cubeb: Show default device option even if enumerating devices fails (#1515)

This allows a savy user, developer or modder to change the comment field of a logging breakpoint to include placeholders such as {r3} or {f3} to log the register values whenever that code is hit.

.github/workflows/build.yml

@@ -39,7 +39,7 @@ jobs:
     - name: "Install system dependencies"
       run: |
         sudo apt update -qq
-        sudo apt install -y clang-15 cmake freeglut3-dev libgcrypt20-dev libglm-dev libgtk-3-dev libpulse-dev libsecret-1-dev libsystemd-dev libudev-dev nasm ninja-build
+        sudo apt install -y clang-15 cmake freeglut3-dev libgcrypt20-dev libglm-dev libgtk-3-dev libpulse-dev libsecret-1-dev libsystemd-dev libudev-dev nasm ninja-build libbluetooth-dev
 
     - name: "Setup cmake"
       uses: jwlawson/actions-setup-cmake@v2
@@ -96,7 +96,7 @@ jobs:
     - name: "Install system dependencies"
       run: |
         sudo apt update -qq
-        sudo apt install -y clang-15 cmake freeglut3-dev libgcrypt20-dev libglm-dev libgtk-3-dev libpulse-dev libsecret-1-dev libsystemd-dev nasm ninja-build appstream
+        sudo apt install -y clang-15 cmake freeglut3-dev libgcrypt20-dev libglm-dev libgtk-3-dev libpulse-dev libsecret-1-dev libsystemd-dev nasm ninja-build appstream libbluetooth-dev
 
     - name: "Build AppImage"
       run: |

.github/workflows/deploy_release.yml

@@ -1,4 +1,4 @@
-name: Deploy experimental release
+name: Deploy release
 on: 
   workflow_dispatch:
     inputs:
@@ -54,7 +54,7 @@ jobs:
       next_version_major: ${{ needs.calculate-version.outputs.next_version_major }}
       next_version_minor: ${{ needs.calculate-version.outputs.next_version_minor }}
   deploy:
-    name: Deploy experimental release
+    name: Deploy release
     runs-on: ubuntu-22.04
     needs: [call-release-build, calculate-version]
     steps:

.github/workflows/deploy_stable_release.yml

@@ -1,85 +0,0 @@
-name: Create new release
-on: 
-  workflow_dispatch:
-    inputs:
-      PlaceholderInput:
-        description: PlaceholderInput
-        required: false
-jobs:
-  call-release-build:
-    uses: ./.github/workflows/build.yml
-    with:
-      deploymode: release
-  deploy:
-    name: Deploy release
-    runs-on: ubuntu-20.04
-    needs: call-release-build
-    steps:
-      - uses: actions/checkout@v3
- 
-      - uses: actions/download-artifact@v4
-        with:
-          name: cemu-bin-linux-x64
-          path: cemu-bin-linux-x64
-
-      - uses: actions/download-artifact@v4
-        with:
-          name: cemu-appimage-x64
-          path: cemu-appimage-x64
-          
-      - uses: actions/download-artifact@v4
-        with:
-          name: cemu-bin-windows-x64
-          path: cemu-bin-windows-x64
-
-      - uses: actions/download-artifact@v4
-        with:
-          name: cemu-bin-macos-x64
-          path: cemu-bin-macos-x64
-
-      - name: Initialize
-        run: |
-          mkdir upload
-          sudo apt update -qq
-          sudo apt install -y zip
-          
-      - name: Get Cemu release version
-        run: |
-          gcc -o getversion .github/getversion.cpp
-          echo "Cemu CI version: $(./getversion)"
-          echo "CEMU_FOLDER_NAME=Cemu_$(./getversion)" >> $GITHUB_ENV
-          echo "CEMU_VERSION=$(./getversion)" >> $GITHUB_ENV
-          
-      - name: Create release from windows-bin
-        run: |
-          ls ./
-          ls ./bin/
-          cp -R ./bin ./${{ env.CEMU_FOLDER_NAME }}
-          mv cemu-bin-windows-x64/Cemu.exe ./${{ env.CEMU_FOLDER_NAME }}/Cemu.exe
-          zip -9 -r upload/cemu-${{ env.CEMU_VERSION }}-windows-x64.zip ${{ env.CEMU_FOLDER_NAME }}
-          rm -r ./${{ env.CEMU_FOLDER_NAME }}
-
-      - name: Create appimage
-        run: |
-          VERSION=${{ env.CEMU_VERSION }}
-          echo "Cemu Version is $VERSION"
-          ls cemu-appimage-x64
-          mv cemu-appimage-x64/Cemu-*-x86_64.AppImage upload/Cemu-$VERSION-x86_64.AppImage
-          
-      - name: Create release from ubuntu-bin
-        run: |
-          ls ./
-          ls ./bin/
-          cp -R ./bin ./${{ env.CEMU_FOLDER_NAME }}
-          mv cemu-bin-linux-x64/Cemu ./${{ env.CEMU_FOLDER_NAME }}/Cemu
-          zip -9 -r upload/cemu-${{ env.CEMU_VERSION }}-ubuntu-20.04-x64.zip ${{ env.CEMU_FOLDER_NAME }}
-          rm -r ./${{ env.CEMU_FOLDER_NAME }}
-          
-      - name: Create release from macos-bin
-        run: cp cemu-bin-macos-x64/Cemu.dmg upload/cemu-${{ env.CEMU_VERSION }}-macos-12-x64.dmg
-          
-      - name: Create release
-        run: |
-          wget -O ghr.tar.gz https://github.com/tcnksm/ghr/releases/download/v0.15.0/ghr_v0.15.0_linux_amd64.tar.gz
-          tar xvzf ghr.tar.gz; rm ghr.tar.gz
-          ghr_v0.15.0_linux_amd64/ghr -t ${{ secrets.GITHUB_TOKEN }} -n "Cemu ${{ env.CEMU_VERSION }}" -b "Changelog:" v${{ env.CEMU_VERSION }} ./upload

.github/workflows/generate_pot.yml

@@ -35,7 +35,7 @@ jobs:
           -o cemu.pot
 
       - name: Upload artifact
-        uses: actions/upload-artifact@v3
+        uses: actions/upload-artifact@v4
         with:
           name: POT file
           path: ./cemu.pot

BUILD.md

@@ -46,10 +46,10 @@ To compile Cemu, a recent enough compiler and STL with C++20 support is required
 ### Dependencies
 
 #### For Arch and derivatives:
-`sudo pacman -S --needed base-devel clang cmake freeglut git glm gtk3 libgcrypt libpulse libsecret linux-headers llvm nasm ninja systemd unzip zip`
+`sudo pacman -S --needed base-devel bluez-libs clang cmake freeglut git glm gtk3 libgcrypt libpulse libsecret linux-headers llvm nasm ninja systemd unzip zip`
 
 #### For Debian, Ubuntu and derivatives:
-`sudo apt install -y cmake curl clang-15 freeglut3-dev git libgcrypt20-dev libglm-dev libgtk-3-dev libpulse-dev libsecret-1-dev libsystemd-dev libtool nasm ninja-build`
+`sudo apt install -y cmake curl clang-15 freeglut3-dev git libbluetooth-dev libgcrypt20-dev libglm-dev libgtk-3-dev libpulse-dev libsecret-1-dev libsystemd-dev libtool nasm ninja-build`
 
 You may also need to install `libusb-1.0-0-dev` as a workaround for an issue with the vcpkg hidapi package.
 
@@ -57,7 +57,7 @@ At Step 3 in [Build Cemu using cmake and clang](#build-cemu-using-cmake-and-clan
    `cmake -S . -B build -DCMAKE_BUILD_TYPE=release -DCMAKE_C_COMPILER=/usr/bin/clang-15 -DCMAKE_CXX_COMPILER=/usr/bin/clang++-15 -G Ninja -DCMAKE_MAKE_PROGRAM=/usr/bin/ninja`
 
 #### For Fedora and derivatives:
-`sudo dnf install clang cmake cubeb-devel freeglut-devel git glm-devel gtk3-devel kernel-headers libgcrypt-devel libsecret-devel libtool libusb1-devel llvm nasm ninja-build perl-core systemd-devel zlib-devel zlib-static`
+`sudo dnf install bluez-libs-devel clang cmake cubeb-devel freeglut-devel git glm-devel gtk3-devel kernel-headers libgcrypt-devel libsecret-devel libtool libusb1-devel llvm nasm ninja-build perl-core systemd-devel wayland-protocols-devel zlib-devel zlib-static`
 
 ### Build Cemu
 
@@ -120,6 +120,9 @@ This section refers to running `cmake -S...` (truncated).
 * Compiling failed during rebuild after `git pull` with an error that mentions RPATH
    * Add the following and try running the command again:
       * `-DCMAKE_BUILD_WITH_INSTALL_RPATH=ON`
+* Environment variable `VCPKG_FORCE_SYSTEM_BINARIES` must be set.
+   * Execute the folowing and then try running the command again:
+      * `export VCPKG_FORCE_SYSTEM_BINARIES=1`
 * If you are getting a random error, read the [package-name-and-platform]-out.log and [package-name-and-platform]-err.log for the actual reason to see if you might be lacking the headers from a dependency.
 
 
@@ -189,3 +192,41 @@ Then install the dependencies:
 
 If CMake complains about Cemu already being compiled or another similar error, try deleting the `CMakeCache.txt` file inside the `build` folder and retry building.
 
+## CMake configure flags
+Some flags can be passed during CMake configure to customise which features are enabled on build.
+
+Example usage: `cmake -S . -B build -DCMAKE_BUILD_TYPE=release -DENABLE_SDL=ON -DENABLE_VULKAN=OFF`
+
+### All platforms
+| Flag               |   | Description                                                                 | Default | Note               |
+|--------------------|:--|-----------------------------------------------------------------------------|---------|--------------------|
+| ALLOW_PORTABLE     |   | Allow Cemu to use the `portable` directory to store configs and data        | ON      |                    |
+| CEMU_CXX_FLAGS     |   | Flags passed straight to the compiler, e.g. `-march=native`, `-Wall`, `/W3` | ""      |                    |
+| ENABLE_CUBEB       |   | Enable cubeb audio backend                                                  | ON      |                    |
+| ENABLE_DISCORD_RPC |   | Enable Discord Rich presence support                                        | ON      |                    |
+| ENABLE_OPENGL      |   | Enable OpenGL graphics backend                                              | ON      | Currently required |
+| ENABLE_HIDAPI      |   | Enable HIDAPI (used for Wiimote controller API)                             | ON      |                    |
+| ENABLE_SDL         |   | Enable SDLController controller API                                         | ON      | Currently required |
+| ENABLE_VCPKG       |   | Use VCPKG package manager to obtain dependencies                            | ON      |                    |
+| ENABLE_VULKAN      |   | Enable the Vulkan graphics backend                                          | ON      |                    |
+| ENABLE_WXWIDGETS   |   | Enable wxWidgets UI                                                         | ON      | Currently required |
+
+### Windows
+| Flag               | Description                       | Default | Note               |
+|--------------------|-----------------------------------|---------|--------------------|
+| ENABLE_DIRECTAUDIO | Enable DirectAudio audio backend  | ON      | Currently required |
+| ENABLE_DIRECTINPUT | Enable DirectInput controller API | ON      | Currently required |
+| ENABLE_XAUDIO      | Enable XAudio audio backend       | ON      |                    |
+| ENABLE_XINPUT      | Enable XInput controller API      | ON      |                    |
+
+### Linux
+| Flag                  | Description                                        | Default |
+|-----------------------|----------------------------------------------------|---------|
+| ENABLE_BLUEZ          | Build with Bluez (used for Wiimote controller API) | ON      |
+| ENABLE_FERAL_GAMEMODE | Enable Feral Interactive GameMode support          | ON      |
+| ENABLE_WAYLAND        | Enable Wayland support                             | ON      |
+
+### macOS
+| Flag         | Description                                    | Default |
+|--------------|------------------------------------------------|---------|
+| MACOS_BUNDLE | MacOS executable will be an application bundle | OFF     |

CMakeLists.txt

@@ -2,6 +2,7 @@ cmake_minimum_required(VERSION 3.21.1)
 
 option(ENABLE_VCPKG "Enable the vcpkg package manager" ON)
 option(MACOS_BUNDLE "The executable when built on macOS will be created as an application bundle" OFF)
+option(ALLOW_PORTABLE "Allow Cemu to be run in portable mode" ON)
 
 # used by CI script to set version:
 set(EMULATOR_VERSION_MAJOR "0" CACHE STRING "")
@@ -98,6 +99,7 @@ endif()
 if (UNIX AND NOT APPLE)
 	option(ENABLE_WAYLAND "Build with Wayland support" ON)
 	option(ENABLE_FERAL_GAMEMODE "Enables Feral Interactive GameMode Support" ON)
+	option(ENABLE_BLUEZ "Build with Bluez support" ON)
 endif()
 
 option(ENABLE_OPENGL "Enables the OpenGL backend" ON)
@@ -122,23 +124,6 @@ if (WIN32)
 endif()
 option(ENABLE_CUBEB "Enabled cubeb backend" ON)
 
-# usb hid backends
-if (WIN32)
-	option(ENABLE_NSYSHID_WINDOWS_HID "Enables the native Windows HID backend for nsyshid" ON)
-endif ()
-# libusb and windows hid backends shouldn't be active at the same time; otherwise we'd see all devices twice!
-if (NOT ENABLE_NSYSHID_WINDOWS_HID)
-	option(ENABLE_NSYSHID_LIBUSB "Enables the libusb backend for nsyshid" ON)
-else ()
-	set(ENABLE_NSYSHID_LIBUSB OFF CACHE BOOL "" FORCE)
-endif ()
-if (ENABLE_NSYSHID_WINDOWS_HID)
-	add_compile_definitions(NSYSHID_ENABLE_BACKEND_WINDOWS_HID)
-endif ()
-if (ENABLE_NSYSHID_LIBUSB)
-	add_compile_definitions(NSYSHID_ENABLE_BACKEND_LIBUSB)
-endif ()
-
 option(ENABLE_WXWIDGETS "Build with wxWidgets UI (Currently required)" ON)
 
 set(THREADS_PREFER_PTHREAD_FLAG true)
@@ -179,6 +164,12 @@ if (UNIX AND NOT APPLE)
 	endif()
 	find_package(GTK3 REQUIRED)
 
+	if(ENABLE_BLUEZ)
+		find_package(bluez REQUIRED)
+		set(ENABLE_WIIMOTE ON)
+		add_compile_definitions(HAS_BLUEZ)
+	endif()
+
 endif()
 
 if (ENABLE_VULKAN)

cmake/Findbluez.cmake

@@ -0,0 +1,20 @@
+# SPDX-FileCopyrightText: 2022 Andrea Pappacoda <andrea@pappacoda.it>
+# SPDX-License-Identifier: ISC
+
+find_package(bluez CONFIG)
+if (NOT bluez_FOUND)
+  find_package(PkgConfig)
+  if (PKG_CONFIG_FOUND)
+    pkg_search_module(bluez IMPORTED_TARGET GLOBAL bluez-1.0 bluez)
+    if (bluez_FOUND)
+      add_library(bluez::bluez ALIAS PkgConfig::bluez)
+    endif ()
+  endif ()
+endif ()
+
+find_package_handle_standard_args(bluez
+    REQUIRED_VARS
+    bluez_LINK_LIBRARIES
+    bluez_FOUND
+    VERSION_VAR bluez_VERSION
+)

src/CMakeLists.txt

@@ -82,8 +82,8 @@ if (MACOS_BUNDLE)
 	set(MACOSX_BUNDLE_ICON_FILE "cemu.icns")
 	set(MACOSX_BUNDLE_GUI_IDENTIFIER "info.cemu.Cemu")
 	set(MACOSX_BUNDLE_BUNDLE_NAME "Cemu")
-	set(MACOSX_BUNDLE_SHORT_VERSION_STRING ${CMAKE_PROJECT_VERSION})
-	set(MACOSX_BUNDLE_BUNDLE_VERSION ${CMAKE_PROJECT_VERSION})
+	set(MACOSX_BUNDLE_SHORT_VERSION_STRING "${EMULATOR_VERSION_MAJOR}.${EMULATOR_VERSION_MINOR}.${EMULATOR_VERSION_PATCH}")
+	set(MACOSX_BUNDLE_BUNDLE_VERSION "${EMULATOR_VERSION_MAJOR}.${EMULATOR_VERSION_MINOR}.${EMULATOR_VERSION_PATCH}")
 	set(MACOSX_BUNDLE_COPYRIGHT "Copyright © 2024 Cemu Project")
 
 	set(MACOSX_BUNDLE_CATEGORY "public.app-category.games")
@@ -101,12 +101,18 @@ if (MACOS_BUNDLE)
 			COMMAND ${CMAKE_COMMAND} ARGS -E copy_directory "${CMAKE_SOURCE_DIR}/bin/${folder}" "${CMAKE_SOURCE_DIR}/bin/${OUTPUT_NAME}.app/Contents/SharedSupport/${folder}")
 	endforeach(folder)
 
+	if(CMAKE_BUILD_TYPE STREQUAL "Debug")
+	    set(LIBUSB_PATH "${CMAKE_BINARY_DIR}/vcpkg_installed/x64-osx/debug/lib/libusb-1.0.0.dylib")
+	else()
+	    set(LIBUSB_PATH "${CMAKE_BINARY_DIR}/vcpkg_installed/x64-osx/lib/libusb-1.0.0.dylib")
+	endif()
+
 	add_custom_command (TARGET CemuBin POST_BUILD
 		COMMAND ${CMAKE_COMMAND} ARGS -E copy "/usr/local/lib/libMoltenVK.dylib" "${CMAKE_SOURCE_DIR}/bin/${OUTPUT_NAME}.app/Contents/Frameworks/libMoltenVK.dylib"
-		COMMAND ${CMAKE_COMMAND} ARGS -E copy "${CMAKE_BINARY_DIR}/vcpkg_installed/x64-osx/lib/libusb-1.0.0.dylib" "${CMAKE_SOURCE_DIR}/bin/${OUTPUT_NAME}.app/Contents/Frameworks/libusb-1.0.0.dylib"
+		COMMAND ${CMAKE_COMMAND} ARGS -E copy "${LIBUSB_PATH}" "${CMAKE_SOURCE_DIR}/bin/${OUTPUT_NAME}.app/Contents/Frameworks/libusb-1.0.0.dylib"
 		COMMAND ${CMAKE_COMMAND} ARGS -E copy "${CMAKE_SOURCE_DIR}/src/resource/update.sh" "${CMAKE_SOURCE_DIR}/bin/${OUTPUT_NAME}.app/Contents/MacOS/update.sh"
 		COMMAND bash -c "install_name_tool -add_rpath @executable_path/../Frameworks ${CMAKE_SOURCE_DIR}/bin/${OUTPUT_NAME}.app/Contents/MacOS/${OUTPUT_NAME}"
-		COMMAND bash -c "install_name_tool -change /Users/runner/work/Cemu/Cemu/build/vcpkg_installed/x64-osx/lib/libusb-1.0.0.dylib @executable_path/../Frameworks/libusb-1.0.0.dylib ${CMAKE_SOURCE_DIR}/bin/${OUTPUT_NAME}.app/Contents/MacOS/${OUTPUT_NAME}")
+		COMMAND bash -c "install_name_tool -change ${LIBUSB_PATH} @executable_path/../Frameworks/libusb-1.0.0.dylib ${CMAKE_SOURCE_DIR}/bin/${OUTPUT_NAME}.app/Contents/MacOS/${OUTPUT_NAME}")
 endif()
 
 set_target_properties(CemuBin PROPERTIES

src/Cafe/CMakeLists.txt

@@ -463,8 +463,8 @@ add_library(CemuCafe
   OS/libs/nsyshid/BackendEmulated.h
   OS/libs/nsyshid/BackendLibusb.cpp
   OS/libs/nsyshid/BackendLibusb.h
-  OS/libs/nsyshid/BackendWindowsHID.cpp
-  OS/libs/nsyshid/BackendWindowsHID.h
+  OS/libs/nsyshid/Dimensions.cpp
+  OS/libs/nsyshid/Dimensions.h
   OS/libs/nsyshid/Infinity.cpp
   OS/libs/nsyshid/Infinity.h
   OS/libs/nsyshid/Skylander.cpp
@@ -530,6 +530,12 @@ set_property(TARGET CemuCafe PROPERTY MSVC_RUNTIME_LIBRARY "MultiThreaded$<$<CON
 
 target_include_directories(CemuCafe PUBLIC "../")
 
+if (glslang_VERSION VERSION_LESS "15.0.0")
+	set(glslang_target "glslang::SPIRV")
+else()
+	set(glslang_target "glslang")
+endif()
+
 target_link_libraries(CemuCafe PRIVATE
 	CemuAsm
 	CemuAudio
@@ -545,7 +551,7 @@ target_link_libraries(CemuCafe PRIVATE
 	Boost::nowide
 	CURL::libcurl
 	fmt::fmt
-	glslang::SPIRV
+	${glslang_target}
 	ih264d
 	OpenSSL::Crypto
 	OpenSSL::SSL
@@ -561,15 +567,16 @@ if (ENABLE_WAYLAND)
   target_link_libraries(CemuCafe PUBLIC Wayland::Client)
 endif()
 
-if (ENABLE_NSYSHID_LIBUSB)
-	if (ENABLE_VCPKG)
-		find_package(PkgConfig REQUIRED)
-		pkg_check_modules(libusb REQUIRED IMPORTED_TARGET libusb-1.0)
-		target_link_libraries(CemuCafe PRIVATE PkgConfig::libusb)
-	else ()
-		find_package(libusb MODULE REQUIRED)
-		target_link_libraries(CemuCafe PRIVATE libusb::libusb)
-	endif ()
+if (ENABLE_VCPKG)
+  if(WIN32)
+    set(PKG_CONFIG_EXECUTABLE "${VCPKG_INSTALLED_DIR}/x64-windows/tools/pkgconf/pkgconf.exe")
+  endif()
+  find_package(PkgConfig REQUIRED)
+  pkg_check_modules(libusb REQUIRED IMPORTED_TARGET libusb-1.0)
+  target_link_libraries(CemuCafe PRIVATE PkgConfig::libusb)
+else ()
+  find_package(libusb MODULE REQUIRED)
+  target_link_libraries(CemuCafe PRIVATE libusb::libusb)
 endif ()
 
 if (ENABLE_WXWIDGETS)

src/Cafe/CafeSystem.cpp

@@ -9,6 +9,7 @@
 #include "audio/IAudioAPI.h"
 #include "audio/IAudioInputAPI.h"
 #include "config/ActiveSettings.h"
+#include "config/LaunchSettings.h"
 #include "Cafe/TitleList/GameInfo.h"
 #include "Cafe/GraphicPack/GraphicPack2.h"
 #include "util/helpers/SystemException.h"
@@ -637,40 +638,40 @@ namespace CafeSystem
         fsc_unmount("/cemuBossStorage/", FSC_PRIORITY_BASE);
     }
 
-	STATUS_CODE LoadAndMountForegroundTitle(TitleId titleId)
+	PREPARE_STATUS_CODE LoadAndMountForegroundTitle(TitleId titleId)
 	{
         cemuLog_log(LogType::Force, "Mounting title {:016x}", (uint64)titleId);
 		sGameInfo_ForegroundTitle = CafeTitleList::GetGameInfo(titleId);
 		if (!sGameInfo_ForegroundTitle.IsValid())
 		{
 			cemuLog_log(LogType::Force, "Mounting failed: Game meta information is either missing, inaccessible or not valid (missing or invalid .xml files in code and meta folder)");
-			return STATUS_CODE::UNABLE_TO_MOUNT;
+			return PREPARE_STATUS_CODE::UNABLE_TO_MOUNT;
 		}
 		// check base
 		TitleInfo& titleBase = sGameInfo_ForegroundTitle.GetBase();
 		if (!titleBase.IsValid())
-			return STATUS_CODE::UNABLE_TO_MOUNT;
+			return PREPARE_STATUS_CODE::UNABLE_TO_MOUNT;
 		if(!titleBase.ParseXmlInfo())
-			return STATUS_CODE::UNABLE_TO_MOUNT;
+			return PREPARE_STATUS_CODE::UNABLE_TO_MOUNT;
 		cemuLog_log(LogType::Force, "Base: {}", titleBase.GetPrintPath());
 		// mount base
 		if (!titleBase.Mount("/vol/content", "content", FSC_PRIORITY_BASE) || !titleBase.Mount(GetInternalVirtualCodeFolder(), "code", FSC_PRIORITY_BASE))
 		{
 			cemuLog_log(LogType::Force, "Mounting failed");
-			return STATUS_CODE::UNABLE_TO_MOUNT;
+			return PREPARE_STATUS_CODE::UNABLE_TO_MOUNT;
 		}
 		// check update
 		TitleInfo& titleUpdate = sGameInfo_ForegroundTitle.GetUpdate();
 		if (titleUpdate.IsValid())
 		{
 			if (!titleUpdate.ParseXmlInfo())
-				return STATUS_CODE::UNABLE_TO_MOUNT;
+				return PREPARE_STATUS_CODE::UNABLE_TO_MOUNT;
 			cemuLog_log(LogType::Force, "Update: {}", titleUpdate.GetPrintPath());
 			// mount update
 			if (!titleUpdate.Mount("/vol/content", "content", FSC_PRIORITY_PATCH) || !titleUpdate.Mount(GetInternalVirtualCodeFolder(), "code", FSC_PRIORITY_PATCH))
 			{
 				cemuLog_log(LogType::Force, "Mounting failed");
-				return STATUS_CODE::UNABLE_TO_MOUNT;
+				return PREPARE_STATUS_CODE::UNABLE_TO_MOUNT;
 			}
 		}
 		else
@@ -682,20 +683,20 @@ namespace CafeSystem
 			// todo - support for multi-title AOC
 			TitleInfo& titleAOC = aocList[0];
 			if (!titleAOC.ParseXmlInfo())
-				return STATUS_CODE::UNABLE_TO_MOUNT;
+				return PREPARE_STATUS_CODE::UNABLE_TO_MOUNT;
 			cemu_assert_debug(titleAOC.IsValid());
 			cemuLog_log(LogType::Force, "DLC: {}", titleAOC.GetPrintPath());
 			// mount AOC
 			if (!titleAOC.Mount(fmt::format("/vol/aoc{:016x}", titleAOC.GetAppTitleId()), "content", FSC_PRIORITY_PATCH))
 			{
 				cemuLog_log(LogType::Force, "Mounting failed");
-				return STATUS_CODE::UNABLE_TO_MOUNT;
+				return PREPARE_STATUS_CODE::UNABLE_TO_MOUNT;
 			}
 		}
 		else
 			cemuLog_log(LogType::Force, "DLC: Not present");
 		sForegroundTitleId = titleId;
-		return STATUS_CODE::SUCCESS;
+		return PREPARE_STATUS_CODE::SUCCESS;
 	}
 
     void UnmountForegroundTitle()
@@ -723,7 +724,7 @@ namespace CafeSystem
         }
     }
 
-	STATUS_CODE SetupExecutable()
+	PREPARE_STATUS_CODE SetupExecutable()
 	{
 		// set rpx path from cos.xml if available
 		_pathToBaseExecutable = _pathToExecutable;
@@ -755,7 +756,7 @@ namespace CafeSystem
 			}
 		}
 		LoadMainExecutable();
-		return STATUS_CODE::SUCCESS;
+		return PREPARE_STATUS_CODE::SUCCESS;
 	}
 
     void SetupMemorySpace()
@@ -769,7 +770,7 @@ namespace CafeSystem
         memory_unmapForCurrentTitle();
     }
 
-	STATUS_CODE PrepareForegroundTitle(TitleId titleId)
+	PREPARE_STATUS_CODE PrepareForegroundTitle(TitleId titleId)
 	{
 		CafeTitleList::WaitForMandatoryScan();
 		sLaunchModeIsStandalone = false;
@@ -780,21 +781,21 @@ namespace CafeSystem
         // mount mlc storage
         MountBaseDirectories();
         // mount title folders
-		STATUS_CODE r = LoadAndMountForegroundTitle(titleId);
-		if (r != STATUS_CODE::SUCCESS)
+		PREPARE_STATUS_CODE r = LoadAndMountForegroundTitle(titleId);
+		if (r != PREPARE_STATUS_CODE::SUCCESS)
 			return r;
 		gameProfile_load();
 		// setup memory space and PPC recompiler
         SetupMemorySpace();
         PPCRecompiler_init();
 		r = SetupExecutable(); // load RPX
-		if (r != STATUS_CODE::SUCCESS)
+		if (r != PREPARE_STATUS_CODE::SUCCESS)
 			return r;
 		InitVirtualMlcStorage();
-		return STATUS_CODE::SUCCESS;
+		return PREPARE_STATUS_CODE::SUCCESS;
 	}
 
-	STATUS_CODE PrepareForegroundTitleFromStandaloneRPX(const fs::path& path)
+	PREPARE_STATUS_CODE PrepareForegroundTitleFromStandaloneRPX(const fs::path& path)
 	{
 		sLaunchModeIsStandalone = true;
 		cemuLog_log(LogType::Force, "Launching executable in standalone mode due to incorrect layout or missing meta files");
@@ -812,7 +813,7 @@ namespace CafeSystem
 				if (!r)
 				{
 					cemuLog_log(LogType::Force, "Failed to mount {}", _pathToUtf8(contentPath));
-					return STATUS_CODE::UNABLE_TO_MOUNT;
+					return PREPARE_STATUS_CODE::UNABLE_TO_MOUNT;
 				}
 			}
 		}
@@ -824,7 +825,7 @@ namespace CafeSystem
 		// since a lot of systems (including save folder location) rely on a TitleId, we derive a placeholder id from the executable hash
 		auto execData = fsc_extractFile(_pathToExecutable.c_str());
 		if (!execData)
-			return STATUS_CODE::INVALID_RPX;
+			return PREPARE_STATUS_CODE::INVALID_RPX;
 		uint32 h = generateHashFromRawRPXData(execData->data(), execData->size());
 		sForegroundTitleId = 0xFFFFFFFF00000000ULL | (uint64)h;
 		cemuLog_log(LogType::Force, "Generated placeholder TitleId: {:016x}", sForegroundTitleId);
@@ -834,7 +835,7 @@ namespace CafeSystem
         // load executable
         SetupExecutable();
 		InitVirtualMlcStorage();
-		return STATUS_CODE::SUCCESS;
+		return PREPARE_STATUS_CODE::SUCCESS;
 	}
 
 	void _LaunchTitleThread()
@@ -843,7 +844,7 @@ namespace CafeSystem
 			module->TitleStart();
 		cemu_initForGame();
 		// enter scheduler
-		if (ActiveSettings::GetCPUMode() == CPUMode::MulticoreRecompiler)
+		if (ActiveSettings::GetCPUMode() == CPUMode::MulticoreRecompiler && !LaunchSettings::ForceInterpreter())
 			coreinit::OSSchedulerBegin(3);
 		else
 			coreinit::OSSchedulerBegin(1);

src/Cafe/CafeSystem.h

@@ -15,20 +15,19 @@ namespace CafeSystem
 		virtual void CafeRecreateCanvas() = 0;
 	};
 
-	enum class STATUS_CODE
+	enum class PREPARE_STATUS_CODE
 	{
 		SUCCESS,
 		INVALID_RPX,
 		UNABLE_TO_MOUNT, // failed to mount through TitleInfo (most likely caused by an invalid or outdated path)
-		//BAD_META_DATA, - the title list only stores titles with valid meta, so this error code is impossible
 	};
 
 	void Initialize();
 	void SetImplementation(SystemImplementation* impl);
     void Shutdown();
 
-	STATUS_CODE PrepareForegroundTitle(TitleId titleId);
-	STATUS_CODE PrepareForegroundTitleFromStandaloneRPX(const fs::path& path);
+	PREPARE_STATUS_CODE PrepareForegroundTitle(TitleId titleId);
+	PREPARE_STATUS_CODE PrepareForegroundTitleFromStandaloneRPX(const fs::path& path);
 	void LaunchForegroundTitle();
 	bool IsTitleRunning();
 

src/Cafe/Filesystem/FST/FST.cpp

@@ -3,8 +3,7 @@
 #include "Cemu/ncrypto/ncrypto.h"
 #include "Cafe/Filesystem/WUD/wud.h"
 #include "util/crypto/aes128.h"
-#include "openssl/evp.h" /* EVP_Digest */
-#include "openssl/sha.h" /* SHA1 / SHA256_DIGEST_LENGTH */
+#include "openssl/sha.h" /* SHA1 / SHA256 */
 #include "fstUtil.h"
 
 #include "FST.h"
@@ -141,7 +140,7 @@ struct DiscPartitionTableHeader
 	static constexpr uint32 MAGIC_VALUE = 0xCCA6E67B;
 
 	/* +0x00 */ uint32be magic;
-	/* +0x04 */ uint32be sectorSize; // must be 0x8000?
+	/* +0x04 */ uint32be blockSize; // must be 0x8000?
 	/* +0x08 */ uint8 partitionTableHash[20]; // hash of the data range at +0x800 to end of sector (0x8000)
 	/* +0x1C */ uint32be numPartitions;
 };
@@ -164,10 +163,10 @@ struct DiscPartitionHeader
 	static constexpr uint32 MAGIC_VALUE = 0xCC93A4F5;
 
 	/* +0x00 */ uint32be magic;
-	/* +0x04 */ uint32be sectorSize; // must match DISC_SECTOR_SIZE
+	/* +0x04 */ uint32be sectorSize; // must match DISC_SECTOR_SIZE for hashed blocks
 
 	/* +0x08 */ uint32be ukn008;
-	/* +0x0C */ uint32be ukn00C;
+	/* +0x0C */ uint32be ukn00C; // h3 array size?
 	/* +0x10 */ uint32be h3HashNum;
 	/* +0x14 */ uint32be fstSize; // in bytes
 	/* +0x18 */ uint32be fstSector; // relative to partition start
@@ -178,13 +177,15 @@ struct DiscPartitionHeader
 	/* +0x24 */ uint8 fstHashType;
 	/* +0x25 */ uint8 fstEncryptionType; // purpose of this isn't really understood. Maybe it controls which key is being used? (1 -> disc key, 2 -> partition key)
 
-	/* +0x26 */ uint8 versionA;
-	/* +0x27 */ uint8 ukn027; // also a version field?
+	/* +0x26 */ uint8be versionA;
+	/* +0x27 */ uint8be ukn027; // also a version field?
 
 	// there is an array at +0x40 ? Related to H3 list. Also related to value at +0x0C and h3HashNum
+	/* +0x28 */ uint8be _uknOrPadding028[0x18];
+	/* +0x40 */ uint8be h3HashArray[32]; // dynamic size. Only present if fstHashType != 0
 };
 
-static_assert(sizeof(DiscPartitionHeader) == 0x28);
+static_assert(sizeof(DiscPartitionHeader) == 0x40+0x20);
 
 bool FSTVolume::FindDiscKey(const fs::path& path, NCrypto::AesKey& discTitleKey)
 {
@@ -269,7 +270,7 @@ FSTVolume* FSTVolume::OpenFromDiscImage(const fs::path& path, NCrypto::AesKey& d
 		cemuLog_log(LogType::Force, "Disc image rejected because decryption failed");
 		return nullptr;
 	}
-	if (partitionHeader->sectorSize != DISC_SECTOR_SIZE)
+	if (partitionHeader->blockSize != DISC_SECTOR_SIZE)
 	{
 		cemuLog_log(LogType::Force, "Disc image rejected because partition sector size is invalid");
 		return nullptr;
@@ -336,6 +337,9 @@ FSTVolume* FSTVolume::OpenFromDiscImage(const fs::path& path, NCrypto::AesKey& d
 	cemu_assert_debug(partitionHeaderSI.fstEncryptionType == 1);
 	// todo - check other fields?
 
+	if(partitionHeaderSI.fstHashType == 0 && partitionHeaderSI.h3HashNum != 0)
+		cemuLog_log(LogType::Force, "FST: Partition uses unhashed blocks but stores a non-zero amount of H3 hashes");
+
 	// GM partition
 	DiscPartitionHeader partitionHeaderGM{};
 	if (!readPartitionHeader(partitionHeaderGM, gmPartitionIndex))
@@ -349,9 +353,10 @@ FSTVolume* FSTVolume::OpenFromDiscImage(const fs::path& path, NCrypto::AesKey& d
 	// if decryption is necessary
 	// load SI FST
 	dataSource->SetBaseOffset((uint64)partitionArray[siPartitionIndex].partitionAddress * DISC_SECTOR_SIZE);
-	auto siFST = OpenFST(dataSource.get(), (uint64)partitionHeaderSI.fstSector * DISC_SECTOR_SIZE, partitionHeaderSI.fstSize, &discTitleKey, static_cast<FSTVolume::ClusterHashMode>(partitionHeaderSI.fstHashType));
+	auto siFST = OpenFST(dataSource.get(), (uint64)partitionHeaderSI.fstSector * DISC_SECTOR_SIZE, partitionHeaderSI.fstSize, &discTitleKey, static_cast<FSTVolume::ClusterHashMode>(partitionHeaderSI.fstHashType), nullptr);
 	if (!siFST)
 		return nullptr;
+	cemu_assert_debug(!(siFST->HashIsDisabled() && partitionHeaderSI.h3HashNum != 0)); // if hash is disabled, no H3 data may be present
 	// load ticket file for partition that we want to decrypt
 	NCrypto::ETicketParser ticketParser;
 	std::vector<uint8> ticketData = siFST->ExtractFile(fmt::format("{:02x}/title.tik", gmPartitionIndex));
@@ -360,16 +365,32 @@ FSTVolume* FSTVolume::OpenFromDiscImage(const fs::path& path, NCrypto::AesKey& d
 		cemuLog_log(LogType::Force, "Disc image ticket file is invalid");
 		return nullptr;
 	}
+#if 0
+	// each SI partition seems to contain a title.tmd that we could parse and which should have information about the associated GM partition
+	// but the console seems to ignore this file for disc images, at least when mounting, so we shouldn't rely on it either
+	std::vector<uint8> tmdData = siFST->ExtractFile(fmt::format("{:02x}/title.tmd", gmPartitionIndex));
+	if (tmdData.empty())
+	{
+		cemuLog_log(LogType::Force, "Disc image TMD file is missing");
+		return nullptr;
+	}
+	// parse TMD
+	NCrypto::TMDParser tmdParser;
+	if (!tmdParser.parse(tmdData.data(), tmdData.size()))
+	{
+		cemuLog_log(LogType::Force, "Disc image TMD file is invalid");
+		return nullptr;
+	}
+#endif
 	delete siFST;
-
 	NCrypto::AesKey gmTitleKey;
 	ticketParser.GetTitleKey(gmTitleKey);
-
 	// load GM partition
 	dataSource->SetBaseOffset((uint64)partitionArray[gmPartitionIndex].partitionAddress * DISC_SECTOR_SIZE);
-	FSTVolume* r = OpenFST(std::move(dataSource), (uint64)partitionHeaderGM.fstSector * DISC_SECTOR_SIZE, partitionHeaderGM.fstSize, &gmTitleKey, static_cast<FSTVolume::ClusterHashMode>(partitionHeaderGM.fstHashType));
+	FSTVolume* r = OpenFST(std::move(dataSource), (uint64)partitionHeaderGM.fstSector * DISC_SECTOR_SIZE, partitionHeaderGM.fstSize, &gmTitleKey, static_cast<FSTVolume::ClusterHashMode>(partitionHeaderGM.fstHashType), nullptr);
 	if (r)
 		SET_FST_ERROR(OK);
+	cemu_assert_debug(!(r->HashIsDisabled() && partitionHeaderGM.h3HashNum != 0)); // if hash is disabled, no H3 data may be present
 	return r;
 }
 
@@ -426,15 +447,15 @@ FSTVolume* FSTVolume::OpenFromContentFolder(fs::path folderPath, ErrorCode* erro
 	}
 	// load FST
 	// fstSize = size of first cluster?
-	FSTVolume* fstVolume = FSTVolume::OpenFST(std::move(dataSource), 0, fstSize, &titleKey, fstHashMode);
+	FSTVolume* fstVolume = FSTVolume::OpenFST(std::move(dataSource), 0, fstSize, &titleKey, fstHashMode, &tmdParser);
 	if (fstVolume)
 		SET_FST_ERROR(OK);
 	return fstVolume;
 }
 
-FSTVolume* FSTVolume::OpenFST(FSTDataSource* dataSource, uint64 fstOffset, uint32 fstSize, NCrypto::AesKey* partitionTitleKey, ClusterHashMode fstHashMode)
+FSTVolume* FSTVolume::OpenFST(FSTDataSource* dataSource, uint64 fstOffset, uint32 fstSize, NCrypto::AesKey* partitionTitleKey, ClusterHashMode fstHashMode, NCrypto::TMDParser* optionalTMD)
 {
-	cemu_assert_debug(fstHashMode != ClusterHashMode::RAW || fstHashMode != ClusterHashMode::RAW2);
+	cemu_assert_debug(fstHashMode != ClusterHashMode::RAW || fstHashMode != ClusterHashMode::RAW_STREAM);
 	if (fstSize < sizeof(FSTHeader))
 		return nullptr;
 	constexpr uint64 FST_CLUSTER_OFFSET = 0;
@@ -465,6 +486,34 @@ FSTVolume* FSTVolume::OpenFST(FSTDataSource* dataSource, uint64 fstOffset, uint3
 		clusterTable[i].offset = clusterDataTable[i].offset;
 		clusterTable[i].size = clusterDataTable[i].size;
 		clusterTable[i].hashMode = static_cast<FSTVolume::ClusterHashMode>((uint8)clusterDataTable[i].hashMode);
+		clusterTable[i].hasContentHash = false; // from the TMD file (H4?)
+	}
+	// if the TMD is available (when opening .app files) we can use the extra info from it to validate unhashed clusters
+	// each content entry in the TMD corresponds to one cluster used by the FST
+	if(optionalTMD)
+	{
+		if(numCluster != optionalTMD->GetContentList().size())
+		{
+			cemuLog_log(LogType::Force, "FST: Number of clusters does not match TMD content list");
+			return nullptr;
+		}
+		auto& contentList = optionalTMD->GetContentList();
+		for(size_t i=0; i<contentList.size(); i++)
+		{
+			auto& cluster = clusterTable[i];
+			auto& content = contentList[i];
+			cluster.hasContentHash = true;
+			cluster.contentHashIsSHA1 = HAS_FLAG(contentList[i].contentFlags, NCrypto::TMDParser::TMDContentFlags::FLAG_SHA1);
+			cluster.contentSize = content.size;
+			static_assert(sizeof(content.hash32) == sizeof(cluster.contentHash32));
+			memcpy(cluster.contentHash32, content.hash32, sizeof(cluster.contentHash32));
+			// if unhashed mode, then initialize the hash context
+			if(cluster.hashMode == ClusterHashMode::RAW || cluster.hashMode == ClusterHashMode::RAW_STREAM)
+			{
+				cluster.singleHashCtx.reset(EVP_MD_CTX_new());
+				EVP_DigestInit_ex(cluster.singleHashCtx.get(), cluster.contentHashIsSHA1 ? EVP_sha1() : EVP_sha256(), nullptr);
+			}
+		}
 	}
 	// preprocess FST table
 	FSTHeader_FileEntry* fileTable = (FSTHeader_FileEntry*)(clusterDataTable + numCluster);
@@ -491,16 +540,17 @@ FSTVolume* FSTVolume::OpenFST(FSTDataSource* dataSource, uint64 fstOffset, uint3
 	fstVolume->m_offsetFactor = fstHeader->offsetFactor;
 	fstVolume->m_sectorSize = DISC_SECTOR_SIZE;
 	fstVolume->m_partitionTitlekey = *partitionTitleKey;
-	std::swap(fstVolume->m_cluster, clusterTable);
-	std::swap(fstVolume->m_entries, fstEntries);
-	std::swap(fstVolume->m_nameStringTable, nameStringTable);
+	fstVolume->m_hashIsDisabled = fstHeader->hashIsDisabled != 0;
+	fstVolume->m_cluster = std::move(clusterTable);
+	fstVolume->m_entries = std::move(fstEntries);
+	fstVolume->m_nameStringTable = std::move(nameStringTable);
 	return fstVolume;
 }
 
-FSTVolume* FSTVolume::OpenFST(std::unique_ptr<FSTDataSource> dataSource, uint64 fstOffset, uint32 fstSize, NCrypto::AesKey* partitionTitleKey, ClusterHashMode fstHashMode)
+FSTVolume* FSTVolume::OpenFST(std::unique_ptr<FSTDataSource> dataSource, uint64 fstOffset, uint32 fstSize, NCrypto::AesKey* partitionTitleKey, ClusterHashMode fstHashMode, NCrypto::TMDParser* optionalTMD)
 {
 	FSTDataSource* ds = dataSource.release();
-	FSTVolume* fstVolume = OpenFST(ds, fstOffset, fstSize, partitionTitleKey, fstHashMode);
+	FSTVolume* fstVolume = OpenFST(ds, fstOffset, fstSize, partitionTitleKey, fstHashMode, optionalTMD);
 	if (!fstVolume)
 	{
 		delete ds;
@@ -757,7 +807,7 @@ uint32 FSTVolume::ReadFile(FSTFileHandle& fileHandle, uint32 offset, uint32 size
 		return 0;
 	cemu_assert_debug(!HAS_FLAG(entry.GetFlags(), FSTEntry::FLAGS::FLAG_LINK));
 	FSTCluster& cluster = m_cluster[entry.fileInfo.clusterIndex];
-	if (cluster.hashMode == ClusterHashMode::RAW || cluster.hashMode == ClusterHashMode::RAW2)
+	if (cluster.hashMode == ClusterHashMode::RAW || cluster.hashMode == ClusterHashMode::RAW_STREAM)
 		return ReadFile_HashModeRaw(entry.fileInfo.clusterIndex, entry, offset, size, dataOut);
 	else if (cluster.hashMode == ClusterHashMode::HASH_INTERLEAVED)
 		return ReadFile_HashModeHashed(entry.fileInfo.clusterIndex, entry, offset, size, dataOut);
@@ -765,87 +815,15 @@ uint32 FSTVolume::ReadFile(FSTFileHandle& fileHandle, uint32 offset, uint32 size
 	return 0;
 }
 
-uint32 FSTVolume::ReadFile_HashModeRaw(uint32 clusterIndex, FSTEntry& entry, uint32 readOffset, uint32 readSize, void* dataOut)
-{
-	const uint32 readSizeInput = readSize;
-	uint8* dataOutU8 = (uint8*)dataOut;
-	if (readOffset >= entry.fileInfo.fileSize)
-		return 0;
-	else if ((readOffset + readSize) >= entry.fileInfo.fileSize)
-		readSize = (entry.fileInfo.fileSize - readOffset);
-
-	const FSTCluster& cluster = m_cluster[clusterIndex];
-	uint64 clusterOffset = (uint64)cluster.offset * m_sectorSize;
-	uint64 absFileOffset = entry.fileInfo.fileOffset * m_offsetFactor + readOffset;
-
-	// make sure the raw range we read is aligned to AES block size (16)
-	uint64 readAddrStart = absFileOffset & ~0xF;
-	uint64 readAddrEnd = (absFileOffset + readSize + 0xF) & ~0xF;
-
-	bool usesInitialIV = readOffset < 16;
-	if (!usesInitialIV)
-		readAddrStart -= 16; // read previous AES block since we require it for the IV
-	uint32 prePadding = (uint32)(absFileOffset - readAddrStart); // number of extra bytes we read before readOffset (for AES alignment and IV calculation)
-	uint32 postPadding = (uint32)(readAddrEnd - (absFileOffset + readSize));
-
-	uint8 readBuffer[64 * 1024];
-	// read first chunk
-	// if file read offset (readOffset) is within the first AES-block then use initial IV calculated from cluster index
-	// otherwise read previous AES-block is the IV (AES-CBC)
-	uint64 readAddrCurrent = readAddrStart;
-	uint32 rawBytesToRead = (uint32)std::min((readAddrEnd - readAddrStart), (uint64)sizeof(readBuffer));
-	if (m_dataSource->readData(clusterIndex, clusterOffset, readAddrCurrent, readBuffer, rawBytesToRead) != rawBytesToRead)
-	{
-		cemuLog_log(LogType::Force, "FST read error in raw content");
-		return 0;
-	}
-	readAddrCurrent += rawBytesToRead;
-
-	uint8 iv[16]{};
-	if (usesInitialIV)
-	{
-		// for the first AES block, the IV is initialized from cluster index
-		iv[0] = (uint8)(clusterIndex >> 8);
-		iv[1] = (uint8)(clusterIndex >> 0);
-		AES128_CBC_decrypt_updateIV(readBuffer, readBuffer, rawBytesToRead, m_partitionTitlekey.b, iv);
-		std::memcpy(dataOutU8, readBuffer + prePadding, rawBytesToRead - prePadding - postPadding);
-		dataOutU8 += (rawBytesToRead - prePadding - postPadding);
-		readSize -= (rawBytesToRead - prePadding - postPadding);
-	}
-	else
-	{
-		// IV is initialized from previous AES block (AES-CBC)
-		std::memcpy(iv, readBuffer, 16);
-		AES128_CBC_decrypt_updateIV(readBuffer + 16, readBuffer + 16, rawBytesToRead - 16, m_partitionTitlekey.b, iv);
-		std::memcpy(dataOutU8, readBuffer + prePadding, rawBytesToRead - prePadding - postPadding);
-		dataOutU8 += (rawBytesToRead - prePadding - postPadding);
-		readSize -= (rawBytesToRead - prePadding - postPadding);
-	}
-
-	// read remaining chunks
-	while (readSize > 0)
-	{
-		uint32 bytesToRead = (uint32)std::min((uint32)sizeof(readBuffer), readSize);
-		uint32 alignedBytesToRead = (bytesToRead + 15) & ~0xF;
-		if (m_dataSource->readData(clusterIndex, clusterOffset, readAddrCurrent, readBuffer, alignedBytesToRead) != alignedBytesToRead)
-		{
-			cemuLog_log(LogType::Force, "FST read error in raw content");
-			return 0;
-		}
-		AES128_CBC_decrypt_updateIV(readBuffer, readBuffer, alignedBytesToRead, m_partitionTitlekey.b, iv);
-		std::memcpy(dataOutU8, readBuffer, bytesToRead);
-		dataOutU8 += bytesToRead;
-		readSize -= bytesToRead;
-		readAddrCurrent += alignedBytesToRead;
-	}
-
-	return readSizeInput - readSize;
-}
-
 constexpr size_t BLOCK_SIZE = 0x10000;
 constexpr size_t BLOCK_HASH_SIZE = 0x0400;
 constexpr size_t BLOCK_FILE_SIZE = 0xFC00;
 
+struct FSTRawBlock
+{
+	std::vector<uint8> rawData; // unhashed block size depends on sector size field in partition header
+};
+
 struct FSTHashedBlock
 {
 	uint8 rawData[BLOCK_SIZE];
@@ -887,12 +865,160 @@ struct FSTHashedBlock
 
 static_assert(sizeof(FSTHashedBlock) == BLOCK_SIZE);
 
+struct FSTCachedRawBlock
+{
+	FSTRawBlock blockData;
+	uint8 ivForNextBlock[16];
+	uint64 lastAccess;
+};
+
 struct FSTCachedHashedBlock 
 {
 	FSTHashedBlock blockData;
 	uint64 lastAccess;
 };
 
+// Checks cache fill state and if necessary drops least recently accessed block from the cache. Optionally allows to recycle the released cache entry to cut down cost of memory allocation and clearing
+void FSTVolume::TrimCacheIfRequired(FSTCachedRawBlock** droppedRawBlock, FSTCachedHashedBlock** droppedHashedBlock)
+{
+	// calculate size used by cache
+	size_t cacheSize = 0;
+	for (auto& itr : m_cacheDecryptedRawBlocks)
+		cacheSize += itr.second->blockData.rawData.size();
+	for (auto& itr : m_cacheDecryptedHashedBlocks)
+		cacheSize += sizeof(FSTCachedHashedBlock) + sizeof(FSTHashedBlock);
+	// only trim if cache is full (larger than 2MB)
+	if (cacheSize < 2*1024*1024) // 2MB
+		return;
+	// scan both cache lists to find least recently accessed block to drop
+	auto dropRawItr = std::min_element(m_cacheDecryptedRawBlocks.begin(), m_cacheDecryptedRawBlocks.end(), [](const auto& a, const auto& b) -> bool
+										{ return a.second->lastAccess < b.second->lastAccess; });
+	auto dropHashedItr = std::min_element(m_cacheDecryptedHashedBlocks.begin(), m_cacheDecryptedHashedBlocks.end(), [](const auto& a, const auto& b) -> bool
+										{ return a.second->lastAccess < b.second->lastAccess; });
+	uint64 lastAccess = std::numeric_limits<uint64>::max();
+	if(dropRawItr != m_cacheDecryptedRawBlocks.end())
+		lastAccess = dropRawItr->second->lastAccess;
+	if(dropHashedItr != m_cacheDecryptedHashedBlocks.end())
+		lastAccess = std::min<uint64>(lastAccess, dropHashedItr->second->lastAccess);
+	if(dropRawItr != m_cacheDecryptedRawBlocks.end() && dropRawItr->second->lastAccess == lastAccess)
+	{
+		if (droppedRawBlock)
+			*droppedRawBlock = dropRawItr->second;
+		else
+			delete dropRawItr->second;
+		m_cacheDecryptedRawBlocks.erase(dropRawItr);
+		return;
+	}
+	else if(dropHashedItr != m_cacheDecryptedHashedBlocks.end() && dropHashedItr->second->lastAccess == lastAccess)
+	{
+		if (droppedHashedBlock)
+			*droppedHashedBlock = dropHashedItr->second;
+		else
+			delete dropHashedItr->second;
+		m_cacheDecryptedHashedBlocks.erase(dropHashedItr);
+	}
+}
+
+void FSTVolume::DetermineUnhashedBlockIV(uint32 clusterIndex, uint32 blockIndex, uint8 ivOut[16])
+{
+	memset(ivOut, 0, sizeof(ivOut));
+	if(blockIndex == 0)
+	{
+		ivOut[0] = (uint8)(clusterIndex >> 8);
+		ivOut[1] = (uint8)(clusterIndex >> 0);
+	}
+	else
+	{
+		// the last 16 encrypted bytes of the previous block are the IV (AES CBC)
+		// if the previous block is cached we can grab the IV from there. Otherwise we have to read the 16 bytes from the data source
+		uint32 prevBlockIndex = blockIndex - 1;
+		uint64 cacheBlockId = ((uint64)clusterIndex << (64 - 16)) | (uint64)prevBlockIndex;
+		auto itr = m_cacheDecryptedRawBlocks.find(cacheBlockId);
+		if (itr != m_cacheDecryptedRawBlocks.end())
+		{
+			memcpy(ivOut, itr->second->ivForNextBlock, 16);
+		}
+		else
+		{
+			cemu_assert(m_sectorSize >= 16);
+			uint64 clusterOffset = (uint64)m_cluster[clusterIndex].offset * m_sectorSize;
+			uint8 prevIV[16];
+			if (m_dataSource->readData(clusterIndex, clusterOffset, blockIndex * m_sectorSize - 16, prevIV, 16) != 16)
+			{
+				cemuLog_log(LogType::Force, "Failed to read IV for raw FST block");
+				m_detectedCorruption = true;
+				return;
+			}
+			memcpy(ivOut, prevIV, 16);
+		}
+	}
+}
+
+FSTCachedRawBlock* FSTVolume::GetDecryptedRawBlock(uint32 clusterIndex, uint32 blockIndex)
+{
+	FSTCluster& cluster = m_cluster[clusterIndex];
+	uint64 clusterOffset = (uint64)cluster.offset * m_sectorSize;
+	// generate id for cache
+	uint64 cacheBlockId = ((uint64)clusterIndex << (64 - 16)) | (uint64)blockIndex;
+	// lookup block in cache
+	FSTCachedRawBlock* block = nullptr;
+	auto itr = m_cacheDecryptedRawBlocks.find(cacheBlockId);
+	if (itr != m_cacheDecryptedRawBlocks.end())
+	{
+		block = itr->second;
+		block->lastAccess = ++m_cacheAccessCounter;
+		return block;
+	}
+	// if cache already full, drop least recently accessed block and recycle FSTCachedRawBlock object if possible
+	TrimCacheIfRequired(&block, nullptr);
+	if (!block)
+		block = new FSTCachedRawBlock();
+	block->blockData.rawData.resize(m_sectorSize);
+	// block not cached, read new
+	block->lastAccess = ++m_cacheAccessCounter;
+	if (m_dataSource->readData(clusterIndex, clusterOffset, blockIndex * m_sectorSize, block->blockData.rawData.data(), m_sectorSize) != m_sectorSize)
+	{
+		cemuLog_log(LogType::Force, "Failed to read raw FST block");
+		delete block;
+		m_detectedCorruption = true;
+		return nullptr;
+	}
+	// decrypt hash data
+	uint8 iv[16]{};
+	DetermineUnhashedBlockIV(clusterIndex, blockIndex, iv);
+	memcpy(block->ivForNextBlock, block->blockData.rawData.data() + m_sectorSize - 16, 16);
+	AES128_CBC_decrypt(block->blockData.rawData.data(), block->blockData.rawData.data(), m_sectorSize, m_partitionTitlekey.b, iv);
+	// if this is the next block, then hash it
+	if(cluster.hasContentHash)
+	{
+		if(cluster.singleHashNumBlocksHashed == blockIndex)
+		{
+			cemu_assert_debug(!(cluster.contentSize % m_sectorSize)); // size should be multiple of sector size? Regardless, the hashing code below can handle non-aligned sizes
+			bool isLastBlock = blockIndex == (std::max<uint32>(cluster.contentSize / m_sectorSize, 1) - 1);
+			uint32 hashSize = m_sectorSize;
+			if(isLastBlock)
+				hashSize = cluster.contentSize - (uint64)blockIndex*m_sectorSize;
+			EVP_DigestUpdate(cluster.singleHashCtx.get(), block->blockData.rawData.data(), hashSize);
+			cluster.singleHashNumBlocksHashed++;
+			if(isLastBlock)
+			{
+				uint8 hash[32];
+				EVP_DigestFinal_ex(cluster.singleHashCtx.get(), hash, nullptr);
+				if(memcmp(hash, cluster.contentHash32, cluster.contentHashIsSHA1 ? 20 : 32) != 0)
+				{
+					cemuLog_log(LogType::Force, "FST: Raw section hash mismatch");
+					delete block;
+					m_detectedCorruption = true;
+					return nullptr;
+				}
+			}
+		}
+	}
+	// register in cache
+	m_cacheDecryptedRawBlocks.emplace(cacheBlockId, block);
+	return block;
+}
+
 FSTCachedHashedBlock* FSTVolume::GetDecryptedHashedBlock(uint32 clusterIndex, uint32 blockIndex)
 {
 	const FSTCluster& cluster = m_cluster[clusterIndex];
@@ -908,22 +1034,17 @@ FSTCachedHashedBlock* FSTVolume::GetDecryptedHashedBlock(uint32 clusterIndex, ui
 		block->lastAccess = ++m_cacheAccessCounter;
 		return block;
 	}
-	// if cache already full, drop least recently accessed block (but recycle the FSTHashedBlock* object)
-	if (m_cacheDecryptedHashedBlocks.size() >= 16)
-	{
-		auto dropItr = std::min_element(m_cacheDecryptedHashedBlocks.begin(), m_cacheDecryptedHashedBlocks.end(), [](const auto& a, const auto& b) -> bool
-		{ return a.second->lastAccess < b.second->lastAccess; });
-		block = dropItr->second;
-		m_cacheDecryptedHashedBlocks.erase(dropItr);
-	}
-	else
+	// if cache already full, drop least recently accessed block and recycle FSTCachedHashedBlock object if possible
+	TrimCacheIfRequired(nullptr, &block);
+	if (!block)
 		block = new FSTCachedHashedBlock();
 	// block not cached, read new
 	block->lastAccess = ++m_cacheAccessCounter;
 	if (m_dataSource->readData(clusterIndex, clusterOffset, blockIndex * BLOCK_SIZE, block->blockData.rawData, BLOCK_SIZE) != BLOCK_SIZE)
 	{
-		cemuLog_log(LogType::Force, "Failed to read FST block");
+		cemuLog_log(LogType::Force, "Failed to read hashed FST block");
 		delete block;
+		m_detectedCorruption = true;
 		return nullptr;
 	}
 	// decrypt hash data
@@ -931,11 +1052,46 @@ FSTCachedHashedBlock* FSTVolume::GetDecryptedHashedBlock(uint32 clusterIndex, ui
 	AES128_CBC_decrypt(block->blockData.getHashData(), block->blockData.getHashData(), BLOCK_HASH_SIZE, m_partitionTitlekey.b, iv);
 	// decrypt file data
 	AES128_CBC_decrypt(block->blockData.getFileData(), block->blockData.getFileData(), BLOCK_FILE_SIZE, m_partitionTitlekey.b, block->blockData.getH0Hash(blockIndex%16));
+	// compare with H0 to verify data integrity
+	NCrypto::CHash160 h0;
+	SHA1(block->blockData.getFileData(), BLOCK_FILE_SIZE, h0.b);
+	uint32 h0Index = (blockIndex % 4096);
+	if (memcmp(h0.b, block->blockData.getH0Hash(h0Index & 0xF), sizeof(h0.b)) != 0)
+	{
+		cemuLog_log(LogType::Force, "FST: Hash H0 mismatch in hashed block (section {} index {})", clusterIndex, blockIndex);
+		delete block;
+		m_detectedCorruption = true;
+		return nullptr;
+	}
 	// register in cache
 	m_cacheDecryptedHashedBlocks.emplace(cacheBlockId, block);
 	return block;
 }
 
+uint32 FSTVolume::ReadFile_HashModeRaw(uint32 clusterIndex, FSTEntry& entry, uint32 readOffset, uint32 readSize, void* dataOut)
+{
+	uint8* dataOutU8 = (uint8*)dataOut;
+	if (readOffset >= entry.fileInfo.fileSize)
+		return 0;
+	else if ((readOffset + readSize) >= entry.fileInfo.fileSize)
+		readSize = (entry.fileInfo.fileSize - readOffset);
+	uint64 absFileOffset = entry.fileInfo.fileOffset * m_offsetFactor + readOffset;
+	uint32 remainingReadSize = readSize;
+	while (remainingReadSize > 0)
+	{
+		const FSTCachedRawBlock* rawBlock = this->GetDecryptedRawBlock(clusterIndex, absFileOffset/m_sectorSize);
+		if (!rawBlock)
+			break;
+		uint32 blockOffset = (uint32)(absFileOffset % m_sectorSize);
+		uint32 bytesToRead = std::min<uint32>(remainingReadSize, m_sectorSize - blockOffset);
+		std::memcpy(dataOutU8, rawBlock->blockData.rawData.data() + blockOffset, bytesToRead);
+		dataOutU8 += bytesToRead;
+		remainingReadSize -= bytesToRead;
+		absFileOffset += bytesToRead;
+	}
+	return readSize - remainingReadSize;
+}
+
 uint32 FSTVolume::ReadFile_HashModeHashed(uint32 clusterIndex, FSTEntry& entry, uint32 readOffset, uint32 readSize, void* dataOut)
 {
 	/*
@@ -966,7 +1122,6 @@ uint32 FSTVolume::ReadFile_HashModeHashed(uint32 clusterIndex, FSTEntry& entry,
 	*/
 
 	const FSTCluster& cluster = m_cluster[clusterIndex];
-	uint64 clusterBaseOffset = (uint64)cluster.offset * m_sectorSize;
 	uint64 fileReadOffset = entry.fileInfo.fileOffset * m_offsetFactor + readOffset;
 	uint32 blockIndex = (uint32)(fileReadOffset / BLOCK_FILE_SIZE);
 	uint32 bytesRemaining = readSize;
@@ -1019,6 +1174,8 @@ bool FSTVolume::Next(FSTDirectoryIterator& directoryIterator, FSTFileHandle& fil
 
 FSTVolume::~FSTVolume()
 {
+	for (auto& itr : m_cacheDecryptedRawBlocks)
+		delete itr.second;
 	for (auto& itr : m_cacheDecryptedHashedBlocks)
 		delete itr.second;
 	if (m_sourceIsOwned)

src/Cafe/Filesystem/FST/FST.h

@@ -1,5 +1,6 @@
 #pragma once
 #include "Cemu/ncrypto/ncrypto.h"
+#include "openssl/evp.h"
 
 struct FSTFileHandle
 {
@@ -45,6 +46,7 @@ public:
 	~FSTVolume();
 
 	uint32 GetFileCount() const;
+	bool HasCorruption() const { return m_detectedCorruption; }
 
 	bool OpenFile(std::string_view path, FSTFileHandle& fileHandleOut, bool openOnlyFiles = false);
 
@@ -86,15 +88,25 @@ private:
 	enum class ClusterHashMode : uint8
 	{
 		RAW = 0, // raw data + encryption, no hashing?
-		RAW2 = 1, // raw data + encryption, with hash stored in tmd?
+	  	RAW_STREAM = 1, // raw data + encryption, with hash stored in tmd?
 		HASH_INTERLEAVED = 2, // hashes + raw interleaved in 0x10000 blocks (0x400 bytes of hashes at the beginning, followed by 0xFC00 bytes of data)
 	};
 
 	struct FSTCluster
 	{
+		FSTCluster() : singleHashCtx(nullptr, &EVP_MD_CTX_free) {}
+
 		uint32 offset;
 		uint32 size;
 		ClusterHashMode hashMode;
+		// extra data if TMD is available
+		bool hasContentHash;
+		uint8 contentHash32[32];
+		bool contentHashIsSHA1; // if true then it's SHA1 (with extra bytes zeroed out), otherwise it's SHA256
+		uint64 contentSize; // size of the content (in blocks)
+		// hash context for single hash mode (content hash must be available)
+		std::unique_ptr<EVP_MD_CTX, decltype(&EVP_MD_CTX_free)> singleHashCtx; // unique_ptr to make this move-only
+		uint32 singleHashNumBlocksHashed{0};
 	};
 
 	struct FSTEntry
@@ -164,17 +176,30 @@ private:
 	bool m_sourceIsOwned{};
 	uint32 m_sectorSize{}; // for cluster offsets
 	uint32 m_offsetFactor{}; // for file offsets
+	bool m_hashIsDisabled{}; // disables hash verification (for all clusters of this volume?)
 	std::vector<FSTCluster> m_cluster;
 	std::vector<FSTEntry> m_entries;
 	std::vector<char> m_nameStringTable;
 	NCrypto::AesKey m_partitionTitlekey;
+	bool m_detectedCorruption{false};
 
-	/* Cache for decrypted hashed blocks */
+	bool HashIsDisabled() const
+	{
+		return m_hashIsDisabled;
+	}
+
+	/* Cache for decrypted raw and hashed blocks */
+	std::unordered_map<uint64, struct FSTCachedRawBlock*> m_cacheDecryptedRawBlocks;
 	std::unordered_map<uint64, struct FSTCachedHashedBlock*> m_cacheDecryptedHashedBlocks;
 	uint64 m_cacheAccessCounter{};
 
+	void DetermineUnhashedBlockIV(uint32 clusterIndex, uint32 blockIndex, uint8 ivOut[16]);
+
+	struct FSTCachedRawBlock* GetDecryptedRawBlock(uint32 clusterIndex, uint32 blockIndex);
 	struct FSTCachedHashedBlock* GetDecryptedHashedBlock(uint32 clusterIndex, uint32 blockIndex);
 
+	void TrimCacheIfRequired(struct FSTCachedRawBlock** droppedRawBlock, struct FSTCachedHashedBlock** droppedHashedBlock);
+
 	/* File reading */
 	uint32 ReadFile_HashModeRaw(uint32 clusterIndex, FSTEntry& entry, uint32 readOffset, uint32 readSize, void* dataOut);
 	uint32 ReadFile_HashModeHashed(uint32 clusterIndex, FSTEntry& entry, uint32 readOffset, uint32 readSize, void* dataOut);
@@ -185,7 +210,10 @@ private:
 		/* +0x00 */ uint32be magic;
 		/* +0x04 */ uint32be offsetFactor;
 		/* +0x08 */ uint32be numCluster;
-		/* +0x0C */ uint32be ukn0C;
+		/* +0x0C */ uint8be hashIsDisabled;
+		/* +0x0D */ uint8be ukn0D;
+		/* +0x0E */ uint8be ukn0E;
+		/* +0x0F */ uint8be ukn0F;
 		/* +0x10 */ uint32be ukn10;
 		/* +0x14 */ uint32be ukn14;
 		/* +0x18 */ uint32be ukn18;
@@ -262,8 +290,8 @@ private:
 
 	static_assert(sizeof(FSTHeader_FileEntry) == 0x10);
 
-	static FSTVolume* OpenFST(FSTDataSource* dataSource, uint64 fstOffset, uint32 fstSize, NCrypto::AesKey* partitionTitleKey, ClusterHashMode fstHashMode);
-	static FSTVolume* OpenFST(std::unique_ptr<FSTDataSource> dataSource, uint64 fstOffset, uint32 fstSize, NCrypto::AesKey* partitionTitleKey, ClusterHashMode fstHashMode);
+	static FSTVolume* OpenFST(FSTDataSource* dataSource, uint64 fstOffset, uint32 fstSize, NCrypto::AesKey* partitionTitleKey, ClusterHashMode fstHashMode, NCrypto::TMDParser* optionalTMD);
+	static FSTVolume* OpenFST(std::unique_ptr<FSTDataSource> dataSource, uint64 fstOffset, uint32 fstSize, NCrypto::AesKey* partitionTitleKey, ClusterHashMode fstHashMode, NCrypto::TMDParser* optionalTMD);
 	static bool ProcessFST(FSTHeader_FileEntry* fileTable, uint32 numFileEntries, uint32 numCluster, std::vector<char>& nameStringTable, std::vector<FSTEntry>& fstEntries);
 
 	bool MatchFSTEntryName(FSTEntry& entry, std::string_view comparedName)

src/Cafe/GameProfile/GameProfile.cpp

@@ -140,7 +140,7 @@ bool gameProfile_loadEnumOption(IniParser& iniParser, const char* optionName, T&
 	for(const T& v : T())
 	{
 		// test integer option
-		if (boost::iequals(fmt::format("{}", static_cast<typename std::underlying_type<T>::type>(v)), *option_value))
+		if (boost::iequals(fmt::format("{}", fmt::underlying(v)), *option_value))
 		{
 			option = v;
 			return true;

src/Cafe/GraphicPack/GraphicPack2.cpp

@@ -345,7 +345,7 @@ GraphicPack2::GraphicPack2(fs::path rulesPath, IniParser& rules)
 			const auto preset_name = rules.FindOption("name");
 			if (!preset_name)
 			{
-				cemuLog_log(LogType::Force, "Graphic pack \"{}\": Preset in line {} skipped because it has no name option defined", m_name, rules.GetCurrentSectionLineNumber());
+				cemuLog_log(LogType::Force, "Graphic pack \"{}\": Preset in line {} skipped because it has no name option defined", GetNormalizedPathString(), rules.GetCurrentSectionLineNumber());
 				continue;
 			}
 			
@@ -369,7 +369,7 @@ GraphicPack2::GraphicPack2(fs::path rulesPath, IniParser& rules)
 			}
 			catch (const std::exception & ex)
 			{
-				cemuLog_log(LogType::Force, "Graphic pack \"{}\": Can't parse preset \"{}\": {}", m_name, *preset_name, ex.what());
+				cemuLog_log(LogType::Force, "Graphic pack \"{}\": Can't parse preset \"{}\": {}", GetNormalizedPathString(), *preset_name, ex.what());
 			}
 		}
 		else if (boost::iequals(currentSectionName, "RAM"))
@@ -383,7 +383,7 @@ GraphicPack2::GraphicPack2(fs::path rulesPath, IniParser& rules)
 				{
 					if (m_version <= 5)
 					{
-						cemuLog_log(LogType::Force, "Graphic pack \"{}\": [RAM] options are only available for graphic pack version 6 or higher", m_name, optionNameBuf);
+						cemuLog_log(LogType::Force, "Graphic pack \"{}\": [RAM] options are only available for graphic pack version 6 or higher", GetNormalizedPathString(), optionNameBuf);
 						throw std::exception();
 					}
 
@@ -393,12 +393,12 @@ GraphicPack2::GraphicPack2(fs::path rulesPath, IniParser& rules)
 					{
 						if (addrEnd <= addrStart)
 						{
-							cemuLog_log(LogType::Force, "Graphic pack \"{}\": start address (0x{:08x}) must be greater than end address (0x{:08x}) for {}", m_name, addrStart, addrEnd, optionNameBuf);
+							cemuLog_log(LogType::Force, "Graphic pack \"{}\": start address (0x{:08x}) must be greater than end address (0x{:08x}) for {}", GetNormalizedPathString(), addrStart, addrEnd, optionNameBuf);
 							throw std::exception();
 						}
 						else if ((addrStart & 0xFFF) != 0 || (addrEnd & 0xFFF) != 0)
 						{
-							cemuLog_log(LogType::Force, "Graphic pack \"{}\": addresses for %s are not aligned to 0x1000", m_name, optionNameBuf);
+							cemuLog_log(LogType::Force, "Graphic pack \"{}\": addresses for %s are not aligned to 0x1000", GetNormalizedPathString(), optionNameBuf);
 							throw std::exception();
 						}
 						else
@@ -408,7 +408,7 @@ GraphicPack2::GraphicPack2(fs::path rulesPath, IniParser& rules)
 					}
 					else
 					{
-						cemuLog_log(LogType::Force, "Graphic pack \"{}\": has invalid syntax for option {}", m_name, optionNameBuf);
+						cemuLog_log(LogType::Force, "Graphic pack \"{}\": has invalid syntax for option {}", GetNormalizedPathString(), optionNameBuf);
 						throw std::exception();
 					}
 				}
@@ -422,24 +422,32 @@ GraphicPack2::GraphicPack2(fs::path rulesPath, IniParser& rules)
 		std::unordered_map<std::string, std::vector<PresetPtr>> tmp_map;
 		
 		// all vars must be defined in the default preset vars before
-		for (const auto& entry : m_presets)
+		std::vector<std::pair<std::string, std::string>> mismatchingPresetVars;
+		for (const auto& presetEntry : m_presets)
 		{
-			tmp_map[entry->category].emplace_back(entry);
+			tmp_map[presetEntry->category].emplace_back(presetEntry);
 			
-			for (auto& kv : entry->variables)
+			for (auto& presetVar : presetEntry->variables)
 			{
-				const auto it = m_preset_vars.find(kv.first);
+				const auto it = m_preset_vars.find(presetVar.first);
 				if (it == m_preset_vars.cend())
 				{
-					cemuLog_log(LogType::Force, "Graphic pack: \"{}\" contains preset variables which are not defined in the default section", m_name);
-					throw std::exception();
+					mismatchingPresetVars.emplace_back(presetEntry->name, presetVar.first);
+					continue;
 				}
-
 				// overwrite var type with default var type
-				kv.second.first = it->second.first;
+				presetVar.second.first = it->second.first;
 			}
 		}
 
+		if(!mismatchingPresetVars.empty())
+		{
+			cemuLog_log(LogType::Force, "Graphic pack \"{}\" contains preset variables which are not defined in the [Default] section:", GetNormalizedPathString());
+			for (const auto& [presetName, varName] : mismatchingPresetVars)
+				cemuLog_log(LogType::Force, "Preset: {} Variable: {}", presetName, varName);
+			throw std::exception();
+		}
+
 		// have first entry be default active for every category if no default= is set
 		for(auto entry : get_values(tmp_map))
 		{
@@ -469,7 +477,7 @@ GraphicPack2::GraphicPack2(fs::path rulesPath, IniParser& rules)
 				auto& p2 = kv.second[i + 1];
 				if (p1->variables.size() != p2->variables.size())
 				{
-					cemuLog_log(LogType::Force, "Graphic pack: \"{}\" contains inconsistent preset variables", m_name);
+					cemuLog_log(LogType::Force, "Graphic pack: \"{}\" contains inconsistent preset variables", GetNormalizedPathString());
 					throw std::exception();
 				}
 
@@ -477,14 +485,14 @@ GraphicPack2::GraphicPack2(fs::path rulesPath, IniParser& rules)
 				std::set<std::string> keys2(get_keys(p2->variables).begin(), get_keys(p2->variables).end());
 				if (keys1 != keys2)
 				{
-					cemuLog_log(LogType::Force, "Graphic pack: \"{}\" contains inconsistent preset variables", m_name);
+					cemuLog_log(LogType::Force, "Graphic pack: \"{}\" contains inconsistent preset variables", GetNormalizedPathString());
 					throw std::exception();
 				}
 
 				if(p1->is_default)
 				{
 					if(has_default)
-						cemuLog_log(LogType::Force, "Graphic pack: \"{}\" has more than one preset with the default key set for the same category \"{}\"", m_name, p1->name);
+						cemuLog_log(LogType::Force, "Graphic pack: \"{}\" has more than one preset with the default key set for the same category \"{}\"", GetNormalizedPathString(), p1->name);
 					p1->active = true;
 					has_default = true;
 				}
@@ -960,7 +968,7 @@ bool GraphicPack2::Activate()
 				auto option_upscale = rules.FindOption("upscaleMagFilter");
 				if(option_upscale && boost::iequals(*option_upscale, "NearestNeighbor"))
 					m_output_settings.upscale_filter = LatteTextureView::MagFilter::kNearestNeighbor;
-				auto option_downscale = rules.FindOption("NearestNeighbor");
+				auto option_downscale = rules.FindOption("downscaleMinFilter");
 				if (option_downscale && boost::iequals(*option_downscale, "NearestNeighbor"))
 					m_output_settings.downscale_filter = LatteTextureView::MagFilter::kNearestNeighbor;
 			}

src/Cafe/HW/Espresso/Debugger/Debugger.cpp

@@ -8,6 +8,7 @@
 #include "gui/debugger/DebuggerWindow2.h"
 
 #include "Cafe/OS/libs/coreinit/coreinit.h"
+#include "util/helpers/helpers.h"
 
 #if BOOST_OS_WINDOWS
 #include <Windows.h>
@@ -136,11 +137,6 @@ void debugger_createCodeBreakpoint(uint32 address, uint8 bpType)
 	debugger_updateExecutionBreakpoint(address);
 }
 
-void debugger_createExecuteBreakpoint(uint32 address)
-{
-	debugger_createCodeBreakpoint(address, DEBUGGER_BP_T_NORMAL);
-}
-
 namespace coreinit
 {
 	std::vector<std::thread::native_handle_type>& OSGetSchedulerThreads();
@@ -294,8 +290,23 @@ void debugger_toggleExecuteBreakpoint(uint32 address)
 	}
 	else
 	{
-		// create new breakpoint
-		debugger_createExecuteBreakpoint(address);
+		// create new execution breakpoint
+		debugger_createCodeBreakpoint(address, DEBUGGER_BP_T_NORMAL);
+	}
+}
+
+void debugger_toggleLoggingBreakpoint(uint32 address)
+{
+	auto existingBP = debugger_getFirstBP(address, DEBUGGER_BP_T_LOGGING);
+	if (existingBP)
+	{
+		// delete existing breakpoint
+		debugger_deleteBreakpoint(existingBP);
+	}
+	else
+	{
+		// create new logging breakpoint
+		debugger_createCodeBreakpoint(address, DEBUGGER_BP_T_LOGGING);
 	}
 }
 
@@ -447,6 +458,34 @@ bool debugger_hasPatch(uint32 address)
 	return false;
 }
 
+void debugger_removePatch(uint32 address)
+{
+	for (sint32 i = 0; i < debuggerState.patches.size(); i++)
+	{
+		auto& patch = debuggerState.patches[i];
+		if (address < patch->address || address >= (patch->address + patch->length))
+			continue;
+		MPTR startAddress = patch->address;
+		MPTR endAddress = patch->address + patch->length;
+		// remove any breakpoints overlapping with the patch
+		for (auto& bp : debuggerState.breakpoints)
+		{
+			if (bp->address + 4 > startAddress && bp->address < endAddress)
+			{
+				bp->enabled = false;
+				debugger_updateExecutionBreakpoint(bp->address);
+			}
+		}
+		// restore original data
+		memcpy(MEMPTR<void>(startAddress).GetPtr(), patch->origData.data(), patch->length);
+		PPCRecompiler_invalidateRange(startAddress, endAddress);
+		// remove patch
+		delete patch;
+		debuggerState.patches.erase(debuggerState.patches.begin() + i);
+		return;
+	}
+}
+
 void debugger_stepInto(PPCInterpreter_t* hCPU, bool updateDebuggerWindow = true)
 {
 	bool isRecEnabled = ppcRecompilerEnabled;
@@ -510,7 +549,48 @@ void debugger_enterTW(PPCInterpreter_t* hCPU)
 	{
 		if (bp->bpType == DEBUGGER_BP_T_LOGGING && bp->enabled)
 		{
-			std::string logName = !bp->comment.empty() ? "Breakpoint '"+boost::nowide::narrow(bp->comment)+"'" : fmt::format("Breakpoint at 0x{:08X} (no comment)", bp->address);
+			std::string comment = !bp->comment.empty() ? boost::nowide::narrow(bp->comment) : fmt::format("Breakpoint at 0x{:08X} (no comment)", bp->address);
+
+			auto replacePlaceholders = [&](const std::string& prefix, const auto& formatFunc)
+			{
+				size_t pos = 0;
+				while ((pos = comment.find(prefix, pos)) != std::string::npos)
+				{
+					size_t endPos = comment.find('}', pos);
+					if (endPos == std::string::npos)
+						break;
+
+					try
+					{
+						if (int regNum = ConvertString<int>(comment.substr(pos + prefix.length(), endPos - pos - prefix.length())); regNum >= 0 && regNum < 32)
+						{
+							std::string replacement = formatFunc(regNum);
+							comment.replace(pos, endPos - pos + 1, replacement);
+							pos += replacement.length();
+						}
+						else
+						{
+							pos = endPos + 1;
+						}
+					}
+					catch (...)
+					{
+						pos = endPos + 1;
+					}
+				}
+			};
+
+			// Replace integer register placeholders {rX}
+			replacePlaceholders("{r", [&](int regNum) {
+				return fmt::format("0x{:08X}", hCPU->gpr[regNum]);
+			});
+
+			// Replace floating point register placeholders {fX}
+			replacePlaceholders("{f", [&](int regNum) {
+				return fmt::format("{}", hCPU->fpr[regNum].fpr);
+			});
+
+			std::string logName = "Breakpoint '" + comment + "'";
 			std::string logContext = fmt::format("Thread: {:08x} LR: 0x{:08x}", MEMPTR<OSThread_t>(coreinit::OSGetCurrentThread()).GetMPTR(), hCPU->spr.LR, cemuLog_advancedPPCLoggingEnabled() ? " Stack Trace:" : "");
 			cemuLog_log(LogType::Force, "[Debugger] {} was executed! {}", logName, logContext);
 			if (cemuLog_advancedPPCLoggingEnabled())
@@ -547,7 +627,7 @@ void debugger_enterTW(PPCInterpreter_t* hCPU)
 	debuggerState.debugSession.stepInto = false;
 	debuggerState.debugSession.stepOver = false;
 	debuggerState.debugSession.run = false;
-	while (true)
+	while (debuggerState.debugSession.isTrapped)
 	{
 		std::this_thread::sleep_for(std::chrono::milliseconds(1));
 		// check for step commands

src/Cafe/HW/Espresso/Debugger/Debugger.h

@@ -100,8 +100,8 @@ extern debuggerState_t debuggerState;
 // new API
 DebuggerBreakpoint* debugger_getFirstBP(uint32 address);
 void debugger_createCodeBreakpoint(uint32 address, uint8 bpType);
-void debugger_createExecuteBreakpoint(uint32 address);
 void debugger_toggleExecuteBreakpoint(uint32 address); // create/remove execute breakpoint
+void debugger_toggleLoggingBreakpoint(uint32 address); // create/remove logging breakpoint
 void debugger_toggleBreakpoint(uint32 address, bool state, DebuggerBreakpoint* bp);
 
 void debugger_createMemoryBreakpoint(uint32 address, bool onRead, bool onWrite);
@@ -114,6 +114,7 @@ void debugger_updateExecutionBreakpoint(uint32 address, bool forceRestore = fals
 
 void debugger_createPatch(uint32 address, std::span<uint8> patchData);
 bool debugger_hasPatch(uint32 address);
+void debugger_removePatch(uint32 address);
 
 void debugger_forceBreak(); // force breakpoint at the next possible instruction
 bool debugger_isTrapped();

src/Cafe/HW/Espresso/Recompiler/PPCRecompilerX64.cpp

@@ -114,13 +114,13 @@ void* ATTR_MS_ABI PPCRecompiler_virtualHLE(PPCInterpreter_t* hCPU, uint32 hleFun
 
 void ATTR_MS_ABI PPCRecompiler_getTBL(PPCInterpreter_t* hCPU, uint32 gprIndex)
 {
-	uint64 coreTime = coreinit::coreinit_getTimerTick();
+	uint64 coreTime = coreinit::OSGetSystemTime();
 	hCPU->gpr[gprIndex] = (uint32)(coreTime&0xFFFFFFFF);
 }
 
 void ATTR_MS_ABI PPCRecompiler_getTBU(PPCInterpreter_t* hCPU, uint32 gprIndex)
 {
-	uint64 coreTime = coreinit::coreinit_getTimerTick();
+	uint64 coreTime = coreinit::OSGetSystemTime();
 	hCPU->gpr[gprIndex] = (uint32)((coreTime>>32)&0xFFFFFFFF);
 }
 

src/Cafe/HW/Latte/Core/LatteCommandProcessor.cpp

@@ -141,6 +141,14 @@ private:
 
 void LatteCP_processCommandBuffer(DrawPassContext& drawPassCtx);
 
+// called whenever the GPU runs out of commands or hits a wait condition (semaphores, HLE waits)
+void LatteCP_signalEnterWait()
+{
+	// based on the assumption that games won't do a rugpull and swap out buffer data in the middle of an uninterrupted sequence of drawcalls,
+	// we only flush caches when the GPU goes idle or has to wait for any operation
+	LatteIndices_invalidateAll();
+}
+
 /*
 * Read a U32 from the command buffer
 * If no data is available then wait in a busy loop
@@ -466,6 +474,8 @@ LatteCMDPtr LatteCP_itWaitRegMem(LatteCMDPtr cmd, uint32 nWords)
 	const uint32 GPU7_WAIT_MEM_OP_GREATER = 6;
 	const uint32 GPU7_WAIT_MEM_OP_NEVER = 7;
 
+	LatteCP_signalEnterWait();
+
 	bool stalls = false;
 	if ((word0 & 0x10) != 0)
 	{
@@ -594,6 +604,7 @@ LatteCMDPtr LatteCP_itMemSemaphore(LatteCMDPtr cmd, uint32 nWords)
 	else if(SEM_SIGNAL == 7)
 	{
 		// wait
+		LatteCP_signalEnterWait();
 		size_t loopCount = 0;
 		while (true)
 		{
@@ -788,7 +799,7 @@ LatteCMDPtr LatteCP_itHLESampleTimer(LatteCMDPtr cmd, uint32 nWords)
 {
 	cemu_assert_debug(nWords == 1);
 	MPTR timerMPTR = (MPTR)LatteReadCMD();
-	memory_writeU64(timerMPTR, coreinit::coreinit_getTimerTick());
+	memory_writeU64(timerMPTR, coreinit::OSGetSystemTime());
 	return cmd;
 }
 
@@ -1305,11 +1316,13 @@ void LatteCP_processCommandBuffer(DrawPassContext& drawPassCtx)
 				}
 				case IT_HLE_TRIGGER_SCANBUFFER_SWAP:
 				{
+					LatteCP_signalEnterWait();
 					LatteCP_itHLESwapScanBuffer(cmdData, nWords);
 					break;
 				}
 				case IT_HLE_WAIT_FOR_FLIP:
 				{
+					LatteCP_signalEnterWait();
 					LatteCP_itHLEWaitForFlip(cmdData, nWords);
 					break;
 				}
@@ -1594,12 +1607,14 @@ void LatteCP_ProcessRingbuffer()
 			}
 			case IT_HLE_TRIGGER_SCANBUFFER_SWAP:
 			{
+				LatteCP_signalEnterWait();
 				LatteCP_itHLESwapScanBuffer(cmd, nWords);
 				timerRecheck += CP_TIMER_RECHECK / 64;
 				break;
 			}
 			case IT_HLE_WAIT_FOR_FLIP:
 			{
+				LatteCP_signalEnterWait();
 				LatteCP_itHLEWaitForFlip(cmd, nWords);
 				timerRecheck += CP_TIMER_RECHECK / 1;
 				break;

src/Cafe/HW/Latte/Core/LatteIndices.cpp

@@ -1,6 +1,7 @@
 #include "Cafe/HW/Latte/Core/LatteConst.h"
 #include "Cafe/HW/Latte/Renderer/Renderer.h"
 #include "Cafe/HW/Latte/ISA/RegDefines.h"
+#include "Cafe/HW/Latte/Core/LattePerformanceMonitor.h"
 #include "Common/cpu_features.h"
 
 #if defined(ARCH_X86_64) && defined(__GNUC__)
@@ -9,32 +10,53 @@
 
 struct  
 {
-	const void* lastPtr;
-	uint32 lastCount;
-	LattePrimitiveMode lastPrimitiveMode;
-	LatteIndexType lastIndexType;
-	// output
-	uint32 indexMin;
-	uint32 indexMax;
-	Renderer::INDEX_TYPE renderIndexType;
-	uint32 outputCount;
-	uint32 indexBufferOffset;
-	uint32 indexBufferIndex;
+	struct CacheEntry
+	{
+		// input data
+		const void* lastPtr;
+		uint32 lastCount;
+		LattePrimitiveMode lastPrimitiveMode;
+		LatteIndexType lastIndexType;
+		uint64 lastUsed;
+		// output
+		uint32 indexMin;
+		uint32 indexMax;
+		Renderer::INDEX_TYPE renderIndexType;
+		uint32 outputCount;
+		Renderer::IndexAllocation indexAllocation;
+	};
+	std::array<CacheEntry, 8> entry;
+	uint64 currentUsageCounter{0};
 }LatteIndexCache{};
 
 void LatteIndices_invalidate(const void* memPtr, uint32 size)
 {
-	if (LatteIndexCache.lastPtr >= memPtr && (LatteIndexCache.lastPtr < ((uint8*)memPtr + size)) )
+	for(auto& entry : LatteIndexCache.entry)
 	{
-		LatteIndexCache.lastPtr = nullptr;
-		LatteIndexCache.lastCount = 0;
+		if (entry.lastPtr >= memPtr && (entry.lastPtr < ((uint8*)memPtr + size)) )
+		{
+			if(entry.lastPtr != nullptr)
+				g_renderer->indexData_releaseIndexMemory(entry.indexAllocation);
+			entry.lastPtr = nullptr;
+			entry.lastCount = 0;
+		}
 	}
 }
 
 void LatteIndices_invalidateAll()
 {
-	LatteIndexCache.lastPtr = nullptr;
-	LatteIndexCache.lastCount = 0;
+	for(auto& entry : LatteIndexCache.entry)
+	{
+		if (entry.lastPtr != nullptr)
+			g_renderer->indexData_releaseIndexMemory(entry.indexAllocation);
+		entry.lastPtr = nullptr;
+		entry.lastCount = 0;
+	}
+}
+
+uint64 LatteIndices_GetNextUsageIndex()
+{
+	return LatteIndexCache.currentUsageCounter++;
 }
 
 uint32 LatteIndices_calculateIndexOutputSize(LattePrimitiveMode primitiveMode, LatteIndexType indexType, uint32 count)
@@ -532,7 +554,7 @@ void LatteIndices_alternativeCalculateIndexMinMax(const void* indexData, LatteIn
 	}
 }
 
-void LatteIndices_decode(const void* indexData, LatteIndexType indexType, uint32 count, LattePrimitiveMode primitiveMode, uint32& indexMin, uint32& indexMax, Renderer::INDEX_TYPE& renderIndexType, uint32& outputCount, uint32& indexBufferOffset, uint32& indexBufferIndex)
+void LatteIndices_decode(const void* indexData, LatteIndexType indexType, uint32 count, LattePrimitiveMode primitiveMode, uint32& indexMin, uint32& indexMax, Renderer::INDEX_TYPE& renderIndexType, uint32& outputCount, Renderer::IndexAllocation& indexAllocation)
 {
 	// what this should do:
 	// [x] use fast SIMD-based index decoding
@@ -542,17 +564,18 @@ void LatteIndices_decode(const void* indexData, LatteIndexType indexType, uint32
 	// [ ] better cache implementation, allow to cache across frames
 
 	// reuse from cache if data didn't change
-	if (LatteIndexCache.lastPtr == indexData &&
-		LatteIndexCache.lastCount == count &&
-		LatteIndexCache.lastPrimitiveMode == primitiveMode &&
-		LatteIndexCache.lastIndexType == indexType)
+	auto cacheEntry = std::find_if(LatteIndexCache.entry.begin(), LatteIndexCache.entry.end(), [indexData, count, primitiveMode, indexType](const auto& entry)
 	{
-		indexMin = LatteIndexCache.indexMin;
-		indexMax = LatteIndexCache.indexMax;
-		renderIndexType = LatteIndexCache.renderIndexType;
-		outputCount = LatteIndexCache.outputCount;
-		indexBufferOffset = LatteIndexCache.indexBufferOffset;
-		indexBufferIndex = LatteIndexCache.indexBufferIndex;
+		return entry.lastPtr == indexData && entry.lastCount == count && entry.lastPrimitiveMode == primitiveMode && entry.lastIndexType == indexType;
+	});
+	if (cacheEntry != LatteIndexCache.entry.end())
+	{
+		indexMin = cacheEntry->indexMin;
+		indexMax = cacheEntry->indexMax;
+		renderIndexType = cacheEntry->renderIndexType;
+		outputCount = cacheEntry->outputCount;
+		indexAllocation = cacheEntry->indexAllocation;
+		cacheEntry->lastUsed = LatteIndices_GetNextUsageIndex();
 		return;
 	}
 
@@ -576,10 +599,12 @@ void LatteIndices_decode(const void* indexData, LatteIndexType indexType, uint32
 		indexMin = 0;
 		indexMax = std::max(count, 1u)-1;
 		renderIndexType = Renderer::INDEX_TYPE::NONE;
+		indexAllocation = {};
 		return; // no indices
 	}
 	// query index buffer from renderer
-	void* indexOutputPtr = g_renderer->indexData_reserveIndexMemory(indexOutputSize, indexBufferOffset, indexBufferIndex);
+	indexAllocation = g_renderer->indexData_reserveIndexMemory(indexOutputSize);
+	void* indexOutputPtr = indexAllocation.mem;
 
 	// decode indices
 	indexMin = std::numeric_limits<uint32>::max();
@@ -704,16 +729,25 @@ void LatteIndices_decode(const void* indexData, LatteIndexType indexType, uint32
 		// recalculate index range but filter out primitive restart index
 		LatteIndices_alternativeCalculateIndexMinMax(indexData, indexType, count, indexMin, indexMax);
 	}
-	g_renderer->indexData_uploadIndexMemory(indexBufferOffset, indexOutputSize);
+	g_renderer->indexData_uploadIndexMemory(indexAllocation);
+	performanceMonitor.cycle[performanceMonitor.cycleIndex].indexDataUploaded += indexOutputSize;
+	// get least recently used cache entry
+	auto lruEntry = std::min_element(LatteIndexCache.entry.begin(), LatteIndexCache.entry.end(), [](const auto& a, const auto& b)
+	{
+		return a.lastUsed < b.lastUsed;
+	});
+	// invalidate previous allocation
+	if(lruEntry->lastPtr != nullptr)
+		g_renderer->indexData_releaseIndexMemory(lruEntry->indexAllocation);
 	// update cache
-	LatteIndexCache.lastPtr = indexData;
-	LatteIndexCache.lastCount = count;
-	LatteIndexCache.lastPrimitiveMode = primitiveMode;
-	LatteIndexCache.lastIndexType = indexType;
-	LatteIndexCache.indexMin = indexMin;
-	LatteIndexCache.indexMax = indexMax;
-	LatteIndexCache.renderIndexType = renderIndexType;
-	LatteIndexCache.outputCount = outputCount;
-	LatteIndexCache.indexBufferOffset = indexBufferOffset;
-	LatteIndexCache.indexBufferIndex = indexBufferIndex;
+	lruEntry->lastPtr = indexData;
+	lruEntry->lastCount = count;
+	lruEntry->lastPrimitiveMode = primitiveMode;
+	lruEntry->lastIndexType = indexType;
+	lruEntry->indexMin = indexMin;
+	lruEntry->indexMax = indexMax;
+	lruEntry->renderIndexType = renderIndexType;
+	lruEntry->outputCount = outputCount;
+	lruEntry->indexAllocation = indexAllocation;
+	lruEntry->lastUsed = LatteIndices_GetNextUsageIndex();
 }

src/Cafe/HW/Latte/Core/LatteIndices.h

@@ -4,4 +4,4 @@
 
 void LatteIndices_invalidate(const void* memPtr, uint32 size);
 void LatteIndices_invalidateAll();
-void LatteIndices_decode(const void* indexData, LatteIndexType indexType, uint32 count, LattePrimitiveMode primitiveMode, uint32& indexMin, uint32& indexMax, Renderer::INDEX_TYPE& renderIndexType, uint32& outputCount, uint32& indexBufferOffset, uint32& indexBufferIndex);
+void LatteIndices_decode(const void* indexData, LatteIndexType indexType, uint32 count, LattePrimitiveMode primitiveMode, uint32& indexMin, uint32& indexMax, Renderer::INDEX_TYPE& renderIndexType, uint32& outputCount, Renderer::IndexAllocation& indexAllocation);

src/Cafe/HW/Latte/Core/LatteOverlay.cpp

@@ -107,7 +107,13 @@ void LatteOverlay_renderOverlay(ImVec2& position, ImVec2& pivot, sint32 directio
 				ImGui::Text("VRAM: %dMB / %dMB", g_state.vramUsage, g_state.vramTotal);
 
 			if (config.overlay.debug)
+			{
+				// general debug info
+				ImGui::Text("--- Debug info ---");
+				ImGui::Text("IndexUploadPerFrame: %dKB", (performanceMonitor.stats.indexDataUploadPerFrame+1023)/1024);
+				// backend specific info
 				g_renderer->AppendOverlayDebugInfo();
+			}
 
 			position.y += (ImGui::GetWindowSize().y + 10.0f) * direction;
 		}

src/Cafe/HW/Latte/Core/LattePerformanceMonitor.cpp

@@ -74,7 +74,6 @@ void LattePerformanceMonitor_frameEnd()
 		uniformBankDataUploadedPerFrame /= 1024ULL;
 		uint32 uniformBankCountUploadedPerFrame = (uint32)(uniformBankUploadedCount / (uint64)elapsedFrames);
 		uint64 indexDataUploadPerFrame = (indexDataUploaded / (uint64)elapsedFrames);
-		indexDataUploadPerFrame /= 1024ULL;
 
 		double fps = (double)elapsedFrames2S * 1000.0 / (double)totalElapsedTimeFPS;
 		uint32 shaderBindsPerFrame = shaderBindCounter / elapsedFrames;
@@ -82,7 +81,7 @@ void LattePerformanceMonitor_frameEnd()
 		uint32 rlps = (uint32)((uint64)recompilerLeaveCount * 1000ULL / (uint64)totalElapsedTime);
 		uint32 tlps = (uint32)((uint64)threadLeaveCount * 1000ULL / (uint64)totalElapsedTime);
 		// set stats
-
+		performanceMonitor.stats.indexDataUploadPerFrame = indexDataUploadPerFrame;
 		// next counter cycle
 		sint32 nextCycleIndex = (performanceMonitor.cycleIndex + 1) % PERFORMANCE_MONITOR_TRACK_CYCLES;
 		performanceMonitor.cycle[nextCycleIndex].drawCallCounter = 0;

src/Cafe/HW/Latte/Core/LattePerformanceMonitor.h

@@ -124,6 +124,7 @@ typedef struct
 		LattePerfStatCounter numGraphicPipelines;
 		LattePerfStatCounter numImages;
 		LattePerfStatCounter numImageViews;
+		LattePerfStatCounter numSamplers;
 		LattePerfStatCounter numRenderPass;
 		LattePerfStatCounter numFramebuffer;
 
@@ -131,6 +132,12 @@ typedef struct
 		LattePerfStatCounter numDrawBarriersPerFrame;
 		LattePerfStatCounter numBeginRenderpassPerFrame;
 	}vk;
+
+	// calculated stats (per frame)
+	struct
+	{
+		uint32 indexDataUploadPerFrame;
+	}stats;
 }performanceMonitor_t;
 
 extern performanceMonitor_t performanceMonitor;

src/Cafe/HW/Latte/Core/LatteRenderTarget.cpp

@@ -11,7 +11,6 @@
 #include "Cafe/HW/Latte/Core/LattePerformanceMonitor.h"
 #include "Cafe/GraphicPack/GraphicPack2.h"
 #include "config/ActiveSettings.h"
-#include "Cafe/HW/Latte/Renderer/Vulkan/VulkanRenderer.h"
 #include "gui/guiWrapper.h"
 #include "Cafe/OS/libs/erreula/erreula.h"
 #include "input/InputManager.h"
@@ -934,13 +933,6 @@ void LatteRenderTarget_copyToBackbuffer(LatteTextureView* textureView, bool isPa
 	{
 		sint32 scaling_filter = downscaling ? GetConfig().downscale_filter : GetConfig().upscale_filter;
 
-		if (g_renderer->GetType() == RendererAPI::Vulkan)
-		{
-			// force linear or nearest neighbor filter
-			if(scaling_filter != kLinearFilter && scaling_filter != kNearestNeighborFilter)
-				scaling_filter = kLinearFilter;
-		}
-
 		if (scaling_filter == kLinearFilter)
 		{
 			if(renderUpsideDown)
@@ -957,7 +949,7 @@ void LatteRenderTarget_copyToBackbuffer(LatteTextureView* textureView, bool isPa
 			else
 				shader = RendererOutputShader::s_bicubic_shader;
 
-			filter = LatteTextureView::MagFilter::kNearestNeighbor;
+			filter = LatteTextureView::MagFilter::kLinear;
 		}
 		else if (scaling_filter == kBicubicHermiteFilter)
 		{

src/Cafe/HW/Latte/Core/LatteShader.cpp

@@ -451,9 +451,8 @@ void LatteShader_DumpShader(uint64 baseHash, uint64 auxHash, LatteDecompilerShad
 		suffix = "gs";
 	else if (shader->shaderType == LatteConst::ShaderType::Pixel)
 		suffix = "ps";
-	fs::path dumpPath = "dump/shaders";
-	dumpPath /= fmt::format("{:016x}_{:016x}_{}.txt", baseHash, auxHash, suffix);
-	FileStream* fs = FileStream::createFile2(dumpPath);
+
+	FileStream* fs = FileStream::createFile2(ActiveSettings::GetUserDataPath("dump/shaders/{:016x}_{:016x}_{}.txt", baseHash, auxHash, suffix));
 	if (fs)
 	{
 		if (shader->strBuf_shaderSource)
@@ -479,9 +478,8 @@ void LatteShader_DumpRawShader(uint64 baseHash, uint64 auxHash, uint32 type, uin
 		suffix = "copy";
 	else if (type == SHADER_DUMP_TYPE_COMPUTE)
 		suffix = "compute";
-	fs::path dumpPath = "dump/shaders";
-	dumpPath /= fmt::format("{:016x}_{:016x}_{}.bin", baseHash, auxHash, suffix);
-	FileStream* fs = FileStream::createFile2(dumpPath);
+
+	FileStream* fs = FileStream::createFile2(ActiveSettings::GetUserDataPath("dump/shaders/{:016x}_{:016x}_{}.bin", baseHash, auxHash, suffix));
 	if (fs)
 	{
 		fs->writeData(programCode, programLen);

src/Cafe/HW/Latte/Core/LatteShaderCache.cpp

@@ -25,6 +25,9 @@
 #include "util/helpers/Serializer.h"
 
 #include <wx/msgdlg.h>
+#include <audio/IAudioAPI.h>
+#include <util/bootSound/BootSoundReader.h>
+#include <thread>
 
 #if BOOST_OS_WINDOWS
 #include <psapi.h>
@@ -155,6 +158,118 @@ bool LoadTGAFile(const std::vector<uint8>& buffer, TGAFILE *tgaFile)
 	return true;
 }
 
+class BootSoundPlayer
+{
+  public:
+	BootSoundPlayer() = default;
+	~BootSoundPlayer()
+	{
+		m_stopRequested = true;
+	}
+
+	void StartSound()
+	{
+		if (!m_bootSndPlayThread.joinable())
+		{
+			m_fadeOutRequested = false;
+			m_stopRequested = false;
+			m_bootSndPlayThread = std::thread{[this]() {
+				StreamBootSound();
+			}};
+		}
+	}
+
+	void FadeOutSound()
+	{
+		m_fadeOutRequested = true;
+	}
+
+	void ApplyFadeOutEffect(std::span<sint16> samples, uint64& fadeOutSample, uint64 fadeOutDuration)
+	{
+		for (size_t i = 0; i < samples.size(); i += 2)
+		{
+			const float decibel = (float)fadeOutSample / fadeOutDuration * -60.0f;
+			const float volumeFactor = pow(10, decibel / 20);
+			samples[i] *= volumeFactor;
+			samples[i + 1] *= volumeFactor;
+			fadeOutSample++;
+		}
+	}
+
+	void StreamBootSound()
+	{
+		SetThreadName("bootsnd");
+		constexpr sint32 sampleRate = 48'000;
+		constexpr sint32 bitsPerSample = 16;
+		constexpr sint32 samplesPerBlock = sampleRate / 10; // block is 1/10th of a second
+		constexpr sint32 nChannels = 2;
+		static_assert(bitsPerSample % 8 == 0, "bits per sample is not a multiple of 8");
+
+		AudioAPIPtr bootSndAudioDev;
+
+		try
+		{
+			bootSndAudioDev = IAudioAPI::CreateDeviceFromConfig(true, sampleRate, nChannels, samplesPerBlock, bitsPerSample);
+			if(!bootSndAudioDev)
+				return;
+		}
+		catch (const std::runtime_error& ex)
+		{
+			cemuLog_log(LogType::Force, "Failed to initialise audio device for bootup sound");
+			return;
+		}
+		bootSndAudioDev->SetAudioDelayOverride(4);
+		bootSndAudioDev->Play();
+
+		std::string sndPath = fmt::format("{}/meta/{}", CafeSystem::GetMlcStoragePath(CafeSystem::GetForegroundTitleId()), "bootSound.btsnd");
+		sint32 fscStatus = FSC_STATUS_UNDEFINED;
+
+		if(!fsc_doesFileExist(sndPath.c_str()))
+			return;
+
+		FSCVirtualFile* bootSndFileHandle = fsc_open(sndPath.c_str(), FSC_ACCESS_FLAG::OPEN_FILE | FSC_ACCESS_FLAG::READ_PERMISSION, &fscStatus);
+		if(!bootSndFileHandle)
+		{
+			cemuLog_log(LogType::Force, "failed to open bootSound.btsnd");
+			return;
+		}
+
+		constexpr sint32 audioBlockSize = samplesPerBlock * (bitsPerSample/8) * nChannels;
+		BootSoundReader bootSndFileReader(bootSndFileHandle, audioBlockSize);
+
+		uint64 fadeOutSample = 0; // track how far into the fadeout
+		constexpr uint64 fadeOutDuration = sampleRate * 2; // fadeout should last 2 seconds
+		while(fadeOutSample < fadeOutDuration && !m_stopRequested)
+		{
+			while (bootSndAudioDev->NeedAdditionalBlocks())
+			{
+				sint16* data = bootSndFileReader.getSamples();
+				if(data == nullptr)
+				{
+					// break outer loop
+					m_stopRequested = true;
+					break;
+				}
+				if(m_fadeOutRequested)
+					ApplyFadeOutEffect({data, samplesPerBlock * nChannels}, fadeOutSample, fadeOutDuration);
+
+				bootSndAudioDev->FeedBlock(data);
+			}
+			// sleep for the duration of a single block
+			std::this_thread::sleep_for(std::chrono::milliseconds(samplesPerBlock / (sampleRate/ 1'000)));
+		}
+
+		if(bootSndFileHandle)
+			fsc_close(bootSndFileHandle);
+	}
+
+  private:
+	std::thread m_bootSndPlayThread;
+	std::atomic_bool m_fadeOutRequested = false;
+	std::atomic_bool m_stopRequested = false;
+};
+static BootSoundPlayer g_bootSndPlayer;
+
 void LatteShaderCache_finish()
 {
 	if (g_renderer->GetType() == RendererAPI::Vulkan)
@@ -299,6 +414,9 @@ void LatteShaderCache_Load()
 	loadBackgroundTexture(true, g_shaderCacheLoaderState.textureTVId);
 	loadBackgroundTexture(false, g_shaderCacheLoaderState.textureDRCId);
 
+	if(GetConfig().play_boot_sound)
+		g_bootSndPlayer.StartSound();
+
 	sint32 numLoadedShaders = 0;
 	uint32 loadIndex = 0;
 
@@ -365,6 +483,11 @@ void LatteShaderCache_Load()
 		g_renderer->DeleteTexture(g_shaderCacheLoaderState.textureTVId);
 	if (g_shaderCacheLoaderState.textureDRCId)
 		g_renderer->DeleteTexture(g_shaderCacheLoaderState.textureDRCId);
+
+	g_bootSndPlayer.FadeOutSound();
+
+	if(Latte_GetStopSignal())
+		LatteThread_Exit();
 }
 
 void LatteShaderCache_ShowProgress(const std::function <bool(void)>& loadUpdateFunc, bool isPipelines)
@@ -505,8 +628,6 @@ void LatteShaderCache_LoadVulkanPipelineCache(uint64 cacheTitleId)
 	g_shaderCacheLoaderState.loadedPipelines = 0;
 	LatteShaderCache_ShowProgress(LatteShaderCache_updatePipelineLoadingProgress, true);
 	pipelineCache.EndLoading();
-    if(Latte_GetStopSignal())
-        LatteThread_Exit();
 }
 
 bool LatteShaderCache_updatePipelineLoadingProgress()

src/Cafe/HW/Latte/Core/LatteThread.cpp

@@ -257,6 +257,7 @@ void LatteThread_Exit()
     LatteSHRC_UnloadAll();
     // close disk cache
     LatteShaderCache_Close();
+	RendererOutputShader::ShutdownStatic();
     // destroy renderer but make sure that g_renderer remains valid until the destructor has finished
 	if (g_renderer)
 	{

src/Cafe/HW/Latte/LegacyShaderDecompiler/LatteDecompiler.cpp

@@ -370,6 +370,8 @@ bool LatteDecompiler_IsALUTransInstruction(bool isOP3, uint32 opcode)
 		opcode == ALU_OP2_INST_LSHR_INT ||
 		opcode == ALU_OP2_INST_MAX_INT ||
 		opcode == ALU_OP2_INST_MIN_INT ||
+		opcode == ALU_OP2_INST_MAX_UINT ||
+		opcode == ALU_OP2_INST_MIN_UINT ||
 		opcode == ALU_OP2_INST_MOVA_FLOOR ||
 		opcode == ALU_OP2_INST_MOVA_INT ||
 		opcode == ALU_OP2_INST_SETE_DX10 ||

src/Cafe/HW/Latte/LegacyShaderDecompiler/LatteDecompilerAnalyzer.cpp

@@ -140,6 +140,8 @@ bool _isIntegerInstruction(const LatteDecompilerALUInstruction& aluInstruction)
 		case ALU_OP2_INST_SUB_INT:
 		case ALU_OP2_INST_MAX_INT:
 		case ALU_OP2_INST_MIN_INT:
+		case ALU_OP2_INST_MAX_UINT:
+		case ALU_OP2_INST_MIN_UINT:
 		case ALU_OP2_INST_SETE_INT:
 		case ALU_OP2_INST_SETGT_INT:
 		case ALU_OP2_INST_SETGE_INT:

src/Cafe/HW/Latte/LegacyShaderDecompiler/LatteDecompilerEmitGLSL.cpp

@@ -1415,19 +1415,23 @@ void _emitALUOP2InstructionCode(LatteDecompilerShaderContext* shaderContext, Lat
 	}
 	else if( aluInstruction->opcode == ALU_OP2_INST_ADD_INT )
 		_emitALUOperationBinary<LATTE_DECOMPILER_DTYPE_SIGNED_INT>(shaderContext, aluInstruction, " + ");
-	else if( aluInstruction->opcode == ALU_OP2_INST_MAX_INT || aluInstruction->opcode == ALU_OP2_INST_MIN_INT )
+	else if( aluInstruction->opcode == ALU_OP2_INST_MAX_INT || aluInstruction->opcode == ALU_OP2_INST_MIN_INT ||
+			 aluInstruction->opcode == ALU_OP2_INST_MAX_UINT || aluInstruction->opcode == ALU_OP2_INST_MIN_UINT)
 	{
 		// not verified
+		bool isUnsigned = aluInstruction->opcode == ALU_OP2_INST_MAX_UINT || aluInstruction->opcode == ALU_OP2_INST_MIN_UINT;
+		auto opType = isUnsigned ? LATTE_DECOMPILER_DTYPE_UNSIGNED_INT : LATTE_DECOMPILER_DTYPE_SIGNED_INT;
 		_emitInstructionOutputVariableName(shaderContext, aluInstruction);
-		if( aluInstruction->opcode == ALU_OP2_INST_MAX_INT )
-			src->add(" = max(");
+		src->add(" = ");
+		_emitTypeConversionPrefix(shaderContext, opType, outputType);
+		if( aluInstruction->opcode == ALU_OP2_INST_MAX_INT || aluInstruction->opcode == ALU_OP2_INST_MAX_UINT )
+			src->add("max(");
 		else
-			src->add(" = min(");
-		_emitTypeConversionPrefix(shaderContext, LATTE_DECOMPILER_DTYPE_SIGNED_INT, outputType);
-		_emitOperandInputCode(shaderContext, aluInstruction, 0, LATTE_DECOMPILER_DTYPE_SIGNED_INT);
+			src->add("min(");
+		_emitOperandInputCode(shaderContext, aluInstruction, 0, opType);
 		src->add(", ");
-		_emitOperandInputCode(shaderContext, aluInstruction, 1, LATTE_DECOMPILER_DTYPE_SIGNED_INT);
-		_emitTypeConversionSuffix(shaderContext, LATTE_DECOMPILER_DTYPE_SIGNED_INT, outputType);
+		_emitOperandInputCode(shaderContext, aluInstruction, 1, opType);
+		_emitTypeConversionSuffix(shaderContext, opType, outputType);
 		src->add(");" _CRLF);
 	}
 	else if( aluInstruction->opcode == ALU_OP2_INST_SUB_INT )

src/Cafe/HW/Latte/LegacyShaderDecompiler/LatteDecompilerInstructions.h

@@ -60,6 +60,8 @@
 #define ALU_OP2_INST_SUB_INT		(0x035) // integer instruction
 #define ALU_OP2_INST_MAX_INT		(0x036) // integer instruction
 #define ALU_OP2_INST_MIN_INT		(0x037) // integer instruction
+#define ALU_OP2_INST_MAX_UINT		(0x038) // integer instruction
+#define ALU_OP2_INST_MIN_UINT		(0x039) // integer instruction
 #define ALU_OP2_INST_SETE_INT		(0x03A) // integer instruction
 #define ALU_OP2_INST_SETGT_INT		(0x03B) // integer instruction
 #define ALU_OP2_INST_SETGE_INT		(0x03C) // integer instruction

src/Cafe/HW/Latte/Renderer/OpenGL/OpenGLRenderer.cpp

@@ -570,13 +570,10 @@ void OpenGLRenderer::DrawBackbufferQuad(LatteTextureView* texView, RendererOutpu
 		g_renderer->ClearColorbuffer(padView);
 	}
 
-	sint32 effectiveWidth, effectiveHeight;
-	texView->baseTexture->GetEffectiveSize(effectiveWidth, effectiveHeight, 0);
-
 	shader_unbind(RendererShader::ShaderType::kGeometry);
 	shader_bind(shader->GetVertexShader());
 	shader_bind(shader->GetFragmentShader());
-	shader->SetUniformParameters(*texView, { effectiveWidth, effectiveHeight }, { imageWidth, imageHeight });
+	shader->SetUniformParameters(*texView, {imageWidth, imageHeight});
 
 	// set viewport
 	glViewportIndexedf(0, imageX, imageY, imageWidth, imageHeight);
@@ -584,6 +581,12 @@ void OpenGLRenderer::DrawBackbufferQuad(LatteTextureView* texView, RendererOutpu
 	LatteTextureViewGL* texViewGL = (LatteTextureViewGL*)texView;
 	texture_bindAndActivate(texView, 0);
 
+	glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE);
+	glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE);
+	texViewGL->samplerState.clampS = texViewGL->samplerState.clampT = 0xFF;
+
+	glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, useLinearTexFilter ? GL_LINEAR : GL_NEAREST);
+	texViewGL->samplerState.filterMin = 0xFFFFFFFF;
 	glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, useLinearTexFilter ? GL_LINEAR : GL_NEAREST);
 	texViewGL->samplerState.filterMag = 0xFFFFFFFF;
 

src/Cafe/HW/Latte/Renderer/OpenGL/OpenGLRenderer.h

@@ -102,16 +102,21 @@ public:
 	static void SetAttributeArrayState(uint32 index, bool isEnabled, sint32 aluDivisor);
 	static void SetArrayElementBuffer(GLuint arrayElementBuffer);
 
-	// index
-	void* indexData_reserveIndexMemory(uint32 size, uint32& offset, uint32& bufferIndex) override
+	// index (not used by OpenGL renderer yet)
+	IndexAllocation indexData_reserveIndexMemory(uint32 size) override
 	{
-		assert_dbg();
-		return nullptr;
+		cemu_assert_unimplemented();
+		return {};
 	}
 
-	void indexData_uploadIndexMemory(uint32 offset, uint32 size) override
+	void indexData_releaseIndexMemory(IndexAllocation& allocation) override
 	{
-		assert_dbg();
+		cemu_assert_unimplemented();
+	}
+
+	void indexData_uploadIndexMemory(IndexAllocation& allocation) override
+	{
+		cemu_assert_unimplemented();
 	}
 
 	// uniform

src/Cafe/HW/Latte/Renderer/Renderer.h

@@ -138,8 +138,15 @@ public:
 	virtual void draw_endSequence() = 0;
 
 	// index
-	virtual void* indexData_reserveIndexMemory(uint32 size, uint32& offset, uint32& bufferIndex) = 0;
-	virtual void indexData_uploadIndexMemory(uint32 offset, uint32 size) = 0;
+	struct IndexAllocation
+	{
+		void* mem; // pointer to index data inside buffer
+		void* rendererInternal; // for renderer use
+	};
+
+	virtual IndexAllocation indexData_reserveIndexMemory(uint32 size) = 0;
+	virtual void indexData_releaseIndexMemory(IndexAllocation& allocation) = 0;
+	virtual void indexData_uploadIndexMemory(IndexAllocation& allocation) = 0;
 
 	// occlusion queries
 	virtual LatteQueryObject* occlusionQuery_create() = 0;

src/Cafe/HW/Latte/Renderer/RendererOuputShader.cpp

@@ -2,18 +2,7 @@
 #include "Cafe/HW/Latte/Renderer/OpenGL/OpenGLRenderer.h"
 
 const std::string RendererOutputShader::s_copy_shader_source =
-R"(#version 420
-
-#ifdef VULKAN
-layout(location = 0) in vec2 passUV;
-layout(binding = 0) uniform sampler2D textureSrc;
-layout(location = 0) out vec4 colorOut0;
-#else
-in vec2 passUV;
-layout(binding=0) uniform sampler2D textureSrc;
-layout(location = 0) out vec4 colorOut0;
-#endif
-
+R"(
 void main()
 {
 	colorOut0 = vec4(texture(textureSrc, passUV).rgb,1.0);
@@ -22,20 +11,6 @@ void main()
 
 const std::string RendererOutputShader::s_bicubic_shader_source =
 R"(
-#version 420
-
-#ifdef VULKAN
-layout(location = 0) in vec2 passUV;
-layout(binding = 0)  uniform sampler2D textureSrc;
-layout(binding = 1)  uniform vec2 textureSrcResolution;
-layout(location = 0) out vec4 colorOut0;
-#else
-in vec2 passUV;
-layout(binding=0) uniform sampler2D textureSrc;
-uniform vec2 textureSrcResolution;
-layout(location = 0) out vec4 colorOut0;
-#endif
-
 vec4 cubic(float x)
 {
 	float x2 = x * x;
@@ -48,24 +23,23 @@ vec4 cubic(float x)
 	return w / 6.0;
 }
 
-vec4 bcFilter(vec2 texcoord, vec2 texscale)
+vec4 bcFilter(vec2 uv, vec4 texelSize)
 {
-	float fx = fract(texcoord.x);
-	float fy = fract(texcoord.y);
-	texcoord.x -= fx;
-	texcoord.y -= fy;
+	vec2 pixel = uv*texelSize.zw - 0.5;
+	vec2 pixelFrac = fract(pixel);
+	vec2 pixelInt = pixel - pixelFrac;
 
-	vec4 xcubic = cubic(fx);
-	vec4 ycubic = cubic(fy);
+	vec4 xcubic = cubic(pixelFrac.x);
+	vec4 ycubic = cubic(pixelFrac.y);
 
-	vec4 c = vec4(texcoord.x - 0.5, texcoord.x + 1.5, texcoord.y - 0.5, texcoord.y + 1.5);
+	vec4 c = vec4(pixelInt.x - 0.5, pixelInt.x + 1.5, pixelInt.y - 0.5, pixelInt.y + 1.5);
 	vec4 s = vec4(xcubic.x + xcubic.y, xcubic.z + xcubic.w, ycubic.x + ycubic.y, ycubic.z + ycubic.w);
 	vec4 offset = c + vec4(xcubic.y, xcubic.w, ycubic.y, ycubic.w) / s;
 
-	vec4 sample0 = texture(textureSrc, vec2(offset.x, offset.z) * texscale);
-	vec4 sample1 = texture(textureSrc, vec2(offset.y, offset.z) * texscale);
-	vec4 sample2 = texture(textureSrc, vec2(offset.x, offset.w) * texscale);
-	vec4 sample3 = texture(textureSrc, vec2(offset.y, offset.w) * texscale);
+	vec4 sample0 = texture(textureSrc, vec2(offset.x, offset.z) * texelSize.xy);
+	vec4 sample1 = texture(textureSrc, vec2(offset.y, offset.z) * texelSize.xy);
+	vec4 sample2 = texture(textureSrc, vec2(offset.x, offset.w) * texelSize.xy);
+	vec4 sample3 = texture(textureSrc, vec2(offset.y, offset.w) * texelSize.xy);
 
 	float sx = s.x / (s.x + s.y);
 	float sy = s.z / (s.z + s.w);
@@ -76,20 +50,13 @@ vec4 bcFilter(vec2 texcoord, vec2 texscale)
 }
 
 void main(){
-	colorOut0 = vec4(bcFilter(passUV*textureSrcResolution, vec2(1.0,1.0)/textureSrcResolution).rgb,1.0);
+	vec4 texelSize = vec4( 1.0 / textureSrcResolution.xy, textureSrcResolution.xy);
+	colorOut0 = vec4(bcFilter(passUV, texelSize).rgb,1.0);
 }
 )";
 
 const std::string RendererOutputShader::s_hermite_shader_source =
-R"(#version 420
-
-in vec4 gl_FragCoord;	
-in vec2 passUV;
-layout(binding=0) uniform sampler2D textureSrc;
-uniform vec2 textureSrcResolution;
-uniform vec2 outputResolution;
-layout(location = 0) out vec4 colorOut0;
-
+R"(
 // https://www.shadertoy.com/view/MllSzX
 
 vec3 CubicHermite (vec3 A, vec3 B, vec3 C, vec3 D, float t)
@@ -111,7 +78,7 @@ vec3 BicubicHermiteTexture(vec2 uv, vec4 texelSize)
 	vec2 frac = fract(pixel);	
     pixel = floor(pixel) / texelSize.zw - vec2(texelSize.xy/2.0);
 	
-	vec4 doubleSize = texelSize*texelSize;
+	vec4 doubleSize = texelSize*2.0;
 
 	vec3 C00 = texture(textureSrc, pixel + vec2(-texelSize.x ,-texelSize.y)).rgb;
     vec3 C10 = texture(textureSrc, pixel + vec2( 0.0        ,-texelSize.y)).rgb;
@@ -142,15 +109,17 @@ vec3 BicubicHermiteTexture(vec2 uv, vec4 texelSize)
 }
 
 void main(){
-	vec4 texelSize = vec4( 1.0 / outputResolution.xy, outputResolution.xy);
+	vec4 texelSize = vec4( 1.0 / textureSrcResolution.xy, textureSrcResolution.xy);
 	colorOut0 = vec4(BicubicHermiteTexture(passUV, texelSize), 1.0);
 }
 )";
 
 RendererOutputShader::RendererOutputShader(const std::string& vertex_source, const std::string& fragment_source)
 {
-	m_vertex_shader = g_renderer->shader_create(RendererShader::ShaderType::kVertex, 0, 0, vertex_source, false, false);
-	m_fragment_shader = g_renderer->shader_create(RendererShader::ShaderType::kFragment, 0, 0, fragment_source, false, false);
+	auto finalFragmentSrc = PrependFragmentPreamble(fragment_source);
+
+	m_vertex_shader.reset(g_renderer->shader_create(RendererShader::ShaderType::kVertex, 0, 0, vertex_source, false, false));
+	m_fragment_shader.reset(g_renderer->shader_create(RendererShader::ShaderType::kFragment, 0, 0, finalFragmentSrc, false, false));
 
 	m_vertex_shader->PreponeCompilation(true);
 	m_fragment_shader->PreponeCompilation(true);
@@ -163,74 +132,45 @@ RendererOutputShader::RendererOutputShader(const std::string& vertex_source, con
 
 	if (g_renderer->GetType() == RendererAPI::OpenGL)
 	{
-		m_attributes[0].m_loc_texture_src_resolution = m_vertex_shader->GetUniformLocation("textureSrcResolution");
-		m_attributes[0].m_loc_input_resolution = m_vertex_shader->GetUniformLocation("inputResolution");
-		m_attributes[0].m_loc_output_resolution = m_vertex_shader->GetUniformLocation("outputResolution");
+		m_uniformLocations[0].m_loc_textureSrcResolution = m_vertex_shader->GetUniformLocation("textureSrcResolution");
+		m_uniformLocations[0].m_loc_nativeResolution = m_vertex_shader->GetUniformLocation("nativeResolution");
+		m_uniformLocations[0].m_loc_outputResolution = m_vertex_shader->GetUniformLocation("outputResolution");
 
-		m_attributes[1].m_loc_texture_src_resolution = m_fragment_shader->GetUniformLocation("textureSrcResolution");
-		m_attributes[1].m_loc_input_resolution = m_fragment_shader->GetUniformLocation("inputResolution");
-		m_attributes[1].m_loc_output_resolution = m_fragment_shader->GetUniformLocation("outputResolution");
+		m_uniformLocations[1].m_loc_textureSrcResolution = m_fragment_shader->GetUniformLocation("textureSrcResolution");
+		m_uniformLocations[1].m_loc_nativeResolution = m_fragment_shader->GetUniformLocation("nativeResolution");
+		m_uniformLocations[1].m_loc_outputResolution = m_fragment_shader->GetUniformLocation("outputResolution");
 	}
-	else
-	{
-		cemuLog_logDebug(LogType::Force, "RendererOutputShader() - todo for Vulkan");
-		m_attributes[0].m_loc_texture_src_resolution = -1;
-		m_attributes[0].m_loc_input_resolution = -1;
-		m_attributes[0].m_loc_output_resolution = -1;
-
-		m_attributes[1].m_loc_texture_src_resolution = -1;
-		m_attributes[1].m_loc_input_resolution = -1;
-		m_attributes[1].m_loc_output_resolution = -1;
-	}
-
 }
 
-void RendererOutputShader::SetUniformParameters(const LatteTextureView& texture_view, const Vector2i& input_res, const Vector2i& output_res) const
+void RendererOutputShader::SetUniformParameters(const LatteTextureView& texture_view, const Vector2i& output_res) const
 {
-	float res[2];
-	// vertex shader
-	if (m_attributes[0].m_loc_texture_src_resolution != -1)
-	{ 
-		res[0] = (float)texture_view.baseTexture->width;
-		res[1] = (float)texture_view.baseTexture->height;
-		m_vertex_shader->SetUniform2fv(m_attributes[0].m_loc_texture_src_resolution, res, 1);
-	}
+	sint32 effectiveWidth, effectiveHeight;
+	texture_view.baseTexture->GetEffectiveSize(effectiveWidth, effectiveHeight, 0);
+	auto setUniforms = [&](RendererShader* shader, const UniformLocations& locations){
+	  float res[2];
+	  if (locations.m_loc_textureSrcResolution != -1)
+	  {
+		  res[0] = (float)effectiveWidth;
+		  res[1] = (float)effectiveHeight;
+		  shader->SetUniform2fv(locations.m_loc_textureSrcResolution, res, 1);
+	  }
 
-	if (m_attributes[0].m_loc_input_resolution != -1)
-	{
-		res[0] = (float)input_res.x;
-		res[1] = (float)input_res.y;
-		m_vertex_shader->SetUniform2fv(m_attributes[0].m_loc_input_resolution, res, 1);
-	}
+	  if (locations.m_loc_nativeResolution != -1)
+	  {
+		  res[0] = (float)texture_view.baseTexture->width;
+		  res[1] = (float)texture_view.baseTexture->height;
+		  shader->SetUniform2fv(locations.m_loc_nativeResolution, res, 1);
+	  }
 
-	if (m_attributes[0].m_loc_output_resolution != -1)
-	{
-		res[0] = (float)output_res.x;
-		res[1] = (float)output_res.y;
-		m_vertex_shader->SetUniform2fv(m_attributes[0].m_loc_output_resolution, res, 1);
-	}
-
-	// fragment shader
-	if (m_attributes[1].m_loc_texture_src_resolution != -1)
-	{
-		res[0] = (float)texture_view.baseTexture->width;
-		res[1] = (float)texture_view.baseTexture->height;
-		m_fragment_shader->SetUniform2fv(m_attributes[1].m_loc_texture_src_resolution, res, 1);
-	}
-
-	if (m_attributes[1].m_loc_input_resolution != -1)
-	{
-		res[0] = (float)input_res.x;
-		res[1] = (float)input_res.y;
-		m_fragment_shader->SetUniform2fv(m_attributes[1].m_loc_input_resolution, res, 1);
-	}
-
-	if (m_attributes[1].m_loc_output_resolution != -1)
-	{
-		res[0] = (float)output_res.x;
-		res[1] = (float)output_res.y;
-		m_fragment_shader->SetUniform2fv(m_attributes[1].m_loc_output_resolution, res, 1);
-	}
+	  if (locations.m_loc_outputResolution != -1)
+	  {
+		  res[0] = (float)output_res.x;
+		  res[1] = (float)output_res.y;
+		  shader->SetUniform2fv(locations.m_loc_outputResolution, res, 1);
+	  }
+	};
+	setUniforms(m_vertex_shader.get(), m_uniformLocations[0]);
+	setUniforms(m_fragment_shader.get(), m_uniformLocations[1]);
 }
 
 RendererOutputShader* RendererOutputShader::s_copy_shader;
@@ -247,8 +187,8 @@ std::string RendererOutputShader::GetOpenGlVertexSource(bool render_upside_down)
 	// vertex shader
 	std::ostringstream vertex_source;
 		vertex_source <<
-			R"(#version 400
-out vec2 passUV;
+			R"(#version 420
+layout(location = 0) smooth out vec2 passUV;
 
 out gl_PerVertex 
 { 
@@ -341,6 +281,27 @@ void main(){
 )";
 		return vertex_source.str();
 }
+
+std::string RendererOutputShader::PrependFragmentPreamble(const std::string& shaderSrc)
+{
+	return R"(#version 430
+#ifdef VULKAN
+layout(push_constant) uniform pc {
+	vec2 textureSrcResolution;
+	vec2 nativeResolution;
+	vec2 outputResolution;
+};
+#else
+uniform vec2 textureSrcResolution;
+uniform vec2 nativeResolution;
+uniform vec2 outputResolution;
+#endif
+
+layout(location = 0) smooth in vec2 passUV;
+layout(binding = 0) uniform sampler2D textureSrc;
+layout(location = 0) out vec4 colorOut0;
+)" + shaderSrc;
+}
 void RendererOutputShader::InitializeStatic()
 {
 	std::string vertex_source, vertex_source_ud;
@@ -349,28 +310,30 @@ void RendererOutputShader::InitializeStatic()
 	{
 		vertex_source = GetOpenGlVertexSource(false);
 		vertex_source_ud = GetOpenGlVertexSource(true);
-
-		s_copy_shader = new RendererOutputShader(vertex_source, s_copy_shader_source);
-		s_copy_shader_ud = new RendererOutputShader(vertex_source_ud, s_copy_shader_source);
-
-		s_bicubic_shader = new RendererOutputShader(vertex_source, s_bicubic_shader_source);
-		s_bicubic_shader_ud = new RendererOutputShader(vertex_source_ud, s_bicubic_shader_source);
-
-		s_hermit_shader = new RendererOutputShader(vertex_source, s_hermite_shader_source);
-		s_hermit_shader_ud = new RendererOutputShader(vertex_source_ud, s_hermite_shader_source);
 	}
 	else
 	{
 		vertex_source = GetVulkanVertexSource(false);
 		vertex_source_ud = GetVulkanVertexSource(true);
-
-		s_copy_shader = new RendererOutputShader(vertex_source, s_copy_shader_source);
-		s_copy_shader_ud = new RendererOutputShader(vertex_source_ud, s_copy_shader_source);
-
-	/*	s_bicubic_shader = new RendererOutputShader(vertex_source, s_bicubic_shader_source); TODO
-		s_bicubic_shader_ud = new RendererOutputShader(vertex_source_ud, s_bicubic_shader_source);
-
-		s_hermit_shader = new RendererOutputShader(vertex_source, s_hermite_shader_source);
-		s_hermit_shader_ud = new RendererOutputShader(vertex_source_ud, s_hermite_shader_source);*/
 	}
+	s_copy_shader = new RendererOutputShader(vertex_source, s_copy_shader_source);
+	s_copy_shader_ud = new RendererOutputShader(vertex_source_ud, s_copy_shader_source);
+
+	s_bicubic_shader = new RendererOutputShader(vertex_source, s_bicubic_shader_source);
+	s_bicubic_shader_ud = new RendererOutputShader(vertex_source_ud, s_bicubic_shader_source);
+
+	s_hermit_shader = new RendererOutputShader(vertex_source, s_hermite_shader_source);
+	s_hermit_shader_ud = new RendererOutputShader(vertex_source_ud, s_hermite_shader_source);
+}
+
+void RendererOutputShader::ShutdownStatic()
+{
+	delete s_copy_shader;
+	delete s_copy_shader_ud;
+
+	delete s_bicubic_shader;
+	delete s_bicubic_shader_ud;
+
+	delete s_hermit_shader;
+	delete s_hermit_shader_ud;
 }

src/Cafe/HW/Latte/Renderer/RendererOuputShader.h

@@ -17,19 +17,20 @@ public:
 	RendererOutputShader(const std::string& vertex_source, const std::string& fragment_source);
 	virtual ~RendererOutputShader() = default;
 
-	void SetUniformParameters(const LatteTextureView& texture_view, const Vector2i& input_res, const Vector2i& output_res) const;
+	void SetUniformParameters(const LatteTextureView& texture_view, const Vector2i& output_res) const;
 
 	RendererShader* GetVertexShader() const
 	{
-		return m_vertex_shader;
+		return m_vertex_shader.get();
 	}
 
 	RendererShader* GetFragmentShader() const
 	{
-		return m_fragment_shader;
+		return m_fragment_shader.get();
 	}
 
 	static void InitializeStatic();
+	static void ShutdownStatic();
 
 	static RendererOutputShader* s_copy_shader;
 	static RendererOutputShader* s_copy_shader_ud;
@@ -43,16 +44,18 @@ public:
 	static std::string GetVulkanVertexSource(bool render_upside_down);
 	static std::string GetOpenGlVertexSource(bool render_upside_down);
 
-protected:
-	RendererShader* m_vertex_shader;
-	RendererShader* m_fragment_shader;
+	static std::string PrependFragmentPreamble(const std::string& shaderSrc);
 
-	struct
+protected:
+	std::unique_ptr<RendererShader> m_vertex_shader;
+	std::unique_ptr<RendererShader> m_fragment_shader;
+
+	struct UniformLocations
 	{
-		sint32 m_loc_texture_src_resolution = -1;
-		sint32 m_loc_input_resolution = -1;
-		sint32 m_loc_output_resolution = -1;
-	} m_attributes[2]{};
+		sint32 m_loc_textureSrcResolution = -1;
+		sint32 m_loc_nativeResolution = -1;
+		sint32 m_loc_outputResolution = -1;
+	} m_uniformLocations[2]{};
 
 private:
 	static const std::string s_copy_shader_source;

src/Cafe/HW/Latte/Renderer/Vulkan/LatteTextureViewVk.cpp

@@ -202,6 +202,13 @@ VkSampler LatteTextureViewVk::GetDefaultTextureSampler(bool useLinearTexFilter)
 	VkSamplerCreateInfo samplerInfo{};
 	samplerInfo.sType = VK_STRUCTURE_TYPE_SAMPLER_CREATE_INFO;
 
+	// emulate OpenGL minFilters
+	// see note under: https://docs.vulkan.org/spec/latest/chapters/samplers.html#VkSamplerCreateInfo
+	// if maxLod = 0 then magnification is always performed
+	samplerInfo.mipmapMode = VK_SAMPLER_MIPMAP_MODE_NEAREST;
+	samplerInfo.minLod = 0.0f;
+	samplerInfo.maxLod = 0.25f;
+
 	if (useLinearTexFilter)
 	{
 		samplerInfo.magFilter = VK_FILTER_LINEAR;
@@ -212,6 +219,9 @@ VkSampler LatteTextureViewVk::GetDefaultTextureSampler(bool useLinearTexFilter)
 		samplerInfo.magFilter = VK_FILTER_NEAREST;
 		samplerInfo.minFilter = VK_FILTER_NEAREST;
 	}
+	samplerInfo.addressModeU = VK_SAMPLER_ADDRESS_MODE_CLAMP_TO_EDGE;
+	samplerInfo.addressModeV = VK_SAMPLER_ADDRESS_MODE_CLAMP_TO_EDGE;
+	samplerInfo.addressModeW = VK_SAMPLER_ADDRESS_MODE_CLAMP_TO_EDGE;
 
 	if (vkCreateSampler(m_device, &samplerInfo, nullptr, &sampler) != VK_SUCCESS)
 	{

src/Cafe/HW/Latte/Renderer/Vulkan/RendererShaderVk.cpp

@@ -211,6 +211,9 @@ RendererShaderVk::~RendererShaderVk()
 {
 	while (!list_pipelineInfo.empty())
 		delete list_pipelineInfo[0];
+
+	VkDevice vkDev = VulkanRenderer::GetInstance()->GetLogicalDevice();
+	vkDestroyShaderModule(vkDev, m_shader_module, nullptr);
 }
 
 void RendererShaderVk::Init()

src/Cafe/HW/Latte/Renderer/Vulkan/SwapchainInfoVk.cpp

@@ -60,7 +60,7 @@ void SwapchainInfoVk::Create()
 	VkAttachmentDescription colorAttachment = {};
 	colorAttachment.format = m_surfaceFormat.format;
 	colorAttachment.samples = VK_SAMPLE_COUNT_1_BIT;
-	colorAttachment.loadOp = VK_ATTACHMENT_LOAD_OP_LOAD;
+	colorAttachment.loadOp = VK_ATTACHMENT_LOAD_OP_DONT_CARE;
 	colorAttachment.storeOp = VK_ATTACHMENT_STORE_OP_STORE;
 	colorAttachment.stencilLoadOp = VK_ATTACHMENT_LOAD_OP_DONT_CARE;
 	colorAttachment.stencilStoreOp = VK_ATTACHMENT_STORE_OP_DONT_CARE;

src/Cafe/HW/Latte/Renderer/Vulkan/SwapchainInfoVk.h

@@ -70,6 +70,7 @@ struct SwapchainInfoVk
 	VkSurfaceFormatKHR m_surfaceFormat{};
 	VkSwapchainKHR m_swapchain{};
 	Vector2i m_desiredExtent{};
+	VkExtent2D m_actualExtent{};
 	uint32 swapchainImageIndex = (uint32)-1;
 	uint64 m_presentId = 1;
 	uint64 m_queueDepth = 0; // number of frames with pending presentation requests
@@ -92,5 +93,4 @@ private:
 	VkSemaphore m_currentSemaphore = VK_NULL_HANDLE;
 
 	std::array<uint32, 2> m_swapchainQueueFamilyIndices;
-	VkExtent2D m_actualExtent{};
 };

src/Cafe/HW/Latte/Renderer/Vulkan/TextureReadbackVk.cpp

@@ -22,7 +22,7 @@ uint32 LatteTextureReadbackInfoVk::GetImageSize(LatteTextureView* textureView)
 		cemu_assert(textureFormat == VK_FORMAT_R8G8B8A8_UNORM);
 		return baseTexture->width * baseTexture->height * 4;
 	}
-	else if (textureView->format == Latte::E_GX2SURFFMT::R8_UNORM)
+	else if (textureView->format == Latte::E_GX2SURFFMT::R8_UNORM )
 	{
 		cemu_assert(textureFormat == VK_FORMAT_R8_UNORM);
 		return baseTexture->width * baseTexture->height * 1;
@@ -79,6 +79,13 @@ uint32 LatteTextureReadbackInfoVk::GetImageSize(LatteTextureView* textureView)
 		// todo - if driver does not support VK_FORMAT_D24_UNORM_S8_UINT this is represented as VK_FORMAT_D32_SFLOAT_S8_UINT which is 8 bytes
 		return baseTexture->width * baseTexture->height * 4;
 	}
+	else if (textureView->format == Latte::E_GX2SURFFMT::R5_G6_B5_UNORM )
+	{
+		if(textureFormat == VK_FORMAT_R5G6B5_UNORM_PACK16){
+			return baseTexture->width * baseTexture->height * 2;
+		}	
+		return 0;
+	}
 	else
 	{
 		cemuLog_log(LogType::Force, "Unsupported texture readback format {:04x}", (uint32)textureView->format);

src/Cafe/HW/Latte/Renderer/Vulkan/VKRBase.h

@@ -19,7 +19,7 @@ public:
 
 	virtual ~VKRMoveableRefCounter()
 	{
-		cemu_assert_debug(refCount == 0);
+		cemu_assert_debug(m_refCount == 0);
 
 		// remove references
 #ifdef CEMU_DEBUG_ASSERT
@@ -30,7 +30,11 @@ public:
 		}
 #endif
 		for (auto itr : refs)
-			itr->ref->refCount--;
+		{
+			itr->ref->m_refCount--;
+			if (itr->ref->m_refCount == 0)
+				itr->ref->RefCountReachedZero();
+		}
 		refs.clear();
 		delete selfRef;
 		selfRef = nullptr;
@@ -41,8 +45,8 @@ public:
 	VKRMoveableRefCounter(VKRMoveableRefCounter&& rhs) noexcept
 	{
 		this->refs = std::move(rhs.refs);
-		this->refCount = rhs.refCount;
-		rhs.refCount = 0;
+		this->m_refCount = rhs.m_refCount;
+		rhs.m_refCount = 0;
 		this->selfRef = rhs.selfRef;
 		rhs.selfRef = nullptr;
 		this->selfRef->ref = this;
@@ -57,7 +61,7 @@ public:
 	void addRef(VKRMoveableRefCounter* refTarget)
 	{
 		this->refs.emplace_back(refTarget->selfRef);
-		refTarget->refCount++;
+		refTarget->m_refCount++;
 
 #ifdef CEMU_DEBUG_ASSERT
 		// add reverse ref
@@ -68,16 +72,23 @@ public:
 	// methods to directly increment/decrement ref counter (for situations where no external object is available)
 	void incRef()
 	{
-		this->refCount++;
+		m_refCount++;
 	}
 
 	void decRef()
 	{
-		this->refCount--;
+		m_refCount--;
+		if (m_refCount == 0)
+			RefCountReachedZero();
 	}
 
 protected:
-	int refCount{};
+  	virtual void RefCountReachedZero()
+	{
+		// does nothing by default
+	}
+
+	int m_refCount{};
 private:
 	VKRMoveableRefCounterRef* selfRef;
 	std::vector<VKRMoveableRefCounterRef*> refs;
@@ -88,7 +99,7 @@ private:
 	void moveObj(VKRMoveableRefCounter&& rhs)
 	{
 		this->refs = std::move(rhs.refs);
-		this->refCount = rhs.refCount;
+		this->m_refCount = rhs.m_refCount;
 		this->selfRef = rhs.selfRef;
 		this->selfRef->ref = this;
 	}
@@ -131,6 +142,25 @@ public:
 	VkSampler m_textureDefaultSampler[2] = { VK_NULL_HANDLE, VK_NULL_HANDLE }; // relict from LatteTextureViewVk, get rid of it eventually
 };
 
+
+class VKRObjectSampler : public VKRDestructibleObject
+{
+  public:
+	VKRObjectSampler(VkSamplerCreateInfo* samplerInfo);
+	~VKRObjectSampler() override;
+
+	static VKRObjectSampler* GetOrCreateSampler(VkSamplerCreateInfo* samplerInfo);
+	static void DestroyCache();
+
+	void RefCountReachedZero() override; // sampler objects are destroyed when not referenced anymore
+
+	VkSampler GetSampler() const { return m_sampler; }
+  private:
+	static std::unordered_map<uint64, VKRObjectSampler*> s_samplerCache;
+	VkSampler m_sampler{ VK_NULL_HANDLE };
+	uint64 m_hash;
+};
+
 class VKRObjectRenderPass : public VKRDestructibleObject
 {
 public:

src/Cafe/HW/Latte/Renderer/Vulkan/VKRMemoryManager.cpp

@@ -4,6 +4,14 @@
 
 /* VKRSynchronizedMemoryBuffer */
 
+VKRSynchronizedRingAllocator::~VKRSynchronizedRingAllocator()
+{
+	for(auto& buf : m_buffers)
+	{
+		m_vkrMemMgr->DeleteBuffer(buf.vk_buffer, buf.vk_mem);
+	}
+}
+
 void VKRSynchronizedRingAllocator::addUploadBufferSyncPoint(AllocatorBuffer_t& buffer, uint32 offset)
 {
 	auto cmdBufferId = m_vkr->GetCurrentCommandBufferId();
@@ -23,11 +31,11 @@ void VKRSynchronizedRingAllocator::allocateAdditionalUploadBuffer(uint32 sizeReq
 	AllocatorBuffer_t newBuffer{};
 	newBuffer.writeIndex = 0;
 	newBuffer.basePtr = nullptr;
-	if (m_bufferType == BUFFER_TYPE::STAGING)
+	if (m_bufferType == VKR_BUFFER_TYPE::STAGING)
 		m_vkrMemMgr->CreateBuffer(bufferAllocSize, VK_BUFFER_USAGE_TRANSFER_SRC_BIT, VK_MEMORY_PROPERTY_HOST_VISIBLE_BIT, newBuffer.vk_buffer, newBuffer.vk_mem);
-	else if (m_bufferType == BUFFER_TYPE::INDEX)
+	else if (m_bufferType == VKR_BUFFER_TYPE::INDEX)
 		m_vkrMemMgr->CreateBuffer(bufferAllocSize, VK_BUFFER_USAGE_INDEX_BUFFER_BIT, VK_MEMORY_PROPERTY_HOST_VISIBLE_BIT | VK_MEMORY_PROPERTY_HOST_COHERENT_BIT, newBuffer.vk_buffer, newBuffer.vk_mem);
-	else if (m_bufferType == BUFFER_TYPE::STRIDE)
+	else if (m_bufferType == VKR_BUFFER_TYPE::STRIDE)
 		m_vkrMemMgr->CreateBuffer(bufferAllocSize, VK_BUFFER_USAGE_VERTEX_BUFFER_BIT, VK_MEMORY_PROPERTY_HOST_VISIBLE_BIT | VK_MEMORY_PROPERTY_HOST_COHERENT_BIT, newBuffer.vk_buffer, newBuffer.vk_mem);
 	else
 		cemu_assert_debug(false);
@@ -53,7 +61,7 @@ VKRSynchronizedRingAllocator::AllocatorReservation_t VKRSynchronizedRingAllocato
 		uint32 distanceToSyncPoint;
 		if (!itr.queue_syncPoints.empty())
 		{
-			if(itr.queue_syncPoints.front().offset < itr.writeIndex)
+			if (itr.queue_syncPoints.front().offset < itr.writeIndex)
 				distanceToSyncPoint = 0xFFFFFFFF;
 			else
 				distanceToSyncPoint = itr.queue_syncPoints.front().offset - itr.writeIndex;
@@ -100,7 +108,7 @@ VKRSynchronizedRingAllocator::AllocatorReservation_t VKRSynchronizedRingAllocato
 
 void VKRSynchronizedRingAllocator::FlushReservation(AllocatorReservation_t& uploadReservation)
 {
-	cemu_assert_debug(m_bufferType == BUFFER_TYPE::STAGING); // only the staging buffer isn't coherent
+	cemu_assert_debug(m_bufferType == VKR_BUFFER_TYPE::STAGING); // only the staging buffer isn't coherent
 	// todo - use nonCoherentAtomSize for flush size (instead of hardcoded constant)
 	VkMappedMemoryRange flushedRange{};
 	flushedRange.sType = VK_STRUCTURE_TYPE_MAPPED_MEMORY_RANGE;
@@ -167,15 +175,88 @@ void VKRSynchronizedRingAllocator::GetStats(uint32& numBuffers, size_t& totalBuf
 	}
 }
 
+/* VKRSynchronizedHeapAllocator */
+
+VKRSynchronizedHeapAllocator::VKRSynchronizedHeapAllocator(class VKRMemoryManager* vkMemoryManager, VKR_BUFFER_TYPE bufferType, size_t minimumBufferAllocSize)
+	: m_vkrMemMgr(vkMemoryManager), m_chunkedHeap(bufferType, minimumBufferAllocSize) {};
+
+VKRSynchronizedHeapAllocator::AllocatorReservation* VKRSynchronizedHeapAllocator::AllocateBufferMemory(uint32 size, uint32 alignment)
+{
+	CHAddr addr = m_chunkedHeap.alloc(size, alignment);
+	m_activeAllocations.emplace_back(addr);
+	AllocatorReservation* res = m_poolAllocatorReservation.allocObj();
+	res->bufferIndex = addr.chunkIndex;
+	res->bufferOffset = addr.offset;
+	res->size = size;
+	res->memPtr = m_chunkedHeap.GetChunkPtr(addr.chunkIndex) + addr.offset;
+	m_chunkedHeap.GetChunkVkMemInfo(addr.chunkIndex, res->vkBuffer, res->vkMem);
+	return res;
+}
+
+void VKRSynchronizedHeapAllocator::FreeReservation(AllocatorReservation* uploadReservation)
+{
+	// put the allocation on a delayed release queue for the current command buffer
+	uint64 currentCommandBufferId = VulkanRenderer::GetInstance()->GetCurrentCommandBufferId();
+	auto it = std::find_if(m_activeAllocations.begin(), m_activeAllocations.end(), [&uploadReservation](const TrackedAllocation& allocation) { return allocation.allocation.chunkIndex == uploadReservation->bufferIndex && allocation.allocation.offset == uploadReservation->bufferOffset; });
+	cemu_assert_debug(it != m_activeAllocations.end());
+	m_releaseQueue[currentCommandBufferId].emplace_back(it->allocation);
+	m_activeAllocations.erase(it);
+	m_poolAllocatorReservation.freeObj(uploadReservation);
+}
+
+void VKRSynchronizedHeapAllocator::FlushReservation(AllocatorReservation* uploadReservation)
+{
+	if (m_chunkedHeap.RequiresFlush(uploadReservation->bufferIndex))
+	{
+		VkMappedMemoryRange flushedRange{};
+		flushedRange.sType = VK_STRUCTURE_TYPE_MAPPED_MEMORY_RANGE;
+		flushedRange.memory = uploadReservation->vkMem;
+		flushedRange.offset = uploadReservation->bufferOffset;
+		flushedRange.size = uploadReservation->size;
+		vkFlushMappedMemoryRanges(VulkanRenderer::GetInstance()->GetLogicalDevice(), 1, &flushedRange);
+	}
+}
+
+void VKRSynchronizedHeapAllocator::CleanupBuffer(uint64 latestFinishedCommandBufferId)
+{
+	auto it = m_releaseQueue.begin();
+	while (it != m_releaseQueue.end())
+	{
+		if (it->first <= latestFinishedCommandBufferId)
+		{
+			// release allocations
+			for(auto& addr : it->second)
+				m_chunkedHeap.free(addr);
+			it = m_releaseQueue.erase(it);
+			continue;
+		}
+		it++;
+	}
+}
+
+void VKRSynchronizedHeapAllocator::GetStats(uint32& numBuffers, size_t& totalBufferSize, size_t& freeBufferSize) const
+{
+	m_chunkedHeap.GetStats(numBuffers, totalBufferSize, freeBufferSize);
+}
+
 /* VkTextureChunkedHeap */
 
+VkTextureChunkedHeap::~VkTextureChunkedHeap()
+{
+	VkDevice device = VulkanRenderer::GetInstance()->GetLogicalDevice();
+	for (auto& i : m_list_chunkInfo)
+	{
+		vkFreeMemory(device, i.mem, nullptr);
+	}
+}
+
 uint32 VkTextureChunkedHeap::allocateNewChunk(uint32 chunkIndex, uint32 minimumAllocationSize)
 {
 	cemu_assert_debug(m_list_chunkInfo.size() == chunkIndex);
 	m_list_chunkInfo.resize(m_list_chunkInfo.size() + 1);
 
 	// pad minimumAllocationSize to 32KB alignment
-	minimumAllocationSize = (minimumAllocationSize + (32*1024-1)) & ~(32 * 1024 - 1);
+	minimumAllocationSize = (minimumAllocationSize + (32 * 1024 - 1)) & ~(32 * 1024 - 1);
 
 	uint32 allocationSize = 1024 * 1024 * 128;
 	if (chunkIndex == 0)
@@ -189,8 +270,7 @@ uint32 VkTextureChunkedHeap::allocateNewChunk(uint32 chunkIndex, uint32 minimumA
 	std::vector<uint32> deviceLocalMemoryTypeIndices = m_vkrMemoryManager->FindMemoryTypes(m_typeFilter, VK_MEMORY_PROPERTY_DEVICE_LOCAL_BIT);
 	std::vector<uint32> hostLocalMemoryTypeIndices = m_vkrMemoryManager->FindMemoryTypes(m_typeFilter, 0);
 	// remove device local memory types from host local vector
-	auto pred = [&deviceLocalMemoryTypeIndices](const uint32& v) ->bool
-	{
+	auto pred = [&deviceLocalMemoryTypeIndices](const uint32& v) -> bool {
 		return std::find(deviceLocalMemoryTypeIndices.begin(), deviceLocalMemoryTypeIndices.end(), v) != deviceLocalMemoryTypeIndices.end();
 	};
 	hostLocalMemoryTypeIndices.erase(std::remove_if(hostLocalMemoryTypeIndices.begin(), hostLocalMemoryTypeIndices.end(), pred), hostLocalMemoryTypeIndices.end());
@@ -206,7 +286,7 @@ uint32 VkTextureChunkedHeap::allocateNewChunk(uint32 chunkIndex, uint32 minimumA
 			allocInfo.memoryTypeIndex = memType;
 
 			VkDeviceMemory imageMemory;
-			VkResult r = vkAllocateMemory(m_device, &allocInfo, nullptr, &imageMemory);
+			VkResult r = vkAllocateMemory(VulkanRenderer::GetInstance()->GetLogicalDevice(), &allocInfo, nullptr, &imageMemory);
 			if (r != VK_SUCCESS)
 				continue;
 			m_list_chunkInfo[chunkIndex].mem = imageMemory;
@@ -221,7 +301,7 @@ uint32 VkTextureChunkedHeap::allocateNewChunk(uint32 chunkIndex, uint32 minimumA
 			allocInfo.memoryTypeIndex = memType;
 
 			VkDeviceMemory imageMemory;
-			VkResult r = vkAllocateMemory(m_device, &allocInfo, nullptr, &imageMemory);
+			VkResult r = vkAllocateMemory(VulkanRenderer::GetInstance()->GetLogicalDevice(), &allocInfo, nullptr, &imageMemory);
 			if (r != VK_SUCCESS)
 				continue;
 			m_list_chunkInfo[chunkIndex].mem = imageMemory;
@@ -238,28 +318,76 @@ uint32 VkTextureChunkedHeap::allocateNewChunk(uint32 chunkIndex, uint32 minimumA
 	return 0;
 }
 
-uint32_t VKRMemoryManager::FindMemoryType(uint32_t typeFilter, VkMemoryPropertyFlags properties) const
-{
-	VkPhysicalDeviceMemoryProperties memProperties;
-	vkGetPhysicalDeviceMemoryProperties(m_vkr->GetPhysicalDevice(), &memProperties);
+/* VkBufferChunkedHeap */
 
-	for (uint32 i = 0; i < memProperties.memoryTypeCount; i++)
+VKRBuffer* VKRBuffer::Create(VKR_BUFFER_TYPE bufferType, size_t bufferSize, VkMemoryPropertyFlags properties)
+{
+	auto* memMgr = VulkanRenderer::GetInstance()->GetMemoryManager();
+	VkBuffer buffer;
+	VkDeviceMemory bufferMemory;
+	bool allocSuccess;
+	if (bufferType == VKR_BUFFER_TYPE::STAGING)
+		allocSuccess = memMgr->CreateBuffer(bufferSize, VK_BUFFER_USAGE_TRANSFER_SRC_BIT, properties, buffer, bufferMemory);
+	else if (bufferType == VKR_BUFFER_TYPE::INDEX)
+		allocSuccess = memMgr->CreateBuffer(bufferSize, VK_BUFFER_USAGE_INDEX_BUFFER_BIT, properties, buffer, bufferMemory);
+	else if (bufferType == VKR_BUFFER_TYPE::STRIDE)
+		allocSuccess = memMgr->CreateBuffer(bufferSize, VK_BUFFER_USAGE_VERTEX_BUFFER_BIT, properties, buffer, bufferMemory);
+	else
+		cemu_assert_debug(false);
+	if (!allocSuccess)
+		return nullptr;
+
+	VKRBuffer* bufferObj = new VKRBuffer(buffer, bufferMemory);
+	// if host visible, then map buffer
+	void* data = nullptr;
+	if (properties & VK_MEMORY_PROPERTY_HOST_VISIBLE_BIT)
 	{
-		if ((typeFilter & (1 << i)) != 0 && (memProperties.memoryTypes[i].propertyFlags & properties) == properties)
-			return i;
+		vkMapMemory(VulkanRenderer::GetInstance()->GetLogicalDevice(), bufferMemory, 0, bufferSize, 0, &data);
+		bufferObj->m_requiresFlush = !HAS_FLAG(properties, VK_MEMORY_PROPERTY_HOST_COHERENT_BIT);
 	}
-	m_vkr->UnrecoverableError(fmt::format("failed to find suitable memory type ({0:#08x} {1:#08x})", typeFilter, properties).c_str());
-	return 0;
+	bufferObj->m_mappedMemory = (uint8*)data;
+	return bufferObj;
 }
 
-bool VKRMemoryManager::FindMemoryType2(uint32 typeFilter, VkMemoryPropertyFlags properties, uint32& memoryIndex) const
+VKRBuffer::~VKRBuffer()
+{
+	if (m_mappedMemory)
+		vkUnmapMemory(VulkanRenderer::GetInstance()->GetLogicalDevice(), m_bufferMemory);
+	if (m_bufferMemory != VK_NULL_HANDLE)
+		vkFreeMemory(VulkanRenderer::GetInstance()->GetLogicalDevice(), m_bufferMemory, nullptr);
+	if (m_buffer != VK_NULL_HANDLE)
+		vkDestroyBuffer(VulkanRenderer::GetInstance()->GetLogicalDevice(), m_buffer, nullptr);
+}
+
+VkBufferChunkedHeap::~VkBufferChunkedHeap()
+{
+	for (auto& chunk : m_chunkBuffers)
+		delete chunk;
+}
+
+uint32 VkBufferChunkedHeap::allocateNewChunk(uint32 chunkIndex, uint32 minimumAllocationSize)
+{
+	size_t allocationSize = std::max<size_t>(m_minimumBufferAllocationSize, minimumAllocationSize);
+	VKRBuffer* buffer = VKRBuffer::Create(m_bufferType, allocationSize, VK_MEMORY_PROPERTY_HOST_VISIBLE_BIT | VK_MEMORY_PROPERTY_HOST_COHERENT_BIT);
+	if(!buffer)
+		buffer = VKRBuffer::Create(m_bufferType, allocationSize, VK_MEMORY_PROPERTY_HOST_VISIBLE_BIT);
+	if(!buffer)
+		VulkanRenderer::GetInstance()->UnrecoverableError("Failed to allocate buffer memory for VkBufferChunkedHeap");
+	cemu_assert_debug(buffer);
+	cemu_assert_debug(m_chunkBuffers.size() == chunkIndex);
+	m_chunkBuffers.emplace_back(buffer);
+	// todo - VK_MEMORY_PROPERTY_DEVICE_LOCAL_BIT might be worth it?
+	return allocationSize;
+}
+
+bool VKRMemoryManager::FindMemoryType(uint32 typeFilter, VkMemoryPropertyFlags properties, uint32& memoryIndex) const
 {
 	VkPhysicalDeviceMemoryProperties memProperties;
 	vkGetPhysicalDeviceMemoryProperties(m_vkr->GetPhysicalDevice(), &memProperties);
 
 	for (uint32_t i = 0; i < memProperties.memoryTypeCount; i++)
 	{
-		if (typeFilter & (1 << i) && memProperties.memoryTypes[i].propertyFlags == properties)
+		if (typeFilter & (1 << i) && (memProperties.memoryTypes[i].propertyFlags & properties) == properties)
 		{
 			memoryIndex = i;
 			return true;
@@ -330,31 +458,7 @@ size_t VKRMemoryManager::GetTotalMemoryForBufferType(VkBufferUsageFlags usage, V
 	return total;
 }
 
-void VKRMemoryManager::CreateBuffer(VkDeviceSize size, VkBufferUsageFlags usage, VkMemoryPropertyFlags properties, VkBuffer& buffer, VkDeviceMemory& bufferMemory) const
-{
-	VkBufferCreateInfo bufferInfo{};
-	bufferInfo.sType = VK_STRUCTURE_TYPE_BUFFER_CREATE_INFO;
-	bufferInfo.usage = usage;
-	bufferInfo.size = size;
-	bufferInfo.sharingMode = VK_SHARING_MODE_EXCLUSIVE;
-	if (vkCreateBuffer(m_vkr->GetLogicalDevice(), &bufferInfo, nullptr, &buffer) != VK_SUCCESS)
-		m_vkr->UnrecoverableError("Failed to create buffer");
-
-	VkMemoryRequirements memRequirements;
-	vkGetBufferMemoryRequirements(m_vkr->GetLogicalDevice(), buffer, &memRequirements);
-
-	VkMemoryAllocateInfo allocInfo{};
-	allocInfo.sType = VK_STRUCTURE_TYPE_MEMORY_ALLOCATE_INFO;
-	allocInfo.allocationSize = memRequirements.size;
-	allocInfo.memoryTypeIndex = FindMemoryType(memRequirements.memoryTypeBits, properties);
-
-	if (vkAllocateMemory(m_vkr->GetLogicalDevice(), &allocInfo, nullptr, &bufferMemory) != VK_SUCCESS)
-		m_vkr->UnrecoverableError("Failed to allocate buffer memory");
-	if (vkBindBufferMemory(m_vkr->GetLogicalDevice(), buffer, bufferMemory, 0) != VK_SUCCESS)
-		m_vkr->UnrecoverableError("Failed to bind buffer memory");
-}
-
-bool VKRMemoryManager::CreateBuffer2(VkDeviceSize size, VkBufferUsageFlags usage, VkMemoryPropertyFlags properties, VkBuffer& buffer, VkDeviceMemory& bufferMemory) const
+bool VKRMemoryManager::CreateBuffer(VkDeviceSize size, VkBufferUsageFlags usage, VkMemoryPropertyFlags properties, VkBuffer& buffer, VkDeviceMemory& bufferMemory) const
 {
 	VkBufferCreateInfo bufferInfo{};
 	bufferInfo.sType = VK_STRUCTURE_TYPE_BUFFER_CREATE_INFO;
@@ -363,7 +467,7 @@ bool VKRMemoryManager::CreateBuffer2(VkDeviceSize size, VkBufferUsageFlags usage
 	bufferInfo.sharingMode = VK_SHARING_MODE_EXCLUSIVE;
 	if (vkCreateBuffer(m_vkr->GetLogicalDevice(), &bufferInfo, nullptr, &buffer) != VK_SUCCESS)
 	{
-		cemuLog_log(LogType::Force, "Failed to create buffer (CreateBuffer2)");
+		cemuLog_log(LogType::Force, "Failed to create buffer (CreateBuffer)");
 		return false;
 	}
 
@@ -373,7 +477,7 @@ bool VKRMemoryManager::CreateBuffer2(VkDeviceSize size, VkBufferUsageFlags usage
 	VkMemoryAllocateInfo allocInfo{};
 	allocInfo.sType = VK_STRUCTURE_TYPE_MEMORY_ALLOCATE_INFO;
 	allocInfo.allocationSize = memRequirements.size;
-	if (!FindMemoryType2(memRequirements.memoryTypeBits, properties, allocInfo.memoryTypeIndex))
+	if (!FindMemoryType(memRequirements.memoryTypeBits, properties, allocInfo.memoryTypeIndex))
 	{
 		vkDestroyBuffer(m_vkr->GetLogicalDevice(), buffer, nullptr);
 		return false;
@@ -386,7 +490,7 @@ bool VKRMemoryManager::CreateBuffer2(VkDeviceSize size, VkBufferUsageFlags usage
 	if (vkBindBufferMemory(m_vkr->GetLogicalDevice(), buffer, bufferMemory, 0) != VK_SUCCESS)
 	{
 		vkDestroyBuffer(m_vkr->GetLogicalDevice(), buffer, nullptr);
-		cemuLog_log(LogType::Force, "Failed to bind buffer (CreateBuffer2)");
+		cemuLog_log(LogType::Force, "Failed to bind buffer (CreateBuffer)");
 		return false;
 	}
 	return true;
@@ -408,7 +512,7 @@ bool VKRMemoryManager::CreateBufferFromHostMemory(void* hostPointer, VkDeviceSiz
 
 	if (vkCreateBuffer(m_vkr->GetLogicalDevice(), &bufferInfo, nullptr, &buffer) != VK_SUCCESS)
 	{
-		cemuLog_log(LogType::Force, "Failed to create buffer (CreateBuffer2)");
+		cemuLog_log(LogType::Force, "Failed to create buffer (CreateBuffer)");
 		return false;
 	}
 
@@ -429,7 +533,7 @@ bool VKRMemoryManager::CreateBufferFromHostMemory(void* hostPointer, VkDeviceSiz
 
 	allocInfo.pNext = &importHostMem;
 
-	if (!FindMemoryType2(memRequirements.memoryTypeBits, properties, allocInfo.memoryTypeIndex))
+	if (!FindMemoryType(memRequirements.memoryTypeBits, properties, allocInfo.memoryTypeIndex))
 	{
 		vkDestroyBuffer(m_vkr->GetLogicalDevice(), buffer, nullptr);
 		return false;
@@ -469,11 +573,11 @@ VkImageMemAllocation* VKRMemoryManager::imageMemoryAllocate(VkImage image)
 	auto it = map_textureHeap.find(typeFilter);
 	if (it == map_textureHeap.end())
 	{
-		texHeap = new VkTextureChunkedHeap(this, typeFilter, m_vkr->GetLogicalDevice());
+		texHeap = new VkTextureChunkedHeap(this, typeFilter);
 		map_textureHeap.emplace(typeFilter, texHeap);
 	}
 	else
-		texHeap = it->second;
+		texHeap = it->second.get();
 
 	// alloc mem from heap
 	uint32 allocationSize = (uint32)memRequirements.size;

src/Cafe/HW/Latte/Renderer/Vulkan/VKRMemoryManager.h

@@ -2,6 +2,36 @@
 #include "Cafe/HW/Latte/Renderer/Renderer.h"
 #include "Cafe/HW/Latte/Renderer/Vulkan/VulkanAPI.h"
 #include "util/ChunkedHeap/ChunkedHeap.h"
+#include "util/helpers/MemoryPool.h"
+
+enum class VKR_BUFFER_TYPE
+{
+	STAGING, // staging upload buffer
+	INDEX, // buffer for index data
+	STRIDE, // buffer for stride-adjusted vertex data
+};
+
+class VKRBuffer
+{
+  public:
+	static VKRBuffer* Create(VKR_BUFFER_TYPE bufferType, size_t bufferSize, VkMemoryPropertyFlags properties);
+	~VKRBuffer();
+
+	VkBuffer GetVkBuffer() const { return m_buffer; }
+	VkDeviceMemory GetVkBufferMemory() const { return m_bufferMemory; }
+
+	uint8* GetPtr() const { return m_mappedMemory; }
+
+	bool RequiresFlush() const { return m_requiresFlush; }
+
+  private:
+	VKRBuffer(VkBuffer buffer, VkDeviceMemory bufferMem) : m_buffer(buffer), m_bufferMemory(bufferMem) { };
+
+	VkBuffer m_buffer;
+	VkDeviceMemory m_bufferMemory;
+	uint8* m_mappedMemory;
+	bool m_requiresFlush{false};
+};
 
 struct VkImageMemAllocation
 {
@@ -14,18 +44,17 @@ struct VkImageMemAllocation
 	uint32 getAllocationSize() { return allocationSize; }
 };
 
-class VkTextureChunkedHeap : private ChunkedHeap
+class VkTextureChunkedHeap : private ChunkedHeap<>
 {
 public:
-	VkTextureChunkedHeap(class VKRMemoryManager* memoryManager, uint32 typeFilter, VkDevice device) : m_vkrMemoryManager(memoryManager), m_typeFilter(typeFilter), m_device(device) { };
+	VkTextureChunkedHeap(class VKRMemoryManager* memoryManager, uint32 typeFilter) : m_vkrMemoryManager(memoryManager), m_typeFilter(typeFilter) { };
+	~VkTextureChunkedHeap();
 
 	struct ChunkInfo
 	{
 		VkDeviceMemory mem;
 	};
 
-	uint32 allocateNewChunk(uint32 chunkIndex, uint32 minimumAllocationSize) override;
-
 	CHAddr allocMem(uint32 size, uint32 alignment)
 	{
 		if (alignment < 4)
@@ -43,11 +72,6 @@ public:
 		this->free(addr);
 	}
 
-	void setDevice(VkDevice dev)
-	{
-		m_device = dev;
-	}
-
 	VkDeviceMemory getChunkMem(uint32 index)
 	{
 		if (index >= m_list_chunkInfo.size())
@@ -57,29 +81,75 @@ public:
 
 	void getStatistics(uint32& totalHeapSize, uint32& allocatedBytes) const
 	{
-		totalHeapSize = numHeapBytes;
-		allocatedBytes = numAllocatedBytes;
+		totalHeapSize = m_numHeapBytes;
+		allocatedBytes = m_numAllocatedBytes;
 	}
 
-	VkDevice m_device;
+  private:
+	uint32 allocateNewChunk(uint32 chunkIndex, uint32 minimumAllocationSize) override;
+
 	uint32 m_typeFilter{ 0xFFFFFFFF };
 	class VKRMemoryManager* m_vkrMemoryManager;
 	std::vector<ChunkInfo> m_list_chunkInfo;
 };
 
+class VkBufferChunkedHeap : private ChunkedHeap<>
+{
+  public:
+	VkBufferChunkedHeap(VKR_BUFFER_TYPE bufferType, size_t minimumBufferAllocationSize) : m_bufferType(bufferType), m_minimumBufferAllocationSize(minimumBufferAllocationSize) { };
+	~VkBufferChunkedHeap();
+
+	using ChunkedHeap::alloc;
+	using ChunkedHeap::free;
+
+	uint8* GetChunkPtr(uint32 index) const
+	{
+		if (index >= m_chunkBuffers.size())
+			return nullptr;
+		return m_chunkBuffers[index]->GetPtr();
+	}
+
+	void GetChunkVkMemInfo(uint32 index, VkBuffer& buffer, VkDeviceMemory& mem)
+	{
+		if (index >= m_chunkBuffers.size())
+		{
+			buffer = VK_NULL_HANDLE;
+			mem = VK_NULL_HANDLE;
+			return;
+		}
+		buffer = m_chunkBuffers[index]->GetVkBuffer();
+		mem = m_chunkBuffers[index]->GetVkBufferMemory();
+	}
+
+	void GetStats(uint32& numBuffers, size_t& totalBufferSize, size_t& freeBufferSize) const
+	{
+		numBuffers = m_chunkBuffers.size();
+		totalBufferSize = m_numHeapBytes;
+		freeBufferSize = m_numHeapBytes - m_numAllocatedBytes;
+	}
+
+	bool RequiresFlush(uint32 index) const
+	{
+		if (index >= m_chunkBuffers.size())
+			return false;
+		return m_chunkBuffers[index]->RequiresFlush();
+	}
+
+  private:
+	uint32 allocateNewChunk(uint32 chunkIndex, uint32 minimumAllocationSize) override;
+
+	VKR_BUFFER_TYPE m_bufferType;
+	std::vector<VKRBuffer*> m_chunkBuffers;
+	size_t m_minimumBufferAllocationSize;
+};
+
 // a circular ring-buffer which tracks and releases memory per command-buffer
 class VKRSynchronizedRingAllocator
 {
 public:
-	enum class BUFFER_TYPE
-	{
-		STAGING, // staging upload buffer
-		INDEX, // buffer for index data
-		STRIDE, // buffer for stride-adjusted vertex data
-	};
-
-	VKRSynchronizedRingAllocator(class VulkanRenderer* vkRenderer, class VKRMemoryManager* vkMemoryManager, BUFFER_TYPE bufferType, uint32 minimumBufferAllocSize) : m_vkr(vkRenderer), m_vkrMemMgr(vkMemoryManager), m_bufferType(bufferType), m_minimumBufferAllocSize(minimumBufferAllocSize) {};
+	VKRSynchronizedRingAllocator(class VulkanRenderer* vkRenderer, class VKRMemoryManager* vkMemoryManager, VKR_BUFFER_TYPE bufferType, uint32 minimumBufferAllocSize) : m_vkr(vkRenderer), m_vkrMemMgr(vkMemoryManager), m_bufferType(bufferType), m_minimumBufferAllocSize(minimumBufferAllocSize) {};
 	VKRSynchronizedRingAllocator(const VKRSynchronizedRingAllocator&) = delete; // disallow copy
+	~VKRSynchronizedRingAllocator();
 
 	struct BufferSyncPoint_t
 	{
@@ -126,13 +196,53 @@ private:
 
 	const class VulkanRenderer* m_vkr;
 	const class VKRMemoryManager* m_vkrMemMgr;
-	const BUFFER_TYPE m_bufferType;
+	const VKR_BUFFER_TYPE m_bufferType;
 	const uint32 m_minimumBufferAllocSize;
 
 	std::vector<AllocatorBuffer_t> m_buffers;
 
 };
 
+// heap style allocator with released memory being freed after the current command buffer finishes
+class VKRSynchronizedHeapAllocator
+{
+	struct TrackedAllocation
+	{
+		TrackedAllocation(CHAddr allocation) : allocation(allocation) {};
+		CHAddr allocation;
+	};
+
+  public:
+	VKRSynchronizedHeapAllocator(class VKRMemoryManager* vkMemoryManager, VKR_BUFFER_TYPE bufferType, size_t minimumBufferAllocSize);
+	VKRSynchronizedHeapAllocator(const VKRSynchronizedHeapAllocator&) = delete; // disallow copy
+
+	struct AllocatorReservation
+	{
+		VkBuffer vkBuffer;
+		VkDeviceMemory vkMem;
+		uint8* memPtr;
+		uint32 bufferOffset;
+		uint32 size;
+		uint32 bufferIndex;
+	};
+
+	AllocatorReservation* AllocateBufferMemory(uint32 size, uint32 alignment);
+	void FreeReservation(AllocatorReservation* uploadReservation);
+	void FlushReservation(AllocatorReservation* uploadReservation);
+
+	void CleanupBuffer(uint64 latestFinishedCommandBufferId);
+
+	void GetStats(uint32& numBuffers, size_t& totalBufferSize, size_t& freeBufferSize) const;
+  private:
+	const class VKRMemoryManager* m_vkrMemMgr;
+	VkBufferChunkedHeap m_chunkedHeap;
+	// allocations
+	std::vector<TrackedAllocation> m_activeAllocations;
+	MemoryPool<AllocatorReservation> m_poolAllocatorReservation{32};
+	// release queue
+	std::unordered_map<uint64, std::vector<CHAddr>> m_releaseQueue;
+};
+
 void LatteIndices_invalidateAll();
 
 class VKRMemoryManager
@@ -140,15 +250,15 @@ class VKRMemoryManager
 	friend class VKRSynchronizedRingAllocator;
 public:
 	VKRMemoryManager(class VulkanRenderer* renderer) :
-			m_stagingBuffer(renderer, this, VKRSynchronizedRingAllocator::BUFFER_TYPE::STAGING, 32u * 1024 * 1024),
-			m_indexBuffer(renderer, this, VKRSynchronizedRingAllocator::BUFFER_TYPE::INDEX, 4u * 1024 * 1024),
-			m_vertexStrideMetalBuffer(renderer, this, VKRSynchronizedRingAllocator::BUFFER_TYPE::STRIDE, 4u * 1024 * 1024)
+			m_stagingBuffer(renderer, this, VKR_BUFFER_TYPE::STAGING, 32u * 1024 * 1024),
+			m_indexBuffer(this, VKR_BUFFER_TYPE::INDEX, 4u * 1024 * 1024),
+			m_vertexStrideMetalBuffer(renderer, this, VKR_BUFFER_TYPE::STRIDE, 4u * 1024 * 1024)
 	{
 		m_vkr = renderer;
 	}
 
 	// texture memory management
-	std::unordered_map<uint32, VkTextureChunkedHeap*> map_textureHeap; // one heap per memory type
+	std::unordered_map<uint32, std::unique_ptr<VkTextureChunkedHeap>> map_textureHeap; // one heap per memory type
 	std::vector<uint8> m_textureUploadBuffer;
 
 	// texture upload buffer
@@ -167,7 +277,7 @@ public:
 	}
 
 	VKRSynchronizedRingAllocator& getStagingAllocator() { return m_stagingBuffer; }; // allocator for texture/attribute/uniform uploads
-	VKRSynchronizedRingAllocator& getIndexAllocator() { return m_indexBuffer; }; // allocator for index data
+	VKRSynchronizedHeapAllocator& GetIndexAllocator() { return m_indexBuffer; }; // allocator for index data
 	VKRSynchronizedRingAllocator& getMetalStrideWorkaroundAllocator() { return m_vertexStrideMetalBuffer; }; // allocator for stride-adjusted vertex data
 
 	void cleanupBuffers(uint64 latestFinishedCommandBufferId)
@@ -178,9 +288,7 @@ public:
 		m_vertexStrideMetalBuffer.CleanupBuffer(latestFinishedCommandBufferId);
 	}
 
-	// memory helpers
-	uint32_t FindMemoryType(uint32_t typeFilter, VkMemoryPropertyFlags properties) const;
-	bool FindMemoryType2(uint32 typeFilter, VkMemoryPropertyFlags properties, uint32& memoryIndex) const; // searches for exact properties. Can gracefully fail without throwing exception (returns false)
+	bool FindMemoryType(uint32 typeFilter, VkMemoryPropertyFlags properties, uint32& memoryIndex) const; // searches for exact properties. Can gracefully fail without throwing exception (returns false)
 	std::vector<uint32> FindMemoryTypes(uint32_t typeFilter, VkMemoryPropertyFlags properties) const;
 
 	// image memory allocation
@@ -190,8 +298,7 @@ public:
 	// buffer management
 	size_t GetTotalMemoryForBufferType(VkBufferUsageFlags usage, VkMemoryPropertyFlags properties, size_t minimumBufferSize = 16 * 1024 * 1024);
 
-	void CreateBuffer(VkDeviceSize size, VkBufferUsageFlags usage, VkMemoryPropertyFlags properties, VkBuffer& buffer, VkDeviceMemory& bufferMemory) const;
-	bool CreateBuffer2(VkDeviceSize size, VkBufferUsageFlags usage, VkMemoryPropertyFlags properties, VkBuffer& buffer, VkDeviceMemory& bufferMemory) const; // same as CreateBuffer but doesn't throw exception on failure
+	bool CreateBuffer(VkDeviceSize size, VkBufferUsageFlags usage, VkMemoryPropertyFlags properties, VkBuffer& buffer, VkDeviceMemory& bufferMemory) const; // same as CreateBuffer but doesn't throw exception on failure
 	bool CreateBufferFromHostMemory(void* hostPointer, VkDeviceSize size, VkBufferUsageFlags usage, VkMemoryPropertyFlags properties, VkBuffer& buffer, VkDeviceMemory& bufferMemory) const;
 
 	void DeleteBuffer(VkBuffer& buffer, VkDeviceMemory& deviceMem) const;
@@ -202,6 +309,6 @@ public:
 	private:
 		class VulkanRenderer* m_vkr;
 		VKRSynchronizedRingAllocator m_stagingBuffer;
-		VKRSynchronizedRingAllocator m_indexBuffer;
+		VKRSynchronizedHeapAllocator m_indexBuffer;
 		VKRSynchronizedRingAllocator m_vertexStrideMetalBuffer;
 };

src/Cafe/HW/Latte/Renderer/Vulkan/VulkanAPI.h

@@ -165,6 +165,7 @@ VKFUNC_DEVICE(vkCmdDraw);
 VKFUNC_DEVICE(vkCmdCopyBufferToImage);
 VKFUNC_DEVICE(vkCmdCopyImageToBuffer);
 VKFUNC_DEVICE(vkCmdClearColorImage);
+VKFUNC_DEVICE(vkCmdClearAttachments);
 VKFUNC_DEVICE(vkCmdBindIndexBuffer);
 VKFUNC_DEVICE(vkCmdBindVertexBuffers);
 VKFUNC_DEVICE(vkCmdDrawIndexed);
@@ -198,6 +199,7 @@ VKFUNC_DEVICE(vkCmdEndTransformFeedbackEXT);
 
 // query
 VKFUNC_DEVICE(vkCreateQueryPool);
+VKFUNC_DEVICE(vkDestroyQueryPool);
 VKFUNC_DEVICE(vkCmdResetQueryPool);
 VKFUNC_DEVICE(vkCmdBeginQuery);
 VKFUNC_DEVICE(vkCmdEndQuery);
@@ -236,6 +238,7 @@ VKFUNC_DEVICE(vkAllocateDescriptorSets);
 VKFUNC_DEVICE(vkFreeDescriptorSets);
 VKFUNC_DEVICE(vkUpdateDescriptorSets);
 VKFUNC_DEVICE(vkCreateDescriptorPool);
+VKFUNC_DEVICE(vkDestroyDescriptorPool);
 VKFUNC_DEVICE(vkDestroyDescriptorSetLayout);
 
 #undef VKFUNC_INIT

src/Cafe/HW/Latte/Renderer/Vulkan/VulkanRenderer.cpp

@@ -439,7 +439,7 @@ VulkanRenderer::VulkanRenderer()
 	GetDeviceFeatures();
 
 	// init memory manager
-	memoryManager = new VKRMemoryManager(this);
+	memoryManager.reset(new VKRMemoryManager(this));
 
 	try
 	{
@@ -577,15 +577,15 @@ VulkanRenderer::VulkanRenderer()
 	void* bufferPtr;
 	// init ringbuffer for uniform vars
 	m_uniformVarBufferMemoryIsCoherent = false;
-	if (memoryManager->CreateBuffer2(UNIFORMVAR_RINGBUFFER_SIZE, VK_BUFFER_USAGE_UNIFORM_BUFFER_BIT, VK_MEMORY_PROPERTY_HOST_VISIBLE_BIT | VK_MEMORY_PROPERTY_HOST_COHERENT_BIT | VK_MEMORY_PROPERTY_HOST_CACHED_BIT, m_uniformVarBuffer, m_uniformVarBufferMemory))
+	if (memoryManager->CreateBuffer(UNIFORMVAR_RINGBUFFER_SIZE, VK_BUFFER_USAGE_UNIFORM_BUFFER_BIT, VK_MEMORY_PROPERTY_HOST_VISIBLE_BIT | VK_MEMORY_PROPERTY_HOST_COHERENT_BIT | VK_MEMORY_PROPERTY_HOST_CACHED_BIT, m_uniformVarBuffer, m_uniformVarBufferMemory))
 		m_uniformVarBufferMemoryIsCoherent = true;
-	else if (memoryManager->CreateBuffer2(UNIFORMVAR_RINGBUFFER_SIZE, VK_BUFFER_USAGE_UNIFORM_BUFFER_BIT, VK_MEMORY_PROPERTY_HOST_VISIBLE_BIT | VK_MEMORY_PROPERTY_HOST_COHERENT_BIT | VK_MEMORY_PROPERTY_HOST_CACHED_BIT | VK_MEMORY_PROPERTY_DEVICE_LOCAL_BIT, m_uniformVarBuffer, m_uniformVarBufferMemory))
+	else if (memoryManager->CreateBuffer(UNIFORMVAR_RINGBUFFER_SIZE, VK_BUFFER_USAGE_UNIFORM_BUFFER_BIT, VK_MEMORY_PROPERTY_HOST_VISIBLE_BIT | VK_MEMORY_PROPERTY_HOST_COHERENT_BIT | VK_MEMORY_PROPERTY_HOST_CACHED_BIT | VK_MEMORY_PROPERTY_DEVICE_LOCAL_BIT, m_uniformVarBuffer, m_uniformVarBufferMemory))
 		m_uniformVarBufferMemoryIsCoherent = true; // unified memory
-	else if (memoryManager->CreateBuffer2(UNIFORMVAR_RINGBUFFER_SIZE, VK_BUFFER_USAGE_UNIFORM_BUFFER_BIT, VK_MEMORY_PROPERTY_HOST_VISIBLE_BIT | VK_MEMORY_PROPERTY_HOST_COHERENT_BIT, m_uniformVarBuffer, m_uniformVarBufferMemory))
+	else if (memoryManager->CreateBuffer(UNIFORMVAR_RINGBUFFER_SIZE, VK_BUFFER_USAGE_UNIFORM_BUFFER_BIT, VK_MEMORY_PROPERTY_HOST_VISIBLE_BIT | VK_MEMORY_PROPERTY_HOST_COHERENT_BIT, m_uniformVarBuffer, m_uniformVarBufferMemory))
 		m_uniformVarBufferMemoryIsCoherent = true;
 	else
 	{
-		memoryManager->CreateBuffer2(UNIFORMVAR_RINGBUFFER_SIZE, VK_BUFFER_USAGE_UNIFORM_BUFFER_BIT, VK_MEMORY_PROPERTY_HOST_VISIBLE_BIT, m_uniformVarBuffer, m_uniformVarBufferMemory);
+		memoryManager->CreateBuffer(UNIFORMVAR_RINGBUFFER_SIZE, VK_BUFFER_USAGE_UNIFORM_BUFFER_BIT, VK_MEMORY_PROPERTY_HOST_VISIBLE_BIT, m_uniformVarBuffer, m_uniformVarBufferMemory);
 	}
 
 	if (!m_uniformVarBufferMemoryIsCoherent)
@@ -628,6 +628,31 @@ VulkanRenderer::~VulkanRenderer()
 	m_pipeline_cache_semaphore.notify();
 	m_pipeline_cache_save_thread.join();
 
+	vkDestroyPipelineCache(m_logicalDevice, m_pipeline_cache, nullptr);
+
+	if(!m_backbufferBlitDescriptorSetCache.empty())
+	{
+		std::vector<VkDescriptorSet> freeVector;
+		freeVector.reserve(m_backbufferBlitDescriptorSetCache.size());
+		std::transform(m_backbufferBlitDescriptorSetCache.begin(), m_backbufferBlitDescriptorSetCache.end(), std::back_inserter(freeVector), [](auto& i) {
+		  return i.second;
+		});
+		vkFreeDescriptorSets(m_logicalDevice, m_descriptorPool, freeVector.size(), freeVector.data());
+	}
+
+	vkDestroyDescriptorPool(m_logicalDevice, m_descriptorPool, nullptr);
+
+	for(auto& i : m_backbufferBlitPipelineCache)
+	{
+		vkDestroyPipeline(m_logicalDevice, i.second, nullptr);
+	}
+	m_backbufferBlitPipelineCache = {};
+
+	if(m_occlusionQueries.queryPool != VK_NULL_HANDLE)
+		vkDestroyQueryPool(m_logicalDevice, m_occlusionQueries.queryPool, nullptr);
+
+	vkDestroyDescriptorSetLayout(m_logicalDevice, m_swapchainDescriptorSetLayout, nullptr);
+
 	// shut down imgui
 	ImGui_ImplVulkan_Shutdown();
 
@@ -640,10 +665,6 @@ VulkanRenderer::~VulkanRenderer()
 	memoryManager->DeleteBuffer(m_xfbRingBuffer, m_xfbRingBufferMemory);
 	memoryManager->DeleteBuffer(m_occlusionQueries.bufferQueryResults, m_occlusionQueries.memoryQueryResults);
 	memoryManager->DeleteBuffer(m_bufferCache, m_bufferCacheMemory);
-	// texture memory
-	// todo
-	// upload buffers
-	// todo
 
 	m_padSwapchainInfo = nullptr;
 	m_mainSwapchainInfo = nullptr;
@@ -666,12 +687,20 @@ VulkanRenderer::~VulkanRenderer()
 		it = VK_NULL_HANDLE;
 	}
 
+	for(auto& sem : m_commandBufferSemaphores)
+	{
+		vkDestroySemaphore(m_logicalDevice, sem, nullptr);
+		sem = VK_NULL_HANDLE;
+	}
+
 	if (m_pipelineLayout != VK_NULL_HANDLE)
 		vkDestroyPipelineLayout(m_logicalDevice, m_pipelineLayout, nullptr);
 
 	if (m_commandPool != VK_NULL_HANDLE)
 		vkDestroyCommandPool(m_logicalDevice, m_commandPool, nullptr);
 
+	VKRObjectSampler::DestroyCache();
+
 	// destroy debug callback
 	if (m_debugCallback)
 	{
@@ -679,6 +708,12 @@ VulkanRenderer::~VulkanRenderer()
 		vkDestroyDebugUtilsMessengerEXT(m_instance, m_debugCallback, nullptr);
 	}
 
+	while(!m_destructionQueue.empty())
+		ProcessDestructionQueue();
+
+	// destroy memory manager
+	memoryManager.reset();
+
 	// destroy instance, devices
 	if (m_instance != VK_NULL_HANDLE)
 	{
@@ -690,9 +725,6 @@ VulkanRenderer::~VulkanRenderer()
 		vkDestroyInstance(m_instance, nullptr);
 	}
 
-	// destroy memory manager
-	delete memoryManager;
-
 	// crashes?
 	//glslang::FinalizeProcess();
 }
@@ -823,7 +855,14 @@ void VulkanRenderer::HandleScreenshotRequest(LatteTextureView* texView, bool pad
 			VkMemoryAllocateInfo allocInfo{};
 			allocInfo.sType = VK_STRUCTURE_TYPE_MEMORY_ALLOCATE_INFO;
 			allocInfo.allocationSize = memRequirements.size;
-			allocInfo.memoryTypeIndex = memoryManager->FindMemoryType(memRequirements.memoryTypeBits, VK_MEMORY_PROPERTY_DEVICE_LOCAL_BIT);
+			uint32 memIndex;
+			bool foundMemory = memoryManager->FindMemoryType(memRequirements.memoryTypeBits, VK_MEMORY_PROPERTY_DEVICE_LOCAL_BIT, memIndex);
+			if(!foundMemory)
+			{
+				cemuLog_log(LogType::Force, "Screenshot request failed due to incompatible vulkan memory types.");
+				return;
+			}
+			allocInfo.memoryTypeIndex = memIndex;
 
 			if (vkAllocateMemory(m_logicalDevice, &allocInfo, nullptr, &imageMemory) != VK_SUCCESS)
 			{
@@ -1606,6 +1645,7 @@ void VulkanRenderer::Initialize()
 
 void VulkanRenderer::Shutdown()
 {
+	DeleteFontTextures();
 	Renderer::Shutdown();
 	SubmitCommandBuffer();
 	WaitDeviceIdle();
@@ -1806,7 +1846,6 @@ void VulkanRenderer::ImguiEnd()
 	vkCmdEndRenderPass(m_state.currentCommandBuffer);
 }
 
-std::vector<LatteTextureVk*> g_imgui_textures; // TODO manage better
 ImTextureID VulkanRenderer::GenerateTexture(const std::vector<uint8>& data, const Vector2i& size)
 {
 	try
@@ -1836,6 +1875,7 @@ void VulkanRenderer::DeleteTexture(ImTextureID id)
 
 void VulkanRenderer::DeleteFontTextures()
 {
+	WaitDeviceIdle();
 	ImGui_ImplVulkan_DestroyFontsTexture();
 }
 
@@ -1874,7 +1914,7 @@ void VulkanRenderer::InitFirstCommandBuffer()
 	vkResetFences(m_logicalDevice, 1, &m_cmd_buffer_fences[m_commandBufferIndex]);
 	VkCommandBufferBeginInfo beginInfo{};
 	beginInfo.sType = VK_STRUCTURE_TYPE_COMMAND_BUFFER_BEGIN_INFO;
-	beginInfo.flags = VK_COMMAND_BUFFER_USAGE_SIMULTANEOUS_USE_BIT;
+	beginInfo.flags = VK_COMMAND_BUFFER_USAGE_ONE_TIME_SUBMIT_BIT;
 	vkBeginCommandBuffer(m_state.currentCommandBuffer, &beginInfo);
 
 	vkCmdSetViewport(m_state.currentCommandBuffer, 0, 1, &m_state.currentViewport);
@@ -1996,7 +2036,7 @@ void VulkanRenderer::SubmitCommandBuffer(VkSemaphore signalSemaphore, VkSemaphor
 
 	VkCommandBufferBeginInfo beginInfo{};
 	beginInfo.sType = VK_STRUCTURE_TYPE_COMMAND_BUFFER_BEGIN_INFO;
-	beginInfo.flags = VK_COMMAND_BUFFER_USAGE_SIMULTANEOUS_USE_BIT;
+	beginInfo.flags = VK_COMMAND_BUFFER_USAGE_ONE_TIME_SUBMIT_BIT;
 	vkBeginCommandBuffer(m_state.currentCommandBuffer, &beginInfo);
 
 	// make sure some states are set for this command buffer
@@ -2517,9 +2557,8 @@ VkPipeline VulkanRenderer::backbufferBlit_createGraphicsPipeline(VkDescriptorSet
 	hash += (uint64)(chainInfo.m_usesSRGB);
 	hash += ((uint64)padView) << 1;
 
-	static std::unordered_map<uint64, VkPipeline> s_pipeline_cache;
-	const auto it = s_pipeline_cache.find(hash);
-	if (it != s_pipeline_cache.cend())
+	const auto it = m_backbufferBlitPipelineCache.find(hash);
+	if (it != m_backbufferBlitPipelineCache.cend())
 		return it->second;
 
 	std::vector<VkPipelineShaderStageCreateInfo> shaderStages;
@@ -2581,10 +2620,18 @@ VkPipeline VulkanRenderer::backbufferBlit_createGraphicsPipeline(VkDescriptorSet
 	colorBlending.blendConstants[2] = 0.0f;
 	colorBlending.blendConstants[3] = 0.0f;
 
+	VkPushConstantRange pushConstantRange{
+		.stageFlags = VK_SHADER_STAGE_FRAGMENT_BIT,
+		.offset = 0,
+		.size = 3 * sizeof(float) * 2 // 3 vec2's
+	};
+
 	VkPipelineLayoutCreateInfo pipelineLayoutInfo{};
 	pipelineLayoutInfo.sType = VK_STRUCTURE_TYPE_PIPELINE_LAYOUT_CREATE_INFO;
 	pipelineLayoutInfo.setLayoutCount = 1;
 	pipelineLayoutInfo.pSetLayouts = &descriptorLayout;
+	pipelineLayoutInfo.pushConstantRangeCount = 1;
+	pipelineLayoutInfo.pPushConstantRanges = &pushConstantRange;
 
 	VkResult result = vkCreatePipelineLayout(m_logicalDevice, &pipelineLayoutInfo, nullptr, &m_pipelineLayout);
 	if (result != VK_SUCCESS)
@@ -2615,7 +2662,7 @@ VkPipeline VulkanRenderer::backbufferBlit_createGraphicsPipeline(VkDescriptorSet
 		throw std::runtime_error(fmt::format("Failed to create graphics pipeline: {}", result));
 	}
 
-	s_pipeline_cache[hash] = pipeline;
+	m_backbufferBlitPipelineCache[hash] = pipeline;
 	m_pipeline_cache_semaphore.notify();
 
 	return pipeline;
@@ -2912,9 +2959,6 @@ void VulkanRenderer::DrawBackbufferQuad(LatteTextureView* texView, RendererOutpu
 	LatteTextureViewVk* texViewVk = (LatteTextureViewVk*)texView;
 	draw_endRenderPass();
 
-	if (clearBackground)
-		ClearColorbuffer(padView);
-
 	// barrier for input texture
 	VkMemoryBarrier memoryBarrier{};
 	memoryBarrier.sType = VK_STRUCTURE_TYPE_MEMORY_BARRIER;
@@ -2951,11 +2995,40 @@ void VulkanRenderer::DrawBackbufferQuad(LatteTextureView* texView, RendererOutpu
 
 	vkCmdBeginRenderPass(m_state.currentCommandBuffer, &renderPassInfo, VK_SUBPASS_CONTENTS_INLINE);
 
+	if (clearBackground)
+	{
+		VkClearAttachment clearAttachment{};
+		clearAttachment.clearValue = {0,0,0,0};
+		clearAttachment.colorAttachment = 0;
+		clearAttachment.aspectMask = VK_IMAGE_ASPECT_COLOR_BIT;
+		VkClearRect clearExtent = {{{0,0},chainInfo.m_actualExtent}, 0, 1};
+		vkCmdClearAttachments(m_state.currentCommandBuffer, 1, &clearAttachment, 1, &clearExtent);
+	}
+
 	vkCmdBindPipeline(m_state.currentCommandBuffer, VK_PIPELINE_BIND_POINT_GRAPHICS, pipeline);
 	m_state.currentPipeline = pipeline;
 
 	vkCmdBindDescriptorSets(m_state.currentCommandBuffer, VK_PIPELINE_BIND_POINT_GRAPHICS, m_pipelineLayout, 0, 1, &descriptSet, 0, nullptr);
 
+	// update push constants
+	Vector2f pushData[3];
+
+	// textureSrcResolution
+	sint32 effectiveWidth, effectiveHeight;
+	texView->baseTexture->GetEffectiveSize(effectiveWidth, effectiveHeight, 0);
+	pushData[0] = {(float)effectiveWidth, (float)effectiveHeight};
+
+	// nativeResolution
+	pushData[1] = {
+		(float)texViewVk->baseTexture->width,
+		(float)texViewVk->baseTexture->height,
+	};
+
+	// outputResolution
+	pushData[2] = {(float)imageWidth,(float)imageHeight};
+
+	vkCmdPushConstants(m_state.currentCommandBuffer, m_pipelineLayout, VK_SHADER_STAGE_FRAGMENT_BIT, 0, sizeof(float) * 2 * 3, &pushData);
+
 	vkCmdDraw(m_state.currentCommandBuffer, 6, 1, 0, 0);
 
 	vkCmdEndRenderPass(m_state.currentCommandBuffer);
@@ -2996,9 +3069,8 @@ VkDescriptorSet VulkanRenderer::backbufferBlit_createDescriptorSet(VkDescriptorS
 	hash += (uint64)texViewVk->GetViewRGBA();
 	hash += (uint64)texViewVk->GetDefaultTextureSampler(useLinearTexFilter);
 
-	static std::unordered_map<uint64, VkDescriptorSet> s_set_cache;
-	const auto it = s_set_cache.find(hash);
-	if (it != s_set_cache.cend())
+	const auto it = m_backbufferBlitDescriptorSetCache.find(hash);
+	if (it != m_backbufferBlitDescriptorSetCache.cend())
 		return it->second;
 
 	VkDescriptorSetAllocateInfo allocInfo = {};
@@ -3029,7 +3101,7 @@ VkDescriptorSet VulkanRenderer::backbufferBlit_createDescriptorSet(VkDescriptorS
 	vkUpdateDescriptorSets(m_logicalDevice, 1, &descriptorWrites, 0, nullptr);
 	performanceMonitor.vk.numDescriptorSamplerTextures.increment();
 
-	s_set_cache[hash] = result;
+	m_backbufferBlitDescriptorSetCache[hash] = result;
 	return result;
 }
 
@@ -3162,7 +3234,8 @@ VkDescriptorSetInfo::~VkDescriptorSetInfo()
 	performanceMonitor.vk.numDescriptorDynUniformBuffers.decrement(statsNumDynUniformBuffers);
 	performanceMonitor.vk.numDescriptorStorageBuffers.decrement(statsNumStorageBuffers);
 
-	VulkanRenderer::GetInstance()->ReleaseDestructibleObject(m_vkObjDescriptorSet);
+	auto renderer = VulkanRenderer::GetInstance();
+	renderer->ReleaseDestructibleObject(m_vkObjDescriptorSet);
 	m_vkObjDescriptorSet = nullptr;
 }
 
@@ -3672,7 +3745,7 @@ void VulkanRenderer::bufferCache_copyStreamoutToMainBuffer(uint32 srcOffset, uin
 
 void VulkanRenderer::AppendOverlayDebugInfo()
 {
-	ImGui::Text("--- Vulkan info ---");
+	ImGui::Text("--- Vulkan debug info ---");
 	ImGui::Text("GfxPipelines   %u", performanceMonitor.vk.numGraphicPipelines.get());
 	ImGui::Text("DescriptorSets %u", performanceMonitor.vk.numDescriptorSets.get());
 	ImGui::Text("DS ImgSamplers %u", performanceMonitor.vk.numDescriptorSamplerTextures.get());
@@ -3680,6 +3753,7 @@ void VulkanRenderer::AppendOverlayDebugInfo()
 	ImGui::Text("DS StorageBuf  %u", performanceMonitor.vk.numDescriptorStorageBuffers.get());
 	ImGui::Text("Images         %u", performanceMonitor.vk.numImages.get());
 	ImGui::Text("ImageView      %u", performanceMonitor.vk.numImageViews.get());
+	ImGui::Text("ImageSampler   %u", performanceMonitor.vk.numSamplers.get());
 	ImGui::Text("RenderPass     %u", performanceMonitor.vk.numRenderPass.get());
 	ImGui::Text("Framebuffer    %u", performanceMonitor.vk.numFramebuffer.get());
 	m_spinlockDestructionQueue.lock();
@@ -3689,7 +3763,7 @@ void VulkanRenderer::AppendOverlayDebugInfo()
 
 	ImGui::Text("BeginRP/f      %u", performanceMonitor.vk.numBeginRenderpassPerFrame.get());
 	ImGui::Text("Barriers/f     %u", performanceMonitor.vk.numDrawBarriersPerFrame.get());
-	ImGui::Text("--- Cache info ---");
+	ImGui::Text("--- Cache debug info ---");
 
 	uint32 bufferCacheHeapSize = 0;
 	uint32 bufferCacheAllocationSize = 0;
@@ -3709,7 +3783,7 @@ void VulkanRenderer::AppendOverlayDebugInfo()
 	ImGui::SameLine(60.0f);
 	ImGui::Text("%06uKB / %06uKB Buffers: %u", ((uint32)(totalSize - freeSize) + 1023) / 1024, ((uint32)totalSize + 1023) / 1024, (uint32)numBuffers);
 
-	memoryManager->getIndexAllocator().GetStats(numBuffers, totalSize, freeSize);
+	memoryManager->GetIndexAllocator().GetStats(numBuffers, totalSize, freeSize);
 	ImGui::Text("Index");
 	ImGui::SameLine(60.0f);
 	ImGui::Text("%06uKB / %06uKB Buffers: %u", ((uint32)(totalSize - freeSize) + 1023) / 1024, ((uint32)totalSize + 1023) / 1024, (uint32)numBuffers);
@@ -3725,7 +3799,7 @@ void VKRDestructibleObject::flagForCurrentCommandBuffer()
 
 bool VKRDestructibleObject::canDestroy()
 {
-	if (refCount > 0)
+	if (m_refCount > 0)
 		return false;
 	return VulkanRenderer::GetInstance()->HasCommandBufferFinished(m_lastCmdBufferId);
 }
@@ -3766,6 +3840,111 @@ VKRObjectTextureView::~VKRObjectTextureView()
 	performanceMonitor.vk.numImageViews.decrement();
 }
 
+static uint64 CalcHashSamplerCreateInfo(const VkSamplerCreateInfo& info)
+{
+	uint64 h = 0xcbf29ce484222325ULL;
+	auto fnvHashCombine = [](uint64_t &h, auto val) {
+		using T = decltype(val);
+		static_assert(sizeof(T) <= 8);
+		uint64_t val64 = 0;
+		std::memcpy(&val64, &val, sizeof(val));
+		h ^= val64;
+		h *= 0x100000001b3ULL;
+	};
+	cemu_assert_debug(info.sType == VK_STRUCTURE_TYPE_SAMPLER_CREATE_INFO);
+	fnvHashCombine(h, info.flags);
+	fnvHashCombine(h, info.magFilter);
+	fnvHashCombine(h, info.minFilter);
+	fnvHashCombine(h, info.mipmapMode);
+	fnvHashCombine(h, info.addressModeU);
+	fnvHashCombine(h, info.addressModeV);
+	fnvHashCombine(h, info.addressModeW);
+	fnvHashCombine(h, info.mipLodBias);
+	fnvHashCombine(h, info.anisotropyEnable);
+	if(info.anisotropyEnable == VK_TRUE)
+		fnvHashCombine(h, info.maxAnisotropy);
+	fnvHashCombine(h, info.compareEnable);
+	if(info.compareEnable == VK_TRUE)
+		fnvHashCombine(h, info.compareOp);
+	fnvHashCombine(h, info.minLod);
+	fnvHashCombine(h, info.maxLod);
+	fnvHashCombine(h, info.borderColor);
+	fnvHashCombine(h, info.unnormalizedCoordinates);
+	// handle custom border color
+	VkBaseOutStructure* ext = (VkBaseOutStructure*)info.pNext;
+	while(ext)
+	{
+		if(ext->sType == VK_STRUCTURE_TYPE_SAMPLER_CUSTOM_BORDER_COLOR_CREATE_INFO_EXT)
+		{
+			auto* extInfo = (VkSamplerCustomBorderColorCreateInfoEXT*)ext;
+			fnvHashCombine(h, extInfo->customBorderColor.uint32[0]);
+			fnvHashCombine(h, extInfo->customBorderColor.uint32[1]);
+			fnvHashCombine(h, extInfo->customBorderColor.uint32[2]);
+			fnvHashCombine(h, extInfo->customBorderColor.uint32[3]);
+		}
+		else
+		{
+			cemu_assert_unimplemented();
+		}
+		ext = ext->pNext;
+	}
+	return h;
+}
+
+std::unordered_map<uint64, VKRObjectSampler*> VKRObjectSampler::s_samplerCache;
+
+VKRObjectSampler::VKRObjectSampler(VkSamplerCreateInfo* samplerInfo)
+{
+	auto* vulkanRenderer = VulkanRenderer::GetInstance();
+	if (vkCreateSampler(vulkanRenderer->GetLogicalDevice(), samplerInfo, nullptr, &m_sampler) != VK_SUCCESS)
+		vulkanRenderer->UnrecoverableError("Failed to create texture sampler");
+	performanceMonitor.vk.numSamplers.increment();
+	m_hash = CalcHashSamplerCreateInfo(*samplerInfo);
+}
+
+VKRObjectSampler::~VKRObjectSampler()
+{
+	vkDestroySampler(VulkanRenderer::GetInstance()->GetLogicalDevice(), m_sampler, nullptr);
+	performanceMonitor.vk.numSamplers.decrement();
+	// remove from cache
+	auto it = s_samplerCache.find(m_hash);
+	if(it != s_samplerCache.end())
+		s_samplerCache.erase(it);
+}
+
+void VKRObjectSampler::RefCountReachedZero()
+{
+	VulkanRenderer::GetInstance()->ReleaseDestructibleObject(this);
+}
+
+VKRObjectSampler* VKRObjectSampler::GetOrCreateSampler(VkSamplerCreateInfo* samplerInfo)
+{
+	auto* vulkanRenderer = VulkanRenderer::GetInstance();
+	uint64 hash = CalcHashSamplerCreateInfo(*samplerInfo);
+	auto it = s_samplerCache.find(hash);
+	if (it != s_samplerCache.end())
+	{
+		auto* sampler = it->second;
+		return sampler;
+	}
+	auto* sampler = new VKRObjectSampler(samplerInfo);
+	s_samplerCache[hash] = sampler;
+	return sampler;
+}
+
+void VKRObjectSampler::DestroyCache()
+{
+	// assuming all other objects which depend on vkSampler are destroyed, this cache should also have been emptied already
+	// but just to be sure lets still clear the cache
+	cemu_assert_debug(s_samplerCache.empty());
+	for(auto& sampler : s_samplerCache)
+	{
+		cemu_assert_debug(sampler.second->m_refCount == 0);
+		delete sampler.second;
+	}
+	s_samplerCache.clear();
+}
+
 VKRObjectRenderPass::VKRObjectRenderPass(AttachmentInfo_t& attachmentInfo, sint32 colorAttachmentCount)
 {
 	// generate helper hash for pipeline state

src/Cafe/HW/Latte/Renderer/Vulkan/VulkanRenderer.h

@@ -137,8 +137,8 @@ class VulkanRenderer : public Renderer
 public:
 
 	// memory management
-	VKRMemoryManager* memoryManager{};
-	VKRMemoryManager* GetMemoryManager() const { return memoryManager; };
+	std::unique_ptr<VKRMemoryManager> memoryManager;
+	VKRMemoryManager* GetMemoryManager() const { return memoryManager.get(); };
 
 	VkSupportedFormatInfo_t m_supportedFormatInfo;
 
@@ -328,8 +328,9 @@ public:
 
 	RendererShader* shader_create(RendererShader::ShaderType type, uint64 baseHash, uint64 auxHash, const std::string& source, bool isGameShader, bool isGfxPackShader) override;
 
-	void* indexData_reserveIndexMemory(uint32 size, uint32& offset, uint32& bufferIndex) override;
-	void indexData_uploadIndexMemory(uint32 offset, uint32 size) override;
+	IndexAllocation indexData_reserveIndexMemory(uint32 size) override;
+	void indexData_releaseIndexMemory(IndexAllocation& allocation) override;
+	void indexData_uploadIndexMemory(IndexAllocation& allocation) override;
 
 	// externally callable
 	void GetTextureFormatInfoVK(Latte::E_GX2SURFFMT format, bool isDepth, Latte::E_DIM dim, sint32 width, sint32 height, FormatInfoVK* formatInfoOut);
@@ -582,6 +583,8 @@ private:
 	std::shared_mutex m_pipeline_cache_save_mutex;
 	std::thread m_pipeline_cache_save_thread;
 	VkPipelineCache m_pipeline_cache{ nullptr };
+	std::unordered_map<uint64, VkPipeline> m_backbufferBlitPipelineCache;
+	std::unordered_map<uint64, VkDescriptorSet> m_backbufferBlitDescriptorSetCache;
 	VkPipelineLayout m_pipelineLayout{nullptr};
 	VkCommandPool m_commandPool{ nullptr };
 	
@@ -859,7 +862,7 @@ private:
 		memBarrier.pNext = nullptr;
 
 		VkPipelineStageFlags srcStages = VK_PIPELINE_STAGE_TRANSFER_BIT;
-		VkPipelineStageFlags dstStages = VK_PIPELINE_STAGE_TOP_OF_PIPE_BIT;
+		VkPipelineStageFlags dstStages = VK_PIPELINE_STAGE_ALL_COMMANDS_BIT;
 
 		memBarrier.srcAccessMask = VK_ACCESS_TRANSFER_READ_BIT | VK_ACCESS_TRANSFER_WRITE_BIT;
 		memBarrier.dstAccessMask = 0;

src/Cafe/HW/Latte/Renderer/Vulkan/VulkanRendererCore.cpp

@@ -357,18 +357,20 @@ PipelineInfo* VulkanRenderer::draw_getOrCreateGraphicsPipeline(uint32 indexCount
 	return draw_createGraphicsPipeline(indexCount);
 }
 
-void* VulkanRenderer::indexData_reserveIndexMemory(uint32 size, uint32& offset, uint32& bufferIndex)
+Renderer::IndexAllocation VulkanRenderer::indexData_reserveIndexMemory(uint32 size)
 {
-	auto& indexAllocator = this->memoryManager->getIndexAllocator();
-	auto resv = indexAllocator.AllocateBufferMemory(size, 32);
-	offset = resv.bufferOffset;
-	bufferIndex = resv.bufferIndex;
-	return resv.memPtr;
+	VKRSynchronizedHeapAllocator::AllocatorReservation* resv = memoryManager->GetIndexAllocator().AllocateBufferMemory(size, 32);
+	return { resv->memPtr, resv };
 }
 
-void VulkanRenderer::indexData_uploadIndexMemory(uint32 offset, uint32 size)
+void VulkanRenderer::indexData_releaseIndexMemory(IndexAllocation& allocation)
 {
-	// does nothing since the index buffer memory is coherent
+	memoryManager->GetIndexAllocator().FreeReservation((VKRSynchronizedHeapAllocator::AllocatorReservation*)allocation.rendererInternal);
+}
+
+void VulkanRenderer::indexData_uploadIndexMemory(IndexAllocation& allocation)
+{
+	memoryManager->GetIndexAllocator().FlushReservation((VKRSynchronizedHeapAllocator::AllocatorReservation*)allocation.rendererInternal);
 }
 
 float s_vkUniformData[512 * 4];
@@ -727,7 +729,6 @@ VkDescriptorSetInfo* VulkanRenderer::draw_getOrCreateDescriptorSet(PipelineInfo*
 
 		VkSamplerCustomBorderColorCreateInfoEXT samplerCustomBorderColor{};
 
-		VkSampler sampler;
 		VkSamplerCreateInfo samplerInfo{};
 		samplerInfo.sType = VK_STRUCTURE_TYPE_SAMPLER_CREATE_INFO;
 
@@ -900,9 +901,9 @@ VkDescriptorSetInfo* VulkanRenderer::draw_getOrCreateDescriptorSet(PipelineInfo*
 			}
 		}
 
-		if (vkCreateSampler(m_logicalDevice, &samplerInfo, nullptr, &sampler) != VK_SUCCESS)
-			UnrecoverableError("Failed to create texture sampler");
-		info.sampler = sampler;
+		VKRObjectSampler* samplerObj = VKRObjectSampler::GetOrCreateSampler(&samplerInfo);
+		vkObjDS->addRef(samplerObj);
+		info.sampler = samplerObj->GetSampler();
 		textureArray.emplace_back(info);
 	}
 
@@ -1163,28 +1164,17 @@ void VulkanRenderer::draw_prepareDescriptorSets(PipelineInfo* pipeline_info, VkD
 	const auto geometryShader = LatteSHRC_GetActiveGeometryShader();
 	const auto pixelShader = LatteSHRC_GetActivePixelShader();
 
-
-	if (vertexShader)
-	{
-		auto descriptorSetInfo = draw_getOrCreateDescriptorSet(pipeline_info, vertexShader);
+	auto prepareShaderDescriptors = [this, &pipeline_info](LatteDecompilerShader* shader) -> VkDescriptorSetInfo* {
+		if (!shader)
+			return nullptr;
+		auto descriptorSetInfo = draw_getOrCreateDescriptorSet(pipeline_info, shader);
 		descriptorSetInfo->m_vkObjDescriptorSet->flagForCurrentCommandBuffer();
-		vertexDS = descriptorSetInfo;
-	}
+		return descriptorSetInfo;
+	};
 
-	if (pixelShader)
-	{
-		auto descriptorSetInfo = draw_getOrCreateDescriptorSet(pipeline_info, pixelShader);
-		descriptorSetInfo->m_vkObjDescriptorSet->flagForCurrentCommandBuffer();
-		pixelDS = descriptorSetInfo;
-
-	}
-
-	if (geometryShader)
-	{
-		auto descriptorSetInfo = draw_getOrCreateDescriptorSet(pipeline_info, geometryShader);
-		descriptorSetInfo->m_vkObjDescriptorSet->flagForCurrentCommandBuffer();
-		geometryDS = descriptorSetInfo;
-	}
+	vertexDS = prepareShaderDescriptors(vertexShader);
+	pixelDS = prepareShaderDescriptors(pixelShader);
+	geometryDS = prepareShaderDescriptors(geometryShader);
 }
 
 void VulkanRenderer::draw_updateVkBlendConstants()
@@ -1415,14 +1405,15 @@ void VulkanRenderer::draw_execute(uint32 baseVertex, uint32 baseInstance, uint32
 	uint32 hostIndexCount;
 	uint32 indexMin = 0;
 	uint32 indexMax = 0;
-	uint32 indexBufferOffset = 0;
-	uint32 indexBufferIndex = 0;
-	LatteIndices_decode(memory_getPointerFromVirtualOffset(indexDataMPTR), indexType, count, primitiveMode, indexMin, indexMax, hostIndexType, hostIndexCount, indexBufferOffset, indexBufferIndex);
-
+	Renderer::IndexAllocation indexAllocation;
+	LatteIndices_decode(memory_getPointerFromVirtualOffset(indexDataMPTR), indexType, count, primitiveMode, indexMin, indexMax, hostIndexType, hostIndexCount, indexAllocation);
+	VKRSynchronizedHeapAllocator::AllocatorReservation* indexReservation = (VKRSynchronizedHeapAllocator::AllocatorReservation*)indexAllocation.rendererInternal;
 	// update index binding
 	bool isPrevIndexData = false;
 	if (hostIndexType != INDEX_TYPE::NONE)
 	{
+		uint32 indexBufferIndex = indexReservation->bufferIndex;
+		uint32 indexBufferOffset = indexReservation->bufferOffset;
 		if (m_state.activeIndexBufferOffset != indexBufferOffset || m_state.activeIndexBufferIndex != indexBufferIndex || m_state.activeIndexType != hostIndexType)
 		{
 			m_state.activeIndexType = hostIndexType;
@@ -1435,7 +1426,7 @@ void VulkanRenderer::draw_execute(uint32 baseVertex, uint32 baseInstance, uint32
 				vkType = VK_INDEX_TYPE_UINT32;
 			else
 				cemu_assert(false);
-			vkCmdBindIndexBuffer(m_state.currentCommandBuffer, memoryManager->getIndexAllocator().GetBufferByIndex(indexBufferIndex), indexBufferOffset, vkType);
+			vkCmdBindIndexBuffer(m_state.currentCommandBuffer, indexReservation->vkBuffer, indexBufferOffset, vkType);
 		}
 		else
 			isPrevIndexData = true;

src/Cafe/HW/Latte/Renderer/Vulkan/VulkanSurfaceCopy.cpp

@@ -76,6 +76,30 @@ struct CopySurfacePipelineInfo
 	CopySurfacePipelineInfo() = default;
 	CopySurfacePipelineInfo(VkDevice device) : m_device(device) {}
 	CopySurfacePipelineInfo(const CopySurfacePipelineInfo& info) = delete;
+	~CopySurfacePipelineInfo()
+	{
+		auto renderer = VulkanRenderer::GetInstance();
+		renderer->ReleaseDestructibleObject(vkObjRenderPass);
+		renderer->ReleaseDestructibleObject(vkObjPipeline);
+
+		for(auto& i : map_framebuffers)
+		{
+			for(auto& fb : i.second.m_array)
+			{
+				renderer->ReleaseDestructibleObject(fb->vkObjFramebuffer);
+				renderer->ReleaseDestructibleObject(fb->vkObjImageView);
+			}
+		}
+
+		for(auto& i : map_descriptors)
+		{
+			for(auto& descriptor : i.second.m_array)
+			{
+				renderer->ReleaseDestructibleObject(descriptor->vkObjImageView);
+				renderer->ReleaseDestructibleObject(descriptor->vkObjDescriptorSet);
+			}
+		}
+	}
 
 	VkDevice m_device = nullptr;
 
@@ -842,5 +866,9 @@ void VulkanRenderer::surfaceCopy_notifyTextureRelease(LatteTextureVk* hostTextur
 
 void VulkanRenderer::surfaceCopy_cleanup()
 {
-	// todo - release m_copySurfacePipelineCache etc
+	for(auto& i : m_copySurfacePipelineCache)
+	{
+		delete i.second;
+	}
+	m_copySurfacePipelineCache = {};
 }

src/Cafe/IOSU/legacy/iosu_acp.cpp

@@ -469,7 +469,7 @@ namespace iosu
 				entry->ukn0C = 0;
 				entry->sizeA = _swapEndianU64(0); // ukn
 				entry->sizeB = _swapEndianU64(dirSize);
-				entry->time = _swapEndianU64((coreinit::coreinit_getOSTime() / ESPRESSO_TIMER_CLOCK));
+				entry->time = _swapEndianU64((coreinit::OSGetTime() / ESPRESSO_TIMER_CLOCK));
 				sprintf(entry->path, "%susr/save/%08x/%08x/meta/", devicePath, (uint32)(titleId >> 32), (uint32)(titleId & 0xFFFFFFFF));
 				count++;
 			}
@@ -504,7 +504,7 @@ namespace iosu
 					entry->ukn0C = 0;
 					entry->sizeA = _swapEndianU64(0);
 					entry->sizeB = _swapEndianU64(0);
-					entry->time = _swapEndianU64((coreinit::coreinit_getOSTime() / ESPRESSO_TIMER_CLOCK));
+					entry->time = _swapEndianU64((coreinit::OSGetTime() / ESPRESSO_TIMER_CLOCK));
 					sprintf(entry->path, "%susr/save/%08x/%08x/meta/", devicePath, (uint32)(titleId >> 32), (uint32)(titleId & 0xFFFFFFFF));
 					count++;
 				}
@@ -584,7 +584,7 @@ namespace iosu
 
 		uint64 _ACPGetTimestamp()
 		{
-			return coreinit::coreinit_getOSTime() / ESPRESSO_TIMER_CLOCK;
+			return coreinit::OSGetTime() / ESPRESSO_TIMER_CLOCK;
 		}
 
 		nnResult ACPUpdateSaveTimeStamp(uint32 persistentId, uint64 titleId, ACPDeviceType deviceType)

src/Cafe/OS/RPL/rpl_structs.h

@@ -116,7 +116,7 @@ typedef struct
 	/* +0x34 */ uint32be ukn34;
 	/* +0x38 */ uint32be ukn38;
 	/* +0x3C */ uint32be ukn3C;
-	/* +0x40 */ uint32be toolkitVersion;
+	/* +0x40 */ uint32be minimumToolkitVersion;
 	/* +0x44 */ uint32be ukn44;
 	/* +0x48 */ uint32be ukn48;
 	/* +0x4C */ uint32be ukn4C;

src/Cafe/OS/libs/camera/camera.cpp

@@ -181,12 +181,12 @@ namespace camera
 	sint32 CAMInit(uint32 cameraId, CAMInitInfo_t* camInitInfo, uint32be* error)
 	{
 		CameraInstance* camInstance = new CameraInstance(camInitInfo->width, camInitInfo->height, camInitInfo->handlerFuncPtr);
-
+		*error = 0; // Hunter's Trophy 2 will fail to boot if we don't set this
 		std::unique_lock<std::recursive_mutex> _lock(g_mutex_camera);
 		if (g_cameraCounter == 0)
 		{
 			coreinit::OSCreateAlarm(g_alarm_camera.GetPtr());
-			coreinit::OSSetPeriodicAlarm(g_alarm_camera.GetPtr(), coreinit::coreinit_getOSTime(), (uint64)ESPRESSO_TIMER_CLOCK / 60ull, RPLLoader_MakePPCCallable(ppcCAMUpdate60));
+			coreinit::OSSetPeriodicAlarm(g_alarm_camera.GetPtr(), coreinit::OSGetTime(), (uint64)ESPRESSO_TIMER_CLOCK / 60ull, RPLLoader_MakePPCCallable(ppcCAMUpdate60));
 		}
 		g_cameraCounter++;
 

src/Cafe/OS/libs/coreinit/coreinit_Alarm.cpp

@@ -166,7 +166,7 @@ namespace coreinit
 	void alarm_update()
 	{	
 		cemu_assert_debug(!__OSHasSchedulerLock());
-		uint64 currentTick = coreinit::coreinit_getOSTime();
+		uint64 currentTick = coreinit::OSGetTime();
 		if (!OSHostAlarm::quickCheckForAlarm(currentTick))
 			return;
 		__OSLockScheduler();
@@ -233,7 +233,7 @@ namespace coreinit
 			if (period == 0)
 				return;
 
-			uint64 currentTime = coreinit_getOSTime();
+			uint64 currentTime = OSGetTime();
 
 			uint64 ticksSinceStart = currentTime - startTime;
 			uint64 numPeriods = ticksSinceStart / period;
@@ -267,7 +267,7 @@ namespace coreinit
 	void OSSetAlarm(OSAlarm_t* alarm, uint64 delayInTicks, MPTR handlerFunc)
 	{
 		__OSLockScheduler();
-		__OSInitiateAlarm(alarm, coreinit_getOSTime() + delayInTicks, 0, handlerFunc, false);
+		__OSInitiateAlarm(alarm, OSGetTime() + delayInTicks, 0, handlerFunc, false);
 		__OSUnlockScheduler();
 	}
 
@@ -310,7 +310,7 @@ namespace coreinit
 		while( true )
 		{
 			OSWaitEvent(g_alarmEvent.GetPtr());
-			uint64 currentTick = coreinit_getOSTime();
+			uint64 currentTick = OSGetTime();
 			while (true)
 			{
 				// get alarm to fire

src/Cafe/OS/libs/coreinit/coreinit_GHS.cpp

@@ -156,12 +156,22 @@ namespace coreinit
 		return &currentThread->crt.eh_mem_manage;
 	}
 
-	void* __gh_errno_ptr()
+	sint32be* __gh_errno_ptr()
 	{
 		OSThread_t* currentThread = coreinit::OSGetCurrentThread();
 		return &currentThread->context.ghs_errno;
 	}
 
+	void __gh_set_errno(sint32 errNo)
+	{
+		*__gh_errno_ptr() = errNo;
+	}
+
+	sint32 __gh_get_errno()
+	{
+		return *__gh_errno_ptr();
+	}
+
 	void* __get_eh_store_globals()
 	{
 		OSThread_t* currentThread = coreinit::OSGetCurrentThread();
@@ -272,6 +282,8 @@ namespace coreinit
 		cafeExportRegister("coreinit", __get_eh_globals, LogType::Placeholder);
 		cafeExportRegister("coreinit", __get_eh_mem_manage, LogType::Placeholder);
 		cafeExportRegister("coreinit", __gh_errno_ptr, LogType::Placeholder);
+		cafeExportRegister("coreinit", __gh_set_errno, LogType::Placeholder);
+		cafeExportRegister("coreinit", __gh_get_errno, LogType::Placeholder);
 		cafeExportRegister("coreinit", __get_eh_store_globals, LogType::Placeholder);
 		cafeExportRegister("coreinit", __get_eh_store_globals_tdeh, LogType::Placeholder);
 

src/Cafe/OS/libs/coreinit/coreinit_GHS.h

@@ -4,5 +4,9 @@ namespace coreinit
 {
 	void PrepareGHSRuntime();
 
+	sint32be* __gh_errno_ptr();
+	void __gh_set_errno(sint32 errNo);
+	sint32 __gh_get_errno();
+
 	void InitializeGHS();
 };

src/Cafe/OS/libs/coreinit/coreinit_Spinlock.cpp

@@ -86,11 +86,11 @@ namespace coreinit
 		else
 		{
 			// loop until lock acquired or timeout occurred
-			uint64 timeoutValue = coreinit_getTimerTick() + coreinit::EspressoTime::ConvertNsToTimerTicks(timeout);
+			uint64 timeoutValue = OSGetSystemTime() + coreinit::EspressoTime::ConvertNsToTimerTicks(timeout);
 			while (!spinlock->ownerThread.atomic_compare_exchange(nullptr, currentThread))
 			{
 				OSYieldThread();
-				if (coreinit_getTimerTick() >= timeoutValue)
+				if (OSGetSystemTime() >= timeoutValue)
 				{
 					return false;
 				}
@@ -182,11 +182,11 @@ namespace coreinit
 		else
 		{
 			// loop until lock acquired or timeout occurred
-			uint64 timeoutValue = coreinit_getTimerTick() + coreinit::EspressoTime::ConvertNsToTimerTicks(timeout);
+			uint64 timeoutValue = OSGetSystemTime() + coreinit::EspressoTime::ConvertNsToTimerTicks(timeout);
 			while (!spinlock->ownerThread.atomic_compare_exchange(nullptr, currentThread))
 			{
 				OSYieldThread();
-				if (coreinit_getTimerTick() >= timeoutValue)
+				if (OSGetSystemTime() >= timeoutValue)
 				{
 					return false;
 				}

src/Cafe/OS/libs/coreinit/coreinit_Synchronization.cpp

@@ -73,8 +73,6 @@ namespace coreinit
 		}
 	}
 
-	uint64 coreinit_getOSTime();
-
 	bool OSWaitEventWithTimeout(OSEvent* event, uint64 timeout)
 	{
 		__OSLockScheduler();
@@ -95,14 +93,14 @@ namespace coreinit
 
 			// workaround for a bad implementation in some Unity games (like Qube Directors Cut, see FEventWiiU::Wait)
 			// where the the return value of OSWaitEventWithTimeout is ignored and instead the game measures the elapsed time to determine if a timeout occurred
-			timeout = timeout * 98ULL / 100ULL; // 98% (we want the function to return slightly before the actual timeout)
+			if (timeout < 0x00FFFFFFFFFFFFFFULL)
+				timeout = timeout * 98ULL / 100ULL; // 98% (we want the function to return slightly before the actual timeout)
 
 			WaitEventWithTimeoutData data;
 			data.thread = OSGetCurrentThread();
 			data.threadQueue = &event->threadQueue;
 			data.hasTimeout = false;
-
-			auto hostAlarm = coreinit::OSHostAlarmCreate(coreinit::coreinit_getOSTime() + coreinit::EspressoTime::ConvertNsToTimerTicks(timeout), 0, _OSWaitEventWithTimeoutHandler, &data);
+			auto hostAlarm = coreinit::OSHostAlarmCreate(OSGetTime() + coreinit::EspressoTime::ConvertNsToTimerTicks(timeout), 0, _OSWaitEventWithTimeoutHandler, &data);
 			event->threadQueue.queueAndWait(OSGetCurrentThread());
 			coreinit::OSHostAlarmDestroy(hostAlarm);
 			if (data.hasTimeout)

src/Cafe/OS/libs/coreinit/coreinit_Thread.cpp

@@ -655,7 +655,7 @@ namespace coreinit
 		StackAllocator<OSThreadQueue> _threadQueue;
 		OSInitThreadQueue(_threadQueue.GetPointer());
 		__OSLockScheduler();
-		OSHostAlarm* hostAlarm = OSHostAlarmCreate(coreinit_getOSTime() + ticks, 0, _OSSleepTicks_alarmHandler, _threadQueue.GetPointer());
+		OSHostAlarm* hostAlarm = OSHostAlarmCreate(OSGetTime() + ticks, 0, _OSSleepTicks_alarmHandler, _threadQueue.GetPointer());
 		_threadQueue.GetPointer()->queueAndWait(OSGetCurrentThread());
 		OSHostAlarmDestroy(hostAlarm);
 		__OSUnlockScheduler();
@@ -1114,13 +1114,13 @@ namespace coreinit
 		thread->requestFlags = (OSThread_t::REQUEST_FLAG_BIT)(thread->requestFlags & OSThread_t::REQUEST_FLAG_CANCEL); // remove all flags except cancel flag
 
 		// update total cycles
-		uint64 remainingCycles = std::min((uint64)hCPU->remainingCycles, (uint64)thread->quantumTicks);
-		uint64 executedCycles = thread->quantumTicks - remainingCycles;
-		if (executedCycles < hCPU->skippedCycles)
+		sint64 executedCycles = (sint64)thread->quantumTicks - (sint64)hCPU->remainingCycles;
+		executedCycles = std::max<sint64>(executedCycles, 0);
+		if (executedCycles < (sint64)hCPU->skippedCycles)
 			executedCycles = 0;
 		else
 			executedCycles -= hCPU->skippedCycles;
-		thread->totalCycles += executedCycles;
+		thread->totalCycles += (uint64)executedCycles;
 		// store context and set current thread to null
 		__OSThreadStoreContext(hCPU, thread);
 		OSSetCurrentThread(OSGetCoreId(), nullptr);

src/Cafe/OS/libs/coreinit/coreinit_Thread.h

@@ -38,7 +38,7 @@ struct OSContext_t
 	/* +0x1E0 */ uint64be fp_ps1[32];
 	/* +0x2E0 */ uint64be coretime[3];
 	/* +0x2F8 */ uint64be starttime;
-	/* +0x300 */ uint32be ghs_errno;         // returned by __gh_errno_ptr() (used by socketlasterr)
+	/* +0x300 */ sint32be ghs_errno;         // returned by __gh_errno_ptr() (used by socketlasterr)
 	/* +0x304 */ uint32be affinity;
 	/* +0x308 */ uint32be upmc1;
 	/* +0x30C */ uint32be upmc2;

src/Cafe/OS/libs/coreinit/coreinit_Time.cpp

@@ -3,38 +3,32 @@
 
 namespace coreinit
 {
-
-	uint64 coreinit_getTimerTick()
+	uint64 coreinit_GetMFTB()
 	{
 		// bus clock is 1/5th of core clock
 		// timer clock is 1/4th of bus clock
 		return PPCInterpreter_getMainCoreCycleCounter() / 20ULL;
 	}
 
-	uint64 coreinit_getOSTime()
+	uint64 OSGetSystemTime()
 	{
-		return coreinit_getTimerTick() + ppcCyclesSince2000TimerClock;
-	}
-
-	void export_OSGetTick(PPCInterpreter_t* hCPU)
-	{
-		uint64 osTime = coreinit_getOSTime();
-		osLib_returnFromFunction(hCPU, (uint32)osTime);
+		return coreinit_GetMFTB();
 	}
 
 	uint64 OSGetTime()
 	{
-		return coreinit_getOSTime();
+		return OSGetSystemTime() + ppcCyclesSince2000TimerClock;
 	}
 
-	void export_OSGetSystemTime(PPCInterpreter_t* hCPU)
+	uint32 OSGetSystemTick()
 	{
-		osLib_returnFromFunction64(hCPU, coreinit_getTimerTick());
+		return static_cast<uint32>(coreinit_GetMFTB());
 	}
 
-	void export_OSGetSystemTick(PPCInterpreter_t* hCPU)
+	uint32 OSGetTick()
 	{
-		osLib_returnFromFunction(hCPU, (uint32)coreinit_getTimerTick());
+		uint64 osTime = OSGetTime();
+		return static_cast<uint32>(osTime);
 	}
 
 	uint32 getLeapDaysUntilYear(uint32 year)
@@ -360,14 +354,13 @@ namespace coreinit
 	void InitializeTimeAndCalendar()
 	{
 		cafeExportRegister("coreinit", OSGetTime, LogType::Placeholder);
-		osLib_addFunction("coreinit", "OSGetSystemTime", export_OSGetSystemTime);
-		osLib_addFunction("coreinit", "OSGetTick", export_OSGetTick);
-		osLib_addFunction("coreinit", "OSGetSystemTick", export_OSGetSystemTick);
+		cafeExportRegister("coreinit", OSGetSystemTime, LogType::Placeholder);
+		cafeExportRegister("coreinit", OSGetTick, LogType::Placeholder);
+		cafeExportRegister("coreinit", OSGetSystemTick, LogType::Placeholder);
 
 		cafeExportRegister("coreinit", OSTicksToCalendarTime, LogType::Placeholder);
 		cafeExportRegister("coreinit", OSCalendarTimeToTicks, LogType::Placeholder);
 
-
 		//timeTest();
 	}
 };

src/Cafe/OS/libs/coreinit/coreinit_Time.h

@@ -40,20 +40,21 @@ namespace coreinit
 
 		inline TimerTicks ConvertNsToTimerTicks(uint64 ns)
 		{
-			return ((GetTimerClock() / 31250LL) * ((ns)) / 32000LL);
+			return static_cast<TimerTicks>((static_cast<uint64>(GetTimerClock()) / 31250ULL) * (ns) / 32000ULL);
+		}
+
+		inline TimerTicks ConvertMsToTimerTicks(uint64 ms)
+		{
+			return static_cast<TimerTicks>(ms * static_cast<uint64>(GetTimerClock()) / 1000ULL);
 		}
 	};
 
 	void OSTicksToCalendarTime(uint64 ticks, OSCalendarTime_t* calenderStruct);
+
+	uint64 OSGetSystemTime();
 	uint64 OSGetTime();
-
-	uint64 coreinit_getOSTime();
-	uint64 coreinit_getTimerTick();
-
-	static uint64 OSGetSystemTime()
-	{
-		return coreinit_getTimerTick();
-	}
+	uint32 OSGetSystemTick();
+	uint32 OSGetTick();
 
 	void InitializeTimeAndCalendar();
 };

src/Cafe/OS/libs/dmae/dmae.cpp

@@ -11,7 +11,7 @@ uint64 dmaeRetiredTimestamp = 0;
 
 uint64 dmae_getTimestamp()
 {
-	return coreinit::coreinit_getTimerTick();
+	return coreinit::OSGetSystemTime();
 }
 
 void dmae_setRetiredTimestamp(uint64 timestamp)

src/Cafe/OS/libs/erreula/erreula.cpp

@@ -9,32 +9,45 @@
 #include <wx/msgdlg.h>
 
 #include "Cafe/OS/libs/coreinit/coreinit_FS.h"
+#include "Cafe/OS/libs/coreinit/coreinit_Time.h"
 #include "Cafe/OS/libs/vpad/vpad.h"
 
 namespace nn
 {
 namespace erreula
 {
-#define RESULTTYPE_NONE		0
-#define RESULTTYPE_FINISH	1
-#define RESULTTYPE_NEXT		2
-#define RESULTTYPE_JUMP		3
-#define RESULTTYPE_PASSWORD 4
 
-#define ERRORTYPE_CODE 0
-#define ERRORTYPE_TEXT 1
-#define ERRORTYPE_TEXT_ONE_BUTTON 2
-#define ERRORTYPE_TEXT_TWO_BUTTON 3
-
-#define ERREULA_STATE_HIDDEN		0
-#define ERREULA_STATE_APPEARING		1
-#define ERREULA_STATE_VISIBLE		2
-#define ERREULA_STATE_DISAPPEARING	3
-
-	struct AppearArg_t
+	enum class ErrorDialogType : uint32
 	{
-		AppearArg_t() = default;
-		AppearArg_t(const AppearArg_t& o)
+		Code = 0,
+		Text = 1,
+		TextOneButton = 2,
+		TextTwoButton = 3
+	};
+
+	static const sint32 FADE_TIME = 80;
+
+	enum class ErrEulaState : uint32
+	{
+		Hidden = 0,
+		Appearing = 1,
+		Visible = 2,
+		Disappearing = 3
+	};
+
+	enum class ResultType : uint32
+	{
+		None = 0,
+		Finish = 1,
+		Next = 2,
+		Jump = 3,
+		Password = 4
+	};
+
+	struct AppearError
+	{
+		AppearError() = default;
+		AppearError(const AppearError& o)
 		{
 			errorType = o.errorType;
 			screenType = o.screenType;
@@ -49,7 +62,7 @@ namespace erreula
 			drawCursor = o.drawCursor;
 		}
 
-		uint32be errorType;
+		betype<ErrorDialogType> errorType;
 		uint32be screenType;
 		uint32be controllerType;
 		uint32be holdType;
@@ -63,7 +76,9 @@ namespace erreula
 		bool drawCursor{};
 	};
 
-	static_assert(sizeof(AppearArg_t) == 0x2C); // maybe larger
+	using AppearArg = AppearError;
+
+	static_assert(sizeof(AppearError) == 0x2C); // maybe larger
 
 	struct HomeNixSignArg_t
 	{
@@ -80,6 +95,132 @@ namespace erreula
 
 	static_assert(sizeof(ControllerInfo_t) == 0x14); // maybe larger
 
+	class ErrEulaInstance
+	{
+	  public:
+		enum class BUTTON_SELECTION : uint32
+		{
+			NONE = 0xFFFFFFFF,
+			LEFT = 0,
+			RIGHT = 1,
+		};
+
+		void Init()
+		{
+			m_buttonSelection = BUTTON_SELECTION::NONE;
+			m_resultCode = -1;
+			m_resultCodeForLeftButton = 0;
+			m_resultCodeForRightButton = 0;
+			SetState(ErrEulaState::Hidden);
+		}
+
+		void DoAppearError(AppearArg* arg)
+		{
+			m_buttonSelection = BUTTON_SELECTION::NONE;
+			m_resultCode = -1;
+			m_resultCodeForLeftButton = -1;
+			m_resultCodeForRightButton = -1;
+			// for standard dialog its 0 and 1?
+			m_resultCodeForLeftButton = 0;
+			m_resultCodeForRightButton = 1;
+			SetState(ErrEulaState::Appearing);
+		}
+
+		void DoDisappearError()
+		{
+			if(m_state != ErrEulaState::Visible)
+				return;
+			SetState(ErrEulaState::Disappearing);
+		}
+
+		void DoCalc()
+		{
+			// appearing and disappearing state will automatically advance after some time
+			if (m_state == ErrEulaState::Appearing || m_state == ErrEulaState::Disappearing)
+			{
+				uint32 elapsedTick = coreinit::OSGetTime() - m_lastStateChange;
+				if (elapsedTick > coreinit::EspressoTime::ConvertMsToTimerTicks(FADE_TIME))
+				{
+					SetState(m_state == ErrEulaState::Appearing ? ErrEulaState::Visible : ErrEulaState::Hidden);
+				}
+			}
+		}
+
+		bool IsDecideSelectButtonError() const
+		{
+			return m_buttonSelection != BUTTON_SELECTION::NONE;
+		}
+
+		bool IsDecideSelectLeftButtonError() const
+		{
+			return m_buttonSelection != BUTTON_SELECTION::LEFT;
+		}
+
+		bool IsDecideSelectRightButtonError() const
+		{
+			return m_buttonSelection != BUTTON_SELECTION::RIGHT;
+		}
+
+		void SetButtonSelection(BUTTON_SELECTION selection)
+		{
+			cemu_assert_debug(m_buttonSelection == BUTTON_SELECTION::NONE);
+			m_buttonSelection = selection;
+			cemu_assert_debug(selection == BUTTON_SELECTION::LEFT || selection == BUTTON_SELECTION::RIGHT);
+			m_resultCode = selection == BUTTON_SELECTION::LEFT ? m_resultCodeForLeftButton : m_resultCodeForRightButton;
+		}
+
+		ErrEulaState GetState() const
+		{
+			return m_state;
+		}
+
+		sint32 GetResultCode() const
+		{
+			return m_resultCode;
+		}
+
+		ResultType GetResultType() const
+		{
+		  if(m_resultCode == -1)
+			  return ResultType::None;
+		  if(m_resultCode < 10)
+			  return ResultType::Finish;
+		  if(m_resultCode >= 9999)
+			  return ResultType::Next;
+		  if(m_resultCode == 40)
+			  return ResultType::Password;
+		  return ResultType::Jump;
+		}
+
+		float GetFadeTransparency() const
+		{
+			if(m_state == ErrEulaState::Appearing || m_state == ErrEulaState::Disappearing)
+			{
+				uint32 elapsedTick = coreinit::OSGetTime() - m_lastStateChange;
+				if(m_state == ErrEulaState::Appearing)
+					return std::min<float>(1.0f, (float)elapsedTick / (float)coreinit::EspressoTime::ConvertMsToTimerTicks(FADE_TIME));
+				else
+					return std::max<float>(0.0f, 1.0f - (float)elapsedTick / (float)coreinit::EspressoTime::ConvertMsToTimerTicks(FADE_TIME));
+			}
+			return 1.0f;
+		}
+
+	  private:
+		void SetState(ErrEulaState state)
+		{
+			m_state = state;
+			m_lastStateChange = coreinit::OSGetTime();
+		}
+
+		ErrEulaState m_state;
+		uint32 m_lastStateChange;
+
+		/* +0x30 */ betype<sint32> m_resultCode;
+		/* +0x239C */ betype<BUTTON_SELECTION> m_buttonSelection;
+		/* +0x23A0 */ betype<sint32> m_resultCodeForLeftButton;
+		/* +0x23A4 */ betype<sint32> m_resultCodeForRightButton;
+	};
+
 	struct ErrEula_t
 	{
 		SysAllocator<coreinit::OSMutex> mutex;
@@ -87,18 +228,12 @@ namespace erreula
 		uint32 langType;
 		MEMPTR<coreinit::FSClient_t> fsClient;
 
-		AppearArg_t currentDialog;
-		uint32 state;
-		bool buttonPressed;
-		bool rightButtonPressed;
+		std::unique_ptr<ErrEulaInstance> errEulaInstance;
 
+		AppearError currentDialog;
 		bool homeNixSignVisible;
-
-		std::chrono::steady_clock::time_point stateTimer{};
 	} g_errEula = {};
 	
-
-	
 	std::wstring GetText(uint16be* text)
 	{
 		std::wstringstream result;
@@ -113,22 +248,61 @@ namespace erreula
 	}
 	
 
-	void export_ErrEulaCreate(PPCInterpreter_t* hCPU)
+	void ErrEulaCreate(void* workmem, uint32 regionType, uint32 langType, coreinit::FSClient_t* fsClient)
 	{
-		ppcDefineParamMEMPTR(thisptr, uint8, 0);
-		ppcDefineParamU32(regionType, 1);
-		ppcDefineParamU32(langType, 2);
-		ppcDefineParamMEMPTR(fsClient, coreinit::FSClient_t, 3);
-
 		coreinit::OSLockMutex(&g_errEula.mutex);
 
 		g_errEula.regionType = regionType;
 		g_errEula.langType = langType;
 		g_errEula.fsClient = fsClient;
+		cemu_assert_debug(!g_errEula.errEulaInstance);
+		g_errEula.errEulaInstance = std::make_unique<ErrEulaInstance>();
+		g_errEula.errEulaInstance->Init();
 
 		coreinit::OSUnlockMutex(&g_errEula.mutex);
+	}
 
-		osLib_returnFromFunction(hCPU, 0);
+	void ErrEulaDestroy()
+	{
+		g_errEula.errEulaInstance.reset();
+	}
+
+	// check if any dialog button was selected
+	bool IsDecideSelectButtonError()
+	{
+		if(!g_errEula.errEulaInstance)
+			return false;
+		return g_errEula.errEulaInstance->IsDecideSelectButtonError();
+	}
+
+	// check if left dialog button was selected
+	bool IsDecideSelectLeftButtonError()
+	{
+		if(!g_errEula.errEulaInstance)
+			return false;
+		return g_errEula.errEulaInstance->IsDecideSelectLeftButtonError();
+	}
+
+	// check if right dialog button was selected
+	bool IsDecideSelectRightButtonError()
+	{
+		if(!g_errEula.errEulaInstance)
+			return false;
+		return g_errEula.errEulaInstance->IsDecideSelectRightButtonError();
+	}
+
+	sint32 GetResultCode()
+	{
+		if(!g_errEula.errEulaInstance)
+			return -1;
+		return g_errEula.errEulaInstance->GetResultCode();
+	}
+
+	ResultType GetResultType()
+	{
+		if(!g_errEula.errEulaInstance)
+			return ResultType::None;
+		return g_errEula.errEulaInstance->GetResultType();
 	}
 
 	void export_AppearHomeNixSign(PPCInterpreter_t* hCPU)
@@ -137,28 +311,24 @@ namespace erreula
 		osLib_returnFromFunction(hCPU, 0);
 	}
 
-	void export_AppearError(PPCInterpreter_t* hCPU)
+	void ErrEulaAppearError(AppearArg* arg)
 	{
-		ppcDefineParamMEMPTR(arg, AppearArg_t, 0);
-
-		g_errEula.currentDialog = *arg.GetPtr();
-		g_errEula.state = ERREULA_STATE_APPEARING;
-		g_errEula.buttonPressed = false;
-		g_errEula.rightButtonPressed = false;
-
-		g_errEula.stateTimer = tick_cached();
-
-		osLib_returnFromFunction(hCPU, 0);
+		g_errEula.currentDialog = *arg;
+		if(g_errEula.errEulaInstance)
+			g_errEula.errEulaInstance->DoAppearError(arg);
 	}
 
-	void export_GetStateErrorViewer(PPCInterpreter_t* hCPU)
+	void ErrEulaDisappearError()
 	{
-		osLib_returnFromFunction(hCPU, g_errEula.state);
+		if(g_errEula.errEulaInstance)
+			g_errEula.errEulaInstance->DoDisappearError();
 	}
-	void export_DisappearError(PPCInterpreter_t* hCPU)
+
+	ErrEulaState ErrEulaGetStateErrorViewer()
 	{
-		g_errEula.state = ERREULA_STATE_HIDDEN;
-		osLib_returnFromFunction(hCPU, 0);
+		if(!g_errEula.errEulaInstance)
+			return ErrEulaState::Hidden;
+		return g_errEula.errEulaInstance->GetState();
 	}
 
 	void export_ChangeLang(PPCInterpreter_t* hCPU)
@@ -168,27 +338,6 @@ namespace erreula
 		osLib_returnFromFunction(hCPU, 0);
 	}
 
-	void export_IsDecideSelectButtonError(PPCInterpreter_t* hCPU)
-	{
-		if (g_errEula.buttonPressed)
-			cemuLog_logDebug(LogType::Force, "IsDecideSelectButtonError: TRUE");
-		osLib_returnFromFunction(hCPU, g_errEula.buttonPressed);
-	}
-
-	void export_IsDecideSelectLeftButtonError(PPCInterpreter_t* hCPU)
-	{
-		if (g_errEula.buttonPressed)
-			cemuLog_logDebug(LogType::Force, "IsDecideSelectLeftButtonError: TRUE");
-		osLib_returnFromFunction(hCPU, g_errEula.buttonPressed);
-	}
-
-	void export_IsDecideSelectRightButtonError(PPCInterpreter_t* hCPU)
-	{
-		if (g_errEula.rightButtonPressed)
-			cemuLog_logDebug(LogType::Force, "IsDecideSelectRightButtonError: TRUE");
-		osLib_returnFromFunction(hCPU, g_errEula.rightButtonPressed);
-	}
-
 	void export_IsAppearHomeNixSign(PPCInterpreter_t* hCPU)
 	{
 		osLib_returnFromFunction(hCPU, g_errEula.homeNixSignVisible);
@@ -200,61 +349,19 @@ namespace erreula
 		osLib_returnFromFunction(hCPU, 0);
 	}
 
-	void export_GetResultType(PPCInterpreter_t* hCPU)
+	void ErrEulaCalc(ControllerInfo_t* controllerInfo)
 	{
-		uint32 result = RESULTTYPE_NONE;
-		if (g_errEula.buttonPressed || g_errEula.rightButtonPressed)
-		{
-			cemuLog_logDebug(LogType::Force, "GetResultType: FINISH");
-			result = RESULTTYPE_FINISH;
-		}
-
-		osLib_returnFromFunction(hCPU, result);
-	}
-
-	void export_Calc(PPCInterpreter_t* hCPU)
-	{
-		ppcDefineParamMEMPTR(controllerInfo, ControllerInfo_t, 0);
-		// TODO: check controller buttons bla to accept dialog?
-		osLib_returnFromFunction(hCPU, 0);
+		if(g_errEula.errEulaInstance)
+			g_errEula.errEulaInstance->DoCalc();
 	}
 
 	void render(bool mainWindow)
 	{
-		if(g_errEula.state == ERREULA_STATE_HIDDEN)
+		if(!g_errEula.errEulaInstance)
 			return;
-
-		if(g_errEula.state == ERREULA_STATE_APPEARING)
-		{
-			if(std::chrono::duration_cast<std::chrono::milliseconds>(tick_cached() - g_errEula.stateTimer).count() <= 1000)
-			{
-				return;
-			}
-
-			g_errEula.state = ERREULA_STATE_VISIBLE;
-			g_errEula.stateTimer = tick_cached();
-		}
-		/*else if(g_errEula.state == STATE_VISIBLE)
-		{
-			if (std::chrono::duration_cast<std::chrono::milliseconds>(tick_cached() - g_errEula.stateTimer).count() >= 1000)
-			{
-				g_errEula.state = STATE_DISAPPEARING;
-				g_errEula.stateTimer = tick_cached();
-				return;
-			}
-		}*/
-		else if(g_errEula.state == ERREULA_STATE_DISAPPEARING)
-		{
-			if (std::chrono::duration_cast<std::chrono::milliseconds>(tick_cached() - g_errEula.stateTimer).count() >= 2000)
-			{
-				g_errEula.state = ERREULA_STATE_HIDDEN;
-				g_errEula.stateTimer = tick_cached();
-			}
-
+		if(g_errEula.errEulaInstance->GetState() != ErrEulaState::Visible && g_errEula.errEulaInstance->GetState() != ErrEulaState::Appearing && g_errEula.errEulaInstance->GetState() != ErrEulaState::Disappearing)
 			return;
-		}
-		
-		const AppearArg_t& appearArg = g_errEula.currentDialog;
+		const AppearError& appearArg = g_errEula.currentDialog;
 		std::string text;
 		const uint32 errorCode = (uint32)appearArg.errorCode;
 		if (errorCode != 0)
@@ -280,13 +387,24 @@ namespace erreula
 		std::string title;
 		if (appearArg.title)
 			title = boost::nowide::narrow(GetText(appearArg.title.GetPtr()));
-		if(title.empty()) // ImGui doesn't allow empty titles, so set one if appearArg.title is not set or empty
+		if (title.empty()) // ImGui doesn't allow empty titles, so set one if appearArg.title is not set or empty
 			title = "ErrEula";
+
+		float fadeTransparency = 1.0f;
+		if (g_errEula.errEulaInstance->GetState() == ErrEulaState::Appearing || g_errEula.errEulaInstance->GetState() == ErrEulaState::Disappearing)
+		{
+			fadeTransparency = g_errEula.errEulaInstance->GetFadeTransparency();
+		}
+
+		float originalAlpha = ImGui::GetStyle().Alpha;
+		ImGui::GetStyle().Alpha = fadeTransparency;
+		ImGui::SetNextWindowBgAlpha(0.9f * fadeTransparency);
 		if (ImGui::Begin(title.c_str(), nullptr, kPopupFlags))
 		{
 			const float startx = ImGui::GetWindowSize().x / 2.0f;
+			bool hasLeftButtonPressed = false, hasRightButtonPressed = false;
 
-			switch ((uint32)appearArg.errorType)
+			switch (appearArg.errorType)
 			{
 			default:
 			{
@@ -294,11 +412,10 @@ namespace erreula
 				ImGui::TextUnformatted(text.c_str(), text.c_str() + text.size());
 				ImGui::Spacing();
 				ImGui::SetCursorPosX(startx - 50);
-				g_errEula.buttonPressed |= ImGui::Button("OK", {100, 0});
-				
+				hasLeftButtonPressed = ImGui::Button("OK", {100, 0});
 				break;
 			}
-			case ERRORTYPE_TEXT:
+			case ErrorDialogType::Text:
 			{
 				std::string txtTmp = "Unknown Error";
 				if (appearArg.text)
@@ -309,10 +426,10 @@ namespace erreula
 				ImGui::Spacing();
 					
 				ImGui::SetCursorPosX(startx - 50);
-				g_errEula.buttonPressed |= ImGui::Button("OK", { 100, 0 });
+				hasLeftButtonPressed = ImGui::Button("OK", { 100, 0 });
 				break;
 			}
-			case ERRORTYPE_TEXT_ONE_BUTTON:
+			case ErrorDialogType::TextOneButton:
 			{
 				std::string txtTmp = "Unknown Error";
 				if (appearArg.text)
@@ -328,10 +445,10 @@ namespace erreula
 
 				float width = std::max(100.0f, ImGui::CalcTextSize(button1.c_str()).x + 10.0f);
 				ImGui::SetCursorPosX(startx - (width / 2.0f));
-				g_errEula.buttonPressed |= ImGui::Button(button1.c_str(), { width, 0 });
+				hasLeftButtonPressed = ImGui::Button(button1.c_str(), { width, 0 });
 				break;
 			}
-			case ERRORTYPE_TEXT_TWO_BUTTON:
+			case ErrorDialogType::TextTwoButton:
 			{
 				std::string txtTmp = "Unknown Error";
 				if (appearArg.text)
@@ -352,42 +469,52 @@ namespace erreula
 				float width2 = std::max(100.0f, ImGui::CalcTextSize(button2.c_str()).x + 10.0f);
 				ImGui::SetCursorPosX(startx - (width1 / 2.0f) - (width2 / 2.0f) - 10);
 					
-				g_errEula.buttonPressed |= ImGui::Button(button1.c_str(), { width1, 0 });
+				hasLeftButtonPressed = ImGui::Button(button1.c_str(), { width1, 0 });
 				ImGui::SameLine();
 					
-				g_errEula.rightButtonPressed |= ImGui::Button(button2.c_str(), { width2, 0 });
+				hasRightButtonPressed = ImGui::Button(button2.c_str(), { width2, 0 });
 				break;
 			}
 			}
+			if (!g_errEula.errEulaInstance->IsDecideSelectButtonError())
+			{
+				if (hasLeftButtonPressed)
+					g_errEula.errEulaInstance->SetButtonSelection(ErrEulaInstance::BUTTON_SELECTION::LEFT);
+				if (hasRightButtonPressed)
+					g_errEula.errEulaInstance->SetButtonSelection(ErrEulaInstance::BUTTON_SELECTION::RIGHT);
+			}
 		}
 		ImGui::End();
 		ImGui::PopFont();
-
-		if(g_errEula.buttonPressed || g_errEula.rightButtonPressed)
-		{
-			g_errEula.state = ERREULA_STATE_DISAPPEARING;
-			g_errEula.stateTimer = tick_cached();
-		}
+		ImGui::GetStyle().Alpha = originalAlpha;
 	}
 
 	void load()
 	{
+		g_errEula.errEulaInstance.reset();
+
 		OSInitMutexEx(&g_errEula.mutex, nullptr);
 
-		//osLib_addFunction("erreula", "ErrEulaCreate__3RplFPUcQ3_2nn7erreula10", export_ErrEulaCreate); // copy ctor?
-		osLib_addFunction("erreula", "ErrEulaCreate__3RplFPUcQ3_2nn7erreula10RegionTypeQ3_2nn7erreula8LangTypeP8FSClient", export_ErrEulaCreate);
+		cafeExportRegisterFunc(ErrEulaCreate, "erreula", "ErrEulaCreate__3RplFPUcQ3_2nn7erreula10RegionTypeQ3_2nn7erreula8LangTypeP8FSClient", LogType::Placeholder);
+		cafeExportRegisterFunc(ErrEulaDestroy, "erreula", "ErrEulaDestroy__3RplFv", LogType::Placeholder);
+
+		cafeExportRegisterFunc(IsDecideSelectButtonError, "erreula", "ErrEulaIsDecideSelectButtonError__3RplFv", LogType::Placeholder);
+		cafeExportRegisterFunc(IsDecideSelectLeftButtonError, "erreula", "ErrEulaIsDecideSelectLeftButtonError__3RplFv", LogType::Placeholder);
+		cafeExportRegisterFunc(IsDecideSelectRightButtonError, "erreula", "ErrEulaIsDecideSelectRightButtonError__3RplFv", LogType::Placeholder);
+
+		cafeExportRegisterFunc(GetResultCode, "erreula", "ErrEulaGetResultCode__3RplFv", LogType::Placeholder);
+		cafeExportRegisterFunc(GetResultType, "erreula", "ErrEulaGetResultType__3RplFv", LogType::Placeholder);
+
+		cafeExportRegisterFunc(ErrEulaAppearError, "erreula", "ErrEulaAppearError__3RplFRCQ3_2nn7erreula9AppearArg", LogType::Placeholder);
+		cafeExportRegisterFunc(ErrEulaDisappearError, "erreula", "ErrEulaDisappearError__3RplFv", LogType::Placeholder);
+		cafeExportRegisterFunc(ErrEulaGetStateErrorViewer, "erreula", "ErrEulaGetStateErrorViewer__3RplFv", LogType::Placeholder);
+
+		cafeExportRegisterFunc(ErrEulaCalc, "erreula", "ErrEulaCalc__3RplFRCQ3_2nn7erreula14ControllerInfo", LogType::Placeholder);
+
 		osLib_addFunction("erreula", "ErrEulaAppearHomeNixSign__3RplFRCQ3_2nn7erreula14HomeNixSignArg", export_AppearHomeNixSign);
-		osLib_addFunction("erreula", "ErrEulaAppearError__3RplFRCQ3_2nn7erreula9AppearArg", export_AppearError);
-		osLib_addFunction("erreula", "ErrEulaGetStateErrorViewer__3RplFv", export_GetStateErrorViewer);
 		osLib_addFunction("erreula", "ErrEulaChangeLang__3RplFQ3_2nn7erreula8LangType", export_ChangeLang);
-		osLib_addFunction("erreula", "ErrEulaIsDecideSelectButtonError__3RplFv", export_IsDecideSelectButtonError);
-		osLib_addFunction("erreula", "ErrEulaCalc__3RplFRCQ3_2nn7erreula14ControllerInfo", export_Calc);
-		osLib_addFunction("erreula", "ErrEulaIsDecideSelectLeftButtonError__3RplFv", export_IsDecideSelectLeftButtonError);
-		osLib_addFunction("erreula", "ErrEulaIsDecideSelectRightButtonError__3RplFv", export_IsDecideSelectRightButtonError);
 		osLib_addFunction("erreula", "ErrEulaIsAppearHomeNixSign__3RplFv", export_IsAppearHomeNixSign);
 		osLib_addFunction("erreula", "ErrEulaDisappearHomeNixSign__3RplFv", export_DisappearHomeNixSign);
-		osLib_addFunction("erreula", "ErrEulaGetResultType__3RplFv", export_GetResultType);
-		osLib_addFunction("erreula", "ErrEulaDisappearError__3RplFv", export_DisappearError);
 	}
 }
 }

src/Cafe/OS/libs/gx2/GX2.cpp

@@ -322,7 +322,7 @@ uint64 _prevReturnedGPUTime = 0;
 
 uint64 Latte_GetTime()
 {
-	uint64 gpuTime = coreinit::coreinit_getTimerTick();
+	uint64 gpuTime = coreinit::OSGetSystemTime();
 	gpuTime *= 20000ULL;
 	if (gpuTime <= _prevReturnedGPUTime)
 		gpuTime = _prevReturnedGPUTime + 1; // avoid ever returning identical timestamps

src/Cafe/OS/libs/gx2/GX2_Misc.cpp

@@ -54,7 +54,7 @@ void gx2Export_GX2GetGPUTimeout(PPCInterpreter_t* hCPU)
 void gx2Export_GX2SampleTopGPUCycle(PPCInterpreter_t* hCPU)
 {
 	cemuLog_log(LogType::GX2, "GX2SampleTopGPUCycle(0x{:08x})", hCPU->gpr[3]);
-	memory_writeU64(hCPU->gpr[3], coreinit::coreinit_getTimerTick());
+	memory_writeU64(hCPU->gpr[3], coreinit::OSGetSystemTime());
 	osLib_returnFromFunction(hCPU, 0);
 }
 

src/Cafe/OS/libs/gx2/GX2_Resource.cpp

@@ -87,6 +87,11 @@ namespace GX2
 		return true;
 	}
 
+	void GX2RSetBufferName(GX2RBuffer* buffer, const char* name)
+	{
+		// no-op in production builds
+	}
+
 	void* GX2RLockBufferEx(GX2RBuffer* buffer, uint32 resFlags)
 	{
 		return buffer->GetPtr();
@@ -226,6 +231,7 @@ namespace GX2
 		cafeExportRegister("gx2", GX2RCreateBufferUserMemory, LogType::GX2);
 		cafeExportRegister("gx2", GX2RDestroyBufferEx, LogType::GX2);
 		cafeExportRegister("gx2", GX2RBufferExists, LogType::GX2);
+		cafeExportRegister("gx2", GX2RSetBufferName, LogType::GX2);
 		cafeExportRegister("gx2", GX2RLockBufferEx, LogType::GX2);
 		cafeExportRegister("gx2", GX2RUnlockBufferEx, LogType::GX2);
 		cafeExportRegister("gx2", GX2RInvalidateBuffer, LogType::GX2);

src/Cafe/OS/libs/gx2/GX2_Shader.cpp

@@ -421,7 +421,7 @@ namespace GX2
 	{
 		if(aluRegisterOffset&0x8000)
 		{
-			cemuLog_logDebug(LogType::Force, "_GX2SubmitUniformReg(): Unhandled loop const special case or invalid offset");
+			cemuLog_logDebugOnce(LogType::Force, "_GX2SubmitUniformReg(): Unhandled loop const special case or invalid offset");
 			return;
 		}
 		if((aluRegisterOffset+sizeInU32s) > 0x400)

src/Cafe/OS/libs/nn_acp/nn_acp.cpp

@@ -315,7 +315,7 @@ namespace acp
 		ppcDefineParamU32BEPtr(timestamp64, 0);
 		ppcDefineParamU32BEPtr(ukn, 1); // probably timezone or offset? Could also be a bool for success/failed
 
-		uint64 t = coreinit::coreinit_getOSTime() + (uint64)((sint64)(ppcCyclesSince2000_UTC - ppcCyclesSince2000) / 20LL);
+		uint64 t = coreinit::OSGetTime() + (uint64)((sint64)(ppcCyclesSince2000_UTC - ppcCyclesSince2000) / 20LL);
 
 		timestamp64[0] = (uint32)(t >> 32);
 		timestamp64[1] = (uint32)(t & 0xFFFFFFFF);

src/Cafe/OS/libs/nsyshid/AttachDefaultBackends.cpp

@@ -1,24 +1,12 @@
 #include "nsyshid.h"
 #include "Backend.h"
 #include "BackendEmulated.h"
-
-#if NSYSHID_ENABLE_BACKEND_LIBUSB
-
 #include "BackendLibusb.h"
 
-#endif
-
-#if NSYSHID_ENABLE_BACKEND_WINDOWS_HID
-
-#include "BackendWindowsHID.h"
-
-#endif
-
 namespace nsyshid::backend
 {
 	void AttachDefaultBackends()
 	{
-#if NSYSHID_ENABLE_BACKEND_LIBUSB
 		// add libusb backend
 		{
 			auto backendLibusb = std::make_shared<backend::libusb::BackendLibusb>();
@@ -27,17 +15,6 @@ namespace nsyshid::backend
 				AttachBackend(backendLibusb);
 			}
 		}
-#endif // NSYSHID_ENABLE_BACKEND_LIBUSB
-#if NSYSHID_ENABLE_BACKEND_WINDOWS_HID
-		// add windows hid backend
-		{
-			auto backendWindowsHID = std::make_shared<backend::windows::BackendWindowsHID>();
-			if (backendWindowsHID->IsInitialisedOk())
-			{
-				AttachBackend(backendWindowsHID);
-			}
-		}
-#endif // NSYSHID_ENABLE_BACKEND_WINDOWS_HID
 	   // add emulated backend
 		{
 			auto backendEmulated = std::make_shared<backend::emulated::BackendEmulated>();

src/Cafe/OS/libs/nsyshid/Backend.h

@@ -1,5 +1,4 @@
-#ifndef CEMU_NSYSHID_BACKEND_H
-#define CEMU_NSYSHID_BACKEND_H
+#pragma once
 
 #include <list>
 #include <memory>
@@ -26,9 +25,9 @@ namespace nsyshid
 	struct TransferCommand
 	{
 		uint8* data;
-		sint32 length;
+		uint32 length;
 
-		TransferCommand(uint8* data, sint32 length)
+		TransferCommand(uint8* data, uint32 length)
 			: data(data), length(length)
 		{
 		}
@@ -39,7 +38,7 @@ namespace nsyshid
 	{
 		sint32 bytesRead;
 
-		ReadMessage(uint8* data, sint32 length, sint32 bytesRead)
+		ReadMessage(uint8* data, uint32 length, sint32 bytesRead)
 			: bytesRead(bytesRead), TransferCommand(data, length)
 		{
 		}
@@ -50,7 +49,7 @@ namespace nsyshid
 	{
 		sint32 bytesWritten;
 
-		WriteMessage(uint8* data, sint32 length, sint32 bytesWritten)
+		WriteMessage(uint8* data, uint32 length, sint32 bytesWritten)
 			: bytesWritten(bytesWritten), TransferCommand(data, length)
 		{
 		}
@@ -59,14 +58,11 @@ namespace nsyshid
 
 	struct ReportMessage final : TransferCommand
 	{
-		uint8* reportData;
-		sint32 length;
-		uint8* originalData;
-		sint32 originalLength;
+		uint8 reportType;
+		uint8 reportId;
 
-		ReportMessage(uint8* reportData, sint32 length, uint8* originalData, sint32 originalLength)
-			: reportData(reportData), length(length), originalData(originalData),
-			  originalLength(originalLength), TransferCommand(reportData, length)
+		ReportMessage(uint8 reportType, uint8 reportId, uint8* data, uint32 length)
+			: reportType(reportType), reportId(reportId), TransferCommand(data, length)
 		{
 		}
 		using TransferCommand::TransferCommand;
@@ -77,7 +73,8 @@ namespace nsyshid
 	static_assert(offsetof(HID_t, ifIndex) == 0xC, "");
 	static_assert(offsetof(HID_t, protocol) == 0xE, "");
 
-	class Device {
+	class Device
+	{
 	  public:
 		Device() = delete;
 
@@ -131,16 +128,21 @@ namespace nsyshid
 
 		virtual bool GetDescriptor(uint8 descType,
 								   uint8 descIndex,
-								   uint8 lang,
+								   uint16 lang,
 								   uint8* output,
 								   uint32 outputMaxLength) = 0;
 
+		virtual bool SetIdle(uint8 ifIndex,
+							 uint8 reportId,
+							 uint8 duration) = 0;
+
 		virtual bool SetProtocol(uint8 ifIndex, uint8 protocol) = 0;
 
 		virtual bool SetReport(ReportMessage* message) = 0;
 	};
 
-	class Backend {
+	class Backend
+	{
 	  public:
 		Backend();
 
@@ -188,5 +190,3 @@ namespace nsyshid
 		void AttachDefaultBackends();
 	}
 } // namespace nsyshid
-
-#endif // CEMU_NSYSHID_BACKEND_H

src/Cafe/OS/libs/nsyshid/BackendEmulated.cpp

@@ -1,4 +1,6 @@
 #include "BackendEmulated.h"
+
+#include "Dimensions.h"
 #include "Infinity.h"
 #include "Skylander.h"
 #include "config/CemuConfig.h"
@@ -33,5 +35,12 @@ namespace nsyshid::backend::emulated
 			auto device = std::make_shared<InfinityBaseDevice>();
 			AttachDevice(device);
 		}
+		if (GetConfig().emulated_usb_devices.emulate_dimensions_toypad && !FindDeviceById(0x0E6F, 0x0241))
+		{
+			cemuLog_logDebug(LogType::Force, "Attaching Emulated Toypad");
+			// Add Dimensions Toypad
+			auto device = std::make_shared<DimensionsToypadDevice>();
+			AttachDevice(device);
+		}
 	}
 } // namespace nsyshid::backend::emulated

src/Cafe/OS/libs/nsyshid/BackendLibusb.cpp

@@ -1,7 +1,5 @@
 #include "BackendLibusb.h"
 
-#if NSYSHID_ENABLE_BACKEND_LIBUSB
-
 namespace nsyshid::backend::libusb
 {
 	BackendLibusb::BackendLibusb()
@@ -15,8 +13,8 @@ namespace nsyshid::backend::libusb
 		if (m_initReturnCode < 0)
 		{
 			m_ctx = nullptr;
-			cemuLog_logDebug(LogType::Force, "nsyshid::BackendLibusb: failed to initialize libusb with return code %i",
-							 m_initReturnCode);
+			cemuLog_logDebug(LogType::Force, "nsyshid::BackendLibusb: failed to initialize libusb, return code: {}",
+						m_initReturnCode);
 			return;
 		}
 
@@ -35,8 +33,8 @@ namespace nsyshid::backend::libusb
 			if (ret != LIBUSB_SUCCESS)
 			{
 				cemuLog_logDebug(LogType::Force,
-								 "nsyshid::BackendLibusb: failed to register hotplug callback with return code %i",
-								 ret);
+							"nsyshid::BackendLibusb: failed to register hotplug callback with return code {}",
+							ret);
 			}
 			else
 			{
@@ -53,8 +51,8 @@ namespace nsyshid::backend::libusb
 						if (ret != 0)
 						{
 							cemuLog_logDebug(LogType::Force,
-											 "nsyshid::BackendLibusb: hotplug thread: error handling events: {}",
-											 ret);
+										"nsyshid::BackendLibusb: hotplug thread: error handling events: {}",
+										ret);
 							std::this_thread::sleep_for(std::chrono::milliseconds(1000));
 						}
 					}
@@ -139,8 +137,8 @@ namespace nsyshid::backend::libusb
 		case LIBUSB_HOTPLUG_EVENT_DEVICE_ARRIVED:
 		{
 			cemuLog_logDebug(LogType::Force, "nsyshid::BackendLibusb::OnHotplug(): device arrived: {:04x}:{:04x}",
-							 desc.idVendor,
-							 desc.idProduct);
+						desc.idVendor,
+						desc.idProduct);
 			auto device = CheckAndCreateDevice(dev);
 			if (device != nullptr)
 			{
@@ -167,8 +165,8 @@ namespace nsyshid::backend::libusb
 		case LIBUSB_HOTPLUG_EVENT_DEVICE_LEFT:
 		{
 			cemuLog_logDebug(LogType::Force, "nsyshid::BackendLibusb::OnHotplug(): device left: {:04x}:{:04x}",
-							 desc.idVendor,
-							 desc.idProduct);
+						desc.idVendor,
+						desc.idProduct);
 			auto device = FindLibusbDevice(dev);
 			if (device != nullptr)
 			{
@@ -204,7 +202,7 @@ namespace nsyshid::backend::libusb
 		if (ret < 0)
 		{
 			cemuLog_logDebug(LogType::Force,
-							 "nsyshid::BackendLibusb::FindLibusbDevice(): failed to get device descriptor");
+						"nsyshid::BackendLibusb::FindLibusbDevice(): failed to get device descriptor");
 			return nullptr;
 		}
 		uint8 busNumber = libusb_get_bus_number(dev);
@@ -269,12 +267,12 @@ namespace nsyshid::backend::libusb
 		if (desc.idVendor == 0x0e6f && desc.idProduct == 0x0241)
 		{
 			cemuLog_logDebug(LogType::Force,
-							 "nsyshid::BackendLibusb::CheckAndCreateDevice(): lego dimensions portal detected");
+						"nsyshid::BackendLibusb::CheckAndCreateDevice(): lego dimensions portal detected");
 		}
 		auto device = std::make_shared<DeviceLibusb>(m_ctx,
 													 desc.idVendor,
 													 desc.idProduct,
-													 1,
+													 0,
 													 2,
 													 0,
 													 libusb_get_bus_number(dev),
@@ -415,7 +413,7 @@ namespace nsyshid::backend::libusb
 			if (ret < 0)
 			{
 				cemuLog_log(LogType::Force,
-							"nsyshid::DeviceLibusb::open(): failed to get device descriptor; return code: %i",
+							"nsyshid::DeviceLibusb::open(): failed to get device descriptor, return code: {}",
 							ret);
 				libusb_free_device_list(devices, 1);
 				return false;
@@ -439,19 +437,20 @@ namespace nsyshid::backend::libusb
 				{
 					this->m_libusbHandle = nullptr;
 					cemuLog_log(LogType::Force,
-								"nsyshid::DeviceLibusb::open(): failed to open device; return code: %i",
-								ret);
+								"nsyshid::DeviceLibusb::open(): failed to open device: {}",
+								libusb_strerror(ret));
 					libusb_free_device_list(devices, 1);
 					return false;
 				}
 				this->m_handleInUseCounter = 0;
 			}
+
+			int ret = ClaimAllInterfaces(0);
+
+			if (ret != 0)
 			{
-				int ret = ClaimAllInterfaces(0);
-				if (ret != 0)
-				{
-					cemuLog_logDebug(LogType::Force, "nsyshid::DeviceLibusb::open(): cannot claim interface");
-				}
+				cemuLog_log(LogType::Force, "nsyshid::DeviceLibusb::open(): cannot claim interface for config 0");
+				return false;
 			}
 		}
 
@@ -475,7 +474,7 @@ namespace nsyshid::backend::libusb
 			{
 				m_handleInUseCounterDecremented.wait(lock);
 			}
-			libusb_release_interface(handle, 0);
+			ReleaseAllInterfacesForCurrentConfig();
 			libusb_close(handle);
 			m_handleInUseCounter = -1;
 			m_handleInUseCounterDecremented.notify_all();
@@ -493,21 +492,26 @@ namespace nsyshid::backend::libusb
 		if (!handleLock->IsValid())
 		{
 			cemuLog_logDebug(LogType::Force,
-							 "nsyshid::DeviceLibusb::read(): cannot read from a non-opened device\n");
+						"nsyshid::DeviceLibusb::read(): cannot read from a non-opened device\n");
 			return ReadResult::Error;
 		}
 
+		for (int i = 0; i < m_config_descriptors.size(); i++)
+		{
+			ClaimAllInterfaces(i);
+		}
+
 		const unsigned int timeout = 50;
 		int actualLength = 0;
 		int ret = 0;
 		do
 		{
-			ret = libusb_bulk_transfer(handleLock->GetHandle(),
-									   this->m_libusbEndpointIn,
-									   message->data,
-									   message->length,
-									   &actualLength,
-									   timeout);
+			ret = libusb_interrupt_transfer(handleLock->GetHandle(),
+											this->m_libusbEndpointIn,
+											message->data,
+											message->length,
+											&actualLength,
+											timeout);
 		}
 		while (ret == LIBUSB_ERROR_TIMEOUT && actualLength == 0 && IsOpened());
 
@@ -521,8 +525,8 @@ namespace nsyshid::backend::libusb
 			return ReadResult::Success;
 		}
 		cemuLog_logDebug(LogType::Force,
-						 "nsyshid::DeviceLibusb::read(): failed with error code: {}",
-						 ret);
+					"nsyshid::DeviceLibusb::read(): failed at endpoint 0x{:02x} with error message: {}", this->m_libusbEndpointIn,
+					libusb_error_name(ret));
 		return ReadResult::Error;
 	}
 
@@ -532,18 +536,23 @@ namespace nsyshid::backend::libusb
 		if (!handleLock->IsValid())
 		{
 			cemuLog_logDebug(LogType::Force,
-							 "nsyshid::DeviceLibusb::write(): cannot write to a non-opened device\n");
+						"nsyshid::DeviceLibusb::write(): cannot write to a non-opened device\n");
 			return WriteResult::Error;
 		}
 
+		for (int i = 0; i < m_config_descriptors.size(); i++)
+		{
+			ClaimAllInterfaces(i);
+		}
+
 		message->bytesWritten = 0;
 		int actualLength = 0;
-		int ret = libusb_bulk_transfer(handleLock->GetHandle(),
-									   this->m_libusbEndpointOut,
-									   message->data,
-									   message->length,
-									   &actualLength,
-									   0);
+		int ret = libusb_interrupt_transfer(handleLock->GetHandle(),
+											this->m_libusbEndpointOut,
+											message->data,
+											message->length,
+											&actualLength,
+											0);
 
 		if (ret == 0)
 		{
@@ -556,14 +565,14 @@ namespace nsyshid::backend::libusb
 			return WriteResult::Success;
 		}
 		cemuLog_logDebug(LogType::Force,
-						 "nsyshid::DeviceLibusb::write(): failed with error code: {}",
-						 ret);
+					"nsyshid::DeviceLibusb::write(): failed with error code: {}",
+					ret);
 		return WriteResult::Error;
 	}
 
 	bool DeviceLibusb::GetDescriptor(uint8 descType,
 									 uint8 descIndex,
-									 uint8 lang,
+									 uint16 lang,
 									 uint8* output,
 									 uint32 outputMaxLength)
 	{
@@ -579,7 +588,6 @@ namespace nsyshid::backend::libusb
 			struct libusb_config_descriptor* conf = nullptr;
 			libusb_device* dev = libusb_get_device(handleLock->GetHandle());
 			int ret = libusb_get_active_config_descriptor(dev, &conf);
-
 			if (ret == 0)
 			{
 				std::vector<uint8> configurationDescriptor(conf->wTotalLength);
@@ -656,7 +664,6 @@ namespace nsyshid::backend::libusb
 								extraReadPointer += bLength;
 							}
 						}
-
 						for (int endpointIndex = 0; endpointIndex < altsetting.bNumEndpoints; endpointIndex++)
 						{
 							// endpoint descriptor
@@ -681,24 +688,61 @@ namespace nsyshid::backend::libusb
 				uint32 bytesWritten = currentWritePtr - &configurationDescriptor[0];
 				libusb_free_config_descriptor(conf);
 				cemu_assert_debug(bytesWritten <= conf->wTotalLength);
-
 				memcpy(output, &configurationDescriptor[0],
 					   std::min<uint32>(outputMaxLength, bytesWritten));
 				return true;
 			}
-			else
-			{
-				cemuLog_logDebug(LogType::Force,
-								 "nsyshid::DeviceLibusb::getDescriptor(): failed to get config descriptor with error code: {}",
-								 ret);
-				return false;
-			}
 		}
 		else
 		{
-			cemu_assert_unimplemented();
+			uint16 wValue = uint16(descType) << 8 | uint16(descIndex);
+			// HID Get_Descriptor requests are handled via libusb_control_transfer
+			int ret = libusb_control_transfer(handleLock->GetHandle(),
+											  LIBUSB_REQUEST_TYPE_STANDARD | LIBUSB_ENDPOINT_IN,
+											  LIBUSB_REQUEST_GET_DESCRIPTOR,
+											  wValue,
+											  lang,
+											  output,
+											  outputMaxLength,
+											  0);
+			if (ret != outputMaxLength)
+			{
+				cemuLog_logDebug(LogType::Force, "nsyshid::DeviceLibusb::GetDescriptor(): Control Transfer Failed: {}", libusb_error_name(ret));
+				return false;
+			}
 		}
-		return false;
+		return true;
+	}
+
+	bool DeviceLibusb::SetIdle(uint8 ifIndex,
+							   uint8 reportId,
+							   uint8 duration)
+	{
+		auto handleLock = AquireHandleLock();
+		if (!handleLock->IsValid())
+		{
+			cemuLog_log(LogType::Force, "nsyshid::DeviceLibusb::SetIdle(): device is not opened");
+			return false;
+		}
+
+		uint16 wValue = uint16(duration) << 8 | uint16(reportId);
+
+		// HID Set_Idle requests are handled via libusb_control_transfer
+		int ret = libusb_control_transfer(handleLock->GetHandle(),
+										  LIBUSB_REQUEST_TYPE_CLASS | LIBUSB_RECIPIENT_INTERFACE | LIBUSB_ENDPOINT_OUT,
+										  HID_CLASS_SET_IDLE, // Defined in HID Class Specific Requests (7.2)
+										  wValue,
+										  ifIndex,
+										  nullptr,
+										  0,
+										  0);
+
+		if (ret != 0)
+		{
+			cemuLog_logDebug(LogType::Force, "nsyshid::DeviceLibusb::SetIdle(): Control Transfer Failed: {}", libusb_error_name(ret));
+			return false;
+		}
+		return true;
 	}
 
 	template<typename Configs, typename Function>
@@ -767,18 +811,22 @@ namespace nsyshid::backend::libusb
 			cemuLog_logDebug(LogType::Force, "nsyshid::DeviceLibusb::SetProtocol(): device is not opened");
 			return false;
 		}
-		if (m_interfaceIndex != ifIndex)
-			m_interfaceIndex = ifIndex;
 
-		ReleaseAllInterfacesForCurrentConfig();
-		int ret = libusb_set_configuration(AquireHandleLock()->GetHandle(), protocol);
-		if (ret == LIBUSB_SUCCESS)
-			ret = ClaimAllInterfaces(protocol);
+		int ret = libusb_control_transfer(handleLock->GetHandle(),
+										  LIBUSB_REQUEST_TYPE_CLASS | LIBUSB_RECIPIENT_INTERFACE | LIBUSB_ENDPOINT_OUT,
+										  HID_CLASS_SET_PROTOCOL, // Defined in HID Class Specific Requests (7.2)
+										  protocol,
+										  ifIndex,
+										  nullptr,
+										  0,
+										  0);
 
-		if (ret == LIBUSB_SUCCESS)
-			return true;
-
-		return false;
+		if (ret != 0)
+		{
+			cemuLog_logDebug(LogType::Force, "nsyshid::DeviceLibusb::SetProtocol(): Control Transfer Failed: {}", libusb_error_name(ret));
+			return false;
+		}
+		return true;
 	}
 
 	bool DeviceLibusb::SetReport(ReportMessage* message)
@@ -790,18 +838,20 @@ namespace nsyshid::backend::libusb
 			return false;
 		}
 
+		uint16 wValue = uint16(message->reportType) << 8 | uint16(message->reportId);
+
 		int ret = libusb_control_transfer(handleLock->GetHandle(),
 										  LIBUSB_REQUEST_TYPE_CLASS | LIBUSB_RECIPIENT_INTERFACE | LIBUSB_ENDPOINT_OUT,
-										  LIBUSB_REQUEST_SET_CONFIGURATION,
-										  512,
-										  0,
-										  message->originalData,
-										  message->originalLength,
+										  HID_CLASS_SET_REPORT, // Defined in HID Class Specific Requests (7.2)
+										  wValue,
+										  m_interfaceIndex,
+										  message->data,
+										  uint16(message->length & 0xFFFF),
 										  0);
 
-		if (ret != message->originalLength)
+		if (ret != message->length)
 		{
-			cemuLog_logDebug(LogType::Force, "nsyshid::DeviceLibusb::SetReport(): Control Transfer Failed: {}", libusb_error_name(ret));
+			cemuLog_logDebug(LogType::Force, "nsyshid::DeviceLibusb::SetReport(): Control Transfer Failed at interface {} : {}", m_interfaceIndex, libusb_error_name(ret));
 			return false;
 		}
 		return true;
@@ -854,5 +904,3 @@ namespace nsyshid::backend::libusb
 		return m_handle;
 	}
 } // namespace nsyshid::backend::libusb
-
-#endif // NSYSHID_ENABLE_BACKEND_LIBUSB

src/Cafe/OS/libs/nsyshid/BackendLibusb.h

@@ -1,15 +1,20 @@
-#ifndef CEMU_NSYSHID_BACKEND_LIBUSB_H
-#define CEMU_NSYSHID_BACKEND_LIBUSB_H
-
 #include "nsyshid.h"
 
-#if NSYSHID_ENABLE_BACKEND_LIBUSB
-
 #include <libusb-1.0/libusb.h>
 #include "Backend.h"
 
 namespace nsyshid::backend::libusb
 {
+	enum : uint8
+	{
+		HID_CLASS_GET_REPORT   = 0x01,
+		HID_CLASS_GET_IDLE     = 0x02,
+		HID_CLASS_GET_PROTOCOL = 0x03,
+		HID_CLASS_SET_REPORT   = 0x09,
+		HID_CLASS_SET_IDLE     = 0x0A,
+		HID_CLASS_SET_PROTOCOL = 0x0B
+	};
+
 	class BackendLibusb : public nsyshid::Backend {
 	  public:
 		BackendLibusb();
@@ -75,10 +80,14 @@ namespace nsyshid::backend::libusb
 
 		bool GetDescriptor(uint8 descType,
 						   uint8 descIndex,
-						   uint8 lang,
+						   uint16 lang,
 						   uint8* output,
 						   uint32 outputMaxLength) override;
 
+		bool SetIdle(uint8 ifIndex,
+					 uint8 reportId,
+					 uint8 duration) override;
+
 		bool SetProtocol(uint8 ifIndex, uint8 protocol) override;
 
 		int ClaimAllInterfaces(uint8 config_num);
@@ -134,7 +143,3 @@ namespace nsyshid::backend::libusb
 		std::unique_ptr<HandleLock> AquireHandleLock();
 	};
 } // namespace nsyshid::backend::libusb
-
-#endif // NSYSHID_ENABLE_BACKEND_LIBUSB
-
-#endif // CEMU_NSYSHID_BACKEND_LIBUSB_H

src/Cafe/OS/libs/nsyshid/BackendWindowsHID.cpp

@@ -1,444 +0,0 @@
-#include "BackendWindowsHID.h"
-
-#if NSYSHID_ENABLE_BACKEND_WINDOWS_HID
-
-#include <setupapi.h>
-#include <initguid.h>
-#include <hidsdi.h>
-
-#pragma comment(lib, "Setupapi.lib")
-#pragma comment(lib, "hid.lib")
-
-DEFINE_GUID(GUID_DEVINTERFACE_HID,
-			0x4D1E55B2L, 0xF16F, 0x11CF, 0x88, 0xCB, 0x00, 0x11, 0x11, 0x00, 0x00, 0x30);
-
-namespace nsyshid::backend::windows
-{
-	BackendWindowsHID::BackendWindowsHID()
-	{
-	}
-
-	void BackendWindowsHID::AttachVisibleDevices()
-	{
-		// add all currently connected devices
-		HDEVINFO hDevInfo;
-		SP_DEVICE_INTERFACE_DATA DevIntfData;
-		PSP_DEVICE_INTERFACE_DETAIL_DATA DevIntfDetailData;
-		SP_DEVINFO_DATA DevData;
-
-		DWORD dwSize, dwMemberIdx;
-
-		hDevInfo = SetupDiGetClassDevs(&GUID_DEVINTERFACE_HID, NULL, 0, DIGCF_DEVICEINTERFACE | DIGCF_PRESENT);
-
-		if (hDevInfo != INVALID_HANDLE_VALUE)
-		{
-			DevIntfData.cbSize = sizeof(SP_DEVICE_INTERFACE_DATA);
-			dwMemberIdx = 0;
-
-			SetupDiEnumDeviceInterfaces(hDevInfo, NULL, &GUID_DEVINTERFACE_HID,
-										dwMemberIdx, &DevIntfData);
-
-			while (GetLastError() != ERROR_NO_MORE_ITEMS)
-			{
-				DevData.cbSize = sizeof(DevData);
-				SetupDiGetDeviceInterfaceDetail(
-					hDevInfo, &DevIntfData, NULL, 0, &dwSize, NULL);
-
-				DevIntfDetailData = (PSP_DEVICE_INTERFACE_DETAIL_DATA)HeapAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY,
-																				dwSize);
-				DevIntfDetailData->cbSize = sizeof(SP_DEVICE_INTERFACE_DETAIL_DATA);
-
-				if (SetupDiGetDeviceInterfaceDetail(hDevInfo, &DevIntfData,
-													DevIntfDetailData, dwSize, &dwSize, &DevData))
-				{
-					HANDLE hHIDDevice = OpenDevice(DevIntfDetailData->DevicePath);
-					if (hHIDDevice != INVALID_HANDLE_VALUE)
-					{
-						auto device = CheckAndCreateDevice(DevIntfDetailData->DevicePath, hHIDDevice);
-						if (device != nullptr)
-						{
-							if (IsDeviceWhitelisted(device->m_vendorId, device->m_productId))
-							{
-								if (!AttachDevice(device))
-								{
-									cemuLog_log(LogType::Force,
-												"nsyshid::BackendWindowsHID: failed to attach device: {:04x}:{:04x}",
-												device->m_vendorId,
-												device->m_productId);
-								}
-							}
-						}
-						CloseHandle(hHIDDevice);
-					}
-				}
-				HeapFree(GetProcessHeap(), 0, DevIntfDetailData);
-				// next
-				SetupDiEnumDeviceInterfaces(hDevInfo, NULL, &GUID_DEVINTERFACE_HID, ++dwMemberIdx, &DevIntfData);
-			}
-			SetupDiDestroyDeviceInfoList(hDevInfo);
-		}
-	}
-
-	BackendWindowsHID::~BackendWindowsHID()
-	{
-	}
-
-	bool BackendWindowsHID::IsInitialisedOk()
-	{
-		return true;
-	}
-
-	std::shared_ptr<Device> BackendWindowsHID::CheckAndCreateDevice(wchar_t* devicePath, HANDLE hDevice)
-	{
-		HIDD_ATTRIBUTES hidAttr;
-		hidAttr.Size = sizeof(HIDD_ATTRIBUTES);
-		if (HidD_GetAttributes(hDevice, &hidAttr) == FALSE)
-			return nullptr;
-
-		auto device = std::make_shared<DeviceWindowsHID>(hidAttr.VendorID,
-														 hidAttr.ProductID,
-														 1,
-														 2,
-														 0,
-														 _wcsdup(devicePath));
-		// get additional device info
-		sint32 maxPacketInputLength = -1;
-		sint32 maxPacketOutputLength = -1;
-		PHIDP_PREPARSED_DATA ppData = nullptr;
-		if (HidD_GetPreparsedData(hDevice, &ppData))
-		{
-			HIDP_CAPS caps;
-			if (HidP_GetCaps(ppData, &caps) == HIDP_STATUS_SUCCESS)
-			{
-				// length includes the report id byte
-				maxPacketInputLength = caps.InputReportByteLength - 1;
-				maxPacketOutputLength = caps.OutputReportByteLength - 1;
-			}
-			HidD_FreePreparsedData(ppData);
-		}
-		if (maxPacketInputLength <= 0 || maxPacketInputLength >= 0xF000)
-		{
-			cemuLog_logDebug(LogType::Force, "HID: Input packet length not available or out of range (length = {})", maxPacketInputLength);
-			maxPacketInputLength = 0x20;
-		}
-		if (maxPacketOutputLength <= 0 || maxPacketOutputLength >= 0xF000)
-		{
-			cemuLog_logDebug(LogType::Force, "HID: Output packet length not available or out of range (length = {})", maxPacketOutputLength);
-			maxPacketOutputLength = 0x20;
-		}
-
-		device->m_maxPacketSizeRX = maxPacketInputLength;
-		device->m_maxPacketSizeTX = maxPacketOutputLength;
-
-		return device;
-	}
-
-	DeviceWindowsHID::DeviceWindowsHID(uint16 vendorId,
-									   uint16 productId,
-									   uint8 interfaceIndex,
-									   uint8 interfaceSubClass,
-									   uint8 protocol,
-									   wchar_t* devicePath)
-		: Device(vendorId,
-				 productId,
-				 interfaceIndex,
-				 interfaceSubClass,
-				 protocol),
-		  m_devicePath(devicePath),
-		  m_hFile(INVALID_HANDLE_VALUE)
-	{
-	}
-
-	DeviceWindowsHID::~DeviceWindowsHID()
-	{
-		if (m_hFile != INVALID_HANDLE_VALUE)
-		{
-			CloseHandle(m_hFile);
-			m_hFile = INVALID_HANDLE_VALUE;
-		}
-	}
-
-	bool DeviceWindowsHID::Open()
-	{
-		if (IsOpened())
-		{
-			return true;
-		}
-		m_hFile = OpenDevice(m_devicePath);
-		if (m_hFile == INVALID_HANDLE_VALUE)
-		{
-			return false;
-		}
-		HidD_SetNumInputBuffers(m_hFile, 2); // don't cache too many reports
-		return true;
-	}
-
-	void DeviceWindowsHID::Close()
-	{
-		if (m_hFile != INVALID_HANDLE_VALUE)
-		{
-			CloseHandle(m_hFile);
-			m_hFile = INVALID_HANDLE_VALUE;
-		}
-	}
-
-	bool DeviceWindowsHID::IsOpened()
-	{
-		return m_hFile != INVALID_HANDLE_VALUE;
-	}
-
-	Device::ReadResult DeviceWindowsHID::Read(ReadMessage* message)
-	{
-		message->bytesRead = 0;
-		DWORD bt;
-		OVERLAPPED ovlp = {0};
-		ovlp.hEvent = CreateEvent(NULL, TRUE, FALSE, NULL);
-
-		uint8* tempBuffer = (uint8*)malloc(message->length + 1);
-		sint32 transferLength = 0; // minus report byte
-
-		_debugPrintHex("HID_READ_BEFORE", message->data, message->length);
-
-		cemuLog_logDebug(LogType::Force, "HidRead Begin (Length 0x{:08x})", message->length);
-		BOOL readResult = ReadFile(this->m_hFile, tempBuffer, message->length + 1, &bt, &ovlp);
-		if (readResult != FALSE)
-		{
-			// sometimes we get the result immediately
-			if (bt == 0)
-				transferLength = 0;
-			else
-				transferLength = bt - 1;
-			cemuLog_logDebug(LogType::Force, "HidRead Result received immediately (error 0x{:08x}) Length 0x{:08x}",
-							 GetLastError(), transferLength);
-		}
-		else
-		{
-			// wait for result
-			cemuLog_logDebug(LogType::Force, "HidRead WaitForResult (error 0x{:08x})", GetLastError());
-			// async hid read is never supposed to return unless there is a response? Lego Dimensions stops HIDRead calls as soon as one of them fails with a non-zero error (which includes time out)
-			DWORD r = WaitForSingleObject(ovlp.hEvent, 2000 * 100);
-			if (r == WAIT_TIMEOUT)
-			{
-				cemuLog_logDebug(LogType::Force, "HidRead internal timeout (error 0x{:08x})", GetLastError());
-				// return -108 in case of timeout
-				free(tempBuffer);
-				CloseHandle(ovlp.hEvent);
-				return ReadResult::ErrorTimeout;
-			}
-
-			cemuLog_logDebug(LogType::Force, "HidRead WaitHalfComplete");
-			GetOverlappedResult(this->m_hFile, &ovlp, &bt, false);
-			if (bt == 0)
-				transferLength = 0;
-			else
-				transferLength = bt - 1;
-			cemuLog_logDebug(LogType::Force, "HidRead WaitComplete Length: 0x{:08x}", transferLength);
-		}
-		sint32 returnCode = 0;
-		ReadResult result = ReadResult::Success;
-		if (bt != 0)
-		{
-			memcpy(message->data, tempBuffer + 1, transferLength);
-			sint32 hidReadLength = transferLength;
-
-			char debugOutput[1024] = {0};
-			for (sint32 i = 0; i < transferLength; i++)
-			{
-				sprintf(debugOutput + i * 3, "%02x ", tempBuffer[1 + i]);
-			}
-			cemuLog_logDebug(LogType::Force, "HIDRead data: {}", debugOutput);
-
-			message->bytesRead = transferLength;
-			result = ReadResult::Success;
-		}
-		else
-		{
-			cemuLog_log(LogType::Force, "Failed HID read");
-			result = ReadResult::Error;
-		}
-		free(tempBuffer);
-		CloseHandle(ovlp.hEvent);
-		return result;
-	}
-
-	Device::WriteResult DeviceWindowsHID::Write(WriteMessage* message)
-	{
-		message->bytesWritten = 0;
-		DWORD bt;
-		OVERLAPPED ovlp = {0};
-		ovlp.hEvent = CreateEvent(NULL, TRUE, FALSE, NULL);
-
-		uint8* tempBuffer = (uint8*)malloc(message->length + 1);
-		memcpy(tempBuffer + 1, message->data, message->length);
-		tempBuffer[0] = 0; // report byte?
-
-		cemuLog_logDebug(LogType::Force, "HidWrite Begin (Length 0x{:08x})", message->length);
-		BOOL writeResult = WriteFile(this->m_hFile, tempBuffer, message->length + 1, &bt, &ovlp);
-		if (writeResult != FALSE)
-		{
-			// sometimes we get the result immediately
-			cemuLog_logDebug(LogType::Force, "HidWrite Result received immediately (error 0x{:08x}) Length 0x{:08x}",
-							 GetLastError());
-		}
-		else
-		{
-			// wait for result
-			cemuLog_logDebug(LogType::Force, "HidWrite WaitForResult (error 0x{:08x})", GetLastError());
-			// todo - check for error type
-			DWORD r = WaitForSingleObject(ovlp.hEvent, 2000);
-			if (r == WAIT_TIMEOUT)
-			{
-				cemuLog_logDebug(LogType::Force, "HidWrite internal timeout");
-				// return -108 in case of timeout
-				free(tempBuffer);
-				CloseHandle(ovlp.hEvent);
-				return WriteResult::ErrorTimeout;
-			}
-
-			cemuLog_logDebug(LogType::Force, "HidWrite WaitHalfComplete");
-			GetOverlappedResult(this->m_hFile, &ovlp, &bt, false);
-			cemuLog_logDebug(LogType::Force, "HidWrite WaitComplete");
-		}
-
-		free(tempBuffer);
-		CloseHandle(ovlp.hEvent);
-
-		if (bt != 0)
-		{
-			message->bytesWritten = message->length;
-			return WriteResult::Success;
-		}
-		return WriteResult::Error;
-	}
-
-	bool DeviceWindowsHID::GetDescriptor(uint8 descType,
-										 uint8 descIndex,
-										 uint8 lang,
-										 uint8* output,
-										 uint32 outputMaxLength)
-	{
-		if (!IsOpened())
-		{
-			cemuLog_logDebug(LogType::Force, "nsyshid::DeviceWindowsHID::getDescriptor(): device is not opened");
-			return false;
-		}
-		if (descType == 0x02)
-		{
-			uint8 configurationDescriptor[0x29];
-
-			uint8* currentWritePtr;
-
-			// configuration descriptor
-			currentWritePtr = configurationDescriptor + 0;
-			*(uint8*)(currentWritePtr + 0) = 9;			// bLength
-			*(uint8*)(currentWritePtr + 1) = 2;			// bDescriptorType
-			*(uint16be*)(currentWritePtr + 2) = 0x0029; // wTotalLength
-			*(uint8*)(currentWritePtr + 4) = 1;			// bNumInterfaces
-			*(uint8*)(currentWritePtr + 5) = 1;			// bConfigurationValue
-			*(uint8*)(currentWritePtr + 6) = 0;			// iConfiguration
-			*(uint8*)(currentWritePtr + 7) = 0x80;		// bmAttributes
-			*(uint8*)(currentWritePtr + 8) = 0xFA;		// MaxPower
-			currentWritePtr = currentWritePtr + 9;
-			// configuration descriptor
-			*(uint8*)(currentWritePtr + 0) = 9;	   // bLength
-			*(uint8*)(currentWritePtr + 1) = 0x04; // bDescriptorType
-			*(uint8*)(currentWritePtr + 2) = 0;	   // bInterfaceNumber
-			*(uint8*)(currentWritePtr + 3) = 0;	   // bAlternateSetting
-			*(uint8*)(currentWritePtr + 4) = 2;	   // bNumEndpoints
-			*(uint8*)(currentWritePtr + 5) = 3;	   // bInterfaceClass
-			*(uint8*)(currentWritePtr + 6) = 0;	   // bInterfaceSubClass
-			*(uint8*)(currentWritePtr + 7) = 0;	   // bInterfaceProtocol
-			*(uint8*)(currentWritePtr + 8) = 0;	   // iInterface
-			currentWritePtr = currentWritePtr + 9;
-			// configuration descriptor
-			*(uint8*)(currentWritePtr + 0) = 9;			// bLength
-			*(uint8*)(currentWritePtr + 1) = 0x21;		// bDescriptorType
-			*(uint16be*)(currentWritePtr + 2) = 0x0111; // bcdHID
-			*(uint8*)(currentWritePtr + 4) = 0x00;		// bCountryCode
-			*(uint8*)(currentWritePtr + 5) = 0x01;		// bNumDescriptors
-			*(uint8*)(currentWritePtr + 6) = 0x22;		// bDescriptorType
-			*(uint16be*)(currentWritePtr + 7) = 0x001D; // wDescriptorLength
-			currentWritePtr = currentWritePtr + 9;
-			// endpoint descriptor 1
-			*(uint8*)(currentWritePtr + 0) = 7;	   // bLength
-			*(uint8*)(currentWritePtr + 1) = 0x05; // bDescriptorType
-			*(uint8*)(currentWritePtr + 2) = 0x81; // bEndpointAddress
-			*(uint8*)(currentWritePtr + 3) = 0x03; // bmAttributes
-			*(uint16be*)(currentWritePtr + 4) =
-				this->m_maxPacketSizeRX;		   // wMaxPacketSize
-			*(uint8*)(currentWritePtr + 6) = 0x01; // bInterval
-			currentWritePtr = currentWritePtr + 7;
-			// endpoint descriptor 2
-			*(uint8*)(currentWritePtr + 0) = 7;	   // bLength
-			*(uint8*)(currentWritePtr + 1) = 0x05; // bDescriptorType
-			*(uint8*)(currentWritePtr + 2) = 0x02; // bEndpointAddress
-			*(uint8*)(currentWritePtr + 3) = 0x03; // bmAttributes
-			*(uint16be*)(currentWritePtr + 4) =
-				this->m_maxPacketSizeTX;		   // wMaxPacketSize
-			*(uint8*)(currentWritePtr + 6) = 0x01; // bInterval
-			currentWritePtr = currentWritePtr + 7;
-
-			cemu_assert_debug((currentWritePtr - configurationDescriptor) == 0x29);
-
-			memcpy(output, configurationDescriptor,
-				   std::min<uint32>(outputMaxLength, sizeof(configurationDescriptor)));
-			return true;
-		}
-		else
-		{
-			cemu_assert_unimplemented();
-		}
-		return false;
-	}
-
-	bool DeviceWindowsHID::SetProtocol(uint8 ifIndex, uint8 protocol)
-	{
-		// ToDo: implement this
-		// pretend that everything is fine
-		return true;
-	}
-
-	bool DeviceWindowsHID::SetReport(ReportMessage* message)
-	{
-		sint32 retryCount = 0;
-		while (true)
-		{
-			BOOL r = HidD_SetOutputReport(this->m_hFile, message->reportData, message->length);
-			if (r != FALSE)
-				break;
-			Sleep(20); // retry
-			retryCount++;
-			if (retryCount >= 50)
-			{
-				cemuLog_log(LogType::Force, "nsyshid::DeviceWindowsHID::SetReport(): HID SetReport failed");
-				return false;
-			}
-		}
-		return true;
-	}
-
-	HANDLE OpenDevice(wchar_t* devicePath)
-	{
-		return CreateFile(devicePath,
-						  GENERIC_READ | GENERIC_WRITE,
-						  FILE_SHARE_READ |
-							  FILE_SHARE_WRITE,
-						  NULL,
-						  OPEN_EXISTING,
-						  FILE_FLAG_OVERLAPPED,
-						  NULL);
-	}
-
-	void _debugPrintHex(std::string prefix, uint8* data, size_t len)
-	{
-		char debugOutput[1024] = {0};
-		len = std::min(len, (size_t)100);
-		for (sint32 i = 0; i < len; i++)
-		{
-			sprintf(debugOutput + i * 3, "%02x ", data[i]);
-		}
-		cemuLog_logDebug(LogType::Force, "[{}] Data: {}", prefix, debugOutput);
-	}
-} // namespace nsyshid::backend::windows
-
-#endif // NSYSHID_ENABLE_BACKEND_WINDOWS_HID

src/Cafe/OS/libs/nsyshid/BackendWindowsHID.h

@@ -1,66 +0,0 @@
-#ifndef CEMU_NSYSHID_BACKEND_WINDOWS_HID_H
-#define CEMU_NSYSHID_BACKEND_WINDOWS_HID_H
-
-#include "nsyshid.h"
-
-#if NSYSHID_ENABLE_BACKEND_WINDOWS_HID
-
-#include "Backend.h"
-
-namespace nsyshid::backend::windows
-{
-	class BackendWindowsHID : public nsyshid::Backend {
-	  public:
-		BackendWindowsHID();
-
-		~BackendWindowsHID();
-
-		bool IsInitialisedOk() override;
-
-	  protected:
-		void AttachVisibleDevices() override;
-
-	  private:
-		std::shared_ptr<Device> CheckAndCreateDevice(wchar_t* devicePath, HANDLE hDevice);
-	};
-
-	class DeviceWindowsHID : public nsyshid::Device {
-	  public:
-		DeviceWindowsHID(uint16 vendorId,
-						 uint16 productId,
-						 uint8 interfaceIndex,
-						 uint8 interfaceSubClass,
-						 uint8 protocol,
-						 wchar_t* devicePath);
-
-		~DeviceWindowsHID();
-
-		bool Open() override;
-
-		void Close() override;
-
-		bool IsOpened() override;
-
-		ReadResult Read(ReadMessage* message) override;
-
-		WriteResult Write(WriteMessage* message) override;
-
-		bool GetDescriptor(uint8 descType, uint8 descIndex, uint8 lang, uint8* output, uint32 outputMaxLength) override;
-
-		bool SetProtocol(uint8 ifIndex, uint8 protocol) override;
-
-		bool SetReport(ReportMessage* message) override;
-
-	  private:
-		wchar_t* m_devicePath;
-		HANDLE m_hFile;
-	};
-
-	HANDLE OpenDevice(wchar_t* devicePath);
-
-	void _debugPrintHex(std::string prefix, uint8* data, size_t len);
-} // namespace nsyshid::backend::windows
-
-#endif // NSYSHID_ENABLE_BACKEND_WINDOWS_HID
-
-#endif // CEMU_NSYSHID_BACKEND_WINDOWS_HID_H

src/Cafe/OS/libs/nsyshid/Dimensions.h

@@ -0,0 +1,112 @@
+#include <mutex>
+
+#include "nsyshid.h"
+#include "Backend.h"
+
+#include "Common/FileStream.h"
+
+namespace nsyshid
+{
+	class DimensionsToypadDevice final : public Device
+	{
+	  public:
+		DimensionsToypadDevice();
+		~DimensionsToypadDevice() = default;
+
+		bool Open() override;
+
+		void Close() override;
+
+		bool IsOpened() override;
+
+		ReadResult Read(ReadMessage* message) override;
+
+		WriteResult Write(WriteMessage* message) override;
+
+		bool GetDescriptor(uint8 descType,
+						   uint8 descIndex,
+						   uint16 lang,
+						   uint8* output,
+						   uint32 outputMaxLength) override;
+
+		bool SetIdle(uint8 ifIndex,
+					 uint8 reportId,
+					 uint8 duration) override;
+
+		bool SetProtocol(uint8 ifIndex, uint8 protocol) override;
+
+		bool SetReport(ReportMessage* message) override;
+
+	  private:
+		bool m_IsOpened;
+	};
+
+	class DimensionsUSB
+	{
+	  public:
+		struct DimensionsMini final
+		{
+			std::unique_ptr<FileStream> dimFile;
+			std::array<uint8, 0x2D * 0x04> data{};
+			uint8 index = 255;
+			uint8 pad = 255;
+			uint32 id = 0;
+			void Save();
+		};
+
+		void SendCommand(std::span<const uint8, 32> buf);
+		std::array<uint8, 32> GetStatus();
+
+		void GenerateRandomNumber(std::span<const uint8, 8> buf, uint8 sequence,
+								  std::array<uint8, 32>& replyBuf);
+		void InitializeRNG(uint32 seed);
+		void GetChallengeResponse(std::span<const uint8, 8> buf, uint8 sequence,
+								  std::array<uint8, 32>& replyBuf);
+		void QueryBlock(uint8 index, uint8 page, std::array<uint8, 32>& replyBuf,
+						uint8 sequence);
+		void WriteBlock(uint8 index, uint8 page, std::span<const uint8, 4> toWriteBuf, std::array<uint8, 32>& replyBuf,
+						uint8 sequence);
+		void GetModel(std::span<const uint8, 8> buf, uint8 sequence,
+					  std::array<uint8, 32>& replyBuf);
+
+		bool RemoveFigure(uint8 pad, uint8 index, bool fullRemove);
+		bool TempRemove(uint8 index);
+		bool CancelRemove(uint8 index);
+		uint32 LoadFigure(const std::array<uint8, 0x2D * 0x04>& buf, std::unique_ptr<FileStream> file, uint8 pad, uint8 index);
+		bool CreateFigure(fs::path pathName, uint32 id);
+		bool MoveFigure(uint8 pad, uint8 index, uint8 oldPad, uint8 oldIndex);
+		static std::map<const uint32, const char*> GetListMinifigs();
+		static std::map<const uint32, const char*> GetListTokens();
+		std::string FindFigure(uint32 figNum);
+
+	  protected:
+		std::mutex m_dimensionsMutex;
+		std::array<DimensionsMini, 7> m_figures{};
+
+	  private:
+		void RandomUID(std::array<uint8, 0x2D * 0x04>& uidBuffer);
+		uint8 GenerateChecksum(const std::array<uint8, 32>& data,
+							   int numOfBytes) const;
+		std::array<uint8, 8> Decrypt(std::span<const uint8, 8> buf, std::optional<std::array<uint8, 16>> key);
+		std::array<uint8, 8> Encrypt(std::span<const uint8, 8> buf, std::optional<std::array<uint8, 16>> key);
+		std::array<uint8, 16> GenerateFigureKey(const std::array<uint8, 0x2D * 0x04>& uid);
+		std::array<uint8, 4> PWDGenerate(const std::array<uint8, 0x2D * 0x04>& uid);
+		std::array<uint8, 4> DimensionsRandomize(const std::vector<uint8> key, uint8 count);
+		uint32 GetFigureId(const std::array<uint8, 0x2D * 0x04>& buf);
+		uint32 Scramble(const std::array<uint8, 7>& uid, uint8 count);
+		uint32 GetNext();
+		DimensionsMini& GetFigureByIndex(uint8 index);
+
+		uint32 m_randomA;
+		uint32 m_randomB;
+		uint32 m_randomC;
+		uint32 m_randomD;
+
+		bool m_isAwake = false;
+
+		std::queue<std::array<uint8, 32>> m_figureAddedRemovedResponses;
+		std::queue<std::array<uint8, 32>> m_queries;
+	};
+	extern DimensionsUSB g_dimensionstoypad;
+
+} // namespace nsyshid

src/Cafe/OS/libs/nsyshid/Infinity.cpp

@@ -387,7 +387,7 @@ namespace nsyshid
 
 	bool InfinityBaseDevice::GetDescriptor(uint8 descType,
 										   uint8 descIndex,
-										   uint8 lang,
+										   uint16 lang,
 										   uint8* output,
 										   uint32 outputMaxLength)
 	{
@@ -450,6 +450,13 @@ namespace nsyshid
 		return true;
 	}
 
+	bool InfinityBaseDevice::SetIdle(uint8 ifIndex,
+										 uint8 reportId,
+										 uint8 duration)
+	{
+		return true;
+	}
+
 	bool InfinityBaseDevice::SetProtocol(uint8 ifIndex, uint8 protocol)
 	{
 		return true;
@@ -492,7 +499,7 @@ namespace nsyshid
 		return response;
 	}
 
-	void InfinityUSB::SendCommand(uint8* buf, sint32 originalLength)
+	void InfinityUSB::SendCommand(uint8* buf, uint32 length)
 	{
 		const uint8 command = buf[2];
 		const uint8 sequence = buf[3];

src/Cafe/OS/libs/nsyshid/Infinity.h

@@ -26,10 +26,14 @@ namespace nsyshid
 
 		bool GetDescriptor(uint8 descType,
 						   uint8 descIndex,
-						   uint8 lang,
+						   uint16 lang,
 						   uint8* output,
 						   uint32 outputMaxLength) override;
 
+		bool SetIdle(uint8 ifIndex,
+					 uint8 reportId,
+					 uint8 duration) override;
+
 		bool SetProtocol(uint8 ifIndex, uint8 protocol) override;
 
 		bool SetReport(ReportMessage* message) override;
@@ -53,7 +57,7 @@ namespace nsyshid
 			void Save();
 		};
 
-		void SendCommand(uint8* buf, sint32 originalLength);
+		void SendCommand(uint8* buf, uint32 length);
 		std::array<uint8, 32> GetStatus();
 
 		void GetBlankResponse(uint8 sequence, std::array<uint8, 32>& replyBuf);

src/Cafe/OS/libs/nsyshid/Skylander.cpp

@@ -564,7 +564,7 @@ namespace nsyshid
 
 	bool SkylanderPortalDevice::GetDescriptor(uint8 descType,
 											  uint8 descIndex,
-											  uint8 lang,
+											  uint16 lang,
 											  uint8* output,
 											  uint32 outputMaxLength)
 	{
@@ -583,7 +583,7 @@ namespace nsyshid
 		*(uint8*)(currentWritePtr + 7) = 0x80;		// bmAttributes
 		*(uint8*)(currentWritePtr + 8) = 0xFA;		// MaxPower
 		currentWritePtr = currentWritePtr + 9;
-		// configuration descriptor
+		// interface descriptor
 		*(uint8*)(currentWritePtr + 0) = 9;	   // bLength
 		*(uint8*)(currentWritePtr + 1) = 0x04; // bDescriptorType
 		*(uint8*)(currentWritePtr + 2) = 0;	   // bInterfaceNumber
@@ -594,7 +594,7 @@ namespace nsyshid
 		*(uint8*)(currentWritePtr + 7) = 0;	   // bInterfaceProtocol
 		*(uint8*)(currentWritePtr + 8) = 0;	   // iInterface
 		currentWritePtr = currentWritePtr + 9;
-		// configuration descriptor
+		// HID descriptor
 		*(uint8*)(currentWritePtr + 0) = 9;			// bLength
 		*(uint8*)(currentWritePtr + 1) = 0x21;		// bDescriptorType
 		*(uint16be*)(currentWritePtr + 2) = 0x0111; // bcdHID
@@ -608,7 +608,7 @@ namespace nsyshid
 		*(uint8*)(currentWritePtr + 1) = 0x05;	  // bDescriptorType
 		*(uint8*)(currentWritePtr + 2) = 0x81;	  // bEndpointAddress
 		*(uint8*)(currentWritePtr + 3) = 0x03;	  // bmAttributes
-		*(uint16be*)(currentWritePtr + 4) = 0x40; // wMaxPacketSize
+		*(uint16be*)(currentWritePtr + 4) = 0x0040; // wMaxPacketSize
 		*(uint8*)(currentWritePtr + 6) = 0x01;	  // bInterval
 		currentWritePtr = currentWritePtr + 7;
 		// endpoint descriptor 2
@@ -616,7 +616,7 @@ namespace nsyshid
 		*(uint8*)(currentWritePtr + 1) = 0x05;	  // bDescriptorType
 		*(uint8*)(currentWritePtr + 2) = 0x02;	  // bEndpointAddress
 		*(uint8*)(currentWritePtr + 3) = 0x03;	  // bmAttributes
-		*(uint16be*)(currentWritePtr + 4) = 0x40; // wMaxPacketSize
+		*(uint16be*)(currentWritePtr + 4) = 0x0040; // wMaxPacketSize
 		*(uint8*)(currentWritePtr + 6) = 0x01;	  // bInterval
 		currentWritePtr = currentWritePtr + 7;
 
@@ -627,6 +627,13 @@ namespace nsyshid
 		return true;
 	}
 
+	bool SkylanderPortalDevice::SetIdle(uint8 ifIndex,
+									 uint8 reportId,
+									 uint8 duration)
+	{
+		return true;
+	}
+
 	bool SkylanderPortalDevice::SetProtocol(uint8 ifIndex, uint8 protocol)
 	{
 		return true;
@@ -634,12 +641,12 @@ namespace nsyshid
 
 	bool SkylanderPortalDevice::SetReport(ReportMessage* message)
 	{
-		g_skyportal.ControlTransfer(message->originalData, message->originalLength);
+		g_skyportal.ControlTransfer(message->data, message->length);
 		std::this_thread::sleep_for(std::chrono::milliseconds(1));
 		return true;
 	}
 
-	void SkylanderUSB::ControlTransfer(uint8* buf, sint32 originalLength)
+	void SkylanderUSB::ControlTransfer(uint8* buf, uint32 length)
 	{
 		std::array<uint8, 64> interruptResponse = {};
 		switch (buf[0])

src/Cafe/OS/libs/nsyshid/Skylander.h

@@ -26,10 +26,14 @@ namespace nsyshid
 
 		bool GetDescriptor(uint8 descType,
 						   uint8 descIndex,
-						   uint8 lang,
+						   uint16 lang,
 						   uint8* output,
 						   uint32 outputMaxLength) override;
 
+		bool SetIdle(uint8 ifIndex,
+					 uint8 reportId,
+					 uint8 duration) override;
+
 		bool SetProtocol(uint8 ifIndex, uint8 protocol) override;
 
 		bool SetReport(ReportMessage* message) override;
@@ -70,7 +74,7 @@ namespace nsyshid
 			uint8 blue = 0;
 		};
 
-		void ControlTransfer(uint8* buf, sint32 originalLength);
+		void ControlTransfer(uint8* buf, uint32 length);
 
 		void Activate();
 		void Deactivate();

src/Cafe/OS/libs/nsyshid/nsyshid.cpp

@@ -305,47 +305,37 @@ namespace nsyshid
 		osLib_returnFromFunction(hCPU, 0);
 	}
 
-	void export_HIDGetDescriptor(PPCInterpreter_t* hCPU)
+	void _debugPrintHex(const std::string prefix, const uint8* data, size_t size)
 	{
-		ppcDefineParamU32(hidHandle, 0);	   // r3
-		ppcDefineParamU8(descType, 1);		   // r4
-		ppcDefineParamU8(descIndex, 2);		   // r5
-		ppcDefineParamU8(lang, 3);			   // r6
-		ppcDefineParamUStr(output, 4);		   // r7
-		ppcDefineParamU32(outputMaxLength, 5); // r8
-		ppcDefineParamMPTR(cbFuncMPTR, 6);	   // r9
-		ppcDefineParamMPTR(cbParamMPTR, 7);	   // r10
+		constexpr size_t BYTES_PER_LINE = 16;
 
-		int returnValue = -1;
-		std::shared_ptr<Device> device = GetDeviceByHandle(hidHandle, true);
-		if (device)
+		std::string out;
+		for (size_t row_start = 0; row_start < size; row_start += BYTES_PER_LINE)
 		{
-			memset(output, 0, outputMaxLength);
-			if (device->GetDescriptor(descType, descIndex, lang, output, outputMaxLength))
+			out += fmt::format("{:06x}: ", row_start);
+			for (size_t i = 0; i < BYTES_PER_LINE; ++i)
 			{
-				returnValue = 0;
+				if (row_start + i < size)
+				{
+					out += fmt::format("{:02x} ", data[row_start + i]);
+				}
+				else
+				{
+					out += "   ";
+				}
 			}
-			else
+			out += " ";
+			for (size_t i = 0; i < BYTES_PER_LINE; ++i)
 			{
-				returnValue = -1;
+				if (row_start + i < size)
+				{
+					char c = static_cast<char>(data[row_start + i]);
+					out += std::isprint(c, std::locale::classic()) ? c : '.';
+				}
 			}
+			out += "\n";
 		}
-		else
-		{
-			cemu_assert_suspicious();
-		}
-		osLib_returnFromFunction(hCPU, returnValue);
-	}
-
-	void _debugPrintHex(std::string prefix, uint8* data, size_t len)
-	{
-		char debugOutput[1024] = {0};
-		len = std::min(len, (size_t)100);
-		for (sint32 i = 0; i < len; i++)
-		{
-			sprintf(debugOutput + i * 3, "%02x ", data[i]);
-		}
-		cemuLog_logDebug(LogType::Force, "[{}] Data: {}", prefix, debugOutput);
+		cemuLog_logDebug(LogType::Force, "[{}] Data: \n{}", prefix, out);
 	}
 
 	void DoHIDTransferCallback(MPTR callbackFuncMPTR, MPTR callbackParamMPTR, uint32 hidHandle, uint32 errorCode,
@@ -354,26 +344,152 @@ namespace nsyshid
 		coreinitAsyncCallback_add(callbackFuncMPTR, 5, hidHandle, errorCode, buffer, length, callbackParamMPTR);
 	}
 
-	void export_HIDSetIdle(PPCInterpreter_t* hCPU)
+	void _hidGetDescriptorAsync(std::shared_ptr<Device> device, uint8 descType, uint8 descIndex, uint16 lang, uint8* output, uint32 outputMaxLength, MPTR callbackFuncMPTR, MPTR callbackParamMPTR)
 	{
-		ppcDefineParamU32(hidHandle, 0);		  // r3
-		ppcDefineParamU32(ifIndex, 1);			  // r4
-		ppcDefineParamU32(ukn, 2);				  // r5
-		ppcDefineParamU32(duration, 3);			  // r6
-		ppcDefineParamMPTR(callbackFuncMPTR, 4);  // r7
-		ppcDefineParamMPTR(callbackParamMPTR, 5); // r8
-		cemuLog_logDebug(LogType::Force, "nsyshid.HIDSetIdle(...)");
-
-		// todo
-		if (callbackFuncMPTR)
+		if (device->GetDescriptor(descType, descIndex, lang, output, outputMaxLength))
 		{
-			DoHIDTransferCallback(callbackFuncMPTR, callbackParamMPTR, hidHandle, 0, MPTR_NULL, 0);
+			DoHIDTransferCallback(callbackFuncMPTR,
+								  callbackParamMPTR,
+								  device->m_hid->handle,
+								  0,
+								  0,
+								  0);
 		}
 		else
 		{
-			cemu_assert_unimplemented();
+			DoHIDTransferCallback(callbackFuncMPTR,
+								  callbackParamMPTR,
+								  device->m_hid->handle,
+								  -1,
+								  0,
+								  0);
+		}
+	}
+
+	void export_HIDGetDescriptor(PPCInterpreter_t* hCPU)
+	{
+		ppcDefineParamU32(hidHandle, 0);	   // r3
+		ppcDefineParamU8(descType, 1);		   // r4
+		ppcDefineParamU8(descIndex, 2);		   // r5
+		ppcDefineParamU16(lang, 3);			   // r6
+		ppcDefineParamUStr(output, 4);		   // r7
+		ppcDefineParamU32(outputMaxLength, 5); // r8
+		ppcDefineParamMPTR(cbFuncMPTR, 6);	   // r9
+		ppcDefineParamMPTR(cbParamMPTR, 7);	   // r10
+		cemuLog_logDebug(LogType::Force, "nsyshid.HIDGetDescriptor(0x{:08x}, 0x{:02x}, 0x{:02x}, 0x{:04x}, 0x{:x}, 0x{:08x}, 0x{:08x}, 0x{:08x})",
+					hCPU->gpr[3], hCPU->gpr[4], hCPU->gpr[5], hCPU->gpr[6], hCPU->gpr[7], hCPU->gpr[8], hCPU->gpr[9], hCPU->gpr[10]);
+
+		std::shared_ptr<Device> device = GetDeviceByHandle(hidHandle, true);
+		if (device == nullptr)
+		{
+			cemuLog_log(LogType::Force, "nsyshid.HIDGetDescriptor(): Unable to find device with hid handle {}", hidHandle);
+			osLib_returnFromFunction(hCPU, -1);
+			return;
+		}
+
+		// issue request (synchronous or asynchronous)
+		sint32 returnCode = 0;
+		if (cbFuncMPTR == MPTR_NULL)
+		{
+			// synchronous
+			returnCode = -1;
+			if (device->GetDescriptor(descType, descIndex, lang, output, outputMaxLength))
+			{
+				returnCode = outputMaxLength;
+			}
+		}
+		else
+		{
+			// asynchronous
+			std::thread(&_hidGetDescriptorAsync, device, descType, descIndex, lang, output, outputMaxLength, cbFuncMPTR, cbParamMPTR)
+				.detach();
+			returnCode = 0;
+		}
+		osLib_returnFromFunction(hCPU, returnCode);
+	}
+
+	void _hidSetIdleAsync(std::shared_ptr<Device> device, uint8 ifIndex, uint8 reportId, uint8 duration, MPTR callbackFuncMPTR, MPTR callbackParamMPTR)
+	{
+		if (device->SetIdle(ifIndex, reportId, duration))
+		{
+			DoHIDTransferCallback(callbackFuncMPTR,
+								  callbackParamMPTR,
+								  device->m_hid->handle,
+								  0,
+								  0,
+								  0);
+		}
+		else
+		{
+			DoHIDTransferCallback(callbackFuncMPTR,
+								  callbackParamMPTR,
+								  device->m_hid->handle,
+								  -1,
+								  0,
+								  0);
+		}
+	}
+
+	void export_HIDSetIdle(PPCInterpreter_t* hCPU)
+	{
+		ppcDefineParamU32(hidHandle, 0);		  // r3
+		ppcDefineParamU8(ifIndex, 1);			  // r4
+		ppcDefineParamU8(reportId, 2);			  // r5
+		ppcDefineParamU8(duration, 3);			  // r6
+		ppcDefineParamMPTR(callbackFuncMPTR, 4);  // r7
+		ppcDefineParamMPTR(callbackParamMPTR, 5); // r8
+		cemuLog_logDebug(LogType::Force, "nsyshid.HIDSetIdle(0x{:08x}, 0x{:02x}, 0x{:02x}, 0x{:02x}, 0x{:08x}, 0x{:08x})", hCPU->gpr[3],
+					hCPU->gpr[4], hCPU->gpr[5], hCPU->gpr[6], hCPU->gpr[7], hCPU->gpr[8]);
+
+		std::shared_ptr<Device> device = GetDeviceByHandle(hidHandle, true);
+		if (device == nullptr)
+		{
+			cemuLog_log(LogType::Force, "nsyshid.HIDSetIdle(): Unable to find device with hid handle {}", hidHandle);
+			osLib_returnFromFunction(hCPU, -1);
+			return;
+		}
+
+		// issue request (synchronous or asynchronous)
+		sint32 returnCode = 0;
+		if (callbackFuncMPTR == MPTR_NULL)
+		{
+			// synchronous
+			returnCode = -1;
+			if (device->SetIdle(ifIndex, reportId, duration))
+			{
+				returnCode = 0;
+			}
+		}
+		else
+		{
+			// asynchronous
+			std::thread(&_hidSetIdleAsync, device, ifIndex, reportId, duration, callbackFuncMPTR, callbackParamMPTR)
+				.detach();
+			returnCode = 0;
+		}
+		osLib_returnFromFunction(hCPU, returnCode);
+	}
+
+	void _hidSetProtocolAsync(std::shared_ptr<Device> device, uint8 ifIndex, uint8 protocol, MPTR callbackFuncMPTR, MPTR callbackParamMPTR)
+	{
+		if (device->SetProtocol(ifIndex, protocol))
+		{
+			DoHIDTransferCallback(callbackFuncMPTR,
+								  callbackParamMPTR,
+								  device->m_hid->handle,
+								  0,
+								  0,
+								  0);
+		}
+		else
+		{
+			DoHIDTransferCallback(callbackFuncMPTR,
+								  callbackParamMPTR,
+								  device->m_hid->handle,
+								  -1,
+								  0,
+								  0);
 		}
-		osLib_returnFromFunction(hCPU, 0); // for non-async version, return number of bytes transferred
 	}
 
 	void export_HIDSetProtocol(PPCInterpreter_t* hCPU)
@@ -383,51 +499,51 @@ namespace nsyshid
 		ppcDefineParamU8(protocol, 2);			  // r5
 		ppcDefineParamMPTR(callbackFuncMPTR, 3);  // r6
 		ppcDefineParamMPTR(callbackParamMPTR, 4); // r7
-		cemuLog_logDebug(LogType::Force, "nsyshid.HIDSetProtocol(...)");
+		cemuLog_logDebug(LogType::Force, "nsyshid.HIDSetProtocol(0x{:08x}, 0x{:02x}, 0x{:02x}, 0x{:08x}, 0x{:08x})", hCPU->gpr[3],
+					hCPU->gpr[4], hCPU->gpr[5], hCPU->gpr[6], hCPU->gpr[7]);
 
 		std::shared_ptr<Device> device = GetDeviceByHandle(hidHandle, true);
-		sint32 returnCode = -1;
-		if (device)
+		if (device == nullptr)
 		{
-			if (!device->IsOpened())
+			cemuLog_log(LogType::Force, "nsyshid.HIDSetProtocol(): Unable to find device with hid handle {}", hidHandle);
+			osLib_returnFromFunction(hCPU, -1);
+			return;
+		}
+		// issue request (synchronous or asynchronous)
+		sint32 returnCode = 0;
+		if (callbackFuncMPTR == MPTR_NULL)
+		{
+			// synchronous
+			returnCode = -1;
+			if (device->SetProtocol(ifIndex, protocol))
 			{
-				cemuLog_logDebug(LogType::Force, "nsyshid.HIDSetProtocol(): error: device is not opened");
-			}
-			else
-			{
-				if (device->SetProtocol(ifIndex, protocol))
-				{
-					returnCode = 0;
-				}
+				returnCode = 0;
 			}
 		}
 		else
 		{
-			cemu_assert_suspicious();
-		}
-
-		if (callbackFuncMPTR)
-		{
-			DoHIDTransferCallback(callbackFuncMPTR, callbackParamMPTR, hidHandle, 0, MPTR_NULL, 0);
+			// asynchronous
+			std::thread(&_hidSetProtocolAsync, device, ifIndex, protocol, callbackFuncMPTR, callbackParamMPTR)
+				.detach();
+			returnCode = 0;
 		}
 		osLib_returnFromFunction(hCPU, returnCode);
 	}
 
 	// handler for async HIDSetReport transfers
-	void _hidSetReportAsync(std::shared_ptr<Device> device, uint8* reportData, sint32 length,
-							uint8* originalData,
-							sint32 originalLength, MPTR callbackFuncMPTR, MPTR callbackParamMPTR)
+	void _hidSetReportAsync(std::shared_ptr<Device> device, uint8 reportType, uint8 reportId, uint8* data, uint32 length,
+							MPTR callbackFuncMPTR, MPTR callbackParamMPTR)
 	{
 		cemuLog_logDebug(LogType::Force, "_hidSetReportAsync begin");
-		ReportMessage message(reportData, length, originalData, originalLength);
+		ReportMessage message(reportType, reportId, data, length);
 		if (device->SetReport(&message))
 		{
 			DoHIDTransferCallback(callbackFuncMPTR,
 								  callbackParamMPTR,
 								  device->m_hid->handle,
 								  0,
-								  memory_getVirtualOffsetFromPointer(originalData),
-								  originalLength);
+								  memory_getVirtualOffsetFromPointer(data),
+								  length);
 		}
 		else
 		{
@@ -435,24 +551,22 @@ namespace nsyshid
 								  callbackParamMPTR,
 								  device->m_hid->handle,
 								  -1,
-								  memory_getVirtualOffsetFromPointer(originalData),
-								  0);
+								  memory_getVirtualOffsetFromPointer(data),
+								  length);
 		}
-		free(reportData);
 	}
 
 	// handler for synchronous HIDSetReport transfers
-	sint32 _hidSetReportSync(std::shared_ptr<Device> device, uint8* reportData, sint32 length,
-							 uint8* originalData, sint32 originalLength, coreinit::OSEvent* event)
+	sint32 _hidSetReportSync(std::shared_ptr<Device> device, uint8 reportType, uint8 reportId,
+							 uint8* data, uint32 length, coreinit::OSEvent* event)
 	{
-		_debugPrintHex("_hidSetReportSync Begin", reportData, length);
+		_debugPrintHex("_hidSetReportSync Begin", data, length);
 		sint32 returnCode = 0;
-		ReportMessage message(reportData, length, originalData, originalLength);
+		ReportMessage message(reportType, reportId, data, length);
 		if (device->SetReport(&message))
 		{
-			returnCode = originalLength;
+			returnCode = length;
 		}
-		free(reportData);
 		cemuLog_logDebug(LogType::Force, "_hidSetReportSync end. returnCode: {}", returnCode);
 		coreinit::OSSignalEvent(event);
 		return returnCode;
@@ -461,19 +575,19 @@ namespace nsyshid
 	void export_HIDSetReport(PPCInterpreter_t* hCPU)
 	{
 		ppcDefineParamU32(hidHandle, 0);		  // r3
-		ppcDefineParamU32(reportRelatedUkn, 1);	  // r4
-		ppcDefineParamU32(reportId, 2);			  // r5
+		ppcDefineParamU8(reportType, 1);		  // r4
+		ppcDefineParamU8(reportId, 2);			  // r5
 		ppcDefineParamUStr(data, 3);			  // r6
 		ppcDefineParamU32(dataLength, 4);		  // r7
 		ppcDefineParamMPTR(callbackFuncMPTR, 5);  // r8
 		ppcDefineParamMPTR(callbackParamMPTR, 6); // r9
-		cemuLog_logDebug(LogType::Force, "nsyshid.HIDSetReport({},0x{:02x},0x{:02x},...)", hidHandle, reportRelatedUkn,
-						 reportId);
+		cemuLog_logDebug(LogType::Force, "nsyshid.HIDSetReport(0x{:08x}, 0x{:02x}, 0x{:02x}, 0x{:08x}, 0x{:08x}, 0x{:08x}, 0x{:08x})", hCPU->gpr[3],
+					hCPU->gpr[4], hCPU->gpr[5], hCPU->gpr[6], hCPU->gpr[7], hCPU->gpr[8], hCPU->gpr[9]);
 
 		_debugPrintHex("HIDSetReport", data, dataLength);
 
 #ifdef CEMU_DEBUG_ASSERT
-		if (reportRelatedUkn != 2 || reportId != 0)
+		if (reportType != 2 || reportId != 0)
 			assert_dbg();
 #endif
 
@@ -485,15 +599,6 @@ namespace nsyshid
 			return;
 		}
 
-		// prepare report data
-		// note: Currently we need to pad the data to 0x20 bytes for it to work (plus one extra byte for HidD_SetOutputReport)
-		// Does IOSU pad data to 0x20 byte? Also check if this is specific to Skylanders portal
-		sint32 paddedLength = (dataLength + 0x1F) & ~0x1F;
-		uint8* reportData = (uint8*)malloc(paddedLength + 1);
-		memset(reportData, 0, paddedLength + 1);
-		reportData[0] = 0;
-		memcpy(reportData + 1, data, dataLength);
-
 		// issue request (synchronous or asynchronous)
 		sint32 returnCode = 0;
 		if (callbackFuncMPTR == MPTR_NULL)
@@ -501,15 +606,14 @@ namespace nsyshid
 			// synchronous
 			StackAllocator<coreinit::OSEvent> event;
 			coreinit::OSInitEvent(&event, coreinit::OSEvent::EVENT_STATE::STATE_NOT_SIGNALED, coreinit::OSEvent::EVENT_MODE::MODE_AUTO);
-			std::future<sint32> res = std::async(std::launch::async, &_hidSetReportSync, device, reportData,
-												 paddedLength + 1, data, dataLength, &event);
+			std::future<sint32> res = std::async(std::launch::async, &_hidSetReportSync, device, reportType, reportId, data, dataLength, &event);
 			coreinit::OSWaitEvent(&event);
 			returnCode = res.get();
 		}
 		else
 		{
 			// asynchronous
-			std::thread(&_hidSetReportAsync, device, reportData, paddedLength + 1, data, dataLength,
+			std::thread(&_hidSetReportAsync, device, reportType, reportId, data, dataLength,
 						callbackFuncMPTR, callbackParamMPTR)
 				.detach();
 			returnCode = 0;
@@ -586,7 +690,7 @@ namespace nsyshid
 		ppcDefineParamMPTR(callbackFuncMPTR, 3);  // r6
 		ppcDefineParamMPTR(callbackParamMPTR, 4); // r7
 		cemuLog_logDebug(LogType::Force, "nsyshid.HIDRead(0x{:x},0x{:08x},0x{:08x},0x{:08x},0x{:08x})", hCPU->gpr[3],
-						 hCPU->gpr[4], hCPU->gpr[5], hCPU->gpr[6], hCPU->gpr[7]);
+					hCPU->gpr[4], hCPU->gpr[5], hCPU->gpr[6], hCPU->gpr[7]);
 
 		std::shared_ptr<Device> device = GetDeviceByHandle(hidHandle, true);
 		if (device == nullptr)
@@ -683,7 +787,7 @@ namespace nsyshid
 		ppcDefineParamMPTR(callbackFuncMPTR, 3);  // r6
 		ppcDefineParamMPTR(callbackParamMPTR, 4); // r7
 		cemuLog_logDebug(LogType::Force, "nsyshid.HIDWrite(0x{:x},0x{:08x},0x{:08x},0x{:08x},0x{:08x})", hCPU->gpr[3],
-						 hCPU->gpr[4], hCPU->gpr[5], hCPU->gpr[6], hCPU->gpr[7]);
+					hCPU->gpr[4], hCPU->gpr[5], hCPU->gpr[6], hCPU->gpr[7]);
 
 		std::shared_ptr<Device> device = GetDeviceByHandle(hidHandle, true);
 		if (device == nullptr)
@@ -718,7 +822,7 @@ namespace nsyshid
 		ppcDefineParamTypePtr(ukn0, uint32be, 1);
 		ppcDefineParamTypePtr(ukn1, uint32be, 2);
 		cemuLog_logDebug(LogType::Force, "nsyshid.HIDDecodeError(0x{:08x},0x{:08x},0x{:08x})", hCPU->gpr[3],
-						 hCPU->gpr[4], hCPU->gpr[5]);
+					hCPU->gpr[4], hCPU->gpr[5]);
 
 		// todo
 		*ukn0 = 0x3FF;

src/Cafe/OS/libs/nsysnet/nsysnet.cpp

@@ -3,6 +3,7 @@
 #include "Cafe/OS/libs/coreinit/coreinit_Thread.h"
 #include "Cafe/IOSU/legacy/iosu_crypto.h"
 #include "Cafe/OS/libs/coreinit/coreinit_Time.h"
+#include "Cafe/OS/libs/coreinit/coreinit_GHS.h"
 
 #include "Common/socket.h"
 
@@ -117,20 +118,14 @@ void nsysnetExport_socket_lib_finish(PPCInterpreter_t* hCPU)
 	osLib_returnFromFunction(hCPU, 0); // 0 -> Success
 }
 
-static uint32be* __gh_errno_ptr()
-{
-	OSThread_t* osThread = coreinit::OSGetCurrentThread();
-	return &osThread->context.ghs_errno;
-}
-
 void _setSockError(sint32 errCode)
 {
-	*(uint32be*)__gh_errno_ptr() = (uint32)errCode;
+	coreinit::__gh_set_errno(errCode);
 }
 
 sint32 _getSockError()
 {
-	return (sint32)*(uint32be*)__gh_errno_ptr();
+	return coreinit::__gh_get_errno();
 }
 
 // error translation modes for _translateError

src/Cafe/OS/libs/padscore/padscore.cpp

@@ -760,7 +760,7 @@ namespace padscore
 	void start()
 	{
 		OSCreateAlarm(&g_padscore.alarm);
-		const uint64 start_tick = coreinit::coreinit_getOSTime();
+		const uint64 start_tick = coreinit::OSGetTime();
 		const uint64 period_tick = coreinit::EspressoTime::GetTimerClock() / 200; // every 5ms
 		MPTR handler = PPCInterpreter_makeCallableExportDepr(TickFunction);
 		OSSetPeriodicAlarm(&g_padscore.alarm, start_tick, period_tick, handler);

src/Cafe/OS/libs/proc_ui/proc_ui.cpp

@@ -427,7 +427,7 @@ namespace proc_ui
 		}
 		if(callbackType != ProcUICallbackId::AcquireForeground)
 			priority = -priority;
-		AddCallbackInternal(funcPtr, userParam, priority, 0, s_CallbackTables[stdx::to_underlying(callbackType)][coreIndex]);
+		AddCallbackInternal(funcPtr, userParam, 0, priority, s_CallbackTables[stdx::to_underlying(callbackType)][coreIndex]);
 	}
 
 	void ProcUIRegisterCallback(ProcUICallbackId callbackType, void* funcPtr, void* userParam, sint32 priority)
@@ -437,7 +437,7 @@ namespace proc_ui
 
 	void ProcUIRegisterBackgroundCallback(void* funcPtr, void* userParam, uint64 tickDelay)
 	{
-		AddCallbackInternal(funcPtr, userParam, 0, tickDelay, s_backgroundCallbackList);
+		AddCallbackInternal(funcPtr, userParam, tickDelay, 0, s_backgroundCallbackList);
 	}
 
 	void FreeCallbackChain(ProcUICallbackList& callbackList)

src/Cafe/OS/libs/snd_core/ax_out.cpp

@@ -396,90 +396,35 @@ namespace snd_core
 
 	void AXOut_init()
 	{
-		auto& config = GetConfig();
-		const auto audio_api = (IAudioAPI::AudioAPI)config.audio_api;
 
 		numQueuedFramesSndGeneric = 0;
 
 		std::unique_lock lock(g_audioMutex);
 		if (!g_tvAudio)
 		{
-			sint32 channels;
-			switch (config.tv_channels)
+			try
 			{
-			case 0:
-				channels = 1; // will mix mono sound on both output channels
-				break;
-			case 2:
-				channels = 6;
-				break;
-			default: // stereo
-				channels = 2;
-				break;
+				g_tvAudio = IAudioAPI::CreateDeviceFromConfig(true, 48000, snd_core::AX_SAMPLES_PER_3MS_48KHZ * AX_FRAMES_PER_GROUP, 16);
 			}
-
-			IAudioAPI::DeviceDescriptionPtr device_description;
-			if (IAudioAPI::IsAudioAPIAvailable(audio_api))
+			catch (std::runtime_error& ex)
 			{
-				auto devices = IAudioAPI::GetDevices(audio_api);
-				const auto it = std::find_if(devices.begin(), devices.end(), [&config](const auto& d) {return d->GetIdentifier() == config.tv_device; });
-				if (it != devices.end())
-					device_description = *it;
-			}
-
-			if (device_description)
-			{
-				try
-				{
-					g_tvAudio = IAudioAPI::CreateDevice((IAudioAPI::AudioAPI)config.audio_api, device_description, 48000, channels, snd_core::AX_SAMPLES_PER_3MS_48KHZ * AX_FRAMES_PER_GROUP, 16);
-					g_tvAudio->SetVolume(config.tv_volume);
-				}
-				catch (std::runtime_error& ex)
-				{
-					cemuLog_log(LogType::Force, "can't initialize tv audio: {}", ex.what());
-					exit(0);
-				}
+				cemuLog_log(LogType::Force, "can't initialize tv audio: {}", ex.what());
+				exit(0);
 			}
 		}
 
 		if (!g_padAudio)
 		{
-			sint32 channels;
-			switch (config.pad_channels)
+			try
 			{
-			case 0:
-				channels = 1; // will mix mono sound on both output channels
-				break;
-			case 2:
-				channels = 6;
-				break;
-			default: // stereo
-				channels = 2;
-				break;
+				g_padAudio = IAudioAPI::CreateDeviceFromConfig(false, 48000, snd_core::AX_SAMPLES_PER_3MS_48KHZ * AX_FRAMES_PER_GROUP, 16);
+				if(g_padAudio)
+					g_padVolume = g_padAudio->GetVolume();
 			}
-
-			IAudioAPI::DeviceDescriptionPtr device_description;
-			if (IAudioAPI::IsAudioAPIAvailable(audio_api))
+			catch (std::runtime_error& ex)
 			{
-				auto devices = IAudioAPI::GetDevices(audio_api);
-				const auto it = std::find_if(devices.begin(), devices.end(), [&config](const auto& d) {return d->GetIdentifier() == config.pad_device; });
-				if (it != devices.end())
-					device_description = *it;
-			}
-
-			if (device_description)
-			{
-				try
-				{
-					g_padAudio = IAudioAPI::CreateDevice((IAudioAPI::AudioAPI)config.audio_api, device_description, 48000, channels, snd_core::AX_SAMPLES_PER_3MS_48KHZ * AX_FRAMES_PER_GROUP, 16);
-					g_padAudio->SetVolume(config.pad_volume);
-					g_padVolume = config.pad_volume;
-				}
-				catch (std::runtime_error& ex)
-				{
-					cemuLog_log(LogType::Force, "can't initialize pad audio: {}", ex.what());
-					exit(0);
-				}
+				cemuLog_log(LogType::Force, "can't initialize pad audio: {}", ex.what());
+				exit(0);
 			}
 		}
 	}

src/Cafe/OS/libs/vpad/vpad.cpp

@@ -267,7 +267,7 @@ namespace vpad
 		{
 			if (channel <= 1 && vpadDelayEnabled)
 			{
-				uint64 currentTime = coreinit::coreinit_getOSTime();
+				uint64 currentTime = coreinit::OSGetTime();
 				const auto dif = currentTime - vpad::g_vpad.controller_data[channel].drcLastCallTime;
 				if (dif <= (ESPRESSO_TIMER_CLOCK / 60ull))
 				{
@@ -1149,7 +1149,7 @@ namespace vpad
 	void start()
 	{
 		coreinit::OSCreateAlarm(&g_vpad.alarm);
-		const uint64 start_tick = coreinit::coreinit_getOSTime();
+		const uint64 start_tick = coreinit::OSGetTime();
 		const uint64 period_tick = coreinit::EspressoTime::GetTimerClock() * 5 / 1000;
 		const MPTR handler = PPCInterpreter_makeCallableExportDepr(TickFunction);
 		coreinit::OSSetPeriodicAlarm(&g_vpad.alarm, start_tick, period_tick, handler);

src/Common/precompiled.h

@@ -274,6 +274,25 @@ inline uint64 _udiv128(uint64 highDividend, uint64 lowDividend, uint64 divisor,
 	#define NOEXPORT __attribute__ ((visibility ("hidden")))
 #endif
 
+#if defined(_MSC_VER)
+#define FORCE_INLINE __forceinline
+#elif defined(__GNUC__) || defined(__clang__)
+#define FORCE_INLINE inline __attribute__((always_inline))
+#else
+#define FORCE_INLINE inline
+#endif
+
+FORCE_INLINE int BSF(uint32 v) // returns index of first bit set, counting from LSB. If v is 0 then result is undefined
+{
+#if defined(_MSC_VER)
+	return _tzcnt_u32(v); // TZCNT requires BMI1. But if not supported it will execute as BSF
+#elif defined(__GNUC__) || defined(__clang__)
+	return __builtin_ctz(v);
+#else
+	return std::countr_zero(v);
+#endif
+}
+
 // On aarch64 we handle some of the x86 intrinsics by implementing them as wrappers
 #if defined(__aarch64__)
 

src/audio/CubebAPI.cpp

@@ -114,7 +114,7 @@ CubebAPI::~CubebAPI()
 bool CubebAPI::NeedAdditionalBlocks() const
 {
 	std::shared_lock lock(m_mutex);
-	return m_buffer.size() < s_audioDelay * m_bytesPerBlock;
+	return m_buffer.size() < GetAudioDelay() * m_bytesPerBlock;
 }
 
 bool CubebAPI::FeedBlock(sint16* data)
@@ -183,17 +183,17 @@ void CubebAPI::Destroy()
 
 std::vector<IAudioAPI::DeviceDescriptionPtr> CubebAPI::GetDevices()
 {
-	cubeb_device_collection devices;
-	if (cubeb_enumerate_devices(s_context, CUBEB_DEVICE_TYPE_OUTPUT, &devices) != CUBEB_OK)
-		return {};
-
 	std::vector<DeviceDescriptionPtr> result;
-	result.reserve(devices.count + 1); // Reserve space for the default device
-
 	// Add the default device to the list
 	auto defaultDevice = std::make_shared<CubebDeviceDescription>(nullptr, "default", L"Default Device");
 	result.emplace_back(defaultDevice);
 
+	cubeb_device_collection devices;
+	if (cubeb_enumerate_devices(s_context, CUBEB_DEVICE_TYPE_OUTPUT, &devices) != CUBEB_OK)
+		return result;
+
+	result.reserve(devices.count + 1); // The default device already occupies one element
+
 	for (size_t i = 0; i < devices.count; ++i)
 	{
 		// const auto& device = devices.device[i];

src/audio/CubebInputAPI.cpp

@@ -175,17 +175,17 @@ void CubebInputAPI::Destroy()
 
 std::vector<IAudioInputAPI::DeviceDescriptionPtr> CubebInputAPI::GetDevices()
 {
-	cubeb_device_collection devices;
-	if (cubeb_enumerate_devices(s_context, CUBEB_DEVICE_TYPE_INPUT, &devices) != CUBEB_OK)
-		return {};
-
 	std::vector<DeviceDescriptionPtr> result;
-	result.reserve(devices.count + 1); // Reserve space for the default device
-
 	// Add the default device to the list
 	auto defaultDevice = std::make_shared<CubebDeviceDescription>(nullptr, "default", L"Default Device");
 	result.emplace_back(defaultDevice);
 
+	cubeb_device_collection devices;
+	if (cubeb_enumerate_devices(s_context, CUBEB_DEVICE_TYPE_INPUT, &devices) != CUBEB_OK)
+		return result;
+
+	result.reserve(devices.count + 1); // The default device already occupies one element
+
 	for (size_t i = 0; i < devices.count; ++i)
 	{
 		// const auto& device = devices.device[i];

src/audio/DirectSoundAPI.cpp

@@ -210,7 +210,7 @@ void DirectSoundAPI::SetVolume(sint32 volume)
 bool DirectSoundAPI::NeedAdditionalBlocks() const
 {
 	std::shared_lock lock(m_mutex);
-	return m_buffer.size() < s_audioDelay;
+	return m_buffer.size() < GetAudioDelay();
 }
 
 std::vector<DirectSoundAPI::DeviceDescriptionPtr> DirectSoundAPI::GetDevices()

src/audio/IAudioAPI.cpp

@@ -97,7 +97,40 @@ bool IAudioAPI::IsAudioAPIAvailable(AudioAPI api)
 	return false;
 }
 
+AudioAPIPtr IAudioAPI::CreateDeviceFromConfig(bool TV, sint32 rate, sint32 samples_per_block, sint32 bits_per_sample)
+{
+	auto& config = GetConfig();
+	sint32 channels = CemuConfig::AudioChannelsToNChannels(TV ? config.tv_channels : config.pad_channels);
+	return CreateDeviceFromConfig(TV, rate, channels, samples_per_block, bits_per_sample);
+}
 
+AudioAPIPtr IAudioAPI::CreateDeviceFromConfig(bool TV, sint32 rate, sint32 channels, sint32 samples_per_block, sint32 bits_per_sample)
+{
+	AudioAPIPtr audioAPIDev;
+
+	auto& config = GetConfig();
+
+	const auto audio_api = (IAudioAPI::AudioAPI)config.audio_api;
+	auto& selectedDevice = TV ? config.tv_device : config.pad_device;
+
+	if(selectedDevice.empty())
+		return {};
+
+	IAudioAPI::DeviceDescriptionPtr device_description;
+	if (IAudioAPI::IsAudioAPIAvailable(audio_api))
+	{
+		auto devices = IAudioAPI::GetDevices(audio_api);
+		const auto it = std::find_if(devices.begin(), devices.end(), [&selectedDevice](const auto& d) {return d->GetIdentifier() == selectedDevice; });
+		if (it != devices.end())
+			device_description = *it;
+	}
+	if (!device_description)
+		throw std::runtime_error("failed to find selected device while trying to create audio device");
+
+	audioAPIDev = CreateDevice(audio_api, device_description, rate, channels, samples_per_block, bits_per_sample);
+	audioAPIDev->SetVolume(TV ? config.tv_volume : config.pad_volume);
+	return audioAPIDev;
+}
 
 AudioAPIPtr IAudioAPI::CreateDevice(AudioAPI api, const DeviceDescriptionPtr& device, sint32 samplerate, sint32 channels, sint32 samples_per_block, sint32 bits_per_sample)
 {
@@ -167,3 +200,12 @@ std::vector<IAudioAPI::DeviceDescriptionPtr> IAudioAPI::GetDevices(AudioAPI api)
 	}
 }
 
+void IAudioAPI::SetAudioDelayOverride(uint32 delay)
+{
+	m_audioDelayOverride = delay;
+}
+
+uint32 IAudioAPI::GetAudioDelay() const
+{
+	return m_audioDelayOverride > 0 ? m_audioDelayOverride : s_audioDelay;
+}

src/audio/IAudioAPI.h

@@ -55,11 +55,15 @@ public:
 	virtual bool FeedBlock(sint16* data) = 0;
 	virtual bool Play() = 0;
 	virtual bool Stop() = 0;
+	void SetAudioDelayOverride(uint32 delay);
+	uint32 GetAudioDelay() const;
 
 	static void PrintLogging();
 	static void InitializeStatic();
 	static bool IsAudioAPIAvailable(AudioAPI api);
 
+	static std::unique_ptr<IAudioAPI> CreateDeviceFromConfig(bool TV, sint32 rate, sint32 samples_per_block, sint32 bits_per_sample);
+	static std::unique_ptr<IAudioAPI> CreateDeviceFromConfig(bool TV, sint32 rate, sint32 channels, sint32 samples_per_block, sint32 bits_per_sample);
 	static std::unique_ptr<IAudioAPI> CreateDevice(AudioAPI api, const DeviceDescriptionPtr& device, sint32 samplerate, sint32 channels, sint32 samples_per_block, sint32 bits_per_sample);
 	static std::vector<DeviceDescriptionPtr> GetDevices(AudioAPI api);
 
@@ -75,9 +79,10 @@ protected:
 	bool m_playing = false;
 
 	static std::array<bool, AudioAPIEnd> s_availableApis;
-	static uint32 s_audioDelay;
+	uint32 m_audioDelayOverride = 0;
 
 private:
+	static uint32 s_audioDelay;
 	void InitWFX(sint32 samplerate, sint32 channels, sint32 bits_per_sample);
 	
 };

src/audio/XAudio27API.cpp

@@ -33,8 +33,8 @@ XAudio27API::XAudio27API(uint32 device_id, uint32 samplerate, uint32 channels, u
 	m_wfx.Format.nChannels = channels;
 	m_wfx.Format.nSamplesPerSec = samplerate;
 	m_wfx.Format.wBitsPerSample = bits_per_sample;
-	m_wfx.Format.nBlockAlign = (m_wfx.Format.nChannels * m_wfx.Format.wBitsPerSample) / 8; // must equal (nChannels × wBitsPerSample) / 8
-	m_wfx.Format.nAvgBytesPerSec = m_wfx.Format.nSamplesPerSec * m_wfx.Format.nBlockAlign; // must equal nSamplesPerSec × nBlockAlign.
+	m_wfx.Format.nBlockAlign = (m_wfx.Format.nChannels * m_wfx.Format.wBitsPerSample) / 8; // must equal (nChannels × wBitsPerSample) / 8
+	m_wfx.Format.nAvgBytesPerSec = m_wfx.Format.nSamplesPerSec * m_wfx.Format.nBlockAlign; // must equal nSamplesPerSec × nBlockAlign.
 	m_wfx.Format.cbSize = sizeof(WAVEFORMATEXTENSIBLE) - sizeof(WAVEFORMATEX);
 
 	m_wfx.SubFormat = KSDATAFORMAT_SUBTYPE_PCM;
@@ -199,9 +199,7 @@ bool XAudio27API::FeedBlock(sint16* data)
 	// check if we queued too many blocks
 	if(m_blocks_queued >= kBlockCount)
 	{
-		XAUDIO2_VOICE_STATE state{};
-		m_source_voice->GetState(&state);
-		m_blocks_queued = state.BuffersQueued;
+		m_blocks_queued = GetQueuedBuffers();
 
 		if (m_blocks_queued >= kBlockCount)
 		{
@@ -222,7 +220,14 @@ bool XAudio27API::FeedBlock(sint16* data)
 	return true;
 }
 
+uint32 XAudio27API::GetQueuedBuffers() const
+{
+	XAUDIO2_VOICE_STATE state{};
+	m_source_voice->GetState(&state);
+	return state.BuffersQueued;
+}
+
 bool XAudio27API::NeedAdditionalBlocks() const
 {
-	return m_blocks_queued < s_audioDelay;
+	return GetQueuedBuffers() < GetAudioDelay();
 }

src/audio/XAudio27API.h

@@ -47,6 +47,8 @@ public:
 	static std::vector<DeviceDescriptionPtr> GetDevices();
 
 private:
+	uint32 GetQueuedBuffers() const;
+
 	struct XAudioDeleter
 	{
 		void operator()(IXAudio2* ptr) const;