android: Fix input overlay version check

yuzu: add opengl shader backend info in status bar

.github/workflows/verify.yml

@@ -165,24 +165,3 @@ jobs:
         with:
           name: android
           path: artifacts/
-  release:
-    runs-on: ubuntu-latest
-    needs: [ android ]
-    if: ${{ startsWith(github.ref, 'refs/tags/') }}
-    steps:
-      - uses: actions/download-artifact@v3
-      - name: Create release
-        uses: actions/create-release@v1
-        env:
-          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
-        with:
-          tag_name: ${{ github.ref_name }}
-          release_name: ${{ github.ref_name }}
-          draft: false
-          prerelease: false
-      - name: Upload artifacts
-        uses: alexellis/upload-assets@0.4.0
-        env:
-          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
-        with:
-          asset_paths: '["./**/*.tar.*","./**/*.AppImage","./**/*.7z","./**/*.zip","./**/*.apk","./**/*.aab"]'

.gitmodules

@@ -49,6 +49,6 @@
 [submodule "cpp-jwt"]
 	path = externals/cpp-jwt
 	url = https://github.com/arun11299/cpp-jwt.git
-[submodule "externals/libadrenotools"]
+[submodule "libadrenotools"]
 	path = externals/libadrenotools
 	url = https://github.com/bylaws/libadrenotools

CMakeLists.txt

@@ -255,7 +255,7 @@ endif()
 # boost asio's concept usage doesn't play nicely with some compilers yet.
 add_definitions(-DBOOST_ASIO_DISABLE_CONCEPTS)
 if (MSVC)
-    add_compile_options($<$<COMPILE_LANGUAGE:CXX>:/std:c++latest>)
+    add_compile_options($<$<COMPILE_LANGUAGE:CXX>:/std:c++20>)
 
     # boost still makes use of deprecated result_of.
     add_definitions(-D_HAS_DEPRECATED_RESULT_OF)

src/CMakeLists.txt

@@ -43,7 +43,7 @@ if (MSVC)
         /Zo
         /permissive-
         /EHsc
-        /std:c++latest
+        /std:c++20
         /utf-8
         /volatile:iso
         /Zc:externConstexpr
@@ -51,8 +51,10 @@ if (MSVC)
         /Zc:throwingNew
         /GT
 
+        # Modules
+        /experimental:module- # Disable module support explicitly due to conflicts with precompiled headers
+
         # External headers diagnostics
-        /experimental:external  # Enables the external headers options. This option isn't required in Visual Studio 2019 version 16.10 and later
         /external:anglebrackets # Treats all headers included by #include <header>, where the header file is enclosed in angle brackets (< >), as external headers
         /external:W0            # Sets the default warning level to 0 for external headers, effectively turning off warnings for external headers
 

src/android/app/build.gradle.kts

@@ -57,6 +57,7 @@ android {
         applicationId = "org.yuzu.yuzu_emu"
         minSdk = 30
         targetSdk = 33
+        versionCode = 1
         versionName = getGitVersion()
 
         ndk {

src/android/app/src/main/AndroidManifest.xml

@@ -6,17 +6,10 @@ SPDX-License-Identifier: GPL-3.0-or-later
 -->
 
 <manifest xmlns:android="http://schemas.android.com/apk/res/android">
-    <uses-feature
-        android:name="android.hardware.touchscreen"
-        android:required="false"/>
-    <uses-feature
-        android:name="android.hardware.gamepad"
-        android:required="false"/>
-
-    <uses-feature
-        android:name="android.hardware.vulkan.version"
-        android:version="0x401000"
-        android:required="true" />
+    <uses-feature android:name="android.hardware.touchscreen" android:required="false" />
+    <uses-feature android:name="android.hardware.gamepad" android:required="false" />
+    <uses-feature android:name="android.software.leanback" android:required="false" />
+    <uses-feature android:name="android.hardware.vulkan.version" android:version="0x401000" android:required="true" />
 
     <uses-permission android:name="android.permission.INTERNET" />
     <uses-permission android:name="android.permission.FOREGROUND_SERVICE" />
@@ -31,7 +24,7 @@ SPDX-License-Identifier: GPL-3.0-or-later
         android:hasFragileUserData="true"
         android:supportsRtl="true"
         android:isGame="true"
-        android:banner="@drawable/ic_launcher"
+        android:banner="@drawable/tv_banner"
         android:extractNativeLibs="true"
         android:fullBackupContent="@xml/data_extraction_rules"
         android:dataExtractionRules="@xml/data_extraction_rules_api_31"
@@ -44,9 +37,10 @@ SPDX-License-Identifier: GPL-3.0-or-later
 
             <!-- This intentfilter marks this Activity as the one that gets launched from Home screen. -->
             <intent-filter>
-                <action android:name="android.intent.action.MAIN"/>
+                <action android:name="android.intent.action.MAIN" />
 
-                <category android:name="android.intent.category.LAUNCHER"/>
+                <category android:name="android.intent.category.LAUNCHER" />
+                <category android:name="android.intent.category.LEANBACK_LAUNCHER" />
             </intent-filter>
         </activity>
 

src/android/app/src/main/java/org/yuzu/yuzu_emu/activities/EmulationActivity.kt

@@ -263,7 +263,8 @@ class EmulationActivity : AppCompatActivity(), SensorEventListener {
         val config: Configuration = resources.configuration
 
         if ((config.screenLayout and Configuration.SCREENLAYOUT_LONG_YES) != 0 ||
-            (config.screenLayout and Configuration.SCREENLAYOUT_LONG_NO) == 0) {
+            (config.screenLayout and Configuration.SCREENLAYOUT_LONG_NO) == 0 ||
+            (config.screenLayout and Configuration.SCREENLAYOUT_SIZE_SMALL) != 0) {
             return rotation
         }
         when (rotation) {

src/android/app/src/main/java/org/yuzu/yuzu_emu/overlay/InputOverlay.kt

@@ -765,18 +765,20 @@ class InputOverlay(context: Context, attrs: AttributeSet?) : SurfaceView(context
             // If we have API access, calculate the safe area to draw the overlay
             var cutoutLeft = 0
             var cutoutBottom = 0
-            val insets = context.windowManager.currentWindowMetrics.windowInsets.displayCutout
-            if (insets != null && Build.VERSION.SDK_INT >= Build.VERSION_CODES.Q) {
-                if (insets.boundingRectTop.bottom != 0 && insets.boundingRectTop.bottom > maxY / 2)
-                    insets.boundingRectTop.bottom.toFloat() else maxY
-                if (insets.boundingRectRight.left != 0 && insets.boundingRectRight.left > maxX / 2)
-                    insets.boundingRectRight.left.toFloat() else maxX
+            if (Build.VERSION.SDK_INT >= Build.VERSION_CODES.R) {
+                val insets = context.windowManager.currentWindowMetrics.windowInsets.displayCutout
+                if (insets != null) {
+                    if (insets.boundingRectTop.bottom != 0 && insets.boundingRectTop.bottom > maxY / 2)
+                        insets.boundingRectTop.bottom.toFloat() else maxY
+                    if (insets.boundingRectRight.left != 0 && insets.boundingRectRight.left > maxX / 2)
+                        insets.boundingRectRight.left.toFloat() else maxX
 
-                minX = insets.boundingRectLeft.right - insets.boundingRectLeft.left
-                minY = insets.boundingRectBottom.top - insets.boundingRectBottom.bottom
+                    minX = insets.boundingRectLeft.right - insets.boundingRectLeft.left
+                    minY = insets.boundingRectBottom.top - insets.boundingRectBottom.bottom
 
-                cutoutLeft = insets.boundingRectRight.right - insets.boundingRectRight.left
-                cutoutBottom = insets.boundingRectTop.top - insets.boundingRectTop.bottom
+                    cutoutLeft = insets.boundingRectRight.right - insets.boundingRectRight.left
+                    cutoutBottom = insets.boundingRectTop.top - insets.boundingRectTop.bottom
+                }
             }
 
             // This makes sure that if we have an inset on one side of the screen, we mirror it on

src/android/gradle.properties

@@ -9,8 +9,9 @@
 # http://www.gradle.org/docs/current/userguide/build_environment.html
 # Specifies the JVM arguments used for the daemon process.
 # The setting is particularly useful for tweaking memory settings.
-org.gradle.jvmargs=-Xmx2048m -XX:MaxPermSize=512m -XX:+HeapDumpOnOutOfMemoryError -Dfile.encoding=UTF-8
+org.gradle.jvmargs=-Xms512m -XX:MaxMetaspaceSize=512m -XX:+HeapDumpOnOutOfMemoryError -Dfile.encoding=UTF-8
 android.useAndroidX=true
 # Kotlin code style for this project: "official" or "obsolete":
 kotlin.code.style=official
+kotlin.parallel.tasks.in.project=true
 android.defaults.buildfeatures.buildconfig=true

src/audio_core/audio_core.cpp

@@ -47,12 +47,4 @@ AudioRenderer::ADSP::ADSP& AudioCore::GetADSP() {
     return *adsp;
 }
 
-void AudioCore::SetNVDECActive(bool active) {
-    nvdec_active = active;
-}
-
-bool AudioCore::IsNVDECActive() const {
-    return nvdec_active;
-}
-
 } // namespace AudioCore

src/audio_core/audio_core.h

@@ -57,18 +57,6 @@ public:
      */
     AudioRenderer::ADSP::ADSP& GetADSP();
 
-    /**
-     * Toggle NVDEC state, used to avoid stall in playback.
-     *
-     * @param active - Set true if nvdec is active, otherwise false.
-     */
-    void SetNVDECActive(bool active);
-
-    /**
-     * Get NVDEC state.
-     */
-    bool IsNVDECActive() const;
-
 private:
     /**
      * Create the sinks on startup.
@@ -83,8 +71,6 @@ private:
     std::unique_ptr<Sink::Sink> input_sink;
     /// The ADSP in the sysmodule
     std::unique_ptr<AudioRenderer::ADSP::ADSP> adsp;
-    /// Is NVDec currently active?
-    bool nvdec_active{false};
 };
 
 } // namespace AudioCore

src/common/fs/fs_paths.h

@@ -10,6 +10,7 @@
 
 // Sub-directories contained within a yuzu data directory
 
+#define AMIIBO_DIR "amiibo"
 #define CACHE_DIR "cache"
 #define CONFIG_DIR "config"
 #define DUMP_DIR "dump"

src/common/fs/path_util.cpp

@@ -114,6 +114,7 @@ public:
 #endif
 
         GenerateYuzuPath(YuzuPath::YuzuDir, yuzu_path);
+        GenerateYuzuPath(YuzuPath::AmiiboDir, yuzu_path / AMIIBO_DIR);
         GenerateYuzuPath(YuzuPath::CacheDir, yuzu_path_cache);
         GenerateYuzuPath(YuzuPath::ConfigDir, yuzu_path_config);
         GenerateYuzuPath(YuzuPath::DumpDir, yuzu_path / DUMP_DIR);

src/common/fs/path_util.h

@@ -12,6 +12,7 @@ namespace Common::FS {
 
 enum class YuzuPath {
     YuzuDir,        // Where yuzu stores its data.
+    AmiiboDir,      // Where Amiibo backups are stored.
     CacheDir,       // Where cached filesystem data is stored.
     ConfigDir,      // Where config files are stored.
     DumpDir,        // Where dumped data is stored.

src/common/settings.cpp

@@ -235,6 +235,7 @@ void RestoreGlobalState(bool is_powered_on) {
     values.bg_green.SetGlobal(true);
     values.bg_blue.SetGlobal(true);
     values.enable_compute_pipelines.SetGlobal(true);
+    values.use_video_framerate.SetGlobal(true);
 
     // System
     values.language_index.SetGlobal(true);

src/common/settings.h

@@ -482,6 +482,7 @@ struct Values {
     SwitchableSetting<AstcRecompression, true> astc_recompression{
         AstcRecompression::Uncompressed, AstcRecompression::Uncompressed, AstcRecompression::Bc3,
         "astc_recompression"};
+    SwitchableSetting<bool> use_video_framerate{false, "use_video_framerate"};
 
     SwitchableSetting<u8> bg_red{0, "bg_red"};
     SwitchableSetting<u8> bg_green{0, "bg_green"};

src/common/uuid.cpp

@@ -48,7 +48,7 @@ std::array<u8, 0x10> ConstructFromRawString(std::string_view raw_string) {
 }
 
 std::array<u8, 0x10> ConstructFromFormattedString(std::string_view formatted_string) {
-    std::array<u8, 0x10> uuid;
+    std::array<u8, 0x10> uuid{};
 
     size_t i = 0;
 

src/core/core.cpp

@@ -216,6 +216,14 @@ struct System::Impl {
         }
     }
 
+    void SetNVDECActive(bool is_nvdec_active) {
+        nvdec_active = is_nvdec_active;
+    }
+
+    bool GetNVDECActive() {
+        return nvdec_active;
+    }
+
     void InitializeDebugger(System& system, u16 port) {
         debugger = std::make_unique<Debugger>(system, port);
     }
@@ -485,6 +493,8 @@ struct System::Impl {
     std::atomic_bool is_powered_on{};
     bool exit_lock = false;
 
+    bool nvdec_active{};
+
     Reporter reporter;
     std::unique_ptr<Memory::CheatEngine> cheat_engine;
     std::unique_ptr<Tools::Freezer> memory_freezer;
@@ -594,6 +604,14 @@ void System::UnstallApplication() {
     impl->UnstallApplication();
 }
 
+void System::SetNVDECActive(bool is_nvdec_active) {
+    impl->SetNVDECActive(is_nvdec_active);
+}
+
+bool System::GetNVDECActive() {
+    return impl->GetNVDECActive();
+}
+
 void System::InitializeDebugger() {
     impl->InitializeDebugger(*this, Settings::values.gdbstub_port.GetValue());
 }

src/core/core.h

@@ -189,6 +189,9 @@ public:
     std::unique_lock<std::mutex> StallApplication();
     void UnstallApplication();
 
+    void SetNVDECActive(bool is_nvdec_active);
+    [[nodiscard]] bool GetNVDECActive();
+
     /**
      * Initialize the debugger.
      */

src/core/file_sys/control_metadata.cpp

@@ -23,8 +23,8 @@ const std::array<const char*, 16> LANGUAGE_NAMES{{
     "Portuguese",
     "Russian",
     "Korean",
-    "Taiwanese",
-    "Chinese",
+    "TraditionalChinese",
+    "SimplifiedChinese",
     "BrazilianPortuguese",
 }};
 
@@ -45,17 +45,17 @@ constexpr std::array<Language, 18> language_to_codes = {{
     Language::German,
     Language::Italian,
     Language::Spanish,
-    Language::Chinese,
+    Language::SimplifiedChinese,
     Language::Korean,
     Language::Dutch,
     Language::Portuguese,
     Language::Russian,
-    Language::Taiwanese,
+    Language::TraditionalChinese,
     Language::BritishEnglish,
     Language::CanadianFrench,
     Language::LatinAmericanSpanish,
-    Language::Chinese,
-    Language::Taiwanese,
+    Language::SimplifiedChinese,
+    Language::TraditionalChinese,
     Language::BrazilianPortuguese,
 }};
 

src/core/file_sys/control_metadata.h

@@ -84,8 +84,8 @@ enum class Language : u8 {
     Portuguese = 10,
     Russian = 11,
     Korean = 12,
-    Taiwanese = 13,
-    Chinese = 14,
+    TraditionalChinese = 13,
+    SimplifiedChinese = 14,
     BrazilianPortuguese = 15,
 
     Default = 255,

src/core/file_sys/patch_manager.cpp

@@ -25,6 +25,8 @@
 #include "core/file_sys/vfs_layered.h"
 #include "core/file_sys/vfs_vector.h"
 #include "core/hle/service/filesystem/filesystem.h"
+#include "core/hle/service/ns/language.h"
+#include "core/hle/service/set/set.h"
 #include "core/loader/loader.h"
 #include "core/loader/nso.h"
 #include "core/memory/cheat_engine.h"
@@ -624,8 +626,37 @@ PatchManager::Metadata PatchManager::ParseControlNCA(const NCA& nca) const {
 
     auto nacp = nacp_file == nullptr ? nullptr : std::make_unique<NACP>(nacp_file);
 
+    // Get language code from settings
+    const auto language_code =
+        Service::Set::GetLanguageCodeFromIndex(Settings::values.language_index.GetValue());
+
+    // Convert to application language and get priority list
+    const auto application_language =
+        Service::NS::ConvertToApplicationLanguage(language_code)
+            .value_or(Service::NS::ApplicationLanguage::AmericanEnglish);
+    const auto language_priority_list =
+        Service::NS::GetApplicationLanguagePriorityList(application_language);
+
+    // Convert to language names
+    auto priority_language_names = FileSys::LANGUAGE_NAMES; // Copy
+    if (language_priority_list) {
+        for (size_t i = 0; i < priority_language_names.size(); ++i) {
+            // Relies on FileSys::LANGUAGE_NAMES being in the same order as
+            // Service::NS::ApplicationLanguage
+            const auto language_index = static_cast<u8>(language_priority_list->at(i));
+
+            if (language_index < FileSys::LANGUAGE_NAMES.size()) {
+                priority_language_names[i] = FileSys::LANGUAGE_NAMES[language_index];
+            } else {
+                // Not a catastrophe, unlikely to happen
+                LOG_WARNING(Loader, "Invalid language index {}", language_index);
+            }
+        }
+    }
+
+    // Get first matching icon
     VirtualFile icon_file;
-    for (const auto& language : FileSys::LANGUAGE_NAMES) {
+    for (const auto& language : priority_language_names) {
         icon_file = extracted->GetFile(std::string("icon_").append(language).append(".dat"));
         if (icon_file != nullptr) {
             break;

src/core/hle/service/nfc/common/device.cpp

@@ -12,6 +12,11 @@
 #pragma warning(pop)
 #endif
 
+#include <fmt/format.h>
+
+#include "common/fs/file.h"
+#include "common/fs/fs.h"
+#include "common/fs/path_util.h"
 #include "common/input.h"
 #include "common/logging/log.h"
 #include "common/string_util.h"
@@ -136,7 +141,7 @@ bool NfcDevice::LoadNfcTag(std::span<const u8> data) {
     if (!NFP::AmiiboCrypto::IsKeyAvailable()) {
         LOG_INFO(Service_NFC, "Loading amiibo without keys");
         memcpy(&encrypted_tag_data, data.data(), sizeof(NFP::EncryptedNTAG215File));
-        BuildAmiiboWithoutKeys();
+        BuildAmiiboWithoutKeys(tag_data, encrypted_tag_data);
         is_plain_amiibo = true;
         is_write_protected = true;
         return true;
@@ -366,16 +371,25 @@ Result NfcDevice::Mount(NFP::ModelType model_type, NFP::MountTarget mount_target
 
     // The loaded amiibo is not encrypted
     if (is_plain_amiibo) {
+        std::vector<u8> data(sizeof(NFP::NTAG215File));
+        memcpy(data.data(), &tag_data, sizeof(tag_data));
+        WriteBackupData(tag_data.uid, data);
+
         device_state = DeviceState::TagMounted;
         mount_target = mount_target_;
         return ResultSuccess;
     }
 
     if (!NFP::AmiiboCrypto::DecodeAmiibo(encrypted_tag_data, tag_data)) {
-        LOG_ERROR(Service_NFP, "Can't decode amiibo {}", device_state);
-        return ResultCorruptedData;
+        bool has_backup = HasBackup(encrypted_tag_data.uuid.uid).IsSuccess();
+        LOG_ERROR(Service_NFP, "Can't decode amiibo, has_backup= {}", has_backup);
+        return has_backup ? ResultCorruptedDataWithBackup : ResultCorruptedData;
     }
 
+    std::vector<u8> data(sizeof(NFP::EncryptedNTAG215File));
+    memcpy(data.data(), &encrypted_tag_data, sizeof(encrypted_tag_data));
+    WriteBackupData(encrypted_tag_data.uuid.uid, data);
+
     device_state = DeviceState::TagMounted;
     mount_target = mount_target_;
     return ResultSuccess;
@@ -470,6 +484,7 @@ Result NfcDevice::FlushWithBreak(NFP::BreakType break_type) {
     std::vector<u8> data(sizeof(NFP::EncryptedNTAG215File));
     if (is_plain_amiibo) {
         memcpy(data.data(), &tag_data, sizeof(tag_data));
+        WriteBackupData(tag_data.uid, data);
     } else {
         if (!NFP::AmiiboCrypto::EncodeAmiibo(tag_data, encrypted_tag_data)) {
             LOG_ERROR(Service_NFP, "Failed to encode data");
@@ -477,6 +492,7 @@ Result NfcDevice::FlushWithBreak(NFP::BreakType break_type) {
         }
 
         memcpy(data.data(), &encrypted_tag_data, sizeof(encrypted_tag_data));
+        WriteBackupData(encrypted_tag_data.uuid.uid, data);
     }
 
     if (!npad_device->WriteNfc(data)) {
@@ -488,7 +504,7 @@ Result NfcDevice::FlushWithBreak(NFP::BreakType break_type) {
 }
 
 Result NfcDevice::Restore() {
-    if (device_state != DeviceState::TagMounted) {
+    if (device_state != DeviceState::TagFound) {
         LOG_ERROR(Service_NFP, "Wrong device state {}", device_state);
         if (device_state == DeviceState::TagRemoved) {
             return ResultTagRemoved;
@@ -496,13 +512,59 @@ Result NfcDevice::Restore() {
         return ResultWrongDeviceState;
     }
 
-    if (mount_target == NFP::MountTarget::None || mount_target == NFP::MountTarget::Rom) {
-        LOG_ERROR(Service_NFC, "Amiibo is read only", device_state);
-        return ResultWrongDeviceState;
+    NFC::TagInfo tag_info{};
+    std::array<u8, sizeof(NFP::EncryptedNTAG215File)> data{};
+    Result result = GetTagInfo(tag_info, false);
+
+    if (result.IsError()) {
+        return result;
     }
 
-    // TODO: Load amiibo from backup on system
-    LOG_ERROR(Service_NFP, "Not Implemented");
+    result = ReadBackupData(tag_info.uuid, data);
+
+    if (result.IsError()) {
+        return result;
+    }
+
+    NFP::NTAG215File temporary_tag_data{};
+    NFP::EncryptedNTAG215File temporary_encrypted_tag_data{};
+
+    // Fallback for encrypted amiibos without keys
+    if (is_write_protected) {
+        return ResultWriteAmiiboFailed;
+    }
+
+    // Fallback for plain amiibos
+    if (is_plain_amiibo) {
+        LOG_INFO(Service_NFP, "Restoring backup of plain amiibo");
+        memcpy(&temporary_tag_data, data.data(), sizeof(NFP::EncryptedNTAG215File));
+        temporary_encrypted_tag_data = NFP::AmiiboCrypto::EncodedDataToNfcData(temporary_tag_data);
+    }
+
+    if (!is_plain_amiibo) {
+        LOG_INFO(Service_NFP, "Restoring backup of encrypted amiibo");
+        temporary_tag_data = {};
+        memcpy(&temporary_encrypted_tag_data, data.data(), sizeof(NFP::EncryptedNTAG215File));
+    }
+
+    if (!NFP::AmiiboCrypto::IsAmiiboValid(temporary_encrypted_tag_data)) {
+        return ResultNotAnAmiibo;
+    }
+
+    if (!is_plain_amiibo) {
+        if (!NFP::AmiiboCrypto::DecodeAmiibo(temporary_encrypted_tag_data, temporary_tag_data)) {
+            LOG_ERROR(Service_NFP, "Can't decode amiibo");
+            return ResultCorruptedData;
+        }
+    }
+
+    // Overwrite tag contents with backup and mount the tag
+    tag_data = temporary_tag_data;
+    encrypted_tag_data = temporary_encrypted_tag_data;
+    device_state = DeviceState::TagMounted;
+    mount_target = NFP::MountTarget::All;
+    is_data_moddified = true;
+
     return ResultSuccess;
 }
 
@@ -1132,13 +1194,69 @@ Result NfcDevice::BreakTag(NFP::BreakType break_type) {
     return FlushWithBreak(break_type);
 }
 
-Result NfcDevice::ReadBackupData(std::span<u8> data) const {
-    // Not implemented
+Result NfcDevice::HasBackup(const NFC::UniqueSerialNumber& uid) const {
+    constexpr auto backup_dir = "backup";
+    const auto yuzu_amiibo_dir = Common::FS::GetYuzuPath(Common::FS::YuzuPath::AmiiboDir);
+    const auto file_name = fmt::format("{0:02x}.bin", fmt::join(uid, ""));
+
+    if (!Common::FS::Exists(yuzu_amiibo_dir / backup_dir / file_name)) {
+        return ResultUnableToAccessBackupFile;
+    }
+
     return ResultSuccess;
 }
 
-Result NfcDevice::WriteBackupData(std::span<const u8> data) {
-    // Not implemented
+Result NfcDevice::ReadBackupData(const NFC::UniqueSerialNumber& uid, std::span<u8> data) const {
+    constexpr auto backup_dir = "backup";
+    const auto yuzu_amiibo_dir = Common::FS::GetYuzuPath(Common::FS::YuzuPath::AmiiboDir);
+    const auto file_name = fmt::format("{0:02x}.bin", fmt::join(uid, ""));
+
+    const Common::FS::IOFile keys_file{yuzu_amiibo_dir / backup_dir / file_name,
+                                       Common::FS::FileAccessMode::Read,
+                                       Common::FS::FileType::BinaryFile};
+
+    if (!keys_file.IsOpen()) {
+        LOG_ERROR(Service_NFP, "Failed to open amiibo backup");
+        return ResultUnableToAccessBackupFile;
+    }
+
+    if (keys_file.Read(data) != data.size()) {
+        LOG_ERROR(Service_NFP, "Failed to read amiibo backup");
+        return ResultUnableToAccessBackupFile;
+    }
+
+    return ResultSuccess;
+}
+
+Result NfcDevice::WriteBackupData(const NFC::UniqueSerialNumber& uid, std::span<const u8> data) {
+    constexpr auto backup_dir = "backup";
+    const auto yuzu_amiibo_dir = Common::FS::GetYuzuPath(Common::FS::YuzuPath::AmiiboDir);
+    const auto file_name = fmt::format("{0:02x}.bin", fmt::join(uid, ""));
+
+    if (HasBackup(uid).IsError()) {
+        if (!Common::FS::CreateDir(yuzu_amiibo_dir / backup_dir)) {
+            return ResultBackupPathAlreadyExist;
+        }
+
+        if (!Common::FS::NewFile(yuzu_amiibo_dir / backup_dir / file_name)) {
+            return ResultBackupPathAlreadyExist;
+        }
+    }
+
+    const Common::FS::IOFile keys_file{yuzu_amiibo_dir / backup_dir / file_name,
+                                       Common::FS::FileAccessMode::ReadWrite,
+                                       Common::FS::FileType::BinaryFile};
+
+    if (!keys_file.IsOpen()) {
+        LOG_ERROR(Service_NFP, "Failed to open amiibo backup");
+        return ResultUnableToAccessBackupFile;
+    }
+
+    if (keys_file.Write(data) != data.size()) {
+        LOG_ERROR(Service_NFP, "Failed to write amiibo backup");
+        return ResultUnableToAccessBackupFile;
+    }
+
     return ResultSuccess;
 }
 
@@ -1177,7 +1295,8 @@ NFP::AmiiboName NfcDevice::GetAmiiboName(const NFP::AmiiboSettings& settings) co
     return amiibo_name;
 }
 
-void NfcDevice::SetAmiiboName(NFP::AmiiboSettings& settings, const NFP::AmiiboName& amiibo_name) {
+void NfcDevice::SetAmiiboName(NFP::AmiiboSettings& settings,
+                              const NFP::AmiiboName& amiibo_name) const {
     std::array<char16_t, NFP::amiibo_name_length> settings_amiibo_name{};
 
     // Convert from utf8 to utf16
@@ -1258,22 +1377,23 @@ void NfcDevice::UpdateRegisterInfoCrc() {
     tag_data.register_info_crc = crc.checksum();
 }
 
-void NfcDevice::BuildAmiiboWithoutKeys() {
+void NfcDevice::BuildAmiiboWithoutKeys(NFP::NTAG215File& stubbed_tag_data,
+                                       const NFP::EncryptedNTAG215File& encrypted_file) const {
     Service::Mii::MiiManager manager;
-    auto& settings = tag_data.settings;
+    auto& settings = stubbed_tag_data.settings;
 
-    tag_data = NFP::AmiiboCrypto::NfcDataToEncodedData(encrypted_tag_data);
+    stubbed_tag_data = NFP::AmiiboCrypto::NfcDataToEncodedData(encrypted_file);
 
     // Common info
-    tag_data.write_counter = 0;
-    tag_data.amiibo_version = 0;
+    stubbed_tag_data.write_counter = 0;
+    stubbed_tag_data.amiibo_version = 0;
     settings.write_date = GetAmiiboDate(GetCurrentPosixTime());
 
     // Register info
     SetAmiiboName(settings, {'y', 'u', 'z', 'u', 'A', 'm', 'i', 'i', 'b', 'o'});
     settings.settings.font_region.Assign(0);
     settings.init_date = GetAmiiboDate(GetCurrentPosixTime());
-    tag_data.owner_mii = manager.BuildFromStoreData(manager.BuildDefault(0));
+    stubbed_tag_data.owner_mii = manager.BuildFromStoreData(manager.BuildDefault(0));
 
     // Admin info
     settings.settings.amiibo_initialized.Assign(1);

src/core/hle/service/nfc/common/device.h

@@ -86,8 +86,9 @@ public:
     Result GetAll(NFP::NfpData& data) const;
     Result SetAll(const NFP::NfpData& data);
     Result BreakTag(NFP::BreakType break_type);
-    Result ReadBackupData(std::span<u8> data) const;
-    Result WriteBackupData(std::span<const u8> data);
+    Result HasBackup(const NFC::UniqueSerialNumber& uid) const;
+    Result ReadBackupData(const NFC::UniqueSerialNumber& uid, std::span<u8> data) const;
+    Result WriteBackupData(const NFC::UniqueSerialNumber& uid, std::span<const u8> data);
     Result WriteNtf(std::span<const u8> data);
 
     u64 GetHandle() const;
@@ -103,14 +104,15 @@ private:
     void CloseNfcTag();
 
     NFP::AmiiboName GetAmiiboName(const NFP::AmiiboSettings& settings) const;
-    void SetAmiiboName(NFP::AmiiboSettings& settings, const NFP::AmiiboName& amiibo_name);
+    void SetAmiiboName(NFP::AmiiboSettings& settings, const NFP::AmiiboName& amiibo_name) const;
     NFP::AmiiboDate GetAmiiboDate(s64 posix_time) const;
     u64 GetCurrentPosixTime() const;
     u64 RemoveVersionByte(u64 application_id) const;
     void UpdateSettingsCrc();
     void UpdateRegisterInfoCrc();
 
-    void BuildAmiiboWithoutKeys();
+    void BuildAmiiboWithoutKeys(NFP::NTAG215File& stubbed_tag_data,
+                                const NFP::EncryptedNTAG215File& encrypted_file) const;
 
     bool is_controller_set{};
     int callback_key;

src/core/hle/service/nfc/common/device_manager.cpp

@@ -543,9 +543,14 @@ Result DeviceManager::ReadBackupData(u64 device_handle, std::span<u8> data) cons
 
     std::shared_ptr<NfcDevice> device = nullptr;
     auto result = GetDeviceHandle(device_handle, device);
+    NFC::TagInfo tag_info{};
 
     if (result.IsSuccess()) {
-        result = device->ReadBackupData(data);
+        result = device->GetTagInfo(tag_info, false);
+    }
+
+    if (result.IsSuccess()) {
+        result = device->ReadBackupData(tag_info.uuid, data);
         result = VerifyDeviceResult(device, result);
     }
 
@@ -557,9 +562,14 @@ Result DeviceManager::WriteBackupData(u64 device_handle, std::span<const u8> dat
 
     std::shared_ptr<NfcDevice> device = nullptr;
     auto result = GetDeviceHandle(device_handle, device);
+    NFC::TagInfo tag_info{};
 
     if (result.IsSuccess()) {
-        result = device->WriteBackupData(data);
+        result = device->GetTagInfo(tag_info, false);
+    }
+
+    if (result.IsSuccess()) {
+        result = device->WriteBackupData(tag_info.uuid, data);
         result = VerifyDeviceResult(device, result);
     }
 

src/core/hle/service/nfc/nfc_interface.cpp

@@ -302,7 +302,7 @@ Result NfcInterface::TranslateResultToServiceError(Result result) const {
         return TranslateResultToNfp(result);
     }
     default:
-        if (result != ResultUnknown216) {
+        if (result != ResultBackupPathAlreadyExist) {
             return result;
         }
         return ResultUnknown74;
@@ -343,6 +343,9 @@ Result NfcInterface::TranslateResultToNfp(Result result) const {
     if (result == ResultApplicationAreaIsNotInitialized) {
         return NFP::ResultApplicationAreaIsNotInitialized;
     }
+    if (result == ResultCorruptedDataWithBackup) {
+        return NFP::ResultCorruptedDataWithBackup;
+    }
     if (result == ResultCorruptedData) {
         return NFP::ResultCorruptedData;
     }
@@ -355,6 +358,9 @@ Result NfcInterface::TranslateResultToNfp(Result result) const {
     if (result == ResultNotAnAmiibo) {
         return NFP::ResultNotAnAmiibo;
     }
+    if (result == ResultUnableToAccessBackupFile) {
+        return NFP::ResultUnableToAccessBackupFile;
+    }
     LOG_WARNING(Service_NFC, "Result conversion not handled");
     return result;
 }

src/core/hle/service/nfc/nfc_result.h

@@ -9,20 +9,22 @@ namespace Service::NFC {
 
 constexpr Result ResultDeviceNotFound(ErrorModule::NFC, 64);
 constexpr Result ResultInvalidArgument(ErrorModule::NFC, 65);
-constexpr Result ResultWrongApplicationAreaSize(ErrorModule::NFP, 68);
+constexpr Result ResultWrongApplicationAreaSize(ErrorModule::NFC, 68);
 constexpr Result ResultWrongDeviceState(ErrorModule::NFC, 73);
 constexpr Result ResultUnknown74(ErrorModule::NFC, 74);
 constexpr Result ResultUnknown76(ErrorModule::NFC, 76);
 constexpr Result ResultNfcNotInitialized(ErrorModule::NFC, 77);
 constexpr Result ResultNfcDisabled(ErrorModule::NFC, 80);
-constexpr Result ResultWriteAmiiboFailed(ErrorModule::NFP, 88);
+constexpr Result ResultWriteAmiiboFailed(ErrorModule::NFC, 88);
 constexpr Result ResultTagRemoved(ErrorModule::NFC, 97);
-constexpr Result ResultRegistrationIsNotInitialized(ErrorModule::NFP, 120);
-constexpr Result ResultApplicationAreaIsNotInitialized(ErrorModule::NFP, 128);
-constexpr Result ResultCorruptedData(ErrorModule::NFP, 144);
-constexpr Result ResultWrongApplicationAreaId(ErrorModule::NFP, 152);
-constexpr Result ResultApplicationAreaExist(ErrorModule::NFP, 168);
-constexpr Result ResultNotAnAmiibo(ErrorModule::NFP, 178);
-constexpr Result ResultUnknown216(ErrorModule::NFC, 216);
+constexpr Result ResultUnableToAccessBackupFile(ErrorModule::NFC, 113);
+constexpr Result ResultRegistrationIsNotInitialized(ErrorModule::NFC, 120);
+constexpr Result ResultApplicationAreaIsNotInitialized(ErrorModule::NFC, 128);
+constexpr Result ResultCorruptedDataWithBackup(ErrorModule::NFC, 136);
+constexpr Result ResultCorruptedData(ErrorModule::NFC, 144);
+constexpr Result ResultWrongApplicationAreaId(ErrorModule::NFC, 152);
+constexpr Result ResultApplicationAreaExist(ErrorModule::NFC, 168);
+constexpr Result ResultNotAnAmiibo(ErrorModule::NFC, 178);
+constexpr Result ResultBackupPathAlreadyExist(ErrorModule::NFC, 216);
 
 } // namespace Service::NFC

src/core/hle/service/nfp/nfp_interface.cpp

@@ -126,7 +126,7 @@ void Interface::Flush(HLERequestContext& ctx) {
 void Interface::Restore(HLERequestContext& ctx) {
     IPC::RequestParser rp{ctx};
     const auto device_handle{rp.Pop<u64>()};
-    LOG_WARNING(Service_NFP, "(STUBBED) called, device_handle={}", device_handle);
+    LOG_INFO(Service_NFP, "called, device_handle={}", device_handle);
 
     auto result = GetManager()->Restore(device_handle);
     result = TranslateResultToServiceError(result);
@@ -394,7 +394,7 @@ void Interface::BreakTag(HLERequestContext& ctx) {
 void Interface::ReadBackupData(HLERequestContext& ctx) {
     IPC::RequestParser rp{ctx};
     const auto device_handle{rp.Pop<u64>()};
-    LOG_WARNING(Service_NFP, "(STUBBED) called, device_handle={}", device_handle);
+    LOG_INFO(Service_NFP, "called, device_handle={}", device_handle);
 
     std::vector<u8> backup_data{};
     auto result = GetManager()->ReadBackupData(device_handle, backup_data);
@@ -412,7 +412,7 @@ void Interface::WriteBackupData(HLERequestContext& ctx) {
     IPC::RequestParser rp{ctx};
     const auto device_handle{rp.Pop<u64>()};
     const auto backup_data_buffer{ctx.ReadBuffer()};
-    LOG_WARNING(Service_NFP, "(STUBBED) called, device_handle={}", device_handle);
+    LOG_INFO(Service_NFP, "called, device_handle={}", device_handle);
 
     auto result = GetManager()->WriteBackupData(device_handle, backup_data_buffer);
     result = TranslateResultToServiceError(result);

src/core/hle/service/nfp/nfp_result.h

@@ -17,9 +17,11 @@ constexpr Result ResultWriteAmiiboFailed(ErrorModule::NFP, 88);
 constexpr Result ResultTagRemoved(ErrorModule::NFP, 97);
 constexpr Result ResultRegistrationIsNotInitialized(ErrorModule::NFP, 120);
 constexpr Result ResultApplicationAreaIsNotInitialized(ErrorModule::NFP, 128);
+constexpr Result ResultCorruptedDataWithBackup(ErrorModule::NFP, 136);
 constexpr Result ResultCorruptedData(ErrorModule::NFP, 144);
 constexpr Result ResultWrongApplicationAreaId(ErrorModule::NFP, 152);
 constexpr Result ResultApplicationAreaExist(ErrorModule::NFP, 168);
 constexpr Result ResultNotAnAmiibo(ErrorModule::NFP, 178);
+constexpr Result ResultUnableToAccessBackupFile(ErrorModule::NFP, 200);
 
 } // namespace Service::NFP

src/core/hle/service/nvdrv/devices/nvhost_nvdec.cpp

@@ -69,7 +69,7 @@ NvResult nvhost_nvdec::Ioctl3(DeviceFD fd, Ioctl command, std::span<const u8> in
 
 void nvhost_nvdec::OnOpen(DeviceFD fd) {
     LOG_INFO(Service_NVDRV, "NVDEC video stream started");
-    system.AudioCore().SetNVDECActive(true);
+    system.SetNVDECActive(true);
 }
 
 void nvhost_nvdec::OnClose(DeviceFD fd) {
@@ -79,7 +79,7 @@ void nvhost_nvdec::OnClose(DeviceFD fd) {
     if (iter != host1x_file.fd_to_id.end()) {
         system.GPU().ClearCdmaInstance(iter->second);
     }
-    system.AudioCore().SetNVDECActive(false);
+    system.SetNVDECActive(false);
 }
 
 } // namespace Service::Nvidia::Devices

src/core/hle/service/nvnflinger/nvnflinger.cpp

@@ -324,6 +324,10 @@ s64 Nvnflinger::GetNextTicks() const {
             speed_scale = 0.01f;
         }
     }
+    if (system.GetNVDECActive() && settings.use_video_framerate.GetValue()) {
+        // Run at intended presentation rate during video playback.
+        speed_scale = 1.f;
+    }
 
     // As an extension, treat nonpositive swap interval as framerate multiplier.
     const f32 effective_fps = swap_interval <= 0 ? 120.f * static_cast<f32>(1 - swap_interval)

src/video_core/CMakeLists.txt

@@ -133,8 +133,8 @@ add_library(video_core STATIC
     renderer_opengl/gl_shader_util.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_staging_buffer_pool.cpp
+    renderer_opengl/gl_staging_buffer_pool.h
     renderer_opengl/gl_texture_cache.cpp
     renderer_opengl/gl_texture_cache.h
     renderer_opengl/gl_texture_cache_base.cpp

src/video_core/buffer_cache/buffer_cache.h

@@ -478,7 +478,6 @@ void BufferCache<P>::CommitAsyncFlushesHigh() {
 
     if (committed_ranges.empty()) {
         if constexpr (IMPLEMENTS_ASYNC_DOWNLOADS) {
-
             async_buffers.emplace_back(std::optional<Async_Buffer>{});
         }
         return;
@@ -539,7 +538,6 @@ void BufferCache<P>::CommitAsyncFlushesHigh() {
     committed_ranges.clear();
     if (downloads.empty()) {
         if constexpr (IMPLEMENTS_ASYNC_DOWNLOADS) {
-
             async_buffers.emplace_back(std::optional<Async_Buffer>{});
         }
         return;
@@ -691,7 +689,7 @@ void BufferCache<P>::BindHostIndexBuffer() {
     const u32 size = channel_state->index_buffer.size;
     const auto& draw_state = maxwell3d->draw_manager->GetDrawState();
     if (!draw_state.inline_index_draw_indexes.empty()) [[unlikely]] {
-        if constexpr (USE_MEMORY_MAPS) {
+        if constexpr (USE_MEMORY_MAPS_FOR_UPLOADS) {
             auto upload_staging = runtime.UploadStagingBuffer(size);
             std::array<BufferCopy, 1> copies{
                 {BufferCopy{.src_offset = upload_staging.offset, .dst_offset = 0, .size = size}}};
@@ -1462,7 +1460,7 @@ bool BufferCache<P>::SynchronizeBufferNoModified(Buffer& buffer, VAddr cpu_addr,
 template <class P>
 void BufferCache<P>::UploadMemory(Buffer& buffer, u64 total_size_bytes, u64 largest_copy,
                                   std::span<BufferCopy> copies) {
-    if constexpr (USE_MEMORY_MAPS) {
+    if constexpr (USE_MEMORY_MAPS_FOR_UPLOADS) {
         MappedUploadMemory(buffer, total_size_bytes, copies);
     } else {
         ImmediateUploadMemory(buffer, largest_copy, copies);
@@ -1473,7 +1471,7 @@ template <class P>
 void BufferCache<P>::ImmediateUploadMemory([[maybe_unused]] Buffer& buffer,
                                            [[maybe_unused]] u64 largest_copy,
                                            [[maybe_unused]] std::span<const BufferCopy> copies) {
-    if constexpr (!USE_MEMORY_MAPS) {
+    if constexpr (!USE_MEMORY_MAPS_FOR_UPLOADS) {
         std::span<u8> immediate_buffer;
         for (const BufferCopy& copy : copies) {
             std::span<const u8> upload_span;
@@ -1532,7 +1530,7 @@ bool BufferCache<P>::InlineMemory(VAddr dest_address, size_t copy_size,
     auto& buffer = slot_buffers[buffer_id];
     SynchronizeBuffer(buffer, dest_address, static_cast<u32>(copy_size));
 
-    if constexpr (USE_MEMORY_MAPS) {
+    if constexpr (USE_MEMORY_MAPS_FOR_UPLOADS) {
         auto upload_staging = runtime.UploadStagingBuffer(copy_size);
         std::array copies{BufferCopy{
             .src_offset = upload_staging.offset,

src/video_core/buffer_cache/buffer_cache_base.h

@@ -173,6 +173,7 @@ class BufferCache : public VideoCommon::ChannelSetupCaches<BufferCacheChannelInf
     static constexpr bool USE_MEMORY_MAPS = P::USE_MEMORY_MAPS;
     static constexpr bool SEPARATE_IMAGE_BUFFERS_BINDINGS = P::SEPARATE_IMAGE_BUFFER_BINDINGS;
     static constexpr bool IMPLEMENTS_ASYNC_DOWNLOADS = P::IMPLEMENTS_ASYNC_DOWNLOADS;
+    static constexpr bool USE_MEMORY_MAPS_FOR_UPLOADS = P::USE_MEMORY_MAPS_FOR_UPLOADS;
 
     static constexpr s64 DEFAULT_EXPECTED_MEMORY = 512_MiB;
     static constexpr s64 DEFAULT_CRITICAL_MEMORY = 1_GiB;

src/video_core/renderer_opengl/gl_buffer_cache.cpp

@@ -106,8 +106,10 @@ GLuint Buffer::View(u32 offset, u32 size, PixelFormat format) {
     return views.back().texture.handle;
 }
 
-BufferCacheRuntime::BufferCacheRuntime(const Device& device_)
-    : device{device_}, has_fast_buffer_sub_data{device.HasFastBufferSubData()},
+BufferCacheRuntime::BufferCacheRuntime(const Device& device_,
+                                       StagingBufferPool& staging_buffer_pool_)
+    : device{device_}, staging_buffer_pool{staging_buffer_pool_},
+      has_fast_buffer_sub_data{device.HasFastBufferSubData()},
       use_assembly_shaders{device.UseAssemblyShaders()},
       has_unified_vertex_buffers{device.HasVertexBufferUnifiedMemory()},
       stream_buffer{has_fast_buffer_sub_data ? std::nullopt : std::make_optional<StreamBuffer>()} {
@@ -140,6 +142,14 @@ BufferCacheRuntime::BufferCacheRuntime(const Device& device_)
     }();
 }
 
+StagingBufferMap BufferCacheRuntime::UploadStagingBuffer(size_t size) {
+    return staging_buffer_pool.RequestUploadBuffer(size);
+}
+
+StagingBufferMap BufferCacheRuntime::DownloadStagingBuffer(size_t size) {
+    return staging_buffer_pool.RequestDownloadBuffer(size);
+}
+
 u64 BufferCacheRuntime::GetDeviceMemoryUsage() const {
     if (device.CanReportMemoryUsage()) {
         return device_access_memory - device.GetCurrentDedicatedVideoMemory();
@@ -147,13 +157,47 @@ u64 BufferCacheRuntime::GetDeviceMemoryUsage() const {
     return 2_GiB;
 }
 
+void BufferCacheRuntime::CopyBuffer(GLuint dst_buffer, GLuint src_buffer,
+                                    std::span<const VideoCommon::BufferCopy> copies, bool barrier) {
+    if (barrier) {
+        PreCopyBarrier();
+    }
+    for (const VideoCommon::BufferCopy& copy : copies) {
+        glCopyNamedBufferSubData(src_buffer, dst_buffer, static_cast<GLintptr>(copy.src_offset),
+                                 static_cast<GLintptr>(copy.dst_offset),
+                                 static_cast<GLsizeiptr>(copy.size));
+    }
+    if (barrier) {
+        PostCopyBarrier();
+    }
+}
+
+void BufferCacheRuntime::CopyBuffer(GLuint dst_buffer, Buffer& src_buffer,
+                                    std::span<const VideoCommon::BufferCopy> copies, bool barrier) {
+    CopyBuffer(dst_buffer, src_buffer.Handle(), copies, barrier);
+}
+
+void BufferCacheRuntime::CopyBuffer(Buffer& dst_buffer, GLuint src_buffer,
+                                    std::span<const VideoCommon::BufferCopy> copies, bool barrier) {
+    CopyBuffer(dst_buffer.Handle(), src_buffer, copies, barrier);
+}
+
 void BufferCacheRuntime::CopyBuffer(Buffer& dst_buffer, Buffer& src_buffer,
                                     std::span<const VideoCommon::BufferCopy> copies) {
-    for (const VideoCommon::BufferCopy& copy : copies) {
-        glCopyNamedBufferSubData(
-            src_buffer.Handle(), dst_buffer.Handle(), static_cast<GLintptr>(copy.src_offset),
-            static_cast<GLintptr>(copy.dst_offset), static_cast<GLsizeiptr>(copy.size));
-    }
+    CopyBuffer(dst_buffer.Handle(), src_buffer.Handle(), copies);
+}
+
+void BufferCacheRuntime::PreCopyBarrier() {
+    // TODO: finer grained barrier?
+    glMemoryBarrier(GL_ALL_BARRIER_BITS);
+}
+
+void BufferCacheRuntime::PostCopyBarrier() {
+    glMemoryBarrier(GL_BUFFER_UPDATE_BARRIER_BIT | GL_CLIENT_MAPPED_BUFFER_BARRIER_BIT);
+}
+
+void BufferCacheRuntime::Finish() {
+    glFinish();
 }
 
 void BufferCacheRuntime::ClearBuffer(Buffer& dest_buffer, u32 offset, size_t size, u32 value) {

src/video_core/renderer_opengl/gl_buffer_cache.h

@@ -12,7 +12,7 @@
 #include "video_core/rasterizer_interface.h"
 #include "video_core/renderer_opengl/gl_device.h"
 #include "video_core/renderer_opengl/gl_resource_manager.h"
-#include "video_core/renderer_opengl/gl_stream_buffer.h"
+#include "video_core/renderer_opengl/gl_staging_buffer_pool.h"
 
 namespace OpenGL {
 
@@ -60,11 +60,28 @@ class BufferCacheRuntime {
 public:
     static constexpr u8 INVALID_BINDING = std::numeric_limits<u8>::max();
 
-    explicit BufferCacheRuntime(const Device& device_);
+    explicit BufferCacheRuntime(const Device& device_, StagingBufferPool& staging_buffer_pool_);
+
+    [[nodiscard]] StagingBufferMap UploadStagingBuffer(size_t size);
+
+    [[nodiscard]] StagingBufferMap DownloadStagingBuffer(size_t size);
+
+    void CopyBuffer(GLuint dst_buffer, GLuint src_buffer,
+                    std::span<const VideoCommon::BufferCopy> copies, bool barrier = true);
+
+    void CopyBuffer(GLuint dst_buffer, Buffer& src_buffer,
+                    std::span<const VideoCommon::BufferCopy> copies, bool barrier = true);
+
+    void CopyBuffer(Buffer& dst_buffer, GLuint src_buffer,
+                    std::span<const VideoCommon::BufferCopy> copies, bool barrier = true);
 
     void CopyBuffer(Buffer& dst_buffer, Buffer& src_buffer,
                     std::span<const VideoCommon::BufferCopy> copies);
 
+    void PreCopyBarrier();
+    void PostCopyBarrier();
+    void Finish();
+
     void ClearBuffer(Buffer& dest_buffer, u32 offset, size_t size, u32 value);
 
     void BindIndexBuffer(Buffer& buffer, u32 offset, u32 size);
@@ -169,6 +186,7 @@ private:
     };
 
     const Device& device;
+    StagingBufferPool& staging_buffer_pool;
 
     bool has_fast_buffer_sub_data = false;
     bool use_assembly_shaders = false;
@@ -201,7 +219,7 @@ private:
 struct BufferCacheParams {
     using Runtime = OpenGL::BufferCacheRuntime;
     using Buffer = OpenGL::Buffer;
-    using Async_Buffer = u32;
+    using Async_Buffer = OpenGL::StagingBufferMap;
     using MemoryTracker = VideoCommon::MemoryTrackerBase<VideoCore::RasterizerInterface>;
 
     static constexpr bool IS_OPENGL = true;
@@ -209,9 +227,12 @@ struct BufferCacheParams {
     static constexpr bool HAS_FULL_INDEX_AND_PRIMITIVE_SUPPORT = true;
     static constexpr bool NEEDS_BIND_UNIFORM_INDEX = true;
     static constexpr bool NEEDS_BIND_STORAGE_INDEX = true;
-    static constexpr bool USE_MEMORY_MAPS = false;
+    static constexpr bool USE_MEMORY_MAPS = true;
     static constexpr bool SEPARATE_IMAGE_BUFFER_BINDINGS = true;
     static constexpr bool IMPLEMENTS_ASYNC_DOWNLOADS = false;
+
+    // TODO: Investigate why OpenGL seems to perform worse with persistently mapped buffer uploads
+    static constexpr bool USE_MEMORY_MAPS_FOR_UPLOADS = false;
 };
 
 using BufferCache = VideoCommon::BufferCache<BufferCacheParams>;

src/video_core/renderer_opengl/gl_rasterizer.cpp

@@ -24,6 +24,7 @@
 #include "video_core/renderer_opengl/gl_query_cache.h"
 #include "video_core/renderer_opengl/gl_rasterizer.h"
 #include "video_core/renderer_opengl/gl_shader_cache.h"
+#include "video_core/renderer_opengl/gl_staging_buffer_pool.h"
 #include "video_core/renderer_opengl/gl_texture_cache.h"
 #include "video_core/renderer_opengl/maxwell_to_gl.h"
 #include "video_core/renderer_opengl/renderer_opengl.h"
@@ -58,8 +59,9 @@ RasterizerOpenGL::RasterizerOpenGL(Core::Frontend::EmuWindow& emu_window_, Tegra
                                    StateTracker& state_tracker_)
     : RasterizerAccelerated(cpu_memory_), gpu(gpu_), device(device_), screen_info(screen_info_),
       program_manager(program_manager_), state_tracker(state_tracker_),
-      texture_cache_runtime(device, program_manager, state_tracker),
-      texture_cache(texture_cache_runtime, *this), buffer_cache_runtime(device),
+      texture_cache_runtime(device, program_manager, state_tracker, staging_buffer_pool),
+      texture_cache(texture_cache_runtime, *this),
+      buffer_cache_runtime(device, staging_buffer_pool),
       buffer_cache(*this, cpu_memory_, buffer_cache_runtime),
       shader_cache(*this, emu_window_, device, texture_cache, buffer_cache, program_manager,
                    state_tracker, gpu.ShaderNotify()),

src/video_core/renderer_opengl/gl_rasterizer.h

@@ -230,6 +230,7 @@ private:
     ProgramManager& program_manager;
     StateTracker& state_tracker;
 
+    StagingBufferPool staging_buffer_pool;
     TextureCacheRuntime texture_cache_runtime;
     TextureCache texture_cache;
     BufferCacheRuntime buffer_cache_runtime;

src/video_core/renderer_opengl/gl_staging_buffer_pool.cpp

@@ -0,0 +1,150 @@
+// SPDX-FileCopyrightText: Copyright 2021 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include <array>
+#include <memory>
+#include <span>
+
+#include <glad/glad.h>
+
+#include "common/alignment.h"
+#include "common/assert.h"
+#include "common/bit_util.h"
+#include "common/microprofile.h"
+#include "video_core/renderer_opengl/gl_staging_buffer_pool.h"
+
+MICROPROFILE_DEFINE(OpenGL_BufferRequest, "OpenGL", "BufferRequest", MP_RGB(128, 128, 192));
+
+namespace OpenGL {
+
+StagingBufferMap::~StagingBufferMap() {
+    if (sync) {
+        sync->Create();
+    }
+}
+
+StagingBuffers::StagingBuffers(GLenum storage_flags_, GLenum map_flags_)
+    : storage_flags{storage_flags_}, map_flags{map_flags_} {}
+
+StagingBuffers::~StagingBuffers() = default;
+
+StagingBufferMap StagingBuffers::RequestMap(size_t requested_size, bool insert_fence) {
+    MICROPROFILE_SCOPE(OpenGL_BufferRequest);
+
+    const size_t index = RequestBuffer(requested_size);
+    OGLSync* const sync = insert_fence ? &syncs[index] : nullptr;
+    sync_indices[index] = insert_fence ? ++current_sync_index : 0;
+    return StagingBufferMap{
+        .mapped_span = std::span(maps[index], requested_size),
+        .sync = sync,
+        .buffer = buffers[index].handle,
+    };
+}
+
+size_t StagingBuffers::RequestBuffer(size_t requested_size) {
+    if (const std::optional<size_t> index = FindBuffer(requested_size); index) {
+        return *index;
+    }
+
+    OGLBuffer& buffer = buffers.emplace_back();
+    buffer.Create();
+    const auto next_pow2_size = Common::NextPow2(requested_size);
+    glNamedBufferStorage(buffer.handle, next_pow2_size, nullptr,
+                         storage_flags | GL_MAP_PERSISTENT_BIT);
+    maps.push_back(static_cast<u8*>(glMapNamedBufferRange(buffer.handle, 0, next_pow2_size,
+                                                          map_flags | GL_MAP_PERSISTENT_BIT)));
+    syncs.emplace_back();
+    sync_indices.emplace_back();
+    sizes.push_back(next_pow2_size);
+
+    ASSERT(syncs.size() == buffers.size() && buffers.size() == maps.size() &&
+           maps.size() == sizes.size());
+
+    return buffers.size() - 1;
+}
+
+std::optional<size_t> StagingBuffers::FindBuffer(size_t requested_size) {
+    size_t known_unsignaled_index = current_sync_index + 1;
+    size_t smallest_buffer = std::numeric_limits<size_t>::max();
+    std::optional<size_t> found;
+    const size_t num_buffers = sizes.size();
+    for (size_t index = 0; index < num_buffers; ++index) {
+        const size_t buffer_size = sizes[index];
+        if (buffer_size < requested_size || buffer_size >= smallest_buffer) {
+            continue;
+        }
+        if (syncs[index].handle != 0) {
+            if (sync_indices[index] >= known_unsignaled_index) {
+                // This fence is later than a fence that is known to not be signaled
+                continue;
+            }
+            if (!syncs[index].IsSignaled()) {
+                // Since this fence hasn't been signaled, it's safe to assume all later
+                // fences haven't been signaled either
+                known_unsignaled_index = std::min(known_unsignaled_index, sync_indices[index]);
+                continue;
+            }
+            syncs[index].Release();
+        }
+        smallest_buffer = buffer_size;
+        found = index;
+    }
+    return found;
+}
+
+StreamBuffer::StreamBuffer() {
+    static constexpr GLenum flags = GL_MAP_WRITE_BIT | GL_MAP_PERSISTENT_BIT | GL_MAP_COHERENT_BIT;
+    buffer.Create();
+    glObjectLabel(GL_BUFFER, buffer.handle, -1, "Stream Buffer");
+    glNamedBufferStorage(buffer.handle, STREAM_BUFFER_SIZE, nullptr, flags);
+    mapped_pointer =
+        static_cast<u8*>(glMapNamedBufferRange(buffer.handle, 0, STREAM_BUFFER_SIZE, flags));
+    for (OGLSync& sync : fences) {
+        sync.Create();
+    }
+}
+
+std::pair<std::span<u8>, size_t> StreamBuffer::Request(size_t size) noexcept {
+    ASSERT(size < REGION_SIZE);
+    for (size_t region = Region(used_iterator), region_end = Region(iterator); region < region_end;
+         ++region) {
+        fences[region].Create();
+    }
+    used_iterator = iterator;
+
+    for (size_t region = Region(free_iterator) + 1,
+                region_end = std::min(Region(iterator + size) + 1, NUM_SYNCS);
+         region < region_end; ++region) {
+        glClientWaitSync(fences[region].handle, 0, GL_TIMEOUT_IGNORED);
+        fences[region].Release();
+    }
+    if (iterator + size >= free_iterator) {
+        free_iterator = iterator + size;
+    }
+    if (iterator + size > STREAM_BUFFER_SIZE) {
+        for (size_t region = Region(used_iterator); region < NUM_SYNCS; ++region) {
+            fences[region].Create();
+        }
+        used_iterator = 0;
+        iterator = 0;
+        free_iterator = size;
+
+        for (size_t region = 0, region_end = Region(size); region <= region_end; ++region) {
+            glClientWaitSync(fences[region].handle, 0, GL_TIMEOUT_IGNORED);
+            fences[region].Release();
+        }
+    }
+    const size_t offset = iterator;
+    iterator = Common::AlignUp(iterator + size, MAX_ALIGNMENT);
+    return {std::span(mapped_pointer + offset, size), offset};
+}
+
+StagingBufferMap StagingBufferPool::RequestUploadBuffer(size_t size) {
+    return upload_buffers.RequestMap(size, true);
+}
+
+StagingBufferMap StagingBufferPool::RequestDownloadBuffer(size_t size) {
+    return download_buffers.RequestMap(size, false);
+}
+
+} // namespace OpenGL

src/video_core/renderer_opengl/gl_staging_buffer_pool.h

@@ -4,8 +4,10 @@
 #pragma once
 
 #include <array>
+#include <optional>
 #include <span>
 #include <utility>
+#include <vector>
 
 #include <glad/glad.h>
 
@@ -17,6 +19,35 @@ namespace OpenGL {
 
 using namespace Common::Literals;
 
+struct StagingBufferMap {
+    ~StagingBufferMap();
+
+    std::span<u8> mapped_span;
+    size_t offset = 0;
+    OGLSync* sync;
+    GLuint buffer;
+};
+
+struct StagingBuffers {
+    explicit StagingBuffers(GLenum storage_flags_, GLenum map_flags_);
+    ~StagingBuffers();
+
+    StagingBufferMap RequestMap(size_t requested_size, bool insert_fence);
+
+    size_t RequestBuffer(size_t requested_size);
+
+    std::optional<size_t> FindBuffer(size_t requested_size);
+
+    std::vector<OGLSync> syncs;
+    std::vector<OGLBuffer> buffers;
+    std::vector<u8*> maps;
+    std::vector<size_t> sizes;
+    std::vector<size_t> sync_indices;
+    GLenum storage_flags;
+    GLenum map_flags;
+    size_t current_sync_index = 0;
+};
+
 class StreamBuffer {
     static constexpr size_t STREAM_BUFFER_SIZE = 64_MiB;
     static constexpr size_t NUM_SYNCS = 16;
@@ -48,4 +79,17 @@ private:
     std::array<OGLSync, NUM_SYNCS> fences;
 };
 
+class StagingBufferPool {
+public:
+    StagingBufferPool() = default;
+    ~StagingBufferPool() = default;
+
+    StagingBufferMap RequestUploadBuffer(size_t size);
+    StagingBufferMap RequestDownloadBuffer(size_t size);
+
+private:
+    StagingBuffers upload_buffers{GL_MAP_WRITE_BIT, GL_MAP_WRITE_BIT | GL_MAP_FLUSH_EXPLICIT_BIT};
+    StagingBuffers download_buffers{GL_MAP_READ_BIT | GL_CLIENT_STORAGE_BIT, GL_MAP_READ_BIT};
+};
+
 } // namespace OpenGL

src/video_core/renderer_opengl/gl_stream_buffer.cpp

@@ -1,63 +0,0 @@
-// SPDX-FileCopyrightText: Copyright 2021 yuzu Emulator Project
-// SPDX-License-Identifier: GPL-2.0-or-later
-
-#include <array>
-#include <memory>
-#include <span>
-
-#include <glad/glad.h>
-
-#include "common/alignment.h"
-#include "common/assert.h"
-#include "video_core/renderer_opengl/gl_stream_buffer.h"
-
-namespace OpenGL {
-
-StreamBuffer::StreamBuffer() {
-    static constexpr GLenum flags = GL_MAP_WRITE_BIT | GL_MAP_PERSISTENT_BIT | GL_MAP_COHERENT_BIT;
-    buffer.Create();
-    glObjectLabel(GL_BUFFER, buffer.handle, -1, "Stream Buffer");
-    glNamedBufferStorage(buffer.handle, STREAM_BUFFER_SIZE, nullptr, flags);
-    mapped_pointer =
-        static_cast<u8*>(glMapNamedBufferRange(buffer.handle, 0, STREAM_BUFFER_SIZE, flags));
-    for (OGLSync& sync : fences) {
-        sync.Create();
-    }
-}
-
-std::pair<std::span<u8>, size_t> StreamBuffer::Request(size_t size) noexcept {
-    ASSERT(size < REGION_SIZE);
-    for (size_t region = Region(used_iterator), region_end = Region(iterator); region < region_end;
-         ++region) {
-        fences[region].Create();
-    }
-    used_iterator = iterator;
-
-    for (size_t region = Region(free_iterator) + 1,
-                region_end = std::min(Region(iterator + size) + 1, NUM_SYNCS);
-         region < region_end; ++region) {
-        glClientWaitSync(fences[region].handle, 0, GL_TIMEOUT_IGNORED);
-        fences[region].Release();
-    }
-    if (iterator + size >= free_iterator) {
-        free_iterator = iterator + size;
-    }
-    if (iterator + size > STREAM_BUFFER_SIZE) {
-        for (size_t region = Region(used_iterator); region < NUM_SYNCS; ++region) {
-            fences[region].Create();
-        }
-        used_iterator = 0;
-        iterator = 0;
-        free_iterator = size;
-
-        for (size_t region = 0, region_end = Region(size); region <= region_end; ++region) {
-            glClientWaitSync(fences[region].handle, 0, GL_TIMEOUT_IGNORED);
-            fences[region].Release();
-        }
-    }
-    const size_t offset = iterator;
-    iterator = Common::AlignUp(iterator + size, MAX_ALIGNMENT);
-    return {std::span(mapped_pointer + offset, size), offset};
-}
-
-} // namespace OpenGL

src/video_core/renderer_opengl/gl_texture_cache.cpp

@@ -456,19 +456,14 @@ OGLTexture MakeImage(const VideoCommon::ImageInfo& info, GLenum gl_internal_form
         return is_srgb ? GL_SRGB8_ALPHA8 : GL_RGBA8;
     }
 }
-
 } // Anonymous namespace
 
-ImageBufferMap::~ImageBufferMap() {
-    if (sync) {
-        sync->Create();
-    }
-}
-
 TextureCacheRuntime::TextureCacheRuntime(const Device& device_, ProgramManager& program_manager,
-                                         StateTracker& state_tracker_)
-    : device{device_}, state_tracker{state_tracker_}, util_shaders(program_manager),
-      format_conversion_pass{util_shaders}, resolution{Settings::values.resolution_info} {
+                                         StateTracker& state_tracker_,
+                                         StagingBufferPool& staging_buffer_pool_)
+    : device{device_}, state_tracker{state_tracker_}, staging_buffer_pool{staging_buffer_pool_},
+      util_shaders(program_manager), format_conversion_pass{util_shaders},
+      resolution{Settings::values.resolution_info} {
     static constexpr std::array TARGETS{GL_TEXTURE_1D_ARRAY, GL_TEXTURE_2D_ARRAY, GL_TEXTURE_3D};
     for (size_t i = 0; i < TARGETS.size(); ++i) {
         const GLenum target = TARGETS[i];
@@ -558,12 +553,12 @@ void TextureCacheRuntime::Finish() {
     glFinish();
 }
 
-ImageBufferMap TextureCacheRuntime::UploadStagingBuffer(size_t size) {
-    return upload_buffers.RequestMap(size, true);
+StagingBufferMap TextureCacheRuntime::UploadStagingBuffer(size_t size) {
+    return staging_buffer_pool.RequestUploadBuffer(size);
 }
 
-ImageBufferMap TextureCacheRuntime::DownloadStagingBuffer(size_t size) {
-    return download_buffers.RequestMap(size, false);
+StagingBufferMap TextureCacheRuntime::DownloadStagingBuffer(size_t size) {
+    return staging_buffer_pool.RequestDownloadBuffer(size);
 }
 
 u64 TextureCacheRuntime::GetDeviceMemoryUsage() const {
@@ -648,7 +643,7 @@ void TextureCacheRuntime::BlitFramebuffer(Framebuffer* dst, Framebuffer* src,
                            is_linear ? GL_LINEAR : GL_NEAREST);
 }
 
-void TextureCacheRuntime::AccelerateImageUpload(Image& image, const ImageBufferMap& map,
+void TextureCacheRuntime::AccelerateImageUpload(Image& image, const StagingBufferMap& map,
                                                 std::span<const SwizzleParameters> swizzles) {
     switch (image.info.type) {
     case ImageType::e2D:
@@ -690,64 +685,6 @@ bool TextureCacheRuntime::HasNativeASTC() const noexcept {
     return device.HasASTC();
 }
 
-TextureCacheRuntime::StagingBuffers::StagingBuffers(GLenum storage_flags_, GLenum map_flags_)
-    : storage_flags{storage_flags_}, map_flags{map_flags_} {}
-
-TextureCacheRuntime::StagingBuffers::~StagingBuffers() = default;
-
-ImageBufferMap TextureCacheRuntime::StagingBuffers::RequestMap(size_t requested_size,
-                                                               bool insert_fence) {
-    const size_t index = RequestBuffer(requested_size);
-    OGLSync* const sync = insert_fence ? &syncs[index] : nullptr;
-    return ImageBufferMap{
-        .mapped_span = std::span(maps[index], requested_size),
-        .sync = sync,
-        .buffer = buffers[index].handle,
-    };
-}
-
-size_t TextureCacheRuntime::StagingBuffers::RequestBuffer(size_t requested_size) {
-    if (const std::optional<size_t> index = FindBuffer(requested_size); index) {
-        return *index;
-    }
-
-    OGLBuffer& buffer = buffers.emplace_back();
-    buffer.Create();
-    glNamedBufferStorage(buffer.handle, requested_size, nullptr,
-                         storage_flags | GL_MAP_PERSISTENT_BIT);
-    maps.push_back(static_cast<u8*>(glMapNamedBufferRange(buffer.handle, 0, requested_size,
-                                                          map_flags | GL_MAP_PERSISTENT_BIT)));
-
-    syncs.emplace_back();
-    sizes.push_back(requested_size);
-
-    ASSERT(syncs.size() == buffers.size() && buffers.size() == maps.size() &&
-           maps.size() == sizes.size());
-
-    return buffers.size() - 1;
-}
-
-std::optional<size_t> TextureCacheRuntime::StagingBuffers::FindBuffer(size_t requested_size) {
-    size_t smallest_buffer = std::numeric_limits<size_t>::max();
-    std::optional<size_t> found;
-    const size_t num_buffers = sizes.size();
-    for (size_t index = 0; index < num_buffers; ++index) {
-        const size_t buffer_size = sizes[index];
-        if (buffer_size < requested_size || buffer_size >= smallest_buffer) {
-            continue;
-        }
-        if (syncs[index].handle != 0) {
-            if (!syncs[index].IsSignaled()) {
-                continue;
-            }
-            syncs[index].Release();
-        }
-        smallest_buffer = buffer_size;
-        found = index;
-    }
-    return found;
-}
-
 Image::Image(TextureCacheRuntime& runtime_, const VideoCommon::ImageInfo& info_, GPUVAddr gpu_addr_,
              VAddr cpu_addr_)
     : VideoCommon::ImageBase(info_, gpu_addr_, cpu_addr_), runtime{&runtime_} {
@@ -823,7 +760,7 @@ void Image::UploadMemory(GLuint buffer_handle, size_t buffer_offset,
     }
 }
 
-void Image::UploadMemory(const ImageBufferMap& map,
+void Image::UploadMemory(const StagingBufferMap& map,
                          std::span<const VideoCommon::BufferImageCopy> copies) {
     UploadMemory(map.buffer, map.offset, copies);
 }
@@ -870,7 +807,7 @@ void Image::DownloadMemory(std::span<GLuint> buffer_handles, std::span<size_t> b
     }
 }
 
-void Image::DownloadMemory(ImageBufferMap& map,
+void Image::DownloadMemory(StagingBufferMap& map,
                            std::span<const VideoCommon::BufferImageCopy> copies) {
     DownloadMemory(map.buffer, map.offset, copies);
 }

src/video_core/renderer_opengl/gl_texture_cache.h

@@ -11,6 +11,7 @@
 #include "shader_recompiler/shader_info.h"
 #include "video_core/renderer_opengl/gl_device.h"
 #include "video_core/renderer_opengl/gl_resource_manager.h"
+#include "video_core/renderer_opengl/gl_staging_buffer_pool.h"
 #include "video_core/renderer_opengl/util_shaders.h"
 #include "video_core/texture_cache/image_view_base.h"
 #include "video_core/texture_cache/texture_cache_base.h"
@@ -37,15 +38,6 @@ using VideoCommon::Region2D;
 using VideoCommon::RenderTargets;
 using VideoCommon::SlotVector;
 
-struct ImageBufferMap {
-    ~ImageBufferMap();
-
-    std::span<u8> mapped_span;
-    size_t offset = 0;
-    OGLSync* sync;
-    GLuint buffer;
-};
-
 struct FormatProperties {
     GLenum compatibility_class;
     bool compatibility_by_size;
@@ -74,14 +66,15 @@ class TextureCacheRuntime {
 
 public:
     explicit TextureCacheRuntime(const Device& device, ProgramManager& program_manager,
-                                 StateTracker& state_tracker);
+                                 StateTracker& state_tracker,
+                                 StagingBufferPool& staging_buffer_pool);
     ~TextureCacheRuntime();
 
     void Finish();
 
-    ImageBufferMap UploadStagingBuffer(size_t size);
+    StagingBufferMap UploadStagingBuffer(size_t size);
 
-    ImageBufferMap DownloadStagingBuffer(size_t size);
+    StagingBufferMap DownloadStagingBuffer(size_t size);
 
     u64 GetDeviceLocalMemory() const {
         return device_access_memory;
@@ -120,7 +113,7 @@ public:
                          const Region2D& src_region, Tegra::Engines::Fermi2D::Filter filter,
                          Tegra::Engines::Fermi2D::Operation operation);
 
-    void AccelerateImageUpload(Image& image, const ImageBufferMap& map,
+    void AccelerateImageUpload(Image& image, const StagingBufferMap& map,
                                std::span<const VideoCommon::SwizzleParameters> swizzles);
 
     void InsertUploadMemoryBarrier();
@@ -149,35 +142,16 @@ public:
     }
 
 private:
-    struct StagingBuffers {
-        explicit StagingBuffers(GLenum storage_flags_, GLenum map_flags_);
-        ~StagingBuffers();
-
-        ImageBufferMap RequestMap(size_t requested_size, bool insert_fence);
-
-        size_t RequestBuffer(size_t requested_size);
-
-        std::optional<size_t> FindBuffer(size_t requested_size);
-
-        std::vector<OGLSync> syncs;
-        std::vector<OGLBuffer> buffers;
-        std::vector<u8*> maps;
-        std::vector<size_t> sizes;
-        GLenum storage_flags;
-        GLenum map_flags;
-    };
-
     const Device& device;
     StateTracker& state_tracker;
+    StagingBufferPool& staging_buffer_pool;
+
     UtilShaders util_shaders;
     FormatConversionPass format_conversion_pass;
 
     std::array<std::unordered_map<GLenum, FormatProperties>, 3> format_properties;
     bool has_broken_texture_view_formats = false;
 
-    StagingBuffers upload_buffers{GL_MAP_WRITE_BIT, GL_MAP_WRITE_BIT | GL_MAP_FLUSH_EXPLICIT_BIT};
-    StagingBuffers download_buffers{GL_MAP_READ_BIT | GL_CLIENT_STORAGE_BIT, GL_MAP_READ_BIT};
-
     OGLTexture null_image_1d_array;
     OGLTexture null_image_cube_array;
     OGLTexture null_image_3d;
@@ -213,7 +187,7 @@ public:
     void UploadMemory(GLuint buffer_handle, size_t buffer_offset,
                       std::span<const VideoCommon::BufferImageCopy> copies);
 
-    void UploadMemory(const ImageBufferMap& map,
+    void UploadMemory(const StagingBufferMap& map,
                       std::span<const VideoCommon::BufferImageCopy> copies);
 
     void DownloadMemory(GLuint buffer_handle, size_t buffer_offset,
@@ -222,7 +196,8 @@ public:
     void DownloadMemory(std::span<GLuint> buffer_handle, std::span<size_t> buffer_offset,
                         std::span<const VideoCommon::BufferImageCopy> copies);
 
-    void DownloadMemory(ImageBufferMap& map, std::span<const VideoCommon::BufferImageCopy> copies);
+    void DownloadMemory(StagingBufferMap& map,
+                        std::span<const VideoCommon::BufferImageCopy> copies);
 
     GLuint StorageHandle() noexcept;
 

src/video_core/renderer_opengl/util_shaders.cpp

@@ -19,6 +19,7 @@
 #include "video_core/host_shaders/pitch_unswizzle_comp.h"
 #include "video_core/renderer_opengl/gl_shader_manager.h"
 #include "video_core/renderer_opengl/gl_shader_util.h"
+#include "video_core/renderer_opengl/gl_staging_buffer_pool.h"
 #include "video_core/renderer_opengl/gl_texture_cache.h"
 #include "video_core/renderer_opengl/util_shaders.h"
 #include "video_core/texture_cache/accelerated_swizzle.h"
@@ -63,7 +64,7 @@ UtilShaders::UtilShaders(ProgramManager& program_manager_)
 
 UtilShaders::~UtilShaders() = default;
 
-void UtilShaders::ASTCDecode(Image& image, const ImageBufferMap& map,
+void UtilShaders::ASTCDecode(Image& image, const StagingBufferMap& map,
                              std::span<const VideoCommon::SwizzleParameters> swizzles) {
     static constexpr GLuint BINDING_INPUT_BUFFER = 0;
     static constexpr GLuint BINDING_OUTPUT_IMAGE = 0;
@@ -111,7 +112,7 @@ void UtilShaders::ASTCDecode(Image& image, const ImageBufferMap& map,
     program_manager.RestoreGuestCompute();
 }
 
-void UtilShaders::BlockLinearUpload2D(Image& image, const ImageBufferMap& map,
+void UtilShaders::BlockLinearUpload2D(Image& image, const StagingBufferMap& map,
                                       std::span<const SwizzleParameters> swizzles) {
     static constexpr Extent3D WORKGROUP_SIZE{32, 32, 1};
     static constexpr GLuint BINDING_SWIZZLE_BUFFER = 0;
@@ -148,7 +149,7 @@ void UtilShaders::BlockLinearUpload2D(Image& image, const ImageBufferMap& map,
     program_manager.RestoreGuestCompute();
 }
 
-void UtilShaders::BlockLinearUpload3D(Image& image, const ImageBufferMap& map,
+void UtilShaders::BlockLinearUpload3D(Image& image, const StagingBufferMap& map,
                                       std::span<const SwizzleParameters> swizzles) {
     static constexpr Extent3D WORKGROUP_SIZE{16, 8, 8};
 
@@ -189,7 +190,7 @@ void UtilShaders::BlockLinearUpload3D(Image& image, const ImageBufferMap& map,
     program_manager.RestoreGuestCompute();
 }
 
-void UtilShaders::PitchUpload(Image& image, const ImageBufferMap& map,
+void UtilShaders::PitchUpload(Image& image, const StagingBufferMap& map,
                               std::span<const SwizzleParameters> swizzles) {
     static constexpr Extent3D WORKGROUP_SIZE{32, 32, 1};
     static constexpr GLuint BINDING_INPUT_BUFFER = 0;

src/video_core/renderer_opengl/util_shaders.h

@@ -16,23 +16,23 @@ namespace OpenGL {
 class Image;
 class ProgramManager;
 
-struct ImageBufferMap;
+struct StagingBufferMap;
 
 class UtilShaders {
 public:
     explicit UtilShaders(ProgramManager& program_manager);
     ~UtilShaders();
 
-    void ASTCDecode(Image& image, const ImageBufferMap& map,
+    void ASTCDecode(Image& image, const StagingBufferMap& map,
                     std::span<const VideoCommon::SwizzleParameters> swizzles);
 
-    void BlockLinearUpload2D(Image& image, const ImageBufferMap& map,
+    void BlockLinearUpload2D(Image& image, const StagingBufferMap& map,
                              std::span<const VideoCommon::SwizzleParameters> swizzles);
 
-    void BlockLinearUpload3D(Image& image, const ImageBufferMap& map,
+    void BlockLinearUpload3D(Image& image, const StagingBufferMap& map,
                              std::span<const VideoCommon::SwizzleParameters> swizzles);
 
-    void PitchUpload(Image& image, const ImageBufferMap& map,
+    void PitchUpload(Image& image, const StagingBufferMap& map,
                      std::span<const VideoCommon::SwizzleParameters> swizzles);
 
     void CopyBC4(Image& dst_image, Image& src_image,

src/video_core/renderer_vulkan/vk_buffer_cache.h

@@ -157,6 +157,7 @@ struct BufferCacheParams {
     static constexpr bool USE_MEMORY_MAPS = true;
     static constexpr bool SEPARATE_IMAGE_BUFFER_BINDINGS = false;
     static constexpr bool IMPLEMENTS_ASYNC_DOWNLOADS = true;
+    static constexpr bool USE_MEMORY_MAPS_FOR_UPLOADS = true;
 };
 
 using BufferCache = VideoCommon::BufferCache<BufferCacheParams>;

src/video_core/texture_cache/texture_cache.h

@@ -850,15 +850,11 @@ void TextureCache<P>::PopAsyncFlushes() {
 template <class P>
 ImageId TextureCache<P>::DmaImageId(const Tegra::DMA::ImageOperand& operand, bool is_upload) {
     const ImageInfo dst_info(operand);
-    const ImageId dst_id = FindDMAImage(dst_info, operand.address);
-    if (!dst_id) {
-        return NULL_IMAGE_ID;
-    }
-    auto& image = slot_images[dst_id];
-    if (False(image.flags & ImageFlagBits::GpuModified)) {
-        // No need to waste time on an image that's synced with guest
+    const ImageId image_id = FindDMAImage(dst_info, operand.address);
+    if (!image_id) {
         return NULL_IMAGE_ID;
     }
+    auto& image = slot_images[image_id];
     if (!is_upload && !image.info.dma_downloaded) {
         // Force a full sync.
         image.info.dma_downloaded = true;
@@ -868,7 +864,7 @@ ImageId TextureCache<P>::DmaImageId(const Tegra::DMA::ImageOperand& operand, boo
     if (!base) {
         return NULL_IMAGE_ID;
     }
-    return dst_id;
+    return image_id;
 }
 
 template <class P>

src/video_core/vulkan_common/vulkan_device.h

@@ -85,7 +85,6 @@
 
 // Define extensions which must be supported.
 #define FOR_EACH_VK_MANDATORY_EXTENSION(EXTENSION_NAME)                                            \
-    EXTENSION_NAME(VK_EXT_ROBUSTNESS_2_EXTENSION_NAME)                                             \
     EXTENSION_NAME(VK_EXT_VERTEX_ATTRIBUTE_DIVISOR_EXTENSION_NAME)                                 \
     EXTENSION_NAME(VK_KHR_DRIVER_PROPERTIES_EXTENSION_NAME)                                        \
     EXTENSION_NAME(VK_KHR_SAMPLER_MIRROR_CLAMP_TO_EDGE_EXTENSION_NAME)                             \
@@ -105,6 +104,7 @@
     EXTENSION_NAME(VK_EXT_EXTENDED_DYNAMIC_STATE_2_EXTENSION_NAME)                                 \
     EXTENSION_NAME(VK_EXT_EXTENDED_DYNAMIC_STATE_3_EXTENSION_NAME)                                 \
     EXTENSION_NAME(VK_EXT_LINE_RASTERIZATION_EXTENSION_NAME)                                       \
+    EXTENSION_NAME(VK_EXT_ROBUSTNESS_2_EXTENSION_NAME)                                             \
     EXTENSION_NAME(VK_EXT_VERTEX_INPUT_DYNAMIC_STATE_EXTENSION_NAME)                               \
     EXTENSION_NAME(VK_NV_GEOMETRY_SHADER_PASSTHROUGH_EXTENSION_NAME)                               \
     EXTENSION_NAME(VK_NV_VIEWPORT_ARRAY2_EXTENSION_NAME)                                           \
@@ -141,9 +141,6 @@
     FEATURE_NAME(features, vertexPipelineStoresAndAtomics)                                         \
     FEATURE_NAME(features, wideLines)                                                              \
     FEATURE_NAME(host_query_reset, hostQueryReset)                                                 \
-    FEATURE_NAME(robustness2, nullDescriptor)                                                      \
-    FEATURE_NAME(robustness2, robustBufferAccess2)                                                 \
-    FEATURE_NAME(robustness2, robustImageAccess2)                                                  \
     FEATURE_NAME(shader_demote_to_helper_invocation, shaderDemoteToHelperInvocation)               \
     FEATURE_NAME(shader_draw_parameters, shaderDrawParameters)                                     \
     FEATURE_NAME(variable_pointer, variablePointers)                                               \
@@ -156,6 +153,9 @@
     FEATURE_NAME(index_type_uint8, indexTypeUint8)                                                 \
     FEATURE_NAME(primitive_topology_list_restart, primitiveTopologyListRestart)                    \
     FEATURE_NAME(provoking_vertex, provokingVertexLast)                                            \
+    FEATURE_NAME(robustness2, nullDescriptor)                                                      \
+    FEATURE_NAME(robustness2, robustBufferAccess2)                                                 \
+    FEATURE_NAME(robustness2, robustImageAccess2)                                                  \
     FEATURE_NAME(shader_float16_int8, shaderFloat16)                                               \
     FEATURE_NAME(shader_float16_int8, shaderInt8)                                                  \
     FEATURE_NAME(timeline_semaphore, timelineSemaphore)                                            \

src/yuzu/configuration/config.cpp

@@ -101,6 +101,12 @@ const std::map<Settings::RendererBackend, QString> Config::renderer_backend_text
     {Settings::RendererBackend::Null, QStringLiteral(QT_TRANSLATE_NOOP("GMainWindow", "Null"))},
 };
 
+const std::map<Settings::ShaderBackend, QString> Config::shader_backend_texts_map = {
+    {Settings::ShaderBackend::GLSL, QStringLiteral(QT_TRANSLATE_NOOP("GMainWindow", "GLSL"))},
+    {Settings::ShaderBackend::GLASM, QStringLiteral(QT_TRANSLATE_NOOP("GMainWindow", "GLASM"))},
+    {Settings::ShaderBackend::SPIRV, QStringLiteral(QT_TRANSLATE_NOOP("GMainWindow", "SPIRV"))},
+};
+
 // This shouldn't have anything except static initializers (no functions). So
 // QKeySequence(...).toString() is NOT ALLOWED HERE.
 // This must be in alphabetical order according to action name as it must have the same order as

src/yuzu/configuration/config.h

@@ -54,6 +54,7 @@ public:
     static const std::map<bool, QString> use_docked_mode_texts_map;
     static const std::map<Settings::GPUAccuracy, QString> gpu_accuracy_texts_map;
     static const std::map<Settings::RendererBackend, QString> renderer_backend_texts_map;
+    static const std::map<Settings::ShaderBackend, QString> shader_backend_texts_map;
 
     static constexpr UISettings::Theme default_theme{
 #ifdef _WIN32

src/yuzu/configuration/configure_graphics_advanced.cpp

@@ -42,6 +42,7 @@ void ConfigureGraphicsAdvanced::SetConfiguration() {
         Settings::values.use_vulkan_driver_pipeline_cache.GetValue());
     ui->enable_compute_pipelines_checkbox->setChecked(
         Settings::values.enable_compute_pipelines.GetValue());
+    ui->use_video_framerate_checkbox->setChecked(Settings::values.use_video_framerate.GetValue());
 
     if (Settings::IsConfiguringGlobal()) {
         ui->gpu_accuracy->setCurrentIndex(
@@ -91,6 +92,8 @@ void ConfigureGraphicsAdvanced::ApplyConfiguration() {
     ConfigurationShared::ApplyPerGameSetting(&Settings::values.enable_compute_pipelines,
                                              ui->enable_compute_pipelines_checkbox,
                                              enable_compute_pipelines);
+    ConfigurationShared::ApplyPerGameSetting(&Settings::values.use_video_framerate,
+                                             ui->use_video_framerate_checkbox, use_video_framerate);
 }
 
 void ConfigureGraphicsAdvanced::changeEvent(QEvent* event) {
@@ -125,6 +128,8 @@ void ConfigureGraphicsAdvanced::SetupPerGameUI() {
             Settings::values.max_anisotropy.UsingGlobal());
         ui->enable_compute_pipelines_checkbox->setEnabled(
             Settings::values.enable_compute_pipelines.UsingGlobal());
+        ui->use_video_framerate_checkbox->setEnabled(
+            Settings::values.use_video_framerate.UsingGlobal());
 
         return;
     }
@@ -149,6 +154,9 @@ void ConfigureGraphicsAdvanced::SetupPerGameUI() {
     ConfigurationShared::SetColoredTristate(ui->enable_compute_pipelines_checkbox,
                                             Settings::values.enable_compute_pipelines,
                                             enable_compute_pipelines);
+    ConfigurationShared::SetColoredTristate(ui->use_video_framerate_checkbox,
+                                            Settings::values.use_video_framerate,
+                                            use_video_framerate);
     ConfigurationShared::SetColoredComboBox(
         ui->gpu_accuracy, ui->label_gpu_accuracy,
         static_cast<int>(Settings::values.gpu_accuracy.GetValue(true)));

src/yuzu/configuration/configure_graphics_advanced.h

@@ -47,6 +47,7 @@ private:
     ConfigurationShared::CheckState use_fast_gpu_time;
     ConfigurationShared::CheckState use_vulkan_driver_pipeline_cache;
     ConfigurationShared::CheckState enable_compute_pipelines;
+    ConfigurationShared::CheckState use_video_framerate;
 
     const Core::System& system;
 };

src/yuzu/configuration/configure_graphics_advanced.ui

@@ -191,6 +191,16 @@ Compute pipelines are always enabled on all other drivers.</string>
           </property>
          </widget>
         </item>
+        <item>
+         <widget class="QCheckBox" name="use_video_framerate_checkbox">
+          <property name="toolTip">
+           <string>Run the game at normal speed during video playback, even when the framerate is unlocked.</string>
+          </property>
+          <property name="text">
+           <string>Sync to framerate of video playback</string>
+          </property>
+         </widget>
+        </item>
         <item>
          <widget class="QWidget" name="af_layout" native="true">
           <layout class="QHBoxLayout" name="horizontalLayout_1">

src/yuzu/main.cpp

@@ -3491,6 +3491,7 @@ void GMainWindow::ResetWindowSize1080() {
 void GMainWindow::OnConfigure() {
     const auto old_theme = UISettings::values.theme;
     const bool old_discord_presence = UISettings::values.enable_discord_presence.GetValue();
+    const auto old_language_index = Settings::values.language_index.GetValue();
 
     Settings::SetConfiguringGlobal(true);
     ConfigureDialog configure_dialog(this, hotkey_registry, input_subsystem.get(), *system,
@@ -3559,7 +3560,7 @@ void GMainWindow::OnConfigure() {
     emit UpdateThemedIcons();
 
     const auto reload = UISettings::values.is_game_list_reload_pending.exchange(false);
-    if (reload) {
+    if (reload || Settings::values.language_index.GetValue() != old_language_index) {
         game_list->PopulateAsync(UISettings::values.game_dirs);
     }
 
@@ -4115,7 +4116,13 @@ void GMainWindow::UpdateDockedButton() {
 void GMainWindow::UpdateAPIText() {
     const auto api = Settings::values.renderer_backend.GetValue();
     const auto renderer_status_text = Config::renderer_backend_texts_map.find(api)->second;
-    renderer_status_button->setText(renderer_status_text.toUpper());
+    renderer_status_button->setText(
+        api == Settings::RendererBackend::OpenGL
+            ? tr("%1 %2").arg(
+                  renderer_status_text.toUpper(),
+                  Config::shader_backend_texts_map.find(Settings::values.shader_backend.GetValue())
+                      ->second)
+            : renderer_status_text.toUpper());
 }
 
 void GMainWindow::UpdateFilterText() {