cache.h
         renderer_vulkan/vk_update_descriptor.cpp
         renderer_vulkan/vk_update_descriptor.h
+        renderer_vulkan/wrapper.cpp
+        renderer_vulkan/wrapper.h
     )
-
-    target_include_directories(video_core PRIVATE sirit ../../externals/Vulkan-Headers/include)
-    target_compile_definitions(video_core PRIVATE HAS_VULKAN)
 endif()
 
 create_target_directory_groups(video_core)
 
 target_link_libraries(video_core PUBLIC common core)
 target_link_libraries(video_core PRIVATE glad)
+
 if (ENABLE_VULKAN)
+    target_include_directories(video_core PRIVATE sirit ../../externals/Vulkan-Headers/include)
+    target_compile_definitions(video_core PRIVATE HAS_VULKAN)
     target_link_libraries(video_core PRIVATE sirit)
 endif()
 
+if (ENABLE_NSIGHT_AFTERMATH)
+    if (NOT DEFINED ENV{NSIGHT_AFTERMATH_SDK})
+        message(ERROR "Environment variable NSIGHT_AFTERMATH_SDK has to be provided")
+    endif()
+    if (NOT WIN32)
+        message(ERROR "Nsight Aftermath doesn't support non-Windows platforms")
+    endif()
+    target_compile_definitions(video_core PRIVATE HAS_NSIGHT_AFTERMATH)
+    target_include_directories(video_core PRIVATE "$ENV{NSIGHT_AFTERMATH_SDK}/include")
+endif()
+
 if (MSVC)
     target_compile_options(video_core PRIVATE /we4267)
 else()

src/video_core/buffer_cache/buffer_block.h

@@ -15,37 +15,29 @@ namespace VideoCommon {
 
 class BufferBlock {
 public:
-    bool Overlaps(const CacheAddr start, const CacheAddr end) const {
-        return (cache_addr < end) && (cache_addr_end > start);
+    bool Overlaps(const VAddr start, const VAddr end) const {
+        return (cpu_addr < end) && (cpu_addr_end > start);
     }
 
-    bool IsInside(const CacheAddr other_start, const CacheAddr other_end) const {
-        return cache_addr <= other_start && other_end <= cache_addr_end;
+    bool IsInside(const VAddr other_start, const VAddr other_end) const {
+        return cpu_addr <= other_start && other_end <= cpu_addr_end;
     }
 
-    u8* GetWritableHostPtr() const {
-        return FromCacheAddr(cache_addr);
+    std::size_t GetOffset(const VAddr in_addr) {
+        return static_cast<std::size_t>(in_addr - cpu_addr);
     }
 
-    u8* GetWritableHostPtr(std::size_t offset) const {
-        return FromCacheAddr(cache_addr + offset);
+    VAddr GetCpuAddr() const {
+        return cpu_addr;
     }
 
-    std::size_t GetOffset(const CacheAddr in_addr) {
-        return static_cast<std::size_t>(in_addr - cache_addr);
+    VAddr GetCpuAddrEnd() const {
+        return cpu_addr_end;
     }
 
-    CacheAddr GetCacheAddr() const {
-        return cache_addr;
-    }
-
-    CacheAddr GetCacheAddrEnd() const {
-        return cache_addr_end;
-    }
-
-    void SetCacheAddr(const CacheAddr new_addr) {
-        cache_addr = new_addr;
-        cache_addr_end = new_addr + size;
+    void SetCpuAddr(const VAddr new_addr) {
+        cpu_addr = new_addr;
+        cpu_addr_end = new_addr + size;
     }
 
     std::size_t GetSize() const {
@@ -61,14 +53,14 @@ public:
     }
 
 protected:
-    explicit BufferBlock(CacheAddr cache_addr, const std::size_t size) : size{size} {
-        SetCacheAddr(cache_addr);
+    explicit BufferBlock(VAddr cpu_addr, const std::size_t size) : size{size} {
+        SetCpuAddr(cpu_addr);
     }
     ~BufferBlock() = default;
 
 private:
-    CacheAddr cache_addr{};
-    CacheAddr cache_addr_end{};
+    VAddr cpu_addr{};
+    VAddr cpu_addr_end{};
     std::size_t size{};
     u64 epoch{};
 };

src/video_core/buffer_cache/buffer_cache.h

@@ -19,6 +19,7 @@
 #include "common/alignment.h"
 #include "common/common_types.h"
 #include "core/core.h"
+#include "core/memory.h"
 #include "video_core/buffer_cache/buffer_block.h"
 #include "video_core/buffer_cache/map_interval.h"
 #include "video_core/memory_manager.h"
@@ -37,28 +38,45 @@ public:
                             bool is_written = false, bool use_fast_cbuf = false) {
         std::lock_guard lock{mutex};
 
-        auto& memory_manager = system.GPU().MemoryManager();
-        const auto host_ptr = memory_manager.GetPointer(gpu_addr);
-        if (!host_ptr) {
+        const std::optional<VAddr> cpu_addr_opt =
+            system.GPU().MemoryManager().GpuToCpuAddress(gpu_addr);
+
+        if (!cpu_addr_opt) {
             return {GetEmptyBuffer(size), 0};
         }
-        const auto cache_addr = ToCacheAddr(host_ptr);
+
+        VAddr cpu_addr = *cpu_addr_opt;
 
         // Cache management is a big overhead, so only cache entries with a given size.
         // TODO: Figure out which size is the best for given games.
         constexpr std::size_t max_stream_size = 0x800;
         if (use_fast_cbuf || size < max_stream_size) {
-            if (!is_written && !IsRegionWritten(cache_addr, cache_addr + size - 1)) {
+            if (!is_written && !IsRegionWritten(cpu_addr, cpu_addr + size - 1)) {
+                auto& memory_manager = system.GPU().MemoryManager();
                 if (use_fast_cbuf) {
-                    return ConstBufferUpload(host_ptr, size);
+                    if (memory_manager.IsGranularRange(gpu_addr, size)) {
+                        const auto host_ptr = memory_manager.GetPointer(gpu_addr);
+                        return ConstBufferUpload(host_ptr, size);
+                    } else {
+                        staging_buffer.resize(size);
+                        memory_manager.ReadBlockUnsafe(gpu_addr, staging_buffer.data(), size);
+                        return ConstBufferUpload(staging_buffer.data(), size);
+                    }
                 } else {
-                    return StreamBufferUpload(host_ptr, size, alignment);
+                    if (memory_manager.IsGranularRange(gpu_addr, size)) {
+                        const auto host_ptr = memory_manager.GetPointer(gpu_addr);
+                        return StreamBufferUpload(host_ptr, size, alignment);
+                    } else {
+                        staging_buffer.resize(size);
+                        memory_manager.ReadBlockUnsafe(gpu_addr, staging_buffer.data(), size);
+                        return StreamBufferUpload(staging_buffer.data(), size, alignment);
+                    }
                 }
             }
         }
 
-        auto block = GetBlock(cache_addr, size);
-        auto map = MapAddress(block, gpu_addr, cache_addr, size);
+        auto block = GetBlock(cpu_addr, size);
+        auto map = MapAddress(block, gpu_addr, cpu_addr, size);
         if (is_written) {
             map->MarkAsModified(true, GetModifiedTicks());
             if (!map->IsWritten()) {
@@ -71,7 +89,7 @@ public:
             }
         }
 
-        const u64 offset = static_cast<u64>(block->GetOffset(cache_addr));
+        const u64 offset = static_cast<u64>(block->GetOffset(cpu_addr));
 
         return {ToHandle(block), offset};
     }
@@ -112,7 +130,7 @@ public:
     }
 
     /// Write any cached resources overlapping the specified region back to memory
-    void FlushRegion(CacheAddr addr, std::size_t size) {
+    void FlushRegion(VAddr addr, std::size_t size) {
         std::lock_guard lock{mutex};
 
         std::vector<MapInterval> objects = GetMapsInRange(addr, size);
@@ -127,7 +145,7 @@ public:
     }
 
     /// Mark the specified region as being invalidated
-    void InvalidateRegion(CacheAddr addr, u64 size) {
+    void InvalidateRegion(VAddr addr, u64 size) {
         std::lock_guard lock{mutex};
 
         std::vector<MapInterval> objects = GetMapsInRange(addr, size);
@@ -152,7 +170,7 @@ protected:
 
     virtual void WriteBarrier() = 0;
 
-    virtual TBuffer CreateBlock(CacheAddr cache_addr, std::size_t size) = 0;
+    virtual TBuffer CreateBlock(VAddr cpu_addr, std::size_t size) = 0;
 
     virtual void UploadBlockData(const TBuffer& buffer, std::size_t offset, std::size_t size,
                                  const u8* data) = 0;
@@ -169,20 +187,17 @@ protected:
 
     /// Register an object into the cache
     void Register(const MapInterval& new_map, bool inherit_written = false) {
-        const CacheAddr cache_ptr = new_map->GetStart();
-        const std::optional<VAddr> cpu_addr =
-            system.GPU().MemoryManager().GpuToCpuAddress(new_map->GetGpuAddress());
-        if (!cache_ptr || !cpu_addr) {
+        const VAddr cpu_addr = new_map->GetStart();
+        if (!cpu_addr) {
             LOG_CRITICAL(HW_GPU, "Failed to register buffer with unmapped gpu_address 0x{:016x}",
                          new_map->GetGpuAddress());
             return;
         }
         const std::size_t size = new_map->GetEnd() - new_map->GetStart();
-        new_map->SetCpuAddress(*cpu_addr);
         new_map->MarkAsRegistered(true);
         const IntervalType interval{new_map->GetStart(), new_map->GetEnd()};
         mapped_addresses.insert({interval, new_map});
-        rasterizer.UpdatePagesCachedCount(*cpu_addr, size, 1);
+        rasterizer.UpdatePagesCachedCount(cpu_addr, size, 1);
         if (inherit_written) {
             MarkRegionAsWritten(new_map->GetStart(), new_map->GetEnd() - 1);
             new_map->MarkAsWritten(true);
@@ -192,7 +207,7 @@ protected:
     /// Unregisters an object from the cache
     void Unregister(MapInterval& map) {
         const std::size_t size = map->GetEnd() - map->GetStart();
-        rasterizer.UpdatePagesCachedCount(map->GetCpuAddress(), size, -1);
+        rasterizer.UpdatePagesCachedCount(map->GetStart(), size, -1);
         map->MarkAsRegistered(false);
         if (map->IsWritten()) {
             UnmarkRegionAsWritten(map->GetStart(), map->GetEnd() - 1);
@@ -202,32 +217,39 @@ protected:
     }
 
 private:
-    MapInterval CreateMap(const CacheAddr start, const CacheAddr end, const GPUVAddr gpu_addr) {
+    MapInterval CreateMap(const VAddr start, const VAddr end, const GPUVAddr gpu_addr) {
         return std::make_shared<MapIntervalBase>(start, end, gpu_addr);
     }
 
-    MapInterval MapAddress(const TBuffer& block, const GPUVAddr gpu_addr,
-                           const CacheAddr cache_addr, const std::size_t size) {
+    MapInterval MapAddress(const TBuffer& block, const GPUVAddr gpu_addr, const VAddr cpu_addr,
+                           const std::size_t size) {
 
-        std::vector<MapInterval> overlaps = GetMapsInRange(cache_addr, size);
+        std::vector<MapInterval> overlaps = GetMapsInRange(cpu_addr, size);
         if (overlaps.empty()) {
-            const CacheAddr cache_addr_end = cache_addr + size;
-            MapInterval new_map = CreateMap(cache_addr, cache_addr_end, gpu_addr);
-            u8* host_ptr = FromCacheAddr(cache_addr);
-            UploadBlockData(block, block->GetOffset(cache_addr), size, host_ptr);
+            auto& memory_manager = system.GPU().MemoryManager();
+            const VAddr cpu_addr_end = cpu_addr + size;
+            MapInterval new_map = CreateMap(cpu_addr, cpu_addr_end, gpu_addr);
+            if (memory_manager.IsGranularRange(gpu_addr, size)) {
+                u8* host_ptr = memory_manager.GetPointer(gpu_addr);
+                UploadBlockData(block, block->GetOffset(cpu_addr), size, host_ptr);
+            } else {
+                staging_buffer.resize(size);
+                memory_manager.ReadBlockUnsafe(gpu_addr, staging_buffer.data(), size);
+                UploadBlockData(block, block->GetOffset(cpu_addr), size, staging_buffer.data());
+            }
             Register(new_map);
             return new_map;
         }
 
-        const CacheAddr cache_addr_end = cache_addr + size;
+        const VAddr cpu_addr_end = cpu_addr + size;
         if (overlaps.size() == 1) {
             MapInterval& current_map = overlaps[0];
-            if (current_map->IsInside(cache_addr, cache_addr_end)) {
+            if (current_map->IsInside(cpu_addr, cpu_addr_end)) {
                 return current_map;
             }
         }
-        CacheAddr new_start = cache_addr;
-        CacheAddr new_end = cache_addr_end;
+        VAddr new_start = cpu_addr;
+        VAddr new_end = cpu_addr_end;
         bool write_inheritance = false;
         bool modified_inheritance = false;
         // Calculate new buffer parameters
@@ -237,7 +259,7 @@ private:
             write_inheritance |= overlap->IsWritten();
             modified_inheritance |= overlap->IsModified();
         }
-        GPUVAddr new_gpu_addr = gpu_addr + new_start - cache_addr;
+        GPUVAddr new_gpu_addr = gpu_addr + new_start - cpu_addr;
         for (auto& overlap : overlaps) {
             Unregister(overlap);
         }
@@ -250,7 +272,7 @@ private:
         return new_map;
     }
 
-    void UpdateBlock(const TBuffer& block, CacheAddr start, CacheAddr end,
+    void UpdateBlock(const TBuffer& block, VAddr start, VAddr end,
                      std::vector<MapInterval>& overlaps) {
         const IntervalType base_interval{start, end};
         IntervalSet interval_set{};
@@ -262,13 +284,15 @@ private:
         for (auto& interval : interval_set) {
             std::size_t size = interval.upper() - interval.lower();
             if (size > 0) {
-                u8* host_ptr = FromCacheAddr(interval.lower());
-                UploadBlockData(block, block->GetOffset(interval.lower()), size, host_ptr);
+                staging_buffer.resize(size);
+                system.Memory().ReadBlockUnsafe(interval.lower(), staging_buffer.data(), size);
+                UploadBlockData(block, block->GetOffset(interval.lower()), size,
+                                staging_buffer.data());
             }
         }
     }
 
-    std::vector<MapInterval> GetMapsInRange(CacheAddr addr, std::size_t size) {
+    std::vector<MapInterval> GetMapsInRange(VAddr addr, std::size_t size) {
         if (size == 0) {
             return {};
         }
@@ -290,8 +314,9 @@ private:
     void FlushMap(MapInterval map) {
         std::size_t size = map->GetEnd() - map->GetStart();
         TBuffer block = blocks[map->GetStart() >> block_page_bits];
-        u8* host_ptr = FromCacheAddr(map->GetStart());
-        DownloadBlockData(block, block->GetOffset(map->GetStart()), size, host_ptr);
+        staging_buffer.resize(size);
+        DownloadBlockData(block, block->GetOffset(map->GetStart()), size, staging_buffer.data());
+        system.Memory().WriteBlockUnsafe(map->GetStart(), staging_buffer.data(), size);
         map->MarkAsModified(false, 0);
     }
 
@@ -316,14 +341,14 @@ private:
     TBuffer EnlargeBlock(TBuffer buffer) {
         const std::size_t old_size = buffer->GetSize();
         const std::size_t new_size = old_size + block_page_size;
-        const CacheAddr cache_addr = buffer->GetCacheAddr();
-        TBuffer new_buffer = CreateBlock(cache_addr, new_size);
+        const VAddr cpu_addr = buffer->GetCpuAddr();
+        TBuffer new_buffer = CreateBlock(cpu_addr, new_size);
         CopyBlock(buffer, new_buffer, 0, 0, old_size);
         buffer->SetEpoch(epoch);
         pending_destruction.push_back(buffer);
-        const CacheAddr cache_addr_end = cache_addr + new_size - 1;
-        u64 page_start = cache_addr >> block_page_bits;
-        const u64 page_end = cache_addr_end >> block_page_bits;
+        const VAddr cpu_addr_end = cpu_addr + new_size - 1;
+        u64 page_start = cpu_addr >> block_page_bits;
+        const u64 page_end = cpu_addr_end >> block_page_bits;
         while (page_start <= page_end) {
             blocks[page_start] = new_buffer;
             ++page_start;
@@ -334,9 +359,9 @@ private:
     TBuffer MergeBlocks(TBuffer first, TBuffer second) {
         const std::size_t size_1 = first->GetSize();
         const std::size_t size_2 = second->GetSize();
-        const CacheAddr first_addr = first->GetCacheAddr();
-        const CacheAddr second_addr = second->GetCacheAddr();
-        const CacheAddr new_addr = std::min(first_addr, second_addr);
+        const VAddr first_addr = first->GetCpuAddr();
+        const VAddr second_addr = second->GetCpuAddr();
+        const VAddr new_addr = std::min(first_addr, second_addr);
         const std::size_t new_size = size_1 + size_2;
         TBuffer new_buffer = CreateBlock(new_addr, new_size);
         CopyBlock(first, new_buffer, 0, new_buffer->GetOffset(first_addr), size_1);
@@ -345,9 +370,9 @@ private:
         second->SetEpoch(epoch);
         pending_destruction.push_back(first);
         pending_destruction.push_back(second);
-        const CacheAddr cache_addr_end = new_addr + new_size - 1;
+        const VAddr cpu_addr_end = new_addr + new_size - 1;
         u64 page_start = new_addr >> block_page_bits;
-        const u64 page_end = cache_addr_end >> block_page_bits;
+        const u64 page_end = cpu_addr_end >> block_page_bits;
         while (page_start <= page_end) {
             blocks[page_start] = new_buffer;
             ++page_start;
@@ -355,18 +380,18 @@ private:
         return new_buffer;
     }
 
-    TBuffer GetBlock(const CacheAddr cache_addr, const std::size_t size) {
+    TBuffer GetBlock(const VAddr cpu_addr, const std::size_t size) {
         TBuffer found{};
-        const CacheAddr cache_addr_end = cache_addr + size - 1;
-        u64 page_start = cache_addr >> block_page_bits;
-        const u64 page_end = cache_addr_end >> block_page_bits;
+        const VAddr cpu_addr_end = cpu_addr + size - 1;
+        u64 page_start = cpu_addr >> block_page_bits;
+        const u64 page_end = cpu_addr_end >> block_page_bits;
         while (page_start <= page_end) {
             auto it = blocks.find(page_start);
             if (it == blocks.end()) {
                 if (found) {
                     found = EnlargeBlock(found);
                 } else {
-                    const CacheAddr start_addr = (page_start << block_page_bits);
+                    const VAddr start_addr = (page_start << block_page_bits);
                     found = CreateBlock(start_addr, block_page_size);
                     blocks[page_start] = found;
                 }
@@ -386,7 +411,7 @@ private:
         return found;
     }
 
-    void MarkRegionAsWritten(const CacheAddr start, const CacheAddr end) {
+    void MarkRegionAsWritten(const VAddr start, const VAddr end) {
         u64 page_start = start >> write_page_bit;
         const u64 page_end = end >> write_page_bit;
         while (page_start <= page_end) {
@@ -400,7 +425,7 @@ private:
         }
     }
 
-    void UnmarkRegionAsWritten(const CacheAddr start, const CacheAddr end) {
+    void UnmarkRegionAsWritten(const VAddr start, const VAddr end) {
         u64 page_start = start >> write_page_bit;
         const u64 page_end = end >> write_page_bit;
         while (page_start <= page_end) {
@@ -416,7 +441,7 @@ private:
         }
     }
 
-    bool IsRegionWritten(const CacheAddr start, const CacheAddr end) const {
+    bool IsRegionWritten(const VAddr start, const VAddr end) const {
         u64 page_start = start >> write_page_bit;
         const u64 page_end = end >> write_page_bit;
         while (page_start <= page_end) {
@@ -440,8 +465,8 @@ private:
     u64 buffer_offset = 0;
     u64 buffer_offset_base = 0;
 
-    using IntervalSet = boost::icl::interval_set<CacheAddr>;
-    using IntervalCache = boost::icl::interval_map<CacheAddr, MapInterval>;
+    using IntervalSet = boost::icl::interval_set<VAddr>;
+    using IntervalCache = boost::icl::interval_map<VAddr, MapInterval>;
     using IntervalType = typename IntervalCache::interval_type;
     IntervalCache mapped_addresses;
 
@@ -456,6 +481,8 @@ private:
     u64 epoch = 0;
     u64 modified_ticks = 0;
 
+    std::vector<u8> staging_buffer;
+
     std::recursive_mutex mutex;
 };
 

src/video_core/buffer_cache/map_interval.h

@@ -11,7 +11,7 @@ namespace VideoCommon {
 
 class MapIntervalBase {
 public:
-    MapIntervalBase(const CacheAddr start, const CacheAddr end, const GPUVAddr gpu_addr)
+    MapIntervalBase(const VAddr start, const VAddr end, const GPUVAddr gpu_addr)
         : start{start}, end{end}, gpu_addr{gpu_addr} {}
 
     void SetCpuAddress(VAddr new_cpu_addr) {
@@ -26,7 +26,7 @@ public:
         return gpu_addr;
     }
 
-    bool IsInside(const CacheAddr other_start, const CacheAddr other_end) const {
+    bool IsInside(const VAddr other_start, const VAddr other_end) const {
         return (start <= other_start && other_end <= end);
     }
 
@@ -46,11 +46,11 @@ public:
         return is_registered;
     }
 
-    CacheAddr GetStart() const {
+    VAddr GetStart() const {
         return start;
     }
 
-    CacheAddr GetEnd() const {
+    VAddr GetEnd() const {
         return end;
     }
 
@@ -76,8 +76,8 @@ public:
     }
 
 private:
-    CacheAddr start;
-    CacheAddr end;
+    VAddr start;
+    VAddr end;
     GPUVAddr gpu_addr;
     VAddr cpu_addr{};
     bool is_written{};

src/video_core/dirty_flags.cpp

@@ -0,0 +1,38 @@
+// Copyright 2019 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <array>
+#include <cstddef>
+
+#include "common/common_types.h"
+#include "video_core/dirty_flags.h"
+
+#define OFF(field_name) MAXWELL3D_REG_INDEX(field_name)
+#define NUM(field_name) (sizeof(::Tegra::Engines::Maxwell3D::Regs::field_name) / sizeof(u32))
+
+namespace VideoCommon::Dirty {
+
+using Tegra::Engines::Maxwell3D;
+
+void SetupDirtyRenderTargets(Tegra::Engines::Maxwell3D::DirtyState::Tables& tables) {
+    static constexpr std::size_t num_per_rt = NUM(rt[0]);
+    static constexpr std::size_t begin = OFF(rt);
+    static constexpr std::size_t num = num_per_rt * Maxwell3D::Regs::NumRenderTargets;
+    for (std::size_t rt = 0; rt < Maxwell3D::Regs::NumRenderTargets; ++rt) {
+        FillBlock(tables[0], begin + rt * num_per_rt, num_per_rt, ColorBuffer0 + rt);
+    }
+    FillBlock(tables[1], begin, num, RenderTargets);
+
+    static constexpr std::array zeta_flags{ZetaBuffer, RenderTargets};
+    for (std::size_t i = 0; i < std::size(zeta_flags); ++i) {
+        const u8 flag = zeta_flags[i];
+        auto& table = tables[i];
+        table[OFF(zeta_enable)] = flag;
+        table[OFF(zeta_width)] = flag;
+        table[OFF(zeta_height)] = flag;
+        FillBlock(table, OFF(zeta), NUM(zeta), flag);
+    }
+}
+
+} // namespace VideoCommon::Dirty

src/video_core/dirty_flags.h

@@ -0,0 +1,49 @@
+// Copyright 2019 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <algorithm>
+#include <cstddef>
+#include <iterator>
+
+#include "common/common_types.h"
+#include "video_core/engines/maxwell_3d.h"
+
+namespace VideoCommon::Dirty {
+
+enum : u8 {
+    NullEntry = 0,
+
+    RenderTargets,
+    ColorBuffer0,
+    ColorBuffer1,
+    ColorBuffer2,
+    ColorBuffer3,
+    ColorBuffer4,
+    ColorBuffer5,
+    ColorBuffer6,
+    ColorBuffer7,
+    ZetaBuffer,
+
+    LastCommonEntry,
+};
+
+template <typename Integer>
+void FillBlock(Tegra::Engines::Maxwell3D::DirtyState::Table& table, std::size_t begin,
+               std::size_t num, Integer dirty_index) {
+    const auto it = std::begin(table) + begin;
+    std::fill(it, it + num, static_cast<u8>(dirty_index));
+}
+
+template <typename Integer1, typename Integer2>
+void FillBlock(Tegra::Engines::Maxwell3D::DirtyState::Tables& tables, std::size_t begin,
+               std::size_t num, Integer1 index_a, Integer2 index_b) {
+    FillBlock(tables[0], begin, num, index_a);
+    FillBlock(tables[1], begin, num, index_b);
+}
+
+void SetupDirtyRenderTargets(Tegra::Engines::Maxwell3D::DirtyState::Tables& tables);
+
+} // namespace VideoCommon::Dirty

src/video_core/dma_pusher.cpp

@@ -22,7 +22,7 @@ void DmaPusher::DispatchCalls() {
     MICROPROFILE_SCOPE(DispatchCalls);
 
     // On entering GPU code, assume all memory may be touched by the ARM core.
-    gpu.Maxwell3D().dirty.OnMemoryWrite();
+    gpu.Maxwell3D().OnMemoryWrite();
 
     dma_pushbuffer_subindex = 0;
 

src/video_core/engines/const_buffer_engine_interface.h

@@ -16,11 +16,16 @@ namespace Tegra::Engines {
 
 struct SamplerDescriptor {
     union {
-        BitField<0, 20, Tegra::Shader::TextureType> texture_type;
-        BitField<20, 1, u32> is_array;
-        BitField<21, 1, u32> is_buffer;
-        BitField<22, 1, u32> is_shadow;
-        u32 raw{};
+        u32 raw = 0;
+        BitField<0, 2, Tegra::Shader::TextureType> texture_type;
+        BitField<2, 3, Tegra::Texture::ComponentType> r_type;
+        BitField<5, 1, u32> is_array;
+        BitField<6, 1, u32> is_buffer;
+        BitField<7, 1, u32> is_shadow;
+        BitField<8, 3, Tegra::Texture::ComponentType> g_type;
+        BitField<11, 3, Tegra::Texture::ComponentType> b_type;
+        BitField<14, 3, Tegra::Texture::ComponentType> a_type;
+        BitField<17, 7, Tegra::Texture::TextureFormat> format;
     };
 
     bool operator==(const SamplerDescriptor& rhs) const noexcept {
@@ -31,68 +36,50 @@ struct SamplerDescriptor {
         return !operator==(rhs);
     }
 
-    static SamplerDescriptor FromTicTexture(Tegra::Texture::TextureType tic_texture_type) {
+    static SamplerDescriptor FromTIC(const Tegra::Texture::TICEntry& tic) {
+        using Tegra::Shader::TextureType;
         SamplerDescriptor result;
-        switch (tic_texture_type) {
+
+        result.format.Assign(tic.format.Value());
+        result.r_type.Assign(tic.r_type.Value());
+        result.g_type.Assign(tic.g_type.Value());
+        result.b_type.Assign(tic.b_type.Value());
+        result.a_type.Assign(tic.a_type.Value());
+
+        switch (tic.texture_type.Value()) {
         case Tegra::Texture::TextureType::Texture1D:
-            result.texture_type.Assign(Tegra::Shader::TextureType::Texture1D);
-            result.is_array.Assign(0);
-            result.is_buffer.Assign(0);
-            result.is_shadow.Assign(0);
+            result.texture_type.Assign(TextureType::Texture1D);
             return result;
         case Tegra::Texture::TextureType::Texture2D:
-            result.texture_type.Assign(Tegra::Shader::TextureType::Texture2D);
-            result.is_array.Assign(0);
-            result.is_buffer.Assign(0);
-            result.is_shadow.Assign(0);
+            result.texture_type.Assign(TextureType::Texture2D);
             return result;
         case Tegra::Texture::TextureType::Texture3D:
-            result.texture_type.Assign(Tegra::Shader::TextureType::Texture3D);
-            result.is_array.Assign(0);
-            result.is_buffer.Assign(0);
-            result.is_shadow.Assign(0);
+            result.texture_type.Assign(TextureType::Texture3D);
             return result;
         case Tegra::Texture::TextureType::TextureCubemap:
-            result.texture_type.Assign(Tegra::Shader::TextureType::TextureCube);
-            result.is_array.Assign(0);
-            result.is_buffer.Assign(0);
-            result.is_shadow.Assign(0);
+            result.texture_type.Assign(TextureType::TextureCube);
             return result;
         case Tegra::Texture::TextureType::Texture1DArray:
-            result.texture_type.Assign(Tegra::Shader::TextureType::Texture1D);
+            result.texture_type.Assign(TextureType::Texture1D);
             result.is_array.Assign(1);
-            result.is_buffer.Assign(0);
-            result.is_shadow.Assign(0);
             return result;
         case Tegra::Texture::TextureType::Texture2DArray:
-            result.texture_type.Assign(Tegra::Shader::TextureType::Texture2D);
+            result.texture_type.Assign(TextureType::Texture2D);
             result.is_array.Assign(1);
-            result.is_buffer.Assign(0);
-            result.is_shadow.Assign(0);
             return result;
         case Tegra::Texture::TextureType::Texture1DBuffer:
-            result.texture_type.Assign(Tegra::Shader::TextureType::Texture1D);
-            result.is_array.Assign(0);
+            result.texture_type.Assign(TextureType::Texture1D);
             result.is_buffer.Assign(1);
-            result.is_shadow.Assign(0);
             return result;
         case Tegra::Texture::TextureType::Texture2DNoMipmap:
-            result.texture_type.Assign(Tegra::Shader::TextureType::Texture2D);
-            result.is_array.Assign(0);
-            result.is_buffer.Assign(0);
-            result.is_shadow.Assign(0);
+            result.texture_type.Assign(TextureType::Texture2D);
             return result;
         case Tegra::Texture::TextureType::TextureCubeArray:
-            result.texture_type.Assign(Tegra::Shader::TextureType::TextureCube);
+            result.texture_type.Assign(TextureType::TextureCube);
             result.is_array.Assign(1);
-            result.is_buffer.Assign(0);
-            result.is_shadow.Assign(0);
             return result;
         default:
-            result.texture_type.Assign(Tegra::Shader::TextureType::Texture2D);
-            result.is_array.Assign(0);
-            result.is_buffer.Assign(0);
-            result.is_shadow.Assign(0);
+            result.texture_type.Assign(TextureType::Texture2D);
             return result;
         }
     }

src/video_core/engines/kepler_compute.cpp

@@ -39,7 +39,7 @@ void KeplerCompute::CallMethod(const GPU::MethodCall& method_call) {
         const bool is_last_call = method_call.IsLastCall();
         upload_state.ProcessData(method_call.argument, is_last_call);
         if (is_last_call) {
-            system.GPU().Maxwell3D().dirty.OnMemoryWrite();
+            system.GPU().Maxwell3D().OnMemoryWrite();
         }
         break;
     }
@@ -89,7 +89,7 @@ SamplerDescriptor KeplerCompute::AccessBindlessSampler(ShaderType stage, u64 con
 
     const Texture::TextureHandle tex_handle{memory_manager.Read<u32>(tex_info_address)};
     const Texture::FullTextureInfo tex_info = GetTextureInfo(tex_handle);
-    SamplerDescriptor result = SamplerDescriptor::FromTicTexture(tex_info.tic.texture_type.Value());
+    SamplerDescriptor result = SamplerDescriptor::FromTIC(tex_info.tic);
     result.is_shadow.Assign(tex_info.tsc.depth_compare_enabled.Value());
     return result;
 }
@@ -119,14 +119,6 @@ Texture::TICEntry KeplerCompute::GetTICEntry(u32 tic_index) const {
     Texture::TICEntry tic_entry;
     memory_manager.ReadBlockUnsafe(tic_address_gpu, &tic_entry, sizeof(Texture::TICEntry));
 
-    const auto r_type{tic_entry.r_type.Value()};
-    const auto g_type{tic_entry.g_type.Value()};
-    const auto b_type{tic_entry.b_type.Value()};
-    const auto a_type{tic_entry.a_type.Value()};
-
-    // TODO(Subv): Different data types for separate components are not supported
-    DEBUG_ASSERT(r_type == g_type && r_type == b_type && r_type == a_type);
-
     return tic_entry;
 }
 

src/video_core/engines/kepler_memory.cpp

@@ -34,7 +34,7 @@ void KeplerMemory::CallMethod(const GPU::MethodCall& method_call) {
         const bool is_last_call = method_call.IsLastCall();
         upload_state.ProcessData(method_call.argument, is_last_call);
         if (is_last_call) {
-            system.GPU().Maxwell3D().dirty.OnMemoryWrite();
+            system.GPU().Maxwell3D().OnMemoryWrite();
         }
         break;
     }

src/video_core/engines/maxwell_3d.cpp

@@ -26,7 +26,8 @@ Maxwell3D::Maxwell3D(Core::System& system, VideoCore::RasterizerInterface& raste
                      MemoryManager& memory_manager)
     : system{system}, rasterizer{rasterizer}, memory_manager{memory_manager},
       macro_interpreter{*this}, upload_state{memory_manager, regs.upload} {
-    InitDirtySettings();
+    dirty.flags.flip();
+
     InitializeRegisterDefaults();
 }
 
@@ -75,8 +76,8 @@ void Maxwell3D::InitializeRegisterDefaults() {
     regs.stencil_back_mask = 0xFFFFFFFF;
 
     regs.depth_test_func = Regs::ComparisonOp::Always;
-    regs.cull.front_face = Regs::Cull::FrontFace::CounterClockWise;
-    regs.cull.cull_face = Regs::Cull::CullFace::Back;
+    regs.front_face = Regs::FrontFace::CounterClockWise;
+    regs.cull_face = Regs::CullFace::Back;
 
     // TODO(Rodrigo): Most games do not set a point size. I think this is a case of a
     // register carrying a default value. Assume it's OpenGL's default (1).
@@ -95,7 +96,9 @@ void Maxwell3D::InitializeRegisterDefaults() {
     regs.rasterize_enable = 1;
     regs.rt_separate_frag_data = 1;
     regs.framebuffer_srgb = 1;
-    regs.cull.front_face = Maxwell3D::Regs::Cull::FrontFace::ClockWise;
+    regs.front_face = Maxwell3D::Regs::FrontFace::ClockWise;
+
+    shadow_state = regs;
 
     mme_inline[MAXWELL3D_REG_INDEX(draw.vertex_end_gl)] = true;
     mme_inline[MAXWELL3D_REG_INDEX(draw.vertex_begin_gl)] = true;
@@ -103,164 +106,6 @@ void Maxwell3D::InitializeRegisterDefaults() {
     mme_inline[MAXWELL3D_REG_INDEX(index_array.count)] = true;
 }
 
-#define DIRTY_REGS_POS(field_name) static_cast<u8>(offsetof(Maxwell3D::DirtyRegs, field_name))
-
-void Maxwell3D::InitDirtySettings() {
-    const auto set_block = [this](std::size_t start, std::size_t range, u8 position) {
-        const auto start_itr = dirty_pointers.begin() + start;
-        const auto end_itr = start_itr + range;
-        std::fill(start_itr, end_itr, position);
-    };
-    dirty.regs.fill(true);
-
-    // Init Render Targets
-    constexpr u32 registers_per_rt = sizeof(regs.rt[0]) / sizeof(u32);
-    constexpr u32 rt_start_reg = MAXWELL3D_REG_INDEX(rt);
-    constexpr u32 rt_end_reg = rt_start_reg + registers_per_rt * 8;
-    u8 rt_dirty_reg = DIRTY_REGS_POS(render_target);
-    for (u32 rt_reg = rt_start_reg; rt_reg < rt_end_reg; rt_reg += registers_per_rt) {
-        set_block(rt_reg, registers_per_rt, rt_dirty_reg);
-        ++rt_dirty_reg;
-    }
-    constexpr u32 depth_buffer_flag = DIRTY_REGS_POS(depth_buffer);
-    dirty_pointers[MAXWELL3D_REG_INDEX(zeta_enable)] = depth_buffer_flag;
-    dirty_pointers[MAXWELL3D_REG_INDEX(zeta_width)] = depth_buffer_flag;
-    dirty_pointers[MAXWELL3D_REG_INDEX(zeta_height)] = depth_buffer_flag;
-    constexpr u32 registers_in_zeta = sizeof(regs.zeta) / sizeof(u32);
-    constexpr u32 zeta_reg = MAXWELL3D_REG_INDEX(zeta);
-    set_block(zeta_reg, registers_in_zeta, depth_buffer_flag);
-
-    // Init Vertex Arrays
-    constexpr u32 vertex_array_start = MAXWELL3D_REG_INDEX(vertex_array);
-    constexpr u32 vertex_array_size = sizeof(regs.vertex_array[0]) / sizeof(u32);
-    constexpr u32 vertex_array_end = vertex_array_start + vertex_array_size * Regs::NumVertexArrays;
-    u8 va_dirty_reg = DIRTY_REGS_POS(vertex_array);
-    u8 vi_dirty_reg = DIRTY_REGS_POS(vertex_instance);
-    for (u32 vertex_reg = vertex_array_start; vertex_reg < vertex_array_end;
-         vertex_reg += vertex_array_size) {
-        set_block(vertex_reg, 3, va_dirty_reg);
-        // The divisor concerns vertex array instances
-        dirty_pointers[static_cast<std::size_t>(vertex_reg) + 3] = vi_dirty_reg;
-        ++va_dirty_reg;
-        ++vi_dirty_reg;
-    }
-    constexpr u32 vertex_limit_start = MAXWELL3D_REG_INDEX(vertex_array_limit);
-    constexpr u32 vertex_limit_size = sizeof(regs.vertex_array_limit[0]) / sizeof(u32);
-    constexpr u32 vertex_limit_end = vertex_limit_start + vertex_limit_size * Regs::NumVertexArrays;
-    va_dirty_reg = DIRTY_REGS_POS(vertex_array);
-    for (u32 vertex_reg = vertex_limit_start; vertex_reg < vertex_limit_end;
-         vertex_reg += vertex_limit_size) {
-        set_block(vertex_reg, vertex_limit_size, va_dirty_reg);
-        va_dirty_reg++;
-    }
-    constexpr u32 vertex_instance_start = MAXWELL3D_REG_INDEX(instanced_arrays);
-    constexpr u32 vertex_instance_size =
-        sizeof(regs.instanced_arrays.is_instanced[0]) / sizeof(u32);
-    constexpr u32 vertex_instance_end =
-        vertex_instance_start + vertex_instance_size * Regs::NumVertexArrays;
-    vi_dirty_reg = DIRTY_REGS_POS(vertex_instance);
-    for (u32 vertex_reg = vertex_instance_start; vertex_reg < vertex_instance_end;
-         vertex_reg += vertex_instance_size) {
-        set_block(vertex_reg, vertex_instance_size, vi_dirty_reg);
-        vi_dirty_reg++;
-    }
-    set_block(MAXWELL3D_REG_INDEX(vertex_attrib_format), regs.vertex_attrib_format.size(),
-              DIRTY_REGS_POS(vertex_attrib_format));
-
-    // Init Shaders
-    constexpr u32 shader_registers_count =
-        sizeof(regs.shader_config[0]) * Regs::MaxShaderProgram / sizeof(u32);
-    set_block(MAXWELL3D_REG_INDEX(shader_config[0]), shader_registers_count,
-              DIRTY_REGS_POS(shaders));
-
-    // State
-
-    // Viewport
-    constexpr u8 viewport_dirty_reg = DIRTY_REGS_POS(viewport);
-    constexpr u32 viewport_start = MAXWELL3D_REG_INDEX(viewports);
-    constexpr u32 viewport_size = sizeof(regs.viewports) / sizeof(u32);
-    set_block(viewport_start, viewport_size, viewport_dirty_reg);
-    constexpr u32 view_volume_start = MAXWELL3D_REG_INDEX(view_volume_clip_control);
-    constexpr u32 view_volume_size = sizeof(regs.view_volume_clip_control) / sizeof(u32);
-    set_block(view_volume_start, view_volume_size, viewport_dirty_reg);
-
-    // Viewport transformation
-    constexpr u32 viewport_trans_start = MAXWELL3D_REG_INDEX(viewport_transform);
-    constexpr u32 viewport_trans_size = sizeof(regs.viewport_transform) / sizeof(u32);
-    set_block(viewport_trans_start, viewport_trans_size, DIRTY_REGS_POS(viewport_transform));
-
-    // Cullmode
-    constexpr u32 cull_mode_start = MAXWELL3D_REG_INDEX(cull);
-    constexpr u32 cull_mode_size = sizeof(regs.cull) / sizeof(u32);
-    set_block(cull_mode_start, cull_mode_size, DIRTY_REGS_POS(cull_mode));
-
-    // Screen y control
-    dirty_pointers[MAXWELL3D_REG_INDEX(screen_y_control)] = DIRTY_REGS_POS(screen_y_control);
-
-    // Primitive Restart
-    constexpr u32 primitive_restart_start = MAXWELL3D_REG_INDEX(primitive_restart);
-    constexpr u32 primitive_restart_size = sizeof(regs.primitive_restart) / sizeof(u32);
-    set_block(primitive_restart_start, primitive_restart_size, DIRTY_REGS_POS(primitive_restart));
-
-    // Depth Test
-    constexpr u8 depth_test_dirty_reg = DIRTY_REGS_POS(depth_test);
-    dirty_pointers[MAXWELL3D_REG_INDEX(depth_test_enable)] = depth_test_dirty_reg;
-    dirty_pointers[MAXWELL3D_REG_INDEX(depth_write_enabled)] = depth_test_dirty_reg;
-    dirty_pointers[MAXWELL3D_REG_INDEX(depth_test_func)] = depth_test_dirty_reg;
-
-    // Stencil Test
-    constexpr u32 stencil_test_dirty_reg = DIRTY_REGS_POS(stencil_test);
-    dirty_pointers[MAXWELL3D_REG_INDEX(stencil_enable)] = stencil_test_dirty_reg;
-    dirty_pointers[MAXWELL3D_REG_INDEX(stencil_front_func_func)] = stencil_test_dirty_reg;
-    dirty_pointers[MAXWELL3D_REG_INDEX(stencil_front_func_ref)] = stencil_test_dirty_reg;
-    dirty_pointers[MAXWELL3D_REG_INDEX(stencil_front_func_mask)] = stencil_test_dirty_reg;
-    dirty_pointers[MAXWELL3D_REG_INDEX(stencil_front_op_fail)] = stencil_test_dirty_reg;
-    dirty_pointers[MAXWELL3D_REG_INDEX(stencil_front_op_zfail)] = stencil_test_dirty_reg;
-    dirty_pointers[MAXWELL3D_REG_INDEX(stencil_front_op_zpass)] = stencil_test_dirty_reg;
-    dirty_pointers[MAXWELL3D_REG_INDEX(stencil_front_mask)] = stencil_test_dirty_reg;
-    dirty_pointers[MAXWELL3D_REG_INDEX(stencil_two_side_enable)] = stencil_test_dirty_reg;
-    dirty_pointers[MAXWELL3D_REG_INDEX(stencil_back_func_func)] = stencil_test_dirty_reg;
-    dirty_pointers[MAXWELL3D_REG_INDEX(stencil_back_func_ref)] = stencil_test_dirty_reg;
-    dirty_pointers[MAXWELL3D_REG_INDEX(stencil_back_func_mask)] = stencil_test_dirty_reg;
-    dirty_pointers[MAXWELL3D_REG_INDEX(stencil_back_op_fail)] = stencil_test_dirty_reg;
-    dirty_pointers[MAXWELL3D_REG_INDEX(stencil_back_op_zfail)] = stencil_test_dirty_reg;
-    dirty_pointers[MAXWELL3D_REG_INDEX(stencil_back_op_zpass)] = stencil_test_dirty_reg;
-    dirty_pointers[MAXWELL3D_REG_INDEX(stencil_back_mask)] = stencil_test_dirty_reg;
-
-    // Color Mask
-    constexpr u8 color_mask_dirty_reg = DIRTY_REGS_POS(color_mask);
-    dirty_pointers[MAXWELL3D_REG_INDEX(color_mask_common)] = color_mask_dirty_reg;
-    set_block(MAXWELL3D_REG_INDEX(color_mask), sizeof(regs.color_mask) / sizeof(u32),
-              color_mask_dirty_reg);
-    // Blend State
-    constexpr u8 blend_state_dirty_reg = DIRTY_REGS_POS(blend_state);
-    set_block(MAXWELL3D_REG_INDEX(blend_color), sizeof(regs.blend_color) / sizeof(u32),
-              blend_state_dirty_reg);
-    dirty_pointers[MAXWELL3D_REG_INDEX(independent_blend_enable)] = blend_state_dirty_reg;
-    set_block(MAXWELL3D_REG_INDEX(blend), sizeof(regs.blend) / sizeof(u32), blend_state_dirty_reg);
-    set_block(MAXWELL3D_REG_INDEX(independent_blend), sizeof(regs.independent_blend) / sizeof(u32),
-              blend_state_dirty_reg);
-
-    // Scissor State
-    constexpr u8 scissor_test_dirty_reg = DIRTY_REGS_POS(scissor_test);
-    set_block(MAXWELL3D_REG_INDEX(scissor_test), sizeof(regs.scissor_test) / sizeof(u32),
-              scissor_test_dirty_reg);
-
-    // Polygon Offset
-    constexpr u8 polygon_offset_dirty_reg = DIRTY_REGS_POS(polygon_offset);
-    dirty_pointers[MAXWELL3D_REG_INDEX(polygon_offset_fill_enable)] = polygon_offset_dirty_reg;
-    dirty_pointers[MAXWELL3D_REG_INDEX(polygon_offset_line_enable)] = polygon_offset_dirty_reg;
-    dirty_pointers[MAXWELL3D_REG_INDEX(polygon_offset_point_enable)] = polygon_offset_dirty_reg;
-    dirty_pointers[MAXWELL3D_REG_INDEX(polygon_offset_units)] = polygon_offset_dirty_reg;
-    dirty_pointers[MAXWELL3D_REG_INDEX(polygon_offset_factor)] = polygon_offset_dirty_reg;
-    dirty_pointers[MAXWELL3D_REG_INDEX(polygon_offset_clamp)] = polygon_offset_dirty_reg;
-
-    // Depth bounds
-    constexpr u8 depth_bounds_values_dirty_reg = DIRTY_REGS_POS(depth_bounds_values);
-    dirty_pointers[MAXWELL3D_REG_INDEX(depth_bounds[0])] = depth_bounds_values_dirty_reg;
-    dirty_pointers[MAXWELL3D_REG_INDEX(depth_bounds[1])] = depth_bounds_values_dirty_reg;
-}
-
 void Maxwell3D::CallMacroMethod(u32 method, std::size_t num_parameters, const u32* parameters) {
     // Reset the current macro.
     executing_macro = 0;
@@ -317,31 +162,34 @@ void Maxwell3D::CallMethod(const GPU::MethodCall& method_call) {
     ASSERT_MSG(method < Regs::NUM_REGS,
                "Invalid Maxwell3D register, increase the size of the Regs structure");
 
-    if (regs.reg_array[method] != method_call.argument) {
-        regs.reg_array[method] = method_call.argument;
-        const std::size_t dirty_reg = dirty_pointers[method];
-        if (dirty_reg) {
-            dirty.regs[dirty_reg] = true;
-            if (dirty_reg >= DIRTY_REGS_POS(vertex_array) &&
-                dirty_reg < DIRTY_REGS_POS(vertex_array_buffers)) {
-                dirty.vertex_array_buffers = true;
-            } else if (dirty_reg >= DIRTY_REGS_POS(vertex_instance) &&
-                       dirty_reg < DIRTY_REGS_POS(vertex_instances)) {
-                dirty.vertex_instances = true;
-            } else if (dirty_reg >= DIRTY_REGS_POS(render_target) &&
-                       dirty_reg < DIRTY_REGS_POS(render_settings)) {
-                dirty.render_settings = true;
-            }
+    u32 arg = method_call.argument;
+    // Keep track of the register value in shadow_state when requested.
+    if (shadow_state.shadow_ram_control == Regs::ShadowRamControl::Track ||
+        shadow_state.shadow_ram_control == Regs::ShadowRamControl::TrackWithFilter) {
+        shadow_state.reg_array[method] = arg;
+    } else if (shadow_state.shadow_ram_control == Regs::ShadowRamControl::Replay) {
+        arg = shadow_state.reg_array[method];
+    }
+
+    if (regs.reg_array[method] != arg) {
+        regs.reg_array[method] = arg;
+
+        for (const auto& table : dirty.tables) {
+            dirty.flags[table[method]] = true;
         }
     }
 
     switch (method) {
+    case MAXWELL3D_REG_INDEX(shadow_ram_control): {
+        shadow_state.shadow_ram_control = static_cast<Regs::ShadowRamControl>(method_call.argument);
+        break;
+    }
     case MAXWELL3D_REG_INDEX(macros.data): {
-        ProcessMacroUpload(method_call.argument);
+        ProcessMacroUpload(arg);
         break;
     }
     case MAXWELL3D_REG_INDEX(macros.bind): {
-        ProcessMacroBind(method_call.argument);
+        ProcessMacroBind(arg);
         break;
     }
     case MAXWELL3D_REG_INDEX(firmware[4]): {
@@ -417,9 +265,9 @@ void Maxwell3D::CallMethod(const GPU::MethodCall& method_call) {
     }
     case MAXWELL3D_REG_INDEX(data_upload): {
         const bool is_last_call = method_call.IsLastCall();
-        upload_state.ProcessData(method_call.argument, is_last_call);
+        upload_state.ProcessData(arg, is_last_call);
         if (is_last_call) {
-            dirty.OnMemoryWrite();
+            OnMemoryWrite();
         }
         break;
     }
@@ -727,7 +575,7 @@ void Maxwell3D::FinishCBData() {
 
     const u32 id = cb_data_state.id;
     memory_manager.WriteBlock(address, cb_data_state.buffer[id].data(), size);
-    dirty.OnMemoryWrite();
+    OnMemoryWrite();
 
     cb_data_state.id = null_cb_data;
     cb_data_state.current = null_cb_data;
@@ -805,7 +653,7 @@ SamplerDescriptor Maxwell3D::AccessBindlessSampler(ShaderType stage, u64 const_b
 
     const Texture::TextureHandle tex_handle{memory_manager.Read<u32>(tex_info_address)};
     const Texture::FullTextureInfo tex_info = GetTextureInfo(tex_handle);
-    SamplerDescriptor result = SamplerDescriptor::FromTicTexture(tex_info.tic.texture_type.Value());
+    SamplerDescriptor result = SamplerDescriptor::FromTIC(tex_info.tic);
     result.is_shadow.Assign(tex_info.tsc.depth_compare_enabled.Value());
     return result;
 }

src/video_core/engines/maxwell_3d.h

@@ -6,6 +6,7 @@
 
 #include <array>
 #include <bitset>
+#include <limits>
 #include <optional>
 #include <type_traits>
 #include <unordered_map>
@@ -66,6 +67,7 @@ public:
         static constexpr std::size_t NumVaryings = 31;
         static constexpr std::size_t NumImages = 8; // TODO(Rodrigo): Investigate this number
         static constexpr std::size_t NumClipDistances = 8;
+        static constexpr std::size_t NumTransformFeedbackBuffers = 4;
         static constexpr std::size_t MaxShaderProgram = 6;
         static constexpr std::size_t MaxShaderStage = 5;
         // Maximum number of const buffers per shader stage.
@@ -310,6 +312,35 @@ public:
             }
         };
 
+        struct MsaaSampleLocation {
+            union {
+                BitField<0, 4, u32> x0;
+                BitField<4, 4, u32> y0;
+                BitField<8, 4, u32> x1;
+                BitField<12, 4, u32> y1;
+                BitField<16, 4, u32> x2;
+                BitField<20, 4, u32> y2;
+                BitField<24, 4, u32> x3;
+                BitField<28, 4, u32> y3;
+            };
+
+            constexpr std::pair<u32, u32> Location(int index) const {
+                switch (index) {
+                case 0:
+                    return {x0, y0};
+                case 1:
+                    return {x1, y1};
+                case 2:
+                    return {x2, y2};
+                case 3:
+                    return {x3, y3};
+                default:
+                    UNREACHABLE();
+                    return {0, 0};
+                }
+            }
+        };
+
         enum class DepthMode : u32 {
             MinusOneToOne = 0,
             ZeroToOne = 1,
@@ -431,21 +462,15 @@ public:
             GeneratedPrimitives = 0x1F,
         };
 
-        struct Cull {
-            enum class FrontFace : u32 {
-                ClockWise = 0x0900,
-                CounterClockWise = 0x0901,
-            };
+        enum class FrontFace : u32 {
+            ClockWise = 0x0900,
+            CounterClockWise = 0x0901,
+        };
 
-            enum class CullFace : u32 {
-                Front = 0x0404,
-                Back = 0x0405,
-                FrontAndBack = 0x0408,
-            };
-
-            u32 enabled;
-            FrontFace front_face;
-            CullFace cull_face;
+        enum class CullFace : u32 {
+            Front = 0x0404,
+            Back = 0x0405,
+            FrontAndBack = 0x0408,
         };
 
         struct Blend {
@@ -529,6 +554,23 @@ public:
             FractionalEven = 2,
         };
 
+        enum class PolygonMode : u32 {
+            Point = 0x1b00,
+            Line = 0x1b01,
+            Fill = 0x1b02,
+        };
+
+        enum class ShadowRamControl : u32 {
+            // write value to shadow ram
+            Track = 0,
+            // write value to shadow ram ( with validation ??? )
+            TrackWithFilter = 1,
+            // only write to real hw register
+            Passthrough = 2,
+            // write value from shadow ram to real hw register
+            Replay = 3,
+        };
+
         struct RenderTargetConfig {
             u32 address_high;
             u32 address_low;
@@ -574,7 +616,7 @@ public:
             f32 translate_z;
             INSERT_UNION_PADDING_WORDS(2);
 
-            Common::Rectangle<s32> GetRect() const {
+            Common::Rectangle<f32> GetRect() const {
                 return {
                     GetX(),               // left
                     GetY() + GetHeight(), // top
@@ -583,20 +625,20 @@ public:
                 };
             };
 
-            s32 GetX() const {
-                return static_cast<s32>(std::max(0.0f, translate_x - std::fabs(scale_x)));
+            f32 GetX() const {
+                return std::max(0.0f, translate_x - std::fabs(scale_x));
             }
 
-            s32 GetY() const {
-                return static_cast<s32>(std::max(0.0f, translate_y - std::fabs(scale_y)));
+            f32 GetY() const {
+                return std::max(0.0f, translate_y - std::fabs(scale_y));
             }
 
-            s32 GetWidth() const {
-                return static_cast<s32>(translate_x + std::fabs(scale_x)) - GetX();
+            f32 GetWidth() const {
+                return translate_x + std::fabs(scale_x) - GetX();
             }
 
-            s32 GetHeight() const {
-                return static_cast<s32>(translate_y + std::fabs(scale_y)) - GetY();
+            f32 GetHeight() const {
+                return translate_y + std::fabs(scale_y) - GetY();
             }
         };
 
@@ -626,6 +668,29 @@ public:
             float depth_range_far;
         };
 
+        struct TransformFeedbackBinding {
+            u32 buffer_enable;
+            u32 address_high;
+            u32 address_low;
+            s32 buffer_size;
+            s32 buffer_offset;
+            INSERT_UNION_PADDING_WORDS(3);
+
+            GPUVAddr Address() const {
+                return static_cast<GPUVAddr>((static_cast<GPUVAddr>(address_high) << 32) |
+                                             address_low);
+            }
+        };
+        static_assert(sizeof(TransformFeedbackBinding) == 32);
+
+        struct TransformFeedbackLayout {
+            u32 stream;
+            u32 varying_count;
+            u32 stride;
+            INSERT_UNION_PADDING_WORDS(1);
+        };
+        static_assert(sizeof(TransformFeedbackLayout) == 16);
+
         bool IsShaderConfigEnabled(std::size_t index) const {
             // The VertexB is always enabled.
             if (index == static_cast<std::size_t>(Regs::ShaderProgram::VertexB)) {
@@ -634,6 +699,10 @@ public:
             return shader_config[index].enable != 0;
         }
 
+        bool IsShaderConfigEnabled(Regs::ShaderProgram type) const {
+            return IsShaderConfigEnabled(static_cast<std::size_t>(type));
+        }
+
         union {
             struct {
                 INSERT_UNION_PADDING_WORDS(0x45);
@@ -645,7 +714,9 @@ public:
                     u32 bind;
                 } macros;
 
-                INSERT_UNION_PADDING_WORDS(0x17);
+                ShadowRamControl shadow_ram_control;
+
+                INSERT_UNION_PADDING_WORDS(0x16);
 
                 Upload::Registers upload;
                 struct {
@@ -682,7 +753,13 @@ public:
 
                 u32 rasterize_enable;
 
-                INSERT_UNION_PADDING_WORDS(0xF1);
+                std::array<TransformFeedbackBinding, NumTransformFeedbackBuffers> tfb_bindings;
+
+                INSERT_UNION_PADDING_WORDS(0xC0);
+
+                std::array<TransformFeedbackLayout, NumTransformFeedbackBuffers> tfb_layouts;
+
+                INSERT_UNION_PADDING_WORDS(0x1);
 
                 u32 tfb_enabled;
 
@@ -710,7 +787,12 @@ public:
 
                 s32 clear_stencil;
 
-                INSERT_UNION_PADDING_WORDS(0x7);
+                INSERT_UNION_PADDING_WORDS(0x2);
+
+                PolygonMode polygon_mode_front;
+                PolygonMode polygon_mode_back;
+
+                INSERT_UNION_PADDING_WORDS(0x3);
 
                 u32 polygon_offset_point_enable;
                 u32 polygon_offset_line_enable;
@@ -740,7 +822,13 @@ public:
 
                 u32 rt_separate_frag_data;
 
-                INSERT_UNION_PADDING_WORDS(0xC);
+                INSERT_UNION_PADDING_WORDS(0x1);
+
+                u32 multisample_raster_enable;
+                u32 multisample_raster_samples;
+                std::array<u32, 4> multisample_sample_mask;
+
+                INSERT_UNION_PADDING_WORDS(0x5);
 
                 struct {
                     u32 address_high;
@@ -769,11 +857,24 @@ public:
                     BitField<12, 4, u32> viewport;
                 } clear_flags;
 
-                INSERT_UNION_PADDING_WORDS(0x19);
+                INSERT_UNION_PADDING_WORDS(0x10);
+
+                u32 fill_rectangle;
+
+                INSERT_UNION_PADDING_WORDS(0x8);
 
                 std::array<VertexAttribute, NumVertexAttributes> vertex_attrib_format;
 
-                INSERT_UNION_PADDING_WORDS(0xF);
+                std::array<MsaaSampleLocation, 4> multisample_sample_locations;
+
+                INSERT_UNION_PADDING_WORDS(0x2);
+
+                union {
+                    BitField<0, 1, u32> enable;
+                    BitField<4, 3, u32> target;
+                } multisample_coverage_to_color;
+
+                INSERT_UNION_PADDING_WORDS(0x8);
 
                 struct {
                     union {
@@ -865,23 +966,17 @@ public:
                     BitField<4, 1, u32> triangle_rast_flip;
                 } screen_y_control;
 
-                INSERT_UNION_PADDING_WORDS(0x21);
+                float line_width_smooth;
+                float line_width_aliased;
+
+                INSERT_UNION_PADDING_WORDS(0x1F);
 
                 u32 vb_element_base;
                 u32 vb_base_instance;
 
                 INSERT_UNION_PADDING_WORDS(0x35);
 
-                union {
-                    BitField<0, 1, u32> c0;
-                    BitField<1, 1, u32> c1;
-                    BitField<2, 1, u32> c2;
-                    BitField<3, 1, u32> c3;
-                    BitField<4, 1, u32> c4;
-                    BitField<5, 1, u32> c5;
-                    BitField<6, 1, u32> c6;
-                    BitField<7, 1, u32> c7;
-                } clip_distance_enabled;
+                u32 clip_distance_enabled;
 
                 u32 samplecnt_enable;
 
@@ -895,7 +990,7 @@ public:
 
                 CounterReset counter_reset;
 
-                INSERT_UNION_PADDING_WORDS(0x1);
+                u32 multisample_enable;
 
                 u32 zeta_enable;
 
@@ -932,7 +1027,7 @@ public:
 
                 float polygon_offset_factor;
 
-                INSERT_UNION_PADDING_WORDS(0x1);
+                u32 line_smooth_enable;
 
                 struct {
                     u32 tic_address_high;
@@ -959,7 +1054,11 @@ public:
 
                 float polygon_offset_units;
 
-                INSERT_UNION_PADDING_WORDS(0x11);
+                INSERT_UNION_PADDING_WORDS(0x4);
+
+                Tegra::Texture::MsaaMode multisample_mode;
+
+                INSERT_UNION_PADDING_WORDS(0xC);
 
                 union {
                     BitField<2, 1, u32> coord_origin;
@@ -1060,7 +1159,9 @@ public:
 
                 INSERT_UNION_PADDING_WORDS(1);
 
-                Cull cull;
+                u32 cull_test_enabled;
+                FrontFace front_face;
+                CullFace cull_face;
 
                 u32 pixel_center_integer;
 
@@ -1199,7 +1300,11 @@ public:
 
                 u32 tex_cb_index;
 
-                INSERT_UNION_PADDING_WORDS(0x395);
+                INSERT_UNION_PADDING_WORDS(0x7D);
+
+                std::array<std::array<u8, 128>, NumTransformFeedbackBuffers> tfb_varying_locs;
+
+                INSERT_UNION_PADDING_WORDS(0x298);
 
                 struct {
                     /// Compressed address of a buffer that holds information about bound SSBOs.
@@ -1222,7 +1327,12 @@ public:
             };
             std::array<u32, NUM_REGS> reg_array;
         };
-    } regs{};
+    };
+
+    Regs regs{};
+
+    /// Store temporary hw register values, used by some calls to restore state after a operation
+    Regs shadow_state;
 
     static_assert(sizeof(Regs) == Regs::NUM_REGS * sizeof(u32), "Maxwell3D Regs has wrong size");
     static_assert(std::is_trivially_copyable_v<Regs>, "Maxwell3D Regs must be trivially copyable");
@@ -1238,79 +1348,6 @@ public:
 
     State state{};
 
-    struct DirtyRegs {
-        static constexpr std::size_t NUM_REGS = 256;
-        static_assert(NUM_REGS - 1 <= std::numeric_limits<u8>::max());
-
-        union {
-            struct {
-                bool null_dirty;
-
-                // Vertex Attributes
-                bool vertex_attrib_format;
-
-                // Vertex Arrays
-                std::array<bool, 32> vertex_array;
-
-                bool vertex_array_buffers;
-
-                // Vertex Instances
-                std::array<bool, 32> vertex_instance;
-
-                bool vertex_instances;
-
-                // Render Targets
-                std::array<bool, 8> render_target;
-                bool depth_buffer;
-
-                bool render_settings;
-
-                // Shaders
-                bool shaders;
-
-                // Rasterizer State
-                bool viewport;
-                bool clip_coefficient;
-                bool cull_mode;
-                bool primitive_restart;
-                bool depth_test;
-                bool stencil_test;
-                bool blend_state;
-                bool scissor_test;
-                bool transform_feedback;
-                bool color_mask;
-                bool polygon_offset;
-                bool depth_bounds_values;
-
-                // Complementary
-                bool viewport_transform;
-                bool screen_y_control;
-
-                bool memory_general;
-            };
-            std::array<bool, NUM_REGS> regs;
-        };
-
-        void ResetVertexArrays() {
-            vertex_array.fill(true);
-            vertex_array_buffers = true;
-        }
-
-        void ResetRenderTargets() {
-            depth_buffer = true;
-            render_target.fill(true);
-            render_settings = true;
-        }
-
-        void OnMemoryWrite() {
-            shaders = true;
-            memory_general = true;
-            ResetRenderTargets();
-            ResetVertexArrays();
-        }
-
-    } dirty{};
-
     /// Reads a register value located at the input method address
     u32 GetRegisterValue(u32 method) const;
 
@@ -1356,6 +1393,11 @@ public:
         return execute_on;
     }
 
+    /// Notify a memory write has happened.
+    void OnMemoryWrite() {
+        dirty.flags |= dirty.on_write_stores;
+    }
+
     enum class MMEDrawMode : u32 {
         Undefined,
         Array,
@@ -1371,6 +1413,16 @@ public:
         u32 gl_end_count{};
     } mme_draw;
 
+    struct DirtyState {
+        using Flags = std::bitset<std::numeric_limits<u8>::max()>;
+        using Table = std::array<u8, Regs::NUM_REGS>;
+        using Tables = std::array<Table, 2>;
+
+        Flags flags;
+        Flags on_write_stores;
+        Tables tables{};
+    } dirty;
+
 private:
     void InitializeRegisterDefaults();
 
@@ -1417,8 +1469,6 @@ private:
     /// Retrieves information about a specific TSC entry from the TSC buffer.
     Texture::TSCEntry GetTSCEntry(u32 tsc_index) const;
 
-    void InitDirtySettings();
-
     /**
      * Call a macro on this engine.
      * @param method Method to call
@@ -1477,6 +1527,7 @@ private:
                   "Field " #field_name " has invalid position")
 
 ASSERT_REG_POSITION(macros, 0x45);
+ASSERT_REG_POSITION(shadow_ram_control, 0x49);
 ASSERT_REG_POSITION(upload, 0x60);
 ASSERT_REG_POSITION(exec_upload, 0x6C);
 ASSERT_REG_POSITION(data_upload, 0x6D);
@@ -1485,6 +1536,8 @@ ASSERT_REG_POSITION(tess_mode, 0xC8);
 ASSERT_REG_POSITION(tess_level_outer, 0xC9);
 ASSERT_REG_POSITION(tess_level_inner, 0xCD);
 ASSERT_REG_POSITION(rasterize_enable, 0xDF);
+ASSERT_REG_POSITION(tfb_bindings, 0xE0);
+ASSERT_REG_POSITION(tfb_layouts, 0x1C0);
 ASSERT_REG_POSITION(tfb_enabled, 0x1D1);
 ASSERT_REG_POSITION(rt, 0x200);
 ASSERT_REG_POSITION(viewport_transform, 0x280);
@@ -1494,6 +1547,8 @@ ASSERT_REG_POSITION(depth_mode, 0x35F);
 ASSERT_REG_POSITION(clear_color[0], 0x360);
 ASSERT_REG_POSITION(clear_depth, 0x364);
 ASSERT_REG_POSITION(clear_stencil, 0x368);
+ASSERT_REG_POSITION(polygon_mode_front, 0x36B);
+ASSERT_REG_POSITION(polygon_mode_back, 0x36C);
 ASSERT_REG_POSITION(polygon_offset_point_enable, 0x370);
 ASSERT_REG_POSITION(polygon_offset_line_enable, 0x371);
 ASSERT_REG_POSITION(polygon_offset_fill_enable, 0x372);
@@ -1503,11 +1558,17 @@ 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(color_mask_common, 0x3E4);
-ASSERT_REG_POSITION(rt_separate_frag_data, 0x3EB);
 ASSERT_REG_POSITION(depth_bounds, 0x3E7);
+ASSERT_REG_POSITION(rt_separate_frag_data, 0x3EB);
+ASSERT_REG_POSITION(multisample_raster_enable, 0x3ED);
+ASSERT_REG_POSITION(multisample_raster_samples, 0x3EE);
+ASSERT_REG_POSITION(multisample_sample_mask, 0x3EF);
 ASSERT_REG_POSITION(zeta, 0x3F8);
 ASSERT_REG_POSITION(clear_flags, 0x43E);
+ASSERT_REG_POSITION(fill_rectangle, 0x44F);
 ASSERT_REG_POSITION(vertex_attrib_format, 0x458);
+ASSERT_REG_POSITION(multisample_sample_locations, 0x478);
+ASSERT_REG_POSITION(multisample_coverage_to_color, 0x47E);
 ASSERT_REG_POSITION(rt_control, 0x487);
 ASSERT_REG_POSITION(zeta_width, 0x48a);
 ASSERT_REG_POSITION(zeta_height, 0x48b);
@@ -1533,6 +1594,8 @@ ASSERT_REG_POSITION(stencil_front_func_mask, 0x4E6);
 ASSERT_REG_POSITION(stencil_front_mask, 0x4E7);
 ASSERT_REG_POSITION(frag_color_clamp, 0x4EA);
 ASSERT_REG_POSITION(screen_y_control, 0x4EB);
+ASSERT_REG_POSITION(line_width_smooth, 0x4EC);
+ASSERT_REG_POSITION(line_width_aliased, 0x4ED);
 ASSERT_REG_POSITION(vb_element_base, 0x50D);
 ASSERT_REG_POSITION(vb_base_instance, 0x50E);
 ASSERT_REG_POSITION(clip_distance_enabled, 0x544);
@@ -1540,11 +1603,13 @@ ASSERT_REG_POSITION(samplecnt_enable, 0x545);
 ASSERT_REG_POSITION(point_size, 0x546);
 ASSERT_REG_POSITION(point_sprite_enable, 0x548);
 ASSERT_REG_POSITION(counter_reset, 0x54C);
+ASSERT_REG_POSITION(multisample_enable, 0x54D);
 ASSERT_REG_POSITION(zeta_enable, 0x54E);
 ASSERT_REG_POSITION(multisample_control, 0x54F);
 ASSERT_REG_POSITION(condition, 0x554);
 ASSERT_REG_POSITION(tsc, 0x557);
-ASSERT_REG_POSITION(polygon_offset_factor, 0x55b);
+ASSERT_REG_POSITION(polygon_offset_factor, 0x55B);
+ASSERT_REG_POSITION(line_smooth_enable, 0x55C);
 ASSERT_REG_POSITION(tic, 0x55D);
 ASSERT_REG_POSITION(stencil_two_side_enable, 0x565);
 ASSERT_REG_POSITION(stencil_back_op_fail, 0x566);
@@ -1553,6 +1618,7 @@ ASSERT_REG_POSITION(stencil_back_op_zpass, 0x568);
 ASSERT_REG_POSITION(stencil_back_func_func, 0x569);
 ASSERT_REG_POSITION(framebuffer_srgb, 0x56E);
 ASSERT_REG_POSITION(polygon_offset_units, 0x56F);
+ASSERT_REG_POSITION(multisample_mode, 0x574);
 ASSERT_REG_POSITION(point_coord_replace, 0x581);
 ASSERT_REG_POSITION(code_address, 0x582);
 ASSERT_REG_POSITION(draw, 0x585);
@@ -1561,7 +1627,9 @@ ASSERT_REG_POSITION(index_array, 0x5F2);
 ASSERT_REG_POSITION(polygon_offset_clamp, 0x61F);
 ASSERT_REG_POSITION(instanced_arrays, 0x620);
 ASSERT_REG_POSITION(vp_point_size, 0x644);
-ASSERT_REG_POSITION(cull, 0x646);
+ASSERT_REG_POSITION(cull_test_enabled, 0x646);
+ASSERT_REG_POSITION(front_face, 0x647);
+ASSERT_REG_POSITION(cull_face, 0x648);
 ASSERT_REG_POSITION(pixel_center_integer, 0x649);
 ASSERT_REG_POSITION(viewport_transform_enabled, 0x64B);
 ASSERT_REG_POSITION(view_volume_clip_control, 0x64F);
@@ -1578,6 +1646,7 @@ ASSERT_REG_POSITION(firmware, 0x8C0);
 ASSERT_REG_POSITION(const_buffer, 0x8E0);
 ASSERT_REG_POSITION(cb_bind[0], 0x904);
 ASSERT_REG_POSITION(tex_cb_index, 0x982);
+ASSERT_REG_POSITION(tfb_varying_locs, 0xA00);
 ASSERT_REG_POSITION(ssbo_info, 0xD18);
 ASSERT_REG_POSITION(tex_info_buffers.address[0], 0xD2A);
 ASSERT_REG_POSITION(tex_info_buffers.size[0], 0xD2F);

src/video_core/engines/maxwell_dma.cpp

@@ -57,7 +57,7 @@ void MaxwellDMA::HandleCopy() {
     }
 
     // All copies here update the main memory, so mark all rasterizer states as invalid.
-    system.GPU().Maxwell3D().dirty.OnMemoryWrite();
+    system.GPU().Maxwell3D().OnMemoryWrite();
 
     if (regs.exec.is_dst_linear && regs.exec.is_src_linear) {
         // When the enable_2d bit is disabled, the copy is performed as if we were copying a 1D

src/video_core/engines/shader_bytecode.h

@@ -82,6 +82,10 @@ union Attribute {
         Position = 7,
         Attribute_0 = 8,
         Attribute_31 = 39,
+        FrontColor = 40,
+        FrontSecondaryColor = 41,
+        BackColor = 42,
+        BackSecondaryColor = 43,
         ClipDistances0123 = 44,
         ClipDistances4567 = 45,
         PointCoord = 46,
@@ -89,6 +93,8 @@ union Attribute {
         // shader, and a tuple of (TessCoord.x, TessCoord.y, TessCoord.z, ~) when inside a Tess Eval
         // shader.
         TessCoordInstanceIDVertexID = 47,
+        TexCoord_0 = 48,
+        TexCoord_7 = 55,
         // This attribute contains a tuple of (Unk, Unk, Unk, gl_FrontFacing) when inside a fragment
         // shader. It is unknown what the other values contain.
         FrontFacing = 63,
@@ -225,18 +231,6 @@ enum class AtomicOp : u64 {
     Or = 6,
     Xor = 7,
     Exch = 8,
-};
-
-enum class GlobalAtomicOp : u64 {
-    Add = 0,
-    Min = 1,
-    Max = 2,
-    Inc = 3,
-    Dec = 4,
-    And = 5,
-    Or = 6,
-    Xor = 7,
-    Exch = 8,
     SafeAdd = 10,
 };
 
@@ -296,6 +290,23 @@ enum class VmadShr : u64 {
     Shr15 = 2,
 };
 
+enum class VmnmxType : u64 {
+    Bits8,
+    Bits16,
+    Bits32,
+};
+
+enum class VmnmxOperation : u64 {
+    Mrg_16H = 0,
+    Mrg_16L = 1,
+    Mrg_8B0 = 2,
+    Mrg_8B2 = 3,
+    Acc = 4,
+    Min = 5,
+    Max = 6,
+    Nop = 7,
+};
+
 enum class XmadMode : u64 {
     None = 0,
     CLo = 1,
@@ -911,14 +922,9 @@ union Instruction {
     } fadd32i;
 
     union {
-        BitField<20, 8, u64> shift_position;
-        BitField<28, 8, u64> shift_length;
-        BitField<48, 1, u64> negate_b;
-        BitField<49, 1, u64> negate_a;
-
-        u64 GetLeftShiftValue() const {
-            return 32 - (shift_position + shift_length);
-        }
+        BitField<40, 1, u64> brev;
+        BitField<47, 1, u64> rd_cc;
+        BitField<48, 1, u64> is_signed;
     } bfe;
 
     union {
@@ -1000,7 +1006,7 @@ union Instruction {
     } stg;
 
     union {
-        BitField<52, 4, GlobalAtomicOp> operation;
+        BitField<52, 4, AtomicOp> operation;
         BitField<49, 3, GlobalAtomicType> type;
         BitField<28, 20, s64> offset;
     } atom;
@@ -1661,6 +1667,42 @@ union Instruction {
         BitField<47, 1, u64> cc;
     } vmad;
 
+    union {
+        BitField<54, 1, u64> is_dest_signed;
+        BitField<48, 1, u64> is_src_a_signed;
+        BitField<49, 1, u64> is_src_b_signed;
+        BitField<37, 2, u64> src_format_a;
+        BitField<29, 2, u64> src_format_b;
+        BitField<56, 1, u64> mx;
+        BitField<55, 1, u64> sat;
+        BitField<36, 2, u64> selector_a;
+        BitField<28, 2, u64> selector_b;
+        BitField<50, 1, u64> is_op_b_register;
+        BitField<51, 3, VmnmxOperation> operation;
+
+        VmnmxType SourceFormatA() const {
+            switch (src_format_a) {
+            case 0b11:
+                return VmnmxType::Bits32;
+            case 0b10:
+                return VmnmxType::Bits16;
+            default:
+                return VmnmxType::Bits8;
+            }
+        }
+
+        VmnmxType SourceFormatB() const {
+            switch (src_format_b) {
+            case 0b11:
+                return VmnmxType::Bits32;
+            case 0b10:
+                return VmnmxType::Bits16;
+            default:
+                return VmnmxType::Bits8;
+            }
+        }
+    } vmnmx;
+
     union {
         BitField<20, 16, u64> imm20_16;
         BitField<35, 1, u64> high_b_rr; // used on RR
@@ -1723,6 +1765,7 @@ public:
         BRK,
         DEPBAR,
         VOTE,
+        VOTE_VTG,
         SHFL,
         FSWZADD,
         BFE_C,
@@ -1769,9 +1812,11 @@ public:
         IPA,
         OUT_R, // Emit vertex/primitive
         ISBERD,
+        BAR,
         MEMBAR,
         VMAD,
         VSETP,
+        VMNMX,
         FFMA_IMM, // Fused Multiply and Add
         FFMA_CR,
         FFMA_RC,
@@ -1853,7 +1898,7 @@ public:
         MOV_C,
         MOV_R,
         MOV_IMM,
-        MOV_SYS,
+        S2R,
         MOV32_IMM,
         SHL_C,
         SHL_R,
@@ -2037,6 +2082,7 @@ private:
             INST("111000110000----", Id::EXIT, Type::Flow, "EXIT"),
             INST("1111000011110---", Id::DEPBAR, Type::Synch, "DEPBAR"),
             INST("0101000011011---", Id::VOTE, Type::Warp, "VOTE"),
+            INST("0101000011100---", Id::VOTE_VTG, Type::Warp, "VOTE_VTG"),
             INST("1110111100010---", Id::SHFL, Type::Warp, "SHFL"),
             INST("0101000011111---", Id::FSWZADD, Type::Warp, "FSWZADD"),
             INST("1110111111011---", Id::LD_A, Type::Memory, "LD_A"),
@@ -2074,9 +2120,11 @@ private:
             INST("11100000--------", Id::IPA, Type::Trivial, "IPA"),
             INST("1111101111100---", Id::OUT_R, Type::Trivial, "OUT_R"),
             INST("1110111111010---", Id::ISBERD, Type::Trivial, "ISBERD"),
+            INST("1111000010101---", Id::BAR, Type::Trivial, "BAR"),
             INST("1110111110011---", Id::MEMBAR, Type::Trivial, "MEMBAR"),
             INST("01011111--------", Id::VMAD, Type::Video, "VMAD"),
             INST("0101000011110---", Id::VSETP, Type::Video, "VSETP"),
+            INST("0011101---------", Id::VMNMX, Type::Video, "VMNMX"),
             INST("0011001-1-------", Id::FFMA_IMM, Type::Ffma, "FFMA_IMM"),
             INST("010010011-------", Id::FFMA_CR, Type::Ffma, "FFMA_CR"),
             INST("010100011-------", Id::FFMA_RC, Type::Ffma, "FFMA_RC"),
@@ -2145,7 +2193,7 @@ private:
             INST("0100110010011---", Id::MOV_C, Type::Arithmetic, "MOV_C"),
             INST("0101110010011---", Id::MOV_R, Type::Arithmetic, "MOV_R"),
             INST("0011100-10011---", Id::MOV_IMM, Type::Arithmetic, "MOV_IMM"),
-            INST("1111000011001---", Id::MOV_SYS, Type::Trivial, "MOV_SYS"),
+            INST("1111000011001---", Id::S2R, Type::Trivial, "S2R"),
             INST("000000010000----", Id::MOV32_IMM, Type::ArithmeticImmediate, "MOV32_IMM"),
             INST("0100110001100---", Id::FMNMX_C, Type::Arithmetic, "FMNMX_C"),
             INST("0101110001100---", Id::FMNMX_R, Type::Arithmetic, "FMNMX_R"),
@@ -2177,7 +2225,7 @@ private:
             INST("0011011-11111---", Id::SHF_LEFT_IMM, Type::Shift, "SHF_LEFT_IMM"),
             INST("0100110011100---", Id::I2I_C, Type::Conversion, "I2I_C"),
             INST("0101110011100---", Id::I2I_R, Type::Conversion, "I2I_R"),
-            INST("0011101-11100---", Id::I2I_IMM, Type::Conversion, "I2I_IMM"),
+            INST("0011100-11100---", Id::I2I_IMM, Type::Conversion, "I2I_IMM"),
             INST("0100110010111---", Id::I2F_C, Type::Conversion, "I2F_C"),
             INST("0101110010111---", Id::I2F_R, Type::Conversion, "I2F_R"),
             INST("0011100-10111---", Id::I2F_IMM, Type::Conversion, "I2F_IMM"),

src/video_core/engines/shader_header.h

@@ -4,6 +4,9 @@
 
 #pragma once
 
+#include <array>
+#include <optional>
+
 #include "common/bit_field.h"
 #include "common/common_funcs.h"
 #include "common/common_types.h"
@@ -16,7 +19,7 @@ enum class OutputTopology : u32 {
     TriangleStrip = 7,
 };
 
-enum class AttributeUse : u8 {
+enum class PixelImap : u8 {
     Unused = 0,
     Constant = 1,
     Perspective = 2,
@@ -24,7 +27,7 @@ enum class AttributeUse : u8 {
 };
 
 // Documentation in:
-// http://download.nvidia.com/open-gpu-doc/Shader-Program-Header/1/Shader-Program-Header.html#ImapTexture
+// http://download.nvidia.com/open-gpu-doc/Shader-Program-Header/1/Shader-Program-Header.html
 struct Header {
     union {
         BitField<0, 5, u32> sph_type;
@@ -59,8 +62,8 @@ struct Header {
     union {
         BitField<0, 12, u32> max_output_vertices;
         BitField<12, 8, u32> store_req_start; // NOTE: not used by geometry shaders.
-        BitField<24, 4, u32> reserved;
-        BitField<12, 8, u32> store_req_end; // NOTE: not used by geometry shaders.
+        BitField<20, 4, u32> reserved;
+        BitField<24, 8, u32> store_req_end; // NOTE: not used by geometry shaders.
     } common4{};
 
     union {
@@ -93,17 +96,20 @@ struct Header {
         struct {
             INSERT_UNION_PADDING_BYTES(3); // ImapSystemValuesA
             INSERT_UNION_PADDING_BYTES(1); // ImapSystemValuesB
+
             union {
-                BitField<0, 2, AttributeUse> x;
-                BitField<2, 2, AttributeUse> y;
-                BitField<4, 2, AttributeUse> w;
-                BitField<6, 2, AttributeUse> z;
+                BitField<0, 2, PixelImap> x;
+                BitField<2, 2, PixelImap> y;
+                BitField<4, 2, PixelImap> z;
+                BitField<6, 2, PixelImap> w;
                 u8 raw;
             } imap_generic_vector[32];
+
             INSERT_UNION_PADDING_BYTES(2);  // ImapColor
             INSERT_UNION_PADDING_BYTES(2);  // ImapSystemValuesC
             INSERT_UNION_PADDING_BYTES(10); // ImapFixedFncTexture[10]
             INSERT_UNION_PADDING_BYTES(2);  // ImapReserved
+
             struct {
                 u32 target;
                 union {
@@ -112,31 +118,30 @@ struct Header {
                     BitField<2, 30, u32> reserved;
                 };
             } omap;
+
             bool IsColorComponentOutputEnabled(u32 render_target, u32 component) const {
                 const u32 bit = render_target * 4 + component;
                 return omap.target & (1 << bit);
             }
-            AttributeUse GetAttributeIndexUse(u32 attribute, u32 index) const {
-                return static_cast<AttributeUse>(
-                    (imap_generic_vector[attribute].raw >> (index * 2)) & 0x03);
-            }
-            AttributeUse GetAttributeUse(u32 attribute) const {
-                AttributeUse result = AttributeUse::Unused;
-                for (u32 i = 0; i < 4; i++) {
-                    const auto index = GetAttributeIndexUse(attribute, i);
-                    if (index == AttributeUse::Unused) {
+
+            PixelImap GetPixelImap(u32 attribute) const {
+                const auto get_index = [this, attribute](u32 index) {
+                    return static_cast<PixelImap>(
+                        (imap_generic_vector[attribute].raw >> (index * 2)) & 3);
+                };
+
+                std::optional<PixelImap> result;
+                for (u32 component = 0; component < 4; ++component) {
+                    const PixelImap index = get_index(component);
+                    if (index == PixelImap::Unused) {
                         continue;
                     }
-                    if (result == AttributeUse::Unused || result == index) {
-                        result = index;
-                        continue;
-                    }
-                    LOG_CRITICAL(HW_GPU, "Generic Attribute Conflict in Interpolation Mode");
-                    if (index == AttributeUse::Perspective) {
-                        result = index;
+                    if (result && result != index) {
+                        LOG_CRITICAL(HW_GPU, "Generic attribute conflict in interpolation mode");
                     }
+                    result = index;
                 }
-                return result;
+                return result.value_or(PixelImap::Unused);
             }
         } ps;
 

src/video_core/gpu.cpp

@@ -7,6 +7,7 @@
 #include "core/core.h"
 #include "core/core_timing.h"
 #include "core/core_timing_util.h"
+#include "core/frontend/emu_window.h"
 #include "core/memory.h"
 #include "video_core/engines/fermi_2d.h"
 #include "video_core/engines/kepler_compute.h"
@@ -16,14 +17,15 @@
 #include "video_core/gpu.h"
 #include "video_core/memory_manager.h"
 #include "video_core/renderer_base.h"
+#include "video_core/video_core.h"
 
 namespace Tegra {
 
 MICROPROFILE_DEFINE(GPU_wait, "GPU", "Wait for the GPU", MP_RGB(128, 128, 192));
 
-GPU::GPU(Core::System& system, VideoCore::RendererBase& renderer, bool is_async)
-    : system{system}, renderer{renderer}, is_async{is_async} {
-    auto& rasterizer{renderer.Rasterizer()};
+GPU::GPU(Core::System& system, std::unique_ptr<VideoCore::RendererBase>&& renderer_, bool is_async)
+    : system{system}, renderer{std::move(renderer_)}, is_async{is_async} {
+    auto& rasterizer{renderer->Rasterizer()};
     memory_manager = std::make_unique<Tegra::MemoryManager>(system, rasterizer);
     dma_pusher = std::make_unique<Tegra::DmaPusher>(*this);
     maxwell_3d = std::make_unique<Engines::Maxwell3D>(system, rasterizer, *memory_manager);
@@ -137,7 +139,7 @@ u64 GPU::GetTicks() const {
 }
 
 void GPU::FlushCommands() {
-    renderer.Rasterizer().FlushCommands();
+    renderer->Rasterizer().FlushCommands();
 }
 
 // Note that, traditionally, methods are treated as 4-byte addressable locations, and hence

src/video_core/gpu.h

@@ -25,8 +25,11 @@ inline u8* FromCacheAddr(CacheAddr cache_addr) {
 }
 
 namespace Core {
-class System;
+namespace Frontend {
+class EmuWindow;
 }
+class System;
+} // namespace Core
 
 namespace VideoCore {
 class RendererBase;
@@ -39,6 +42,7 @@ enum class RenderTargetFormat : u32 {
     RGBA32_FLOAT = 0xC0,
     RGBA32_UINT = 0xC2,
     RGBA16_UNORM = 0xC6,
+    RGBA16_SNORM = 0xC7,
     RGBA16_UINT = 0xC9,
     RGBA16_FLOAT = 0xCA,
     RG32_FLOAT = 0xCB,
@@ -128,7 +132,8 @@ class MemoryManager;
 
 class GPU {
 public:
-    explicit GPU(Core::System& system, VideoCore::RendererBase& renderer, bool is_async);
+    explicit GPU(Core::System& system, std::unique_ptr<VideoCore::RendererBase>&& renderer,
+                 bool is_async);
 
     virtual ~GPU();
 
@@ -173,6 +178,14 @@ public:
     /// Returns a reference to the GPU DMA pusher.
     Tegra::DmaPusher& DmaPusher();
 
+    VideoCore::RendererBase& Renderer() {
+        return *renderer;
+    }
+
+    const VideoCore::RendererBase& Renderer() const {
+        return *renderer;
+    }
+
     // Waits for the GPU to finish working
     virtual void WaitIdle() const = 0;
 
@@ -257,13 +270,13 @@ public:
     virtual void SwapBuffers(const Tegra::FramebufferConfig* framebuffer) = 0;
 
     /// Notify rasterizer that any caches of the specified region should be flushed to Switch memory
-    virtual void FlushRegion(CacheAddr addr, u64 size) = 0;
+    virtual void FlushRegion(VAddr addr, u64 size) = 0;
 
     /// Notify rasterizer that any caches of the specified region should be invalidated
-    virtual void InvalidateRegion(CacheAddr addr, u64 size) = 0;
+    virtual void InvalidateRegion(VAddr addr, u64 size) = 0;
 
     /// Notify rasterizer that any caches of the specified region should be flushed and invalidated
-    virtual void FlushAndInvalidateRegion(CacheAddr addr, u64 size) = 0;
+    virtual void FlushAndInvalidateRegion(VAddr addr, u64 size) = 0;
 
 protected:
     virtual void TriggerCpuInterrupt(u32 syncpoint_id, u32 value) const = 0;
@@ -286,7 +299,7 @@ private:
 protected:
     std::unique_ptr<Tegra::DmaPusher> dma_pusher;
     Core::System& system;
-    VideoCore::RendererBase& renderer;
+    std::unique_ptr<VideoCore::RendererBase> renderer;
 
 private:
     std::unique_ptr<Tegra::MemoryManager> memory_manager;

src/video_core/gpu_asynch.cpp

@@ -10,13 +10,16 @@
 
 namespace VideoCommon {
 
-GPUAsynch::GPUAsynch(Core::System& system, VideoCore::RendererBase& renderer)
-    : GPU(system, renderer, true), gpu_thread{system} {}
+GPUAsynch::GPUAsynch(Core::System& system, std::unique_ptr<VideoCore::RendererBase>&& renderer_,
+                     std::unique_ptr<Core::Frontend::GraphicsContext>&& context)
+    : GPU(system, std::move(renderer_), true), gpu_thread{system}, gpu_context(std::move(context)),
+      cpu_context(renderer->GetRenderWindow().CreateSharedContext()) {}
 
 GPUAsynch::~GPUAsynch() = default;
 
 void GPUAsynch::Start() {
-    gpu_thread.StartThread(renderer, *dma_pusher);
+    cpu_context->MakeCurrent();
+    gpu_thread.StartThread(*renderer, *gpu_context, *dma_pusher);
 }
 
 void GPUAsynch::PushGPUEntries(Tegra::CommandList&& entries) {
@@ -27,15 +30,15 @@ void GPUAsynch::SwapBuffers(const Tegra::FramebufferConfig* framebuffer) {
     gpu_thread.SwapBuffers(framebuffer);
 }
 
-void GPUAsynch::FlushRegion(CacheAddr addr, u64 size) {
+void GPUAsynch::FlushRegion(VAddr addr, u64 size) {
     gpu_thread.FlushRegion(addr, size);
 }
 
-void GPUAsynch::InvalidateRegion(CacheAddr addr, u64 size) {
+void GPUAsynch::InvalidateRegion(VAddr addr, u64 size) {
     gpu_thread.InvalidateRegion(addr, size);
 }
 
-void GPUAsynch::FlushAndInvalidateRegion(CacheAddr addr, u64 size) {
+void GPUAsynch::FlushAndInvalidateRegion(VAddr addr, u64 size) {
     gpu_thread.FlushAndInvalidateRegion(addr, size);
 }
 

src/video_core/gpu_asynch.h

@@ -7,6 +7,10 @@
 #include "video_core/gpu.h"
 #include "video_core/gpu_thread.h"
 
+namespace Core::Frontend {
+class GraphicsContext;
+}
+
 namespace VideoCore {
 class RendererBase;
 } // namespace VideoCore
@@ -16,15 +20,16 @@ namespace VideoCommon {
 /// Implementation of GPU interface that runs the GPU asynchronously
 class GPUAsynch final : public Tegra::GPU {
 public:
-    explicit GPUAsynch(Core::System& system, VideoCore::RendererBase& renderer);
+    explicit GPUAsynch(Core::System& system, std::unique_ptr<VideoCore::RendererBase>&& renderer,
+                       std::unique_ptr<Core::Frontend::GraphicsContext>&& context);
     ~GPUAsynch() override;
 
     void Start() override;
     void PushGPUEntries(Tegra::CommandList&& entries) override;
     void SwapBuffers(const Tegra::FramebufferConfig* framebuffer) override;
-    void FlushRegion(CacheAddr addr, u64 size) override;
-    void InvalidateRegion(CacheAddr addr, u64 size) override;
-    void FlushAndInvalidateRegion(CacheAddr addr, u64 size) override;
+    void FlushRegion(VAddr addr, u64 size) override;
+    void InvalidateRegion(VAddr addr, u64 size) override;
+    void FlushAndInvalidateRegion(VAddr addr, u64 size) override;
     void WaitIdle() const override;
 
 protected:
@@ -32,6 +37,8 @@ protected:
 
 private:
     GPUThread::ThreadManager gpu_thread;
+    std::unique_ptr<Core::Frontend::GraphicsContext> cpu_context;
+    std::unique_ptr<Core::Frontend::GraphicsContext> gpu_context;
 };
 
 } // namespace VideoCommon

src/video_core/gpu_synch.cpp

@@ -7,12 +7,15 @@
 
 namespace VideoCommon {
 
-GPUSynch::GPUSynch(Core::System& system, VideoCore::RendererBase& renderer)
-    : GPU(system, renderer, false) {}
+GPUSynch::GPUSynch(Core::System& system, std::unique_ptr<VideoCore::RendererBase>&& renderer,
+                   std::unique_ptr<Core::Frontend::GraphicsContext>&& context)
+    : GPU(system, std::move(renderer), false), context{std::move(context)} {}
 
 GPUSynch::~GPUSynch() = default;
 
-void GPUSynch::Start() {}
+void GPUSynch::Start() {
+    context->MakeCurrent();
+}
 
 void GPUSynch::PushGPUEntries(Tegra::CommandList&& entries) {
     dma_pusher->Push(std::move(entries));
@@ -20,19 +23,19 @@ void GPUSynch::PushGPUEntries(Tegra::CommandList&& entries) {
 }
 
 void GPUSynch::SwapBuffers(const Tegra::FramebufferConfig* framebuffer) {
-    renderer.SwapBuffers(framebuffer);
+    renderer->SwapBuffers(framebuffer);
 }
 
-void GPUSynch::FlushRegion(CacheAddr addr, u64 size) {
-    renderer.Rasterizer().FlushRegion(addr, size);
+void GPUSynch::FlushRegion(VAddr addr, u64 size) {
+    renderer->Rasterizer().FlushRegion(addr, size);
 }
 
-void GPUSynch::InvalidateRegion(CacheAddr addr, u64 size) {
-    renderer.Rasterizer().InvalidateRegion(addr, size);
+void GPUSynch::InvalidateRegion(VAddr addr, u64 size) {
+    renderer->Rasterizer().InvalidateRegion(addr, size);
 }
 
-void GPUSynch::FlushAndInvalidateRegion(CacheAddr addr, u64 size) {
-    renderer.Rasterizer().FlushAndInvalidateRegion(addr, size);
+void GPUSynch::FlushAndInvalidateRegion(VAddr addr, u64 size) {
+    renderer->Rasterizer().FlushAndInvalidateRegion(addr, size);
 }
 
 } // namespace VideoCommon

src/video_core/gpu_synch.h

@@ -6,6 +6,10 @@
 
 #include "video_core/gpu.h"
 
+namespace Core::Frontend {
+class GraphicsContext;
+}
+
 namespace VideoCore {
 class RendererBase;
 } // namespace VideoCore
@@ -15,20 +19,24 @@ namespace VideoCommon {
 /// Implementation of GPU interface that runs the GPU synchronously
 class GPUSynch final : public Tegra::GPU {
 public:
-    explicit GPUSynch(Core::System& system, VideoCore::RendererBase& renderer);
+    explicit GPUSynch(Core::System& system, std::unique_ptr<VideoCore::RendererBase>&& renderer,
+                      std::unique_ptr<Core::Frontend::GraphicsContext>&& context);
     ~GPUSynch() override;
 
     void Start() override;
     void PushGPUEntries(Tegra::CommandList&& entries) override;
     void SwapBuffers(const Tegra::FramebufferConfig* framebuffer) override;
-    void FlushRegion(CacheAddr addr, u64 size) override;
-    void InvalidateRegion(CacheAddr addr, u64 size) override;
-    void FlushAndInvalidateRegion(CacheAddr addr, u64 size) override;
+    void FlushRegion(VAddr addr, u64 size) override;
+    void InvalidateRegion(VAddr addr, u64 size) override;
+    void FlushAndInvalidateRegion(VAddr addr, u64 size) override;
     void WaitIdle() const override {}
 
 protected:
     void TriggerCpuInterrupt([[maybe_unused]] u32 syncpoint_id,
                              [[maybe_unused]] u32 value) const override {}
+
+private:
+    std::unique_ptr<Core::Frontend::GraphicsContext> context;
 };
 
 } // namespace VideoCommon

src/video_core/gpu_thread.cpp

@@ -5,7 +5,7 @@
 #include "common/assert.h"
 #include "common/microprofile.h"
 #include "core/core.h"
-#include "core/frontend/scope_acquire_context.h"
+#include "core/frontend/emu_window.h"
 #include "video_core/dma_pusher.h"
 #include "video_core/gpu.h"
 #include "video_core/gpu_thread.h"
@@ -14,8 +14,8 @@
 namespace VideoCommon::GPUThread {
 
 /// Runs the GPU thread
-static void RunThread(VideoCore::RendererBase& renderer, Tegra::DmaPusher& dma_pusher,
-                      SynchState& state) {
+static void RunThread(VideoCore::RendererBase& renderer, Core::Frontend::GraphicsContext& context,
+                      Tegra::DmaPusher& dma_pusher, SynchState& state) {
     MicroProfileOnThreadCreate("GpuThread");
 
     // Wait for first GPU command before acquiring the window context
@@ -27,7 +27,7 @@ static void RunThread(VideoCore::RendererBase& renderer, Tegra::DmaPusher& dma_p
         return;
     }
 
-    Core::Frontend::ScopeAcquireContext acquire_context{renderer.GetRenderWindow()};
+    auto current_context = context.Acquire();
 
     CommandDataContainer next;
     while (state.is_running) {
@@ -62,8 +62,11 @@ ThreadManager::~ThreadManager() {
     thread.join();
 }
 
-void ThreadManager::StartThread(VideoCore::RendererBase& renderer, Tegra::DmaPusher& dma_pusher) {
-    thread = std::thread{RunThread, std::ref(renderer), std::ref(dma_pusher), std::ref(state)};
+void ThreadManager::StartThread(VideoCore::RendererBase& renderer,
+                                Core::Frontend::GraphicsContext& context,
+                                Tegra::DmaPusher& dma_pusher) {
+    thread = std::thread{RunThread, std::ref(renderer), std::ref(context), std::ref(dma_pusher),
+                         std::ref(state)};
 }
 
 void ThreadManager::SubmitList(Tegra::CommandList&& entries) {
@@ -74,15 +77,15 @@ void ThreadManager::SwapBuffers(const Tegra::FramebufferConfig* framebuffer) {
     PushCommand(SwapBuffersCommand(framebuffer ? std::make_optional(*framebuffer) : std::nullopt));
 }
 
-void ThreadManager::FlushRegion(CacheAddr addr, u64 size) {
+void ThreadManager::FlushRegion(VAddr addr, u64 size) {
     PushCommand(FlushRegionCommand(addr, size));
 }
 
-void ThreadManager::InvalidateRegion(CacheAddr addr, u64 size) {
+void ThreadManager::InvalidateRegion(VAddr addr, u64 size) {
     system.Renderer().Rasterizer().InvalidateRegion(addr, size);
 }
 
-void ThreadManager::FlushAndInvalidateRegion(CacheAddr addr, u64 size) {
+void ThreadManager::FlushAndInvalidateRegion(VAddr addr, u64 size) {
     // Skip flush on asynch mode, as FlushAndInvalidateRegion is not used for anything too important
     InvalidateRegion(addr, size);
 }

src/video_core/gpu_thread.h

@@ -10,7 +10,6 @@
 #include <optional>
 #include <thread>
 #include <variant>
-
 #include "common/threadsafe_queue.h"
 #include "video_core/gpu.h"
 
@@ -20,6 +19,9 @@ class DmaPusher;
 } // namespace Tegra
 
 namespace Core {
+namespace Frontend {
+class GraphicsContext;
+}
 class System;
 } // namespace Core
 
@@ -45,26 +47,26 @@ struct SwapBuffersCommand final {
 
 /// Command to signal to the GPU thread to flush a region
 struct FlushRegionCommand final {
-    explicit constexpr FlushRegionCommand(CacheAddr addr, u64 size) : addr{addr}, size{size} {}
+    explicit constexpr FlushRegionCommand(VAddr addr, u64 size) : addr{addr}, size{size} {}
 
-    CacheAddr addr;
+    VAddr addr;
     u64 size;
 };
 
 /// Command to signal to the GPU thread to invalidate a region
 struct InvalidateRegionCommand final {
-    explicit constexpr InvalidateRegionCommand(CacheAddr addr, u64 size) : addr{addr}, size{size} {}
+    explicit constexpr InvalidateRegionCommand(VAddr addr, u64 size) : addr{addr}, size{size} {}
 
-    CacheAddr addr;
+    VAddr addr;
     u64 size;
 };
 
 /// Command to signal to the GPU thread to flush and invalidate a region
 struct FlushAndInvalidateRegionCommand final {
-    explicit constexpr FlushAndInvalidateRegionCommand(CacheAddr addr, u64 size)
+    explicit constexpr FlushAndInvalidateRegionCommand(VAddr addr, u64 size)
         : addr{addr}, size{size} {}
 
-    CacheAddr addr;
+    VAddr addr;
     u64 size;
 };
 
@@ -99,7 +101,8 @@ public:
     ~ThreadManager();
 
     /// Creates and starts the GPU thread.
-    void StartThread(VideoCore::RendererBase& renderer, Tegra::DmaPusher& dma_pusher);
+    void StartThread(VideoCore::RendererBase& renderer, Core::Frontend::GraphicsContext& context,
+                     Tegra::DmaPusher& dma_pusher);
 
     /// Push GPU command entries to be processed
     void SubmitList(Tegra::CommandList&& entries);
@@ -108,13 +111,13 @@ public:
     void SwapBuffers(const Tegra::FramebufferConfig* framebuffer);
 
     /// Notify rasterizer that any caches of the specified region should be flushed to Switch memory
-    void FlushRegion(CacheAddr addr, u64 size);
+    void FlushRegion(VAddr addr, u64 size);
 
     /// Notify rasterizer that any caches of the specified region should be invalidated
-    void InvalidateRegion(CacheAddr addr, u64 size);
+    void InvalidateRegion(VAddr addr, u64 size);
 
     /// Notify rasterizer that any caches of the specified region should be flushed and invalidated
-    void FlushAndInvalidateRegion(CacheAddr addr, u64 size);
+    void FlushAndInvalidateRegion(VAddr addr, u64 size);
 
     // Wait until the gpu thread is idle.
     void WaitIdle() const;

src/video_core/guest_driver.cpp

@@ -4,13 +4,15 @@
 
 #include <algorithm>
 #include <limits>
+#include <vector>
 
+#include "common/common_types.h"
 #include "video_core/guest_driver.h"
 
 namespace VideoCore {
 
-void GuestDriverProfile::DeduceTextureHandlerSize(std::vector<u32>&& bound_offsets) {
-    if (texture_handler_size_deduced) {
+void GuestDriverProfile::DeduceTextureHandlerSize(std::vector<u32> bound_offsets) {
+    if (texture_handler_size) {
         return;
     }
     const std::size_t size = bound_offsets.size();
@@ -29,7 +31,6 @@ void GuestDriverProfile::DeduceTextureHandlerSize(std::vector<u32>&& bound_offse
     if (min_val > 2) {
         return;
     }
-    texture_handler_size_deduced = true;
     texture_handler_size = min_texture_handler_size * min_val;
 }
 

src/video_core/guest_driver.h

@@ -4,6 +4,7 @@
 
 #pragma once
 
+#include <optional>
 #include <vector>
 
 #include "common/common_types.h"
@@ -17,25 +18,29 @@ namespace VideoCore {
  */
 class GuestDriverProfile {
 public:
-    void DeduceTextureHandlerSize(std::vector<u32>&& bound_offsets);
+    explicit GuestDriverProfile() = default;
+    explicit GuestDriverProfile(std::optional<u32> texture_handler_size)
+        : texture_handler_size{texture_handler_size} {}
+
+    void DeduceTextureHandlerSize(std::vector<u32> bound_offsets);
 
     u32 GetTextureHandlerSize() const {
-        return texture_handler_size;
+        return texture_handler_size.value_or(default_texture_handler_size);
     }
 
-    bool TextureHandlerSizeKnown() const {
-        return texture_handler_size_deduced;
+    bool IsTextureHandlerSizeKnown() const {
+        return texture_handler_size.has_value();
     }
 
 private:
     // Minimum size of texture handler any driver can use.
     static constexpr u32 min_texture_handler_size = 4;
-    // This goes with Vulkan and OpenGL standards but Nvidia GPUs can easily
-    // use 4 bytes instead. Thus, certain drivers may squish the size.
+
+    // This goes with Vulkan and OpenGL standards but Nvidia GPUs can easily use 4 bytes instead.
+    // Thus, certain drivers may squish the size.
     static constexpr u32 default_texture_handler_size = 8;
 
-    u32 texture_handler_size = default_texture_handler_size;
-    bool texture_handler_size_deduced = false;
+    std::optional<u32> texture_handler_size = default_texture_handler_size;
 };
 
 } // namespace VideoCore

src/video_core/memory_manager.cpp

@@ -81,12 +81,11 @@ GPUVAddr MemoryManager::UnmapBuffer(GPUVAddr gpu_addr, u64 size) {
     ASSERT((gpu_addr & page_mask) == 0);
 
     const u64 aligned_size{Common::AlignUp(size, page_size)};
-    const CacheAddr cache_addr{ToCacheAddr(GetPointer(gpu_addr))};
     const auto cpu_addr = GpuToCpuAddress(gpu_addr);
     ASSERT(cpu_addr);
 
     // Flush and invalidate through the GPU interface, to be asynchronous if possible.
-    system.GPU().FlushAndInvalidateRegion(cache_addr, aligned_size);
+    system.GPU().FlushAndInvalidateRegion(*cpu_addr, aligned_size);
 
     UnmapRange(gpu_addr, aligned_size);
     ASSERT(system.CurrentProcess()
@@ -140,11 +139,11 @@ T MemoryManager::Read(GPUVAddr addr) const {
         return {};
     }
 
-    const u8* page_pointer{page_table.pointers[addr >> page_bits]};
+    const u8* page_pointer{GetPointer(addr)};
     if (page_pointer) {
         // NOTE: Avoid adding any extra logic to this fast-path block
         T value;
-        std::memcpy(&value, &page_pointer[addr & page_mask], sizeof(T));
+        std::memcpy(&value, page_pointer, sizeof(T));
         return value;
     }
 
@@ -167,10 +166,10 @@ void MemoryManager::Write(GPUVAddr addr, T data) {
         return;
     }
 
-    u8* page_pointer{page_table.pointers[addr >> page_bits]};
+    u8* page_pointer{GetPointer(addr)};
     if (page_pointer) {
         // NOTE: Avoid adding any extra logic to this fast-path block
-        std::memcpy(&page_pointer[addr & page_mask], &data, sizeof(T));
+        std::memcpy(page_pointer, &data, sizeof(T));
         return;
     }
 
@@ -201,9 +200,12 @@ u8* MemoryManager::GetPointer(GPUVAddr addr) {
         return {};
     }
 
-    u8* const page_pointer{page_table.pointers[addr >> page_bits]};
-    if (page_pointer != nullptr) {
-        return page_pointer + (addr & page_mask);
+    auto& memory = system.Memory();
+
+    const VAddr page_addr{page_table.backing_addr[addr >> page_bits]};
+
+    if (page_addr != 0) {
+        return memory.GetPointer(page_addr + (addr & page_mask));
     }
 
     LOG_ERROR(HW_GPU, "Unknown GetPointer @ 0x{:016X}", addr);
@@ -215,9 +217,12 @@ const u8* MemoryManager::GetPointer(GPUVAddr addr) const {
         return {};
     }
 
-    const u8* const page_pointer{page_table.pointers[addr >> page_bits]};
-    if (page_pointer != nullptr) {
-        return page_pointer + (addr & page_mask);
+    const auto& memory = system.Memory();
+
+    const VAddr page_addr{page_table.backing_addr[addr >> page_bits]};
+
+    if (page_addr != 0) {
+        return memory.GetPointer(page_addr + (addr & page_mask));
     }
 
     LOG_ERROR(HW_GPU, "Unknown GetPointer @ 0x{:016X}", addr);
@@ -238,17 +243,19 @@ void MemoryManager::ReadBlock(GPUVAddr src_addr, void* dest_buffer, const std::s
     std::size_t page_index{src_addr >> page_bits};
     std::size_t page_offset{src_addr & page_mask};
 
+    auto& memory = system.Memory();
+
     while (remaining_size > 0) {
         const std::size_t copy_amount{
             std::min(static_cast<std::size_t>(page_size) - page_offset, remaining_size)};
 
         switch (page_table.attributes[page_index]) {
         case Common::PageType::Memory: {
-            const u8* src_ptr{page_table.pointers[page_index] + page_offset};
+            const VAddr src_addr{page_table.backing_addr[page_index] + page_offset};
             // Flush must happen on the rasterizer interface, such that memory is always synchronous
             // when it is read (even when in asynchronous GPU mode). Fixes Dead Cells title menu.
-            rasterizer.FlushRegion(ToCacheAddr(src_ptr), copy_amount);
-            std::memcpy(dest_buffer, src_ptr, copy_amount);
+            rasterizer.FlushRegion(src_addr, copy_amount);
+            memory.ReadBlockUnsafe(src_addr, dest_buffer, copy_amount);
             break;
         }
         default:
@@ -268,13 +275,15 @@ void MemoryManager::ReadBlockUnsafe(GPUVAddr src_addr, void* dest_buffer,
     std::size_t page_index{src_addr >> page_bits};
     std::size_t page_offset{src_addr & page_mask};
 
+    auto& memory = system.Memory();
+
     while (remaining_size > 0) {
         const std::size_t copy_amount{
             std::min(static_cast<std::size_t>(page_size) - page_offset, remaining_size)};
         const u8* page_pointer = page_table.pointers[page_index];
         if (page_pointer) {
-            const u8* src_ptr{page_pointer + page_offset};
-            std::memcpy(dest_buffer, src_ptr, copy_amount);
+            const VAddr src_addr{page_table.backing_addr[page_index] + page_offset};
+            memory.ReadBlockUnsafe(src_addr, dest_buffer, copy_amount);
         } else {
             std::memset(dest_buffer, 0, copy_amount);
         }
@@ -290,17 +299,19 @@ void MemoryManager::WriteBlock(GPUVAddr dest_addr, const void* src_buffer, const
     std::size_t page_index{dest_addr >> page_bits};
     std::size_t page_offset{dest_addr & page_mask};
 
+    auto& memory = system.Memory();
+
     while (remaining_size > 0) {
         const std::size_t copy_amount{
             std::min(static_cast<std::size_t>(page_size) - page_offset, remaining_size)};
 
         switch (page_table.attributes[page_index]) {
         case Common::PageType::Memory: {
-            u8* dest_ptr{page_table.pointers[page_index] + page_offset};
+            const VAddr dest_addr{page_table.backing_addr[page_index] + page_offset};
             // Invalidate must happen on the rasterizer interface, such that memory is always
             // synchronous when it is written (even when in asynchronous GPU mode).
-            rasterizer.InvalidateRegion(ToCacheAddr(dest_ptr), copy_amount);
-            std::memcpy(dest_ptr, src_buffer, copy_amount);
+            rasterizer.InvalidateRegion(dest_addr, copy_amount);
+            memory.WriteBlockUnsafe(dest_addr, src_buffer, copy_amount);
             break;
         }
         default:
@@ -320,13 +331,15 @@ void MemoryManager::WriteBlockUnsafe(GPUVAddr dest_addr, const void* src_buffer,
     std::size_t page_index{dest_addr >> page_bits};
     std::size_t page_offset{dest_addr & page_mask};
 
+    auto& memory = system.Memory();
+
     while (remaining_size > 0) {
         const std::size_t copy_amount{
             std::min(static_cast<std::size_t>(page_size) - page_offset, remaining_size)};
         u8* page_pointer = page_table.pointers[page_index];
         if (page_pointer) {
-            u8* dest_ptr{page_pointer + page_offset};
-            std::memcpy(dest_ptr, src_buffer, copy_amount);
+            const VAddr dest_addr{page_table.backing_addr[page_index] + page_offset};
+            memory.WriteBlockUnsafe(dest_addr, src_buffer, copy_amount);
         }
         page_index++;
         page_offset = 0;
@@ -336,33 +349,9 @@ void MemoryManager::WriteBlockUnsafe(GPUVAddr dest_addr, const void* src_buffer,
 }
 
 void MemoryManager::CopyBlock(GPUVAddr dest_addr, GPUVAddr src_addr, const std::size_t size) {
-    std::size_t remaining_size{size};
-    std::size_t page_index{src_addr >> page_bits};
-    std::size_t page_offset{src_addr & page_mask};
-
-    while (remaining_size > 0) {
-        const std::size_t copy_amount{
-            std::min(static_cast<std::size_t>(page_size) - page_offset, remaining_size)};
-
-        switch (page_table.attributes[page_index]) {
-        case Common::PageType::Memory: {
-            // Flush must happen on the rasterizer interface, such that memory is always synchronous
-            // when it is copied (even when in asynchronous GPU mode).
-            const u8* src_ptr{page_table.pointers[page_index] + page_offset};
-            rasterizer.FlushRegion(ToCacheAddr(src_ptr), copy_amount);
-            WriteBlock(dest_addr, src_ptr, copy_amount);
-            break;
-        }
-        default:
-            UNREACHABLE();
-        }
-
-        page_index++;
-        page_offset = 0;
-        dest_addr += static_cast<VAddr>(copy_amount);
-        src_addr += static_cast<VAddr>(copy_amount);
-        remaining_size -= copy_amount;
-    }
+    std::vector<u8> tmp_buffer(size);
+    ReadBlock(src_addr, tmp_buffer.data(), size);
+    WriteBlock(dest_addr, tmp_buffer.data(), size);
 }
 
 void MemoryManager::CopyBlockUnsafe(GPUVAddr dest_addr, GPUVAddr src_addr, const std::size_t size) {
@@ -371,6 +360,12 @@ void MemoryManager::CopyBlockUnsafe(GPUVAddr dest_addr, GPUVAddr src_addr, const
     WriteBlockUnsafe(dest_addr, tmp_buffer.data(), size);
 }
 
+bool MemoryManager::IsGranularRange(GPUVAddr gpu_addr, std::size_t size) {
+    const VAddr addr = page_table.backing_addr[gpu_addr >> page_bits];
+    const std::size_t page = (addr & Memory::PAGE_MASK) + size;
+    return page <= Memory::PAGE_SIZE;
+}
+
 void MemoryManager::MapPages(GPUVAddr base, u64 size, u8* memory, Common::PageType type,
                              VAddr backing_addr) {
     LOG_DEBUG(HW_GPU, "Mapping {} onto {:016X}-{:016X}", fmt::ptr(memory), base * page_size,

src/video_core/memory_manager.h

@@ -97,6 +97,11 @@ public:
     void WriteBlockUnsafe(GPUVAddr dest_addr, const void* src_buffer, std::size_t size);
     void CopyBlockUnsafe(GPUVAddr dest_addr, GPUVAddr src_addr, std::size_t size);
 
+    /**
+     * IsGranularRange checks if a gpu region can be simply read with a pointer
+     */
+    bool IsGranularRange(GPUVAddr gpu_addr, std::size_t size);
+
 private:
     using VMAMap = std::map<GPUVAddr, VirtualMemoryArea>;
     using VMAHandle = VMAMap::const_iterator;
@@ -174,7 +179,7 @@ private:
     /// End of address space, based on address space in bits.
     static constexpr GPUVAddr address_space_end{1ULL << address_space_width};
 
-    Common::PageTable page_table{page_bits};
+    Common::BackingPageTable page_table{page_bits};
     VMAMap vma_map;
     VideoCore::RasterizerInterface& rasterizer;
 

src/video_core/morton.cpp

@@ -51,6 +51,7 @@ static constexpr ConversionArray morton_to_linear_fns = {
     MortonCopy<true, PixelFormat::R8UI>,
     MortonCopy<true, PixelFormat::RGBA16F>,
     MortonCopy<true, PixelFormat::RGBA16U>,
+    MortonCopy<true, PixelFormat::RGBA16S>,
     MortonCopy<true, PixelFormat::RGBA16UI>,
     MortonCopy<true, PixelFormat::R11FG11FB10F>,
     MortonCopy<true, PixelFormat::RGBA32UI>,
@@ -131,6 +132,7 @@ static constexpr ConversionArray linear_to_morton_fns = {
     MortonCopy<false, PixelFormat::R8U>,
     MortonCopy<false, PixelFormat::R8UI>,
     MortonCopy<false, PixelFormat::RGBA16F>,
+    MortonCopy<false, PixelFormat::RGBA16S>,
     MortonCopy<false, PixelFormat::RGBA16U>,
     MortonCopy<false, PixelFormat::RGBA16UI>,
     MortonCopy<false, PixelFormat::R11FG11FB10F>,

src/video_core/query_cache.h

@@ -98,12 +98,12 @@ public:
                                                       static_cast<QueryCache&>(*this),
                                                       VideoCore::QueryType::SamplesPassed}}} {}
 
-    void InvalidateRegion(CacheAddr addr, std::size_t size) {
+    void InvalidateRegion(VAddr addr, std::size_t size) {
         std::unique_lock lock{mutex};
         FlushAndRemoveRegion(addr, size);
     }
 
-    void FlushRegion(CacheAddr addr, std::size_t size) {
+    void FlushRegion(VAddr addr, std::size_t size) {
         std::unique_lock lock{mutex};
         FlushAndRemoveRegion(addr, size);
     }
@@ -117,14 +117,16 @@ public:
     void Query(GPUVAddr gpu_addr, VideoCore::QueryType type, std::optional<u64> timestamp) {
         std::unique_lock lock{mutex};
         auto& memory_manager = system.GPU().MemoryManager();
-        const auto host_ptr = memory_manager.GetPointer(gpu_addr);
+        const std::optional<VAddr> cpu_addr_opt = memory_manager.GpuToCpuAddress(gpu_addr);
+        ASSERT(cpu_addr_opt);
+        VAddr cpu_addr = *cpu_addr_opt;
 
-        CachedQuery* query = TryGet(ToCacheAddr(host_ptr));
+        CachedQuery* query = TryGet(cpu_addr);
         if (!query) {
-            const auto cpu_addr = memory_manager.GpuToCpuAddress(gpu_addr);
-            ASSERT_OR_EXECUTE(cpu_addr, return;);
+            ASSERT_OR_EXECUTE(cpu_addr_opt, return;);
+            const auto host_ptr = memory_manager.GetPointer(gpu_addr);
 
-            query = Register(type, *cpu_addr, host_ptr, timestamp.has_value());
+            query = Register(type, cpu_addr, host_ptr, timestamp.has_value());
         }
 
         query->BindCounter(Stream(type).Current(), timestamp);
@@ -173,11 +175,11 @@ protected:
 
 private:
     /// Flushes a memory range to guest memory and removes it from the cache.
-    void FlushAndRemoveRegion(CacheAddr addr, std::size_t size) {
+    void FlushAndRemoveRegion(VAddr addr, std::size_t size) {
         const u64 addr_begin = static_cast<u64>(addr);
         const u64 addr_end = addr_begin + static_cast<u64>(size);
         const auto in_range = [addr_begin, addr_end](CachedQuery& query) {
-            const u64 cache_begin = query.GetCacheAddr();
+            const u64 cache_begin = query.GetCpuAddr();
             const u64 cache_end = cache_begin + query.SizeInBytes();
             return cache_begin < addr_end && addr_begin < cache_end;
         };
@@ -193,7 +195,7 @@ private:
                 if (!in_range(query)) {
                     continue;
                 }
-                rasterizer.UpdatePagesCachedCount(query.CpuAddr(), query.SizeInBytes(), -1);
+                rasterizer.UpdatePagesCachedCount(query.GetCpuAddr(), query.SizeInBytes(), -1);
                 query.Flush();
             }
             contents.erase(std::remove_if(std::begin(contents), std::end(contents), in_range),
@@ -204,22 +206,21 @@ private:
     /// Registers the passed parameters as cached and returns a pointer to the stored cached query.
     CachedQuery* Register(VideoCore::QueryType type, VAddr cpu_addr, u8* host_ptr, bool timestamp) {
         rasterizer.UpdatePagesCachedCount(cpu_addr, CachedQuery::SizeInBytes(timestamp), 1);
-        const u64 page = static_cast<u64>(ToCacheAddr(host_ptr)) >> PAGE_SHIFT;
+        const u64 page = static_cast<u64>(cpu_addr) >> PAGE_SHIFT;
         return &cached_queries[page].emplace_back(static_cast<QueryCache&>(*this), type, cpu_addr,
                                                   host_ptr);
     }
 
     /// Tries to a get a cached query. Returns nullptr on failure.
-    CachedQuery* TryGet(CacheAddr addr) {
+    CachedQuery* TryGet(VAddr addr) {
         const u64 page = static_cast<u64>(addr) >> PAGE_SHIFT;
         const auto it = cached_queries.find(page);
         if (it == std::end(cached_queries)) {
             return nullptr;
         }
         auto& contents = it->second;
-        const auto found =
-            std::find_if(std::begin(contents), std::end(contents),
-                         [addr](auto& query) { return query.GetCacheAddr() == addr; });
+        const auto found = std::find_if(std::begin(contents), std::end(contents),
+                                        [addr](auto& query) { return query.GetCpuAddr() == addr; });
         return found != std::end(contents) ? &*found : nullptr;
     }
 
@@ -323,14 +324,10 @@ public:
         timestamp = timestamp_;
     }
 
-    VAddr CpuAddr() const noexcept {
+    VAddr GetCpuAddr() const noexcept {
         return cpu_addr;
     }
 
-    CacheAddr GetCacheAddr() const noexcept {
-        return ToCacheAddr(host_ptr);
-    }
-
     u64 SizeInBytes() const noexcept {
         return SizeInBytes(timestamp.has_value());
     }

src/video_core/rasterizer_cache.h

@@ -18,22 +18,14 @@
 
 class RasterizerCacheObject {
 public:
-    explicit RasterizerCacheObject(const u8* host_ptr)
-        : host_ptr{host_ptr}, cache_addr{ToCacheAddr(host_ptr)} {}
+    explicit RasterizerCacheObject(const VAddr cpu_addr) : cpu_addr{cpu_addr} {}
 
     virtual ~RasterizerCacheObject();
 
-    CacheAddr GetCacheAddr() const {
-        return cache_addr;
+    VAddr GetCpuAddr() const {
+        return cpu_addr;
     }
 
-    const u8* GetHostPtr() const {
-        return host_ptr;
-    }
-
-    /// Gets the address of the shader in guest memory, required for cache management
-    virtual VAddr GetCpuAddr() const = 0;
-
     /// Gets the size of the shader in guest memory, required for cache management
     virtual std::size_t GetSizeInBytes() const = 0;
 
@@ -68,8 +60,7 @@ private:
     bool is_registered{};      ///< Whether the object is currently registered with the cache
     bool is_dirty{};           ///< Whether the object is dirty (out of sync with guest memory)
     u64 last_modified_ticks{}; ///< When the object was last modified, used for in-order flushing
-    const u8* host_ptr{};      ///< Pointer to the memory backing this cached region
-    CacheAddr cache_addr{};    ///< Cache address memory, unique from emulated virtual address space
+    VAddr cpu_addr{};          ///< Cpu address memory, unique from emulated virtual address space
 };
 
 template <class T>
@@ -80,7 +71,7 @@ public:
     explicit RasterizerCache(VideoCore::RasterizerInterface& rasterizer) : rasterizer{rasterizer} {}
 
     /// Write any cached resources overlapping the specified region back to memory
-    void FlushRegion(CacheAddr addr, std::size_t size) {
+    void FlushRegion(VAddr addr, std::size_t size) {
         std::lock_guard lock{mutex};
 
         const auto& objects{GetSortedObjectsFromRegion(addr, size)};
@@ -90,7 +81,7 @@ public:
     }
 
     /// Mark the specified region as being invalidated
-    void InvalidateRegion(CacheAddr addr, u64 size) {
+    void InvalidateRegion(VAddr addr, u64 size) {
         std::lock_guard lock{mutex};
 
         const auto& objects{GetSortedObjectsFromRegion(addr, size)};
@@ -114,27 +105,20 @@ public:
 
 protected:
     /// Tries to get an object from the cache with the specified cache address
-    T TryGet(CacheAddr addr) const {
+    T TryGet(VAddr addr) const {
         const auto iter = map_cache.find(addr);
         if (iter != map_cache.end())
             return iter->second;
         return nullptr;
     }
 
-    T TryGet(const void* addr) const {
-        const auto iter = map_cache.find(ToCacheAddr(addr));
-        if (iter != map_cache.end())
-            return iter->second;
-        return nullptr;
-    }
-
     /// Register an object into the cache
     virtual void Register(const T& object) {
         std::lock_guard lock{mutex};
 
         object->SetIsRegistered(true);
         interval_cache.add({GetInterval(object), ObjectSet{object}});
-        map_cache.insert({object->GetCacheAddr(), object});
+        map_cache.insert({object->GetCpuAddr(), object});
         rasterizer.UpdatePagesCachedCount(object->GetCpuAddr(), object->GetSizeInBytes(), 1);
     }
 
@@ -144,7 +128,7 @@ protected:
 
         object->SetIsRegistered(false);
         rasterizer.UpdatePagesCachedCount(object->GetCpuAddr(), object->GetSizeInBytes(), -1);
-        const CacheAddr addr = object->GetCacheAddr();
+        const VAddr addr = object->GetCpuAddr();
         interval_cache.subtract({GetInterval(object), ObjectSet{object}});
         map_cache.erase(addr);
     }
@@ -173,7 +157,7 @@ protected:
 
 private:
     /// Returns a list of cached objects from the specified memory region, ordered by access time
-    std::vector<T> GetSortedObjectsFromRegion(CacheAddr addr, u64 size) {
+    std::vector<T> GetSortedObjectsFromRegion(VAddr addr, u64 size) {
         if (size == 0) {
             return {};
         }
@@ -197,13 +181,13 @@ private:
     }
 
     using ObjectSet = std::set<T>;
-    using ObjectCache = std::unordered_map<CacheAddr, T>;
-    using IntervalCache = boost::icl::interval_map<CacheAddr, ObjectSet>;
+    using ObjectCache = std::unordered_map<VAddr, T>;
+    using IntervalCache = boost::icl::interval_map<VAddr, ObjectSet>;
     using ObjectInterval = typename IntervalCache::interval_type;
 
     static auto GetInterval(const T& object) {
-        return ObjectInterval::right_open(object->GetCacheAddr(),
-                                          object->GetCacheAddr() + object->GetSizeInBytes());
+        return ObjectInterval::right_open(object->GetCpuAddr(),
+                                          object->GetCpuAddr() + object->GetSizeInBytes());
     }
 
     ObjectCache map_cache;