ci: Update various actions from v2 to v3

CI: Use GitHub Actions to validate pull requests

.ci/scripts/clang/docker.sh

@@ -3,14 +3,12 @@
 # Exit on error, rather than continuing with the rest of the script.
 set -e
 
-cd /yuzu
-
 ccache -s
 
 mkdir build || true && cd build
-cmake .. -DDISPLAY_VERSION=$1 -DCMAKE_BUILD_TYPE=Release -DCMAKE_C_COMPILER=/usr/lib/ccache/clang -DCMAKE_CXX_COMPILER=/usr/lib/ccache/clang++ -DYUZU_ENABLE_COMPATIBILITY_REPORTING=${ENABLE_COMPATIBILITY_REPORTING:-"OFF"} -DENABLE_COMPATIBILITY_LIST_DOWNLOAD=ON -DUSE_DISCORD_PRESENCE=ON -DENABLE_QT_TRANSLATION=ON -DCMAKE_INSTALL_PREFIX="/usr"
+cmake .. -GNinja -DDISPLAY_VERSION=$1 -DCMAKE_BUILD_TYPE=Release -DCMAKE_C_COMPILER=/usr/lib/ccache/clang -DCMAKE_CXX_COMPILER=/usr/lib/ccache/clang++ -DYUZU_ENABLE_COMPATIBILITY_REPORTING=${ENABLE_COMPATIBILITY_REPORTING:-"OFF"} -DENABLE_COMPATIBILITY_LIST_DOWNLOAD=ON -DUSE_DISCORD_PRESENCE=ON -DENABLE_QT_TRANSLATION=ON -DCMAKE_INSTALL_PREFIX="/usr"
 
-make -j$(nproc)
+ninja
 
 ccache -s
 

.ci/scripts/clang/exec.sh

@@ -4,5 +4,5 @@ mkdir -p "ccache"  || true
 chmod a+x ./.ci/scripts/clang/docker.sh
 # the UID for the container yuzu user is 1027
 sudo chown -R 1027 ./
-docker run -e ENABLE_COMPATIBILITY_REPORTING -e CCACHE_DIR=/yuzu/ccache -v $(pwd):/yuzu yuzuemu/build-environments:linux-fresh /bin/bash /yuzu/.ci/scripts/clang/docker.sh $1
+docker run -e ENABLE_COMPATIBILITY_REPORTING -e CCACHE_DIR=/yuzu/ccache -v "$(pwd):/yuzu" -w /yuzu yuzuemu/build-environments:linux-fresh /bin/bash /yuzu/.ci/scripts/clang/docker.sh "$1"
 sudo chown -R $UID ./

.ci/scripts/common/post-upload.sh

@@ -4,8 +4,10 @@
 cp license.txt "$DIR_NAME"
 cp README.md "$DIR_NAME"
 
-tar -cJvf "${REV_NAME}-source.tar.xz" src externals CMakeLists.txt README.md license.txt
-cp "${REV_NAME}-source.tar.xz" "$DIR_NAME"
+if [[ -z "${NO_SOURCE_PACK}" ]]; then
+  tar -cJvf "${REV_NAME}-source.tar.xz" src externals CMakeLists.txt README.md license.txt
+  cp -v "${REV_NAME}-source.tar.xz" "$DIR_NAME"
+fi
 
 tar $COMPRESSION_FLAGS "$ARCHIVE_NAME" "$DIR_NAME"
 

.ci/scripts/format/exec.sh

@@ -3,5 +3,5 @@
 chmod a+x ./.ci/scripts/format/docker.sh
 # the UID for the container yuzu user is 1027
 sudo chown -R 1027 ./
-docker run -v $(pwd):/yuzu yuzuemu/build-environments:linux-clang-format /bin/bash -ex /yuzu/.ci/scripts/format/docker.sh
+docker run -v "$(pwd):/yuzu" -w /yuzu yuzuemu/build-environments:linux-clang-format /bin/bash -ex /yuzu/.ci/scripts/format/docker.sh
 sudo chown -R $UID ./

.ci/scripts/linux/docker.sh

@@ -3,8 +3,6 @@
 # Exit on error, rather than continuing with the rest of the script.
 set -e
 
-cd /yuzu
-
 ccache -s
 
 mkdir build || true && cd build
@@ -19,15 +17,16 @@ cmake .. \
       -DENABLE_QT_TRANSLATION=ON \
       -DUSE_DISCORD_PRESENCE=ON \
       -DYUZU_ENABLE_COMPATIBILITY_REPORTING=${ENABLE_COMPATIBILITY_REPORTING:-"OFF"} \
-      -DYUZU_USE_BUNDLED_FFMPEG=ON
+      -DYUZU_USE_BUNDLED_FFMPEG=ON \
+      -GNinja
 
-make -j$(nproc)
+ninja
 
 ccache -s
 
 ctest -VV -C Release
 
-make install DESTDIR=AppDir
+DESTDIR="$PWD/AppDir" ninja install
 rm -vf AppDir/usr/bin/yuzu-cmd AppDir/usr/bin/yuzu-tester
 
 # Download tools needed to build an AppImage

.ci/scripts/linux/exec.sh

@@ -4,5 +4,5 @@ mkdir -p "ccache"  || true
 chmod a+x ./.ci/scripts/linux/docker.sh
 # the UID for the container yuzu user is 1027
 sudo chown -R 1027 ./
-docker run -e ENABLE_COMPATIBILITY_REPORTING -e CCACHE_DIR=/yuzu/ccache -v $(pwd):/yuzu yuzuemu/build-environments:linux-fresh /bin/bash /yuzu/.ci/scripts/linux/docker.sh $1
+docker run -e ENABLE_COMPATIBILITY_REPORTING -e CCACHE_DIR=/yuzu/ccache -v "$(pwd):/yuzu" -w /yuzu yuzuemu/build-environments:linux-fresh /bin/bash /yuzu/.ci/scripts/linux/docker.sh "$1"
 sudo chown -R $UID ./

.ci/scripts/linux/upload.sh

@@ -24,6 +24,11 @@ cd build
 wget -nc https://github.com/yuzu-emu/ext-linux-bin/raw/main/appimage/appimagetool-x86_64.AppImage
 chmod 755 appimagetool-x86_64.AppImage
 
+# if FUSE is not available, then fallback to extract and run
+if ! ./appimagetool-x86_64.AppImage --version; then
+    export APPIMAGE_EXTRACT_AND_RUN=1
+fi
+
 if [ "${RELEASE_NAME}" = "mainline" ]; then
     # Generate update information if releasing to mainline
     ./appimagetool-x86_64.AppImage -u "gh-releases-zsync|yuzu-emu|yuzu-${RELEASE_NAME}|latest|yuzu-*.AppImage.zsync" AppDir "${APPIMAGE_NAME}"

.ci/scripts/windows/docker.sh

@@ -1,14 +1,33 @@
 #!/bin/bash -ex
 
-cd /yuzu
+set -e
 
-ccache -s
+#cd /yuzu
 
-mkdir build || true && cd build
-cmake .. -G Ninja -DDISPLAY_VERSION=$1 -DCMAKE_TOOLCHAIN_FILE="$(pwd)/../CMakeModules/MinGWCross.cmake" -DUSE_CCACHE=ON -DENABLE_COMPATIBILITY_LIST_DOWNLOAD=ON -DCMAKE_BUILD_TYPE=Release -DENABLE_QT_TRANSLATION=ON
-ninja
+ccache -sv
 
-ccache -s
+mkdir -p "$HOME/.conan/profiles"
+wget -c "https://github.com/yuzu-emu/build-environments/raw/master/linux-mingw/default" -O "$HOME/.conan/profiles/default"
+wget -c "https://github.com/yuzu-emu/build-environments/raw/master/linux-mingw/settings.yml" -O "$HOME/.conan/settings.yml"
+
+mkdir -p build && cd build
+export LDFLAGS="-fuse-ld=lld"
+# -femulated-tls required due to an incompatibility between GCC and Clang
+# TODO(lat9nq): If this is widespread, we probably need to add this to CMakeLists where appropriate
+export CXXFLAGS="-femulated-tls"
+cmake .. \
+    -DCMAKE_BUILD_TYPE=Release \
+    -DCMAKE_TOOLCHAIN_FILE="${PWD}/../CMakeModules/MinGWClangCross.cmake" \
+    -DDISPLAY_VERSION="$1" \
+    -DENABLE_COMPATIBILITY_LIST_DOWNLOAD=ON \
+    -DENABLE_QT_TRANSLATION=ON \
+    -DUSE_CCACHE=ON \
+    -DYUZU_USE_BUNDLED_SDL2=OFF \
+    -DYUZU_USE_EXTERNAL_SDL2=OFF \
+    -GNinja
+ninja yuzu yuzu-cmd
+
+ccache -sv
 
 echo "Tests skipped"
 #ctest -VV -C Release

.ci/scripts/windows/exec.sh

@@ -4,5 +4,5 @@ mkdir -p "ccache" || true
 chmod a+x ./.ci/scripts/windows/docker.sh
 # the UID for the container yuzu user is 1027
 sudo chown -R 1027 ./
-docker run -e CCACHE_DIR=/yuzu/ccache -v $(pwd):/yuzu yuzuemu/build-environments:linux-mingw /bin/bash -ex /yuzu/.ci/scripts/windows/docker.sh $1
+docker run -e CCACHE_DIR=/yuzu/ccache -v "$(pwd):/yuzu" -w /yuzu yuzuemu/build-environments:linux-mingw /bin/bash -ex /yuzu/.ci/scripts/windows/docker.sh "$1"
 sudo chown -R $UID ./

.github/workflows/ci.yml

@@ -13,7 +13,7 @@ jobs:
     container: yuzuemu/build-environments:linux-transifex
     if: ${{ github.repository == 'yuzu-emu/yuzu' && !github.head_ref }}
     steps:
-      - uses: actions/checkout@v2
+      - uses: actions/checkout@v3
         with:
           submodules: recursive
           fetch-depth: 0   

.github/workflows/verify.yml

@@ -0,0 +1,115 @@
+name: 'yuzu verify'
+
+on:
+  pull_request:
+    branches: [ master ]
+
+jobs:
+  format:
+    name: 'verify format'
+    runs-on: ubuntu-latest
+    container:
+      image: yuzuemu/build-environments:linux-clang-format
+      options: -u 1001
+    steps:
+      - uses: actions/checkout@v3
+        with:
+          submodules: false
+      - name: 'Verify Formatting'
+        run: bash -ex ./.ci/scripts/format/script.sh
+  build:
+    name: 'test build'
+    needs: format
+    runs-on: ubuntu-latest
+    strategy:
+      fail-fast: false
+      matrix:
+        include:
+          - type: clang
+            image: linux-fresh
+          - type: linux
+            image: linux-fresh
+          - type: windows
+            image: linux-mingw
+    container:
+      image: yuzuemu/build-environments:${{ matrix.image }}
+      options: -u 1001
+    steps:
+      - uses: actions/checkout@v3
+        with:
+          submodules: recursive
+          fetch-depth: 0
+      - name: Set up cache
+        uses: actions/cache@v3
+        id: ccache-restore
+        with:
+          path: ~/.ccache
+          key: ${{ runner.os }}-${{ matrix.type }}-${{ github.sha }}
+          restore-keys: |
+            ${{ runner.os }}-${{ matrix.type }}-
+      - name: Create ccache directory
+        if: steps.ccache-restore.outputs.cache-hit != 'true'
+        run: mkdir -p ~/.ccache
+      - name: Build
+        run: ./.ci/scripts/${{ matrix.type }}/docker.sh
+        env:
+          ENABLE_COMPATIBILITY_REPORTING: "ON"
+      - name: Pack
+        run: ./.ci/scripts/${{ matrix.type }}/upload.sh
+        env:
+          NO_SOURCE_PACK: "YES"
+      - name: Upload
+        uses: actions/upload-artifact@v3
+        with:
+          name: ${{ matrix.type }}
+          path: artifacts/
+  build-msvc:
+    name: 'test build (windows, msvc)'
+    needs: format
+    runs-on: windows-2019
+    steps:
+      - name: Set up cache
+        uses: actions/cache@v3
+        with:
+          path: ~/.buildcache
+          key: ${{ runner.os }}-msvc-${{ github.sha }}
+          restore-keys: |
+            ${{ runner.os }}-msvc-
+      - name: Install dependencies
+        # due to how chocolatey works, only cmd.exe is supported here
+        shell: cmd
+        run: |
+          choco install vulkan-sdk wget
+          python -m pip install --upgrade pip conan
+          call refreshenv
+          wget https://github.com/mbitsnbites/buildcache/releases/download/v0.27.6/buildcache-windows.zip
+          7z x buildcache-windows.zip
+          copy buildcache\bin\buildcache.exe C:\ProgramData\chocolatey\bin
+          rmdir buildcache
+          echo %PATH% >> %GITHUB_PATH%
+      - name: Set up MSVC
+        uses: ilammy/msvc-dev-cmd@v1
+      - uses: actions/checkout@v3
+        with:
+          submodules: recursive
+          fetch-depth: 0
+      - name: Configure
+        env:
+          CC: cl.exe
+          CXX: cl.exe
+        run: |
+          glslangValidator --version
+          mkdir build
+          cmake . -B build -GNinja -DCMAKE_TOOLCHAIN_FILE="CMakeModules/MSVCCache.cmake" -DUSE_CCACHE=ON -DYUZU_USE_BUNDLED_QT=1 -DYUZU_USE_BUNDLED_SDL2=1 -DYUZU_USE_QT_WEB_ENGINE=ON -DENABLE_COMPATIBILITY_LIST_DOWNLOAD=ON -DYUZU_ENABLE_COMPATIBILITY_REPORTING=ON -DUSE_DISCORD_PRESENCE=ON -DENABLE_QT_TRANSLATION=ON -DCMAKE_BUILD_TYPE=Release
+      - name: Build
+        run: cmake --build build
+      - name: Cache Summary
+        run: buildcache -s
+      - name: Pack
+        shell: pwsh
+        run: .\.ci\scripts\windows\upload.ps1
+      - name: Upload
+        uses: actions/upload-artifact@v3
+        with:
+          name: msvc
+          path: artifacts/

CMakeLists.txt

@@ -218,8 +218,13 @@ if (Boost_FOUND)
         list(APPEND Boost_LIBRARIES Boost::context)
     endif()
 else()
-    message(STATUS "Boost 1.73.0 or newer not found, falling back to Conan")
-    list(APPEND CONAN_REQUIRED_LIBS "boost/1.78.0")
+    message(STATUS "Boost 1.79.0 or newer not found, falling back to Conan")
+    list(APPEND CONAN_REQUIRED_LIBS "boost/1.79.0")
+endif()
+
+# boost:asio has functions that require AcceptEx et al
+if (MINGW)
+    find_library(MSWSOCK_LIBRARY mswsock REQUIRED)
 endif()
 
 # Attempt to locate any packages that are required and report the missing ones in CONAN_REQUIRED_LIBS
@@ -622,6 +627,14 @@ add_definitions(-DBOOST_ERROR_CODE_HEADER_ONLY
     -DBOOST_DATE_TIME_NO_LIB
     -DBOOST_REGEX_NO_LIB
 )
+# Adjustments for MSVC + Ninja
+if (MSVC AND CMAKE_GENERATOR STREQUAL "Ninja")
+    add_compile_options(
+        /wd4711 # function 'function' selected for automatic inline expansion
+        /wd4464 # relative include path contains '..'
+        /wd4820 # 'identifier1': '4' bytes padding added after data member 'identifier2'
+    )
+endif()
 
 enable_testing()
 add_subdirectory(externals)

CMakeModules/CopyYuzuFFmpegDeps.cmake

@@ -2,5 +2,6 @@ function(copy_yuzu_FFmpeg_deps target_dir)
     include(WindowsCopyFiles)
     set(DLL_DEST "${CMAKE_BINARY_DIR}/bin/$<CONFIG>/")
     file(READ "${FFmpeg_PATH}/requirements.txt" FFmpeg_REQUIRED_DLLS)
+    string(STRIP "${FFmpeg_REQUIRED_DLLS}" FFmpeg_REQUIRED_DLLS)
     windows_copy_files(${target_dir} ${FFmpeg_DLL_DIR} ${DLL_DEST} ${FFmpeg_REQUIRED_DLLS})
 endfunction(copy_yuzu_FFmpeg_deps)

CMakeModules/MSVCCache.cmake

@@ -0,0 +1,12 @@
+# buildcache wrapper
+OPTION(USE_CCACHE "Use buildcache for compilation" OFF)
+IF(USE_CCACHE)
+    FIND_PROGRAM(CCACHE buildcache)
+    IF (CCACHE)
+        MESSAGE(STATUS "Using buildcache found in PATH")
+        SET_PROPERTY(GLOBAL PROPERTY RULE_LAUNCH_COMPILE ${CCACHE})
+        SET_PROPERTY(GLOBAL PROPERTY RULE_LAUNCH_LINK ${CCACHE})
+    ELSE(CCACHE)
+        MESSAGE(WARNING "USE_CCACHE enabled, but no buildcache executable found")
+    ENDIF(CCACHE)
+ENDIF(USE_CCACHE)

CMakeModules/MinGWClangCross.cmake

@@ -0,0 +1,55 @@
+set(MINGW_PREFIX                /usr/x86_64-w64-mingw32/)
+set(CMAKE_SYSTEM_NAME           Windows)
+set(CMAKE_SYSTEM_PROCESSOR      x86_64)
+
+set(CMAKE_FIND_ROOT_PATH        ${MINGW_PREFIX})
+set(SDL2_PATH                   ${MINGW_PREFIX})
+set(MINGW_TOOL_PREFIX           ${CMAKE_SYSTEM_PROCESSOR}-w64-mingw32-)
+
+# Specify the cross compiler
+set(CMAKE_C_COMPILER            ${MINGW_TOOL_PREFIX}clang)
+set(CMAKE_CXX_COMPILER          ${MINGW_TOOL_PREFIX}clang++)
+set(CMAKE_RC_COMPILER           ${MINGW_TOOL_PREFIX}windres)
+set(CMAKE_C_COMPILER_AR         ${MINGW_TOOL_PREFIX}ar)
+set(CMAKE_CXX_COMPILER_AR       ${MINGW_TOOL_PREFIX}ar)
+set(CMAKE_C_COMPILER_RANLIB     ${MINGW_TOOL_PREFIX}ranlib)
+set(CMAKE_CXX_COMPILER_RANLIB   ${MINGW_TOOL_PREFIX}ranlib)
+
+# Mingw tools
+set(STRIP                       ${MINGW_TOOL_PREFIX}strip)
+set(WINDRES                     ${MINGW_TOOL_PREFIX}windres)
+set(ENV{PKG_CONFIG}             ${MINGW_TOOL_PREFIX}pkg-config)
+
+# ccache wrapper
+option(USE_CCACHE "Use ccache for compilation" OFF)
+if(USE_CCACHE)
+    find_program(CCACHE ccache)
+    if(CCACHE)
+        message(STATUS "Using ccache found in PATH")
+        set_property(GLOBAL PROPERTY RULE_LAUNCH_COMPILE ${CCACHE})
+        set_property(GLOBAL PROPERTY RULE_LAUNCH_LINK ${CCACHE})
+    else(CCACHE)
+        message(WARNING "USE_CCACHE enabled, but no ccache found")
+    endif(CCACHE)
+endif(USE_CCACHE)
+
+# Search for programs in the build host directories
+set(CMAKE_FIND_ROOT_PATH_MODE_PROGRAM NEVER)
+
+
+# Echo modified cmake vars to screen for debugging purposes
+if(NOT DEFINED ENV{MINGW_DEBUG_INFO})
+        message("")
+        message("Custom cmake vars: (blank = system default)")
+        message("-----------------------------------------")
+        message("* CMAKE_C_COMPILER                     : ${CMAKE_C_COMPILER}")
+        message("* CMAKE_CXX_COMPILER                   : ${CMAKE_CXX_COMPILER}")
+        message("* CMAKE_RC_COMPILER                    : ${CMAKE_RC_COMPILER}")
+        message("* WINDRES                              : ${WINDRES}")
+        message("* ENV{PKG_CONFIG}                      : $ENV{PKG_CONFIG}")
+        message("* STRIP                                : ${STRIP}")
+        message("* USE_CCACHE                           : ${USE_CCACHE}")
+        message("")
+        # So that the debug info only appears once
+        set(ENV{MINGW_DEBUG_INFO} SHOWN)
+endif()

dist/org.yuzu_emu.yuzu.metainfo.xml

@@ -32,6 +32,11 @@
     <binary>yuzu</binary>
     <binary>yuzu-cmd</binary>
   </provides>
+  <supports>
+    <control>pointing</control>
+    <control>keyboard</control>
+    <control>gamepad</control>
+  </supports>
   <requires>
     <memory>8192</memory>
   </requires>

dist/qt_themes/default/style.qss

@@ -58,6 +58,19 @@ QPushButton#GPUStatusBarButton:!checked {
     color: #109010;
 }
 
+QPushButton#DockingStatusBarButton {
+    min-width: 0px;
+    color: #000000;
+    border: 1px solid transparent;
+    background-color: transparent;
+    padding: 0px 3px 0px 3px;
+    text-align: center;
+}
+
+QPushButton#DockingStatusBarButton:hover {
+    border: 1px solid #76797C;
+}
+
 QPushButton#buttonRefreshDevices {
     min-width: 21px;
     min-height: 21px;

dist/qt_themes/qdarkstyle/style.qss

@@ -1304,6 +1304,19 @@ QPushButton#GPUStatusBarButton:!checked {
     color: #40dd40;
 }
 
+QPushButton#DockingStatusBarButton {
+    min-width: 0px;
+    color: #ffffff;
+    border: 1px solid transparent;
+    background-color: transparent;
+    padding: 0px 3px 0px 3px;
+    text-align: center;
+}
+
+QPushButton#DockingStatusBarButton:hover {
+    border: 1px solid #76797C;
+}
+
 QPushButton#buttonRefreshDevices {
     min-width: 23px;
     min-height: 23px;

dist/qt_themes/qdarkstyle_midnight_blue/style.qss

@@ -2207,6 +2207,19 @@ QPushButton#GPUStatusBarButton:!checked {
   color: #40dd40;
 }
 
+QPushButton#DockingStatusBarButton {
+    min-width: 0px;
+    color: #ffffff;
+    border: 1px solid transparent;
+    background-color: transparent;
+    padding: 0px 3px 0px 3px;
+    text-align: center;
+}
+
+QPushButton#DockingStatusBarButton:hover {
+    border: 1px solid #76797C;
+}
+
 QPushButton#buttonRefreshDevices {
   min-width: 19px;
   min-height: 19px;

externals/CMakeLists.txt

@@ -40,6 +40,11 @@ target_include_directories(mbedtls PUBLIC ./mbedtls/include)
 add_library(microprofile INTERFACE)
 target_include_directories(microprofile INTERFACE ./microprofile)
 
+# GCC bugs
+if (CMAKE_CXX_COMPILER_VERSION VERSION_GREATER_EQUAL "12" AND CMAKE_CXX_COMPILER_ID STREQUAL "GNU" AND MINGW)
+    target_compile_options(microprofile INTERFACE "-Wno-array-bounds")
+endif()
+
 # libusb
 if (NOT LIBUSB_FOUND OR YUZU_USE_BUNDLED_LIBUSB)
     add_subdirectory(libusb)

externals/microprofile/microprofile.h

@@ -1246,7 +1246,7 @@ struct MicroProfileScopeLock
 {
     bool bUseLock;
     std::recursive_mutex& m;
-    MicroProfileScopeLock(std::recursive_mutex& m) : bUseLock(g_bUseLock), m(m)
+    MicroProfileScopeLock(std::recursive_mutex& m_) : bUseLock(g_bUseLock), m(m_)
     {
         if(bUseLock)
             m.lock();

externals/microprofile/microprofileui.h

@@ -213,8 +213,8 @@ struct MicroProfileCustom
 
 struct SOptionDesc
 {
-    SOptionDesc(){}
-    SOptionDesc(uint8_t nSubType, uint8_t nIndex, const char* fmt, ...):nSubType(nSubType), nIndex(nIndex)
+    SOptionDesc()=default;
+    SOptionDesc(uint8_t nSubType_, uint8_t nIndex_, const char* fmt, ...):nSubType(nSubType_), nIndex(nIndex_)
     {
         va_list args;
         va_start (args, fmt);
@@ -573,10 +573,10 @@ inline void MicroProfileToolTipMeta(MicroProfileStringArray* pToolTip)
                     }
                     else
                     {
-                        for(int i = 0; i < MICROPROFILE_META_MAX; ++i)
+                        for(int k = 0; k < MICROPROFILE_META_MAX; ++k)
                         {
-                            nMetaSumInclusive[i] += nMetaSum[i];
-                            nMetaSum[i] = 0;
+                            nMetaSumInclusive[k] += nMetaSum[k];
+                            nMetaSum[k] = 0;
                         }
                     }
                     break;
@@ -708,10 +708,10 @@ inline void MicroProfileDrawFloatTooltip(uint32_t nX, uint32_t nY, uint32_t nTok
 
     if(UI.nMouseLeftMod)
     {
-        int nIndex = (g_MicroProfileUI.LockedToolTipFront + MICROPROFILE_TOOLTIP_MAX_LOCKED - 1) % MICROPROFILE_TOOLTIP_MAX_LOCKED;
-        g_MicroProfileUI.nLockedToolTipColor[nIndex] = S.TimerInfo[nTimerId].nColor;
-        MicroProfileStringArrayCopy(&g_MicroProfileUI.LockedToolTips[nIndex], &ToolTip);
-        g_MicroProfileUI.LockedToolTipFront = nIndex;
+        int nToolTipIndex = (g_MicroProfileUI.LockedToolTipFront + MICROPROFILE_TOOLTIP_MAX_LOCKED - 1) % MICROPROFILE_TOOLTIP_MAX_LOCKED;
+        g_MicroProfileUI.nLockedToolTipColor[nToolTipIndex] = S.TimerInfo[nTimerId].nColor;
+        MicroProfileStringArrayCopy(&g_MicroProfileUI.LockedToolTips[nToolTipIndex], &ToolTip);
+        g_MicroProfileUI.LockedToolTipFront = nToolTipIndex;
 
     }
 }
@@ -917,9 +917,8 @@ inline void MicroProfileDrawDetailedBars(uint32_t nWidth, uint32_t nHeight, int
         float fStart = floor(fMsBase*fRcpStep) * fStep;
         for(float f = fStart; f < fMsEnd; )
         {
-            float fStart = f;
             float fNext = f + fStep;
-            MicroProfileDrawBox(((fStart-fMsBase) * fMsToScreen), nBaseY, (fNext-fMsBase) * fMsToScreen+1, nBaseY + nHeight, UI.nOpacityBackground | g_nMicroProfileBackColors[nColorIndex++ & 1]);
+            MicroProfileDrawBox(((f-fMsBase) * fMsToScreen), nBaseY, (fNext-fMsBase) * fMsToScreen+1, nBaseY + nHeight, UI.nOpacityBackground | g_nMicroProfileBackColors[nColorIndex++ & 1]);
             f = fNext;
         }
     }
@@ -1116,9 +1115,9 @@ inline void MicroProfileDrawDetailedBars(uint32_t nWidth, uint32_t nHeight, int
 
                         nMaxStackDepth = MicroProfileMax(nMaxStackDepth, nStackPos);
                         float fMsStart = fToMs * MicroProfileLogTickDifference(nBaseTicks, nTickStart);
-                        float fMsEnd = fToMs * MicroProfileLogTickDifference(nBaseTicks, nTickEnd);
+                        float fMsEnd2 = fToMs * MicroProfileLogTickDifference(nBaseTicks, nTickEnd);
                         float fXStart = fMsStart * fMsToScreen;
-                        float fXEnd = fMsEnd * fMsToScreen;
+                        float fXEnd = fMsEnd2 * fMsToScreen;
                         float fYStart = (float)(nY + nStackPos * nYDelta);
                         float fYEnd = fYStart + (MICROPROFILE_DETAILED_BAR_HEIGHT);
                         float fXDist = MicroProfileMax(fXStart - fMouseX, fMouseX - fXEnd);
@@ -1269,22 +1268,22 @@ inline void MicroProfileDrawDetailedBars(uint32_t nWidth, uint32_t nHeight, int
         if(UI.nRangeBegin != UI.nRangeEnd)
         {
             float fMsStart = fToMsCpu * MicroProfileLogTickDifference(nBaseTicksCpu, UI.nRangeBegin);
-            float fMsEnd = fToMsCpu * MicroProfileLogTickDifference(nBaseTicksCpu, UI.nRangeEnd);
+            float fMsEnd3 = fToMsCpu * MicroProfileLogTickDifference(nBaseTicksCpu, UI.nRangeEnd);
             float fXStart = fMsStart * fMsToScreen;
-            float fXEnd = fMsEnd * fMsToScreen;
+            float fXEnd = fMsEnd3 * fMsToScreen;
             MicroProfileDrawBox(fXStart, nBaseY, fXEnd, nHeight, MICROPROFILE_FRAME_COLOR_HIGHTLIGHT, MicroProfileBoxTypeFlat);
             MicroProfileDrawLineVertical(fXStart, nBaseY, nHeight, MICROPROFILE_FRAME_COLOR_HIGHTLIGHT | 0x44000000);
             MicroProfileDrawLineVertical(fXEnd, nBaseY, nHeight, MICROPROFILE_FRAME_COLOR_HIGHTLIGHT | 0x44000000);
 
             fMsStart += fDetailedOffset;
-            fMsEnd += fDetailedOffset;
+            fMsEnd3 += fDetailedOffset;
             char sBuffer[32];
             uint32_t nLenStart = snprintf(sBuffer, sizeof(sBuffer)-1, "%.2fms", fMsStart);
             float fStartTextWidth = (float)((1+MICROPROFILE_TEXT_WIDTH) * nLenStart);
             float fStartTextX = fXStart - fStartTextWidth - 2;
             MicroProfileDrawBox(fStartTextX, nBaseY, fStartTextX + fStartTextWidth + 2, MICROPROFILE_TEXT_HEIGHT + 2 + nBaseY, 0x33000000, MicroProfileBoxTypeFlat);
             MicroProfileDrawText(fStartTextX+1, nBaseY, UINT32_MAX, sBuffer, nLenStart);
-            uint32_t nLenEnd = snprintf(sBuffer, sizeof(sBuffer)-1, "%.2fms", fMsEnd);
+            uint32_t nLenEnd = snprintf(sBuffer, sizeof(sBuffer)-1, "%.2fms", fMsEnd3);
             MicroProfileDrawBox(fXEnd+1, nBaseY, fXEnd+1+(1+MICROPROFILE_TEXT_WIDTH) * nLenEnd + 3, MICROPROFILE_TEXT_HEIGHT + 2 + nBaseY, 0x33000000, MicroProfileBoxTypeFlat);
             MicroProfileDrawText(fXEnd+2, nBaseY+1, UINT32_MAX, sBuffer, nLenEnd);
 
@@ -1297,9 +1296,9 @@ inline void MicroProfileDrawDetailedBars(uint32_t nWidth, uint32_t nHeight, int
         if(UI.nRangeBeginGpu != UI.nRangeEndGpu)
         {
             float fMsStart = fToMsGpu * MicroProfileLogTickDifference(nBaseTicksGpu, UI.nRangeBeginGpu);
-            float fMsEnd = fToMsGpu * MicroProfileLogTickDifference(nBaseTicksGpu, UI.nRangeEndGpu);
+            float fMsEnd4 = fToMsGpu * MicroProfileLogTickDifference(nBaseTicksGpu, UI.nRangeEndGpu);
             float fXStart = fMsStart * fMsToScreen;
-            float fXEnd = fMsEnd * fMsToScreen;
+            float fXEnd = fMsEnd4 * fMsToScreen;
             MicroProfileDrawBox(fXStart, nBaseY, fXEnd, nHeight, MICROPROFILE_FRAME_COLOR_HIGHTLIGHT_GPU, MicroProfileBoxTypeFlat);
             MicroProfileDrawLineVertical(fXStart, nBaseY, nHeight, MICROPROFILE_FRAME_COLOR_HIGHTLIGHT_GPU | 0x44000000);
             MicroProfileDrawLineVertical(fXEnd, nBaseY, nHeight, MICROPROFILE_FRAME_COLOR_HIGHTLIGHT_GPU | 0x44000000);
@@ -1307,14 +1306,14 @@ inline void MicroProfileDrawDetailedBars(uint32_t nWidth, uint32_t nHeight, int
             nBaseY += MICROPROFILE_TEXT_HEIGHT+1;
 
             fMsStart += fDetailedOffset;
-            fMsEnd += fDetailedOffset;
+            fMsEnd4 += fDetailedOffset;
             char sBuffer[32];
             uint32_t nLenStart = snprintf(sBuffer, sizeof(sBuffer)-1, "%.2fms", fMsStart);
             float fStartTextWidth = (float)((1+MICROPROFILE_TEXT_WIDTH) * nLenStart);
             float fStartTextX = fXStart - fStartTextWidth - 2;
             MicroProfileDrawBox(fStartTextX, nBaseY, fStartTextX + fStartTextWidth + 2, MICROPROFILE_TEXT_HEIGHT + 2 + nBaseY, 0x33000000, MicroProfileBoxTypeFlat);
             MicroProfileDrawText(fStartTextX+1, nBaseY, UINT32_MAX, sBuffer, nLenStart);
-            uint32_t nLenEnd = snprintf(sBuffer, sizeof(sBuffer)-1, "%.2fms", fMsEnd);
+            uint32_t nLenEnd = snprintf(sBuffer, sizeof(sBuffer)-1, "%.2fms", fMsEnd4);
             MicroProfileDrawBox(fXEnd+1, nBaseY, fXEnd+1+(1+MICROPROFILE_TEXT_WIDTH) * nLenEnd + 3, MICROPROFILE_TEXT_HEIGHT + 2 + nBaseY, 0x33000000, MicroProfileBoxTypeFlat);
             MicroProfileDrawText(fXEnd+2, nBaseY+1, UINT32_MAX, sBuffer, nLenEnd);
         }
@@ -1716,8 +1715,8 @@ bool MicroProfileDrawGraph(uint32_t nScreenWidth, uint32_t nScreenHeight)
         uint32_t nTextCount = 0;
         uint32_t nGraphIndex = (S.nGraphPut + MICROPROFILE_GRAPH_HISTORY - int(MICROPROFILE_GRAPH_HISTORY*(1.f - fMouseXPrc))) % MICROPROFILE_GRAPH_HISTORY;
 
-        uint32_t nX = UI.nMouseX;
-        uint32_t nY = UI.nMouseY + 20;
+        uint32_t nMouseX = UI.nMouseX;
+        uint32_t nMouseY = UI.nMouseY + 20;
 
         for(uint32_t i = 0; i < MICROPROFILE_MAX_GRAPHS; ++i)
         {
@@ -1736,7 +1735,7 @@ bool MicroProfileDrawGraph(uint32_t nScreenWidth, uint32_t nScreenHeight)
         }
         if(nTextCount)
         {
-            MicroProfileDrawFloatWindow(nX, nY, Strings.ppStrings, Strings.nNumStrings, 0, pColors);
+            MicroProfileDrawFloatWindow(nMouseX, nMouseY, Strings.ppStrings, Strings.nNumStrings, 0, pColors);
         }
 
         if(UI.nMouseRight)
@@ -2321,8 +2320,8 @@ inline void MicroProfileDrawMenu(uint32_t nWidth, uint32_t nHeight)
     uint32_t    nMenuX[MICROPROFILE_MENU_MAX] = {0};
     uint32_t nNumMenuItems = 0;
 
-    int nLen = snprintf(buffer, 127, "MicroProfile");
-    MicroProfileDrawText(nX, nY, UINT32_MAX, buffer, nLen);
+    int nMPTextLen = snprintf(buffer, 127, "MicroProfile");
+    MicroProfileDrawText(nX, nY, UINT32_MAX, buffer, nMPTextLen);
     nX += (sizeof("MicroProfile")+2) * (MICROPROFILE_TEXT_WIDTH+1);
     pMenuText[nNumMenuItems++] = "Mode";
     pMenuText[nNumMenuItems++] = "Groups";
@@ -2438,16 +2437,16 @@ inline void MicroProfileDrawMenu(uint32_t nWidth, uint32_t nHeight)
         int nNumLines = 0;
         bool bSelected = false;
         const char* pString = CB(nNumLines, &bSelected);
-        uint32_t nWidth = 0, nHeight = 0;
+        uint32_t nTextWidth = 0, nTextHeight = 0;
         while(pString)
         {
-            nWidth = MicroProfileMax<int>(nWidth, (int)strlen(pString));
+            nTextWidth = MicroProfileMax<int>(nTextWidth, (int)strlen(pString));
             nNumLines++;
             pString = CB(nNumLines, &bSelected);
         }
-        nWidth = (2+nWidth) * (MICROPROFILE_TEXT_WIDTH+1);
-        nHeight = nNumLines * (MICROPROFILE_TEXT_HEIGHT+1);
-        if(UI.nMouseY <= nY + nHeight+0 && UI.nMouseY >= nY-0 && UI.nMouseX <= nX + nWidth + 0 && UI.nMouseX >= nX - 0)
+        nTextWidth = (2+nTextWidth) * (MICROPROFILE_TEXT_WIDTH+1);
+        nTextHeight = nNumLines * (MICROPROFILE_TEXT_HEIGHT+1);
+        if(UI.nMouseY <= nY + nTextHeight+0 && UI.nMouseY >= nY-0 && UI.nMouseX <= nX + nTextWidth + 0 && UI.nMouseX >= nX - 0)
         {
             UI.nActiveMenu = nMenu;
         }
@@ -2455,21 +2454,21 @@ inline void MicroProfileDrawMenu(uint32_t nWidth, uint32_t nHeight)
         {
             UI.nActiveMenu = UINT32_MAX;
         }
-        MicroProfileDrawBox(nX, nY, nX + nWidth, nY + nHeight, 0xff000000|g_nMicroProfileBackColors[1]);
+        MicroProfileDrawBox(nX, nY, nX + nTextWidth, nY + nTextHeight, 0xff000000|g_nMicroProfileBackColors[1]);
         for(int i = 0; i < nNumLines; ++i)
         {
-            bool bSelected = false;
-            const char* pString = CB(i, &bSelected);
+            bool bSelected2 = false;
+            const char* pString2 = CB(i, &bSelected2);
             if(UI.nMouseY >= nY && UI.nMouseY < nY + MICROPROFILE_TEXT_HEIGHT + 1)
             {
                 if(UI.nMouseLeft || UI.nMouseRight)
                 {
                     CBClick[nMenu](i);
                 }
-                MicroProfileDrawBox(nX, nY, nX + nWidth, nY + MICROPROFILE_TEXT_HEIGHT + 1, 0xff888888);
+                MicroProfileDrawBox(nX, nY, nX + nTextWidth, nY + MICROPROFILE_TEXT_HEIGHT + 1, 0xff888888);
             }
-            int nLen = snprintf(buffer, SBUF_SIZE-1, "%c %s", bSelected ? '*' : ' ' ,pString);
-            MicroProfileDrawText(nX, nY, UINT32_MAX, buffer, nLen);
+            int nTextLen = snprintf(buffer, SBUF_SIZE-1, "%c %s", bSelected2 ? '*' : ' ' ,pString2);
+            MicroProfileDrawText(nX, nY, UINT32_MAX, buffer, nTextLen);
             nY += MICROPROFILE_TEXT_HEIGHT+1;
         }
     }
@@ -2605,7 +2604,7 @@ inline void MicroProfileDrawCustom(uint32_t nWidth, uint32_t nHeight)
             for(uint32_t i = 0; i < nCount; ++i)
             {
                 nOffsetY += (1+MICROPROFILE_TEXT_HEIGHT);
-                uint32_t nWidth = MicroProfileMin(nMaxWidth, (uint32_t)(nMaxWidth * pMs[i] * fRcpReference));
+                nWidth = MicroProfileMin(nMaxWidth, (uint32_t)(nMaxWidth * pMs[i] * fRcpReference));
                 MicroProfileDrawBox(nMaxOffsetX, nOffsetY, nMaxOffsetX+nWidth, nOffsetY+MICROPROFILE_TEXT_HEIGHT, pColors[i]|0xff000000);
             }
         }

src/CMakeLists.txt

@@ -36,7 +36,6 @@ if (MSVC)
     # /GT                 - Supports fiber safety for data allocated using static thread-local storage
     add_compile_options(
         /MP
-        /Zi
         /Zm200
         /Zo
         /permissive-
@@ -65,6 +64,10 @@ if (MSVC)
         /we4305 # 'context': truncation from 'type1' to 'type2'
         /we4388 # 'expression': signed/unsigned mismatch
         /we4389 # 'operator': signed/unsigned mismatch
+        /we4456 # Declaration of 'identifier' hides previous local declaration
+        /we4457 # Declaration of 'identifier' hides function parameter
+        /we4458 # Declaration of 'identifier' hides class member
+        /we4459 # Declaration of 'identifier' hides global declaration
         /we4505 # 'function': unreferenced local function has been removed
         /we4547 # 'operator': operator before comma has no effect; expected operator with side-effect
         /we4549 # 'operator1': operator before comma has no effect; did you intend 'operator2'?
@@ -75,6 +78,13 @@ if (MSVC)
         /we5245 # 'function': unreferenced function with internal linkage has been removed
     )
 
+    if (USE_CCACHE)
+    # when caching, we need to use /Z7 to downgrade debug info to use an older but more cachable format
+        add_compile_options(/Z7)
+    else()
+        add_compile_options(/Zi)
+    endif()
+
     if (ARCHITECTURE_x86_64)
         add_compile_options(/QIntel-jcc-erratum)
     endif()
@@ -92,6 +102,7 @@ else()
         -Werror=missing-declarations
         -Werror=missing-field-initializers
         -Werror=reorder
+        -Werror=shadow
         -Werror=sign-compare
         -Werror=switch
         -Werror=uninitialized

src/audio_core/CMakeLists.txt

@@ -49,9 +49,6 @@ if (NOT MSVC)
     target_compile_options(audio_core PRIVATE
         -Werror=conversion
         -Werror=ignored-qualifiers
-        -Werror=shadow
-        -Werror=unused-parameter
-        -Werror=unused-variable
 
         $<$<CXX_COMPILER_ID:GNU>:-Werror=unused-but-set-parameter>
         $<$<CXX_COMPILER_ID:GNU>:-Werror=unused-but-set-variable>

src/audio_core/audio_renderer.cpp

@@ -98,13 +98,13 @@ AudioRenderer::AudioRenderer(Core::Timing::CoreTiming& core_timing_, Core::Memor
 
 AudioRenderer::~AudioRenderer() = default;
 
-ResultCode AudioRenderer::Start() {
+Result AudioRenderer::Start() {
     audio_out->StartStream(stream);
     ReleaseAndQueueBuffers();
     return ResultSuccess;
 }
 
-ResultCode AudioRenderer::Stop() {
+Result AudioRenderer::Stop() {
     audio_out->StopStream(stream);
     return ResultSuccess;
 }
@@ -125,8 +125,8 @@ Stream::State AudioRenderer::GetStreamState() const {
     return stream->GetState();
 }
 
-ResultCode AudioRenderer::UpdateAudioRenderer(const std::vector<u8>& input_params,
-                                              std::vector<u8>& output_params) {
+Result AudioRenderer::UpdateAudioRenderer(const std::vector<u8>& input_params,
+                                          std::vector<u8>& output_params) {
     std::scoped_lock lock{mutex};
     InfoUpdater info_updater{input_params, output_params, behavior_info};
 

src/audio_core/audio_renderer.h

@@ -43,10 +43,10 @@ public:
                   Stream::ReleaseCallback&& release_callback, std::size_t instance_number);
     ~AudioRenderer();
 
-    [[nodiscard]] ResultCode UpdateAudioRenderer(const std::vector<u8>& input_params,
-                                                 std::vector<u8>& output_params);
-    [[nodiscard]] ResultCode Start();
-    [[nodiscard]] ResultCode Stop();
+    [[nodiscard]] Result UpdateAudioRenderer(const std::vector<u8>& input_params,
+                                             std::vector<u8>& output_params);
+    [[nodiscard]] Result Start();
+    [[nodiscard]] Result Stop();
     void QueueMixedBuffer(Buffer::Tag tag);
     void ReleaseAndQueueBuffers();
     [[nodiscard]] u32 GetSampleRate() const;

src/audio_core/command_generator.cpp

@@ -129,17 +129,17 @@ s32 ToS32(float sample) {
     return static_cast<s32>(rescaled_sample);
 }
 
-constexpr std::array<std::size_t, 20> REVERB_TAP_INDEX_1CH{0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
-                                                           0, 0, 0, 0, 0, 0, 0, 0, 0, 0};
+constexpr std::array<u8, 20> REVERB_TAP_INDEX_1CH{0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+                                                  0, 0, 0, 0, 0, 0, 0, 0, 0, 0};
 
-constexpr std::array<std::size_t, 20> REVERB_TAP_INDEX_2CH{0, 0, 0, 1, 1, 1, 1, 0, 0, 0,
-                                                           1, 1, 1, 0, 0, 0, 0, 1, 1, 1};
+constexpr std::array<u8, 20> REVERB_TAP_INDEX_2CH{0, 0, 0, 1, 1, 1, 1, 0, 0, 0,
+                                                  1, 1, 1, 0, 0, 0, 0, 1, 1, 1};
 
-constexpr std::array<std::size_t, 20> REVERB_TAP_INDEX_4CH{0, 0, 0, 1, 1, 1, 1, 2, 2, 2,
-                                                           1, 1, 1, 0, 0, 0, 0, 3, 3, 3};
+constexpr std::array<u8, 20> REVERB_TAP_INDEX_4CH{0, 0, 0, 1, 1, 1, 1, 2, 2, 2,
+                                                  1, 1, 1, 0, 0, 0, 0, 3, 3, 3};
 
-constexpr std::array<std::size_t, 20> REVERB_TAP_INDEX_6CH{4, 0, 0, 1, 1, 1, 1, 2, 2, 2,
-                                                           1, 1, 1, 0, 0, 0, 0, 3, 3, 3};
+constexpr std::array<u8, 20> REVERB_TAP_INDEX_6CH{4, 0, 0, 1, 1, 1, 1, 2, 2, 2,
+                                                  1, 1, 1, 0, 0, 0, 0, 3, 3, 3};
 
 template <std::size_t CHANNEL_COUNT>
 void ApplyReverbGeneric(
@@ -429,7 +429,7 @@ void CommandGenerator::GenerateDataSourceCommand(ServerVoiceInfo& voice_info, Vo
                                   in_params.node_id);
             break;
         default:
-            UNREACHABLE_MSG("Unimplemented sample format={}", in_params.sample_format);
+            ASSERT_MSG(false, "Unimplemented sample format={}", in_params.sample_format);
         }
     }
 }
@@ -1312,7 +1312,7 @@ void CommandGenerator::DecodeFromWaveBuffers(ServerVoiceInfo& voice_info, std::s
                                 samples_to_read - samples_read, channel, temp_mix_offset);
                 break;
             default:
-                UNREACHABLE_MSG("Unimplemented sample format={}", in_params.sample_format);
+                ASSERT_MSG(false, "Unimplemented sample format={}", in_params.sample_format);
             }
 
             temp_mix_offset += samples_decoded;

src/audio_core/common.h

@@ -10,8 +10,8 @@
 
 namespace AudioCommon {
 namespace Audren {
-constexpr ResultCode ERR_INVALID_PARAMETERS{ErrorModule::Audio, 41};
-constexpr ResultCode ERR_SPLITTER_SORT_FAILED{ErrorModule::Audio, 43};
+constexpr Result ERR_INVALID_PARAMETERS{ErrorModule::Audio, 41};
+constexpr Result ERR_SPLITTER_SORT_FAILED{ErrorModule::Audio, 43};
 } // namespace Audren
 
 constexpr u8 BASE_REVISION = '0';

src/audio_core/effect_context.cpp

@@ -50,7 +50,7 @@ EffectBase* EffectContext::RetargetEffect(std::size_t i, EffectType effect) {
         effects[i] = std::make_unique<EffectBiquadFilter>();
         break;
     default:
-        UNREACHABLE_MSG("Unimplemented effect {}", effect);
+        ASSERT_MSG(false, "Unimplemented effect {}", effect);
         effects[i] = std::make_unique<EffectStubbed>();
     }
     return GetInfo(i);
@@ -104,7 +104,7 @@ void EffectI3dl2Reverb::Update(EffectInfo::InParams& in_params) {
     auto& params = GetParams();
     const auto* reverb_params = reinterpret_cast<I3dl2ReverbParams*>(in_params.raw.data());
     if (!ValidChannelCountForEffect(reverb_params->max_channels)) {
-        UNREACHABLE_MSG("Invalid reverb max channel count {}", reverb_params->max_channels);
+        ASSERT_MSG(false, "Invalid reverb max channel count {}", reverb_params->max_channels);
         return;
     }
 

src/audio_core/info_updater.cpp

@@ -285,9 +285,8 @@ bool InfoUpdater::UpdateSplitterInfo(SplitterContext& splitter_context) {
     return true;
 }
 
-ResultCode InfoUpdater::UpdateMixes(MixContext& mix_context, std::size_t mix_buffer_count,
-                                    SplitterContext& splitter_context,
-                                    EffectContext& effect_context) {
+Result InfoUpdater::UpdateMixes(MixContext& mix_context, std::size_t mix_buffer_count,
+                                SplitterContext& splitter_context, EffectContext& effect_context) {
     std::vector<MixInfo::InParams> mix_in_params;
 
     if (!behavior_info.IsMixInParameterDirtyOnlyUpdateSupported()) {

src/audio_core/info_updater.h

@@ -32,8 +32,8 @@ public:
                       VAddr audio_codec_dsp_addr);
     bool UpdateEffects(EffectContext& effect_context, bool is_active);
     bool UpdateSplitterInfo(SplitterContext& splitter_context);
-    ResultCode UpdateMixes(MixContext& mix_context, std::size_t mix_buffer_count,
-                           SplitterContext& splitter_context, EffectContext& effect_context);
+    Result UpdateMixes(MixContext& mix_context, std::size_t mix_buffer_count,
+                       SplitterContext& splitter_context, EffectContext& effect_context);
     bool UpdateSinks(SinkContext& sink_context);
     bool UpdatePerformanceBuffer();
     bool UpdateErrorInfo(BehaviorInfo& in_behavior_info);

src/audio_core/splitter_context.cpp

@@ -483,7 +483,7 @@ bool NodeStates::DepthFirstSearch(EdgeMatrix& edge_matrix) {
                     // Add more work
                     index_stack.push(j);
                 } else if (node_state == NodeStates::State::InFound) {
-                    UNREACHABLE_MSG("Node start marked as found");
+                    ASSERT_MSG(false, "Node start marked as found");
                     ResetState();
                     return false;
                 }

src/audio_core/voice_context.cpp

@@ -114,7 +114,7 @@ void ServerVoiceInfo::UpdateParameters(const VoiceInfo::InParams& voice_in,
         in_params.current_playstate = ServerPlayState::Play;
         break;
     default:
-        UNREACHABLE_MSG("Unknown playstate {}", voice_in.play_state);
+        ASSERT_MSG(false, "Unknown playstate {}", voice_in.play_state);
         break;
     }
 
@@ -410,7 +410,7 @@ bool ServerVoiceInfo::UpdateParametersForCommandGeneration(
         return in_params.should_depop;
     }
     default:
-        UNREACHABLE_MSG("Invalid playstate {}", in_params.current_playstate);
+        ASSERT_MSG(false, "Invalid playstate {}", in_params.current_playstate);
     }
 
     return false;

src/common/CMakeLists.txt

@@ -58,6 +58,7 @@ add_library(common STATIC
     div_ceil.h
     dynamic_library.cpp
     dynamic_library.h
+    elf.h
     error.cpp
     error.h
     expected.h

src/common/assert.cpp

@@ -6,8 +6,13 @@
 
 #include "common/settings.h"
 
-void assert_handle_failure() {
+void assert_fail_impl() {
     if (Settings::values.use_debug_asserts) {
         Crash();
     }
 }
+
+[[noreturn]] void unreachable_impl() {
+    Crash();
+    throw std::runtime_error("Unreachable code");
+}

src/common/assert.h

@@ -9,44 +9,43 @@
 // Sometimes we want to try to continue even after hitting an assert.
 // However touching this file yields a global recompilation as this header is included almost
 // everywhere. So let's just move the handling of the failed assert to a single cpp file.
-void assert_handle_failure();
 
-// For asserts we'd like to keep all the junk executed when an assert happens away from the
-// important code in the function. One way of doing this is to put all the relevant code inside a
-// lambda and force the compiler to not inline it. Unfortunately, MSVC seems to have no syntax to
-// specify __declspec on lambda functions, so what we do instead is define a noinline wrapper
-// template that calls the lambda. This seems to generate an extra instruction at the call-site
-// compared to the ideal implementation (which wouldn't support ASSERT_MSG parameters), but is good
-// enough for our purposes.
-template <typename Fn>
-#if defined(_MSC_VER)
-[[msvc::noinline]]
-#elif defined(__GNUC__)
-[[gnu::cold, gnu::noinline]]
+void assert_fail_impl();
+[[noreturn]] void unreachable_impl();
+
+#ifdef _MSC_VER
+#define YUZU_NO_INLINE __declspec(noinline)
+#else
+#define YUZU_NO_INLINE __attribute__((noinline))
 #endif
-static void
-assert_noinline_call(const Fn& fn) {
-    fn();
-    assert_handle_failure();
-}
 
 #define ASSERT(_a_)                                                                                \
-    do                                                                                             \
-        if (!(_a_)) {                                                                              \
-            assert_noinline_call([] { LOG_CRITICAL(Debug, "Assertion Failed!"); });                \
+    ([&]() YUZU_NO_INLINE {                                                                        \
+        if (!(_a_)) [[unlikely]] {                                                                 \
+            LOG_CRITICAL(Debug, "Assertion Failed!");                                              \
+            assert_fail_impl();                                                                    \
         }                                                                                          \
-    while (0)
+    }())
 
 #define ASSERT_MSG(_a_, ...)                                                                       \
-    do                                                                                             \
-        if (!(_a_)) {                                                                              \
-            assert_noinline_call([&] { LOG_CRITICAL(Debug, "Assertion Failed!\n" __VA_ARGS__); }); \
+    ([&]() YUZU_NO_INLINE {                                                                        \
+        if (!(_a_)) [[unlikely]] {                                                                 \
+            LOG_CRITICAL(Debug, "Assertion Failed!\n" __VA_ARGS__);                                \
+            assert_fail_impl();                                                                    \
         }                                                                                          \
-    while (0)
+    }())
+
+#define UNREACHABLE()                                                                              \
+    do {                                                                                           \
+        LOG_CRITICAL(Debug, "Unreachable code!");                                                  \
+        unreachable_impl();                                                                        \
+    } while (0)
 
-#define UNREACHABLE() assert_noinline_call([] { LOG_CRITICAL(Debug, "Unreachable code!"); })
 #define UNREACHABLE_MSG(...)                                                                       \
-    assert_noinline_call([&] { LOG_CRITICAL(Debug, "Unreachable code!\n" __VA_ARGS__); })
+    do {                                                                                           \
+        LOG_CRITICAL(Debug, "Unreachable code!\n" __VA_ARGS__);                                    \
+        unreachable_impl();                                                                        \
+    } while (0)
 
 #ifdef _DEBUG
 #define DEBUG_ASSERT(_a_) ASSERT(_a_)

src/common/bounded_threadsafe_queue.h

@@ -0,0 +1,167 @@
+// SPDX-FileCopyrightText: Copyright (c) 2020 Erik Rigtorp <erik@rigtorp.se>
+// SPDX-License-Identifier: MIT
+
+#pragma once
+
+#include <atomic>
+#include <bit>
+#include <condition_variable>
+#include <memory>
+#include <mutex>
+#include <new>
+#include <stop_token>
+#include <type_traits>
+#include <utility>
+
+namespace Common {
+
+#if defined(__cpp_lib_hardware_interference_size)
+constexpr size_t hardware_interference_size = std::hardware_destructive_interference_size;
+#else
+constexpr size_t hardware_interference_size = 64;
+#endif
+
+#ifdef _MSC_VER
+#pragma warning(push)
+#pragma warning(disable : 4324)
+#endif
+
+template <typename T, size_t capacity = 0x400>
+class MPSCQueue {
+public:
+    explicit MPSCQueue() : allocator{std::allocator<Slot<T>>()} {
+        // Allocate one extra slot to prevent false sharing on the last slot
+        slots = allocator.allocate(capacity + 1);
+        // Allocators are not required to honor alignment for over-aligned types
+        // (see http://eel.is/c++draft/allocator.requirements#10) so we verify
+        // alignment here
+        if (reinterpret_cast<uintptr_t>(slots) % alignof(Slot<T>) != 0) {
+            allocator.deallocate(slots, capacity + 1);
+            throw std::bad_alloc();
+        }
+        for (size_t i = 0; i < capacity; ++i) {
+            std::construct_at(&slots[i]);
+        }
+        static_assert(std::has_single_bit(capacity), "capacity must be an integer power of 2");
+        static_assert(alignof(Slot<T>) == hardware_interference_size,
+                      "Slot must be aligned to cache line boundary to prevent false sharing");
+        static_assert(sizeof(Slot<T>) % hardware_interference_size == 0,
+                      "Slot size must be a multiple of cache line size to prevent "
+                      "false sharing between adjacent slots");
+        static_assert(sizeof(MPSCQueue) % hardware_interference_size == 0,
+                      "Queue size must be a multiple of cache line size to "
+                      "prevent false sharing between adjacent queues");
+    }
+
+    ~MPSCQueue() noexcept {
+        for (size_t i = 0; i < capacity; ++i) {
+            std::destroy_at(&slots[i]);
+        }
+        allocator.deallocate(slots, capacity + 1);
+    }
+
+    // The queue must be both non-copyable and non-movable
+    MPSCQueue(const MPSCQueue&) = delete;
+    MPSCQueue& operator=(const MPSCQueue&) = delete;
+
+    MPSCQueue(MPSCQueue&&) = delete;
+    MPSCQueue& operator=(MPSCQueue&&) = delete;
+
+    void Push(const T& v) noexcept {
+        static_assert(std::is_nothrow_copy_constructible_v<T>,
+                      "T must be nothrow copy constructible");
+        emplace(v);
+    }
+
+    template <typename P, typename = std::enable_if_t<std::is_nothrow_constructible_v<T, P&&>>>
+    void Push(P&& v) noexcept {
+        emplace(std::forward<P>(v));
+    }
+
+    void Pop(T& v, std::stop_token stop) noexcept {
+        auto const tail = tail_.fetch_add(1);
+        auto& slot = slots[idx(tail)];
+        if (!slot.turn.test()) {
+            std::unique_lock lock{cv_mutex};
+            cv.wait(lock, stop, [&slot] { return slot.turn.test(); });
+        }
+        v = slot.move();
+        slot.destroy();
+        slot.turn.clear();
+        slot.turn.notify_one();
+    }
+
+private:
+    template <typename U = T>
+    struct Slot {
+        ~Slot() noexcept {
+            if (turn.test()) {
+                destroy();
+            }
+        }
+
+        template <typename... Args>
+        void construct(Args&&... args) noexcept {
+            static_assert(std::is_nothrow_constructible_v<U, Args&&...>,
+                          "T must be nothrow constructible with Args&&...");
+            std::construct_at(reinterpret_cast<U*>(&storage), std::forward<Args>(args)...);
+        }
+
+        void destroy() noexcept {
+            static_assert(std::is_nothrow_destructible_v<U>, "T must be nothrow destructible");
+            std::destroy_at(reinterpret_cast<U*>(&storage));
+        }
+
+        U&& move() noexcept {
+            return reinterpret_cast<U&&>(storage);
+        }
+
+        // Align to avoid false sharing between adjacent slots
+        alignas(hardware_interference_size) std::atomic_flag turn{};
+        struct aligned_store {
+            struct type {
+                alignas(U) unsigned char data[sizeof(U)];
+            };
+        };
+        typename aligned_store::type storage;
+    };
+
+    template <typename... Args>
+    void emplace(Args&&... args) noexcept {
+        static_assert(std::is_nothrow_constructible_v<T, Args&&...>,
+                      "T must be nothrow constructible with Args&&...");
+        auto const head = head_.fetch_add(1);
+        auto& slot = slots[idx(head)];
+        slot.turn.wait(true);
+        slot.construct(std::forward<Args>(args)...);
+        slot.turn.test_and_set();
+        cv.notify_one();
+    }
+
+    constexpr size_t idx(size_t i) const noexcept {
+        return i & mask;
+    }
+
+    static constexpr size_t mask = capacity - 1;
+
+    // Align to avoid false sharing between head_ and tail_
+    alignas(hardware_interference_size) std::atomic<size_t> head_{0};
+    alignas(hardware_interference_size) std::atomic<size_t> tail_{0};
+
+    std::mutex cv_mutex;
+    std::condition_variable_any cv;
+
+    Slot<T>* slots;
+    [[no_unique_address]] std::allocator<Slot<T>> allocator;
+
+    static_assert(std::is_nothrow_copy_assignable_v<T> || std::is_nothrow_move_assignable_v<T>,
+                  "T must be nothrow copy or move assignable");
+
+    static_assert(std::is_nothrow_destructible_v<T>, "T must be nothrow destructible");
+};
+
+#ifdef _MSC_VER
+#pragma warning(pop)
+#endif
+
+} // namespace Common

src/common/detached_tasks.cpp

@@ -33,9 +33,9 @@ void DetachedTasks::AddTask(std::function<void()> task) {
     ++instance->count;
     std::thread([task{std::move(task)}]() {
         task();
-        std::unique_lock lock{instance->mutex};
+        std::unique_lock thread_lock{instance->mutex};
         --instance->count;
-        std::notify_all_at_thread_exit(instance->cv, std::move(lock));
+        std::notify_all_at_thread_exit(instance->cv, std::move(thread_lock));
     }).detach();
 }
 

src/common/elf.h

@@ -0,0 +1,333 @@
+// SPDX-FileCopyrightText: 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <array>
+#include <cstddef>
+
+#include "common_types.h"
+
+namespace Common {
+namespace ELF {
+
+/* Type for a 16-bit quantity.  */
+using Elf32_Half = u16;
+using Elf64_Half = u16;
+
+/* Types for signed and unsigned 32-bit quantities.  */
+using Elf32_Word = u32;
+using Elf32_Sword = s32;
+using Elf64_Word = u32;
+using Elf64_Sword = s32;
+
+/* Types for signed and unsigned 64-bit quantities.  */
+using Elf32_Xword = u64;
+using Elf32_Sxword = s64;
+using Elf64_Xword = u64;
+using Elf64_Sxword = s64;
+
+/* Type of addresses.  */
+using Elf32_Addr = u32;
+using Elf64_Addr = u64;
+
+/* Type of file offsets.  */
+using Elf32_Off = u32;
+using Elf64_Off = u64;
+
+/* Type for section indices, which are 16-bit quantities.  */
+using Elf32_Section = u16;
+using Elf64_Section = u16;
+
+/* Type for version symbol information.  */
+using Elf32_Versym = Elf32_Half;
+using Elf64_Versym = Elf64_Half;
+
+constexpr size_t ElfIdentSize = 16;
+
+/* The ELF file header.  This appears at the start of every ELF file.  */
+
+struct Elf32_Ehdr {
+    std::array<u8, ElfIdentSize> e_ident; /* Magic number and other info */
+    Elf32_Half e_type;                    /* Object file type */
+    Elf32_Half e_machine;                 /* Architecture */
+    Elf32_Word e_version;                 /* Object file version */
+    Elf32_Addr e_entry;                   /* Entry point virtual address */
+    Elf32_Off e_phoff;                    /* Program header table file offset */
+    Elf32_Off e_shoff;                    /* Section header table file offset */
+    Elf32_Word e_flags;                   /* Processor-specific flags */
+    Elf32_Half e_ehsize;                  /* ELF header size in bytes */
+    Elf32_Half e_phentsize;               /* Program header table entry size */
+    Elf32_Half e_phnum;                   /* Program header table entry count */
+    Elf32_Half e_shentsize;               /* Section header table entry size */
+    Elf32_Half e_shnum;                   /* Section header table entry count */
+    Elf32_Half e_shstrndx;                /* Section header string table index */
+};
+
+struct Elf64_Ehdr {
+    std::array<u8, ElfIdentSize> e_ident; /* Magic number and other info */
+    Elf64_Half e_type;                    /* Object file type */
+    Elf64_Half e_machine;                 /* Architecture */
+    Elf64_Word e_version;                 /* Object file version */
+    Elf64_Addr e_entry;                   /* Entry point virtual address */
+    Elf64_Off e_phoff;                    /* Program header table file offset */
+    Elf64_Off e_shoff;                    /* Section header table file offset */
+    Elf64_Word e_flags;                   /* Processor-specific flags */
+    Elf64_Half e_ehsize;                  /* ELF header size in bytes */
+    Elf64_Half e_phentsize;               /* Program header table entry size */
+    Elf64_Half e_phnum;                   /* Program header table entry count */
+    Elf64_Half e_shentsize;               /* Section header table entry size */
+    Elf64_Half e_shnum;                   /* Section header table entry count */
+    Elf64_Half e_shstrndx;                /* Section header string table index */
+};
+
+constexpr u8 ElfClass32 = 1;        /* 32-bit objects */
+constexpr u8 ElfClass64 = 2;        /* 64-bit objects */
+constexpr u8 ElfData2Lsb = 1;       /* 2's complement, little endian */
+constexpr u8 ElfVersionCurrent = 1; /* EV_CURRENT */
+constexpr u8 ElfOsAbiNone = 0;      /* System V ABI */
+
+constexpr u16 ElfTypeNone = 0; /* No file type */
+constexpr u16 ElfTypeRel = 0;  /* Relocatable file */
+constexpr u16 ElfTypeExec = 0; /* Executable file */
+constexpr u16 ElfTypeDyn = 0;  /* Shared object file */
+
+constexpr u16 ElfMachineArm = 40;      /* ARM */
+constexpr u16 ElfMachineAArch64 = 183; /* ARM AARCH64 */
+
+constexpr std::array<u8, ElfIdentSize> Elf32Ident{
+    0x7f, 'E', 'L', 'F', ElfClass32, ElfData2Lsb, ElfVersionCurrent, ElfOsAbiNone};
+
+constexpr std::array<u8, ElfIdentSize> Elf64Ident{
+    0x7f, 'E', 'L', 'F', ElfClass64, ElfData2Lsb, ElfVersionCurrent, ElfOsAbiNone};
+
+/* Section header.  */
+
+struct Elf32_Shdr {
+    Elf32_Word sh_name;      /* Section name (string tbl index) */
+    Elf32_Word sh_type;      /* Section type */
+    Elf32_Word sh_flags;     /* Section flags */
+    Elf32_Addr sh_addr;      /* Section virtual addr at execution */
+    Elf32_Off sh_offset;     /* Section file offset */
+    Elf32_Word sh_size;      /* Section size in bytes */
+    Elf32_Word sh_link;      /* Link to another section */
+    Elf32_Word sh_info;      /* Additional section information */
+    Elf32_Word sh_addralign; /* Section alignment */
+    Elf32_Word sh_entsize;   /* Entry size if section holds table */
+};
+
+struct Elf64_Shdr {
+    Elf64_Word sh_name;       /* Section name (string tbl index) */
+    Elf64_Word sh_type;       /* Section type */
+    Elf64_Xword sh_flags;     /* Section flags */
+    Elf64_Addr sh_addr;       /* Section virtual addr at execution */
+    Elf64_Off sh_offset;      /* Section file offset */
+    Elf64_Xword sh_size;      /* Section size in bytes */
+    Elf64_Word sh_link;       /* Link to another section */
+    Elf64_Word sh_info;       /* Additional section information */
+    Elf64_Xword sh_addralign; /* Section alignment */
+    Elf64_Xword sh_entsize;   /* Entry size if section holds table */
+};
+
+constexpr u32 ElfShnUndef = 0; /* Undefined section */
+
+constexpr u32 ElfShtNull = 0;     /* Section header table entry unused */
+constexpr u32 ElfShtProgBits = 1; /* Program data */
+constexpr u32 ElfShtSymtab = 2;   /* Symbol table */
+constexpr u32 ElfShtStrtab = 3;   /* String table */
+constexpr u32 ElfShtRela = 4;     /* Relocation entries with addends */
+constexpr u32 ElfShtDynamic = 6;  /* Dynamic linking information */
+constexpr u32 ElfShtNobits = 7;   /* Program space with no data (bss) */
+constexpr u32 ElfShtRel = 9;      /* Relocation entries, no addends */
+constexpr u32 ElfShtDynsym = 11;  /* Dynamic linker symbol table */
+
+/* Symbol table entry.  */
+
+struct Elf32_Sym {
+    Elf32_Word st_name;     /* Symbol name (string tbl index) */
+    Elf32_Addr st_value;    /* Symbol value */
+    Elf32_Word st_size;     /* Symbol size */
+    u8 st_info;             /* Symbol type and binding */
+    u8 st_other;            /* Symbol visibility */
+    Elf32_Section st_shndx; /* Section index */
+};
+
+struct Elf64_Sym {
+    Elf64_Word st_name;     /* Symbol name (string tbl index) */
+    u8 st_info;             /* Symbol type and binding */
+    u8 st_other;            /* Symbol visibility */
+    Elf64_Section st_shndx; /* Section index */
+    Elf64_Addr st_value;    /* Symbol value */
+    Elf64_Xword st_size;    /* Symbol size */
+};
+
+/* How to extract and insert information held in the st_info field.  */
+
+static inline u8 ElfStBind(u8 st_info) {
+    return st_info >> 4;
+}
+static inline u8 ElfStType(u8 st_info) {
+    return st_info & 0xf;
+}
+static inline u8 ElfStInfo(u8 st_bind, u8 st_type) {
+    return static_cast<u8>((st_bind << 4) + (st_type & 0xf));
+}
+
+constexpr u8 ElfBindLocal = 0;  /* Local symbol */
+constexpr u8 ElfBindGlobal = 1; /* Global symbol */
+constexpr u8 ElfBindWeak = 2;   /* Weak symbol */
+
+constexpr u8 ElfTypeUnspec = 0; /* Symbol type is unspecified */
+constexpr u8 ElfTypeObject = 1; /* Symbol is a data object */
+constexpr u8 ElfTypeFunc = 2;   /* Symbol is a code object */
+
+static inline u8 ElfStVisibility(u8 st_other) {
+    return static_cast<u8>(st_other & 0x3);
+}
+
+constexpr u8 ElfVisibilityDefault = 0;   /* Default symbol visibility rules */
+constexpr u8 ElfVisibilityInternal = 1;  /* Processor specific hidden class */
+constexpr u8 ElfVisibilityHidden = 2;    /* Sym unavailable in other modules */
+constexpr u8 ElfVisibilityProtected = 3; /* Not preemptible, not exported */
+
+/* Relocation table entry without addend (in section of type ShtRel).  */
+
+struct Elf32_Rel {
+    Elf32_Addr r_offset; /* Address */
+    Elf32_Word r_info;   /* Relocation type and symbol index */
+};
+
+/* Relocation table entry with addend (in section of type ShtRela).  */
+
+struct Elf32_Rela {
+    Elf32_Addr r_offset;  /* Address */
+    Elf32_Word r_info;    /* Relocation type and symbol index */
+    Elf32_Sword r_addend; /* Addend */
+};
+
+struct Elf64_Rela {
+    Elf64_Addr r_offset;   /* Address */
+    Elf64_Xword r_info;    /* Relocation type and symbol index */
+    Elf64_Sxword r_addend; /* Addend */
+};
+
+/* How to extract and insert information held in the r_info field.  */
+
+static inline u32 Elf32RelSymIndex(Elf32_Word r_info) {
+    return r_info >> 8;
+}
+static inline u8 Elf32RelType(Elf32_Word r_info) {
+    return static_cast<u8>(r_info & 0xff);
+}
+static inline Elf32_Word Elf32RelInfo(u32 sym_index, u8 type) {
+    return (sym_index << 8) + type;
+}
+static inline u32 Elf64RelSymIndex(Elf64_Xword r_info) {
+    return static_cast<u32>(r_info >> 32);
+}
+static inline u32 Elf64RelType(Elf64_Xword r_info) {
+    return r_info & 0xffffffff;
+}
+static inline Elf64_Xword Elf64RelInfo(u32 sym_index, u32 type) {
+    return (static_cast<Elf64_Xword>(sym_index) << 32) + type;
+}
+
+constexpr u32 ElfArmCopy = 20;     /* Copy symbol at runtime */
+constexpr u32 ElfArmGlobDat = 21;  /* Create GOT entry */
+constexpr u32 ElfArmJumpSlot = 22; /* Create PLT entry */
+constexpr u32 ElfArmRelative = 23; /* Adjust by program base */
+
+constexpr u32 ElfAArch64Copy = 1024;     /* Copy symbol at runtime */
+constexpr u32 ElfAArch64GlobDat = 1025;  /* Create GOT entry */
+constexpr u32 ElfAArch64JumpSlot = 1026; /* Create PLT entry */
+constexpr u32 ElfAArch64Relative = 1027; /* Adjust by program base */
+
+/* Program segment header.  */
+
+struct Elf32_Phdr {
+    Elf32_Word p_type;   /* Segment type */
+    Elf32_Off p_offset;  /* Segment file offset */
+    Elf32_Addr p_vaddr;  /* Segment virtual address */
+    Elf32_Addr p_paddr;  /* Segment physical address */
+    Elf32_Word p_filesz; /* Segment size in file */
+    Elf32_Word p_memsz;  /* Segment size in memory */
+    Elf32_Word p_flags;  /* Segment flags */
+    Elf32_Word p_align;  /* Segment alignment */
+};
+
+struct Elf64_Phdr {
+    Elf64_Word p_type;    /* Segment type */
+    Elf64_Word p_flags;   /* Segment flags */
+    Elf64_Off p_offset;   /* Segment file offset */
+    Elf64_Addr p_vaddr;   /* Segment virtual address */
+    Elf64_Addr p_paddr;   /* Segment physical address */
+    Elf64_Xword p_filesz; /* Segment size in file */
+    Elf64_Xword p_memsz;  /* Segment size in memory */
+    Elf64_Xword p_align;  /* Segment alignment */
+};
+
+/* Legal values for p_type (segment type).  */
+
+constexpr u32 ElfPtNull = 0;    /* Program header table entry unused */
+constexpr u32 ElfPtLoad = 1;    /* Loadable program segment */
+constexpr u32 ElfPtDynamic = 2; /* Dynamic linking information */
+constexpr u32 ElfPtInterp = 3;  /* Program interpreter */
+constexpr u32 ElfPtNote = 4;    /* Auxiliary information */
+constexpr u32 ElfPtPhdr = 6;    /* Entry for header table itself */
+constexpr u32 ElfPtTls = 7;     /* Thread-local storage segment */
+
+/* Legal values for p_flags (segment flags).  */
+
+constexpr u32 ElfPfExec = 0;  /* Segment is executable */
+constexpr u32 ElfPfWrite = 1; /* Segment is writable */
+constexpr u32 ElfPfRead = 2;  /* Segment is readable */
+
+/* Dynamic section entry.  */
+
+struct Elf32_Dyn {
+    Elf32_Sword d_tag; /* Dynamic entry type */
+    union {
+        Elf32_Word d_val; /* Integer value */
+        Elf32_Addr d_ptr; /* Address value */
+    } d_un;
+};
+
+struct Elf64_Dyn {
+    Elf64_Sxword d_tag; /* Dynamic entry type */
+    union {
+        Elf64_Xword d_val; /* Integer value */
+        Elf64_Addr d_ptr;  /* Address value */
+    } d_un;
+};
+
+/* Legal values for d_tag (dynamic entry type).  */
+
+constexpr u32 ElfDtNull = 0;         /* Marks end of dynamic section */
+constexpr u32 ElfDtNeeded = 1;       /* Name of needed library */
+constexpr u32 ElfDtPltRelSz = 2;     /* Size in bytes of PLT relocs */
+constexpr u32 ElfDtPltGot = 3;       /* Processor defined value */
+constexpr u32 ElfDtHash = 4;         /* Address of symbol hash table */
+constexpr u32 ElfDtStrtab = 5;       /* Address of string table */
+constexpr u32 ElfDtSymtab = 6;       /* Address of symbol table */
+constexpr u32 ElfDtRela = 7;         /* Address of Rela relocs */
+constexpr u32 ElfDtRelasz = 8;       /* Total size of Rela relocs */
+constexpr u32 ElfDtRelaent = 9;      /* Size of one Rela reloc */
+constexpr u32 ElfDtStrsz = 10;       /* Size of string table */
+constexpr u32 ElfDtSyment = 11;      /* Size of one symbol table entry */
+constexpr u32 ElfDtInit = 12;        /* Address of init function */
+constexpr u32 ElfDtFini = 13;        /* Address of termination function */
+constexpr u32 ElfDtRel = 17;         /* Address of Rel relocs */
+constexpr u32 ElfDtRelsz = 18;       /* Total size of Rel relocs */
+constexpr u32 ElfDtRelent = 19;      /* Size of one Rel reloc */
+constexpr u32 ElfDtPltRel = 20;      /* Type of reloc in PLT */
+constexpr u32 ElfDtTextRel = 22;     /* Reloc might modify .text */
+constexpr u32 ElfDtJmpRel = 23;      /* Address of PLT relocs */
+constexpr u32 ElfDtBindNow = 24;     /* Process relocations of object */
+constexpr u32 ElfDtInitArray = 25;   /* Array with addresses of init fct */
+constexpr u32 ElfDtFiniArray = 26;   /* Array with addresses of fini fct */
+constexpr u32 ElfDtInitArraySz = 27; /* Size in bytes of DT_INIT_ARRAY */
+constexpr u32 ElfDtFiniArraySz = 28; /* Size in bytes of DT_FINI_ARRAY */
+constexpr u32 ElfDtSymtabShndx = 34; /* Address of SYMTAB_SHNDX section */
+
+} // namespace ELF
+} // namespace Common

src/common/fiber.cpp

@@ -20,10 +20,8 @@ struct Fiber::FiberImpl {
     VirtualBuffer<u8> rewind_stack;
 
     std::mutex guard;
-    std::function<void(void*)> entry_point;
-    std::function<void(void*)> rewind_point;
-    void* rewind_parameter{};
-    void* start_parameter{};
+    std::function<void()> entry_point;
+    std::function<void()> rewind_point;
     std::shared_ptr<Fiber> previous_fiber;
     bool is_thread_fiber{};
     bool released{};
@@ -34,13 +32,8 @@ struct Fiber::FiberImpl {
     boost::context::detail::fcontext_t rewind_context{};
 };
 
-void Fiber::SetStartParameter(void* new_parameter) {
-    impl->start_parameter = new_parameter;
-}
-
-void Fiber::SetRewindPoint(std::function<void(void*)>&& rewind_func, void* rewind_param) {
+void Fiber::SetRewindPoint(std::function<void()>&& rewind_func) {
     impl->rewind_point = std::move(rewind_func);
-    impl->rewind_parameter = rewind_param;
 }
 
 void Fiber::Start(boost::context::detail::transfer_t& transfer) {
@@ -48,7 +41,7 @@ void Fiber::Start(boost::context::detail::transfer_t& transfer) {
     impl->previous_fiber->impl->context = transfer.fctx;
     impl->previous_fiber->impl->guard.unlock();
     impl->previous_fiber.reset();
-    impl->entry_point(impl->start_parameter);
+    impl->entry_point();
     UNREACHABLE();
 }
 
@@ -59,7 +52,7 @@ void Fiber::OnRewind([[maybe_unused]] boost::context::detail::transfer_t& transf
     u8* tmp = impl->stack_limit;
     impl->stack_limit = impl->rewind_stack_limit;
     impl->rewind_stack_limit = tmp;
-    impl->rewind_point(impl->rewind_parameter);
+    impl->rewind_point();
     UNREACHABLE();
 }
 
@@ -73,10 +66,8 @@ void Fiber::RewindStartFunc(boost::context::detail::transfer_t transfer) {
     fiber->OnRewind(transfer);
 }
 
-Fiber::Fiber(std::function<void(void*)>&& entry_point_func, void* start_parameter)
-    : impl{std::make_unique<FiberImpl>()} {
+Fiber::Fiber(std::function<void()>&& entry_point_func) : impl{std::make_unique<FiberImpl>()} {
     impl->entry_point = std::move(entry_point_func);
-    impl->start_parameter = start_parameter;
     impl->stack_limit = impl->stack.data();
     impl->rewind_stack_limit = impl->rewind_stack.data();
     u8* stack_base = impl->stack_limit + default_stack_size;

src/common/fiber.h

@@ -29,7 +29,7 @@ namespace Common {
  */
 class Fiber {
 public:
-    Fiber(std::function<void(void*)>&& entry_point_func, void* start_parameter);
+    Fiber(std::function<void()>&& entry_point_func);
     ~Fiber();
 
     Fiber(const Fiber&) = delete;
@@ -43,16 +43,13 @@ public:
     static void YieldTo(std::weak_ptr<Fiber> weak_from, Fiber& to);
     [[nodiscard]] static std::shared_ptr<Fiber> ThreadToFiber();
 
-    void SetRewindPoint(std::function<void(void*)>&& rewind_func, void* rewind_param);
+    void SetRewindPoint(std::function<void()>&& rewind_func);
 
     void Rewind();
 
     /// Only call from main thread's fiber
     void Exit();
 
-    /// Changes the start parameter of the fiber. Has no effect if the fiber already started
-    void SetStartParameter(void* new_parameter);
-
 private:
     Fiber();
 

src/common/fs/path_util.cpp

@@ -232,9 +232,7 @@ void SetYuzuPath(YuzuPath yuzu_path, const fs::path& new_path) {
 fs::path GetExeDirectory() {
     wchar_t exe_path[MAX_PATH];
 
-    GetModuleFileNameW(nullptr, exe_path, MAX_PATH);
-
-    if (!exe_path) {
+    if (GetModuleFileNameW(nullptr, exe_path, MAX_PATH) == 0) {
         LOG_ERROR(Common_Filesystem,
                   "Failed to get the path to the executable of the current process");
     }

src/common/input.h

@@ -72,6 +72,7 @@ enum class PollingError {
 enum class VibrationAmplificationType {
     Linear,
     Exponential,
+    Test,
 };
 
 // Analog properties for calibration

src/common/page_table.h

@@ -15,6 +15,9 @@ enum class PageType : u8 {
     Unmapped,
     /// Page is mapped to regular memory. This is the only type you can get pointers to.
     Memory,
+    /// Page is mapped to regular memory, but inaccessible from CPU fastmem and must use
+    /// the callbacks.
+    DebugMemory,
     /// Page is mapped to regular memory, but also needs to check for rasterizer cache flushing and
     /// invalidation
     RasterizerCachedMemory,

src/common/param_package.cpp

@@ -76,7 +76,7 @@ std::string ParamPackage::Serialize() const {
 std::string ParamPackage::Get(const std::string& key, const std::string& default_value) const {
     auto pair = data.find(key);
     if (pair == data.end()) {
-        LOG_DEBUG(Common, "key '{}' not found", key);
+        LOG_TRACE(Common, "key '{}' not found", key);
         return default_value;
     }
 
@@ -86,7 +86,7 @@ std::string ParamPackage::Get(const std::string& key, const std::string& default
 int ParamPackage::Get(const std::string& key, int default_value) const {
     auto pair = data.find(key);
     if (pair == data.end()) {
-        LOG_DEBUG(Common, "key '{}' not found", key);
+        LOG_TRACE(Common, "key '{}' not found", key);
         return default_value;
     }
 
@@ -101,7 +101,7 @@ int ParamPackage::Get(const std::string& key, int default_value) const {
 float ParamPackage::Get(const std::string& key, float default_value) const {
     auto pair = data.find(key);
     if (pair == data.end()) {
-        LOG_DEBUG(Common, "key {} not found", key);
+        LOG_TRACE(Common, "key {} not found", key);
         return default_value;
     }
 

src/common/settings.cpp

@@ -70,6 +70,7 @@ void LogSettings() {
     log_path("DataStorage_NANDDir", Common::FS::GetYuzuPath(Common::FS::YuzuPath::NANDDir));
     log_path("DataStorage_SDMCDir", Common::FS::GetYuzuPath(Common::FS::YuzuPath::SDMCDir));
     log_setting("Debugging_ProgramArgs", values.program_args.GetValue());
+    log_setting("Debugging_GDBStub", values.use_gdbstub.GetValue());
     log_setting("Input_EnableMotion", values.motion_enabled.GetValue());
     log_setting("Input_EnableVibration", values.vibration_enabled.GetValue());
     log_setting("Input_EnableRawInput", values.enable_raw_input.GetValue());
@@ -146,7 +147,7 @@ void UpdateRescalingInfo() {
         info.down_shift = 0;
         break;
     default:
-        UNREACHABLE();
+        ASSERT(false);
         info.up_scale = 1;
         info.down_shift = 0;
     }

src/common/settings.h

@@ -496,7 +496,7 @@ struct Values {
 
     // Renderer
     RangedSetting<RendererBackend> renderer_backend{
-        RendererBackend::OpenGL, RendererBackend::OpenGL, RendererBackend::Vulkan, "backend"};
+        RendererBackend::Vulkan, RendererBackend::OpenGL, RendererBackend::Vulkan, "backend"};
     BasicSetting<bool> renderer_debug{false, "debug"};
     BasicSetting<bool> renderer_shader_feedback{false, "shader_feedback"};
     BasicSetting<bool> enable_nsight_aftermath{false, "nsight_aftermath"};
@@ -601,7 +601,7 @@ struct Values {
     // Debugging
     bool record_frame_times;
     BasicSetting<bool> use_gdbstub{false, "use_gdbstub"};
-    BasicSetting<u16> gdbstub_port{0, "gdbstub_port"};
+    BasicSetting<u16> gdbstub_port{6543, "gdbstub_port"};
     BasicSetting<std::string> program_args{std::string(), "program_args"};
     BasicSetting<bool> dump_exefs{false, "dump_exefs"};
     BasicSetting<bool> dump_nso{false, "dump_nso"};

src/common/string_util.cpp

@@ -178,6 +178,10 @@ std::wstring UTF8ToUTF16W(const std::string& input) {
 
 #endif
 
+std::u16string U16StringFromBuffer(const u16* input, std::size_t length) {
+    return std::u16string(reinterpret_cast<const char16_t*>(input), length);
+}
+
 std::string StringFromFixedZeroTerminatedBuffer(std::string_view buffer, std::size_t max_len) {
     std::size_t len = 0;
     while (len < buffer.length() && len < max_len && buffer[len] != '\0') {

src/common/string_util.h

@@ -44,6 +44,8 @@ bool SplitPath(const std::string& full_path, std::string* _pPath, std::string* _
 
 #endif
 
+[[nodiscard]] std::u16string U16StringFromBuffer(const u16* input, std::size_t length);
+
 /**
  * Compares the string defined by the range [`begin`, `end`) to the null-terminated C-string
  * `other` for equality.

src/common/thread.cpp

@@ -47,6 +47,9 @@ void SetCurrentThreadPriority(ThreadPriority new_priority) {
     case ThreadPriority::VeryHigh:
         windows_priority = THREAD_PRIORITY_HIGHEST;
         break;
+    case ThreadPriority::Critical:
+        windows_priority = THREAD_PRIORITY_TIME_CRITICAL;
+        break;
     default:
         windows_priority = THREAD_PRIORITY_NORMAL;
         break;
@@ -59,9 +62,10 @@ void SetCurrentThreadPriority(ThreadPriority new_priority) {
 void SetCurrentThreadPriority(ThreadPriority new_priority) {
     pthread_t this_thread = pthread_self();
 
-    s32 max_prio = sched_get_priority_max(SCHED_OTHER);
-    s32 min_prio = sched_get_priority_min(SCHED_OTHER);
-    u32 level = static_cast<u32>(new_priority) + 1;
+    const auto scheduling_type = SCHED_OTHER;
+    s32 max_prio = sched_get_priority_max(scheduling_type);
+    s32 min_prio = sched_get_priority_min(scheduling_type);
+    u32 level = std::max(static_cast<u32>(new_priority) + 1, 4U);
 
     struct sched_param params;
     if (max_prio > min_prio) {
@@ -70,7 +74,7 @@ void SetCurrentThreadPriority(ThreadPriority new_priority) {
         params.sched_priority = min_prio - ((min_prio - max_prio) * level) / 4;
     }
 
-    pthread_setschedparam(this_thread, SCHED_OTHER, &params);
+    pthread_setschedparam(this_thread, scheduling_type, &params);
 }
 
 #endif

src/common/thread.h

@@ -92,6 +92,7 @@ enum class ThreadPriority : u32 {
     Normal = 1,
     High = 2,
     VeryHigh = 3,
+    Critical = 4,
 };
 
 void SetCurrentThreadPriority(ThreadPriority new_priority);

src/common/uint128.h

@@ -30,6 +30,10 @@ namespace Common {
 #else
     return _udiv128(r[1], r[0], d, &remainder);
 #endif
+#else
+#ifdef __SIZEOF_INT128__
+    const auto product = static_cast<unsigned __int128>(a) * static_cast<unsigned __int128>(b);
+    return static_cast<u64>(product / d);
 #else
     const u64 diva = a / d;
     const u64 moda = a % d;
@@ -37,6 +41,7 @@ namespace Common {
     const u64 modb = b % d;
     return diva * b + moda * divb + moda * modb / d;
 #endif
+#endif
 }
 
 // This function multiplies 2 u64 values and produces a u128 value;

src/common/x64/native_clock.cpp

@@ -75,8 +75,8 @@ NativeClock::NativeClock(u64 emulated_cpu_frequency_, u64 emulated_clock_frequen
 }
 
 u64 NativeClock::GetRTSC() {
-    TimePoint new_time_point{};
     TimePoint current_time_point{};
+    TimePoint new_time_point{};
 
     current_time_point.pack = Common::AtomicLoad128(time_point.pack.data());
     do {
@@ -89,8 +89,7 @@ u64 NativeClock::GetRTSC() {
         new_time_point.inner.accumulated_ticks = current_time_point.inner.accumulated_ticks + diff;
     } while (!Common::AtomicCompareAndSwap(time_point.pack.data(), new_time_point.pack,
                                            current_time_point.pack, current_time_point.pack));
-    /// The clock cannot be more precise than the guest timer, remove the lower bits
-    return new_time_point.inner.accumulated_ticks & inaccuracy_mask;
+    return new_time_point.inner.accumulated_ticks;
 }
 
 void NativeClock::Pause(bool is_paused) {

src/common/x64/native_clock.h

@@ -37,12 +37,8 @@ private:
         } inner;
     };
 
-    /// value used to reduce the native clocks accuracy as some apss rely on
-    /// undefined behavior where the level of accuracy in the clock shouldn't
-    /// be higher.
-    static constexpr u64 inaccuracy_mask = ~(UINT64_C(0x400) - 1);
-
     TimePoint time_point;
+
     // factors
     u64 clock_rtsc_factor{};
     u64 cpu_rtsc_factor{};

src/core/CMakeLists.txt

@@ -36,6 +36,13 @@ add_library(core STATIC
     crypto/ctr_encryption_layer.h
     crypto/xts_encryption_layer.cpp
     crypto/xts_encryption_layer.h
+    debugger/debugger_interface.h
+    debugger/debugger.cpp
+    debugger/debugger.h
+    debugger/gdbstub_arch.cpp
+    debugger/gdbstub_arch.h
+    debugger/gdbstub.cpp
+    debugger/gdbstub.h
     device_memory.cpp
     device_memory.h
     file_sys/bis_factory.cpp
@@ -215,7 +222,7 @@ add_library(core STATIC
     hle/kernel/k_page_buffer.h
     hle/kernel/k_page_heap.cpp
     hle/kernel/k_page_heap.h
-    hle/kernel/k_page_linked_list.h
+    hle/kernel/k_page_group.h
     hle/kernel/k_page_table.cpp
     hle/kernel/k_page_table.h
     hle/kernel/k_port.cpp
@@ -736,16 +743,11 @@ if (MSVC)
         /we4244 # 'conversion': conversion from 'type1' to 'type2', possible loss of data
         /we4245 # 'conversion': conversion from 'type1' to 'type2', signed/unsigned mismatch
         /we4254 # 'operator': conversion from 'type1:field_bits' to 'type2:field_bits', possible loss of data
-        /we4456 # Declaration of 'identifier' hides previous local declaration
-        /we4457 # Declaration of 'identifier' hides function parameter
-        /we4458 # Declaration of 'identifier' hides class member
-        /we4459 # Declaration of 'identifier' hides global declaration
     )
 else()
     target_compile_options(core PRIVATE
         -Werror=conversion
         -Werror=ignored-qualifiers
-        -Werror=shadow
 
         $<$<CXX_COMPILER_ID:GNU>:-Werror=class-memaccess>
         $<$<CXX_COMPILER_ID:GNU>:-Werror=unused-but-set-parameter>
@@ -761,6 +763,9 @@ create_target_directory_groups(core)
 
 target_link_libraries(core PUBLIC common PRIVATE audio_core video_core)
 target_link_libraries(core PUBLIC Boost::boost PRIVATE fmt::fmt nlohmann_json::nlohmann_json mbedtls Opus::Opus)
+if (MINGW)
+    target_link_libraries(core PRIVATE ${MSWSOCK_LIBRARY})
+endif()
 
 if (ENABLE_WEB_SERVICE)
     target_compile_definitions(core PRIVATE -DENABLE_WEB_SERVICE)

src/core/arm/arm_interface.cpp

@@ -9,7 +9,9 @@
 #include "core/arm/arm_interface.h"
 #include "core/arm/symbols.h"
 #include "core/core.h"
+#include "core/debugger/debugger.h"
 #include "core/hle/kernel/k_process.h"
+#include "core/hle/kernel/svc.h"
 #include "core/loader/loader.h"
 #include "core/memory.h"
 
@@ -88,4 +90,96 @@ void ARM_Interface::LogBacktrace() const {
     }
 }
 
+void ARM_Interface::Run() {
+    using Kernel::StepState;
+    using Kernel::SuspendType;
+
+    while (true) {
+        Kernel::KThread* current_thread{Kernel::GetCurrentThreadPointer(system.Kernel())};
+        Dynarmic::HaltReason hr{};
+
+        // Notify the debugger and go to sleep if a step was performed
+        // and this thread has been scheduled again.
+        if (current_thread->GetStepState() == StepState::StepPerformed) {
+            system.GetDebugger().NotifyThreadStopped(current_thread);
+            current_thread->RequestSuspend(SuspendType::Debug);
+            break;
+        }
+
+        // Otherwise, run the thread.
+        system.EnterDynarmicProfile();
+        if (current_thread->GetStepState() == StepState::StepPending) {
+            hr = StepJit();
+
+            if (Has(hr, step_thread)) {
+                current_thread->SetStepState(StepState::StepPerformed);
+            }
+        } else {
+            hr = RunJit();
+        }
+        system.ExitDynarmicProfile();
+
+        // Notify the debugger and go to sleep if a breakpoint was hit,
+        // or if the thread is unable to continue for any reason.
+        if (Has(hr, breakpoint) || Has(hr, no_execute)) {
+            RewindBreakpointInstruction();
+            if (system.DebuggerEnabled()) {
+                system.GetDebugger().NotifyThreadStopped(current_thread);
+            }
+            current_thread->RequestSuspend(Kernel::SuspendType::Debug);
+            break;
+        }
+
+        // Notify the debugger and go to sleep if a watchpoint was hit.
+        if (Has(hr, watchpoint)) {
+            RewindBreakpointInstruction();
+            if (system.DebuggerEnabled()) {
+                system.GetDebugger().NotifyThreadWatchpoint(current_thread, *HaltedWatchpoint());
+            }
+            current_thread->RequestSuspend(SuspendType::Debug);
+            break;
+        }
+
+        // Handle syscalls and scheduling (this may change the current thread)
+        if (Has(hr, svc_call)) {
+            Kernel::Svc::Call(system, GetSvcNumber());
+        }
+        if (Has(hr, break_loop) || !uses_wall_clock) {
+            break;
+        }
+    }
+}
+
+void ARM_Interface::LoadWatchpointArray(const WatchpointArray& wp) {
+    watchpoints = ℘
+}
+
+const Kernel::DebugWatchpoint* ARM_Interface::MatchingWatchpoint(
+    VAddr addr, u64 size, Kernel::DebugWatchpointType access_type) const {
+    if (!watchpoints) {
+        return nullptr;
+    }
+
+    const VAddr start_address{addr};
+    const VAddr end_address{addr + size};
+
+    for (size_t i = 0; i < Core::Hardware::NUM_WATCHPOINTS; i++) {
+        const auto& watch{(*watchpoints)[i]};
+
+        if (end_address <= watch.start_address) {
+            continue;
+        }
+        if (start_address >= watch.end_address) {
+            continue;
+        }
+        if ((access_type & watch.type) == Kernel::DebugWatchpointType::None) {
+            continue;
+        }
+
+        return &watch;
+    }
+
+    return nullptr;
+}
+
 } // namespace Core

src/core/arm/arm_interface.h

@@ -5,7 +5,11 @@
 #pragma once
 
 #include <array>
+#include <span>
 #include <vector>
+
+#include <dynarmic/interface/halt_reason.h>
+
 #include "common/common_funcs.h"
 #include "common/common_types.h"
 #include "core/hardware_properties.h"
@@ -16,13 +20,16 @@ struct PageTable;
 
 namespace Kernel {
 enum class VMAPermission : u8;
-}
+enum class DebugWatchpointType : u8;
+struct DebugWatchpoint;
+} // namespace Kernel
 
 namespace Core {
 class System;
 class CPUInterruptHandler;
 
 using CPUInterrupts = std::array<CPUInterruptHandler, Core::Hardware::NUM_CPU_CORES>;
+using WatchpointArray = std::array<Kernel::DebugWatchpoint, Core::Hardware::NUM_WATCHPOINTS>;
 
 /// Generic ARMv8 CPU interface
 class ARM_Interface {
@@ -64,10 +71,7 @@ public:
     static_assert(sizeof(ThreadContext64) == 0x320);
 
     /// Runs the CPU until an event happens
-    virtual void Run() = 0;
-
-    /// Step CPU by one instruction
-    virtual void Step() = 0;
+    void Run();
 
     /// Clear all instruction cache
     virtual void ClearInstructionCache() = 0;
@@ -170,6 +174,7 @@ public:
     virtual void SaveContext(ThreadContext64& ctx) = 0;
     virtual void LoadContext(const ThreadContext32& ctx) = 0;
     virtual void LoadContext(const ThreadContext64& ctx) = 0;
+    void LoadWatchpointArray(const WatchpointArray& wp);
 
     /// Clears the exclusive monitor's state.
     virtual void ClearExclusiveState() = 0;
@@ -194,13 +199,29 @@ public:
 
     void LogBacktrace() const;
 
+    static constexpr Dynarmic::HaltReason step_thread = Dynarmic::HaltReason::Step;
+    static constexpr Dynarmic::HaltReason break_loop = Dynarmic::HaltReason::UserDefined2;
+    static constexpr Dynarmic::HaltReason svc_call = Dynarmic::HaltReason::UserDefined3;
+    static constexpr Dynarmic::HaltReason breakpoint = Dynarmic::HaltReason::UserDefined4;
+    static constexpr Dynarmic::HaltReason watchpoint = Dynarmic::HaltReason::UserDefined5;
+    static constexpr Dynarmic::HaltReason no_execute = Dynarmic::HaltReason::UserDefined6;
+
 protected:
     /// System context that this ARM interface is running under.
     System& system;
     CPUInterrupts& interrupt_handlers;
+    const WatchpointArray* watchpoints;
     bool uses_wall_clock;
 
     static void SymbolicateBacktrace(Core::System& system, std::vector<BacktraceEntry>& out);
+    const Kernel::DebugWatchpoint* MatchingWatchpoint(
+        VAddr addr, u64 size, Kernel::DebugWatchpointType access_type) const;
+
+    virtual Dynarmic::HaltReason RunJit() = 0;
+    virtual Dynarmic::HaltReason StepJit() = 0;
+    virtual u32 GetSvcNumber() const = 0;
+    virtual const Kernel::DebugWatchpoint* HaltedWatchpoint() const = 0;
+    virtual void RewindBreakpointInstruction() = 0;
 };
 
 } // namespace Core

src/core/arm/dynarmic/arm_dynarmic_32.cpp

@@ -17,6 +17,8 @@
 #include "core/arm/dynarmic/arm_exclusive_monitor.h"
 #include "core/core.h"
 #include "core/core_timing.h"
+#include "core/debugger/debugger.h"
+#include "core/hle/kernel/k_process.h"
 #include "core/hle/kernel/svc.h"
 #include "core/memory.h"
 
@@ -24,70 +26,102 @@ namespace Core {
 
 using namespace Common::Literals;
 
-constexpr Dynarmic::HaltReason break_loop = Dynarmic::HaltReason::UserDefined2;
-constexpr Dynarmic::HaltReason svc_call = Dynarmic::HaltReason::UserDefined3;
-
 class DynarmicCallbacks32 : public Dynarmic::A32::UserCallbacks {
 public:
     explicit DynarmicCallbacks32(ARM_Dynarmic_32& parent_)
-        : parent{parent_}, memory(parent.system.Memory()) {}
+        : parent{parent_},
+          memory(parent.system.Memory()), debugger_enabled{parent.system.DebuggerEnabled()} {}
 
     u8 MemoryRead8(u32 vaddr) override {
+        CheckMemoryAccess(vaddr, 1, Kernel::DebugWatchpointType::Read);
         return memory.Read8(vaddr);
     }
     u16 MemoryRead16(u32 vaddr) override {
+        CheckMemoryAccess(vaddr, 2, Kernel::DebugWatchpointType::Read);
         return memory.Read16(vaddr);
     }
     u32 MemoryRead32(u32 vaddr) override {
+        CheckMemoryAccess(vaddr, 4, Kernel::DebugWatchpointType::Read);
         return memory.Read32(vaddr);
     }
     u64 MemoryRead64(u32 vaddr) override {
+        CheckMemoryAccess(vaddr, 8, Kernel::DebugWatchpointType::Read);
         return memory.Read64(vaddr);
     }
+    std::optional<u32> MemoryReadCode(u32 vaddr) override {
+        if (!memory.IsValidVirtualAddressRange(vaddr, sizeof(u32))) {
+            return std::nullopt;
+        }
+        return MemoryRead32(vaddr);
+    }
 
     void MemoryWrite8(u32 vaddr, u8 value) override {
-        memory.Write8(vaddr, value);
+        if (CheckMemoryAccess(vaddr, 1, Kernel::DebugWatchpointType::Write)) {
+            memory.Write8(vaddr, value);
+        }
     }
     void MemoryWrite16(u32 vaddr, u16 value) override {
-        memory.Write16(vaddr, value);
+        if (CheckMemoryAccess(vaddr, 2, Kernel::DebugWatchpointType::Write)) {
+            memory.Write16(vaddr, value);
+        }
     }
     void MemoryWrite32(u32 vaddr, u32 value) override {
-        memory.Write32(vaddr, value);
+        if (CheckMemoryAccess(vaddr, 4, Kernel::DebugWatchpointType::Write)) {
+            memory.Write32(vaddr, value);
+        }
     }
     void MemoryWrite64(u32 vaddr, u64 value) override {
-        memory.Write64(vaddr, value);
+        if (CheckMemoryAccess(vaddr, 8, Kernel::DebugWatchpointType::Write)) {
+            memory.Write64(vaddr, value);
+        }
     }
 
     bool MemoryWriteExclusive8(u32 vaddr, u8 value, u8 expected) override {
-        return memory.WriteExclusive8(vaddr, value, expected);
+        return CheckMemoryAccess(vaddr, 1, Kernel::DebugWatchpointType::Write) &&
+               memory.WriteExclusive8(vaddr, value, expected);
     }
     bool MemoryWriteExclusive16(u32 vaddr, u16 value, u16 expected) override {
-        return memory.WriteExclusive16(vaddr, value, expected);
+        return CheckMemoryAccess(vaddr, 2, Kernel::DebugWatchpointType::Write) &&
+               memory.WriteExclusive16(vaddr, value, expected);
     }
     bool MemoryWriteExclusive32(u32 vaddr, u32 value, u32 expected) override {
-        return memory.WriteExclusive32(vaddr, value, expected);
+        return CheckMemoryAccess(vaddr, 4, Kernel::DebugWatchpointType::Write) &&
+               memory.WriteExclusive32(vaddr, value, expected);
     }
     bool MemoryWriteExclusive64(u32 vaddr, u64 value, u64 expected) override {
-        return memory.WriteExclusive64(vaddr, value, expected);
+        return CheckMemoryAccess(vaddr, 8, Kernel::DebugWatchpointType::Write) &&
+               memory.WriteExclusive64(vaddr, value, expected);
     }
 
     void InterpreterFallback(u32 pc, std::size_t num_instructions) override {
         parent.LogBacktrace();
-        UNIMPLEMENTED_MSG("This should never happen, pc = {:08X}, code = {:08X}", pc,
-                          MemoryReadCode(pc));
+        LOG_ERROR(Core_ARM,
+                  "Unimplemented instruction @ 0x{:X} for {} instructions (instr = {:08X})", pc,
+                  num_instructions, MemoryRead32(pc));
     }
 
     void ExceptionRaised(u32 pc, Dynarmic::A32::Exception exception) override {
-        parent.LogBacktrace();
-        LOG_CRITICAL(Core_ARM,
-                     "ExceptionRaised(exception = {}, pc = {:08X}, code = {:08X}, thumb = {})",
-                     exception, pc, MemoryReadCode(pc), parent.IsInThumbMode());
-        UNIMPLEMENTED();
+        switch (exception) {
+        case Dynarmic::A32::Exception::NoExecuteFault:
+            LOG_CRITICAL(Core_ARM, "Cannot execute instruction at unmapped address {:#08x}", pc);
+            ReturnException(pc, ARM_Interface::no_execute);
+            return;
+        default:
+            if (debugger_enabled) {
+                ReturnException(pc, ARM_Interface::breakpoint);
+                return;
+            }
+
+            parent.LogBacktrace();
+            LOG_CRITICAL(Core_ARM,
+                         "ExceptionRaised(exception = {}, pc = {:08X}, code = {:08X}, thumb = {})",
+                         exception, pc, MemoryRead32(pc), parent.IsInThumbMode());
+        }
     }
 
     void CallSVC(u32 swi) override {
         parent.svc_swi = swi;
-        parent.jit.load()->HaltExecution(svc_call);
+        parent.jit.load()->HaltExecution(ARM_Interface::svc_call);
     }
 
     void AddTicks(u64 ticks) override {
@@ -113,9 +147,31 @@ public:
         return std::max<s64>(parent.system.CoreTiming().GetDowncount(), 0);
     }
 
+    bool CheckMemoryAccess(VAddr addr, u64 size, Kernel::DebugWatchpointType type) {
+        if (!debugger_enabled) {
+            return true;
+        }
+
+        const auto match{parent.MatchingWatchpoint(addr, size, type)};
+        if (match) {
+            parent.halted_watchpoint = match;
+            ReturnException(parent.jit.load()->Regs()[15], ARM_Interface::watchpoint);
+            return false;
+        }
+
+        return true;
+    }
+
+    void ReturnException(u32 pc, Dynarmic::HaltReason hr) {
+        parent.SaveContext(parent.breakpoint_context);
+        parent.breakpoint_context.cpu_registers[15] = pc;
+        parent.jit.load()->HaltExecution(hr);
+    }
+
     ARM_Dynarmic_32& parent;
     Core::Memory::Memory& memory;
     std::size_t num_interpreted_instructions{};
+    bool debugger_enabled{};
     static constexpr u64 minimum_run_cycles = 1000U;
 };
 
@@ -150,6 +206,11 @@ std::shared_ptr<Dynarmic::A32::Jit> ARM_Dynarmic_32::MakeJit(Common::PageTable*
     config.code_cache_size = 512_MiB;
     config.far_code_offset = 400_MiB;
 
+    // Allow memory fault handling to work
+    if (system.DebuggerEnabled()) {
+        config.check_halt_on_memory_access = true;
+    }
+
     // null_jit
     if (!page_table) {
         // Don't waste too much memory on null_jit
@@ -232,20 +293,24 @@ std::shared_ptr<Dynarmic::A32::Jit> ARM_Dynarmic_32::MakeJit(Common::PageTable*
     return std::make_unique<Dynarmic::A32::Jit>(config);
 }
 
-void ARM_Dynarmic_32::Run() {
-    while (true) {
-        const auto hr = jit.load()->Run();
-        if (Has(hr, svc_call)) {
-            Kernel::Svc::Call(system, svc_swi);
-        }
-        if (Has(hr, break_loop) || !uses_wall_clock) {
-            break;
-        }
-    }
+Dynarmic::HaltReason ARM_Dynarmic_32::RunJit() {
+    return jit.load()->Run();
 }
 
-void ARM_Dynarmic_32::Step() {
-    jit.load()->Step();
+Dynarmic::HaltReason ARM_Dynarmic_32::StepJit() {
+    return jit.load()->Step();
+}
+
+u32 ARM_Dynarmic_32::GetSvcNumber() const {
+    return svc_swi;
+}
+
+const Kernel::DebugWatchpoint* ARM_Dynarmic_32::HaltedWatchpoint() const {
+    return halted_watchpoint;
+}
+
+void ARM_Dynarmic_32::RewindBreakpointInstruction() {
+    LoadContext(breakpoint_context);
 }
 
 ARM_Dynarmic_32::ARM_Dynarmic_32(System& system_, CPUInterrupts& interrupt_handlers_,

src/core/arm/dynarmic/arm_dynarmic_32.h

@@ -41,8 +41,6 @@ public:
     void SetVectorReg(int index, u128 value) override;
     u32 GetPSTATE() const override;
     void SetPSTATE(u32 pstate) override;
-    void Run() override;
-    void Step() override;
     VAddr GetTlsAddress() const override;
     void SetTlsAddress(VAddr address) override;
     void SetTPIDR_EL0(u64 value) override;
@@ -70,6 +68,13 @@ public:
 
     std::vector<BacktraceEntry> GetBacktrace() const override;
 
+protected:
+    Dynarmic::HaltReason RunJit() override;
+    Dynarmic::HaltReason StepJit() override;
+    u32 GetSvcNumber() const override;
+    const Kernel::DebugWatchpoint* HaltedWatchpoint() const override;
+    void RewindBreakpointInstruction() override;
+
 private:
     std::shared_ptr<Dynarmic::A32::Jit> MakeJit(Common::PageTable* page_table) const;
 
@@ -95,6 +100,10 @@ private:
 
     // SVC callback
     u32 svc_swi{};
+
+    // Watchpoint info
+    const Kernel::DebugWatchpoint* halted_watchpoint;
+    ThreadContext32 breakpoint_context;
 };
 
 } // namespace Core

src/core/arm/dynarmic/arm_dynarmic_64.cpp

@@ -15,6 +15,7 @@
 #include "core/arm/dynarmic/arm_exclusive_monitor.h"
 #include "core/core.h"
 #include "core/core_timing.h"
+#include "core/debugger/debugger.h"
 #include "core/hardware_properties.h"
 #include "core/hle/kernel/k_process.h"
 #include "core/hle/kernel/svc.h"
@@ -25,68 +26,92 @@ namespace Core {
 using Vector = Dynarmic::A64::Vector;
 using namespace Common::Literals;
 
-constexpr Dynarmic::HaltReason break_loop = Dynarmic::HaltReason::UserDefined2;
-constexpr Dynarmic::HaltReason svc_call = Dynarmic::HaltReason::UserDefined3;
-
 class DynarmicCallbacks64 : public Dynarmic::A64::UserCallbacks {
 public:
     explicit DynarmicCallbacks64(ARM_Dynarmic_64& parent_)
-        : parent{parent_}, memory(parent.system.Memory()) {}
+        : parent{parent_},
+          memory(parent.system.Memory()), debugger_enabled{parent.system.DebuggerEnabled()} {}
 
     u8 MemoryRead8(u64 vaddr) override {
+        CheckMemoryAccess(vaddr, 1, Kernel::DebugWatchpointType::Read);
         return memory.Read8(vaddr);
     }
     u16 MemoryRead16(u64 vaddr) override {
+        CheckMemoryAccess(vaddr, 2, Kernel::DebugWatchpointType::Read);
         return memory.Read16(vaddr);
     }
     u32 MemoryRead32(u64 vaddr) override {
+        CheckMemoryAccess(vaddr, 4, Kernel::DebugWatchpointType::Read);
         return memory.Read32(vaddr);
     }
     u64 MemoryRead64(u64 vaddr) override {
+        CheckMemoryAccess(vaddr, 8, Kernel::DebugWatchpointType::Read);
         return memory.Read64(vaddr);
     }
     Vector MemoryRead128(u64 vaddr) override {
+        CheckMemoryAccess(vaddr, 16, Kernel::DebugWatchpointType::Read);
         return {memory.Read64(vaddr), memory.Read64(vaddr + 8)};
     }
+    std::optional<u32> MemoryReadCode(u64 vaddr) override {
+        if (!memory.IsValidVirtualAddressRange(vaddr, sizeof(u32))) {
+            return std::nullopt;
+        }
+        return MemoryRead32(vaddr);
+    }
 
     void MemoryWrite8(u64 vaddr, u8 value) override {
-        memory.Write8(vaddr, value);
+        if (CheckMemoryAccess(vaddr, 1, Kernel::DebugWatchpointType::Write)) {
+            memory.Write8(vaddr, value);
+        }
     }
     void MemoryWrite16(u64 vaddr, u16 value) override {
-        memory.Write16(vaddr, value);
+        if (CheckMemoryAccess(vaddr, 2, Kernel::DebugWatchpointType::Write)) {
+            memory.Write16(vaddr, value);
+        }
     }
     void MemoryWrite32(u64 vaddr, u32 value) override {
-        memory.Write32(vaddr, value);
+        if (CheckMemoryAccess(vaddr, 4, Kernel::DebugWatchpointType::Write)) {
+            memory.Write32(vaddr, value);
+        }
     }
     void MemoryWrite64(u64 vaddr, u64 value) override {
-        memory.Write64(vaddr, value);
+        if (CheckMemoryAccess(vaddr, 8, Kernel::DebugWatchpointType::Write)) {
+            memory.Write64(vaddr, value);
+        }
     }
     void MemoryWrite128(u64 vaddr, Vector value) override {
-        memory.Write64(vaddr, value[0]);
-        memory.Write64(vaddr + 8, value[1]);
+        if (CheckMemoryAccess(vaddr, 16, Kernel::DebugWatchpointType::Write)) {
+            memory.Write64(vaddr, value[0]);
+            memory.Write64(vaddr + 8, value[1]);
+        }
     }
 
     bool MemoryWriteExclusive8(u64 vaddr, std::uint8_t value, std::uint8_t expected) override {
-        return memory.WriteExclusive8(vaddr, value, expected);
+        return CheckMemoryAccess(vaddr, 1, Kernel::DebugWatchpointType::Write) &&
+               memory.WriteExclusive8(vaddr, value, expected);
     }
     bool MemoryWriteExclusive16(u64 vaddr, std::uint16_t value, std::uint16_t expected) override {
-        return memory.WriteExclusive16(vaddr, value, expected);
+        return CheckMemoryAccess(vaddr, 2, Kernel::DebugWatchpointType::Write) &&
+               memory.WriteExclusive16(vaddr, value, expected);
     }
     bool MemoryWriteExclusive32(u64 vaddr, std::uint32_t value, std::uint32_t expected) override {
-        return memory.WriteExclusive32(vaddr, value, expected);
+        return CheckMemoryAccess(vaddr, 4, Kernel::DebugWatchpointType::Write) &&
+               memory.WriteExclusive32(vaddr, value, expected);
     }
     bool MemoryWriteExclusive64(u64 vaddr, std::uint64_t value, std::uint64_t expected) override {
-        return memory.WriteExclusive64(vaddr, value, expected);
+        return CheckMemoryAccess(vaddr, 8, Kernel::DebugWatchpointType::Write) &&
+               memory.WriteExclusive64(vaddr, value, expected);
     }
     bool MemoryWriteExclusive128(u64 vaddr, Vector value, Vector expected) override {
-        return memory.WriteExclusive128(vaddr, value, expected);
+        return CheckMemoryAccess(vaddr, 16, Kernel::DebugWatchpointType::Write) &&
+               memory.WriteExclusive128(vaddr, value, expected);
     }
 
     void InterpreterFallback(u64 pc, std::size_t num_instructions) override {
         parent.LogBacktrace();
         LOG_ERROR(Core_ARM,
                   "Unimplemented instruction @ 0x{:X} for {} instructions (instr = {:08X})", pc,
-                  num_instructions, MemoryReadCode(pc));
+                  num_instructions, MemoryRead32(pc));
     }
 
     void InstructionCacheOperationRaised(Dynarmic::A64::InstructionCacheOperation op,
@@ -119,17 +144,25 @@ public:
         case Dynarmic::A64::Exception::SendEventLocal:
         case Dynarmic::A64::Exception::Yield:
             return;
-        case Dynarmic::A64::Exception::Breakpoint:
+        case Dynarmic::A64::Exception::NoExecuteFault:
+            LOG_CRITICAL(Core_ARM, "Cannot execute instruction at unmapped address {:#016x}", pc);
+            ReturnException(pc, ARM_Interface::no_execute);
+            return;
         default:
+            if (debugger_enabled) {
+                ReturnException(pc, ARM_Interface::breakpoint);
+                return;
+            }
+
             parent.LogBacktrace();
-            ASSERT_MSG(false, "ExceptionRaised(exception = {}, pc = {:08X}, code = {:08X})",
-                       static_cast<std::size_t>(exception), pc, MemoryReadCode(pc));
+            LOG_CRITICAL(Core_ARM, "ExceptionRaised(exception = {}, pc = {:08X}, code = {:08X})",
+                         static_cast<std::size_t>(exception), pc, MemoryRead32(pc));
         }
     }
 
     void CallSVC(u32 swi) override {
         parent.svc_swi = swi;
-        parent.jit.load()->HaltExecution(svc_call);
+        parent.jit.load()->HaltExecution(ARM_Interface::svc_call);
     }
 
     void AddTicks(u64 ticks) override {
@@ -157,10 +190,32 @@ public:
         return parent.system.CoreTiming().GetClockTicks();
     }
 
+    bool CheckMemoryAccess(VAddr addr, u64 size, Kernel::DebugWatchpointType type) {
+        if (!debugger_enabled) {
+            return true;
+        }
+
+        const auto match{parent.MatchingWatchpoint(addr, size, type)};
+        if (match) {
+            parent.halted_watchpoint = match;
+            ReturnException(parent.jit.load()->GetPC(), ARM_Interface::watchpoint);
+            return false;
+        }
+
+        return true;
+    }
+
+    void ReturnException(u64 pc, Dynarmic::HaltReason hr) {
+        parent.SaveContext(parent.breakpoint_context);
+        parent.breakpoint_context.pc = pc;
+        parent.jit.load()->HaltExecution(hr);
+    }
+
     ARM_Dynarmic_64& parent;
     Core::Memory::Memory& memory;
     u64 tpidrro_el0 = 0;
     u64 tpidr_el0 = 0;
+    bool debugger_enabled{};
     static constexpr u64 minimum_run_cycles = 1000U;
 };
 
@@ -211,6 +266,11 @@ std::shared_ptr<Dynarmic::A64::Jit> ARM_Dynarmic_64::MakeJit(Common::PageTable*
     config.code_cache_size = 512_MiB;
     config.far_code_offset = 400_MiB;
 
+    // Allow memory fault handling to work
+    if (system.DebuggerEnabled()) {
+        config.check_halt_on_memory_access = true;
+    }
+
     // null_jit
     if (!page_table) {
         // Don't waste too much memory on null_jit
@@ -293,20 +353,24 @@ std::shared_ptr<Dynarmic::A64::Jit> ARM_Dynarmic_64::MakeJit(Common::PageTable*
     return std::make_shared<Dynarmic::A64::Jit>(config);
 }
 
-void ARM_Dynarmic_64::Run() {
-    while (true) {
-        const auto hr = jit.load()->Run();
-        if (Has(hr, svc_call)) {
-            Kernel::Svc::Call(system, svc_swi);
-        }
-        if (Has(hr, break_loop) || !uses_wall_clock) {
-            break;
-        }
-    }
+Dynarmic::HaltReason ARM_Dynarmic_64::RunJit() {
+    return jit.load()->Run();
 }
 
-void ARM_Dynarmic_64::Step() {
-    jit.load()->Step();
+Dynarmic::HaltReason ARM_Dynarmic_64::StepJit() {
+    return jit.load()->Step();
+}
+
+u32 ARM_Dynarmic_64::GetSvcNumber() const {
+    return svc_swi;
+}
+
+const Kernel::DebugWatchpoint* ARM_Dynarmic_64::HaltedWatchpoint() const {
+    return halted_watchpoint;
+}
+
+void ARM_Dynarmic_64::RewindBreakpointInstruction() {
+    LoadContext(breakpoint_context);
 }
 
 ARM_Dynarmic_64::ARM_Dynarmic_64(System& system_, CPUInterrupts& interrupt_handlers_,

src/core/arm/dynarmic/arm_dynarmic_64.h

@@ -39,8 +39,6 @@ public:
     void SetVectorReg(int index, u128 value) override;
     u32 GetPSTATE() const override;
     void SetPSTATE(u32 pstate) override;
-    void Run() override;
-    void Step() override;
     VAddr GetTlsAddress() const override;
     void SetTlsAddress(VAddr address) override;
     void SetTPIDR_EL0(u64 value) override;
@@ -64,6 +62,13 @@ public:
 
     std::vector<BacktraceEntry> GetBacktrace() const override;
 
+protected:
+    Dynarmic::HaltReason RunJit() override;
+    Dynarmic::HaltReason StepJit() override;
+    u32 GetSvcNumber() const override;
+    const Kernel::DebugWatchpoint* HaltedWatchpoint() const override;
+    void RewindBreakpointInstruction() override;
+
 private:
     std::shared_ptr<Dynarmic::A64::Jit> MakeJit(Common::PageTable* page_table,
                                                 std::size_t address_space_bits) const;
@@ -88,6 +93,10 @@ private:
 
     // SVC callback
     u32 svc_swi{};
+
+    // Breakpoint info
+    const Kernel::DebugWatchpoint* halted_watchpoint;
+    ThreadContext64 breakpoint_context;
 };
 
 } // namespace Core

src/core/arm/dynarmic/arm_dynarmic_cp15.cpp

@@ -20,7 +20,7 @@ struct fmt::formatter<Dynarmic::A32::CoprocReg> {
     }
     template <typename FormatContext>
     auto format(const Dynarmic::A32::CoprocReg& reg, FormatContext& ctx) {
-        return format_to(ctx.out(), "cp{}", static_cast<size_t>(reg));
+        return fmt::format_to(ctx.out(), "cp{}", static_cast<size_t>(reg));
     }
 };
 

src/core/arm/symbols.cpp

@@ -3,73 +3,14 @@
 
 #include "common/bit_field.h"
 #include "common/common_funcs.h"
+#include "common/elf.h"
 #include "core/arm/symbols.h"
 #include "core/core.h"
 #include "core/memory.h"
 
+using namespace Common::ELF;
+
 namespace Core {
-namespace {
-
-constexpr u64 ELF_DYNAMIC_TAG_NULL = 0;
-constexpr u64 ELF_DYNAMIC_TAG_STRTAB = 5;
-constexpr u64 ELF_DYNAMIC_TAG_SYMTAB = 6;
-constexpr u64 ELF_DYNAMIC_TAG_SYMENT = 11;
-
-enum class ELFSymbolType : u8 {
-    None = 0,
-    Object = 1,
-    Function = 2,
-    Section = 3,
-    File = 4,
-    Common = 5,
-    TLS = 6,
-};
-
-enum class ELFSymbolBinding : u8 {
-    Local = 0,
-    Global = 1,
-    Weak = 2,
-};
-
-enum class ELFSymbolVisibility : u8 {
-    Default = 0,
-    Internal = 1,
-    Hidden = 2,
-    Protected = 3,
-};
-
-struct ELF64Symbol {
-    u32 name_index;
-    union {
-        u8 info;
-
-        BitField<0, 4, ELFSymbolType> type;
-        BitField<4, 4, ELFSymbolBinding> binding;
-    };
-    ELFSymbolVisibility visibility;
-    u16 sh_index;
-    u64 value;
-    u64 size;
-};
-static_assert(sizeof(ELF64Symbol) == 0x18, "ELF64Symbol has incorrect size.");
-
-struct ELF32Symbol {
-    u32 name_index;
-    u32 value;
-    u32 size;
-    union {
-        u8 info;
-
-        BitField<0, 4, ELFSymbolType> type;
-        BitField<4, 4, ELFSymbolBinding> binding;
-    };
-    ELFSymbolVisibility visibility;
-    u16 sh_index;
-};
-static_assert(sizeof(ELF32Symbol) == 0x10, "ELF32Symbol has incorrect size.");
-
-} // Anonymous namespace
-
 namespace Symbols {
 
 template <typename Word, typename ELFSymbol, typename ByteReader>
@@ -110,15 +51,15 @@ static Symbols GetSymbols(ByteReader ReadBytes) {
         const Word value = ReadWord(dynamic_index + sizeof(Word));
         dynamic_index += 2 * sizeof(Word);
 
-        if (tag == ELF_DYNAMIC_TAG_NULL) {
+        if (tag == ElfDtNull) {
             break;
         }
 
-        if (tag == ELF_DYNAMIC_TAG_STRTAB) {
+        if (tag == ElfDtStrtab) {
             string_table_offset = value;
-        } else if (tag == ELF_DYNAMIC_TAG_SYMTAB) {
+        } else if (tag == ElfDtSymtab) {
             symbol_table_offset = value;
-        } else if (tag == ELF_DYNAMIC_TAG_SYMENT) {
+        } else if (tag == ElfDtSyment) {
             symbol_entry_size = value;
         }
     }
@@ -134,14 +75,14 @@ static Symbols GetSymbols(ByteReader ReadBytes) {
         ELFSymbol symbol{};
         ReadBytes(&symbol, symbol_index, sizeof(ELFSymbol));
 
-        VAddr string_offset = string_table_offset + symbol.name_index;
+        VAddr string_offset = string_table_offset + symbol.st_name;
         std::string name;
         for (u8 c = Read8(string_offset); c != 0; c = Read8(++string_offset)) {
             name += static_cast<char>(c);
         }
 
         symbol_index += symbol_entry_size;
-        out[name] = std::make_pair(symbol.value, symbol.size);
+        out[name] = std::make_pair(symbol.st_value, symbol.st_size);
     }
 
     return out;
@@ -152,9 +93,9 @@ Symbols GetSymbols(VAddr base, Core::Memory::Memory& memory, bool is_64) {
         [&](void* ptr, size_t offset, size_t size) { memory.ReadBlock(base + offset, ptr, size); }};
 
     if (is_64) {
-        return GetSymbols<u64, ELF64Symbol>(ReadBytes);
+        return GetSymbols<u64, Elf64_Sym>(ReadBytes);
     } else {
-        return GetSymbols<u32, ELF32Symbol>(ReadBytes);
+        return GetSymbols<u32, Elf32_Sym>(ReadBytes);
     }
 }
 
@@ -164,9 +105,9 @@ Symbols GetSymbols(std::span<const u8> data, bool is_64) {
     }};
 
     if (is_64) {
-        return GetSymbols<u64, ELF64Symbol>(ReadBytes);
+        return GetSymbols<u64, Elf64_Sym>(ReadBytes);
     } else {
-        return GetSymbols<u32, ELF32Symbol>(ReadBytes);
+        return GetSymbols<u32, Elf32_Sym>(ReadBytes);
     }
 }
 

src/core/core.cpp

@@ -17,6 +17,7 @@
 #include "core/core.h"
 #include "core/core_timing.h"
 #include "core/cpu_manager.h"
+#include "core/debugger/debugger.h"
 #include "core/device_memory.h"
 #include "core/file_sys/bis_factory.h"
 #include "core/file_sys/mode.h"
@@ -137,7 +138,6 @@ struct System::Impl {
 
         kernel.Suspend(false);
         core_timing.SyncPause(false);
-        cpu_manager.Pause(false);
         is_paused = false;
 
         return status;
@@ -149,28 +149,29 @@ struct System::Impl {
 
         core_timing.SyncPause(true);
         kernel.Suspend(true);
-        cpu_manager.Pause(true);
         is_paused = true;
 
         return status;
     }
 
-    std::unique_lock<std::mutex> StallCPU() {
+    std::unique_lock<std::mutex> StallProcesses() {
         std::unique_lock<std::mutex> lk(suspend_guard);
         kernel.Suspend(true);
         core_timing.SyncPause(true);
-        cpu_manager.Pause(true);
         return lk;
     }
 
-    void UnstallCPU() {
+    void UnstallProcesses() {
         if (!is_paused) {
             core_timing.SyncPause(false);
             kernel.Suspend(false);
-            cpu_manager.Pause(false);
         }
     }
 
+    void InitializeDebugger(System& system, u16 port) {
+        debugger = std::make_unique<Debugger>(system, port);
+    }
+
     SystemResultStatus Init(System& system, Frontend::EmuWindow& emu_window) {
         LOG_DEBUG(Core, "initialized OK");
 
@@ -329,6 +330,9 @@ struct System::Impl {
             gpu_core->NotifyShutdown();
         }
 
+        kernel.ShutdownCores();
+        cpu_manager.Shutdown();
+        debugger.reset();
         services.reset();
         service_manager.reset();
         cheat_engine.reset();
@@ -436,6 +440,9 @@ struct System::Impl {
     /// Network instance
     Network::NetworkInstance network_instance;
 
+    /// Debugger
+    std::unique_ptr<Core::Debugger> debugger;
+
     SystemResultStatus status = SystemResultStatus::Success;
     std::string status_details = "";
 
@@ -472,10 +479,6 @@ SystemResultStatus System::Pause() {
     return impl->Pause();
 }
 
-SystemResultStatus System::SingleStep() {
-    return SystemResultStatus::Success;
-}
-
 void System::InvalidateCpuInstructionCaches() {
     impl->kernel.InvalidateAllInstructionCaches();
 }
@@ -488,12 +491,22 @@ void System::Shutdown() {
     impl->Shutdown();
 }
 
-std::unique_lock<std::mutex> System::StallCPU() {
-    return impl->StallCPU();
+void System::DetachDebugger() {
+    if (impl->debugger) {
+        impl->debugger->NotifyShutdown();
+    }
 }
 
-void System::UnstallCPU() {
-    impl->UnstallCPU();
+std::unique_lock<std::mutex> System::StallProcesses() {
+    return impl->StallProcesses();
+}
+
+void System::UnstallProcesses() {
+    impl->UnstallProcesses();
+}
+
+void System::InitializeDebugger() {
+    impl->InitializeDebugger(*this, Settings::values.gdbstub_port.GetValue());
 }
 
 SystemResultStatus System::Load(Frontend::EmuWindow& emu_window, const std::string& filepath,
@@ -809,6 +822,18 @@ bool System::IsMulticore() const {
     return impl->is_multicore;
 }
 
+bool System::DebuggerEnabled() const {
+    return Settings::values.use_gdbstub.GetValue();
+}
+
+Core::Debugger& System::GetDebugger() {
+    return *impl->debugger;
+}
+
+const Core::Debugger& System::GetDebugger() const {
+    return *impl->debugger;
+}
+
 void System::RegisterExecuteProgramCallback(ExecuteProgramCallback&& callback) {
     impl->execute_program_callback = std::move(callback);
 }

src/core/core.h

@@ -97,6 +97,7 @@ namespace Core {
 
 class ARM_Interface;
 class CpuManager;
+class Debugger;
 class DeviceMemory;
 class ExclusiveMonitor;
 class SpeedLimiter;
@@ -147,12 +148,6 @@ public:
      */
     [[nodiscard]] SystemResultStatus Pause();
 
-    /**
-     * Step the CPU one instruction
-     * @return Result status, indicating whether or not the operation succeeded.
-     */
-    [[nodiscard]] SystemResultStatus SingleStep();
-
     /**
      * Invalidate the CPU instruction caches
      * This function should only be used by GDB Stub to support breakpoints, memory updates and
@@ -165,8 +160,16 @@ public:
     /// Shutdown the emulated system.
     void Shutdown();
 
-    std::unique_lock<std::mutex> StallCPU();
-    void UnstallCPU();
+    /// Forcibly detach the debugger if it is running.
+    void DetachDebugger();
+
+    std::unique_lock<std::mutex> StallProcesses();
+    void UnstallProcesses();
+
+    /**
+     * Initialize the debugger.
+     */
+    void InitializeDebugger();
 
     /**
      * Load an executable application.
@@ -354,6 +357,9 @@ public:
     [[nodiscard]] Service::Time::TimeManager& GetTimeManager();
     [[nodiscard]] const Service::Time::TimeManager& GetTimeManager() const;
 
+    [[nodiscard]] Core::Debugger& GetDebugger();
+    [[nodiscard]] const Core::Debugger& GetDebugger() const;
+
     void SetExitLock(bool locked);
     [[nodiscard]] bool GetExitLock() const;
 
@@ -375,6 +381,9 @@ public:
     /// Tells if system is running on multicore.
     [[nodiscard]] bool IsMulticore() const;
 
+    /// Tells if the system debugger is enabled.
+    [[nodiscard]] bool DebuggerEnabled() const;
+
     /// Type used for the frontend to designate a callback for System to re-launch the application
     /// using a specified program index.
     using ExecuteProgramCallback = std::function<void(std::size_t)>;

src/core/core_timing.cpp

@@ -6,7 +6,9 @@
 #include <string>
 #include <tuple>
 
+#include "common/logging/log.h"
 #include "common/microprofile.h"
+#include "common/thread.h"
 #include "core/core_timing.h"
 #include "core/core_timing_util.h"
 #include "core/hardware_properties.h"
@@ -41,11 +43,11 @@ CoreTiming::CoreTiming()
 
 CoreTiming::~CoreTiming() = default;
 
-void CoreTiming::ThreadEntry(CoreTiming& instance) {
-    constexpr char name[] = "yuzu:HostTiming";
-    MicroProfileOnThreadCreate(name);
-    Common::SetCurrentThreadName(name);
-    Common::SetCurrentThreadPriority(Common::ThreadPriority::VeryHigh);
+void CoreTiming::ThreadEntry(CoreTiming& instance, size_t id) {
+    const std::string name = "yuzu:HostTiming_" + std::to_string(id);
+    MicroProfileOnThreadCreate(name.c_str());
+    Common::SetCurrentThreadName(name.c_str());
+    Common::SetCurrentThreadPriority(Common::ThreadPriority::Critical);
     instance.on_thread_init();
     instance.ThreadLoop();
     MicroProfileOnThreadExit();
@@ -59,68 +61,97 @@ void CoreTiming::Initialize(std::function<void()>&& on_thread_init_) {
     const auto empty_timed_callback = [](std::uintptr_t, std::chrono::nanoseconds) {};
     ev_lost = CreateEvent("_lost_event", empty_timed_callback);
     if (is_multicore) {
-        timer_thread = std::make_unique<std::thread>(ThreadEntry, std::ref(*this));
+        const auto hardware_concurrency = std::thread::hardware_concurrency();
+        size_t id = 0;
+        worker_threads.emplace_back(ThreadEntry, std::ref(*this), id++);
+        if (hardware_concurrency > 8) {
+            worker_threads.emplace_back(ThreadEntry, std::ref(*this), id++);
+        }
     }
 }
 
 void CoreTiming::Shutdown() {
-    paused = true;
+    is_paused = true;
     shutting_down = true;
-    pause_event.Set();
-    event.Set();
-    if (timer_thread) {
-        timer_thread->join();
+    std::atomic_thread_fence(std::memory_order_release);
+
+    event_cv.notify_all();
+    wait_pause_cv.notify_all();
+    for (auto& thread : worker_threads) {
+        thread.join();
     }
+    worker_threads.clear();
     ClearPendingEvents();
-    timer_thread.reset();
     has_started = false;
 }
 
-void CoreTiming::Pause(bool is_paused) {
-    paused = is_paused;
-    pause_event.Set();
-}
-
-void CoreTiming::SyncPause(bool is_paused) {
-    if (is_paused == paused && paused_set == paused) {
+void CoreTiming::Pause(bool is_paused_) {
+    std::unique_lock main_lock(event_mutex);
+    if (is_paused_ == paused_state.load(std::memory_order_relaxed)) {
         return;
     }
-    Pause(is_paused);
-    if (timer_thread) {
-        if (!is_paused) {
-            pause_event.Set();
+    if (is_multicore) {
+        is_paused = is_paused_;
+        event_cv.notify_all();
+        if (!is_paused_) {
+            wait_pause_cv.notify_all();
+        }
+    }
+    paused_state.store(is_paused_, std::memory_order_relaxed);
+}
+
+void CoreTiming::SyncPause(bool is_paused_) {
+    std::unique_lock main_lock(event_mutex);
+    if (is_paused_ == paused_state.load(std::memory_order_relaxed)) {
+        return;
+    }
+
+    if (is_multicore) {
+        is_paused = is_paused_;
+        event_cv.notify_all();
+        if (!is_paused_) {
+            wait_pause_cv.notify_all();
+        }
+    }
+    paused_state.store(is_paused_, std::memory_order_relaxed);
+    if (is_multicore) {
+        if (is_paused_) {
+            wait_signal_cv.wait(main_lock, [this] { return pause_count == worker_threads.size(); });
+        } else {
+            wait_signal_cv.wait(main_lock, [this] { return pause_count == 0; });
         }
-        event.Set();
-        while (paused_set != is_paused)
-            ;
     }
 }
 
 bool CoreTiming::IsRunning() const {
-    return !paused_set;
+    return !paused_state.load(std::memory_order_acquire);
 }
 
 bool CoreTiming::HasPendingEvents() const {
-    return !(wait_set && event_queue.empty());
+    std::unique_lock main_lock(event_mutex);
+    return !event_queue.empty() || pending_events.load(std::memory_order_relaxed) != 0;
 }
 
 void CoreTiming::ScheduleEvent(std::chrono::nanoseconds ns_into_future,
                                const std::shared_ptr<EventType>& event_type,
                                std::uintptr_t user_data) {
-    {
-        std::scoped_lock scope{basic_lock};
-        const u64 timeout = static_cast<u64>((GetGlobalTimeNs() + ns_into_future).count());
 
-        event_queue.emplace_back(Event{timeout, event_fifo_id++, user_data, event_type});
+    std::unique_lock main_lock(event_mutex);
+    const u64 timeout = static_cast<u64>((GetGlobalTimeNs() + ns_into_future).count());
 
-        std::push_heap(event_queue.begin(), event_queue.end(), std::greater<>());
+    event_queue.emplace_back(Event{timeout, event_fifo_id++, user_data, event_type});
+    pending_events.fetch_add(1, std::memory_order_relaxed);
+
+    std::push_heap(event_queue.begin(), event_queue.end(), std::greater<>());
+
+    if (is_multicore) {
+        event_cv.notify_one();
     }
-    event.Set();
 }
 
 void CoreTiming::UnscheduleEvent(const std::shared_ptr<EventType>& event_type,
                                  std::uintptr_t user_data) {
-    std::scoped_lock scope{basic_lock};
+    std::unique_lock main_lock(event_mutex);
     const auto itr = std::remove_if(event_queue.begin(), event_queue.end(), [&](const Event& e) {
         return e.type.lock().get() == event_type.get() && e.user_data == user_data;
     });
@@ -129,6 +160,7 @@ void CoreTiming::UnscheduleEvent(const std::shared_ptr<EventType>& event_type,
     if (itr != event_queue.end()) {
         event_queue.erase(itr, event_queue.end());
         std::make_heap(event_queue.begin(), event_queue.end(), std::greater<>());
+        pending_events.fetch_sub(1, std::memory_order_relaxed);
     }
 }
 
@@ -168,11 +200,12 @@ u64 CoreTiming::GetClockTicks() const {
 }
 
 void CoreTiming::ClearPendingEvents() {
+    std::unique_lock main_lock(event_mutex);
     event_queue.clear();
 }
 
 void CoreTiming::RemoveEvent(const std::shared_ptr<EventType>& event_type) {
-    std::scoped_lock lock{basic_lock};
+    std::unique_lock main_lock(event_mutex);
 
     const auto itr = std::remove_if(event_queue.begin(), event_queue.end(), [&](const Event& e) {
         return e.type.lock().get() == event_type.get();
@@ -186,21 +219,28 @@ void CoreTiming::RemoveEvent(const std::shared_ptr<EventType>& event_type) {
 }
 
 std::optional<s64> CoreTiming::Advance() {
-    std::scoped_lock lock{advance_lock, basic_lock};
     global_timer = GetGlobalTimeNs().count();
 
+    std::unique_lock main_lock(event_mutex);
     while (!event_queue.empty() && event_queue.front().time <= global_timer) {
         Event evt = std::move(event_queue.front());
         std::pop_heap(event_queue.begin(), event_queue.end(), std::greater<>());
         event_queue.pop_back();
-        basic_lock.unlock();
 
         if (const auto event_type{evt.type.lock()}) {
-            event_type->callback(
-                evt.user_data, std::chrono::nanoseconds{static_cast<s64>(global_timer - evt.time)});
+            sequence_mutex.lock();
+            event_mutex.unlock();
+
+            event_type->guard.lock();
+            sequence_mutex.unlock();
+            const s64 delay = static_cast<s64>(GetGlobalTimeNs().count() - evt.time);
+            event_type->callback(evt.user_data, std::chrono::nanoseconds{delay});
+            event_type->guard.unlock();
+
+            event_mutex.lock();
+            pending_events.fetch_sub(1, std::memory_order_relaxed);
         }
 
-        basic_lock.lock();
         global_timer = GetGlobalTimeNs().count();
     }
 
@@ -213,26 +253,34 @@ std::optional<s64> CoreTiming::Advance() {
 }
 
 void CoreTiming::ThreadLoop() {
+    const auto predicate = [this] { return !event_queue.empty() || is_paused; };
     has_started = true;
     while (!shutting_down) {
-        while (!paused) {
-            paused_set = false;
+        while (!is_paused && !shutting_down) {
             const auto next_time = Advance();
             if (next_time) {
                 if (*next_time > 0) {
                     std::chrono::nanoseconds next_time_ns = std::chrono::nanoseconds(*next_time);
-                    event.WaitFor(next_time_ns);
+                    std::unique_lock main_lock(event_mutex);
+                    event_cv.wait_for(main_lock, next_time_ns, predicate);
                 }
             } else {
-                wait_set = true;
-                event.Wait();
+                std::unique_lock main_lock(event_mutex);
+                event_cv.wait(main_lock, predicate);
             }
-            wait_set = false;
         }
-        paused_set = true;
-        clock->Pause(true);
-        pause_event.Wait();
-        clock->Pause(false);
+        std::unique_lock main_lock(event_mutex);
+        pause_count++;
+        if (pause_count == worker_threads.size()) {
+            clock->Pause(true);
+            wait_signal_cv.notify_all();
+        }
+        wait_pause_cv.wait(main_lock, [this] { return !is_paused || shutting_down; });
+        pause_count--;
+        if (pause_count == 0) {
+            clock->Pause(false);
+            wait_signal_cv.notify_all();
+        }
     }
 }
 

src/core/core_timing.h

@@ -5,6 +5,7 @@
 
 #include <atomic>
 #include <chrono>
+#include <condition_variable>
 #include <functional>
 #include <memory>
 #include <mutex>
@@ -14,7 +15,6 @@
 #include <vector>
 
 #include "common/common_types.h"
-#include "common/thread.h"
 #include "common/wall_clock.h"
 
 namespace Core::Timing {
@@ -32,6 +32,7 @@ struct EventType {
     TimedCallback callback;
     /// A pointer to the name of the event.
     const std::string name;
+    mutable std::mutex guard;
 };
 
 /**
@@ -131,7 +132,7 @@ private:
     /// Clear all pending events. This should ONLY be done on exit.
     void ClearPendingEvents();
 
-    static void ThreadEntry(CoreTiming& instance);
+    static void ThreadEntry(CoreTiming& instance, size_t id);
     void ThreadLoop();
 
     std::unique_ptr<Common::WallClock> clock;
@@ -144,21 +145,25 @@ private:
     // accomodated by the standard adaptor class.
     std::vector<Event> event_queue;
     u64 event_fifo_id = 0;
+    std::atomic<size_t> pending_events{};
 
     std::shared_ptr<EventType> ev_lost;
-    Common::Event event{};
-    Common::Event pause_event{};
-    std::mutex basic_lock;
-    std::mutex advance_lock;
-    std::unique_ptr<std::thread> timer_thread;
-    std::atomic<bool> paused{};
-    std::atomic<bool> paused_set{};
-    std::atomic<bool> wait_set{};
-    std::atomic<bool> shutting_down{};
     std::atomic<bool> has_started{};
     std::function<void()> on_thread_init{};
 
+    std::vector<std::thread> worker_threads;
+
+    std::condition_variable event_cv;
+    std::condition_variable wait_pause_cv;
+    std::condition_variable wait_signal_cv;
+    mutable std::mutex event_mutex;
+    mutable std::mutex sequence_mutex;
+
+    std::atomic<bool> paused_state{};
+    bool is_paused{};
+    bool shutting_down{};
     bool is_multicore{};
+    size_t pause_count{};
 
     /// Cycle timing
     u64 ticks{};