Add OpenCV to macOS builds

Apparently table items need one common role set for internal indexing.

.ci/build-mac-arm64.sh

@@ -6,12 +6,13 @@ export HOMEBREW_NO_INSTALLED_DEPENDENTS_CHECK=1
 export HOMEBREW_NO_ENV_HINTS=1
 export HOMEBREW_NO_INSTALL_CLEANUP=1
 
-brew install -f --overwrite --quiet ccache pipenv googletest ffmpeg@5 "llvm@$LLVM_COMPILER_VER" glew sdl3 vulkan-headers
+brew install -f --overwrite --quiet pipenv googletest opencv@4 ffmpeg@5 "llvm@$LLVM_COMPILER_VER" glew sdl3 vulkan-headers
+brew unlink --quiet ffmpeg qtbase qtsvg qtdeclarative
 brew link -f --quiet "llvm@$LLVM_COMPILER_VER" ffmpeg@5
 
-# moltenvk based on commit for 1.3.0 release
+# moltenvk based on commit for 1.4.0 release
 export HOMEBREW_DEVELOPER=1 # Prevents blocking of local formulae
-wget https://raw.githubusercontent.com/Homebrew/homebrew-core/7255441cbcafabaa8950f67c7ec55ff499dbb2d3/Formula/m/molten-vk.rb
+wget https://raw.githubusercontent.com/Homebrew/homebrew-core/ea2bec5f1f4384e188d7fc0702ab21a20a2ced08/Formula/m/molten-vk.rb
 /opt/homebrew/bin/brew install -f --overwrite --formula --quiet ./molten-vk.rb
 export HOMEBREW_DEVELOPER=0
 
@@ -27,23 +28,25 @@ export CMAKE_EXTRA_OPTS='-DLLVM_TARGETS_TO_BUILD=arm64'
 export WORKDIR;
 WORKDIR="$(pwd)"
 
+# Setup ccache
+if [ ! -d "$CCACHE_DIR" ]; then
+  mkdir -p "$CCACHE_DIR"
+fi
+
 # Get Qt
 if [ ! -d "/tmp/Qt/$QT_VER" ]; then
   mkdir -p "/tmp/Qt"
   git clone https://github.com/engnr/qt-downloader.git
   cd qt-downloader
   git checkout f52efee0f18668c6d6de2dec0234b8c4bc54c597
-  # nested Qt 6.9.3 URL workaround
+  sed -i '' "s/'qt{0}_{0}{1}{2}'.format(major, minor, patch)]))/'qt{0}_{0}{1}{2}'.format(major, minor, patch), 'qt{0}_{0}{1}{2}'.format(major, minor, patch)]))/g" qt-downloader
-  # sed -i '' "s/'qt{0}_{0}{1}{2}'.format(major, minor, patch)]))/'qt{0}_{0}{1}{2}'.format(major, minor, patch), 'qt{0}_{0}{1}{2}'.format(major, minor, patch)]))/g" qt-downloader
+  sed -i '' "s/'{}\/{}\/qt{}_{}\/'/'{0}\/{1}\/qt{2}_{3}\/qt{2}_{3}\/'/g" qt-downloader
-  # sed -i '' "s/'{}\/{}\/qt{}_{}\/'/'{0}\/{1}\/qt{2}_{3}\/qt{2}_{3}\/'/g" qt-downloader
-  # archived Qt 6.7.3 URL workaround
-  sed -i '' "s/official_releases/archive/g" qt-downloader
   cd "/tmp/Qt"
   "$BREW_PATH/bin/pipenv" run pip3 uninstall py7zr requests semantic_version lxml
   "$BREW_PATH/bin/pipenv" run pip3 install py7zr requests semantic_version lxml  --no-cache
   mkdir -p "$QT_VER/macos" ; ln -s "macos" "$QT_VER/clang_64"
-  # sed -i '' 's/args\.version \/ derive_toolchain_dir(args) \/ //g' "$WORKDIR/qt-downloader/qt-downloader" # Qt 6.9.3 workaround
+  sed -i '' 's/args\.version \/ derive_toolchain_dir(args) \/ //g' "$WORKDIR/qt-downloader/qt-downloader"
-  "$BREW_PATH/bin/pipenv" run "$WORKDIR/qt-downloader/qt-downloader" macos desktop "$QT_VER" clang_64 --opensource --addons qtmultimedia qtimageformats # -o "$QT_VER/clang_64"
+  "$BREW_PATH/bin/pipenv" run "$WORKDIR/qt-downloader/qt-downloader" macos desktop "$QT_VER" clang_64 --opensource --addons qtmultimedia qtimageformats -o "$QT_VER/clang_64"
 fi
 
 cd "$WORKDIR"

.ci/build-mac.sh

@@ -8,13 +8,17 @@ export HOMEBREW_NO_INSTALL_CLEANUP=1
 
 brew install -f --overwrite --quiet ccache pipenv "llvm@$LLVM_COMPILER_VER"
 brew link -f --overwrite --quiet "llvm@$LLVM_COMPILER_VER"
+# shellcheck disable=SC3009
+rm /usr/local/bin/{idle3.14,pip3.14,pydoc3.14,python3.14,python3.14-config} && \
+rm /usr/local/bin/{idle3,pip3,pydoc3,python3,python3-config}
 arch -x86_64 /bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh)"
-arch -x86_64 /usr/local/bin/brew install -f --overwrite --quiet ffmpeg@5 "llvm@$LLVM_COMPILER_VER" glew sdl3 vulkan-headers
+arch -x86_64 /usr/local/bin/brew install -f --overwrite --quiet opencv@4 ffmpeg@5 "llvm@$LLVM_COMPILER_VER" glew sdl3 vulkan-headers
+arch -x86_64 /usr/local/bin/brew unlink  --quiet ffmpeg qtbase qtsvg qtdeclarative
 arch -x86_64 /usr/local/bin/brew link -f --overwrite --quiet "llvm@$LLVM_COMPILER_VER" ffmpeg@5
 
-# moltenvk based on commit for 1.3.0 release
+# moltenvk based on commit for 1.4.0 release
 export HOMEBREW_DEVELOPER=1 # Prevents blocking of local formulae
-wget https://raw.githubusercontent.com/Homebrew/homebrew-core/7255441cbcafabaa8950f67c7ec55ff499dbb2d3/Formula/m/molten-vk.rb
+wget https://raw.githubusercontent.com/Homebrew/homebrew-core/ea2bec5f1f4384e188d7fc0702ab21a20a2ced08/Formula/m/molten-vk.rb
 arch -x86_64 /usr/local/bin/brew install -f --overwrite --formula --quiet ./molten-vk.rb
 export HOMEBREW_DEVELOPER=0
 export CXX=clang++
@@ -35,16 +39,13 @@ if [ ! -d "/tmp/Qt/$QT_VER" ]; then
   git clone https://github.com/engnr/qt-downloader.git
   cd qt-downloader
   git checkout f52efee0f18668c6d6de2dec0234b8c4bc54c597
-  # nested Qt 6.9.3 URL workaround
+  sed -i '' "s/'qt{0}_{0}{1}{2}'.format(major, minor, patch)]))/'qt{0}_{0}{1}{2}'.format(major, minor, patch), 'qt{0}_{0}{1}{2}'.format(major, minor, patch)]))/g" qt-downloader
-  # sed -i '' "s/'qt{0}_{0}{1}{2}'.format(major, minor, patch)]))/'qt{0}_{0}{1}{2}'.format(major, minor, patch), 'qt{0}_{0}{1}{2}'.format(major, minor, patch)]))/g" qt-downloader
+  sed -i '' "s/'{}\/{}\/qt{}_{}\/'/'{0}\/{1}\/qt{2}_{3}\/qt{2}_{3}\/'/g" qt-downloader
-  # sed -i '' "s/'{}\/{}\/qt{}_{}\/'/'{0}\/{1}\/qt{2}_{3}\/qt{2}_{3}\/'/g" qt-downloader
-  # archived Qt 6.7.3 URL workaround
-  sed -i '' "s/official_releases/archive/g" qt-downloader
   cd "/tmp/Qt"
   "/opt/homebrew/bin/pipenv" --python "/opt/homebrew/bin/python3" run pip3 install py7zr requests semantic_version lxml
   mkdir -p "$QT_VER/macos" ; ln -s "macos" "$QT_VER/clang_64"
-  # sed -i '' 's/args\.version \/ derive_toolchain_dir(args) \/ //g' "$WORKDIR/qt-downloader/qt-downloader" # Qt 6.9.3 workaround
+  sed -i '' 's/args\.version \/ derive_toolchain_dir(args) \/ //g' "$WORKDIR/qt-downloader/qt-downloader"
-  "/opt/homebrew/bin/pipenv"  --python "/opt/homebrew/bin/python3" run "$WORKDIR/qt-downloader/qt-downloader" macos desktop "$QT_VER" clang_64 --opensource --addons qtmultimedia qtimageformats # -o "$QT_VER/clang_64"
+  "/opt/homebrew/bin/pipenv"  --python "/opt/homebrew/bin/python3" run "$WORKDIR/qt-downloader/qt-downloader" macos desktop "$QT_VER" clang_64 --opensource --addons qtmultimedia qtimageformats -o "$QT_VER/clang_64"
 fi
 
 cd "$WORKDIR"
@@ -54,7 +55,8 @@ export Qt6_DIR="$WORKDIR/qt-downloader/$QT_VER/clang_64/lib/cmake/Qt$QT_VER_MAIN
 export SDL3_DIR="$BREW_X64_PATH/opt/sdl3/lib/cmake/SDL3"
 
 export PATH="/opt/homebrew/opt/llvm@$LLVM_COMPILER_VER/bin:$WORKDIR/qt-downloader/$QT_VER/clang_64/bin:$BREW_BIN:$BREW_SBIN:/usr/bin:/bin:/usr/sbin:/sbin:/opt/X11/bin:/Library/Apple/usr/bin:$PATH"
-export LDFLAGS="-L$BREW_X64_PATH/lib -Wl,-rpath,$BREW_X64_PATH/lib"
+# shellcheck disable=SC2155
+export LDFLAGS="-L$BREW_X64_PATH/lib -Wl,-rpath,$BREW_X64_PATH/lib,-L$(brew --prefix llvm)/lib/c++"
 export CPPFLAGS="-I$BREW_X64_PATH/include -msse -msse2 -mcx16 -no-pie -D__MAC_OS_X_VERSION_MIN_REQUIRED=140000"
 export CFLAGS="-D__MAC_OS_X_VERSION_MIN_REQUIRED=140000"
 export LIBRARY_PATH="$BREW_X64_PATH/opt/llvm@$LLVM_COMPILER_VER/lib:$BREW_X64_PATH/lib"

.ci/deploy-linux.sh

@@ -14,7 +14,7 @@ if [ "$DEPLOY_APPIMAGE" = "true" ]; then
     curl -fsSLo linuxdeploy-plugin-checkrt.sh https://github.com/darealshinji/linuxdeploy-plugin-checkrt/releases/download/continuous/linuxdeploy-plugin-checkrt.sh
     chmod +x ./linuxdeploy-plugin-checkrt.sh
 
-    export EXTRA_PLATFORM_PLUGINS="libqwayland-egl.so;libqwayland-generic.so"
+    export EXTRA_PLATFORM_PLUGINS="libqwayland.so"
     export EXTRA_QT_PLUGINS="svg;wayland-decoration-client;wayland-graphics-integration-client;wayland-shell-integration;waylandcompositor"
 
     APPIMAGE_EXTRACT_AND_RUN=1 linuxdeploy --appdir AppDir --plugin qt --plugin checkrt
@@ -32,6 +32,25 @@ if [ "$DEPLOY_APPIMAGE" = "true" ]; then
     # Remove git directory containing local commit history file
     rm -rf ./AppDir/usr/share/rpcs3/git
 
+    # Download translations
+    mkdir -p "./AppDir/usr/translations"
+    ZIP_URL=$(curl -fsSL "https://api.github.com/repos/RPCS3/rpcs3_translations/releases/latest" \
+      | grep "browser_download_url" \
+      | grep "RPCS3-languages.zip" \
+      | cut -d '"' -f 4)
+    if [ -z "$ZIP_URL" ]; then
+      echo "Failed to find RPCS3-languages.zip in the latest release. Continuing without translations."
+    else
+      echo "Downloading translations from: $ZIP_URL"
+      curl -L -o translations.zip "$ZIP_URL" || {
+        echo "Failed to download translations.zip. Continuing without translations."
+        exit 0
+      }
+      unzip -o translations.zip -d "./AppDir/usr/translations" >/dev/null 2>&1 || \
+        echo "Failed to extract translations.zip. Continuing without translations."
+      rm -f translations.zip
+    fi
+
     curl -fsSLo /uruntime "https://github.com/VHSgunzo/uruntime/releases/download/v0.3.4/uruntime-appimage-dwarfs-$CPU_ARCH"
     chmod +x /uruntime
     /uruntime --appimage-mkdwarfs -f --set-owner 0 --set-group 0 --no-history --no-create-timestamp \

.ci/deploy-mac-arm64.sh

@@ -17,6 +17,7 @@ cd bin
 mkdir "rpcs3.app/Contents/lib/" || true
 
 cp "$(realpath /opt/homebrew/opt/llvm@$LLVM_COMPILER_VER/lib/c++/libc++abi.1.0.dylib)" "rpcs3.app/Contents/Frameworks/libc++abi.1.dylib"
+cp "$(realpath /opt/homebrew/opt/gcc/lib/gcc/current/libgcc_s.1.1.dylib)" "rpcs3.app/Contents/Frameworks/libgcc_s.1.1.dylib"
 cp "$(realpath /opt/homebrew/lib/libsharpyuv.0.dylib)" "rpcs3.app/Contents/lib/libsharpyuv.0.dylib"
 cp "$(realpath /opt/homebrew/lib/libintl.8.dylib)" "rpcs3.app/Contents/lib/libintl.8.dylib"
 
@@ -31,11 +32,28 @@ rm -rf "rpcs3.app/Contents/Frameworks/QtPdf.framework" \
 
 ../../.ci/optimize-mac.sh rpcs3.app
 
+# Download translations
+mkdir -p "rpcs3.app/Contents/translations"
+ZIP_URL=$(curl -fsSL "https://api.github.com/repos/RPCS3/rpcs3_translations/releases/latest" \
+  | grep "browser_download_url" \
+  | grep "RPCS3-languages.zip" \
+  | cut -d '"' -f 4)
+if [ -z "$ZIP_URL" ]; then
+  echo "Failed to find RPCS3-languages.zip in the latest release. Continuing without translations."
+else
+  echo "Downloading translations from: $ZIP_URL"
+  curl -L -o translations.zip "$ZIP_URL" || {
+    echo "Failed to download translations.zip. Continuing without translations."
+    exit 0
+  }
+  unzip -o translations.zip -d "rpcs3.app/Contents/translations" >/dev/null 2>&1 || \
+    echo "Failed to extract translations.zip. Continuing without translations."
+  rm -f translations.zip
+fi
+
 # Hack
-install_name_tool \
+install_name_tool -delete_rpath /opt/homebrew/lib RPCS3.app/Contents/MacOS/rpcs3 || echo "Hack for deleting rpath /opt/homebrew/lib not needed"
--delete_rpath /opt/homebrew/lib \
+install_name_tool -delete_rpath /opt/homebrew/opt/llvm@$LLVM_COMPILER_VER/lib RPCS3.app/Contents/MacOS/rpcs3 || echo "Hack for deleting rpath /opt/homebrew/opt/llvm@$LLVM_COMPILER_VER/lib not needed"
--delete_rpath /opt/homebrew/opt/llvm@$LLVM_COMPILER_VER/lib RPCS3.app/Contents/MacOS/rpcs3
-#-delete_rpath /opt/homebrew1/Cellar/sdl3/3.2.8/lib
 
 # Need to do this rename hack due to case insensitive filesystem
 mv rpcs3.app RPCS3_.app

.ci/deploy-mac.sh

@@ -16,7 +16,9 @@ echo "AVVER=$AVVER" >> ../.ci/ci-vars.env
 cd bin
 mkdir "rpcs3.app/Contents/lib/"
 
-cp "/usr/local/opt/llvm@$LLVM_COMPILER_VER/lib/c++/libc++abi.1.0.dylib" "rpcs3.app/Contents/lib/libc++abi.1.dylib"
+cp "$(realpath /usr/local/opt/llvm@$LLVM_COMPILER_VER/lib/c++/libc++abi.1.0.dylib)" "rpcs3.app/Contents/Frameworks/libc++abi.1.dylib"
+cp "$(realpath /usr/local/opt/llvm@$LLVM_COMPILER_VER/lib/unwind/libunwind.1.dylib)" "rpcs3.app/Contents/Frameworks/libunwind.1.dylib"
+cp "$(realpath /usr/local/opt/gcc/lib/gcc/current/libgcc_s.1.1.dylib)" "rpcs3.app/Contents/Frameworks/libgcc_s.1.1.dylib"
 cp "$(realpath /usr/local/lib/libsharpyuv.0.dylib)" "rpcs3.app/Contents/lib/libsharpyuv.0.dylib"
 cp "$(realpath /usr/local/lib/libintl.8.dylib)" "rpcs3.app/Contents/lib/libintl.8.dylib"
 
@@ -31,6 +33,25 @@ rm -rf "rpcs3.app/Contents/Frameworks/QtPdf.framework" \
 
 ../../.ci/optimize-mac.sh rpcs3.app
 
+# Download translations
+mkdir -p "rpcs3.app/Contents/translations"
+ZIP_URL=$(curl -fsSL "https://api.github.com/repos/RPCS3/rpcs3_translations/releases/latest" \
+  | grep "browser_download_url" \
+  | grep "RPCS3-languages.zip" \
+  | cut -d '"' -f 4)
+if [ -z "$ZIP_URL" ]; then
+  echo "Failed to find RPCS3-languages.zip in the latest release. Continuing without translations."
+else
+  echo "Downloading translations from: $ZIP_URL"
+  curl -L -o translations.zip "$ZIP_URL" || {
+    echo "Failed to download translations.zip. Continuing without translations."
+    exit 0
+  }
+  unzip -o translations.zip -d "rpcs3.app/Contents/translations" >/dev/null 2>&1 || \
+    echo "Failed to extract translations.zip. Continuing without translations."
+  rm -f translations.zip
+fi
+
 # Need to do this rename hack due to case insensitive filesystem
 mv rpcs3.app RPCS3_.app
 mv RPCS3_.app RPCS3.app

.ci/deploy-windows-clang.sh

@@ -24,6 +24,25 @@ mkdir ./bin/config/input_configs
 curl -fsSL 'https://raw.githubusercontent.com/gabomdq/SDL_GameControllerDB/master/gamecontrollerdb.txt' 1> ./bin/config/input_configs/gamecontrollerdb.txt
 curl -fsSL 'https://rpcs3.net/compatibility?api=v1&export' | iconv -t UTF-8 1> ./bin/GuiConfigs/compat_database.dat
 
+# Download translations
+mkdir -p ./bin/share/qt6/translations
+ZIP_URL=$(curl -fsSL "https://api.github.com/repos/RPCS3/rpcs3_translations/releases/latest" \
+  | grep "browser_download_url" \
+  | grep "RPCS3-languages.zip" \
+  | cut -d '"' -f 4)
+if [ -z "$ZIP_URL" ]; then
+  echo "Failed to find RPCS3-languages.zip in the latest release. Continuing without translations."
+else
+  echo "Downloading translations from: $ZIP_URL"
+  curl -L -o translations.zip "$ZIP_URL" || {
+    echo "Failed to download translations.zip. Continuing without translations."
+    exit 0
+  }
+  unzip -o translations.zip -d "./bin/share/qt6/translations" >/dev/null 2>&1 || \
+    echo "Failed to extract translations.zip. Continuing without translations."
+  rm -f translations.zip
+fi
+
 # Package artifacts
 7z a -m0=LZMA2 -mx9 "$BUILD" ./bin/*
 

.ci/deploy-windows.sh

@@ -15,6 +15,25 @@ mkdir ./bin/config/input_configs
 curl -fsSL 'https://raw.githubusercontent.com/gabomdq/SDL_GameControllerDB/master/gamecontrollerdb.txt' 1> ./bin/config/input_configs/gamecontrollerdb.txt
 curl -fsSL 'https://rpcs3.net/compatibility?api=v1&export' | iconv -t UTF-8 1> ./bin/GuiConfigs/compat_database.dat
 
+# Download translations
+mkdir -p ./bin/qt6/translations
+ZIP_URL=$(curl -fsSL "https://api.github.com/repos/RPCS3/rpcs3_translations/releases/latest" \
+  | grep "browser_download_url" \
+  | grep "RPCS3-languages.zip" \
+  | cut -d '"' -f 4)
+if [ -z "$ZIP_URL" ]; then
+  echo "Failed to find RPCS3-languages.zip in the latest release. Continuing without translations."
+else
+  echo "Downloading translations from: $ZIP_URL"
+  curl -L -o translations.zip "$ZIP_URL" || {
+    echo "Failed to download translations.zip. Continuing without translations."
+    exit 0
+  }
+  unzip -o translations.zip -d "./bin/qt6/translations" >/dev/null 2>&1 || \
+    echo "Failed to extract translations.zip. Continuing without translations."
+  rm -f translations.zip
+fi
+
 # Download SSL certificate (not needed with CURLSSLOPT_NATIVE_CA)
 #curl -fsSL 'https://curl.haxx.se/ca/cacert.pem' 1> ./bin/cacert.pem
 

.ci/install-freebsd.sh

@@ -12,7 +12,7 @@ pkg info # debug
 pkg install "llvm$LLVM_COMPILER_VER"
 
 # Mandatory dependencies (qtX-base is pulled via qtX-multimedia)
-pkg install git ccache cmake ninja "qt$QT_VER_MAIN-multimedia" "qt$QT_VER_MAIN-svg" glew openal-soft ffmpeg
+pkg install git ccache cmake ninja "qt$QT_VER_MAIN-multimedia" "qt$QT_VER_MAIN-svg" glew openal-soft ffmpeg pcre2
 
 # Optional dependencies (libevdev is pulled by qtX-base)
 pkg install pkgconf alsa-lib pulseaudio sdl3 evdev-proto vulkan-headers vulkan-loader opencv

.ci/setup-windows.sh

@@ -8,7 +8,7 @@ QT_URL_VER=$(echo "$QT_VER" | sed "s/\.//g")
 QT_VER_MSVC_UP=$(echo "${QT_VER_MSVC}" | tr '[:lower:]' '[:upper:]')
 QT_PREFIX="online/qtsdkrepository/windows_x86/desktop/qt${QT_VER_MAIN}_${QT_URL_VER}/qt${QT_VER_MAIN}_${QT_URL_VER}/qt.qt${QT_VER_MAIN}.${QT_URL_VER}."
 QT_PREFIX_2="win64_${QT_VER_MSVC}_64/${QT_VER}-0-${QT_DATE}"
-QT_SUFFIX="-Windows-Windows_11_23H2-${QT_VER_MSVC_UP}-Windows-Windows_11_23H2-X86_64.7z"
+QT_SUFFIX="-Windows-Windows_11_24H2-${QT_VER_MSVC_UP}-Windows-Windows_11_24H2-X86_64.7z"
 QT_BASE_URL="${QT_HOST}${QT_PREFIX}${QT_PREFIX_2}qtbase${QT_SUFFIX}"
 QT_DECL_URL="${QT_HOST}${QT_PREFIX}${QT_PREFIX_2}qtdeclarative${QT_SUFFIX}"
 QT_TOOL_URL="${QT_HOST}${QT_PREFIX}${QT_PREFIX_2}qttools${QT_SUFFIX}"

.github/workflows/rpcs3.yml

@@ -30,23 +30,23 @@ jobs:
       matrix:
         include:
           - os: ubuntu-24.04
-            docker_img: "rpcs3/rpcs3-ci-jammy:1.6"
+            docker_img: "rpcs3/rpcs3-ci-jammy:1.7"
             build_sh: "/rpcs3/.ci/build-linux.sh"
             compiler: clang
             UPLOAD_COMMIT_HASH: d812f1254a1157c80fd402f94446310560f54e5f
             UPLOAD_REPO_FULL_NAME: "rpcs3/rpcs3-binaries-linux"
           - os: ubuntu-24.04
-            docker_img: "rpcs3/rpcs3-ci-jammy:1.6"
+            docker_img: "rpcs3/rpcs3-ci-jammy:1.7"
             build_sh: "/rpcs3/.ci/build-linux.sh"
             compiler: gcc
           - os: ubuntu-24.04-arm
-            docker_img: "rpcs3/rpcs3-ci-jammy-aarch64:1.6"
+            docker_img: "rpcs3/rpcs3-ci-jammy-aarch64:1.7"
             build_sh: "/rpcs3/.ci/build-linux-aarch64.sh"
             compiler: clang
             UPLOAD_COMMIT_HASH: a1d35836e8d45bfc6f63c26f0a3e5d46ef622fe1
             UPLOAD_REPO_FULL_NAME: "rpcs3/rpcs3-binaries-linux-arm64"
           - os: ubuntu-24.04-arm
-            docker_img: "rpcs3/rpcs3-ci-jammy-aarch64:1.6"
+            docker_img: "rpcs3/rpcs3-ci-jammy-aarch64:1.7"
             build_sh: "/rpcs3/.ci/build-linux-aarch64.sh"
             compiler: gcc
     name: RPCS3 Linux ${{ matrix.os }} ${{ matrix.compiler }}
@@ -134,9 +134,9 @@ jobs:
     runs-on: macos-14
     env:
       CCACHE_DIR: /tmp/ccache_dir
-      QT_VER: '6.7.3'
+      QT_VER: '6.10.1'
       QT_VER_MAIN: '6'
-      LLVM_COMPILER_VER: '19'
+      LLVM_COMPILER_VER: '21'
       RELEASE_MESSAGE: ../GitHubReleaseMessage.txt
       UPLOAD_COMMIT_HASH: ${{ matrix.UPLOAD_COMMIT_HASH }}
       UPLOAD_REPO_FULL_NAME: ${{ matrix.UPLOAD_REPO_FULL_NAME }}
@@ -212,9 +212,9 @@ jobs:
     env:
       COMPILER: msvc
       QT_VER_MAIN: '6'
-      QT_VER: '6.9.3'
+      QT_VER: '6.10.1'
       QT_VER_MSVC: 'msvc2022'
-      QT_DATE: '202509261208'
+      QT_DATE: '202511161843'
       LLVM_VER: '19.1.7'
       VULKAN_VER: '1.3.268.0'
       VULKAN_SDK_SHA: '8459ef49bd06b697115ddd3d97c9aec729e849cd775f5be70897718a9b3b9db5'

.gitmodules

@@ -21,7 +21,7 @@
 	ignore = dirty
 [submodule "3rdparty/hidapi"]
 	path = 3rdparty/hidapi/hidapi
-	url = ../../RPCS3/hidapi.git
+	url = ../../libusb/hidapi.git
 	branch = master
 	ignore = dirty
 [submodule "3rdparty/pugixml"]

3rdparty/CMakeLists.txt

@@ -278,7 +278,7 @@ if(USE_FAUDIO)
 			target_compile_definitions(FAudio-static INTERFACE -DHAVE_FAUDIO)
 			set(FAUDIO_TARGET FAudio-static)
 		else()
-			message(FATAL_ERROR
+			message(WARNING
 				"-- RPCS3: 3rdparty FAudio requires SDL 3.2.0 or newer. Since a valid SDL3"
 				">=3.2.0 version cannot be found, building with FAudio will be skipped.")
 			set(USE_FAUDIO OFF CACHE BOOL "Disabled FAudio with SDL < 3.2.0" FORCE)

3rdparty/FAudio

@@ -1 +1 @@
-Subproject commit 8a87fdc9242a7b507cf5c96c539334a1760c1ec7
+Subproject commit 4ea8afea6ba857c24e40877f487d000d559b196d

3rdparty/MoltenVK/CMakeLists.txt

@@ -3,7 +3,7 @@ include(ExternalProject)
 
 ExternalProject_Add(moltenvk
 	GIT_REPOSITORY https://github.com/KhronosGroup/MoltenVK.git
-	GIT_TAG 49b97f2
+	GIT_TAG 4588705
 	BUILD_IN_SOURCE 1
 	SOURCE_DIR ${CMAKE_CURRENT_SOURCE_DIR}/MoltenVK
 	CONFIGURE_COMMAND "${CMAKE_CURRENT_SOURCE_DIR}/MoltenVK/fetchDependencies" --macos

3rdparty/OpenAL/openal-soft

@@ -1 +1 @@
-Subproject commit dc7d7054a5b4f3bec1dc23a42fd616a0847af948
+Subproject commit 0e5e98e4ac8adae92e4f7653dd6eee17aa9c8791

3rdparty/OpenAL/openal-soft.vcxproj

@@ -49,11 +49,11 @@
   <PropertyGroup Label="UserMacros">
     <CmakeReleaseCLI>call vsdevcmd.bat -arch=amd64
       cd "$(SolutionDir)build\tmp\$(ProjectName)-$(Configuration)-$(Platform)"
-      cmake -G Ninja -DCMAKE_CXX_COMPILER="cl.exe" -DCMAKE_C_COMPILER="cl.exe" -DCMAKE_BUILD_TYPE="Release" -DCMAKE_INSTALL_PREFIX="./Release" -DCMAKE_SYSTEM_VERSION=10.0 -DLIBTYPE=STATIC -DFORCE_STATIC_VCRT=true -DALSOFT_UTILS=false -DALSOFT_EXAMPLES=false -DALSOFT_INSTALL=false -DALSOFT_INSTALL_CONFIG=false -DALSOFT_INSTALL_HRTF_DATA=false -DALSOFT_INSTALL_AMBDEC_PRESETS=false -DALSOFT_INSTALL_EXAMPLES=false -DALSOFT_INSTALL_UTILS=false "$(SolutionDir)3rdparty\OpenAL\openal-soft"
+      cmake -G Ninja -DCMAKE_CXX_COMPILER="cl.exe" -DCMAKE_C_COMPILER="cl.exe" -DCMAKE_BUILD_TYPE="Release" -DCMAKE_INSTALL_PREFIX="./Release" -DCMAKE_MSVC_RUNTIME_LIBRARY=MultiThreaded -DCMAKE_SYSTEM_VERSION=10.0 -DLIBTYPE=STATIC -DFORCE_STATIC_VCRT=true -DALSOFT_UTILS=false -DALSOFT_EXAMPLES=false -DALSOFT_INSTALL=false -DALSOFT_INSTALL_CONFIG=false -DALSOFT_INSTALL_HRTF_DATA=false -DALSOFT_INSTALL_AMBDEC_PRESETS=false -DALSOFT_INSTALL_EXAMPLES=false -DALSOFT_INSTALL_UTILS=false "$(SolutionDir)3rdparty\OpenAL\openal-soft"
     </CmakeReleaseCLI>
     <CmakeDebugCLI>call vsdevcmd.bat -arch=amd64
       cd "$(SolutionDir)build\tmp\$(ProjectName)-$(Configuration)-$(Platform)"
-      cmake -G Ninja -DCMAKE_CXX_COMPILER="cl.exe" -DCMAKE_C_COMPILER="cl.exe" -DCMAKE_BUILD_TYPE="Debug" -DCMAKE_INSTALL_PREFIX="./Debug" -DCMAKE_SYSTEM_VERSION=10.0 -DLIBTYPE=STATIC -DALSOFT_UTILS=false -DALSOFT_EXAMPLES=false -DALSOFT_INSTALL=false -DALSOFT_INSTALL_CONFIG=false -DALSOFT_INSTALL_HRTF_DATA=false -DALSOFT_INSTALL_AMBDEC_PRESETS=false -DALSOFT_INSTALL_EXAMPLES=false -DALSOFT_INSTALL_UTILS=false "$(SolutionDir)3rdparty\OpenAL\openal-soft"
+      cmake -G Ninja -DCMAKE_CXX_COMPILER="cl.exe" -DCMAKE_C_COMPILER="cl.exe" -DCMAKE_BUILD_TYPE="Debug" -DCMAKE_INSTALL_PREFIX="./Debug" -DCMAKE_MSVC_RUNTIME_LIBRARY=MultiThreadedDebug -DCMAKE_SYSTEM_VERSION=10.0 -DLIBTYPE=STATIC -DALSOFT_UTILS=false -DALSOFT_EXAMPLES=false -DALSOFT_INSTALL=false -DALSOFT_INSTALL_CONFIG=false -DALSOFT_INSTALL_HRTF_DATA=false -DALSOFT_INSTALL_AMBDEC_PRESETS=false -DALSOFT_INSTALL_EXAMPLES=false -DALSOFT_INSTALL_UTILS=false "$(SolutionDir)3rdparty\OpenAL\openal-soft"
     </CmakeDebugCLI>
     <CmakeCopyCLI>
       echo Copying..

3rdparty/curl/curl

@@ -1 +1 @@
-Subproject commit 11b991232fbcaa88e2b1faecac224416b0001e35
+Subproject commit 400fffa90f30c7a2dc762fa33009d24851bd2016

3rdparty/curl/libcurl.vcxproj

@@ -80,10 +80,12 @@
     <ClCompile Include="curl\lib\cshutdn.c" />
     <ClCompile Include="curl\lib\curlx\base64.c" />
     <ClCompile Include="curl\lib\curlx\dynbuf.c" />
+    <ClCompile Include="curl\lib\curlx\fopen.c" />
     <ClCompile Include="curl\lib\curlx\inet_ntop.c" />
     <ClCompile Include="curl\lib\curlx\inet_pton.c" />
     <ClCompile Include="curl\lib\curlx\multibyte.c" />
     <ClCompile Include="curl\lib\curlx\nonblock.c" />
+    <ClCompile Include="curl\lib\curlx\strerr.c" />
     <ClCompile Include="curl\lib\curlx\strparse.c" />
     <ClCompile Include="curl\lib\curlx\timediff.c" />
     <ClCompile Include="curl\lib\curlx\timeval.c" />
@@ -92,9 +94,9 @@
     <ClCompile Include="curl\lib\curlx\warnless.c" />
     <ClCompile Include="curl\lib\curlx\winapi.c" />
     <ClCompile Include="curl\lib\curl_addrinfo.c" />
-    <ClCompile Include="curl\lib\curl_des.c" />
     <ClCompile Include="curl\lib\curl_endian.c" />
     <ClCompile Include="curl\lib\curl_fnmatch.c" />
+    <ClCompile Include="curl\lib\curl_fopen.c" />
     <ClCompile Include="curl\lib\curl_gethostname.c" />
     <ClCompile Include="curl\lib\curl_get_line.c" />
     <ClCompile Include="curl\lib\curl_gssapi.c" />
@@ -120,7 +122,6 @@
     <ClCompile Include="curl\lib\fake_addrinfo.c" />
     <ClCompile Include="curl\lib\file.c" />
     <ClCompile Include="curl\lib\fileinfo.c" />
-    <ClCompile Include="curl\lib\fopen.c" />
     <ClCompile Include="curl\lib\formdata.c" />
     <ClCompile Include="curl\lib\ftp.c" />
     <ClCompile Include="curl\lib\ftplistparser.c" />
@@ -147,7 +148,6 @@
     <ClCompile Include="curl\lib\idn.c" />
     <ClCompile Include="curl\lib\if2ip.c" />
     <ClCompile Include="curl\lib\imap.c" />
-    <ClCompile Include="curl\lib\krb5.c" />
     <ClCompile Include="curl\lib\ldap.c" />
     <ClCompile Include="curl\lib\llist.c" />
     <ClCompile Include="curl\lib\macos.c" />
@@ -159,6 +159,7 @@
     <ClCompile Include="curl\lib\mqtt.c" />
     <ClCompile Include="curl\lib\multi.c" />
     <ClCompile Include="curl\lib\multi_ev.c" />
+    <ClCompile Include="curl\lib\multi_ntfy.c" />
     <ClCompile Include="curl\lib\netrc.c" />
     <ClCompile Include="curl\lib\noproxy.c" />
     <ClCompile Include="curl\lib\openldap.c" />
@@ -203,6 +204,7 @@
     <ClCompile Include="curl\lib\version.c" />
     <ClCompile Include="curl\lib\vquic\curl_ngtcp2.c" />
     <ClCompile Include="curl\lib\vquic\curl_quiche.c" />
+    <ClCompile Include="curl\lib\vtls\apple.c" />
     <ClCompile Include="curl\lib\vtls\cipher_suite.c" />
     <ClCompile Include="curl\lib\vtls\hostcheck.c" />
     <ClCompile Include="curl\lib\vtls\rustls.c" />
@@ -225,7 +227,6 @@
     <ClCompile Include="curl\lib\vssh\curl_path.c" />
     <ClCompile Include="curl\lib\vssh\libssh.c" />
     <ClCompile Include="curl\lib\vssh\libssh2.c" />
-    <ClCompile Include="curl\lib\vssh\wolfssh.c" />
     <ClCompile Include="curl\lib\vtls\gtls.c" />
     <ClCompile Include="curl\lib\vtls\keylog.c" />
     <ClCompile Include="curl\lib\vtls\mbedtls.c" />
@@ -274,10 +275,12 @@
     <ClInclude Include="curl\lib\curlx\binmode.h" />
     <ClInclude Include="curl\lib\curlx\curlx.h" />
     <ClInclude Include="curl\lib\curlx\dynbuf.h" />
+    <ClInclude Include="curl\lib\curlx\fopen.h" />
     <ClInclude Include="curl\lib\curlx\inet_ntop.h" />
     <ClInclude Include="curl\lib\curlx\inet_pton.h" />
     <ClInclude Include="curl\lib\curlx\multibyte.h" />
     <ClInclude Include="curl\lib\curlx\nonblock.h" />
+    <ClInclude Include="curl\lib\curlx\strerr.h" />
     <ClInclude Include="curl\lib\curlx\strparse.h" />
     <ClInclude Include="curl\lib\curlx\timediff.h" />
     <ClInclude Include="curl\lib\curlx\timeval.h" />
@@ -287,14 +290,13 @@
     <ClInclude Include="curl\lib\curlx\winapi.h" />
     <ClInclude Include="curl\lib\curl_addrinfo.h" />
     <ClInclude Include="curl\lib\curl_ctype.h" />
-    <ClInclude Include="curl\lib\curl_des.h" />
     <ClInclude Include="curl\lib\curl_endian.h" />
     <ClInclude Include="curl\lib\curl_fnmatch.h" />
+    <ClInclude Include="curl\lib\curl_fopen.h" />
     <ClInclude Include="curl\lib\curl_gethostname.h" />
     <ClInclude Include="curl\lib\curl_get_line.h" />
     <ClInclude Include="curl\lib\curl_gssapi.h" />
     <ClInclude Include="curl\lib\curl_hmac.h" />
-    <ClInclude Include="curl\lib\curl_krb5.h" />
     <ClInclude Include="curl\lib\curl_ldap.h" />
     <ClInclude Include="curl\lib\curl_md4.h" />
     <ClInclude Include="curl\lib\curl_md5.h" />
@@ -325,7 +327,6 @@
     <ClInclude Include="curl\lib\fake_addrinfo.h" />
     <ClInclude Include="curl\lib\file.h" />
     <ClInclude Include="curl\lib\fileinfo.h" />
-    <ClInclude Include="curl\lib\fopen.h" />
     <ClInclude Include="curl\lib\formdata.h" />
     <ClInclude Include="curl\lib\ftp.h" />
     <ClInclude Include="curl\lib\ftplistparser.h" />
@@ -357,6 +358,7 @@
     <ClInclude Include="curl\lib\multihandle.h" />
     <ClInclude Include="curl\lib\multiif.h" />
     <ClInclude Include="curl\lib\multi_ev.h" />
+    <ClInclude Include="curl\lib\multi_ntfy.h" />
     <ClInclude Include="curl\lib\netrc.h" />
     <ClInclude Include="curl\lib\noproxy.h" />
     <ClInclude Include="curl\lib\parsedate.h" />
@@ -403,6 +405,7 @@
     <ClInclude Include="curl\lib\vquic\curl_ngtcp2.h" />
     <ClInclude Include="curl\lib\vquic\curl_quiche.h" />
     <ClInclude Include="curl\lib\vquic\vquic_int.h" />
+    <ClInclude Include="curl\lib\vtls\apple.h" />
     <ClInclude Include="curl\lib\vtls\cipher_suite.h" />
     <ClInclude Include="curl\lib\vtls\hostcheck.h" />
     <ClInclude Include="curl\lib\vtls\rustls.h" />

3rdparty/curl/libcurl.vcxproj.filters

@@ -42,9 +42,6 @@
     <ClCompile Include="curl\lib\curl_addrinfo.c">
       <Filter>Source Files</Filter>
     </ClCompile>
-    <ClCompile Include="curl\lib\curl_des.c">
-      <Filter>Source Files</Filter>
-    </ClCompile>
     <ClCompile Include="curl\lib\curl_endian.c">
       <Filter>Source Files</Filter>
     </ClCompile>
@@ -159,9 +156,6 @@
     <ClCompile Include="curl\lib\imap.c">
       <Filter>Source Files</Filter>
     </ClCompile>
-    <ClCompile Include="curl\lib\krb5.c">
-      <Filter>Source Files</Filter>
-    </ClCompile>
     <ClCompile Include="curl\lib\ldap.c">
       <Filter>Source Files</Filter>
     </ClCompile>
@@ -333,9 +327,6 @@
     <ClCompile Include="curl\lib\vssh\libssh2.c">
       <Filter>Source Files</Filter>
     </ClCompile>
-    <ClCompile Include="curl\lib\vssh\wolfssh.c">
-      <Filter>Source Files</Filter>
-    </ClCompile>
     <ClCompile Include="curl\lib\vtls\gtls.c">
       <Filter>Source Files</Filter>
     </ClCompile>
@@ -396,9 +387,6 @@
     <ClCompile Include="curl\lib\headers.c">
       <Filter>Source Files</Filter>
     </ClCompile>
-    <ClCompile Include="curl\lib\fopen.c">
-      <Filter>Source Files</Filter>
-    </ClCompile>
     <ClCompile Include="curl\lib\noproxy.c">
       <Filter>Source Files</Filter>
     </ClCompile>
@@ -546,6 +534,21 @@
     <ClCompile Include="curl\lib\cf-ip-happy.c">
       <Filter>Source Files</Filter>
     </ClCompile>
+    <ClCompile Include="curl\lib\curl_fopen.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="curl\lib\multi_ntfy.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="curl\lib\vtls\apple.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="curl\lib\curlx\fopen.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="curl\lib\curlx\strerr.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
   </ItemGroup>
   <ItemGroup>
     <ClInclude Include="curl\include\curl\curl.h">
@@ -623,9 +626,6 @@
     <ClInclude Include="curl\lib\curl_ctype.h">
       <Filter>Header Files</Filter>
     </ClInclude>
-    <ClInclude Include="curl\lib\curl_des.h">
-      <Filter>Header Files</Filter>
-    </ClInclude>
     <ClInclude Include="curl\lib\curl_endian.h">
       <Filter>Header Files</Filter>
     </ClInclude>
@@ -926,9 +926,6 @@
     <ClInclude Include="curl\lib\vtls\keylog.h">
       <Filter>Header Files</Filter>
     </ClInclude>
-    <ClInclude Include="curl\lib\curl_krb5.h">
-      <Filter>Header Files</Filter>
-    </ClInclude>
     <ClInclude Include="curl\lib\easyoptions.h">
       <Filter>Header Files</Filter>
     </ClInclude>
@@ -953,9 +950,6 @@
     <ClInclude Include="curl\lib\easy_lock.h">
       <Filter>Header Files</Filter>
     </ClInclude>
-    <ClInclude Include="curl\lib\fopen.h">
-      <Filter>Header Files</Filter>
-    </ClInclude>
     <ClInclude Include="curl\lib\functypes.h">
       <Filter>Header Files</Filter>
     </ClInclude>
@@ -1112,6 +1106,21 @@
     <ClInclude Include="curl\lib\curl_mem_undef.h">
       <Filter>Header Files</Filter>
     </ClInclude>
+    <ClInclude Include="curl\lib\curl_fopen.h">
+      <Filter>Header Files</Filter>
+    </ClInclude>
+    <ClInclude Include="curl\lib\multi_ntfy.h">
+      <Filter>Header Files</Filter>
+    </ClInclude>
+    <ClInclude Include="curl\lib\vtls\apple.h">
+      <Filter>Header Files</Filter>
+    </ClInclude>
+    <ClInclude Include="curl\lib\curlx\fopen.h">
+      <Filter>Header Files</Filter>
+    </ClInclude>
+    <ClInclude Include="curl\lib\curlx\strerr.h">
+      <Filter>Header Files</Filter>
+    </ClInclude>
   </ItemGroup>
   <ItemGroup>
     <ResourceCompile Include="curl\lib\libcurl.rc">

3rdparty/fusion/fusion

@@ -1 +1 @@
-Subproject commit 066d4a63b2c714b20b0a8073a01fda7c5c6763f6
+Subproject commit 759ac5d698baefca53f1975a0bb1d2dcbdb9f836

3rdparty/hidapi/hidapi

@@ -1 +1 @@
-Subproject commit f42423643ec9011c98cccc0bb790722bbbd3f30b
+Subproject commit d6b2a974608dec3b76fb1e36c189f22b9cf3650c

3rdparty/libpng/libpng

@@ -1 +1 @@
-Subproject commit 2b978915d82377df13fcbb1fb56660195ded868a
+Subproject commit 49363adcfaf098748d7a4c8c624ad8c45a8c3a86

3rdparty/libsdl-org/SDL

@@ -1 +1 @@
-Subproject commit a96677bdf6b4acb84af4ec294e5f60a4e8cbbe03
+Subproject commit 7f3ae3d57459e59943a4ecfefc8f6277ec6bf540

3rdparty/libsdl-org/SDL.vcxproj

@@ -384,7 +384,6 @@
     <ClCompile Include="SDL\src\render\opengl\SDL_shaders_gl.c" />
     <ClCompile Include="SDL\src\render\opengles2\SDL_render_gles2.c" />
     <ClCompile Include="SDL\src\render\opengles2\SDL_shaders_gles2.c" />
-    <ClCompile Include="SDL\src\render\SDL_d3dmath.c" />
     <ClCompile Include="SDL\src\render\SDL_render.c" />
     <ClCompile Include="SDL\src\render\SDL_render_unsupported.c" />
     <ClCompile Include="SDL\src\render\SDL_yuv_sw.c" />

3rdparty/libsdl-org/SDL.vcxproj.filters

@@ -172,9 +172,6 @@
     <Filter Include="render\direct3d12">
       <UniqueIdentifier>{f48c2b17-1bee-4fec-a7c8-24cf619abe08}</UniqueIdentifier>
     </Filter>
-    <Filter Include="video\intrin">
-      <UniqueIdentifier>{653672cc-90ae-4eba-a256-6479f2c31804}</UniqueIdentifier>
-    </Filter>
     <Filter Include="main">
       <UniqueIdentifier>{00001967ea2801028a046a722a070000}</UniqueIdentifier>
     </Filter>
@@ -1460,9 +1457,6 @@
     <ClCompile Include="SDL\src\sensor\windows\SDL_windowssensor.c">
       <Filter>sensor\windows</Filter>
     </ClCompile>
-    <ClCompile Include="SDL\src\render\SDL_d3dmath.c">
-      <Filter>render</Filter>
-    </ClCompile>
     <ClCompile Include="SDL\src\render\SDL_render.c">
       <Filter>render</Filter>
     </ClCompile>

3rdparty/qt6.cmake

@@ -6,14 +6,15 @@ find_package(Qt6 ${QT_MIN_VER} CONFIG COMPONENTS Widgets Concurrent Multimedia M
 if(WIN32)
 	target_link_libraries(3rdparty_qt6 INTERFACE Qt6::Widgets Qt6::Concurrent Qt6::Multimedia Qt6::MultimediaWidgets Qt6::Svg Qt6::SvgWidgets)
 else()
-	find_package(Qt6 ${QT_MIN_VER} COMPONENTS DBus Gui)
+	set(QT_NO_PRIVATE_MODULE_WARNING ON)
+	find_package(Qt6 ${QT_MIN_VER} COMPONENTS DBus Gui GuiPrivate)
 	if(Qt6DBus_FOUND)
 		target_link_libraries(3rdparty_qt6 INTERFACE Qt6::Widgets Qt6::DBus Qt6::Concurrent Qt6::Multimedia Qt6::MultimediaWidgets Qt6::Svg Qt6::SvgWidgets)
 		target_compile_definitions(3rdparty_qt6 INTERFACE -DHAVE_QTDBUS)
 	else()
 		target_link_libraries(3rdparty_qt6 INTERFACE Qt6::Widgets Qt6::Concurrent Qt6::Multimedia Qt6::MultimediaWidgets Qt6::Svg Qt6::SvgWidgets)
 	endif()
-	target_include_directories(3rdparty_qt6 INTERFACE ${Qt6Gui_PRIVATE_INCLUDE_DIRS})
+	target_link_libraries(3rdparty_qt6 INTERFACE Qt6::GuiPrivate)
 endif()
 
 if(Qt6Widgets_FOUND)

3rdparty/zstd/CMakeLists.txt

@@ -10,7 +10,7 @@ else()
   option(ZSTD_BUILD_STATIC "BUILD STATIC LIBRARIES" ON)
   option(ZSTD_BUILD_TESTS "BUILD TESTS" OFF)
 
-  add_subdirectory(zstd/build/cmake EXLUDE_FROM_ALL)
+  add_subdirectory(zstd/build/cmake EXCLUDE_FROM_ALL)
   add_library(3rdparty_zstd INTERFACE)
   target_link_libraries(3rdparty_zstd INTERFACE libzstd_static)
 endif()

BUILDING.md

@@ -8,37 +8,38 @@ Other instructions may be found [here](https://wiki.rpcs3.net/index.php?title=Bu
 ### Windows 10 or later
 
 The following tools are required to build RPCS3 on Windows 10 or later:
-- [Visual Studio 2022](https://visualstudio.microsoft.com/thank-you-downloading-visual-studio/?sku=Community)
+- [Visual Studio 2022/2026](https://visualstudio.microsoft.com/thank-you-downloading-visual-studio/?sku=Community)
 - **Optional** - [CMake 3.28.0+](https://www.cmake.org/download/) (add to PATH)
 
   **NOTES:**
-  - **Visual Studio 2022** integrates **CMake 3.29+** and it also supports both the `sln` solution (`.sln`, `.vcxproj`) and `CMake` solution (`CMakeLists.txt`, `CMakePresets.json`).
+  - **Visual Studio 2026** needs at least **CMake 4.2.0+**.
+  - **Visual Studio 2022/2026** integrates **CMake 3.29+** and it also supports both the `sln` solution (`.sln`, `.vcxproj`) and `CMake` solution (`CMakeLists.txt`, `CMakePresets.json`).
     See sections [Building with Visual Studio sln solution](#building-with-visual-studio-sln-solution) and [Building with Visual Studio CMake solution](#building-with-visual-studio-cmake-solution)
     on how to build the project with **Visual Studio**.
   - Install and use this standalone **CMake** tool just in case of your preference. See section [Building with standalone CMake tool](#building-with-standalone-cmake-tool) on how to build the project
     with standalone **CMake** tool.
 
 - [Python 3.6+](https://www.python.org/downloads/) (add to PATH)
-- [Qt 6.9.3](https://www.qt.io/download-qt-installer) In case you can't download from the official installer, you can use [Another Qt installer](https://github.com/miurahr/aqtinstall) (In that case you will need to manually add the "qtmultimedia" module when installing Qt)
+- [Qt 6.10.1](https://www.qt.io/download-qt-installer) In case you can't download from the official installer, you can use [Another Qt installer](https://github.com/miurahr/aqtinstall) (In that case you will need to manually add the "qtmultimedia" module when installing Qt)
 - [Vulkan SDK 1.3.268.0](https://vulkan.lunarg.com/sdk/home) (see "Install the SDK" [here](https://vulkan.lunarg.com/doc/sdk/latest/windows/getting_started.html)) for now future SDKs don't work. You need precisely 1.3.268.0.
 
 The `sln` solution available only on **Visual Studio** is the preferred building solution. It easily allows to build the **RPCS3** application in `Release` and `Debug` mode.
 
 In order to build **RPCS3** with the `sln` solution (with **Visual Studio**), **Qt** libs need to be detected. To detect the libs:
-- add and set the `QTDIR` environment variable, e.g. `<QtInstallFolder>\6.9.3\msvc2022_64\`
+- add and set the `QTDIR` environment variable, e.g. `<QtInstallFolder>\6.10.1\msvc2022_64\`
 - or use the [Visual Studio Qt Plugin](https://marketplace.visualstudio.com/items?itemName=TheQtCompany.QtVisualStudioTools2022)
 
   **NOTE:** If you have issues with the **Visual Studio Qt Plugin**, you may want to uninstall it and install the [Legacy Qt Plugin](https://marketplace.visualstudio.com/items?itemName=TheQtCompany.LEGACYQtVisualStudioTools2022) instead.
 
 In order to build **RPCS3** with the `CMake` solution (with both **Visual Studio** and standalone **CMake** tool):
-- add and set the `Qt6_ROOT` environment variable to the **Qt** libs path, e.g. `<QtInstallFolder>\6.9.3\msvc2022_64\`
+- add and set the `Qt6_ROOT` environment variable to the **Qt** libs path, e.g. `<QtInstallFolder>\6.10.1\msvc2022_64\`
 
 ### Linux
 
 These are the essentials tools to build RPCS3 on Linux. Some of them can be installed through your favorite package manager:
 - Clang 17+ or GCC 13+
 - [CMake 3.28.0+](https://www.cmake.org/download/)
-- [Qt 6.9.3](https://www.qt.io/download-qt-installer)
+- [Qt 6.10.1](https://www.qt.io/download-qt-installer)
 - [Vulkan SDK 1.3.268.0](https://vulkan.lunarg.com/sdk/home) (See "Install the SDK" [here](https://vulkan.lunarg.com/doc/sdk/latest/linux/getting_started.html)) for now future SDKs don't work. You need precisely 1.3.268.0.
 - [SDL3](https://github.com/libsdl-org/SDL/releases) (for the FAudio backend)
 
@@ -121,7 +122,7 @@ Start **Visual Studio**, click on `Open a project or solution` and select the `r
 ##### Configuring the Qt Plugin (if used)
 
 1) go to `Extensions->Qt VS Tools->Qt Versions`
-2) add the path to your Qt installation with compiler e.g. `<QtInstallFolder>\6.9.3\msvc2022_64`, version will fill in automatically
+2) add the path to your Qt installation with compiler e.g. `<QtInstallFolder>\6.10.1\msvc2022_64`, version will fill in automatically
 3) go to `Extensions->Qt VS Tools->Options->Legacy Project Format`. (Only available in the **Legacy Qt Plugin**)
 4) set `Build: Run pre-build setup` to `true`. (Only available in the **Legacy Qt Plugin**)
 

Utilities/Config.cpp

@@ -166,6 +166,55 @@ bool try_to_uint64(u64* out, std::string_view value, u64 min, u64 max)
 	return true;
 }
 
+bool try_to_uint128(u128* out, std::string_view value)
+{
+	if (value.empty())
+	{
+		if (out) cfg_log.error("cfg::try_to_uint128(): called with an empty string");
+		return false;
+	}
+
+	u64 result_low = 0, result_high = 0;
+	const char* start_high64 = value.data();
+	const char* end = value.data() + value.size();
+
+	if (start_high64[0] == '0' && value.size() >= 2 && (start_high64[1] == 'x' || start_high64[1] == 'X'))
+	{
+		// Hex support
+		start_high64 += 2;
+	}
+
+	const char* start_low64 = end - std::min<usz>(end - start_high64, 16);
+
+	// Hexadecimal-only
+	constexpr int base = 16;
+
+	auto ret = std::from_chars(start_low64, end, result_low, base);
+
+	if (ret.ec != std::errc() || ret.ptr != end)
+	{
+		if (out) cfg_log.error("cfg::try_to_uint128('%s'): invalid integer", value);
+		return false;
+	}
+
+	if (start_high64 == start_low64)
+	{
+		if (out) *out = result_low;
+		return true;
+	}
+
+	ret = std::from_chars(start_high64, start_low64, result_high, base);
+
+	if (ret.ec != std::errc() || ret.ptr != start_low64)
+	{
+		if (out) cfg_log.error("cfg::try_to_uint128('%s'): invalid integer", value);
+		return false;
+	}
+
+	if (out) *out = result_low + (u128{result_high} << 64);
+	return true;
+}
+
 std::vector<std::string> cfg::make_float_range(f64 min, f64 max)
 {
 	return {std::to_string(min), std::to_string(max)};
@@ -278,6 +327,19 @@ bool cfg::try_to_enum_value(u64* out, decltype(&fmt_class_string<int>::format) f
 	return true;
 }
 
+std::string cfg::uint128::to_string(u128 value) noexcept
+{
+	std::string result = "0x";
+	result.resize(result.size() + 32);
+
+	for (u32 i = 0; i < 32; i++)
+	{
+		result[result.size() - 1 - i] = "0123456789ABCDEF"[static_cast<u64>(value >> (i * 4)) % 16];
+	}
+
+	return result;
+}
+
 std::vector<std::string> cfg::try_to_enum_list(decltype(&fmt_class_string<int>::format) func)
 {
 	std::vector<std::string> result;

Utilities/Config.h

@@ -38,6 +38,7 @@ namespace cfg
 		_enum, // cfg::_enum type
 		_int, // cfg::_int type
 		uint, // cfg::uint type
+		uint128, // cfg::uint128 type
 		string, // cfg::string type
 		set, // cfg::set_entry type
 		map, // cfg::map_entry type
@@ -578,6 +579,86 @@ namespace cfg
 	// Alias for 64 bit int
 	using uint64 = uint<0, u64{umax}>;
 
+	// Unsigned 128-bit integer entry.
+	class uint128 final : public _base
+	{
+		using int_type = u128;
+
+		atomic_t<int_type> m_value{};
+		int_type original_def = 0;
+
+	public:
+		int_type def;
+
+		uint128(node* owner, const std::string& name, int_type def = 0, bool dynamic = false)
+			: _base(type::uint128, owner, name, dynamic)
+			, m_value(def)
+			, original_def(def)
+			, def(def)
+		{
+		}
+
+		operator int_type() const
+		{
+			return m_value;
+		}
+
+		operator ullong() const
+		{
+			return static_cast<ullong>(m_value.load());
+		}
+
+		int_type get() const
+		{
+			return m_value;
+		}
+
+		void from_default() override
+		{
+			m_value = def;
+		}
+
+		void restore_defaults() override
+		{
+			def = original_def;
+		}
+
+		static std::string to_string(u128 value) noexcept;
+
+		std::string to_string() const override
+		{
+			return to_string(m_value.load());
+		}
+
+		std::string def_to_string() const override
+		{
+			return to_string(def);
+		}
+
+		bool from_string(std::string_view value, bool /*dynamic*/ = false) override
+		{
+			u128 result;
+			if (try_to_uint128(&result, value))
+			{
+				m_value = result;
+				return true;
+			}
+
+			return false;
+		}
+
+		void set(u128 value)
+		{
+			m_value = value;
+		}
+
+		std::vector<std::string> to_list() const override
+		{
+			// Should not be used
+			return make_uint_range(0, 1);
+		}
+	};
+
 	// Simple string entry with mutex
 	class string : public _base
 	{

Utilities/JITLLVM.cpp

@@ -658,7 +658,11 @@ jit_compiler::jit_compiler(const std::unordered_map<std::string, u64>& _link, co
 	std::string result;
 
 	auto null_mod = std::make_unique<llvm::Module> ("null_", *m_context);
+#if LLVM_VERSION_MAJOR >= 21 && (LLVM_VERSION_MINOR >= 1 || LLVM_VERSION_MAJOR >= 22)
+	null_mod->setTargetTriple(llvm::Triple(jit_compiler::triple1()));
+#else
 	null_mod->setTargetTriple(jit_compiler::triple1());
+#endif
 
 	std::unique_ptr<llvm::RTDyldMemoryManager> mem;
 
@@ -672,7 +676,11 @@ jit_compiler::jit_compiler(const std::unordered_map<std::string, u64>& _link, co
 		else
 		{
 			mem = std::make_unique<MemoryManager2>(std::move(symbols_cement));
+#if LLVM_VERSION_MAJOR >= 21 && (LLVM_VERSION_MINOR >= 1 || LLVM_VERSION_MAJOR >= 22)
+			null_mod->setTargetTriple(llvm::Triple(jit_compiler::triple2()));
+#else
 			null_mod->setTargetTriple(jit_compiler::triple2());
+#endif
 		}
 	}
 	else

Utilities/StrFmt.h

@@ -394,7 +394,7 @@ namespace fmt
 		}
 
 #if !defined(_MSC_VER) || defined(__clang__)
-		[[noreturn]] ~throw_exception();
+		[[noreturn]] ~throw_exception() = default;
 #endif
 	};
 

Utilities/StrUtil.h

@@ -28,6 +28,9 @@ bool try_to_int64(s64* out, std::string_view value, s64 min, s64 max);
 // Convert string to unsigned integer
 bool try_to_uint64(u64* out, std::string_view value, u64 min, u64 max);
 
+// Convert string to unsigned int128_t
+bool try_to_uint128(u128* out, std::string_view value);
+
 // Convert string to float
 bool try_to_float(f64* out, std::string_view value, f64 min, f64 max);
 

Utilities/Thread.cpp

@@ -106,6 +106,11 @@ thread_local u64 g_tls_wait_fail = 0;
 thread_local bool g_tls_access_violation_recovered = false;
 extern thread_local std::string(*g_tls_log_prefix)();
 
+namespace stx
+{
+	atomic_t<u32> g_launch_retainer{0};
+}
+
 // Report error and call std::abort(), defined in main.cpp
 [[noreturn]] void report_fatal_error(std::string_view text, bool is_html = false, bool include_help_text = true);
 

Utilities/Thread.h

@@ -465,6 +465,8 @@ public:
 namespace stx
 {
 	struct launch_retainer;
+
+	extern atomic_t<u32> g_launch_retainer;
 }
 
 // Derived from the callable object Context, possibly a lambda
@@ -481,6 +483,11 @@ class named_thread final : public Context, result_storage<Context>, thread_base
 
 	u64 entry_point2()
 	{
+		while (u32 value = stx::g_launch_retainer)
+		{
+			stx::g_launch_retainer.wait(value);
+		}
+
 		thread::initialize([]()
 		{
 			if constexpr (!result::empty)

Utilities/deferred_op.hpp

@@ -0,0 +1,27 @@
+#pragma once
+
+// Generic deferred routine wrapper
+// Use-case is similar to "defer" statement in other languages, just invokes a callback when the object goes out of scope
+
+#include <functional>
+
+namespace utils
+{
+	template <typename F>
+		requires std::is_invocable_v<F>
+	class deferred_op
+	{
+	public:
+		deferred_op(F&& callback)
+			: m_callback(callback)
+		{}
+
+		~deferred_op()
+		{
+			m_callback();
+		}
+
+	private:
+		F m_callback;
+	};
+}

bin/GuiConfigs/YoRHa by Ani.qss

@@ -201,7 +201,7 @@ QPushButton::disabled {
 /* QSpinBox (Settings -> Emulator -> width/height) */
 /* QDoubleSpinBox (Pads -> Mouse Acceleration -> x/y) */
 QSpinBox, QDoubleSpinBox {
-	height: 0.1em;
+	height: 1.50em;
 	background-color: #b3ac98;
 }
 QSpinBox::disabled, QDoubleSpinBox::disabled {

buildfiles/cmake/ConfigureCompiler.cmake

@@ -5,7 +5,6 @@ if(MSVC)
 	add_compile_definitions(
 		_CRT_SECURE_NO_DEPRECATE=1 _CRT_NON_CONFORMING_SWPRINTFS=1 _SCL_SECURE_NO_WARNINGS=1
 		NOMINMAX _ENABLE_EXTENDED_ALIGNED_STORAGE=1 _HAS_EXCEPTIONS=0)
-	add_link_options(/DYNAMICBASE)
 
 	#TODO: Some of these could be cleaned up
 	add_compile_options(/wd4805) # Comparing boolean and int

buildfiles/cmake/FindFFMPEG.cmake

@@ -66,7 +66,7 @@ else ()
 
   find_package_handle_standard_args(FFMPEG
     DEFAULT_MSG
-    FFMPEG_LIBAVCODEC FFMPEG_LIBAVFORMAT FFMPEG_LIBSWSCALE FFMPEG_LIBSWRESAMPLE
+    FFMPEG_LIBAVCODEC FFMPEG_LIBAVFORMAT FFMPEG_LIBAVUTIL FFMPEG_LIBSWSCALE FFMPEG_LIBSWRESAMPLE
   )
 
   if (FFMPEG_FOUND)

objdump.cpp

@@ -25,6 +25,7 @@
 #include <string>
 #include <vector>
 #include <charconv>
+#include <string_view>
 
 std::string to_hex(std::uint64_t value, bool prfx = true)
 {
@@ -85,7 +86,7 @@ int main(int argc, char* argv[])
 			// Decode address and try to find the object
 			std::uint64_t addr = -1;
 
-			std::from_chars(arg.data() + strlen("--start-address=0x"), arg.data() + arg.size(), addr, 16);
+			std::from_chars(arg.data() + ("--start-address=0x"sv).size(), arg.data() + arg.size(), addr, 16);
 
 			for (int j = 0; j < 0x100'0000; j++)
 			{

rpcs3/CMakeLists.txt

@@ -10,16 +10,6 @@ include(CheckFunctionExists)
 
 set(CMAKE_CXX_STANDARD 20)
 
-set(ADDITIONAL_LIBS "")
-if(CMAKE_SYSTEM_NAME STREQUAL "Linux")
-    #on some Linux distros shm_unlink and similar functions are in librt only
-    list(APPEND ADDITIONAL_LIBS "rt")
-elseif(NOT WIN32 AND NOT CMAKE_CXX_FLAGS MATCHES "LIBICONV_PLUG")
-    #it seems like glibc includes the iconv functions we use but other libc
-    #implementations like the one on OSX don't seem implement them
-    list(APPEND ADDITIONAL_LIBS "iconv")
-endif()
-
 if(UNIX AND NOT APPLE AND NOT ANDROID)
     add_compile_definitions(DATADIR="${CMAKE_INSTALL_FULL_DATADIR}/rpcs3")
     # Optionally enable X11 for window management
@@ -78,8 +68,16 @@ if (NOT ANDROID)
             3rdparty::libcurl
             3rdparty::zlib
             3rdparty::opencv
-            3rdparty::fusion
+            3rdparty::fusion)
-            ${ADDITIONAL_LIBS})
+
+    if(CMAKE_SYSTEM_NAME STREQUAL "Linux")
+        #on some Linux distros shm_unlink and similar functions are in librt only
+        target_link_libraries(rpcs3_lib PRIVATE rt)
+    elseif(NOT WIN32 AND NOT CMAKE_CXX_FLAGS MATCHES "LIBICONV_PLUG")
+        #it seems like glibc includes the iconv functions we use but other libc
+        #implementations like the one on OSX don't seem implement them
+        target_link_libraries(rpcs3_lib PRIVATE iconv)
+    endif()
 
     # Unix display manager
     if(X11_FOUND)
@@ -106,19 +104,16 @@ if (NOT ANDROID)
     endif()
 
     # Build rpcs3 executable
+    add_executable(rpcs3 WIN32 MACOSX_BUNDLE)
     if(WIN32)
-        add_executable(rpcs3 WIN32)
         target_sources(rpcs3 PRIVATE rpcs3.rc)
         target_compile_definitions(rpcs3 PRIVATE UNICODE _UNICODE)
     elseif(APPLE)
-        add_executable(rpcs3 MACOSX_BUNDLE)
         target_sources(rpcs3 PRIVATE rpcs3.icns update_helper.sh)
         set_source_files_properties(update_helper.sh PROPERTIES MACOSX_PACKAGE_LOCATION Resources)
         set_target_properties(rpcs3
             PROPERTIES
                 MACOSX_BUNDLE_INFO_PLIST "${CMAKE_CURRENT_SOURCE_DIR}/rpcs3.plist.in")
-    else()
-        add_executable(rpcs3)
     endif()
 
     target_sources(rpcs3
@@ -137,17 +132,12 @@ if (NOT ANDROID)
 
     # Copy icons to executable directory
     if(APPLE)
-        if (CMAKE_BUILD_TYPE MATCHES "Debug" OR CMAKE_BUILD_TYPE MATCHES "RelWithDebInfo")
-            set(QT_DEPLOY_FLAGS "-no-strip")
-        else()
-            set(QT_DEPLOY_FLAGS "")
-        endif()
         qt_finalize_target(rpcs3)
         add_custom_command(TARGET rpcs3 POST_BUILD
             COMMAND ${CMAKE_COMMAND} -E copy ${CMAKE_CURRENT_SOURCE_DIR}/rpcs3.icns $<TARGET_FILE_DIR:rpcs3>/../Resources/rpcs3.icns
             COMMAND ${CMAKE_COMMAND} -E copy_directory ${CMAKE_SOURCE_DIR}/bin/Icons $<TARGET_FILE_DIR:rpcs3>/../Resources/Icons
             COMMAND ${CMAKE_COMMAND} -E copy_directory ${CMAKE_SOURCE_DIR}/bin/GuiConfigs $<TARGET_FILE_DIR:rpcs3>/../Resources/GuiConfigs
-            COMMAND "${MACDEPLOYQT_EXECUTABLE}" "${PROJECT_BINARY_DIR}/bin/rpcs3.app" "${QT_DEPLOY_FLAGS}")
+            COMMAND "${MACDEPLOYQT_EXECUTABLE}" "${PROJECT_BINARY_DIR}/bin/rpcs3.app" "$<$<CONFIG:Debug,RelWithDebInfo>:-no-strip>")
     elseif(UNIX)
         add_custom_command(TARGET rpcs3 POST_BUILD
             COMMAND ${CMAKE_COMMAND} -E copy_directory ${CMAKE_SOURCE_DIR}/bin/Icons $<TARGET_FILE_DIR:rpcs3>/Icons
@@ -198,8 +188,11 @@ if(BUILD_RPCS3_TESTS)
         PRIVATE
             tests/test.cpp
             tests/test_fmt.cpp
+            tests/test_pair.cpp
+            tests/test_tuple.cpp
             tests/test_simple_array.cpp
             tests/test_address_range.cpp
+            tests/test_rsx_cfg.cpp
     )
 
     target_link_libraries(rpcs3_test

rpcs3/Crypto/key_vault.cpp

@@ -11,10 +11,10 @@ SELF_KEY::SELF_KEY(u64 ver_start, u64 ver_end, u16 rev, u32 type, const std::str
 	version_end   = ver_end;
 	revision      = rev;
 	self_type     = type;
-	hex_to_bytes(erk, e.c_str(), 0);
+	hex_to_bytes(erk, e, 0);
-	hex_to_bytes(riv, r.c_str(), 0);
+	hex_to_bytes(riv, r, 0);
-	hex_to_bytes(pub, pb.c_str(), 0);
+	hex_to_bytes(pub, pb, 0);
-	hex_to_bytes(priv, pr.c_str(), 0);
+	hex_to_bytes(priv, pr, 0);
 	curve_type = ct;
 }
 

rpcs3/Crypto/utils.cpp

@@ -7,9 +7,11 @@
 #include "sha1.h"
 #include "sha256.h"
 #include "key_vault.h"
+#include <charconv>
+#include <cstdlib>
 #include <cstring>
-#include <stdio.h>
+#include <cstdio>
-#include <time.h>
+#include <ctime>
 #include "Utilities/StrUtil.h"
 #include "Utilities/File.h"
 
@@ -21,49 +23,23 @@
 // Auxiliary functions (endian swap, xor).
 
 // Hex string conversion auxiliary functions.
-u64 hex_to_u64(const char* hex_str)
+void hex_to_bytes(unsigned char* data, std::string_view hex_str, unsigned int str_length)
 {
-	auto length = std::strlen(hex_str);
+	const auto strn_length = (str_length > 0) ? str_length : hex_str.size();
-	u64 tmp = 0;
-	u64 result = 0;
-	char c;
-
-	while (length--)
-	{
-		c = *hex_str++;
-		if((c >= '0') && (c <= '9'))
-			tmp = c - '0';
-		else if((c >= 'a') && (c <= 'f'))
-			tmp = c - 'a' + 10;
-		else if((c >= 'A') && (c <= 'F'))
-			tmp = c - 'A' + 10;
-		else
-			tmp = 0;
-		result |= (tmp << (length * 4));
-	}
-
-	return result;
-}
-
-void hex_to_bytes(unsigned char* data, const char* hex_str, unsigned int str_length)
-{
-	const auto strn_length = (str_length > 0) ? str_length : std::strlen(hex_str);
-	auto data_length = strn_length / 2;
-	char tmp_buf[3] = {0, 0, 0};
 
 	// Don't convert if the string length is odd.
 	if ((strn_length % 2) == 0)
 	{
-		while (data_length--)
+		for (size_t i = 0; i < strn_length; i += 2)
 		{
-			tmp_buf[0] = *hex_str++;
+			const auto [ptr, err] = std::from_chars(hex_str.data() + i, hex_str.data() + i + 2, *data++, 16);
-			tmp_buf[1] = *hex_str++;
+			if (err != std::errc())
-
+			{
-			*data++ = static_cast<u8>(hex_to_u64(tmp_buf) & 0xFF);
+				fmt::throw_exception("Failed to read hex string: %s", std::make_error_code(err).message());
+			}
 		}
 	}
 }
-
 
 // Crypto functions (AES128-CBC, AES128-ECB, SHA1-HMAC and AES-CMAC).
 void aescbc128_decrypt(unsigned char *key, unsigned char *iv, unsigned char *in, unsigned char *out, usz len)
@@ -181,7 +157,7 @@ std::array<u8, PASSPHRASE_KEY_LEN> sc_combine_laid_paid(s64 laid, s64 paid)
 {
 	const std::string paid_laid = fmt::format("%016llx%016llx", laid, paid);
 	std::array<u8, PASSPHRASE_KEY_LEN> out{};
-	hex_to_bytes(out.data(), paid_laid.c_str(), PASSPHRASE_KEY_LEN * 2);
+	hex_to_bytes(out.data(), paid_laid, PASSPHRASE_KEY_LEN * 2);
 	return out;
 }
 

rpcs3/Crypto/utils.h

@@ -6,7 +6,8 @@
 
 #include "util/types.hpp"
 
-#include <stdlib.h>
+#include <cstdlib>
+#include <string_view>
 
 enum { CRYPTO_MAX_PATH = 4096 };
 
@@ -15,8 +16,7 @@ char* extract_file_name(const char* file_path, char real_file_name[CRYPTO_MAX_PA
 std::string sha256_get_hash(const char* data, usz size, bool lower_case);
 
 // Hex string conversion auxiliary functions.
-u64 hex_to_u64(const char* hex_str);
+void hex_to_bytes(unsigned char* data, std::string_view hex_str, unsigned int str_length);
-void hex_to_bytes(unsigned char *data, const char *hex_str, unsigned int str_length);
 
 // Crypto functions (AES128-CBC, AES128-ECB, SHA1-HMAC and AES-CMAC).
 void aescbc128_decrypt(unsigned char *key, unsigned char *iv, unsigned char *in, unsigned char *out, usz len);

rpcs3/Emu/Audio/audio_utils.cpp

@@ -52,6 +52,6 @@ namespace audio
 		g_cfg.audio.volume.set(std::clamp<s32>(new_volume, g_cfg.audio.volume.min, g_cfg.audio.volume.max));
 		Emu.GetCallbacks().update_emu_settings();
 
-		rsx::overlays::queue_message(get_localized_string(localized_string_id::AUDIO_CHANGED, fmt::format("%d%%", g_cfg.audio.volume.get()).c_str()), 3'000'000);
+		rsx::overlays::queue_message(localized_string(localized_string_id::AUDIO_CHANGED, "%d%%", g_cfg.audio.volume.get()), 3'000'000, {}, rsx::overlays::message_pin_location::top_left, {}, true, true);
 	}
 } // namespace audio

rpcs3/Emu/CMakeLists.txt

@@ -402,6 +402,7 @@ target_sources(rpcs3_emu PRIVATE
     Io/GunCon3.cpp
     Io/Infinity.cpp
     Io/interception.cpp
+    Io/KamenRider.cpp
     Io/KeyboardHandler.cpp
     Io/midi_config_types.cpp
     Io/mouse_config.cpp
@@ -515,6 +516,7 @@ target_sources(rpcs3_emu PRIVATE
     RSX/Overlays/overlay_video.cpp
     RSX/Overlays/Shaders/shader_loading_dialog.cpp
     RSX/Overlays/Shaders/shader_loading_dialog_native.cpp
+    RSX/Program/Assembler/FPToCFG.cpp
     RSX/Program/CgBinaryProgram.cpp
     RSX/Program/CgBinaryFragmentProgram.cpp
     RSX/Program/CgBinaryVertexProgram.cpp
@@ -577,6 +579,7 @@ if(TARGET 3rdparty_vulkan)
         RSX/VK/vkutils/buffer_object.cpp
         RSX/VK/vkutils/chip_class.cpp
         RSX/VK/vkutils/commands.cpp
+        RSX/VK/vkutils/ex.cpp
         RSX/VK/vkutils/data_heap.cpp
         RSX/VK/vkutils/descriptors.cpp
         RSX/VK/vkutils/image.cpp
@@ -589,6 +592,7 @@ if(TARGET 3rdparty_vulkan)
         RSX/VK/vkutils/device.cpp
         RSX/VK/vkutils/sampler.cpp
         RSX/VK/vkutils/shared.cpp
+        RSX/VK/vkutils/unique_resource.cpp
         RSX/VK/VKAsyncScheduler.cpp
         RSX/VK/VKCommandStream.cpp
         RSX/VK/VKCommonDecompiler.cpp

rpcs3/Emu/CPU/CPUThread.cpp

@@ -190,7 +190,7 @@ struct cpu_prof
 			reservation_samples = 0;
 		}
 
-		static std::string format(const std::multimap<u64, u64, std::greater<u64>>& chart, u64 samples, u64 idle, bool extended_print = false)
+		static std::string format(const std::multimap<u64, u64, std::greater<u64>>& chart, u64 samples, u64 idle, u32 type_id, bool extended_print = false)
 		{
 			// Print results
 			std::string results;
@@ -204,11 +204,18 @@ struct cpu_prof
 				const f64 _frac = count / busy / samples;
 
 				// Print only 7 hash characters out of 11 (which covers roughly 48 bits)
+				if (type_id == 2)
+				{
 					fmt::append(results, "\n\t[%s", fmt::base57(be_t<u64>{name}));
 					results.resize(results.size() - 4);
 
 					// Print chunk address from lowest 16 bits
 					fmt::append(results, "...chunk-0x%05x]: %.4f%% (%u)", (name & 0xffff) * 4, _frac * 100., count);
+				}
+				else
+				{
+					fmt::append(results, "\n\t[0x%07x]: %.4f%% (%u)", name, _frac * 100., count);
+				}
 
 				if (results.size() >= (extended_print ? 10000 : 5000))
 				{
@@ -257,27 +264,37 @@ struct cpu_prof
 			}
 
 			// Print results
-			const std::string results = format(chart, samples, idle);
+			const std::string results = format(chart, samples, idle, ptr->id_type());
 			profiler.notice("Thread \"%s\" [0x%08x]: %u samples (%.4f%% idle), %u new, %u reservation (%.4f%%):\n%s", ptr->get_name(), ptr->id, samples, get_percent(idle, samples), new_samples, reservation_samples, get_percent(reservation_samples, samples - idle), results);
 
 			new_samples = 0;
 		}
 
-		static void print_all(std::unordered_map<shared_ptr<cpu_thread>, sample_info>& threads, sample_info& all_info)
+		static void print_all(std::unordered_map<shared_ptr<cpu_thread>, sample_info>& threads, sample_info& all_info, u32 type_id)
 		{
 			u64 new_samples = 0;
 
 			// Print all results and cleanup
 			for (auto& [ptr, info] : threads)
 			{
+				if (ptr->id_type() != type_id)
+				{
+					continue;
+				}
+
 				new_samples += info.new_samples;
 				info.print(ptr);
 			}
 
 			std::multimap<u64, u64, std::greater<u64>> chart;
 
-			for (auto& [_, info] : threads)
+			for (auto& [ptr, info] : threads)
 			{
+				if (ptr->id_type() != type_id)
+				{
+					continue;
+				}
+
 				// This function collects thread information regardless of 'new_samples' member state
 				for (auto& [name, count] : info.freq)
 				{
@@ -301,7 +318,7 @@ struct cpu_prof
 
 			if (new_samples < min_print_all_samples && thread_ctrl::state() != thread_state::aborting)
 			{
-				profiler.notice("All Threads: %u samples (%.4f%% idle), %u new, %u reservation (%.4f%%): Not enough new samples have been collected since the last print.", samples, get_percent(idle, samples), new_samples, reservation, get_percent(reservation, samples - idle));
+				profiler.notice("All %s Threads: %u samples (%.4f%% idle), %u new, %u reservation (%.4f%%): Not enough new samples have been collected since the last print.", type_id == 1 ? "PPU" : "SPU", samples, get_percent(idle, samples), new_samples, reservation, get_percent(reservation, samples - idle));
 				return;
 			}
 
@@ -310,12 +327,13 @@ struct cpu_prof
 				chart.emplace(count, name);
 			}
 
-			const std::string results = format(chart, samples, idle, true);
+			const std::string results = format(chart, samples, idle, type_id, true);
-			profiler.notice("All Threads: %u samples (%.4f%% idle), %u new, %u reservation (%.4f%%):%s", samples, get_percent(idle, samples), new_samples, reservation, get_percent(reservation, samples - idle), results);
+			profiler.notice("All %s Threads: %u samples (%.4f%% idle), %u new, %u reservation (%.4f%%):%s", type_id == 1 ? "PPU" : "SPU", samples, get_percent(idle, samples), new_samples, reservation, get_percent(reservation, samples - idle), results);
 		}
 	};
 
-	sample_info all_threads_info{};
+	sample_info all_spu_threads_info{};
+	sample_info all_ppu_threads_info{};
 
 	void operator()()
 	{
@@ -376,8 +394,11 @@ struct cpu_prof
 			{
 				if (auto state = +ptr->state; cpu_flag::exit - state)
 				{
+					const auto spu = ptr->try_get<spu_thread>();
+					const auto ppu = ptr->try_get<ppu_thread>();
+
 					// Get short function hash
-					const u64 name = atomic_storage<u64>::load(ptr->block_hash);
+					const u64 name = ppu ? atomic_storage<u32>::load(ppu->cia) : atomic_storage<u64>::load(ptr->block_hash);
 
 					// Append occurrence
 					info.samples++;
@@ -387,18 +408,18 @@ struct cpu_prof
 						info.freq[name]++;
 						info.new_samples++;
 
-						if (auto spu = ptr->try_get<spu_thread>())
+						if (spu)
 						{
 							if (spu->raddr)
 							{
 								info.reservation_samples++;
 							}
-						}
 
 							// Append verification time to fixed common name 0000000...chunk-0x3fffc
 							if (name >> 16 && (name & 0xffff) == 0)
 								info.freq[0xffff]++;
 						}
+					}
 					else
 					{
 						if (state & (cpu_flag::dbg_pause + cpu_flag::dbg_global_pause))
@@ -420,8 +441,10 @@ struct cpu_prof
 			{
 				profiler.success("Flushing profiling results...");
 
-				all_threads_info = {};
+				all_ppu_threads_info = {};
-				sample_info::print_all(threads, all_threads_info);
+				all_spu_threads_info = {};
+				sample_info::print_all(threads, all_ppu_threads_info, 1);
+				sample_info::print_all(threads, all_spu_threads_info, 2);
 			}
 
 			if (Emu.IsPaused())
@@ -442,7 +465,8 @@ struct cpu_prof
 		}
 
 		// Print all remaining results
-		sample_info::print_all(threads, all_threads_info);
+		sample_info::print_all(threads, all_ppu_threads_info, 1);
+		sample_info::print_all(threads, all_spu_threads_info, 2);
 	}
 
 	static constexpr auto thread_name = "CPU Profiler"sv;
@@ -459,7 +483,7 @@ extern f64 get_cpu_program_usage_percent(u64 hash)
 		{
 			u64 total = 0;
 
-			for (auto [name, count] : prof->all_threads_info.freq)
+			for (auto [name, count] : prof->all_spu_threads_info.freq)
 			{
 				if ((name & -65536) == hash)
 				{
@@ -472,7 +496,7 @@ extern f64 get_cpu_program_usage_percent(u64 hash)
 				return 0;
 			}
 
-			return std::max<f64>(0.0001, static_cast<f64>(total) * 100 / (prof->all_threads_info.samples - prof->all_threads_info.idle));
+			return std::max<f64>(0.0001, static_cast<f64>(total) * 100 / (prof->all_spu_threads_info.samples - prof->all_spu_threads_info.idle));
 		}
 	}
 
@@ -639,27 +663,22 @@ void cpu_thread::operator()()
 		thread_ctrl::set_thread_affinity_mask(thread_ctrl::get_affinity_mask(get_class()));
 	}
 
-	while (!g_fxo->is_init<cpu_profiler>())
+	ensure(g_fxo->is_init<cpu_profiler>());
-	{
-		if (Emu.IsStopped())
-		{
-			return;
-		}
-
-		// Can we have a little race, right? First thread is started concurrently with g_fxo->init()
-		thread_ctrl::wait_for(1000);
-	}
 
 	switch (get_class())
 	{
 	case thread_class::ppu:
 	{
-		//g_fxo->get<cpu_profiler>().registered.push(id);
+		if (g_cfg.core.ppu_prof)
+		{
+			g_fxo->get<cpu_profiler>().registered.push(id);
+		}
+
 		break;
 	}
 	case thread_class::spu:
 	{
-		if (g_cfg.core.spu_prof)
+		if (g_cfg.core.spu_prof || g_cfg.core.spu_debug)
 		{
 			g_fxo->get<cpu_profiler>().registered.push(id);
 		}
@@ -1546,7 +1565,7 @@ void cpu_thread::flush_profilers() noexcept
 		return;
 	}
 
-	if (g_cfg.core.spu_prof)
+	if (g_cfg.core.spu_prof || g_cfg.core.spu_debug || g_cfg.core.ppu_prof)
 	{
 		g_fxo->get<cpu_profiler>().registered.push(0);
 	}
@@ -1585,22 +1604,30 @@ extern bool try_lock_spu_threads_in_a_state_compatible_with_savestates(bool reve
 
 			idm::select<named_thread<spu_thread>>([&](u32 id, spu_thread& spu)
 			{
-				spu_list.emplace_back(ensure(idm::get_unlocked<named_thread<spu_thread>>(id)));
+				if (give_up)
+				{
+					return;
+				}
 
-				if (spu.current_func && spu.unsavable)
+				if (spu.current_func && spu.unsavable && !force_collect)
 				{
 					const u64 start = spu.start_time;
 
-					// Automatically give up if it is asleep 15 seconds or more
+					// Automatically give up if it is asleep 5 seconds or more
-					if (start && current > start && current - start >= 15'000'000)
+					if (start && current > start && current - start >= 5'000'000)
 					{
 						give_up = true;
+						return;
 					}
 				}
+
+				spu_list.emplace_back(ensure(idm::get_unlocked<named_thread<spu_thread>>(id)));
 			});
 
-			if (!force_collect && give_up)
+			if (give_up)
 			{
+				spu_list.clear();
+				old_counter = umax;
 				return decltype(&spu_list){};
 			}
 
@@ -1625,6 +1652,7 @@ extern bool try_lock_spu_threads_in_a_state_compatible_with_savestates(bool reve
 		}
 		else if (get_system_time() - start >= 150'000)
 		{
+			std::this_thread::sleep_for(1ms);
 			passed_count++;
 			start = 0;
 			continue;
@@ -1636,37 +1664,29 @@ extern bool try_lock_spu_threads_in_a_state_compatible_with_savestates(bool reve
 		if (!spu_list)
 		{
 			// Give up for now
-			std::this_thread::sleep_for(10ms);
+			std::this_thread::sleep_for(50ms);
 			passed_count++;
 			start = 0;
 			continue;
 		}
 
 		// Avoid using suspend_all when more than 2 threads known to be unsavable
-		u32 unsavable_threads = 0;
+		u32 savable_threads = 0;
 
 		for (auto& spu : *spu_list)
 		{
-			if (spu->unsavable)
+			if (!spu->unsavable)
 			{
-				unsavable_threads++;
+				savable_threads++;
-
-				if (unsavable_threads >= 3)
-				{
-					break;
-				}
 			}
 		}
 
-		if (unsavable_threads >= 3)
+		if (!savable_threads)
 		{
 			std::this_thread::yield();
 			continue;
 		}
 
-		// Flag for optimization
-		bool paused_anyone = false;
-
 		if (cpu_thread::suspend_all(nullptr, {}, [&]()
 		{
 			if (!get_spus(false, true))
@@ -1695,19 +1715,13 @@ extern bool try_lock_spu_threads_in_a_state_compatible_with_savestates(bool reve
 						break;
 					}
 				}
-				else
-				{
-					paused_anyone = true;
-					ensure(!spu->state.test_and_set(cpu_flag::dbg_global_pause));
-				}
 			}
 
-			if (failed && paused_anyone)
-			{
-				// For faster signalling, first remove state flags then batch notifications
 			for (auto& spu : *spu_list)
 			{
-					spu->state -= cpu_flag::dbg_global_pause;
+				if (!failed && !is_emu_paused)
+				{
+					ensure(!spu->state.test_and_set(cpu_flag::dbg_global_pause));
 				}
 			}
 
@@ -1719,13 +1733,6 @@ extern bool try_lock_spu_threads_in_a_state_compatible_with_savestates(bool reve
 				return false;
 			}
 
-			if (!paused_anyone)
-			{
-				// Need not do anything
-				std::this_thread::yield();
-				continue;
-			}
-
 			for (auto& spu : *spu_list)
 			{
 				if (spu->state & cpu_flag::wait)
@@ -1755,7 +1762,7 @@ extern bool try_lock_spu_threads_in_a_state_compatible_with_savestates(bool reve
 		{
 			spu->state.notify_one();
 		}
-	};
+	}
 
 	return false;
 }

rpcs3/Emu/CPU/CPUTranslator.h

@@ -4007,7 +4007,7 @@ llvm::CallInst* llvm_asm(
 	const std::string& constraints,
 	llvm::LLVMContext& context)
 {
-	llvm::ArrayRef<llvm::Type*> types_ref = std::nullopt;
+	llvm::ArrayRef<llvm::Type*> types_ref {};
 	std::vector<llvm::Type*> types;
 	types.reserve(args.size());
 

rpcs3/Emu/Cell/Modules/cellAtracXdec.cpp

@@ -111,7 +111,41 @@ void AtracXdecDecoder::alloc_avcodec()
 		fmt::throw_exception("avcodec_find_decoder() failed");
 	}
 
-	ensure(!(codec->capabilities & AV_CODEC_CAP_SUBFRAMES));
+	packet = av_packet_alloc();
+	if (!packet)
+	{
+		fmt::throw_exception("av_packet_alloc() failed");
+	}
+
+	frame = av_frame_alloc();
+	if (!frame)
+	{
+		fmt::throw_exception("av_frame_alloc() failed");
+	}
+}
+
+void AtracXdecDecoder::free_avcodec()
+{
+	if (packet)
+	{
+		av_packet_free(&packet);
+	}
+	if (frame)
+	{
+		av_frame_free(&frame);
+	}
+	if (ctx)
+	{
+		avcodec_free_context(&ctx);
+	}
+}
+
+void AtracXdecDecoder::init_avcodec()
+{
+	if (ctx)
+	{
+		avcodec_free_context(&ctx);
+	}
 
 	ctx = avcodec_alloc_context3(codec);
 	if (!ctx)
@@ -133,34 +167,6 @@ void AtracXdecDecoder::alloc_avcodec()
 		frame->buf[0] = av_buffer_create(frame->data[0], ATXDEC_SAMPLES_PER_FRAME * sizeof(f32) * frame->ch_layout.nb_channels, [](void*, uint8_t*){}, nullptr, 0);
 		return 0;
 	};
-
-	packet = av_packet_alloc();
-	if (!packet)
-	{
-		fmt::throw_exception("av_packet_alloc() failed");
-	}
-
-	frame = av_frame_alloc();
-	if (!frame)
-	{
-		fmt::throw_exception("av_frame_alloc() failed");
-	}
-}
-
-void AtracXdecDecoder::free_avcodec()
-{
-	av_packet_free(&packet);
-	av_frame_free(&frame);
-	avcodec_free_context(&ctx);
-}
-
-void AtracXdecDecoder::init_avcodec()
-{
-	if (int err = avcodec_close(ctx); err)
-	{
-		fmt::throw_exception("avcodec_close() failed (err=0x%x='%s')", err, utils::av_error_to_string(err));
-	}
-
 	ctx->block_align = nbytes;
 	ctx->ch_layout.nb_channels = nch_in;
 	ctx->sample_rate = sampling_freq;

rpcs3/Emu/Cell/Modules/cellAtracXdec.h

@@ -192,10 +192,10 @@ struct AtracXdecDecoder
 
 	// HLE exclusive
 	b8 config_is_set = false; // For savestates
-	const AVCodec* codec;
+	const AVCodec* codec = nullptr;
-	AVCodecContext* ctx;
+	AVCodecContext* ctx = nullptr;
-	AVPacket* packet;
+	AVPacket* packet = nullptr;
-	AVFrame* frame;
+	AVFrame* frame = nullptr;
 
 	u8 spurs_stuff[84]; // 120 bytes on LLE, pointers to CellSpurs, CellSpursTaskset, etc.
 

rpcs3/Emu/Cell/Modules/cellCamera.cpp

@@ -1139,8 +1139,10 @@ error_code cellCameraGetBufferInfo(s32 dev_num, vm::ptr<CellCameraInfo> info)
 	return CELL_OK;
 }
 
-error_code cellCameraGetBufferInfoEx(s32 dev_num, vm::ptr<CellCameraInfoEx> info)
+error_code cellCameraGetBufferInfoEx(ppu_thread& ppu, s32 dev_num, vm::ptr<CellCameraInfoEx> info)
 {
+	ppu.state += cpu_flag::wait;
+
 	cellCamera.notice("cellCameraGetBufferInfoEx(dev_num=%d, info=0x%x)", dev_num, info);
 
 	// calls cellCameraGetBufferInfo
@@ -1151,10 +1153,16 @@ error_code cellCameraGetBufferInfoEx(s32 dev_num, vm::ptr<CellCameraInfoEx> info
 	}
 
 	auto& g_camera = g_fxo->get<camera_thread>();
+
+	CellCameraInfoEx info_out;
+
+	{
 		std::lock_guard lock(g_camera.mutex);
 
-	*info = g_camera.info;
+		info_out = g_camera.info;
+	}
 
+	*info = info_out;
 	return CELL_OK;
 }
 

rpcs3/Emu/Cell/Modules/cellGame.cpp

@@ -464,6 +464,8 @@ error_code cellHddGameCheck(ppu_thread& ppu, u32 version, vm::cptr<char> dirName
 	{
 		get->isNewData = CELL_HDDGAME_ISNEWDATA_NODIR;
 		get->getParam = {};
+
+		cellGame.warning("cellHddGameCheck(): New data.");
 	}
 	else
 	{
@@ -476,13 +478,22 @@ error_code cellHddGameCheck(ppu_thread& ppu, u32 version, vm::cptr<char> dirName
 		if (psf.contains("RESOLUTION")) get->getParam.resolution = ::at32(psf, "RESOLUTION").as_integer();
 		if (psf.contains("SOUND_FORMAT")) get->getParam.soundFormat = ::at32(psf, "SOUND_FORMAT").as_integer();
 		if (psf.contains("TITLE")) strcpy_trunc(get->getParam.title, ::at32(psf, "TITLE").as_string());
-		if (psf.contains("APP_VER")) strcpy_trunc(get->getParam.dataVersion, ::at32(psf, "APP_VER").as_string());
+
-		if (psf.contains("TITLE_ID")) strcpy_trunc(get->getParam.titleId, ::at32(psf, "TITLE_ID").as_string());
+		// Old games do not have APP_VER key
+		strcpy_trunc(get->getParam.dataVersion, psf::get_string(psf, "APP_VER", psf::get_string(sfo, "VERSION", "")));
+
+		if (psf.contains("TITLE_ID"))
+		{
+			strcpy_trunc(get->getParam.titleId, ::at32(psf, "TITLE_ID").as_string());
+		}
 
 		for (u32 i = 0; i < CELL_HDDGAME_SYSP_LANGUAGE_NUM; i++)
 		{
 			strcpy_trunc(get->getParam.titleLang[i], psf::get_string(psf, fmt::format("TITLE_%02d", i)));
 		}
+
+		cellGame.warning("cellHddGameCheck(): Data exists:\nATTRIBUTE: 0x%x, RESOLUTION: 0x%x, RESOLUTION: 0x%x, SOUND_FORMAT: 0x%x, dataVersion: %s"
+			, get->getParam.attribute, get->getParam.resolution, get->getParam.soundFormat, get->getParam.soundFormat, std::span<const u8>(reinterpret_cast<const u8*>(get->getParam.dataVersion), 6));
 	}
 
 	// TODO ?
@@ -550,7 +561,7 @@ error_code cellHddGameCheck(ppu_thread& ppu, u32 version, vm::cptr<char> dirName
 
 	case CELL_HDDGAME_CBRESULT_ERR_NOSPACE:
 		cellGame.error("cellHddGameCheck(): callback returned CELL_HDDGAME_CBRESULT_ERR_NOSPACE. Space Needed: %d KB", result->errNeedSizeKB);
-		error_msg = get_localized_string(localized_string_id::CELL_HDD_GAME_CHECK_NOSPACE, fmt::format("%d", result->errNeedSizeKB).c_str());
+		error_msg = get_localized_string(localized_string_id::CELL_HDD_GAME_CHECK_NOSPACE, "%d", result->errNeedSizeKB);
 		break;
 
 	case CELL_HDDGAME_CBRESULT_ERR_BROKEN:
@@ -565,12 +576,12 @@ error_code cellHddGameCheck(ppu_thread& ppu, u32 version, vm::cptr<char> dirName
 
 	case CELL_HDDGAME_CBRESULT_ERR_INVALID:
 		cellGame.error("cellHddGameCheck(): callback returned CELL_HDDGAME_CBRESULT_ERR_INVALID. Error message: %s", result->invalidMsg);
-		error_msg = get_localized_string(localized_string_id::CELL_HDD_GAME_CHECK_INVALID, fmt::format("%s", result->invalidMsg).c_str());
+		error_msg = get_localized_string(localized_string_id::CELL_HDD_GAME_CHECK_INVALID, "%s", result->invalidMsg);
 		break;
 
 	default:
 		cellGame.error("cellHddGameCheck(): callback returned unknown error (code=0x%x). Error message: %s", result->invalidMsg);
-		error_msg = get_localized_string(localized_string_id::CELL_HDD_GAME_CHECK_INVALID, fmt::format("%s", result->invalidMsg).c_str());
+		error_msg = get_localized_string(localized_string_id::CELL_HDD_GAME_CHECK_INVALID, "%s", result->invalidMsg);
 		break;
 	}
 
@@ -1169,7 +1180,7 @@ error_code cellGameDataCheckCreate2(ppu_thread& ppu, u32 version, vm::cptr<char>
 	}
 	case CELL_GAMEDATA_CBRESULT_ERR_NOSPACE:
 		cellGame.error("cellGameDataCheckCreate2(): callback returned CELL_GAMEDATA_CBRESULT_ERR_NOSPACE. Space Needed: %d KB", cbResult->errNeedSizeKB);
-		error_msg = get_localized_string(localized_string_id::CELL_GAMEDATA_CHECK_NOSPACE, fmt::format("%d", cbResult->errNeedSizeKB).c_str());
+		error_msg = get_localized_string(localized_string_id::CELL_GAMEDATA_CHECK_NOSPACE, "%d", cbResult->errNeedSizeKB);
 		break;
 
 	case CELL_GAMEDATA_CBRESULT_ERR_BROKEN:
@@ -1184,12 +1195,12 @@ error_code cellGameDataCheckCreate2(ppu_thread& ppu, u32 version, vm::cptr<char>
 
 	case CELL_GAMEDATA_CBRESULT_ERR_INVALID:
 		cellGame.error("cellGameDataCheckCreate2(): callback returned CELL_GAMEDATA_CBRESULT_ERR_INVALID. Error message: %s", cbResult->invalidMsg);
-		error_msg = get_localized_string(localized_string_id::CELL_GAMEDATA_CHECK_INVALID, fmt::format("%s", cbResult->invalidMsg).c_str());
+		error_msg = get_localized_string(localized_string_id::CELL_GAMEDATA_CHECK_INVALID, "%s", cbResult->invalidMsg);
 		break;
 
 	default:
 		cellGame.error("cellGameDataCheckCreate2(): callback returned unknown error (code=0x%x). Error message: %s", cbResult->invalidMsg);
-		error_msg = get_localized_string(localized_string_id::CELL_GAMEDATA_CHECK_INVALID, fmt::format("%s", cbResult->invalidMsg).c_str());
+		error_msg = get_localized_string(localized_string_id::CELL_GAMEDATA_CHECK_INVALID, "%s", cbResult->invalidMsg);
 		break;
 	}
 
@@ -1686,7 +1697,7 @@ error_code cellGameContentErrorDialog(s32 type, s32 errNeedSizeKB, vm::cptr<char
 		break;
 	case CELL_GAME_ERRDIALOG_NOSPACE:
 		// Not enough available space. The application will continue.
-		error_msg = get_localized_string(localized_string_id::CELL_GAME_ERROR_NOSPACE, fmt::format("%d", errNeedSizeKB).c_str());
+		error_msg = get_localized_string(localized_string_id::CELL_GAME_ERROR_NOSPACE, "%d", errNeedSizeKB);
 		break;
 	case CELL_GAME_ERRDIALOG_BROKEN_EXIT_GAMEDATA:
 		// Game data is corrupted. The application will be terminated.
@@ -1698,7 +1709,7 @@ error_code cellGameContentErrorDialog(s32 type, s32 errNeedSizeKB, vm::cptr<char
 		break;
 	case CELL_GAME_ERRDIALOG_NOSPACE_EXIT:
 		// Not enough available space. The application will be terminated.
-		error_msg = get_localized_string(localized_string_id::CELL_GAME_ERROR_NOSPACE_EXIT, fmt::format("%d", errNeedSizeKB).c_str());
+		error_msg = get_localized_string(localized_string_id::CELL_GAME_ERROR_NOSPACE_EXIT, "%d", errNeedSizeKB);
 		break;
 	default:
 		return CELL_GAME_ERROR_PARAM;
@@ -1712,7 +1723,7 @@ error_code cellGameContentErrorDialog(s32 type, s32 errNeedSizeKB, vm::cptr<char
 		}
 
 		error_msg += '\n';
-		error_msg += get_localized_string(localized_string_id::CELL_GAME_ERROR_DIR_NAME, fmt::format("%s", dirName).c_str());
+		error_msg += get_localized_string(localized_string_id::CELL_GAME_ERROR_DIR_NAME, "%s", dirName);
 	}
 
 	return open_exit_dialog(error_msg, type > CELL_GAME_ERRDIALOG_NOSPACE, msg_dialog_source::_cellGame);

rpcs3/Emu/Cell/Modules/cellGame.h

@@ -224,8 +224,8 @@ enum
 struct CellGameDataSystemFileParam
 {
 	char title[CELL_GAMEDATA_SYSP_TITLE_SIZE];
-	char titleLang[CELL_GAMEDATA_SYSP_LANGUAGE_NUM][CELL_GAMEDATA_SYSP_TITLE_SIZE];
+	char titleLang[CELL_GAMEDATA_SYSP_LANGUAGE_NUM][CELL_GAMEDATA_SYSP_TITLE_SIZE]; // 0x80
-	char titleId[CELL_GAMEDATA_SYSP_TITLEID_SIZE];
+	char titleId[CELL_GAMEDATA_SYSP_TITLEID_SIZE]; // 0xA80
 	char reserved0[2];
 	char dataVersion[CELL_GAMEDATA_SYSP_VERSION_SIZE];
 	char reserved1[2];
@@ -248,13 +248,13 @@ struct CellGameDataStatGet
 {
 	be_t<s32> hddFreeSizeKB;
 	be_t<u32> isNewData;
-	char contentInfoPath[CELL_GAMEDATA_PATH_MAX];
+	char contentInfoPath[CELL_GAMEDATA_PATH_MAX]; // 0x8
-	char gameDataPath[CELL_GAMEDATA_PATH_MAX];
+	char gameDataPath[CELL_GAMEDATA_PATH_MAX]; // 0x427 
-	char reserved0[2];
+	char reserved0[2]; // 0x846
-	be_t<s64> st_atime_;
+	be_t<s64> st_atime_; // 0x848
-	be_t<s64> st_mtime_;
+	be_t<s64> st_mtime_; // 0x850
-	be_t<s64> st_ctime_;
+	be_t<s64> st_ctime_; // 0x858
-	CellGameDataSystemFileParam getParam;
+	CellGameDataSystemFileParam getParam; // 0x860
 	be_t<s32> sizeKB;
 	be_t<s32> sysSizeKB;
 	char reserved1[68];

rpcs3/Emu/Cell/Modules/cellGem.cpp

@@ -361,7 +361,7 @@ public:
 		case move_handler::mouse:
 		case move_handler::raw_mouse:
 		{
-			auto& handler = g_fxo->get<MouseHandlerBase>();
+			auto& handler = *ensure(g_fxo->try_get<MouseHandlerBase>());
 			std::lock_guard mouse_lock(handler.mutex);
 			const MouseInfo& info = handler.GetInfo();
 
@@ -374,7 +374,7 @@ public:
 #ifdef HAVE_LIBEVDEV
 		case move_handler::gun:
 		{
-			gun_thread& gun = g_fxo->get<gun_thread>();
+			gun_thread& gun = *ensure(g_fxo->try_get<gun_thread>());
 			std::scoped_lock lock(gun.handler.mutex);
 			gun.num_devices = gun.handler.init() ? gun.handler.get_num_guns() : 0;
 
@@ -505,7 +505,7 @@ public:
 		case move_handler::mouse:
 		case move_handler::raw_mouse:
 		{
-			auto& handler = g_fxo->get<MouseHandlerBase>();
+			auto& handler = *ensure(g_fxo->try_get<MouseHandlerBase>());
 			std::lock_guard mouse_lock(handler.mutex);
 
 			// Make sure that the mouse handler is initialized
@@ -522,7 +522,7 @@ public:
 #ifdef HAVE_LIBEVDEV
 		case move_handler::gun:
 		{
-			gun_thread& gun = g_fxo->get<gun_thread>();
+			gun_thread& gun = *ensure(g_fxo->try_get<gun_thread>());
 			std::scoped_lock lock(gun.handler.mutex);
 			gun.num_devices = gun.handler.init() ? gun.handler.get_num_guns() : 0;
 			connected_controllers = std::min<u32>(std::min<u32>(attribute.max_connect, CELL_GEM_MAX_NUM), gun.num_devices);
@@ -2284,6 +2284,8 @@ error_code cellGemClearStatusFlags(u32 gem_num, u64 mask)
 
 error_code cellGemConvertVideoFinish(ppu_thread& ppu)
 {
+	ppu.state += cpu_flag::wait;
+
 	cellGem.warning("cellGemConvertVideoFinish()");
 
 	auto& gem = g_fxo->get<gem_config>();
@@ -2306,8 +2308,10 @@ error_code cellGemConvertVideoFinish(ppu_thread& ppu)
 	return CELL_OK;
 }
 
-error_code cellGemConvertVideoStart(vm::cptr<void> video_frame)
+error_code cellGemConvertVideoStart(ppu_thread& ppu, vm::cptr<void> video_frame)
 {
+	ppu.state += cpu_flag::wait;
+
 	cellGem.warning("cellGemConvertVideoStart(video_frame=*0x%x)", video_frame);
 
 	auto& gem = g_fxo->get<gem_config>();
@@ -2459,6 +2463,8 @@ error_code cellGemEnableMagnetometer2(u32 gem_num, u32 enable)
 
 error_code cellGemEnd(ppu_thread& ppu)
 {
+	ppu.state += cpu_flag::wait;
+
 	cellGem.warning("cellGemEnd()");
 
 	auto& gem = g_fxo->get<gem_config>();
@@ -3263,7 +3269,7 @@ error_code cellGemPrepareCamera(s32 max_exposure, f32 image_quality)
 
 	extern error_code cellCameraGetAttribute(s32 dev_num, s32 attrib, vm::ptr<u32> arg1, vm::ptr<u32> arg2);
 	extern error_code cellCameraSetAttribute(s32 dev_num, s32 attrib, u32 arg1, u32 arg2);
-	extern error_code cellCameraGetBufferInfoEx(s32 dev_num, vm::ptr<CellCameraInfoEx> info);
+	extern error_code cellCameraGetBufferInfoEx(ppu_thread&, s32 dev_num, vm::ptr<CellCameraInfoEx> info);
 
 	vm::var<CellCameraInfoEx> info = vm::make_var<CellCameraInfoEx>({});
 	vm::var<u32> arg1 = vm::make_var<u32>({});
@@ -3271,7 +3277,7 @@ error_code cellGemPrepareCamera(s32 max_exposure, f32 image_quality)
 
 	cellCameraGetAttribute(0, 0x3e6, arg1, arg2);
 	cellCameraSetAttribute(0, 0x3e6, 0x3e, *arg2 | 0x80);
-	cellCameraGetBufferInfoEx(0, info);
+	cellCameraGetBufferInfoEx(*cpu_thread::get_current<ppu_thread>(), 0, info);
 
 	if (info->width == 640)
 	{
@@ -3603,6 +3609,8 @@ error_code cellGemTrackHues(vm::cptr<u32> req_hues, vm::ptr<u32> res_hues)
 
 error_code cellGemUpdateFinish(ppu_thread& ppu)
 {
+	ppu.state += cpu_flag::wait;
+
 	cellGem.warning("cellGemUpdateFinish()");
 
 	auto& gem = g_fxo->get<gem_config>();

rpcs3/Emu/Cell/Modules/cellL10n.cpp

@@ -2498,7 +2498,7 @@ s32 UCS2stoSBCSs(vm::cptr<u16> src, vm::ptr<u32> src_len, vm::ptr<u8> dst, vm::p
 
 	for (u32 src_pos = 0; src_pos < *src_len; src_pos++)
 	{
-		const s16 ucs2 = src[src_pos];
+		const u16 ucs2 = src[src_pos];
 
 		if (ucs2 >= 0xfffe)
 		{

rpcs3/Emu/Cell/Modules/cellMsgDialog.cpp

@@ -506,7 +506,7 @@ error_code cellMsgDialogOpenErrorCode(u32 errorCode, vm::ptr<CellMsgDialogCallba
 	default:         string_id = localized_string_id::CELL_MSG_DIALOG_ERROR_DEFAULT;  break; // An error has occurred.
 	}
 
-	const std::string error = get_localized_string(string_id, fmt::format("%08x", errorCode).c_str());
+	const std::string error = get_localized_string(string_id, "%08x", errorCode);
 
 	return cellMsgDialogOpen2(CELL_MSGDIALOG_TYPE_SE_TYPE_ERROR | CELL_MSGDIALOG_TYPE_BUTTON_TYPE_OK, vm::make_str(error), callback, userData, extParam);
 }

rpcs3/Emu/Cell/Modules/cellPad.cpp

@@ -575,12 +575,12 @@ void pad_get_data(u32 port_no, CellPadData* data, bool get_periph_data = false)
 
 		if (setting & CELL_PAD_SETTING_PRESS_ON)
 		{
-			constexpr u32 copy_size = (CELL_PAD_LEN_CHANGE_PRESS_ON - CELL_PAD_BTN_OFFSET_DIGITAL1) * sizeof(u16);
+			constexpr u32 copy_size = (static_cast<u32>(CELL_PAD_LEN_CHANGE_PRESS_ON) - static_cast<u32>(CELL_PAD_BTN_OFFSET_DIGITAL1)) * sizeof(u16);
 			std::memcpy(&data->button[CELL_PAD_BTN_OFFSET_DIGITAL1], &output.button[CELL_PAD_BTN_OFFSET_DIGITAL1], copy_size);
 		}
 		else
 		{
-			constexpr u32 copy_size = (CELL_PAD_LEN_CHANGE_DEFAULT - CELL_PAD_BTN_OFFSET_DIGITAL1) * sizeof(u16);
+			constexpr u32 copy_size = (static_cast<u32>(CELL_PAD_LEN_CHANGE_DEFAULT) - static_cast<u32>(CELL_PAD_BTN_OFFSET_DIGITAL1)) * sizeof(u16);
 			std::memcpy(&data->button[CELL_PAD_BTN_OFFSET_DIGITAL1], &output.button[CELL_PAD_BTN_OFFSET_DIGITAL1], copy_size);
 
 			// Clear area if setting is not used
@@ -590,7 +590,7 @@ void pad_get_data(u32 port_no, CellPadData* data, bool get_periph_data = false)
 
 		if (data->len == CELL_PAD_LEN_CHANGE_SENSOR_ON)
 		{
-			constexpr u32 copy_size = (CELL_PAD_LEN_CHANGE_SENSOR_ON - CELL_PAD_BTN_OFFSET_SENSOR_X) * sizeof(u16);
+			constexpr u32 copy_size = (static_cast<u32>(CELL_PAD_LEN_CHANGE_SENSOR_ON) - static_cast<u32>(CELL_PAD_BTN_OFFSET_SENSOR_X)) * sizeof(u16);
 			std::memcpy(&data->button[CELL_PAD_BTN_OFFSET_SENSOR_X], &output.button[CELL_PAD_BTN_OFFSET_SENSOR_X], copy_size);
 		}
 	}
@@ -1051,7 +1051,15 @@ error_code cellPadSetPortSetting(u32 port_no, u32 port_setting)
 	if (port_no >= CELL_PAD_MAX_PORT_NUM)
 		return CELL_OK;
 
-	config.port_setting[port_no] = port_setting;
+	if (port_setting & CELL_PAD_SETTING_PRESS_ON)
+		config.port_setting[port_no] |= CELL_PAD_SETTING_PRESS_ON;
+	else
+		config.port_setting[port_no] &= ~CELL_PAD_SETTING_PRESS_ON;
+
+	if (port_setting & CELL_PAD_SETTING_SENSOR_ON)
+		config.port_setting[port_no] |= CELL_PAD_SETTING_SENSOR_ON;
+	else
+		config.port_setting[port_no] &= ~CELL_PAD_SETTING_SENSOR_ON;
 
 	// can also return CELL_PAD_ERROR_UNSUPPORTED_GAMEPAD <- Update: seems to be just internal and ignored
 
@@ -1123,7 +1131,7 @@ error_code cellPadSetPressMode(u32 port_no, u32 mode)
 	if (!config.max_connect)
 		return CELL_PAD_ERROR_UNINITIALIZED;
 
-	if (port_no >= CELL_PAD_MAX_PORT_NUM)
+	if (port_no >= CELL_MAX_PADS || mode > 1)
 		return CELL_PAD_ERROR_INVALID_PARAMETER;
 
 	// CELL_PAD_ERROR_NO_DEVICE is not returned in this case.
@@ -1157,7 +1165,7 @@ error_code cellPadSetSensorMode(u32 port_no, u32 mode)
 	if (!config.max_connect)
 		return CELL_PAD_ERROR_UNINITIALIZED;
 
-	if (port_no >= CELL_MAX_PADS)
+	if (port_no >= CELL_MAX_PADS || mode > 1)
 		return CELL_PAD_ERROR_INVALID_PARAMETER;
 
 	// CELL_PAD_ERROR_NO_DEVICE is not returned in this case.

rpcs3/Emu/Cell/Modules/cellSaveData.cpp

@@ -396,7 +396,7 @@ static error_code display_callback_result_error_message(ppu_thread& ppu, const C
 	switch (result.result)
 	{
 	case CELL_SAVEDATA_CBRESULT_ERR_NOSPACE:
-		msg = get_localized_string(localized_string_id::CELL_SAVEDATA_CB_NO_SPACE, fmt::format("%d", result.errNeedSizeKB).c_str());
+		msg = get_localized_string(localized_string_id::CELL_SAVEDATA_CB_NO_SPACE, "%d", result.errNeedSizeKB);
 		break;
 	case CELL_SAVEDATA_CBRESULT_ERR_FAILURE:
 		msg = get_localized_string(localized_string_id::CELL_SAVEDATA_CB_FAILURE);

rpcs3/Emu/Cell/Modules/cellSpurs.cpp

@@ -1230,7 +1230,7 @@ s32 _spurs::initialize(ppu_thread& ppu, vm::ptr<CellSpurs> spurs, u32 revision,
 	if (flags & SAF_UNKNOWN_FLAG_9)          spuTgAttr->type |= 0x0800;
 	if (flags & SAF_SYSTEM_WORKLOAD_ENABLED) spuTgAttr->type |= SYS_SPU_THREAD_GROUP_TYPE_COOPERATE_WITH_SYSTEM;
 
-	if (s32 rc = sys_spu_thread_group_create(ppu, spurs.ptr(&CellSpurs::spuTG), nSpus, spuPriority, spuTgAttr))
+	if (s32 rc = sys_spu_thread_group_create(ppu, spurs.ptr(&CellSpurs::spuTG), nSpus, spuPriority, vm::unsafe_ptr_cast<reduced_sys_spu_thread_group_attribute>(spuTgAttr)))
 	{
 		ppu_execute<&sys_spu_image_close>(ppu, spurs.ptr(&CellSpurs::spuImg));
 		return rollback(), rc;

rpcs3/Emu/Cell/Modules/sceNp.cpp

@@ -3198,7 +3198,7 @@ error_code sceNpLookupTerm()
 
 error_code sceNpLookupCreateTitleCtx(vm::cptr<SceNpCommunicationId> communicationId, vm::cptr<SceNpId> selfNpId)
 {
-	sceNp.warning("sceNpLookupCreateTitleCtx(communicationId=*0x%x(%s), selfNpId=0x%x)", communicationId, communicationId ? communicationId->data : "", selfNpId);
+	sceNp.warning("sceNpLookupCreateTitleCtx(communicationId=*0x%x(%s), selfNpId=0x%x)", communicationId, communicationId ? std::string_view(communicationId->data, 9) : "", selfNpId);
 
 	auto& nph = g_fxo->get<named_thread<np::np_handler>>();
 
@@ -3930,8 +3930,8 @@ error_code sceNpManagerGetMyLanguages(vm::ptr<SceNpMyLanguages> myLanguages)
 		return SCE_NP_ERROR_INVALID_STATE;
 	}
 
-	myLanguages->language1 = SCE_NP_LANG_ENGLISH_US;
+	myLanguages->language1 = g_cfg.sys.language;
-	myLanguages->language2 = g_cfg.sys.language != CELL_SYSUTIL_LANG_ENGLISH_US ? g_cfg.sys.language : -1;
+	myLanguages->language2 = g_cfg.sys.language != CELL_SYSUTIL_LANG_ENGLISH_US ? CELL_SYSUTIL_LANG_ENGLISH_US : -1;
 	myLanguages->language3 = -1;
 
 	return CELL_OK;
@@ -3968,7 +3968,7 @@ error_code sceNpManagerGetAccountRegion(vm::ptr<SceNpCountryCode> countryCode, v
 	ensure(ccode.size() == sizeof(SceNpCountryCode::data));
 	std::memcpy(countryCode->data, ccode.data(), sizeof(SceNpCountryCode::data));
 
-	*language = CELL_SYSUTIL_LANG_ENGLISH_US;
+	*language = g_cfg.sys.language;
 
 	return CELL_OK;
 }
@@ -6962,7 +6962,7 @@ error_code sceNpSignalingGetConnectionFromPeerAddress(u32 ctx_id, np_in_addr_t p
 	return CELL_OK;
 }
 
-error_code sceNpSignalingGetLocalNetInfo(u32 ctx_id, vm::ptr<SceNpSignalingNetInfo> info)
+error_code sceNpSignalingGetLocalNetInfo(u32 ctx_id, vm::ptr<SceNpSignalingNetInfoDeprecated> info)
 {
 	sceNp.warning("sceNpSignalingGetLocalNetInfo(ctx_id=%d, info=*0x%x)", ctx_id, info);
 
@@ -6973,7 +6973,8 @@ error_code sceNpSignalingGetLocalNetInfo(u32 ctx_id, vm::ptr<SceNpSignalingNetIn
 		return SCE_NP_SIGNALING_ERROR_NOT_INITIALIZED;
 	}
 
-	if (!info || info->size != sizeof(SceNpSignalingNetInfo))
+	// Library has backward support for a version of SceNpSignalingNetInfo without npport
+	if (!info || (info->size != sizeof(SceNpSignalingNetInfo) && info->size != sizeof(SceNpSignalingNetInfoDeprecated)))
 	{
 		return SCE_NP_SIGNALING_ERROR_INVALID_ARGUMENT;
 	}
@@ -6985,7 +6986,12 @@ error_code sceNpSignalingGetLocalNetInfo(u32 ctx_id, vm::ptr<SceNpSignalingNetIn
 	info->nat_status    = SCE_NP_SIGNALING_NETINFO_NAT_STATUS_TYPE2;
 	info->upnp_status   = nph.get_upnp_status();
 	info->npport_status = SCE_NP_SIGNALING_NETINFO_NPPORT_STATUS_OPEN;
-	info->npport        = SCE_NP_PORT;
+
+	if (info->size == sizeof(SceNpSignalingNetInfo))
+	{
+		auto new_info = vm::unsafe_ptr_cast<SceNpSignalingNetInfo>(info);
+		new_info->npport = SCE_NP_PORT;
+	}
 
 	return CELL_OK;
 }

rpcs3/Emu/Cell/Modules/sceNp.h

@@ -1584,6 +1584,16 @@ struct SceNpSignalingNetInfo
 	be_t<u16> npport;
 };
 
+struct SceNpSignalingNetInfoDeprecated
+{
+	be_t<u32> size;
+	be_t<u32> local_addr; // in_addr
+	be_t<u32> mapped_addr; // in_addr
+	be_t<s32> nat_status;
+	be_t<s32> upnp_status;
+	be_t<s32> npport_status;
+};
+
 struct SceNpCustomMenuAction
 {
 	be_t<u32> options;

rpcs3/Emu/Cell/Modules/sceNp2.cpp

@@ -1300,7 +1300,7 @@ error_code sceNpMatching2GrantRoomOwner(
 error_code sceNpMatching2CreateContext(
     vm::cptr<SceNpId> npId, vm::cptr<SceNpCommunicationId> commId, vm::cptr<SceNpCommunicationPassphrase> passPhrase, vm::ptr<SceNpMatching2ContextId> ctxId, s32 option)
 {
-	sceNp2.warning("sceNpMatching2CreateContext(npId=*0x%x, commId=*0x%x(%s), passPhrase=*0x%x, ctxId=*0x%x, option=%d)", npId, commId, commId ? commId->data : "", passPhrase, ctxId, option);
+	sceNp2.warning("sceNpMatching2CreateContext(npId=*0x%x, commId=*0x%x(%s), passPhrase=*0x%x, ctxId=*0x%x, option=%d)", npId, commId, commId ? std::string_view(commId->data, 9) : "", passPhrase, ctxId, option);
 
 	auto& nph = g_fxo->get<named_thread<np::np_handler>>();
 

rpcs3/Emu/Cell/Modules/sceNpTrophy.h

@@ -3,7 +3,6 @@
 #include "util/types.hpp"
 #include "Emu/Memory/vm_ptr.h"
 #include "Emu/Cell/ErrorCodes.h"
-#include <mutex>
 #include <vector>
 #include <mutex>
 

rpcs3/Emu/Cell/PPUDisAsm.cpp

@@ -221,7 +221,7 @@ std::pair<PPUDisAsm::const_op, u64> PPUDisAsm::try_get_const_op_gpr_value(u32 re
 
 			GET_CONST_REG(reg_rs, op.rs);
 
-			return { form, utils::rol64(reg_rs, op.sh64) & (~0ull << (op.mbe64 ^ 63)) };
+			return {form, std::rotl<u64>(reg_rs, op.sh64) & (~0ull << (op.mbe64 ^ 63))};
 		}
 		case ppu_itype::OR:
 		{

rpcs3/Emu/Cell/PPUInterpreter.cpp

@@ -3483,7 +3483,7 @@ auto RLWIMI()
 
 	static const auto exec = [](ppu_thread& ppu, ppu_opcode_t op) {
 	const u64 mask = ppu_rotate_mask(32 + op.mb32, 32 + op.me32);
-	ppu.gpr[op.ra] = (ppu.gpr[op.ra] & ~mask) | (dup32(utils::rol32(static_cast<u32>(ppu.gpr[op.rs]), op.sh32)) & mask);
+	ppu.gpr[op.ra] = (ppu.gpr[op.ra] & ~mask) | (dup32(std::rotl<u32>(static_cast<u32>(ppu.gpr[op.rs]), op.sh32)) & mask);
 	if constexpr (((Flags == has_rc) || ...))
 		ppu_cr_set<s64>(ppu, 0, ppu.gpr[op.ra], 0);
 	};
@@ -3497,7 +3497,7 @@ auto RLWINM()
 		return ppu_exec_select<Flags...>::template select<>();
 
 	static const auto exec = [](ppu_thread& ppu, ppu_opcode_t op) {
-	ppu.gpr[op.ra] = dup32(utils::rol32(static_cast<u32>(ppu.gpr[op.rs]), op.sh32)) & ppu_rotate_mask(32 + op.mb32, 32 + op.me32);
+	ppu.gpr[op.ra] = dup32(std::rotl<u32>(static_cast<u32>(ppu.gpr[op.rs]), op.sh32)) & ppu_rotate_mask(32 + op.mb32, 32 + op.me32);
 	if constexpr (((Flags == has_rc) || ...))
 		ppu_cr_set<s64>(ppu, 0, ppu.gpr[op.ra], 0);
 	};
@@ -3511,7 +3511,7 @@ auto RLWNM()
 		return ppu_exec_select<Flags...>::template select<>();
 
 	static const auto exec = [](ppu_thread& ppu, ppu_opcode_t op) {
-	ppu.gpr[op.ra] = dup32(utils::rol32(static_cast<u32>(ppu.gpr[op.rs]), ppu.gpr[op.rb] & 0x1f)) & ppu_rotate_mask(32 + op.mb32, 32 + op.me32);
+	ppu.gpr[op.ra] = dup32(std::rotl<u32>(static_cast<u32>(ppu.gpr[op.rs]), ppu.gpr[op.rb] & 0x1f)) & ppu_rotate_mask(32 + op.mb32, 32 + op.me32);
 	if constexpr (((Flags == has_rc) || ...))
 		ppu_cr_set<s64>(ppu, 0, ppu.gpr[op.ra], 0);
 	};
@@ -3599,7 +3599,7 @@ auto RLDICL()
 		return ppu_exec_select<Flags...>::template select<>();
 
 	static const auto exec = [](ppu_thread& ppu, ppu_opcode_t op) {
-	ppu.gpr[op.ra] = utils::rol64(ppu.gpr[op.rs], op.sh64) & (~0ull >> op.mbe64);
+	ppu.gpr[op.ra] = std::rotl<u64>(ppu.gpr[op.rs], op.sh64) & (~0ull >> op.mbe64);
 	if constexpr (((Flags == has_rc) || ...))
 		ppu_cr_set<s64>(ppu, 0, ppu.gpr[op.ra], 0);
 	};
@@ -3613,7 +3613,7 @@ auto RLDICR()
 		return ppu_exec_select<Flags...>::template select<>();
 
 	static const auto exec = [](ppu_thread& ppu, ppu_opcode_t op) {
-	ppu.gpr[op.ra] = utils::rol64(ppu.gpr[op.rs], op.sh64) & (~0ull << (op.mbe64 ^ 63));
+	ppu.gpr[op.ra] = std::rotl<u64>(ppu.gpr[op.rs], op.sh64) & (~0ull << (op.mbe64 ^ 63));
 	if constexpr (((Flags == has_rc) || ...))
 		ppu_cr_set<s64>(ppu, 0, ppu.gpr[op.ra], 0);
 	};
@@ -3627,7 +3627,7 @@ auto RLDIC()
 		return ppu_exec_select<Flags...>::template select<>();
 
 	static const auto exec = [](ppu_thread& ppu, ppu_opcode_t op) {
-	ppu.gpr[op.ra] = utils::rol64(ppu.gpr[op.rs], op.sh64) & ppu_rotate_mask(op.mbe64, op.sh64 ^ 63);
+	ppu.gpr[op.ra] = std::rotl<u64>(ppu.gpr[op.rs], op.sh64) & ppu_rotate_mask(op.mbe64, op.sh64 ^ 63);
 	if constexpr (((Flags == has_rc) || ...))
 		ppu_cr_set<s64>(ppu, 0, ppu.gpr[op.ra], 0);
 	};
@@ -3642,7 +3642,7 @@ auto RLDIMI()
 
 	static const auto exec = [](ppu_thread& ppu, ppu_opcode_t op) {
 	const u64 mask = ppu_rotate_mask(op.mbe64, op.sh64 ^ 63);
-	ppu.gpr[op.ra] = (ppu.gpr[op.ra] & ~mask) | (utils::rol64(ppu.gpr[op.rs], op.sh64) & mask);
+	ppu.gpr[op.ra] = (ppu.gpr[op.ra] & ~mask) | (std::rotl<u64>(ppu.gpr[op.rs], op.sh64) & mask);
 	if constexpr (((Flags == has_rc) || ...))
 		ppu_cr_set<s64>(ppu, 0, ppu.gpr[op.ra], 0);
 	};
@@ -3656,7 +3656,7 @@ auto RLDCL()
 		return ppu_exec_select<Flags...>::template select<>();
 
 	static const auto exec = [](ppu_thread& ppu, ppu_opcode_t op) {
-	ppu.gpr[op.ra] = utils::rol64(ppu.gpr[op.rs], ppu.gpr[op.rb] & 0x3f) & (~0ull >> op.mbe64);
+	ppu.gpr[op.ra] = std::rotl<u64>(ppu.gpr[op.rs], ppu.gpr[op.rb] & 0x3f) & (~0ull >> op.mbe64);
 	if constexpr (((Flags == has_rc) || ...))
 		ppu_cr_set<s64>(ppu, 0, ppu.gpr[op.ra], 0);
 	};
@@ -3670,7 +3670,7 @@ auto RLDCR()
 		return ppu_exec_select<Flags...>::template select<>();
 
 	static const auto exec = [](ppu_thread& ppu, ppu_opcode_t op) {
-	ppu.gpr[op.ra] = utils::rol64(ppu.gpr[op.rs], ppu.gpr[op.rb] & 0x3f) & (~0ull << (op.mbe64 ^ 63));
+	ppu.gpr[op.ra] = std::rotl<u64>(ppu.gpr[op.rs], ppu.gpr[op.rb] & 0x3f) & (~0ull << (op.mbe64 ^ 63));
 	if constexpr (((Flags == has_rc) || ...))
 		ppu_cr_set<s64>(ppu, 0, ppu.gpr[op.ra], 0);
 	};

rpcs3/Emu/Cell/PPUThread.cpp

@@ -832,6 +832,9 @@ extern void ppu_register_function_at(u32 addr, u32 size, ppu_intrp_func_t ptr =
 		return;
 	}
 
+	size = utils::align<u32>(size + addr % 4, 4);
+	addr &= -4;
+
 	// Initialize interpreter cache
 	while (size)
 	{
@@ -2308,7 +2311,7 @@ void ppu_thread::cpu_sleep()
 	raddr = 0;
 
 	// Setup wait flag and memory flags to relock itself
-	state += g_use_rtm ? cpu_flag::wait : cpu_flag::wait + cpu_flag::memory;
+	state += cpu_flag::wait + cpu_flag::memory;
 
 	if (auto ptr = vm::g_tls_locked)
 	{
@@ -2454,10 +2457,8 @@ ppu_thread::ppu_thread(const ppu_thread_params& param, std::string_view name, u3
 	// Trigger the scheduler
 	state += cpu_flag::suspend;
 
-	if (!g_use_rtm)
+	// Acquire memory passive lock
-	{
 	state += cpu_flag::memory;
-	}
 
 	call_history.data.resize(g_cfg.core.ppu_call_history ? call_history_max_size : 1);
 	syscall_history.data.resize(g_cfg.core.ppu_call_history ? syscall_history_max_size : 1);
@@ -2703,11 +2704,7 @@ ppu_thread::ppu_thread(utils::serial& ar)
 
 	// Trigger the scheduler
 	state += cpu_flag::suspend;
-
-	if (!g_use_rtm)
-	{
 	state += cpu_flag::memory;
-	}
 
 	ppu_tname = make_single<std::string>(ar.pop<std::string>());
 
@@ -3191,221 +3188,6 @@ extern u64 ppu_ldarx(ppu_thread& ppu, u32 addr)
 	return ppu_load_acquire_reservation<u64>(ppu, addr);
 }
 
-const auto ppu_stcx_accurate_tx = build_function_asm<u64(*)(u32 raddr, u64 rtime, const void* _old, u64 _new)>("ppu_stcx_accurate_tx", [](native_asm& c, auto& args)
-{
-	using namespace asmjit;
-
-#if defined(ARCH_X64)
-	Label fall = c.newLabel();
-	Label fail = c.newLabel();
-	Label _ret = c.newLabel();
-	Label load = c.newLabel();
-
-	//if (utils::has_avx() && !s_tsx_avx)
-	//{
-	//	c.vzeroupper();
-	//}
-
-	// Create stack frame if necessary (Windows ABI has only 6 volatile vector registers)
-	c.push(x86::rbp);
-	c.push(x86::r13);
-	c.push(x86::r14);
-	c.sub(x86::rsp, 48);
-#ifdef _WIN32
-	if (!s_tsx_avx)
-	{
-		c.movups(x86::oword_ptr(x86::rsp, 0), x86::xmm6);
-		c.movups(x86::oword_ptr(x86::rsp, 16), x86::xmm7);
-	}
-#endif
-
-	// Prepare registers
-	build_swap_rdx_with(c, args, x86::r10);
-	c.movabs(x86::rbp, reinterpret_cast<u64>(&vm::g_sudo_addr));
-	c.mov(x86::rbp, x86::qword_ptr(x86::rbp));
-	c.lea(x86::rbp, x86::qword_ptr(x86::rbp, args[0]));
-	c.and_(x86::rbp, -128);
-	c.prefetchw(x86::byte_ptr(x86::rbp, 0));
-	c.prefetchw(x86::byte_ptr(x86::rbp, 64));
-	c.movzx(args[0].r32(), args[0].r16());
-	c.shr(args[0].r32(), 1);
-	c.movabs(x86::r11, reinterpret_cast<u64>(+vm::g_reservations));
-	c.lea(x86::r11, x86::qword_ptr(x86::r11, args[0]));
-	c.and_(x86::r11, -128 / 2);
-	c.and_(args[0].r32(), 63);
-
-	// Prepare data
-	if (s_tsx_avx)
-	{
-		c.vmovups(x86::ymm0, x86::ymmword_ptr(args[2], 0));
-		c.vmovups(x86::ymm1, x86::ymmword_ptr(args[2], 32));
-		c.vmovups(x86::ymm2, x86::ymmword_ptr(args[2], 64));
-		c.vmovups(x86::ymm3, x86::ymmword_ptr(args[2], 96));
-	}
-	else
-	{
-		c.movaps(x86::xmm0, x86::oword_ptr(args[2], 0));
-		c.movaps(x86::xmm1, x86::oword_ptr(args[2], 16));
-		c.movaps(x86::xmm2, x86::oword_ptr(args[2], 32));
-		c.movaps(x86::xmm3, x86::oword_ptr(args[2], 48));
-		c.movaps(x86::xmm4, x86::oword_ptr(args[2], 64));
-		c.movaps(x86::xmm5, x86::oword_ptr(args[2], 80));
-		c.movaps(x86::xmm6, x86::oword_ptr(args[2], 96));
-		c.movaps(x86::xmm7, x86::oword_ptr(args[2], 112));
-	}
-
-	// Alloc r14 to stamp0
-	const auto stamp0 = x86::r14;
-	build_get_tsc(c, stamp0);
-
-	Label fail2 = c.newLabel();
-
-	Label tx1 = build_transaction_enter(c, fall, [&]()
-	{
-		build_get_tsc(c);
-		c.sub(x86::rax, stamp0);
-		c.movabs(x86::r13, reinterpret_cast<u64>(&g_rtm_tx_limit2));
-		c.cmp(x86::rax, x86::qword_ptr(x86::r13));
-		c.jae(fall);
-	});
-
-	// Check pause flag
-	c.bt(x86::dword_ptr(args[2], ::offset32(&ppu_thread::state) - ::offset32(&ppu_thread::rdata)), static_cast<u32>(cpu_flag::pause));
-	c.jc(fall);
-	c.xbegin(tx1);
-
-	if (s_tsx_avx)
-	{
-		c.vxorps(x86::ymm0, x86::ymm0, x86::ymmword_ptr(x86::rbp, 0));
-		c.vxorps(x86::ymm1, x86::ymm1, x86::ymmword_ptr(x86::rbp, 32));
-		c.vxorps(x86::ymm2, x86::ymm2, x86::ymmword_ptr(x86::rbp, 64));
-		c.vxorps(x86::ymm3, x86::ymm3, x86::ymmword_ptr(x86::rbp, 96));
-		c.vorps(x86::ymm0, x86::ymm0, x86::ymm1);
-		c.vorps(x86::ymm1, x86::ymm2, x86::ymm3);
-		c.vorps(x86::ymm0, x86::ymm1, x86::ymm0);
-		c.vptest(x86::ymm0, x86::ymm0);
-	}
-	else
-	{
-		c.xorps(x86::xmm0, x86::oword_ptr(x86::rbp, 0));
-		c.xorps(x86::xmm1, x86::oword_ptr(x86::rbp, 16));
-		c.xorps(x86::xmm2, x86::oword_ptr(x86::rbp, 32));
-		c.xorps(x86::xmm3, x86::oword_ptr(x86::rbp, 48));
-		c.xorps(x86::xmm4, x86::oword_ptr(x86::rbp, 64));
-		c.xorps(x86::xmm5, x86::oword_ptr(x86::rbp, 80));
-		c.xorps(x86::xmm6, x86::oword_ptr(x86::rbp, 96));
-		c.xorps(x86::xmm7, x86::oword_ptr(x86::rbp, 112));
-		c.orps(x86::xmm0, x86::xmm1);
-		c.orps(x86::xmm2, x86::xmm3);
-		c.orps(x86::xmm4, x86::xmm5);
-		c.orps(x86::xmm6, x86::xmm7);
-		c.orps(x86::xmm0, x86::xmm2);
-		c.orps(x86::xmm4, x86::xmm6);
-		c.orps(x86::xmm0, x86::xmm4);
-		c.ptest(x86::xmm0, x86::xmm0);
-	}
-
-	c.jnz(fail);
-
-	// Store 8 bytes
-	c.mov(x86::qword_ptr(x86::rbp, args[0], 1, 0), args[3]);
-
-	c.xend();
-	c.lock().add(x86::qword_ptr(x86::r11), 64);
-	build_get_tsc(c);
-	c.sub(x86::rax, stamp0);
-	c.jmp(_ret);
-
-	// XABORT is expensive so try to finish with xend instead
-	c.bind(fail);
-
-	// Load old data to store back in rdata
-	if (s_tsx_avx)
-	{
-		c.vmovaps(x86::ymm0, x86::ymmword_ptr(x86::rbp, 0));
-		c.vmovaps(x86::ymm1, x86::ymmword_ptr(x86::rbp, 32));
-		c.vmovaps(x86::ymm2, x86::ymmword_ptr(x86::rbp, 64));
-		c.vmovaps(x86::ymm3, x86::ymmword_ptr(x86::rbp, 96));
-	}
-	else
-	{
-		c.movaps(x86::xmm0, x86::oword_ptr(x86::rbp, 0));
-		c.movaps(x86::xmm1, x86::oword_ptr(x86::rbp, 16));
-		c.movaps(x86::xmm2, x86::oword_ptr(x86::rbp, 32));
-		c.movaps(x86::xmm3, x86::oword_ptr(x86::rbp, 48));
-		c.movaps(x86::xmm4, x86::oword_ptr(x86::rbp, 64));
-		c.movaps(x86::xmm5, x86::oword_ptr(x86::rbp, 80));
-		c.movaps(x86::xmm6, x86::oword_ptr(x86::rbp, 96));
-		c.movaps(x86::xmm7, x86::oword_ptr(x86::rbp, 112));
-	}
-
-	c.xend();
-	c.jmp(fail2);
-
-	c.bind(fall);
-	c.mov(x86::rax, -1);
-	c.jmp(_ret);
-
-	c.bind(fail2);
-	c.lock().sub(x86::qword_ptr(x86::r11), 64);
-	c.bind(load);
-
-	// Store previous data back to rdata
-	if (s_tsx_avx)
-	{
-		c.vmovaps(x86::ymmword_ptr(args[2], 0), x86::ymm0);
-		c.vmovaps(x86::ymmword_ptr(args[2], 32), x86::ymm1);
-		c.vmovaps(x86::ymmword_ptr(args[2], 64), x86::ymm2);
-		c.vmovaps(x86::ymmword_ptr(args[2], 96), x86::ymm3);
-	}
-	else
-	{
-		c.movaps(x86::oword_ptr(args[2], 0), x86::xmm0);
-		c.movaps(x86::oword_ptr(args[2], 16), x86::xmm1);
-		c.movaps(x86::oword_ptr(args[2], 32), x86::xmm2);
-		c.movaps(x86::oword_ptr(args[2], 48), x86::xmm3);
-		c.movaps(x86::oword_ptr(args[2], 64), x86::xmm4);
-		c.movaps(x86::oword_ptr(args[2], 80), x86::xmm5);
-		c.movaps(x86::oword_ptr(args[2], 96), x86::xmm6);
-		c.movaps(x86::oword_ptr(args[2], 112), x86::xmm7);
-	}
-
-	c.mov(x86::rax, -1);
-	c.mov(x86::qword_ptr(args[2], ::offset32(&ppu_thread::last_ftime) - ::offset32(&ppu_thread::rdata)), x86::rax);
-	c.xor_(x86::eax, x86::eax);
-	//c.jmp(_ret);
-
-	c.bind(_ret);
-
-#ifdef _WIN32
-	if (!s_tsx_avx)
-	{
-		c.vmovups(x86::xmm6, x86::oword_ptr(x86::rsp, 0));
-		c.vmovups(x86::xmm7, x86::oword_ptr(x86::rsp, 16));
-	}
-#endif
-
-	if (s_tsx_avx)
-	{
-		c.vzeroupper();
-	}
-
-	c.add(x86::rsp, 48);
-	c.pop(x86::r14);
-	c.pop(x86::r13);
-	c.pop(x86::rbp);
-
-	maybe_flush_lbr(c);
-	c.ret();
-#else
-	UNUSED(args);
-
-	// Unimplemented should fail.
-	c.brk(Imm(0x42));
-	c.ret(a64::x30);
-#endif
-});
-
 template <typename T>
 static bool ppu_store_reservation(ppu_thread& ppu, u32 addr, u64 reg_value)
 {
@@ -3486,77 +3268,6 @@ static bool ppu_store_reservation(ppu_thread& ppu, u32 addr, u64 reg_value)
 				return false;
 			}
 
-			if (g_use_rtm) [[likely]]
-			{
-				switch (u64 count = ppu_stcx_accurate_tx(addr & -8, rtime, ppu.rdata, std::bit_cast<u64>(new_data)))
-				{
-				case umax:
-				{
-					auto& all_data = *vm::get_super_ptr<spu_rdata_t>(addr & -128);
-					auto& sdata = *vm::get_super_ptr<atomic_be_t<u64>>(addr & -8);
-
-					const bool ok = cpu_thread::suspend_all<+3>(&ppu, {all_data, all_data + 64, &res}, [&]
-					{
-						if ((res & -128) == rtime && cmp_rdata(ppu.rdata, all_data))
-						{
-							sdata.release(new_data);
-							res += 64;
-							return true;
-						}
-
-						mov_rdata_nt(ppu.rdata, all_data);
-						res -= 64;
-						return false;
-					});
-
-					if (ok)
-					{
-						break;
-					}
-
-					ppu.last_ftime = -1;
-					[[fallthrough]];
-				}
-				case 0:
-				{
-					if (ppu.last_faddr == addr)
-					{
-						ppu.last_fail++;
-					}
-
-					if (ppu.last_ftime != umax)
-					{
-						ppu.last_faddr = 0;
-						return false;
-					}
-
-					utils::prefetch_read(ppu.rdata);
-					utils::prefetch_read(ppu.rdata + 64);
-					ppu.last_faddr = addr;
-					ppu.last_ftime = res.load() & -128;
-					ppu.last_ftsc = utils::get_tsc();
-					return false;
-				}
-				default:
-				{
-					if (count > 20000 && g_cfg.core.perf_report) [[unlikely]]
-					{
-						perf_log.warning("STCX: took too long: %.3fus (%u c)", count / (utils::get_tsc_freq() / 1000'000.), count);
-					}
-
-					break;
-				}
-				}
-
-				if (ppu.last_faddr == addr)
-				{
-					ppu.last_succ++;
-				}
-
-				ppu.last_faddr = 0;
-				return true;
-			}
-
 			// Align address: we do not need the lower 7 bits anymore
 			addr &= -128;
 
@@ -4007,7 +3718,7 @@ extern void ppu_finalize(const ppu_module<lv2_obj>& info, bool force_mem_release
 #endif
 }
 
-extern void ppu_precompile(std::vector<std::string>& dir_queue, std::vector<ppu_module<lv2_obj>*>* loaded_modules)
+extern void ppu_precompile(std::vector<std::string>& dir_queue, std::vector<ppu_module<lv2_obj>*>* loaded_modules, bool is_fast_compilation)
 {
 	if (g_cfg.core.ppu_decoder != ppu_decoder_type::llvm)
 	{
@@ -4455,6 +4166,12 @@ extern void ppu_precompile(std::vector<std::string>& dir_queue, std::vector<ppu_
 						break;
 					}
 
+					if (is_fast_compilation)
+					{
+						// Skip overlays in fast mode
+						break;
+					}
+
 					if (!wait_for_memory())
 					{
 						// Emulation stopped
@@ -4749,7 +4466,7 @@ extern void ppu_initialize()
 
 	progress_dialog.reset();
 
-	ppu_precompile(dir_queue, &module_list);
+	ppu_precompile(dir_queue, &module_list, false);
 
 	if (Emu.IsStopped())
 	{
@@ -5803,7 +5520,11 @@ static void ppu_initialize2(jit_compiler& jit, const ppu_module<lv2_obj>& module
 	std::unique_ptr<Module> _module = std::make_unique<Module>(obj_name, jit.get_context());
 
 	// Initialize target
+#if LLVM_VERSION_MAJOR >= 21 && (LLVM_VERSION_MINOR >= 1 || LLVM_VERSION_MAJOR >= 22)
+	_module->setTargetTriple(Triple(jit_compiler::triple1()));
+#else
 	_module->setTargetTriple(jit_compiler::triple1());
+#endif
 	_module->setDataLayout(jit.get_engine().getTargetMachine()->createDataLayout());
 
 	// Initialize translator

rpcs3/Emu/Cell/PPUTranslator.cpp

@@ -416,7 +416,6 @@ Function* PPUTranslator::GetSymbolResolver(const ppu_module<lv2_obj>& info)
 	assert(ptr_inst->getResultElementType() == m_ir->getPtrTy());
 
 	const auto faddr = m_ir->CreateLoad(ptr_inst->getResultElementType(), ptr_inst);
-	const auto faddr_int = m_ir->CreatePtrToInt(faddr, get_type<uptr>());
 	const auto pos_32 = m_reloc ? m_ir->CreateAdd(func_pc, m_seg0) : func_pc;
 	const auto pos = m_ir->CreateShl(pos_32, 1);
 	const auto ptr = m_ir->CreatePtrAdd(m_exec, pos);
@@ -427,7 +426,7 @@ Function* PPUTranslator::GetSymbolResolver(const ppu_module<lv2_obj>& info)
 	const auto seg_val = m_ir->CreateTrunc(m_ir->CreateLShr(m_seg0, 13), get_type<u16>());
 
 	// Store to jumptable
-	m_ir->CreateStore(faddr_int, ptr);
+	m_ir->CreateStore(faddr, ptr);
 	m_ir->CreateStore(seg_val, seg_ptr);
 
 	// Increment index and branch back to loop
@@ -593,6 +592,11 @@ void PPUTranslator::CallFunction(u64 target, Value* indirect)
 			{
 				callee = m_module->getOrInsertFunction(fmt::format("__0x%x", target_last - base), type);
 				cast<Function>(callee.getCallee())->setCallingConv(CallingConv::GHC);
+
+				if (g_cfg.core.ppu_prof)
+				{
+					m_ir->CreateStore(GetAddr(target_last - m_addr), m_ir->CreateStructGEP(m_thread_type, m_thread, static_cast<uint>(&m_cia - m_locals)));
+				}
 			}
 		}
 		else
@@ -1115,7 +1119,24 @@ void PPUTranslator::VCFSX(ppu_opcode_t op)
 void PPUTranslator::VCFUX(ppu_opcode_t op)
 {
 	const auto b = get_vr<u32[4]>(op.vb);
-	set_vr(op.vd, fpcast<f32[4]>(b) * fsplat<f32[4]>(std::pow(2, -static_cast<int>(op.vuimm))));
+
+#ifdef ARCH_ARM64
+	return set_vr(op.vd, fpcast<f32[4]>(b) * fsplat<f32[4]>(std::pow(2, -static_cast<int>(op.vuimm))));
+#else
+	if (m_use_avx512)
+	{
+		return set_vr(op.vd, fpcast<f32[4]>(b) * fsplat<f32[4]>(std::pow(2, -static_cast<int>(op.vuimm))));
+	}
+
+	constexpr int bit_shift = 9;
+	const auto shifted = (b >> bit_shift);
+	const auto cleared = shifted << bit_shift;
+	const auto low_bits = b - cleared;
+	const auto high_part = fpcast<f32[4]>(noncast<s32[4]>(shifted)) * fsplat<f32[4]>(static_cast<f32>(1u << bit_shift));
+	const auto low_part = fpcast<f32[4]>(noncast<s32[4]>(low_bits));
+	const auto temp = high_part + low_part;
+	set_vr(op.vd, temp * fsplat<f32[4]>(std::pow(2, -static_cast<int>(op.vuimm))));
+#endif
 }
 
 void PPUTranslator::VCMPBFP(ppu_opcode_t op)

rpcs3/Emu/Cell/SPUASMJITRecompiler.cpp

@@ -186,7 +186,7 @@ spu_function_t spu_recompiler::compile(spu_program&& _func)
 	c->cmp(SPU_OFF_32(state), 0);
 	c->jnz(label_stop);
 
-	if (g_cfg.core.spu_prof && g_cfg.core.spu_verification)
+	if ((g_cfg.core.spu_prof || g_cfg.core.spu_debug) && g_cfg.core.spu_verification)
 	{
 		c->mov(x86::rax, m_hash_start & -0xffff);
 		c->mov(SPU_OFF_64(block_hash), x86::rax);
@@ -755,7 +755,7 @@ spu_function_t spu_recompiler::compile(spu_program&& _func)
 	c->add(SPU_OFF_64(block_counter), ::size32(words) / (words_align / 4));
 
 	// Set block hash for profiling (if enabled)
-	if (g_cfg.core.spu_prof)
+	if (g_cfg.core.spu_prof || g_cfg.core.spu_debug)
 	{
 		c->mov(x86::rax, m_hash_start | 0xffff);
 		c->mov(SPU_OFF_64(block_hash), x86::rax);
@@ -1199,7 +1199,7 @@ void spu_recompiler::branch_set_link(u32 target)
 				c->movdqa(x86::dqword_ptr(*cpu, *qw1, 0, ::offset32(&spu_thread::stack_mirror)), x86::xmm0);
 
 				// Set block hash for profiling (if enabled)
-				if (g_cfg.core.spu_prof)
+				if (g_cfg.core.spu_prof || g_cfg.core.spu_debug)
 				{
 					c->mov(x86::rax, m_hash_start | 0xffff);
 					c->mov(SPU_OFF_64(block_hash), x86::rax);
@@ -3212,7 +3212,7 @@ void spu_recompiler::ROTQBYI(spu_opcode_t op)
 	}
 	else if (s == 4 || s == 8 || s == 12)
 	{
-		c->pshufd(va, va, utils::rol8(0xE4, s / 2));
+		c->pshufd(va, va, std::rotl<u8>(0xE4, s / 2));
 	}
 	else if (utils::has_ssse3())
 	{

rpcs3/Emu/Cell/SPUCommonRecompiler.cpp

@@ -38,26 +38,9 @@ constexpr u32 s_reg_max = spu_recompiler_base::s_reg_max;
 template<typename T>
 struct span_less
 {
-	static int compare(const std::span<T>& lhs, const std::span<T>& rhs) noexcept
+	static auto compare(const std::span<T>& lhs, const std::span<T>& rhs) noexcept
 	{
-		// TODO: Replace with std::lexicographical_compare_three_way when it becomes available to all compilers
+		return std::lexicographical_compare_three_way(lhs.begin(), lhs.end(), rhs.begin(), rhs.end());
-		for (usz i = 0, last = std::min(lhs.size(), rhs.size()); i != last; i++)
-		{
-			const T vl = lhs[i];
-			const T vr = rhs[i];
-
-			if (vl != vr)
-			{
-				return vl < vr ? -1 : 1;
-			}
-		}
-
-		if (lhs.size() != rhs.size())
-		{
-			return lhs.size() < rhs.size() ? -1 : 1;
-		}
-
-		return 0;
 	}
 
 	bool operator()(const std::span<T>& lhs, const std::span<T>& rhs) const noexcept
@@ -737,9 +720,19 @@ void spu_cache::initialize(bool build_existing_cache)
 	}
 
 	// SPU cache file (version + block size type)
-	const std::string loc = ppu_cache + "spu-" + fmt::to_lower(g_cfg.core.spu_block_size.to_string()) + "-v1-tane.dat";
+	const std::string filename = "spu-" + fmt::to_lower(g_cfg.core.spu_block_size.to_string()) + "-v1-tane.dat";
+	const std::string loc = ppu_cache + filename;
+	const std::string loc_debug = fs::get_cache_dir() + "DEBUG/" + filename;
 
-	spu_cache cache(loc);
+	bool is_debug = false;
+
+	if (fs::is_file(loc_debug))
+	{
+		spu_log.success("SPU Cache override applied!");
+		is_debug = true;
+	}
+
+	spu_cache cache(is_debug ? loc_debug : loc);
 
 	if (!cache)
 	{
@@ -1309,7 +1302,7 @@ bool spu_program::operator<(const spu_program& rhs) const noexcept
 	std::span<const u32> lhs_data(data.data() + lhs_offs, data.size() - lhs_offs);
 	std::span<const u32> rhs_data(rhs.data.data() + rhs_offs, rhs.data.size() - rhs_offs);
 
-	const int cmp0 = span_less<const u32>::compare(lhs_data, rhs_data);
+	const auto cmp0 = span_less<const u32>::compare(lhs_data, rhs_data);
 
 	if (cmp0 < 0)
 		return true;
@@ -1320,7 +1313,7 @@ bool spu_program::operator<(const spu_program& rhs) const noexcept
 	lhs_data = {data.data(), lhs_offs};
 	rhs_data = {rhs.data.data(), rhs_offs};
 
-	const int cmp1 = span_less<const u32>::compare(lhs_data, rhs_data);
+	const auto cmp1 = span_less<const u32>::compare(lhs_data, rhs_data);
 
 	if (cmp1 < 0)
 		return true;
@@ -2333,7 +2326,7 @@ std::vector<u32> spu_thread::discover_functions(u32 base_addr, std::span<const u
 		// Search for BRSL LR and BRASL LR or BR
 		// TODO: BISL
 		const v128 inst = read_from_ptr<be_t<v128>>(ls.data(), i - base_addr);
-		const v128 cleared_i16 = gv_and32(inst, v128::from32p(utils::rol32(~0xffff, 7)));
+		const v128 cleared_i16 = gv_and32(inst, v128::from32p(std::rotl<u32>(~0xffff, 7)));
 		const v128 eq_brsl = gv_eq32(cleared_i16, v128::from32p(0x66u << 23));
 		const v128 eq_brasl = gv_eq32(cleared_i16, brasl_mask);
 		const v128 eq_br = gv_eq32(cleared_i16, v128::from32p(0x64u << 23));
@@ -3086,6 +3079,39 @@ spu_program spu_recompiler_base::analyse(const be_t<u32>* ls, u32 entry_point, s
 				values[op.rt] = pos + 4;
 			}
 
+			const u32 pos_next = wa;
+
+			bool is_no_return = false;
+
+			if (pos_next >= lsa && pos_next < limit)
+			{
+				const u32 data_next = ls[pos_next / 4];
+				const auto type_next = g_spu_itype.decode(data_next);
+				const auto flag_next = g_spu_iflag.decode(data_next);
+				const auto op_next = spu_opcode_t{data_next};
+
+				if (!(type_next & spu_itype::zregmod) && !(type_next & spu_itype::branch))
+				{
+					if (auto iflags = g_spu_iflag.decode(data_next))
+					{
+						if (+flag_next & +spu_iflag::use_ra)
+						{
+							is_no_return = is_no_return || (op_next.ra >= 4 && op_next.ra < 10);
+						}
+
+						if (+flag_next & +spu_iflag::use_rb)
+						{
+							is_no_return = is_no_return || (op_next.rb >= 4 && op_next.rb < 10);
+						}
+
+						if (type_next & spu_itype::_quadrop && +iflags & +spu_iflag::use_rc)
+						{
+							is_no_return = is_no_return || (op_next.ra >= 4 && op_next.rb < 10);
+						}
+					}
+				}
+			}
+
 			if (af & vf::is_const)
 			{
 				const u32 target = spu_branch_target(av);
@@ -3122,7 +3148,7 @@ spu_program spu_recompiler_base::analyse(const be_t<u32>* ls, u32 entry_point, s
 					limit = std::min<u32>(limit, target);
 				}
 
-				if (sl && g_cfg.core.spu_block_size != spu_block_size_type::safe)
+				if (!is_no_return && sl && g_cfg.core.spu_block_size != spu_block_size_type::safe)
 				{
 					m_ret_info[pos / 4 + 1] = true;
 					m_entry_info[pos / 4 + 1] = true;
@@ -3139,7 +3165,7 @@ spu_program spu_recompiler_base::analyse(const be_t<u32>* ls, u32 entry_point, s
 				u64 dabs = 0;
 				u64 drel = 0;
 
-				for (u32 i = start; i < limit; i += 4)
+				for (u32 i = start, abs_fail = 0, rel_fail = 0; i < limit; i += 4)
 				{
 					const u32 target = ls[i / 4];
 
@@ -3149,17 +3175,40 @@ spu_program spu_recompiler_base::analyse(const be_t<u32>* ls, u32 entry_point, s
 						break;
 					}
 
+					if (target >= SPU_LS_SIZE && target <= 0u - SPU_LS_SIZE)
+					{
+						if (g_spu_itype.decode(target) != spu_itype::UNK)
+						{
+							// End of jumptable: valid instruction
+							break;
+						}
+					}
+
 					if (target >= lsa && target < SPU_LS_SIZE)
 					{
 						// Possible jump table entry (absolute)
+						if (!abs_fail)
+						{
 							jt_abs.push_back(target);
 						}
+					}
+					else
+					{
+						abs_fail++;
+					}
 
 					if (target + start >= lsa && target + start < SPU_LS_SIZE)
 					{
 						// Possible jump table entry (relative)
+						if (!rel_fail)
+						{
 							jt_rel.push_back(target + start);
 						}
+					}
+					else
+					{
+						rel_fail++;
+					}
 
 					if (std::max(jt_abs.size(), jt_rel.size()) * 4 + start <= i)
 					{
@@ -3170,6 +3219,35 @@ spu_program spu_recompiler_base::analyse(const be_t<u32>* ls, u32 entry_point, s
 					}
 				}
 
+				for (usz i = 0; i < jt_abs.size(); i++)
+				{
+					if (jt_abs[i] == start + jt_abs.size() * 4)
+					{
+						// If jumptable contains absolute address of code start after the jumptable itself
+						// It is likely an absolute-type jumptable
+
+						bool is_good_conclusion = true;
+
+						// For verification: make sure there is none like this in relative table
+
+						for (u32 target : jt_rel)
+						{
+							if (target == start + jt_rel.size() * 4)
+							{
+								is_good_conclusion = false;
+								break;
+							}
+						}
+						
+						if (is_good_conclusion)
+						{
+							jt_rel.clear();
+						}
+
+						break;
+					}
+				}
+
 				// Choose position after the jt as an anchor and compute the average distance
 				for (u32 target : jt_abs)
 				{
@@ -3268,9 +3346,9 @@ spu_program spu_recompiler_base::analyse(const be_t<u32>* ls, u32 entry_point, s
 				spu_log.notice("[0x%x] At 0x%x: ignoring indirect branch (SYNC)", entry_point, pos);
 			}
 
-			if (type == spu_itype::BI || sl)
+			if (type == spu_itype::BI || sl || is_no_return)
 			{
-				if (type == spu_itype::BI || g_cfg.core.spu_block_size == spu_block_size_type::safe)
+				if (type == spu_itype::BI || g_cfg.core.spu_block_size == spu_block_size_type::safe || is_no_return)
 				{
 					m_targets[pos];
 				}
@@ -3307,9 +3385,42 @@ spu_program spu_recompiler_base::analyse(const be_t<u32>* ls, u32 entry_point, s
 				break;
 			}
 
+			const u32 pos_next = wa;
+
+			bool is_no_return = false;
+
+			if (pos_next >= lsa && pos_next < limit)
+			{
+				const u32 data_next = ls[pos_next / 4];
+				const auto type_next = g_spu_itype.decode(data_next);
+				const auto flag_next = g_spu_iflag.decode(data_next);
+				const auto op_next = spu_opcode_t{data_next};
+
+				if (!(type_next & spu_itype::zregmod) && !(type_next & spu_itype::branch))
+				{
+					if (auto iflags = g_spu_iflag.decode(data_next))
+					{
+						if (+flag_next & +spu_iflag::use_ra)
+						{
+							is_no_return = is_no_return || (op_next.ra >= 4 && op_next.ra < 10);
+						}
+
+						if (+flag_next & +spu_iflag::use_rb)
+						{
+							is_no_return = is_no_return || (op_next.rb >= 4 && op_next.rb < 10);
+						}
+
+						if (type_next & spu_itype::_quadrop && +iflags & +spu_iflag::use_rc)
+						{
+							is_no_return = is_no_return || (op_next.rc >= 4 && op_next.rc < 10);
+						}
+					}
+				}
+			}
+
 			m_targets[pos].push_back(target);
 
-			if (g_cfg.core.spu_block_size != spu_block_size_type::safe)
+			if (!is_no_return && g_cfg.core.spu_block_size != spu_block_size_type::safe)
 			{
 				m_ret_info[pos / 4 + 1] = true;
 				m_entry_info[pos / 4 + 1] = true;
@@ -3317,7 +3428,7 @@ spu_program spu_recompiler_base::analyse(const be_t<u32>* ls, u32 entry_point, s
 				add_block(pos + 4);
 			}
 
-			if (g_cfg.core.spu_block_size == spu_block_size_type::giga && !sync)
+			if (!is_no_return && g_cfg.core.spu_block_size == spu_block_size_type::giga && !sync)
 			{
 				m_entry_info[target / 4] = true;
 				add_block(target);
@@ -4862,6 +4973,7 @@ spu_program spu_recompiler_base::analyse(const be_t<u32>* ls, u32 entry_point, s
 		u32 lsa_last_pc = SPU_LS_SIZE; // PC of first LSA write
 		u32 get_pc = SPU_LS_SIZE; // PC of GETLLAR
 		u32 put_pc = SPU_LS_SIZE; // PC of PUTLLC
+		u32 rdatomic_pc = SPU_LS_SIZE; // PC of last RdAtomcStat read
 		reg_state_t ls{}; // state of LS load/store address register
 		reg_state_t ls_offs = reg_state_t::from_value(0); // Added value to ls
 		reg_state_t lsa{}; // state of LSA register on GETLLAR
@@ -4877,20 +4989,27 @@ spu_program spu_recompiler_base::analyse(const be_t<u32>* ls, u32 entry_point, s
 		bool select_16_or_0_at_runtime = false;
 		bool put_active = false; // PUTLLC happened
 		bool get_rdatomic = false; // True if MFC_RdAtomicStat was read after GETLLAR
+		u32 required_pc = SPU_LS_SIZE; // Require program to be location specific for this optimization (SPU_LS_SIZE - no requirement)
 		u32 mem_count = 0;
+		u32 break_cause = 100;
+		u32 break_pc = SPU_LS_SIZE;
 
 		// Return old state for error reporting
 		atomic16_t discard()
 		{
 			const u32 pc = lsa_pc;
 			const u32 last_pc = lsa_last_pc;
+			const u32 cause = break_cause;
+			const u32 break_pos = break_pc;
 
 			const atomic16_t old = *this;
 			*this = atomic16_t{};
 
 			// Keep some members
-			lsa_pc = pc;
+			this->lsa_pc = pc;
-			lsa_last_pc = last_pc;
+			this->lsa_last_pc = last_pc;
+			this->break_cause = cause;
+			this->break_pc = break_pos;
 			return old;
 		}
 
@@ -4900,7 +5019,7 @@ spu_program spu_recompiler_base::analyse(const be_t<u32>* ls, u32 entry_point, s
 			ls_invalid = true;
 			ls_write |= write;
 
-			if (write)
+			if (ls_write)
 			{
 				return discard();
 			}
@@ -5097,15 +5216,17 @@ spu_program spu_recompiler_base::analyse(const be_t<u32>* ls, u32 entry_point, s
 		{
 			if (previous.active && likely_putllc_loop && getllar_starts.contains(previous.lsa_pc))
 			{
-				const bool is_first = !std::exchange(getllar_starts[previous.lsa_pc], true);
+				had_putllc_evaluation = true;
 
-				if (!is_first)
+				if (cause != 24)
 				{
+					atomic16->break_cause = cause; 
+					atomic16->break_pc = pos; 
 					return;
 				}
 
-				had_putllc_evaluation = true;
+				cause = atomic16->break_cause;
-
+				getllar_starts[previous.lsa_pc] = true;
 				g_fxo->get<putllc16_statistics_t>().breaking_reason[cause]++;
 
 				if (!spu_log.notice)
@@ -5113,7 +5234,7 @@ spu_program spu_recompiler_base::analyse(const be_t<u32>* ls, u32 entry_point, s
 					return;
 				}
 
-				std::string break_error = fmt::format("PUTLLC pattern breakage [%x mem=%d lsa_const=%d cause=%u] (lsa_pc=0x%x)", pos, previous.mem_count, u32{!previous.ls_offs.is_const()} * 2 + previous.lsa.is_const(), cause, previous.lsa_pc);
+				std::string break_error = fmt::format("PUTLLC pattern breakage [%x mem=%d lsa_const=%d cause=%u] (lsa_pc=0x%x)", atomic16->break_pc, previous.mem_count, u32{!previous.ls_offs.is_const()} * 2 + previous.lsa.is_const(), cause, previous.lsa_pc);
 
 				const auto values = sort_breakig_reasons(g_fxo->get<putllc16_statistics_t>().breaking_reason);
 
@@ -5396,7 +5517,7 @@ spu_program spu_recompiler_base::analyse(const be_t<u32>* ls, u32 entry_point, s
 									const usz block_tail = duplicate_positions[it_begin - it_tail];
 
 									// Check if the distance is precisely two times from the end
-									if (reg_state_it.size() - block_start != utils::rol64(reg_state_it.size() - block_tail, 1))
+									if (reg_state_it.size() - block_start != std::rotl<u64>(reg_state_it.size() - block_tail, 1))
 									{
 										continue;
 									}
@@ -6213,6 +6334,8 @@ spu_program spu_recompiler_base::analyse(const be_t<u32>* ls, u32 entry_point, s
 							break;
 						}
 
+						atomic16->rdatomic_pc = pos;
+
 						const auto it = atomic16_all.find(pos);
 
 						if (it == atomic16_all.end())
@@ -6275,6 +6398,7 @@ spu_program spu_recompiler_base::analyse(const be_t<u32>* ls, u32 entry_point, s
 							existing.ls_invalid |= atomic16->ls_invalid;
 							existing.ls_access |= atomic16->ls_access;
 							existing.mem_count = std::max<u32>(existing.mem_count, atomic16->mem_count);
+							existing.required_pc = std::min<u32>(existing.required_pc, atomic16->required_pc);
 							existing.select_16_or_0_at_runtime |= atomic16->select_16_or_0_at_runtime;
 						}
 
@@ -6289,6 +6413,24 @@ spu_program spu_recompiler_base::analyse(const be_t<u32>* ls, u32 entry_point, s
 						invalidate = false;
 					}
 				}
+				else if (atomic16->break_cause != 100 && atomic16->lsa_pc != SPU_LS_SIZE)
+				{
+					const auto it = atomic16_all.find(pos);
+
+					if (it == atomic16_all.end())
+					{
+						// Ensure future failure
+						atomic16_all.emplace(pos, *atomic16);
+						break_putllc16(24, FN(x.active = true, x)(as_rvalue(*atomic16)));
+					}
+					else if (it->second.active && atomic16->break_cause != 100)
+					{
+						it->second = *atomic16;
+						break_putllc16(24, FN(x.active = true, x)(as_rvalue(*atomic16)));
+					}
+
+					atomic16->break_cause = 100;
+				}
 
 				break;
 			}
@@ -6359,6 +6501,10 @@ spu_program spu_recompiler_base::analyse(const be_t<u32>* ls, u32 entry_point, s
 
 				// Do not clear lower 16 bytes addressing because the program can move on 4-byte basis
 				const u32 offs = spu_branch_target(pos - result.lower_bound, op.si16);
+				const u32 true_offs = spu_branch_target(pos, op.si16);
+
+				// Make this optimization depend on the location of the program
+				atomic16->required_pc = result.lower_bound;
 
 				if (atomic16->lsa.is_const() && [&]()
 				{
@@ -6383,6 +6529,10 @@ spu_program spu_recompiler_base::analyse(const be_t<u32>* ls, u32 entry_point, s
 				{
 					// Ignore memory access in this case
 				}
+				else if (atomic16->lsa.is_const() && !atomic16->lsa.compare_with_mask_indifference(true_offs, SPU_LS_MASK_128))
+				{
+					// Same
+				}
 				else if (atomic16->ls_invalid && is_store)
 				{
 					break_putllc16(35, atomic16->set_invalid_ls(is_store));
@@ -7126,7 +7276,7 @@ spu_program spu_recompiler_base::analyse(const be_t<u32>* ls, u32 entry_point, s
 
 	for (const auto& [pc_commited, pattern] : atomic16_all)
 	{
-		if (!pattern.active)
+		if (!pattern.active || pattern.lsa_pc >= pattern.rdatomic_pc)
 		{
 			continue;
 		}
@@ -7136,26 +7286,43 @@ spu_program spu_recompiler_base::analyse(const be_t<u32>* ls, u32 entry_point, s
 			continue;
 		}
 
+		std::string pattern_hash;
+		{
+			sha1_context ctx;
+			u8 output[20]{};
+
+			sha1_starts(&ctx);
+			sha1_update(&ctx, reinterpret_cast<const u8*>(result.data.data()) + (pattern.lsa_pc - result.lower_bound), pattern.rdatomic_pc - pattern.lsa_pc);
+			sha1_finish(&ctx, output);
+			fmt::append(pattern_hash, "%s", fmt::base57(output));
+		}
+
+		union putllc16_or_0_info
+		{
+			u64 data;
+			bf_t<u64, 32, 18> required_pc;
+			bf_t<u64, 30, 2> type;
+			bf_t<u64, 29, 1> runtime16_select;
+			bf_t<u64, 28, 1> no_notify;
+			bf_t<u64, 18, 8> reg;
+			bf_t<u64, 0, 18> off18;
+			bf_t<u64, 0, 8> reg2;
+		} value{};
+
 		auto& stats = g_fxo->get<putllc16_statistics_t>();
 		had_putllc_evaluation = true;
 
 		if (!pattern.ls_write)
 		{
-			spu_log.success("PUTLLC0 Pattern Detected! (put_pc=0x%x, %s) (putllc0=%d, putllc16+0=%d, all=%d)", pattern.put_pc, func_hash, ++stats.nowrite, ++stats.single, +stats.all);
+			if (pattern.required_pc != SPU_LS_SIZE)
-			add_pattern(false, inst_attr::putllc0, pattern.put_pc - lsa);
+			{
-			continue;
+				value.required_pc = pattern.required_pc;
 			}
 
-		union putllc16_info
+			// spu_log.success("PUTLLC0 Pattern Detected! (put_pc=0x%x, %s) (putllc0=%d, putllc16+0=%d, all=%d)", pattern.put_pc, func_hash, ++stats.nowrite, ++stats.single, +stats.all);
-		{
+			// add_pattern(false, inst_attr::putllc0, pattern.put_pc - lsa, value.data);
-			u32 data;
+			continue;
-			bf_t<u32, 30, 2> type;
+		}
-			bf_t<u32, 29, 1> runtime16_select;
-			bf_t<u32, 28, 1> no_notify;
-			bf_t<u32, 18, 8> reg;
-			bf_t<u32, 0, 18> off18;
-			bf_t<u32, 0, 8> reg2;
-		} value{};
 
 		enum : u32
 		{
@@ -7187,6 +7354,11 @@ spu_program spu_recompiler_base::analyse(const be_t<u32>* ls, u32 entry_point, s
 		value.runtime16_select = pattern.select_16_or_0_at_runtime;
 		value.reg = s_reg_max;
 
+		if (pattern.required_pc != SPU_LS_SIZE)
+		{
+			value.required_pc = pattern.required_pc;
+		}
+
 		if (pattern.ls.is_const())
 		{
 			ensure(pattern.reg == s_reg_max && pattern.reg2 == s_reg_max && pattern.ls_offs.is_const(), "Unexpected register usage");
@@ -7215,16 +7387,35 @@ spu_program spu_recompiler_base::analyse(const be_t<u32>* ls, u32 entry_point, s
 			value.reg2 = pattern.reg2;
 		}
 
+		bool allow_pattern = true;
+
 		if (g_cfg.core.spu_accurate_reservations)
 		{
-			// Because enabling it is a hack, as it turns out
+			// The problem with PUTLLC16 optimization, that it is in theory correct at the bounds of the spu function.
-			continue;
+			// But if the SPU code reuses the cache line data observed, it is not truly atomic.
+			// So we may enable it only for known cases where SPU atomic data is not used after the function leaves.
+
+			// So the two options are:
+
+			// 1. Atomic compare exchange 16 bytes operation. (rest of data is not read) -> good for RPCS3 to optimize.
+			// 2. Fetch 128 bytes (read them later), modify only 16 bytes. -> Bad for RPCS3 to optimize.
+
+			// This difference cannot be known at analyzer time but from observing callers.
+			static constexpr std::initializer_list<std::string_view> allowed_patterns =
+			{
+				"620oYSe8uQqq9eTkhWfMqoEXX0us"sv, // CellSpurs JobChain acquire pattern
+			};
+
+			allow_pattern = std::any_of(allowed_patterns.begin(), allowed_patterns.end(), FN(pattern_hash == x));
 		}
 
+		if (allow_pattern)
+		{
 			add_pattern(false, inst_attr::putllc16, pattern.put_pc - result.entry_point, value.data);
+		}
 
-		spu_log.success("PUTLLC16 Pattern Detected! (mem_count=%d, put_pc=0x%x, pc_rel=%d, offset=0x%x, const=%u, two_regs=%d, reg=%u, runtime=%d, 0x%x-%s) (putllc0=%d, putllc16+0=%d, all=%d)"
+		spu_log.success("PUTLLC16 Pattern Detected! (mem_count=%d, put_pc=0x%x, pc_rel=%d, offset=0x%x, const=%u, two_regs=%d, reg=%u, runtime=%d, 0x%x-%s, pattern-hash=%s) (putllc0=%d, putllc16+0=%d, all=%d)"
-			, pattern.mem_count, pattern.put_pc, value.type == v_relative, value.off18, value.type == v_const, value.type == v_reg2, value.reg, value.runtime16_select, entry_point, func_hash, +stats.nowrite, ++stats.single, +stats.all);
+			, pattern.mem_count, pattern.put_pc, value.type == v_relative, value.off18, value.type == v_const, value.type == v_reg2, value.reg, value.runtime16_select, entry_point, func_hash, pattern_hash, +stats.nowrite, ++stats.single, +stats.all);
 	}
 
 	for (const auto& [read_pc, pattern] : rchcnt_loop_all)
@@ -7242,7 +7433,7 @@ spu_program spu_recompiler_base::analyse(const be_t<u32>* ls, u32 entry_point, s
 
 		if (inst_attr attr = m_inst_attrs[(read_pc - entry_point) / 4]; attr == inst_attr::none)
 		{
-			add_pattern(false, inst_attr::rchcnt_loop, read_pc - result.entry_point);
+			add_pattern(false, inst_attr::rchcnt_loop, read_pc - result.entry_point, 0);
 
 			spu_log.error("Channel Loop Pattern Detected! Report to developers! (read_pc=0x%x, branch_pc=0x%x, branch_target=0x%x, 0x%x-%s)", read_pc, pattern.branch_pc, pattern.branch_target, entry_point, func_hash);
 		}
@@ -7258,6 +7449,26 @@ spu_program spu_recompiler_base::analyse(const be_t<u32>* ls, u32 entry_point, s
 		// Blocks starting from 0x0 or invalid instruction won't be compiled, may need special interpreter fallback
 	}
 
+	if (!m_patterns.empty())
+	{
+		std::string out_dump;
+		dump(result, out_dump);
+		spu_log.notice("Dump SPU Function with pattern(s):\n%s", out_dump);
+	}
+
+	for (u32 i = 0; i < result.data.size(); i++)
+	{
+		const be_t<u32> ls_val = ls[result.lower_bound / 4 + i];
+
+		if (result.data[i] && std::bit_cast<u32>(ls_val) != result.data[i])
+		{
+			std::string out_dump;
+			dump(result, out_dump);
+			spu_log.error("SPU Function Dump:\n%s", out_dump);
+			fmt::throw_exception("SPU Analyzer failed: Instruction mismatch at 0x%x [read: 0x%x vs LS: 0x%x] (i=0x%x)", result.lower_bound + i * 4, std::bit_cast<be_t<u32>>(result.data[i]), ls_val, i);
+		}
+	}
+
 	return result;
 }
 
@@ -8267,8 +8478,9 @@ std::array<reg_state_t, s_reg_max>& block_reg_info::evaluate_start_state(const s
 				{
 					// TODO: The true maximum occurence count need to depend on the amount of branching-outs passed through
 					// Currently allow 2 for short-term code and 1 for long-term code
+					// Ignore large jumptables as well
 					const bool loop_terminator_detected = std::count(been_there.begin(), been_there.end(), prev_pc) >= (qi < 20 ? 2u : 1u);
-					const bool avoid_extensive_analysis = qi >= (extensive_evaluation ? 22 : 16);
+					const bool avoid_extensive_analysis = qi >= (extensive_evaluation ? 22 : 16) || it->state_prev.size() >= 8;
 
 					if (!loop_terminator_detected && !avoid_extensive_analysis)
 					{
@@ -8307,19 +8519,10 @@ std::array<reg_state_t, s_reg_max>& block_reg_info::evaluate_start_state(const s
 	return walkby_state;
 }
 
-void spu_recompiler_base::add_pattern(bool fill_all, inst_attr attr, u32 start, u32 end)
+void spu_recompiler_base::add_pattern(bool fill_all, inst_attr attr, u32 start, u64 info)
 {
-	if (end == umax)
+	m_patterns[start] = pattern_info{info};
-	{
+	m_inst_attrs[start / 4] = attr;
-		end = start;
-	}
-
-	m_patterns[start] = pattern_info{utils::address_range32::start_end(start, end)};
-
-	for (u32 i = start; i <= (fill_all ? end : start); i += 4)
-	{
-		m_inst_attrs[i / 4] = attr;
-	}
 }
 
 extern std::string format_spu_func_info(u32 addr, cpu_thread* spu)

rpcs3/Emu/Cell/SPUInterpreter.cpp

@@ -291,7 +291,7 @@ bool ROT(spu_thread& spu, spu_opcode_t op)
 
 	for (u32 i = 0; i < 4; i++)
 	{
-		spu.gpr[op.rt]._u32[i] = utils::rol32(a._u32[i], b._u32[i]);
+		spu.gpr[op.rt]._u32[i] = std::rotl<u32>(a._u32[i], b._u32[i]);
 	}
 	return true;
 }
@@ -346,7 +346,7 @@ bool ROTH(spu_thread& spu, spu_opcode_t op)
 
 	for (u32 i = 0; i < 8; i++)
 	{
-		spu.gpr[op.rt]._u16[i] = utils::rol16(a._u16[i], b._u16[i]);
+		spu.gpr[op.rt]._u16[i] = std::rotl<u16>(a._u16[i], b._u16[i]);
 	}
 	return true;
 }

rpcs3/Emu/Cell/SPULLVMRecompiler.cpp

@@ -1080,7 +1080,7 @@ class spu_llvm_recompiler : public spu_recompiler_base, public cpu_translator
 		m_ir->SetInsertPoint(_body);
 	}
 
-	void putllc16_pattern(const spu_program& /*prog*/, utils::address_range32 range)
+	void putllc16_pattern(const spu_program& /*prog*/, u64 pattern_info)
 	{
 		// Prevent store elimination
 		m_block->store_context_ctr[s_reg_mfc_eal]++;
@@ -1109,16 +1109,17 @@ class spu_llvm_recompiler : public spu_recompiler_base, public cpu_translator
 			}
 		};
 
-		const union putllc16_info
+		const union putllc16_or_0_info
 		{
-			u32 data;
+			u64 data;
-			bf_t<u32, 30, 2> type;
+			bf_t<u64, 32, 18> required_pc;
-			bf_t<u32, 29, 1> runtime16_select;
+			bf_t<u64, 30, 2> type;
-			bf_t<u32, 28, 1> no_notify;
+			bf_t<u64, 29, 1> runtime16_select;
-			bf_t<u32, 18, 8> reg;
+			bf_t<u64, 28, 1> no_notify;
-			bf_t<u32, 0, 18> off18;
+			bf_t<u64, 18, 8> reg;
-			bf_t<u32, 0, 8> reg2;
+			bf_t<u64, 0, 18> off18;
-		} info = std::bit_cast<putllc16_info>(range.end);
+			bf_t<u64, 0, 8> reg2;
+		} info = std::bit_cast<putllc16_or_0_info>(pattern_info);
 
 		enum : u32
 		{
@@ -1150,8 +1151,10 @@ class spu_llvm_recompiler : public spu_recompiler_base, public cpu_translator
 		value_t<u32> eal_val;
 		eal_val.value = _eal;
 
-		auto get_reg32 = [&](u32 reg)
+		auto get_reg32 = [&](u64 reg_)
 		{
+			const u32 reg = static_cast<u32>(reg_);
+
 			if (get_reg_type(reg) != get_type<u32[4]>())
 			{
 				return get_reg_fixed(reg, get_type<u32>());
@@ -1170,6 +1173,19 @@ class spu_llvm_recompiler : public spu_recompiler_base, public cpu_translator
 		}
 		else if (info.type == v_relative)
 		{
+			if (info.required_pc && info.required_pc != SPU_LS_SIZE)
+			{
+				const auto short_op = llvm::BasicBlock::Create(m_context, "__putllc16_short_op", m_function);
+				const auto heavy_op = llvm::BasicBlock::Create(m_context, "__putllc16_heavy_op", m_function);
+
+				m_ir->CreateCondBr(m_ir->CreateICmpNE(m_ir->getInt32(info.required_pc), m_base_pc), heavy_op, short_op);
+				m_ir->SetInsertPoint(heavy_op);
+				update_pc();
+				call("spu_exec_mfc_cmd", &exec_mfc_cmd<false>, m_thread);
+				m_ir->CreateBr(_final);
+				m_ir->SetInsertPoint(short_op);
+			}
+
 			dest = m_ir->CreateAnd(get_pc(spu_branch_target(info.off18 + m_base)), 0x3fff0);
 		}
 		else if (info.type == v_reg_offs)
@@ -1268,17 +1284,18 @@ class spu_llvm_recompiler : public spu_recompiler_base, public cpu_translator
 		const auto _new = m_ir->CreateAlignedLoad(get_type<u128>(), _ptr(m_lsptr, dest), llvm::MaybeAlign{16});
 		const auto _rdata = m_ir->CreateAlignedLoad(get_type<u128>(), _ptr(spu_ptr(&spu_thread::rdata), m_ir->CreateAnd(diff, 0x70)), llvm::MaybeAlign{16});
 
-		const bool is_accurate_op = !!g_cfg.core.spu_accurate_reservations;
+		const bool is_accurate_op = true || !!g_cfg.core.spu_accurate_reservations;
 
-		const auto compare_data_change_res = is_accurate_op ? m_ir->getTrue() : m_ir->CreateICmpNE(_new, _rdata);
+		const auto compare_data_change_res = m_ir->CreateICmpNE(_new, _rdata);
+		const auto second_test_for_complete_op = is_accurate_op ? m_ir->getTrue() : compare_data_change_res;
 
 		if (info.runtime16_select)
 		{
-			m_ir->CreateCondBr(m_ir->CreateAnd(m_ir->CreateICmpULT(diff, m_ir->getInt64(128)), compare_data_change_res), _begin_op, _inc_res, m_md_likely);
+			m_ir->CreateCondBr(m_ir->CreateAnd(m_ir->CreateICmpULT(diff, m_ir->getInt64(128)), second_test_for_complete_op), _begin_op, _inc_res, m_md_likely);
 		}
 		else
 		{
-			m_ir->CreateCondBr(compare_data_change_res, _begin_op, _inc_res, m_md_unlikely);
+			m_ir->CreateCondBr(second_test_for_complete_op, _begin_op, _inc_res, m_md_unlikely);
 		}
 
 		m_ir->SetInsertPoint(_begin_op);
@@ -1323,7 +1340,14 @@ class spu_llvm_recompiler : public spu_recompiler_base, public cpu_translator
 
 		if (!info.no_notify)
 		{
+			const auto notify_block = llvm::BasicBlock::Create(m_context, "__putllc16_block_notify", m_function);
+			const auto notify_next = llvm::BasicBlock::Create(m_context, "__putllc16_block_notify_next", m_function);
+
+			m_ir->CreateCondBr(compare_data_change_res, notify_block, notify_next);
+			m_ir->SetInsertPoint(notify_block);
 			call("atomic_wait_engine::notify_all", static_cast<void(*)(const void*)>(atomic_wait_engine::notify_all), rptr);
+			m_ir->CreateBr(notify_next);
+			m_ir->SetInsertPoint(notify_next);
 		}
 
 		m_ir->CreateBr(_success);
@@ -1373,7 +1397,7 @@ class spu_llvm_recompiler : public spu_recompiler_base, public cpu_translator
 		m_ir->SetInsertPoint(_final);
 	}
 
-	void putllc0_pattern(const spu_program& /*prog*/, utils::address_range32 /*range*/)
+	void putllc0_pattern(const spu_program& /*prog*/, u64 pattern_info)
 	{
 		// Prevent store elimination
 		m_block->store_context_ctr[s_reg_mfc_eal]++;
@@ -1401,6 +1425,18 @@ class spu_llvm_recompiler : public spu_recompiler_base, public cpu_translator
 			}
 		};
 
+		const union putllc16_or_0_info
+		{
+			u64 data;
+			bf_t<u64, 32, 18> required_pc;
+			bf_t<u64, 30, 2> type;
+			bf_t<u64, 29, 1> runtime16_select;
+			bf_t<u64, 28, 1> no_notify;
+			bf_t<u64, 18, 8> reg;
+			bf_t<u64, 0, 18> off18;
+			bf_t<u64, 0, 8> reg2;
+		} info = std::bit_cast<putllc16_or_0_info>(pattern_info);
+
 		const auto _next = llvm::BasicBlock::Create(m_context, "", m_function);
 		const auto _next0 = llvm::BasicBlock::Create(m_context, "", m_function);
 		const auto _fail = llvm::BasicBlock::Create(m_context, "", m_function);
@@ -1409,6 +1445,19 @@ class spu_llvm_recompiler : public spu_recompiler_base, public cpu_translator
 		const auto _eal = (get_reg_fixed<u32>(s_reg_mfc_eal) & -128).eval(m_ir);
 		const auto _raddr = m_ir->CreateLoad(get_type<u32>(), spu_ptr(&spu_thread::raddr));
 
+		if (info.required_pc && info.required_pc != SPU_LS_SIZE)
+		{
+			const auto short_op = llvm::BasicBlock::Create(m_context, "__putllc0_short_op", m_function);
+			const auto heavy_op = llvm::BasicBlock::Create(m_context, "__putllc0_heavy_op", m_function);
+
+			m_ir->CreateCondBr(m_ir->CreateICmpNE(m_ir->getInt32(info.required_pc), m_base_pc), heavy_op, short_op);
+			m_ir->SetInsertPoint(heavy_op);
+			update_pc();
+			call("spu_exec_mfc_cmd", &exec_mfc_cmd<false>, m_thread);
+			m_ir->CreateBr(_final);
+			m_ir->SetInsertPoint(short_op);
+		}
+
 		m_ir->CreateCondBr(m_ir->CreateAnd(m_ir->CreateICmpEQ(_eal, _raddr), m_ir->CreateIsNotNull(_raddr)), _next, _fail, m_md_likely);
 		m_ir->SetInsertPoint(_next);
 
@@ -1598,7 +1647,11 @@ public:
 
 		// Create LLVM module
 		std::unique_ptr<Module> _module = std::make_unique<Module>(m_hash + ".obj", m_context);
+#if LLVM_VERSION_MAJOR >= 21 && (LLVM_VERSION_MINOR >= 1 || LLVM_VERSION_MAJOR >= 22)
+		_module->setTargetTriple(Triple(jit_compiler::triple2()));
+#else
 		_module->setTargetTriple(jit_compiler::triple2());
+#endif
 		_module->setDataLayout(m_jit.get_engine().getTargetMachine()->createDataLayout());
 		m_module = _module.get();
 
@@ -1632,7 +1685,7 @@ public:
 		m_ir->SetInsertPoint(label_test);
 
 		// Set block hash for profiling (if enabled)
-		if (g_cfg.core.spu_prof && g_cfg.core.spu_verification)
+		if ((g_cfg.core.spu_prof || g_cfg.core.spu_debug) && g_cfg.core.spu_verification)
 			m_ir->CreateStore(m_ir->getInt64((m_hash_start & -65536)), spu_ptr(&spu_thread::block_hash));
 
 		if (!g_cfg.core.spu_verification)
@@ -1989,7 +2042,7 @@ public:
 			set_function(m_functions[m_entry].chunk);
 
 			// Set block hash for profiling (if enabled)
-			if (g_cfg.core.spu_prof)
+			if (g_cfg.core.spu_prof || g_cfg.core.spu_debug)
 				m_ir->CreateStore(m_ir->getInt64((m_hash_start & -65536) | (m_entry >> 2)), spu_ptr(&spu_thread::block_hash));
 
 			m_finfo = &m_functions[m_entry];
@@ -2139,12 +2192,12 @@ public:
 					{
 					case inst_attr::putllc0:
 					{
-						putllc0_pattern(func, m_patterns.at(m_pos - start).range);
+						putllc0_pattern(func, m_patterns.at(m_pos - start).info);
 						continue;
 					}
 					case inst_attr::putllc16:
 					{
-						putllc16_pattern(func, m_patterns.at(m_pos - start).range);
+						putllc16_pattern(func, m_patterns.at(m_pos - start).info);
 						continue;
 					}
 					case inst_attr::omit:
@@ -2227,6 +2280,12 @@ public:
 					{
 						for (auto& [a, b] : m_blocks)
 						{
+							if (has_gpr_memory_barriers)
+							{
+								// Dive deeper and inspect GPR store barriers
+								break;
+							}
+
 							// Check if the store occurs before any barrier in the block
 							if (b.store[i] && b.store[i] != bs && b.store_context_first_id[i] == 1)
 							{
@@ -2873,7 +2932,11 @@ public:
 
 		// Create LLVM module
 		std::unique_ptr<Module> _module = std::make_unique<Module>("spu_interpreter.obj", m_context);
+#if LLVM_VERSION_MAJOR >= 21 && (LLVM_VERSION_MINOR >= 1 || LLVM_VERSION_MAJOR >= 22)
+		_module->setTargetTriple(Triple(jit_compiler::triple2()));
+#else
 		_module->setTargetTriple(jit_compiler::triple2());
+#endif
 		_module->setDataLayout(m_jit.get_engine().getTargetMachine()->createDataLayout());
 		m_module = _module.get();
 
@@ -3977,7 +4040,7 @@ public:
 
 				bool must_use_cpp_functions = !!g_cfg.core.spu_accurate_dma;
 
-				if (u64 cmdh = ci->getZExtValue() & ~(MFC_BARRIER_MASK | MFC_FENCE_MASK | MFC_RESULT_MASK); g_cfg.core.rsx_fifo_accuracy || g_cfg.video.strict_rendering_mode || !g_use_rtm)
+				if (u64 cmdh = ci->getZExtValue() & ~(MFC_BARRIER_MASK | MFC_FENCE_MASK | MFC_RESULT_MASK); g_cfg.core.rsx_fifo_accuracy || g_cfg.video.strict_rendering_mode || /*!g_use_rtm*/ true)
 				{
 					// TODO: don't require TSX (current implementation is TSX-only)
 					if (cmdh == MFC_PUT_CMD || cmdh == MFC_SNDSIG_CMD)

rpcs3/Emu/Cell/SPURecompiler.h

@@ -397,12 +397,12 @@ protected:
 
 	struct pattern_info
 	{
-		utils::address_range32 range;
+		u64 info;
 	};
 
 	std::unordered_map<u32, pattern_info> m_patterns;
 
-	void add_pattern(bool fill_all, inst_attr attr, u32 start, u32 end = -1);
+	void add_pattern(bool fill_all, inst_attr attr, u32 start, u64 info);
 
 private:
 	// For private use

rpcs3/Emu/Cell/SPUThread.h

@@ -901,7 +901,8 @@ public:
 
 	// Returns true if reservation existed but was just discovered to be lost
 	// It is safe to use on any address, even if not directly accessed by SPU (so it's slower)
-	bool reservation_check(u32 addr, const decltype(rdata)& data) const;
+	// Optionally pass a known allocated address for internal optimization (the current Effective-Address of the MFC command)
+	bool reservation_check(u32 addr, const decltype(rdata)& data, u32 current_eal = 0) const;
 	static bool reservation_check(u32 addr, u32 hash, atomic_t<u64, 64>* range_lock);
 	usz register_cache_line_waiter(u32 addr);
 	void deregister_cache_line_waiter(usz index);

rpcs3/Emu/Cell/lv2/sys_fs.cpp

@@ -1036,7 +1036,6 @@ lv2_file::open_result_t lv2_file::open(std::string_view vpath, s32 flags, s32 mo
 error_code sys_fs_open(ppu_thread& ppu, vm::cptr<char> path, s32 flags, vm::ptr<u32> fd, s32 mode, vm::cptr<void> arg, u64 size)
 {
 	ppu.state += cpu_flag::wait;
-	lv2_obj::sleep(ppu);
 
 	sys_fs.warning("sys_fs_open(path=%s, flags=%#o, fd=*0x%x, mode=%#o, arg=*0x%x, size=0x%llx)", path, flags, fd, mode, arg, size);
 
@@ -1085,7 +1084,6 @@ error_code sys_fs_open(ppu_thread& ppu, vm::cptr<char> path, s32 flags, vm::ptr<
 error_code sys_fs_read(ppu_thread& ppu, u32 fd, vm::ptr<void> buf, u64 nbytes, vm::ptr<u64> nread)
 {
 	ppu.state += cpu_flag::wait;
-	lv2_obj::sleep(ppu);
 
 	sys_fs.trace("sys_fs_read(fd=%d, buf=*0x%x, nbytes=0x%llx, nread=*0x%x)", fd, buf, nbytes, nread);
 
@@ -1122,6 +1120,11 @@ error_code sys_fs_read(ppu_thread& ppu, u32 fd, vm::ptr<void> buf, u64 nbytes, v
 		return CELL_OK;
 	}
 
+	if (nbytes >= 0x100000 && file->type != lv2_file_type::regular)
+	{
+		lv2_obj::sleep(ppu);
+	}
+
 	std::unique_lock lock(file->mp->mutex);
 
 	if (!file->file)
@@ -1154,7 +1157,6 @@ error_code sys_fs_read(ppu_thread& ppu, u32 fd, vm::ptr<void> buf, u64 nbytes, v
 error_code sys_fs_write(ppu_thread& ppu, u32 fd, vm::cptr<void> buf, u64 nbytes, vm::ptr<u64> nwrite)
 {
 	ppu.state += cpu_flag::wait;
-	lv2_obj::sleep(ppu);
 
 	sys_fs.trace("sys_fs_write(fd=%d, buf=*0x%x, nbytes=0x%llx, nwrite=*0x%x)", fd, buf, nbytes, nwrite);
 
@@ -1237,7 +1239,6 @@ error_code sys_fs_write(ppu_thread& ppu, u32 fd, vm::cptr<void> buf, u64 nbytes,
 error_code sys_fs_close(ppu_thread& ppu, u32 fd)
 {
 	ppu.state += cpu_flag::wait;
-	lv2_obj::sleep(ppu);
 
 	const auto file = idm::get_unlocked<lv2_fs_object, lv2_file>(fd);
 
@@ -1314,7 +1315,6 @@ error_code sys_fs_close(ppu_thread& ppu, u32 fd)
 error_code sys_fs_opendir(ppu_thread& ppu, vm::cptr<char> path, vm::ptr<u32> fd)
 {
 	ppu.state += cpu_flag::wait;
-	lv2_obj::sleep(ppu);
 
 	sys_fs.warning("sys_fs_opendir(path=%s, fd=*0x%x)", path, fd);
 
@@ -1491,7 +1491,6 @@ error_code sys_fs_readdir(ppu_thread& ppu, u32 fd, vm::ptr<CellFsDirent> dir, vm
 error_code sys_fs_closedir(ppu_thread& ppu, u32 fd)
 {
 	ppu.state += cpu_flag::wait;
-	lv2_obj::sleep(ppu);
 
 	sys_fs.warning("sys_fs_closedir(fd=%d)", fd);
 
@@ -1506,7 +1505,6 @@ error_code sys_fs_closedir(ppu_thread& ppu, u32 fd)
 error_code sys_fs_stat(ppu_thread& ppu, vm::cptr<char> path, vm::ptr<CellFsStat> sb)
 {
 	ppu.state += cpu_flag::wait;
-	lv2_obj::sleep(ppu);
 
 	sys_fs.warning("sys_fs_stat(path=%s, sb=*0x%x)", path, sb);
 
@@ -1610,7 +1608,6 @@ error_code sys_fs_stat(ppu_thread& ppu, vm::cptr<char> path, vm::ptr<CellFsStat>
 error_code sys_fs_fstat(ppu_thread& ppu, u32 fd, vm::ptr<CellFsStat> sb)
 {
 	ppu.state += cpu_flag::wait;
-	lv2_obj::sleep(ppu);
 
 	sys_fs.warning("sys_fs_fstat(fd=%d, sb=*0x%x)", fd, sb);
 
@@ -1666,7 +1663,6 @@ error_code sys_fs_link(ppu_thread&, vm::cptr<char> from, vm::cptr<char> to)
 error_code sys_fs_mkdir(ppu_thread& ppu, vm::cptr<char> path, s32 mode)
 {
 	ppu.state += cpu_flag::wait;
-	lv2_obj::sleep(ppu);
 
 	sys_fs.warning("sys_fs_mkdir(path=%s, mode=%#o)", path, mode);
 
@@ -1728,7 +1724,6 @@ error_code sys_fs_mkdir(ppu_thread& ppu, vm::cptr<char> path, s32 mode)
 error_code sys_fs_rename(ppu_thread& ppu, vm::cptr<char> from, vm::cptr<char> to)
 {
 	ppu.state += cpu_flag::wait;
-	lv2_obj::sleep(ppu);
 
 	sys_fs.warning("sys_fs_rename(from=%s, to=%s)", from, to);
 
@@ -1794,7 +1789,6 @@ error_code sys_fs_rename(ppu_thread& ppu, vm::cptr<char> from, vm::cptr<char> to
 error_code sys_fs_rmdir(ppu_thread& ppu, vm::cptr<char> path)
 {
 	ppu.state += cpu_flag::wait;
-	lv2_obj::sleep(ppu);
 
 	sys_fs.warning("sys_fs_rmdir(path=%s)", path);
 
@@ -1850,7 +1844,6 @@ error_code sys_fs_rmdir(ppu_thread& ppu, vm::cptr<char> path)
 error_code sys_fs_unlink(ppu_thread& ppu, vm::cptr<char> path)
 {
 	ppu.state += cpu_flag::wait;
-	lv2_obj::sleep(ppu);
 
 	sys_fs.warning("sys_fs_unlink(path=%s)", path);
 
@@ -1951,8 +1944,6 @@ error_code sys_fs_fcntl(ppu_thread& ppu, u32 fd, u32 op, vm::ptr<void> _arg, u32
 	case 0x8000000a: // cellFsReadWithOffset
 	case 0x8000000b: // cellFsWriteWithOffset
 	{
-		lv2_obj::sleep(ppu);
-
 		const auto arg = vm::static_ptr_cast<lv2_file_op_rw>(_arg);
 
 		if (_size < arg.size())
@@ -1992,6 +1983,11 @@ error_code sys_fs_fcntl(ppu_thread& ppu, u32 fd, u32 op, vm::ptr<void> _arg, u32
 			sys_fs.error("%s type: Writing %u bytes to FD=%d (path=%s)", file->type, arg->size, file->name.data());
 		}
 
+		if (op == 0x8000000a && file->type != lv2_file_type::regular && arg->size >= 0x100000)
+		{
+			lv2_obj::sleep(ppu);
+		}
+
 		std::unique_lock wlock(file->mp->mutex, std::defer_lock);
 		std::shared_lock rlock(file->mp->mutex, std::defer_lock);
 
@@ -2047,8 +2043,6 @@ error_code sys_fs_fcntl(ppu_thread& ppu, u32 fd, u32 op, vm::ptr<void> _arg, u32
 
 	case 0x80000009: // cellFsSdataOpenByFd
 	{
-		lv2_obj::sleep(ppu);
-
 		const auto arg = vm::static_ptr_cast<lv2_file_op_09>(_arg);
 
 		if (_size < arg.size())
@@ -2102,8 +2096,6 @@ error_code sys_fs_fcntl(ppu_thread& ppu, u32 fd, u32 op, vm::ptr<void> _arg, u32
 
 	case 0xc0000002: // cellFsGetFreeSize (TODO)
 	{
-		lv2_obj::sleep(ppu);
-
 		const auto arg = vm::static_ptr_cast<lv2_file_c0000002>(_arg);
 
 		const auto& mp = g_fxo->get<lv2_fs_mount_info_map>().lookup("/dev_hdd0");
@@ -2418,8 +2410,6 @@ error_code sys_fs_fcntl(ppu_thread& ppu, u32 fd, u32 op, vm::ptr<void> _arg, u32
 
 	case 0xe0000012: // cellFsGetDirectoryEntries
 	{
-		lv2_obj::sleep(ppu);
-
 		const auto arg = vm::static_ptr_cast<lv2_file_op_dir::dir_info>(_arg);
 
 		if (_size < arg.size())
@@ -2434,8 +2424,6 @@ error_code sys_fs_fcntl(ppu_thread& ppu, u32 fd, u32 op, vm::ptr<void> _arg, u32
 			return CELL_EBADF;
 		}
 
-		ppu.check_state();
-
 		u32 read_count = 0;
 
 		// NOTE: This function is actually capable of reading only one entry at a time
@@ -2593,7 +2581,6 @@ error_code sys_fs_fcntl(ppu_thread& ppu, u32 fd, u32 op, vm::ptr<void> _arg, u32
 error_code sys_fs_lseek(ppu_thread& ppu, u32 fd, s64 offset, s32 whence, vm::ptr<u64> pos)
 {
 	ppu.state += cpu_flag::wait;
-	lv2_obj::sleep(ppu);
 
 	sys_fs.trace("sys_fs_lseek(fd=%d, offset=0x%llx, whence=0x%x, pos=*0x%x)", fd, offset, whence, pos);
 
@@ -2639,7 +2626,6 @@ error_code sys_fs_lseek(ppu_thread& ppu, u32 fd, s64 offset, s32 whence, vm::ptr
 error_code sys_fs_fdatasync(ppu_thread& ppu, u32 fd)
 {
 	ppu.state += cpu_flag::wait;
-	lv2_obj::sleep(ppu);
 
 	sys_fs.trace("sys_fs_fdadasync(fd=%d)", fd);
 
@@ -2650,6 +2636,8 @@ error_code sys_fs_fdatasync(ppu_thread& ppu, u32 fd)
 		return CELL_EBADF;
 	}
 
+	lv2_obj::sleep(ppu);
+
 	std::lock_guard lock(file->mp->mutex);
 
 	if (!file->file)
@@ -2664,7 +2652,6 @@ error_code sys_fs_fdatasync(ppu_thread& ppu, u32 fd)
 error_code sys_fs_fsync(ppu_thread& ppu, u32 fd)
 {
 	ppu.state += cpu_flag::wait;
-	lv2_obj::sleep(ppu);
 
 	sys_fs.trace("sys_fs_fsync(fd=%d)", fd);
 
@@ -2675,6 +2662,8 @@ error_code sys_fs_fsync(ppu_thread& ppu, u32 fd)
 		return CELL_EBADF;
 	}
 
+	lv2_obj::sleep(ppu);
+
 	std::lock_guard lock(file->mp->mutex);
 
 	if (!file->file)
@@ -2763,7 +2752,6 @@ error_code sys_fs_get_block_size(ppu_thread& ppu, vm::cptr<char> path, vm::ptr<u
 error_code sys_fs_truncate(ppu_thread& ppu, vm::cptr<char> path, u64 size)
 {
 	ppu.state += cpu_flag::wait;
-	lv2_obj::sleep(ppu);
 
 	sys_fs.warning("sys_fs_truncate(path=%s, size=0x%llx)", path, size);
 
@@ -2815,7 +2803,6 @@ error_code sys_fs_truncate(ppu_thread& ppu, vm::cptr<char> path, u64 size)
 error_code sys_fs_ftruncate(ppu_thread& ppu, u32 fd, u64 size)
 {
 	ppu.state += cpu_flag::wait;
-	lv2_obj::sleep(ppu);
 
 	sys_fs.warning("sys_fs_ftruncate(fd=%d, size=0x%llx)", fd, size);
 
@@ -3021,7 +3008,6 @@ error_code sys_fs_disk_free(ppu_thread& ppu, vm::cptr<char> path, vm::ptr<u64> t
 error_code sys_fs_utime(ppu_thread& ppu, vm::cptr<char> path, vm::cptr<CellFsUtimbuf> timep)
 {
 	ppu.state += cpu_flag::wait;
-	lv2_obj::sleep(ppu);
 
 	sys_fs.warning("sys_fs_utime(path=%s, timep=*0x%x)", path, timep);
 	sys_fs.warning("** actime=%u, modtime=%u", timep->actime, timep->modtime);

rpcs3/Emu/Cell/lv2/sys_memory.cpp

@@ -5,6 +5,7 @@
 #include "Emu/CPU/CPUThread.h"
 #include "Emu/Cell/ErrorCodes.h"
 #include "Emu/Cell/SPUThread.h"
+#include "Emu/Cell/PPUThread.h"
 #include "Emu/IdManager.h"
 
 #include "util/asm.hpp"
@@ -213,6 +214,8 @@ error_code sys_memory_allocate_from_container(cpu_thread& cpu, u64 size, u32 cid
 
 			if (alloc_addr)
 			{
+				sys_memory.notice("sys_memory_allocate_from_container(): Allocated 0x%x address (size=0x%x)", addr, size);
+
 				vm::lock_sudo(addr, static_cast<u32>(size));
 				cpu.check_state();
 				*alloc_addr = addr;
@@ -247,17 +250,37 @@ error_code sys_memory_free(cpu_thread& cpu, u32 addr)
 	return CELL_OK;
 }
 
-error_code sys_memory_get_page_attribute(cpu_thread& cpu, u32 addr, vm::ptr<sys_page_attr_t> attr)
+error_code sys_memory_get_page_attribute(ppu_thread& ppu, u32 addr, vm::ptr<sys_page_attr_t> attr)
 {
-	cpu.state += cpu_flag::wait;
+	ppu.state += cpu_flag::wait;
 
 	sys_memory.trace("sys_memory_get_page_attribute(addr=0x%x, attr=*0x%x)", addr, attr);
 
-	vm::writer_lock rlock;
+	if ((addr >> 28) == (ppu.stack_addr >> 28))
-
-	if (!vm::check_addr(addr) || addr >= SPU_FAKE_BASE_ADDR)
 	{
-		return CELL_EINVAL;
+		// Stack address: fast path
+		if (!(addr >= ppu.stack_addr && addr < ppu.stack_addr + ppu.stack_size) && !vm::check_addr(addr))
+		{
+			return { CELL_EINVAL, addr };
+		}
+
+		if (!vm::check_addr(attr.addr(), vm::page_readable, attr.size()))
+		{
+			return CELL_EFAULT;
+		}
+
+		attr->attribute = 0x40000ull; // SYS_MEMORY_PROT_READ_WRITE
+		attr->access_right = SYS_MEMORY_ACCESS_RIGHT_PPU_THR;
+		attr->page_size = 4096;
+		attr->pad = 0; // Always write 0
+		return CELL_OK;
+	}
+
+	const auto [ok, vm_flags] = vm::get_addr_flags(addr);
+
+	if (!ok || addr >= SPU_FAKE_BASE_ADDR)
+	{
+		return { CELL_EINVAL, addr };
 	}
 
 	if (!vm::check_addr(attr.addr(), vm::page_readable, attr.size()))
@@ -266,19 +289,20 @@ error_code sys_memory_get_page_attribute(cpu_thread& cpu, u32 addr, vm::ptr<sys_
 	}
 
 	attr->attribute = 0x40000ull; // SYS_MEMORY_PROT_READ_WRITE (TODO)
-	attr->access_right = addr >> 28 == 0xdu ? SYS_MEMORY_ACCESS_RIGHT_PPU_THR : SYS_MEMORY_ACCESS_RIGHT_ANY;// (TODO)
+	attr->access_right = SYS_MEMORY_ACCESS_RIGHT_ANY; // TODO: Report accurately
 
-	if (vm::check_addr(addr, vm::page_1m_size))
+	if (vm_flags & vm::page_1m_size)
 	{
 		attr->page_size = 0x100000;
 	}
-	else if (vm::check_addr(addr, vm::page_64k_size))
+	else if (vm_flags & vm::page_64k_size)
 	{
 		attr->page_size = 0x10000;
 	}
 	else
 	{
-		attr->page_size = 4096;
+		//attr->page_size = 4096;
+		fmt::throw_exception("Unreachable");
 	}
 
 	attr->pad = 0; // Always write 0

rpcs3/Emu/Cell/lv2/sys_memory.h

@@ -4,6 +4,7 @@
 #include "Emu/Cell/ErrorCodes.h"
 
 class cpu_thread;
+class ppu_thread;
 
 enum lv2_mem_container_id : u32
 {
@@ -131,7 +132,7 @@ struct sys_memory_user_memory_stat_t
 error_code sys_memory_allocate(cpu_thread& cpu, u64 size, u64 flags, vm::ptr<u32> alloc_addr);
 error_code sys_memory_allocate_from_container(cpu_thread& cpu, u64 size, u32 cid, u64 flags, vm::ptr<u32> alloc_addr);
 error_code sys_memory_free(cpu_thread& cpu, u32 start_addr);
-error_code sys_memory_get_page_attribute(cpu_thread& cpu, u32 addr, vm::ptr<sys_page_attr_t> attr);
+error_code sys_memory_get_page_attribute(ppu_thread& cpu, u32 addr, vm::ptr<sys_page_attr_t> attr);
 error_code sys_memory_get_user_memory_size(cpu_thread& cpu, vm::ptr<sys_memory_info_t> mem_info);
 error_code sys_memory_get_user_memory_stat(cpu_thread& cpu, vm::ptr<sys_memory_user_memory_stat_t> mem_stat);
 error_code sys_memory_container_create(cpu_thread& cpu, vm::ptr<u32> cid, u64 size);

rpcs3/Emu/Cell/lv2/sys_ppu_thread.cpp

@@ -468,11 +468,11 @@ error_code sys_ppu_thread_restart(ppu_thread& ppu)
 	return CELL_OK;
 }
 
-error_code _sys_ppu_thread_create(ppu_thread& ppu, vm::ptr<u64> thread_id, vm::ptr<ppu_thread_param_t> param, u64 arg, u64 unk, s32 prio, u32 _stacksz, u64 flags, vm::cptr<char> threadname)
+error_code _sys_ppu_thread_create(ppu_thread& ppu, vm::ptr<u64> thread_id, vm::ptr<ppu_thread_param_t> param, u64 arg, u64 unk, s32 prio, u64 _stacksz, u64 flags, vm::cptr<char> threadname)
 {
 	ppu.state += cpu_flag::wait;
 
-	sys_ppu_thread.warning("_sys_ppu_thread_create(thread_id=*0x%x, param=*0x%x, arg=0x%llx, unk=0x%llx, prio=%d, stacksize=0x%x, flags=0x%llx, threadname=*0x%x)",
+	sys_ppu_thread.warning("_sys_ppu_thread_create(thread_id=*0x%x, param=*0x%x, arg=0x%llx, unk=0x%llx, prio=%d, stacksize=0x%llx, flags=0x%llx, threadname=*0x%x)",
 		thread_id, param, arg, unk, prio, _stacksz, flags, threadname);
 
 	// thread_id is checked for null in stub -> CELL_ENOMEM
@@ -497,7 +497,8 @@ error_code _sys_ppu_thread_create(ppu_thread& ppu, vm::ptr<u64> thread_id, vm::p
 	const u32 tls = param->tls;
 
 	// Compute actual stack size and allocate
-	const u32 stack_size = utils::align<u32>(std::max<u32>(_stacksz, 4096), 4096);
+	// 0 and UINT64_MAX both convert to 4096
+	const u64 stack_size = FN(x ? x : 4096)(utils::align<u64>(_stacksz, 4096));
 
 	auto& dct = g_fxo->get<lv2_memory_container>();
 
@@ -507,7 +508,7 @@ error_code _sys_ppu_thread_create(ppu_thread& ppu, vm::ptr<u64> thread_id, vm::p
 		return {CELL_ENOMEM, dct.size - dct.used};
 	}
 
-	const vm::addr_t stack_base{vm::alloc(stack_size, vm::stack, 4096)};
+	const vm::addr_t stack_base{vm::alloc(static_cast<u32>(stack_size), vm::stack, 4096)};
 
 	if (!stack_base)
 	{
@@ -532,7 +533,7 @@ error_code _sys_ppu_thread_create(ppu_thread& ppu, vm::ptr<u64> thread_id, vm::p
 	{
 		ppu_thread_params p;
 		p.stack_addr = stack_base;
-		p.stack_size = stack_size;
+		p.stack_size = static_cast<u32>(stack_size);
 		p.tls_addr = tls;
 		p.entry = entry;
 		p.arg0 = arg;

rpcs3/Emu/Cell/lv2/sys_ppu_thread.h

@@ -53,7 +53,7 @@ error_code sys_ppu_thread_get_priority(ppu_thread& ppu, u32 thread_id, vm::ptr<s
 error_code sys_ppu_thread_get_stack_information(ppu_thread& ppu, vm::ptr<sys_ppu_thread_stack_t> sp);
 error_code sys_ppu_thread_stop(ppu_thread& ppu, u32 thread_id);
 error_code sys_ppu_thread_restart(ppu_thread& ppu);
-error_code _sys_ppu_thread_create(ppu_thread& ppu, vm::ptr<u64> thread_id, vm::ptr<ppu_thread_param_t> param, u64 arg, u64 arg4, s32 prio, u32 stacksize, u64 flags, vm::cptr<char> threadname);
+error_code _sys_ppu_thread_create(ppu_thread& ppu, vm::ptr<u64> thread_id, vm::ptr<ppu_thread_param_t> param, u64 arg, u64 arg4, s32 prio, u64 stacksize, u64 flags, vm::cptr<char> threadname);
 error_code sys_ppu_thread_start(ppu_thread& ppu, u32 thread_id);
 error_code sys_ppu_thread_rename(ppu_thread& ppu, u32 thread_id, vm::cptr<char> name);
 error_code sys_ppu_thread_recover_page_fault(ppu_thread& ppu, u32 thread_id);

rpcs3/Emu/Cell/lv2/sys_spu.cpp

@@ -767,7 +767,12 @@ error_code sys_spu_thread_initialize(ppu_thread& ppu, vm::ptr<u32> thread, u32 g
 	}
 
 	// Read thread name
-	const std::string thread_name(attr_data.name.get_ptr(), std::max<u32>(attr_data.name_len, 1) - 1);
+	std::string thread_name;
+
+	if (attr_data.name_len && !vm::read_string(attr_data.name.addr(), attr_data.name_len - 1, thread_name, true))
+	{
+		return { CELL_EFAULT, attr_data.name.addr() };
+	}
 
 	const auto group = idm::get_unlocked<lv2_spu_group>(group_id);
 
@@ -906,7 +911,7 @@ error_code sys_spu_thread_get_exit_status(ppu_thread& ppu, u32 id, vm::ptr<s32>
 	return CELL_ESTAT;
 }
 
-error_code sys_spu_thread_group_create(ppu_thread& ppu, vm::ptr<u32> id, u32 num, s32 prio, vm::ptr<sys_spu_thread_group_attribute> attr)
+error_code sys_spu_thread_group_create(ppu_thread& ppu, vm::ptr<u32> id, u32 num, s32 prio, vm::ptr<reduced_sys_spu_thread_group_attribute> attr)
 {
 	ppu.state += cpu_flag::wait;
 
@@ -914,13 +919,32 @@ error_code sys_spu_thread_group_create(ppu_thread& ppu, vm::ptr<u32> id, u32 num
 
 	const s32 min_prio = g_ps3_process_info.has_root_perm() ? 0 : 16;
 
-	const sys_spu_thread_group_attribute attr_data = *attr;
+	sys_spu_thread_group_attribute attr_data{};
+	{
+		const reduced_sys_spu_thread_group_attribute attr_reduced = *attr;
+		attr_data.name = attr_reduced.name;
+		attr_data.nsize = attr_reduced.nsize;
+		attr_data.type = attr_reduced.type;
+
+		// Read container-id member at offset 12 bytes conditionally (that's what LV2 does)
+		if (attr_data.type & SYS_SPU_THREAD_GROUP_TYPE_MEMORY_FROM_CONTAINER)
+		{
+			attr_data.ct = vm::unsafe_ptr_cast<sys_spu_thread_group_attribute>(attr)->ct;
+		}
+	}
 
 	if (attr_data.nsize > 0x80 || !num)
 	{
 		return CELL_EINVAL;
 	}
 
+	std::string group_name;
+
+	if (attr_data.nsize && !vm::read_string(attr_data.name.addr(), attr_data.nsize - 1, group_name, true))
+	{
+		return { CELL_EFAULT, attr_data.name.addr() };
+	}
+
 	const s32 type = attr_data.type;
 
 	bool use_scheduler = true;
@@ -1075,7 +1099,7 @@ error_code sys_spu_thread_group_create(ppu_thread& ppu, vm::ptr<u32> id, u32 num
 		return CELL_EBUSY;
 	}
 
-	const auto group = idm::make_ptr<lv2_spu_group>(std::string(attr_data.name.get_ptr(), std::max<u32>(attr_data.nsize, 1) - 1), num, prio, type, ct, use_scheduler, mem_size);
+	const auto group = idm::make_ptr<lv2_spu_group>(std::move(group_name), num, prio, type, ct, use_scheduler, mem_size);
 
 	if (!group)
 	{

rpcs3/Emu/Cell/lv2/sys_spu.h

@@ -82,11 +82,18 @@ enum spu_stop_syscall : u32
 	SYS_SPU_THREAD_STOP_SWITCH_SYSTEM_MODULE = 0x0120,
 };
 
-struct sys_spu_thread_group_attribute
+struct reduced_sys_spu_thread_group_attribute
 {
 	be_t<u32> nsize; // name length including NULL terminator
 	vm::bcptr<char> name;
 	be_t<s32> type;
+};
+
+struct sys_spu_thread_group_attribute
+{
+	be_t<u32> nsize;
+	vm::bcptr<char> name;
+	be_t<s32> type;
 	be_t<u32> ct; // memory container id
 };
 
@@ -360,7 +367,7 @@ error_code _sys_spu_image_close(ppu_thread&, vm::ptr<sys_spu_image> img);
 error_code _sys_spu_image_get_segments(ppu_thread&, vm::ptr<sys_spu_image> img, vm::ptr<sys_spu_segment> segments, s32 nseg);
 error_code sys_spu_thread_initialize(ppu_thread&, vm::ptr<u32> thread, u32 group, u32 spu_num, vm::ptr<sys_spu_image>, vm::ptr<sys_spu_thread_attribute>, vm::ptr<sys_spu_thread_argument>);
 error_code sys_spu_thread_set_argument(ppu_thread&, u32 id, vm::ptr<sys_spu_thread_argument> arg);
-error_code sys_spu_thread_group_create(ppu_thread&, vm::ptr<u32> id, u32 num, s32 prio, vm::ptr<sys_spu_thread_group_attribute> attr);
+error_code sys_spu_thread_group_create(ppu_thread&, vm::ptr<u32> id, u32 num, s32 prio, vm::ptr<reduced_sys_spu_thread_group_attribute> attr);
 error_code sys_spu_thread_group_destroy(ppu_thread&, u32 id);
 error_code sys_spu_thread_group_start(ppu_thread&, u32 id);
 error_code sys_spu_thread_group_suspend(ppu_thread&, u32 id);

rpcs3/Emu/Cell/lv2/sys_ss.cpp

@@ -210,8 +210,10 @@ error_code sys_ss_get_open_psid(vm::ptr<CellSsOpenPSID> psid)
 {
 	sys_ss.notice("sys_ss_get_open_psid(psid=*0x%x)", psid);
 
-	psid->high = g_cfg.sys.console_psid_high;
+	const u128 configured_psid = g_cfg.sys.console_psid.get();
-	psid->low = g_cfg.sys.console_psid_low;
+
+	psid->high = static_cast<u64>(configured_psid >> 64);
+	psid->low = static_cast<u64>(configured_psid);
 
 	return CELL_OK;
 }
@@ -259,8 +261,8 @@ error_code sys_ss_appliance_info_manager(u32 code, vm::ptr<u8> buffer)
 	case 0x19005:
 	{
 		// AIM_get_open_ps_id
-		be_t<u64> psid[2] = { +g_cfg.sys.console_psid_high, +g_cfg.sys.console_psid_low };
+		const be_t<u128> psid = g_cfg.sys.console_psid.get();
-		std::memcpy(buffer.get_ptr(), psid, 16);
+		std::memcpy(buffer.get_ptr(), &psid, 16);
 		break;
 	}
 	case 0x19006:
@@ -268,7 +270,11 @@ error_code sys_ss_appliance_info_manager(u32 code, vm::ptr<u8> buffer)
 		// qa values (dex only) ??
 		[[fallthrough]];
 	}
-	default: sys_ss.todo("sys_ss_appliance_info_manager(code=0x%x, buffer=*0x%x)", code, buffer);
+	default:
+	{
+		sys_ss.todo("sys_ss_appliance_info_manager(code=0x%x, buffer=*0x%x)", code, buffer);
+		break;
+	}
 	}
 
 	return CELL_OK;

rpcs3/Emu/Cell/lv2/sys_usbd.cpp

@@ -18,6 +18,7 @@
 #include "Emu/Io/Skylander.h"
 #include "Emu/Io/Infinity.h"
 #include "Emu/Io/Dimensions.h"
+#include "Emu/Io/KamenRider.h"
 #include "Emu/Io/GHLtar.h"
 #include "Emu/Io/ghltar_config.h"
 #include "Emu/Io/guncon3_config.h"
@@ -175,7 +176,7 @@ private:
 		{0x1430, 0x0150, 0x0150, "Skylanders Portal", &usb_device_skylander::get_num_emu_devices, &usb_device_skylander::make_instance},
 		{0x0E6F, 0x0129, 0x0129, "Disney Infinity Base", &usb_device_infinity::get_num_emu_devices, &usb_device_infinity::make_instance},
 		{0x0E6F, 0x0241, 0x0241, "Lego Dimensions Portal", &usb_device_dimensions::get_num_emu_devices, &usb_device_dimensions::make_instance},
-		{0x0E6F, 0x200A, 0x200A, "Kamen Rider Summonride Portal", nullptr, nullptr},
+		{0x0E6F, 0x200A, 0x200A, "Kamen Rider Summonride Portal", &usb_device_kamen_rider::get_num_emu_devices, &usb_device_kamen_rider::make_instance},
 
 		// Cameras
 		// {0x1415, 0x0020, 0x2000, "Sony Playstation Eye", nullptr, nullptr}, // TODO: verifiy

rpcs3/Emu/Io/Dimensions.cpp

@@ -544,9 +544,7 @@ std::optional<std::array<u8, 32>> dimensions_toypad::pop_added_removed_response(
 	std::lock_guard lock(m_dimensions_mutex);
 
 	if (m_figure_added_removed_responses.empty())
-	{
 		return std::nullopt;
-	}
 
 	std::array<u8, 32> response = m_figure_added_removed_responses.front();
 	m_figure_added_removed_responses.pop();
@@ -597,7 +595,6 @@ void usb_device_dimensions::interrupt_transfer(u32 buf_size, u8* buf, u32 endpoi
 	{
 		// Read Endpoint, if a request has not been sent via the write endpoint, set expected result as
 		// EHCI_CC_HALTED so the game doesn't report the Toypad as being disconnected.
-		std::lock_guard lock(m_query_mutex);
 		std::optional<std::array<u8, 32>> response = g_dimensionstoypad.pop_added_removed_response();
 		if (response)
 		{
@@ -696,7 +693,6 @@ void usb_device_dimensions::interrupt_transfer(u32 buf_size, u8* buf, u32 endpoi
 			break;
 		}
 		}
-		std::lock_guard lock(m_query_mutex);
 		m_queries.push(q_result);
 		break;
 	}

rpcs3/Emu/Io/Dimensions.h

@@ -79,6 +79,5 @@ public:
 	void isochronous_transfer(UsbTransfer* transfer) override;
 
 protected:
-	shared_mutex m_query_mutex;
 	std::queue<std::array<u8, 32>> m_queries;
 };

rpcs3/Emu/Io/Infinity.cpp

@@ -258,13 +258,13 @@ void infinity_base::get_figure_identifier(u8 fig_num, u8 sequence, std::array<u8
 	reply_buf[11] = generate_checksum(reply_buf, 11);
 }
 
-bool infinity_base::has_figure_been_added_removed() const
+std::optional<std::array<u8, 32>> infinity_base::pop_added_removed_response()
 {
-	return !m_figure_added_removed_responses.empty();
+	std::lock_guard lock(infinity_mutex);
-}
+
+	if (m_figure_added_removed_responses.empty())
+		return std::nullopt;
 
-std::array<u8, 32> infinity_base::pop_added_removed_response()
-{
 	std::array<u8, 32> response = m_figure_added_removed_responses.front();
 	m_figure_added_removed_responses.pop();
 	return response;
@@ -399,9 +399,10 @@ void usb_device_infinity::interrupt_transfer(u32 buf_size, u8* buf, u32 endpoint
 	{
 		// Respond after FF command
 		transfer->expected_time = get_timestamp() + 1000;
-		if (g_infinitybase.has_figure_been_added_removed())
+		std::optional<std::array<u8, 32>> response = g_infinitybase.pop_added_removed_response();
+		if (response)
 		{
-			memcpy(buf, g_infinitybase.pop_added_removed_response().data(), 0x20);
+			memcpy(buf, response.value().data(), 0x20);
 		}
 		else if (!m_queries.empty())
 		{

rpcs3/Emu/Io/Infinity.h

@@ -3,6 +3,7 @@
 #include "Emu/Io/usb_device.h"
 #include "Utilities/mutex.h"
 #include <array>
+#include <optional>
 #include <queue>
 
 struct infinity_figure
@@ -24,8 +25,7 @@ public:
 	void query_block(u8 fig_num, u8 block, std::array<u8, 32>& reply_buf, u8 sequence);
 	void write_block(u8 fig_num, u8 block, const u8* to_write_buf, std::array<u8, 32>& reply_buf, u8 sequence);
 	void get_figure_identifier(u8 fig_num, u8 sequence, std::array<u8, 32>& reply_buf);
-	bool has_figure_been_added_removed() const;
+	std::optional<std::array<u8, 32>> pop_added_removed_response();
-	std::array<u8, 32> pop_added_removed_response();
 
 	bool remove_figure(u8 position);
 	u32 load_figure(const std::array<u8, 0x14 * 0x10>& buf, fs::file in_file, u8 position);

rpcs3/Emu/Io/KamenRider.cpp

@@ -0,0 +1,291 @@
+#include "stdafx.h"
+#include "KamenRider.h"
+
+LOG_CHANNEL(kamen_rider_log, "kamen_rider");
+
+rider_gate g_ridergate;
+
+void kamen_rider_figure::save()
+{
+	if (!kamen_file)
+	{
+		kamen_rider_log.error("Tried to save kamen rider figure to file but no kamen rider figure is active!");
+		return;
+	}
+	kamen_file.seek(0, fs::seek_set);
+	kamen_file.write(data.data(), 0x14 * 0x10);
+}
+
+u8 rider_gate::generate_checksum(const std::array<u8, 64>& data, u32 num_of_bytes) const
+{
+	ensure(num_of_bytes <= data.size());
+	int checksum = 0;
+	for (u32 i = 0; i < num_of_bytes; i++)
+	{
+		checksum += data[i];
+	}
+	return (checksum & 0xFF);
+}
+
+kamen_rider_figure& rider_gate::get_figure_by_uid(const std::array<u8, 7> uid)
+{
+	for (kamen_rider_figure& figure : figures)
+	{
+		if (figure.uid == uid)
+		{
+			return figure;
+		}
+	}
+	return figures[7];
+}
+
+void rider_gate::get_blank_response(u8 command, u8 sequence, std::array<u8, 64>& reply_buf)
+{
+	reply_buf = {0x55, 0x02, command, sequence};
+	reply_buf[4] = generate_checksum(reply_buf, 4);
+}
+
+void rider_gate::wake_rider_gate(std::array<u8, 64>& reply_buf, u8 command, u8 sequence)
+{
+	std::lock_guard lock(kamen_mutex);
+
+	m_is_awake = true;
+	reply_buf = {0x55, 0x1a, command, sequence, 0x00, 0x07, 0x00, 0x03, 0x02,
+		0x09, 0x20, 0x03, 0xf5, 0x00, 0x19, 0x42, 0x52, 0xb7,
+		0xb9, 0xa1, 0xae, 0x2b, 0x88, 0x42, 0x05, 0xfe, 0xe0, 0x1c, 0xac};
+}
+
+void rider_gate::get_list_tags(std::array<u8, 64>& reply_buf, u8 command, u8 sequence)
+{
+	std::lock_guard lock(kamen_mutex);
+
+	reply_buf = {0x55, 0x02, command, sequence};
+	u8 index = 4;
+	for (const kamen_rider_figure& figure : figures)
+	{
+		if (figure.present)
+		{
+			reply_buf[index] = 0x09;
+			memcpy(&reply_buf[index + 1], figure.data.data(), 7);
+			index += 8;
+			reply_buf[1] += 8;
+		}
+	}
+	reply_buf[index] = generate_checksum(reply_buf, index);
+}
+
+void rider_gate::query_block(std::array<u8, 64>& reply_buf, u8 command, u8 sequence, const u8* uid, u8 sector, u8 block)
+{
+	std::lock_guard lock(kamen_mutex);
+
+	reply_buf = {0x55, 0x13, command, sequence, 0x00};
+
+	const std::array<u8, 7> uid_array = {uid[0], uid[1], uid[2], uid[3], uid[4], uid[5], uid[6]};
+
+	const kamen_rider_figure& figure = get_figure_by_uid(uid_array);
+	if (figure.present)
+	{
+		if (sector < 5 && block < 4)
+		{
+			memcpy(&reply_buf[5], &figure.data[(sector * 4 * 16) + (block * 16)], 16);
+		}
+	}
+	reply_buf[21] = generate_checksum(reply_buf, 21);
+}
+
+void rider_gate::write_block(std::array<u8, 64>& replyBuf, u8 command, u8 sequence, const u8* uid, u8 sector, u8 block, const u8* to_write_buf)
+{
+	std::lock_guard lock(kamen_mutex);
+
+	const std::array<u8, 7> uid_array = {uid[0], uid[1], uid[2], uid[3], uid[4], uid[5], uid[6]};
+
+	kamen_rider_figure& figure = get_figure_by_uid(uid_array);
+	if (figure.present)
+	{
+		if (sector < 5 && block < 4)
+		{
+			memcpy(&figure.data[(sector * 4 * 16) + (block * 16)], to_write_buf, 16);
+		}
+	}
+
+	get_blank_response(command, sequence, replyBuf);
+}
+
+std::optional<std::array<u8, 64>> rider_gate::pop_added_removed_response()
+{
+	std::lock_guard lock(kamen_mutex);
+
+	if (m_figure_added_removed_responses.empty())
+	{
+		return std::nullopt;
+	}
+
+	std::array<u8, 64> response = m_figure_added_removed_responses.front();
+	m_figure_added_removed_responses.pop();
+	return response;
+}
+
+bool rider_gate::remove_figure(u8 index)
+{
+	std::lock_guard lock(kamen_mutex);
+
+	auto& figure = figures[index];
+
+	if (figure.present)
+	{
+		figure.present = false;
+		figure.save();
+		figure.kamen_file.close();
+		if (m_is_awake)
+		{
+			std::array<u8, 64> figure_removed_response = {0x56, 0x09, 0x09, 0x00};
+			memcpy(&figure_removed_response[4], figure.uid.data(), figure.uid.size());
+			figure_removed_response[11] = generate_checksum(figure_removed_response, 11);
+			m_figure_added_removed_responses.push(std::move(figure_removed_response));
+		}
+		figure.uid = {};
+		return true;
+	}
+
+	return false;
+}
+
+u8 rider_gate::load_figure(const std::array<u8, 0x14 * 0x10>& buf, fs::file in_file)
+{
+	std::lock_guard lock(kamen_mutex);
+
+	u8 found_slot = 0xFF;
+
+	// mimics spot retaining on the portal
+	for (auto i = 0; i < 7; i++)
+	{
+		if (!figures[i].present)
+		{
+			if (i < found_slot)
+			{
+				found_slot = i;
+			}
+		}
+	}
+
+	if (found_slot != 0xFF)
+	{
+		auto& figure = figures[found_slot];
+		memcpy(figure.data.data(), buf.data(), buf.size());
+		figure.kamen_file = std::move(in_file);
+		figure.uid = {buf[0], buf[1], buf[2], buf[3], buf[4], buf[5], buf[6]};
+		figure.present = true;
+
+		if (m_is_awake)
+		{
+			std::array<u8, 64> figure_added_response = {0x56, 0x09, 0x09, 0x01};
+			memcpy(&figure_added_response[4], figure.uid.data(), figure.uid.size());
+			figure_added_response[11] = generate_checksum(figure_added_response, 11);
+			m_figure_added_removed_responses.push(std::move(figure_added_response));
+		}
+	}
+	return found_slot;
+}
+
+usb_device_kamen_rider::usb_device_kamen_rider(const std::array<u8, 7>& location)
+	: usb_device_emulated(location)
+{
+	device = UsbDescriptorNode(USB_DESCRIPTOR_DEVICE, UsbDeviceDescriptor{0x200, 0x0, 0x0, 0x0, 0x40, 0x0E6F, 0x200A, 0x100, 0x1, 0x2, 0x3, 0x1});
+	auto& config0 = device.add_node(UsbDescriptorNode(USB_DESCRIPTOR_CONFIG, UsbDeviceConfiguration{0x29, 0x1, 0x1, 0x0, 0x80, 0xFA}));
+	config0.add_node(UsbDescriptorNode(USB_DESCRIPTOR_INTERFACE, UsbDeviceInterface{0x0, 0x0, 0x2, 0x3, 0x0, 0x0, 0x0}));
+	config0.add_node(UsbDescriptorNode(USB_DESCRIPTOR_ENDPOINT, UsbDeviceEndpoint{0x81, 0x3, 0x40, 0x1}));
+	config0.add_node(UsbDescriptorNode(USB_DESCRIPTOR_ENDPOINT, UsbDeviceEndpoint{0x1, 0x3, 0x40, 0x1}));
+}
+
+usb_device_kamen_rider::~usb_device_kamen_rider()
+{
+}
+
+std::shared_ptr<usb_device> usb_device_kamen_rider::make_instance(u32, const std::array<u8, 7>& location)
+{
+	return std::make_shared<usb_device_kamen_rider>(location);
+}
+
+u16 usb_device_kamen_rider::get_num_emu_devices()
+{
+	return 1;
+}
+
+void usb_device_kamen_rider::control_transfer(u8 bmRequestType, u8 bRequest, u16 wValue, u16 wIndex, u16 wLength, u32 buf_size, u8* buf, UsbTransfer* transfer)
+{
+	usb_device_emulated::control_transfer(bmRequestType, bRequest, wValue, wIndex, wLength, buf_size, buf, transfer);
+}
+
+void usb_device_kamen_rider::interrupt_transfer(u32 buf_size, u8* buf, u32 endpoint, UsbTransfer* transfer)
+{
+	ensure(buf_size == 0x40);
+
+	transfer->fake = true;
+	transfer->expected_count = buf_size;
+	transfer->expected_result = HC_CC_NOERR;
+
+	if (endpoint == 0x81)
+	{
+		// Respond after FF command
+		transfer->expected_time = get_timestamp() + 1000;
+		std::optional<std::array<u8, 64>> response = g_ridergate.pop_added_removed_response();
+		if (response)
+		{
+			memcpy(buf, response.value().data(), 0x40);
+		}
+		else if (!m_queries.empty())
+		{
+			memcpy(buf, m_queries.front().data(), 0x20);
+			m_queries.pop();
+		}
+		else
+		{
+			transfer->expected_count = 0;
+			transfer->expected_result = EHCI_CC_HALTED;
+		}
+	}
+	else if (endpoint == 0x01)
+	{
+		const u8 command = buf[2];
+		const u8 sequence = buf[3];
+
+		std::array<u8, 64> q_result{};
+
+		switch (command)
+		{
+		case 0xB0: // Wake
+		{
+			g_ridergate.wake_rider_gate(q_result, command, sequence);
+			break;
+		}
+		case 0xC0:
+		case 0xC3: // Color Commands
+		{
+			g_ridergate.get_blank_response(command, sequence, q_result);
+			break;
+		}
+		case 0xD0: // Tag List
+		{
+			// Return list of figure UIDs, separated by an 09
+			g_ridergate.get_list_tags(q_result, command, sequence);
+			break;
+		}
+		case 0xD2: // Read
+		{
+			// Read 16 bytes from figure with UID buf[4] - buf[10]
+			g_ridergate.query_block(q_result, command, sequence, &buf[4], buf[11], buf[12]);
+			break;
+		}
+		case 0xD3:
+		{
+			// Write 16 bytes to figure with UID buf[4] - buf[10]
+			g_ridergate.write_block(q_result, command, sequence, &buf[4], buf[11], buf[12], &buf[13]);
+			break;
+		}
+		default:
+			kamen_rider_log.error("Unhandled Query Type: 0x%02X", command);
+			break;
+		}
+		m_queries.push(std::move(q_result));
+	}
+}

rpcs3/Emu/Io/KamenRider.h

@@ -0,0 +1,60 @@
+#pragma once
+
+#include "Emu/Io/usb_device.h"
+#include "Utilities/mutex.h"
+#include <array>
+#include <optional>
+#include <queue>
+
+struct kamen_rider_figure
+{
+	fs::file kamen_file;
+	std::array<u8, 0x14 * 0x10> data{};
+	std::array<u8, 7> uid{};
+	bool present = false;
+	void save();
+};
+
+class rider_gate
+{
+public:
+	void get_blank_response(u8 command, u8 sequence, std::array<u8, 64>& reply_buf);
+	void wake_rider_gate(std::array<u8, 64>& replyBuf, u8 command, u8 sequence);
+	void get_list_tags(std::array<u8, 64>& replyBuf, u8 command, u8 sequence);
+	void query_block(std::array<u8, 64>& replyBuf, u8 command, u8 sequence, const u8* uid, u8 sector, u8 block);
+	void write_block(std::array<u8, 64>& replyBuf, u8 command, u8 sequence, const u8* uid, u8 sector, u8 block, const u8* to_write_buf);
+	std::optional<std::array<u8, 64>> pop_added_removed_response();
+
+	bool remove_figure(u8 position);
+	u8 load_figure(const std::array<u8, 0x14 * 0x10>& buf, fs::file in_file);
+
+protected:
+	shared_mutex kamen_mutex;
+	std::array<kamen_rider_figure, 8> figures{};
+
+private:
+	u8 generate_checksum(const std::array<u8, 64>& data, u32 num_of_bytes) const;
+	kamen_rider_figure& get_figure_by_uid(const std::array<u8, 7> uid);
+
+	std::queue<std::array<u8, 64>> m_figure_added_removed_responses;
+
+	bool m_is_awake = false;
+};
+
+extern rider_gate g_ridergate;
+
+class usb_device_kamen_rider : public usb_device_emulated
+{
+public:
+	usb_device_kamen_rider(const std::array<u8, 7>& location);
+	~usb_device_kamen_rider();
+
+	static std::shared_ptr<usb_device> make_instance(u32 controller_index, const std::array<u8, 7>& location);
+	static u16 get_num_emu_devices();
+
+	void control_transfer(u8 bmRequestType, u8 bRequest, u16 wValue, u16 wIndex, u16 wLength, u32 buf_size, u8* buf, UsbTransfer* transfer) override;
+	void interrupt_transfer(u32 buf_size, u8* buf, u32 endpoint, UsbTransfer* transfer) override;
+
+protected:
+	std::queue<std::array<u8, 64>> m_queries;
+};

rpcs3/Emu/Io/PadHandler.cpp

@@ -170,7 +170,7 @@ u16 PadHandlerBase::ConvertAxis(f32 value)
 
 // The DS3, (and i think xbox controllers) give a 'square-ish' type response, so that the corners will give (almost)max x/y instead of the ~30x30 from a perfect circle
 // using a simple scale/sensitivity increase would *work* although it eats a chunk of our usable range in exchange
-// this might be the best for now, in practice it seems to push the corners to max of 20x20, with a squircle_factor of 8000
+// this might be the best for now, in practice it seems to push the corners to max of 20x20, with a squircle_factor of ~4000
 // This function assumes inX and inY is already in 0-255
 void PadHandlerBase::ConvertToSquirclePoint(u16& inX, u16& inY, u32 squircle_factor)
 {
@@ -208,6 +208,21 @@ void PadHandlerBase::init_configs()
 	}
 }
 
+pad_capabilities PadHandlerBase::get_capabilities(const std::string& /*pad_id*/)
+{
+	return pad_capabilities
+	{
+		.has_led = b_has_rgb,
+		.has_mono_led = b_has_led,
+		.has_player_led = b_has_player_led,
+		.has_battery_led = b_has_battery_led,
+		.has_rumble = b_has_rumble,
+		.has_accel = b_has_motion,
+		.has_gyro = b_has_motion,
+		.has_pressure_sensitivity = b_has_pressure_intensity_button
+	};
+}
+
 cfg_pad* PadHandlerBase::get_config(const std::string& pad_id)
 {
 	int index = 0;
@@ -331,12 +346,13 @@ PadHandlerBase::connection PadHandlerBase::get_next_button_press(const std::stri
 	if (callback)
 	{
 		pad_preview_values preview_values = get_preview_values(data);
+		pad_capabilities capabilities = get_capabilities(pad_id);
 		const u32 battery_level = get_battery_level(pad_id);
 
 		if (pressed_button.value > 0)
-			callback(pressed_button.value, pressed_button.name, pad_id, battery_level, std::move(preview_values));
+			callback(pressed_button.value, pressed_button.name, pad_id, battery_level, std::move(preview_values), std::move(capabilities));
 		else
-			callback(0, "", pad_id, battery_level, std::move(preview_values));
+			callback(0, "", pad_id, battery_level, std::move(preview_values), std::move(capabilities));
 	}
 
 	return status;

rpcs3/Emu/Io/PadHandler.h

@@ -81,8 +81,20 @@ struct pad_list_entry
 	{}
 };
 
+struct pad_capabilities
+{
+	bool has_led = false;
+	bool has_mono_led = false;
+	bool has_player_led = false;
+	bool has_battery_led = false;
+	bool has_rumble = false;
+	bool has_accel = false;
+	bool has_gyro = false;
+	bool has_pressure_sensitivity = false;
+};
+
 using pad_preview_values = std::array<int, 6>;
-using pad_callback = std::function<void(u16 /*button_value*/, std::string /*button_name*/, std::string /*pad_name*/, u32 /*battery_level*/, pad_preview_values /*preview_values*/)>;
+using pad_callback = std::function<void(u16 /*button_value*/, std::string /*button_name*/, std::string /*pad_name*/, u32 /*battery_level*/, pad_preview_values, pad_capabilities)>;
 using pad_fail_callback = std::function<void(std::string /*pad_name*/)>;
 
 using motion_preview_values = std::array<u16, 4>;
@@ -326,6 +338,8 @@ public:
 	bool has_analog_limiter_button() const { return b_has_analog_limiter_button; }
 	bool has_orientation() const { return b_has_orientation; }
 
+	virtual pad_capabilities get_capabilities(const std::string& /*pad_id*/);
+
 	u16 NormalizeStickInput(u16 raw_value, s32 threshold, s32 multiplier, bool ignore_threshold = false) const;
 	void convert_stick_values(u16& x_out, u16& y_out, s32 x_in, s32 y_in, u32 deadzone, u32 anti_deadzone, u32 padsquircling) const;
 	void set_trigger_recognition_mode(trigger_recognition_mode mode) { m_trigger_recognition_mode = mode; }

rpcs3/Emu/Io/pad_config.h

@@ -88,8 +88,8 @@ struct cfg_pad final : cfg::node
 	cfg::uint<0, 1000000> rstick_anti_deadzone{ this, "Right Stick Anti-Deadzone", 0 };
 	cfg::uint<0, 1000000> ltriggerthreshold{ this, "Left Trigger Threshold", 0 };
 	cfg::uint<0, 1000000> rtriggerthreshold{ this, "Right Trigger Threshold", 0 };
-	cfg::uint<0, 1000000> lpadsquircling{ this, "Left Pad Squircling Factor", 8000 };
+	cfg::uint<0, 1000000> lpadsquircling{ this, "Left Pad Squircling Factor", 4000 };
-	cfg::uint<0, 1000000> rpadsquircling{ this, "Right Pad Squircling Factor", 8000 };
+	cfg::uint<0, 1000000> rpadsquircling{ this, "Right Pad Squircling Factor", 4000 };
 
 	cfg::uint<0, 255> colorR{ this, "Color Value R", 0 };
 	cfg::uint<0, 255> colorG{ this, "Color Value G", 0 };

rpcs3/Emu/Io/pad_types.cpp

@@ -217,6 +217,12 @@ bool Pad::get_analog_limiter_button_active(bool is_toggle_mode, u32 player_id)
 
 	const Button& analog_limiter_button = m_buttons[m_analog_limiter_button_index];
 
+	if (analog_limiter_button.m_key_codes.empty())
+	{
+		// Active by default if no button was assigned
+		return true;
+	}
+
 	if (is_toggle_mode)
 	{
 		const bool pressed = analog_limiter_button.m_pressed;

rpcs3/Emu/Memory/vm.cpp

@@ -98,7 +98,6 @@ namespace vm
 
 	void reservation_update(u32 addr)
 	{
-		u64 old = -1;
 		const auto cpu = get_current_cpu_thread();
 
 		const bool had_wait = cpu && cpu->state & cpu_flag::wait;
@@ -126,8 +125,6 @@ namespace vm
 
 				return;
 			}
-	
-			old = rtime;
 		}
 	}
 
@@ -550,6 +547,13 @@ namespace vm
 			{
 				to_clear = for_all_range_locks(to_clear & ~get_range_lock_bits(true), [&](u64 addr2, u32 size2)
 				{
+					constexpr u32 range_size_loc = vm::range_pos - 32;
+
+					if ((size2 >> range_size_loc) == (vm::range_readable >> vm::range_pos))
+					{
+						return 0;
+					}
+
 					// Split and check every 64K page separately
 					for (u64 hi = addr2 >> 16, max = (addr2 + size2 - 1) >> 16; hi <= max; hi++)
 					{
@@ -953,7 +957,7 @@ namespace vm
 		return true;
 	}
 
-	static u32 _page_unmap(u32 addr, u32 max_size, u64 bflags, utils::shm* shm, std::vector<std::pair<u64, u64>>& unmap_events)
+	static u32 _page_unmap(u32 addr, u32 max_size, u64 bflags, utils::shm* shm, std::vector<std::pair<u64, u64>>& unmap_events, bool is_block_termination = false)
 	{
 		perf_meter<"PAGE_UNm"_u64> perf0;
 
@@ -1024,7 +1028,11 @@ namespace vm
 		ppu_remove_hle_instructions(addr, size);
 
 		// Actually unmap memory
-		if (is_noop)
+		if (is_block_termination && (!shm || is_noop))
+		{
+			// We can skip it if the block is freed
+		}
+		else if (is_noop)
 		{
 			std::memset(g_sudo_addr + addr, 0, size);
 		}
@@ -1330,7 +1338,17 @@ namespace vm
 				const auto size = it->second.first;
 
 				std::vector<std::pair<u64, u64>> event_data;
-				ensure(size == _page_unmap(it->first, size, this->flags, it->second.second.get(), unmapped ? *unmapped : event_data));
+				ensure(size == _page_unmap(it->first, size, this->flags, it->second.second.get(), unmapped ? *unmapped : event_data, true));
+
+				if (it->second.second && addr < 0xE0000000)
+				{
+					if (it->second.second.use_count() != 1)
+					{
+						fmt::throw_exception("External memory usage at block 0x%x (addr=0x%x, size=0x%x)", this->addr, it->first, size);
+					}
+
+					it->second.second.reset();
+				}
 
 				it = next;
 			}
@@ -1341,6 +1359,8 @@ namespace vm
 #ifdef _WIN32
 				m_common->unmap_critical(vm::get_super_ptr(addr));
 #endif
+				ensure(m_common.use_count() == 1);
+				m_common.reset();
 			}
 
 			return true;
@@ -1352,6 +1372,7 @@ namespace vm
 	block_t::~block_t()
 	{
 		ensure(!is_valid());
+		ensure(!m_common || m_common.use_count() == 1);
 	}
 
 	u32 block_t::alloc(const u32 orig_size, const std::shared_ptr<utils::shm>* src, u32 align, u64 flags)
@@ -2247,7 +2268,11 @@ namespace vm
 
 			for (auto& block : g_locations)
 			{
-				if (block) _unmap_block(block);
+				if (block)
+				{
+					_unmap_block(block);
+					ensure(block.use_count() == 1);
+				}
 			}
 
 			g_locations.clear();

rpcs3/Emu/Memory/vm.h

@@ -81,7 +81,7 @@ namespace vm
 	bool check_addr(u32 addr, u8 flags, u32 size);
 
 	template <u32 Size = 1>
-	bool check_addr(u32 addr, u8 flags = page_readable)
+	inline bool check_addr(u32 addr, u8 flags = page_readable)
 	{
 		extern std::array<memory_page, 0x100000000 / 4096> g_pages;
 
@@ -94,6 +94,16 @@ namespace vm
 		return !(~g_pages[addr / 4096] & (flags | page_allocated));
 	}
 
+	// Like check_addr but should only be used in lock-free context with care
+	inline std::pair<bool, u8> get_addr_flags(u32 addr) noexcept
+	{
+		extern std::array<memory_page, 0x100000000 / 4096> g_pages;
+
+		const u8 flags = g_pages[addr / 4096].load();
+
+		return std::make_pair(!!(flags & page_allocated), flags);
+	}
+
 	// Read string in a safe manner (page aware) (bool true = if null-termination)
 	bool read_string(u32 addr, u32 max_size, std::string& out_string, bool check_pages = true) noexcept;
 

rpcs3/Emu/Memory/vm_reservation.h

@@ -6,9 +6,6 @@
 #include "util/tsc.hpp"
 #include <functional>
 
-extern bool g_use_rtm;
-extern u64 g_rtm_tx_limit2;
-
 #ifdef _MSC_VER
 extern "C"
 {
@@ -143,7 +140,7 @@ namespace vm
 	void reservation_op_internal(u32 addr, std::function<bool()> func);
 
 	template <bool Ack = false, typename CPU, typename T, typename AT = u32, typename F>
-	inline SAFE_BUFFERS(auto) reservation_op(CPU& cpu, _ptr_base<T, AT> ptr, F op)
+	inline SAFE_BUFFERS(auto) reservation_op(CPU& /*cpu*/, _ptr_base<T, AT> ptr, F op)
 	{
 		// Atomic operation will be performed on aligned 128 bytes of data, so the data size and alignment must comply
 		static_assert(sizeof(T) <= 128 && alignof(T) == sizeof(T), "vm::reservation_op: unsupported type");
@@ -162,188 +159,6 @@ namespace vm
 		auto& res = vm::reservation_acquire(addr);
 		//_m_prefetchw(&res);
 
-#if defined(ARCH_X64)
-		if (g_use_rtm)
-		{
-			// Stage 1: single optimistic transaction attempt
-			unsigned status = -1;
-			u64 _old = 0;
-
-			auto stamp0 = utils::get_tsc(), stamp1 = stamp0, stamp2 = stamp0;
-
-#ifndef _MSC_VER
-			__asm__ goto ("xbegin %l[stage2];" ::: "memory" : stage2);
-#else
-			status = _xbegin();
-			if (status == umax)
-#endif
-			{
-				if (res & rsrv_unique_lock)
-				{
-#ifndef _MSC_VER
-					__asm__ volatile ("xend; mov $-1, %%eax;" ::: "memory");
-#else
-					_xend();
-#endif
-					goto stage2;
-				}
-
-				if constexpr (std::is_void_v<std::invoke_result_t<F, T&>>)
-				{
-					std::invoke(op, *sptr);
-					const u64 old_time = res.fetch_add(128);
-#ifndef _MSC_VER
-					__asm__ volatile ("xend;" ::: "memory");
-#else
-					_xend();
-#endif
-					if constexpr (Ack)
-						reservation_notifier_notify(addr, old_time);
-					return;
-				}
-				else
-				{
-					if (auto result = std::invoke(op, *sptr))
-					{
-						const u64 old_time = res.fetch_add(128);
-#ifndef _MSC_VER
-						__asm__ volatile ("xend;" ::: "memory");
-#else
-						_xend();
-#endif
-						if constexpr (Ack)
-							reservation_notifier_notify(addr, old_time);
-						return result;
-					}
-					else
-					{
-#ifndef _MSC_VER
-						__asm__ volatile ("xend;" ::: "memory");
-#else
-						_xend();
-#endif
-						return result;
-					}
-				}
-			}
-
-			stage2:
-#ifndef _MSC_VER
-			__asm__ volatile ("mov %%eax, %0;" : "=r" (status) :: "memory");
-#endif
-			stamp1 = utils::get_tsc();
-
-			// Stage 2: try to lock reservation first
-			_old = res.fetch_add(1);
-
-			// Compute stamps excluding memory touch
-			stamp2 = utils::get_tsc() - (stamp1 - stamp0);
-
-			// Start lightened transaction
-			for (; !(_old & vm::rsrv_unique_lock) && stamp2 - stamp0 <= g_rtm_tx_limit2; stamp2 = utils::get_tsc())
-			{
-				if (cpu.has_pause_flag())
-				{
-					break;
-				}
-
-#ifndef _MSC_VER
-				__asm__ goto ("xbegin %l[retry];" ::: "memory" : retry);
-#else
-				status = _xbegin();
-
-				if (status != umax) [[unlikely]]
-				{
-					goto retry;
-				}
-#endif
-				if constexpr (std::is_void_v<std::invoke_result_t<F, T&>>)
-				{
-					std::invoke(op, *sptr);
-#ifndef _MSC_VER
-					__asm__ volatile ("xend;" ::: "memory");
-#else
-					_xend();
-#endif
-					res += 127;
-					if (Ack)
-						reservation_notifier_notify(addr, _old);
-					return;
-				}
-				else
-				{
-					if (auto result = std::invoke(op, *sptr))
-					{
-#ifndef _MSC_VER
-						__asm__ volatile ("xend;" ::: "memory");
-#else
-						_xend();
-#endif
-						res += 127;
-						if (Ack)
-							reservation_notifier_notify(addr, _old);
-						return result;
-					}
-					else
-					{
-#ifndef _MSC_VER
-						__asm__ volatile ("xend;" ::: "memory");
-#else
-						_xend();
-#endif
-						return result;
-					}
-				}
-
-				retry:
-#ifndef _MSC_VER
-				__asm__ volatile ("mov %%eax, %0;" : "=r" (status) :: "memory");
-#endif
-
-				if (!status)
-				{
-					break;
-				}
-			}
-
-			// Stage 3: all failed, heavyweight fallback (see comments at the bottom)
-			if constexpr (std::is_void_v<std::invoke_result_t<F, T&>>)
-			{
-				vm::reservation_op_internal(addr, [&]
-				{
-					std::invoke(op, *sptr);
-					return true;
-				});
-
-				if constexpr (Ack)
-					reservation_notifier_notify(addr, _old);
-				return;
-			}
-			else
-			{
-				auto result = std::invoke_result_t<F, T&>();
-
-				vm::reservation_op_internal(addr, [&]
-				{
-					if ((result = std::invoke(op, *sptr)))
-					{
-						return true;
-					}
-					else
-					{
-						return false;
-					}
-				});
-
-				if (Ack && result)
-					reservation_notifier_notify(addr, _old);
-				return result;
-			}
-		}
-#else
-		static_cast<void>(cpu);
-#endif /* ARCH_X64 */
-
 		// Lock reservation and perform heavyweight lock
 		reservation_shared_lock_internal(res);
 

rpcs3/Emu/NP/np_contexts.cpp

@@ -4,6 +4,7 @@
 #include "Emu/Cell/PPUCallback.h"
 #include "Emu/IdManager.h"
 #include "Emu/Cell/Modules/cellSysutil.h"
+#include "np_helpers.h"
 
 LOG_CHANNEL(sceNp2);
 
@@ -53,7 +54,7 @@ void generic_async_transaction_context::set_result_and_wake(error_code err)
 
 tus_ctx::tus_ctx(vm::cptr<SceNpCommunicationId> communicationId, vm::cptr<SceNpCommunicationPassphrase> passphrase)
 {
-	ensure(!communicationId->data[9] && strlen(communicationId->data) == 9);
+	ensure(communicationId && np::validate_communication_id(*communicationId), "tus_ctx::tus_ctx: Invalid SceNpCommunicationId");
 	memcpy(&this->communicationId, communicationId.get_ptr(), sizeof(SceNpCommunicationId));
 	memcpy(&this->passphrase, passphrase.get_ptr(), sizeof(SceNpCommunicationPassphrase));
 }
@@ -96,7 +97,7 @@ bool destroy_tus_transaction_context(s32 ctx_id)
 
 score_ctx::score_ctx(vm::cptr<SceNpCommunicationId> communicationId, vm::cptr<SceNpCommunicationPassphrase> passphrase)
 {
-	ensure(!communicationId->data[9] && strlen(communicationId->data) == 9);
+	ensure(communicationId && np::validate_communication_id(*communicationId), "score_ctx::score_ctx: Invalid SceNpCommunicationId");
 	memcpy(&this->communicationId, communicationId.get_ptr(), sizeof(SceNpCommunicationId));
 	memcpy(&this->passphrase, passphrase.get_ptr(), sizeof(SceNpCommunicationPassphrase));
 }
@@ -140,7 +141,7 @@ bool destroy_score_transaction_context(s32 ctx_id)
 
 match2_ctx::match2_ctx(vm::cptr<SceNpCommunicationId> communicationId, vm::cptr<SceNpCommunicationPassphrase> passphrase, s32 option)
 {
-	ensure(!communicationId->data[9] && strlen(communicationId->data) == 9);
+	ensure(communicationId && np::validate_communication_id(*communicationId), "match2_ctx::match2_ctx: Invalid SceNpCommunicationId");
 	memcpy(&this->communicationId, communicationId.get_ptr(), sizeof(SceNpCommunicationId));
 	memcpy(&this->passphrase, passphrase.get_ptr(), sizeof(SceNpCommunicationPassphrase));
 
@@ -149,7 +150,7 @@ match2_ctx::match2_ctx(vm::cptr<SceNpCommunicationId> communicationId, vm::cptr<
 }
 u16 create_match2_context(vm::cptr<SceNpCommunicationId> communicationId, vm::cptr<SceNpCommunicationPassphrase> passphrase, s32 option)
 {
-	sceNp2.notice("Creating match2 context with communicationId: <%s>", static_cast<const char*>(communicationId->data));
+	sceNp2.notice("Creating match2 context with communicationId: <%s>", std::string_view(communicationId->data, 9));
 	return static_cast<u16>(idm::make<match2_ctx>(communicationId, passphrase, option));
 }
 bool destroy_match2_context(u16 ctx_id)
@@ -167,7 +168,7 @@ shared_ptr<match2_ctx> get_match2_context(u16 ctx_id)
 
 lookup_title_ctx::lookup_title_ctx(vm::cptr<SceNpCommunicationId> communicationId)
 {
-	ensure(!communicationId->data[9] && strlen(communicationId->data) == 9);
+	ensure(communicationId && np::validate_communication_id(*communicationId), "lookup_title_ctx::lookup_title_ctx: Invalid SceNpCommunicationId");
 	memcpy(&this->communicationId, communicationId.get_ptr(), sizeof(SceNpCommunicationId));
 }
 s32 create_lookup_title_context(vm::cptr<SceNpCommunicationId> communicationId)

rpcs3/Emu/NP/np_handler.cpp

@@ -646,7 +646,7 @@ namespace np
 
 		for (; it != end; ++it)
 		{
-			strcpy(ifr.ifr_name, it->ifr_name);
+			strcpy_trunc(ifr.ifr_name, it->ifr_name);
 			if (ioctl(sock, SIOCGIFFLAGS, &ifr) == 0)
 			{
 				if (!(ifr.ifr_flags & IFF_LOOPBACK))

rpcs3/Emu/NP/np_helpers.cpp

@@ -1,3 +1,4 @@
+#include "Emu/Cell/Modules/sceNp.h"
 #include "stdafx.h"
 #include "util/types.hpp"
 #include "Utilities/StrUtil.h"
@@ -22,9 +23,15 @@ namespace np
 		return fmt::format("%02X:%02X:%02X:%02X:%02X:%02X", ether[0], ether[1], ether[2], ether[3], ether[4], ether[5]);
 	}
 
+	bool validate_communication_id(const SceNpCommunicationId& com_id)
+	{
+		return std::all_of(com_id.data, com_id.data + 9, [](char c) { return (c >= '0' && c <= '9') || (c >= 'A' && c <= 'Z'); }) && com_id.num <= 99;
+	}
+
 	std::string communication_id_to_string(const SceNpCommunicationId& communicationId)
 	{
-		return fmt::format("%s_%02d", communicationId.data, communicationId.num);
+		std::string_view com_id_data(communicationId.data, 9);
+		return fmt::format("%s_%02d", com_id_data, communicationId.num);
 	}
 
 	void strings_to_userinfo(std::string_view npid, std::string_view online_name, std::string_view avatar_url, SceNpUserInfo& user_info)

rpcs3/Emu/NP/np_helpers.h

@@ -8,6 +8,7 @@ namespace np
 {
 	std::string ip_to_string(u32 addr);
 	std::string ether_to_string(std::array<u8, 6>& ether);
+	bool validate_communication_id(const SceNpCommunicationId& com_id);
 	std::string communication_id_to_string(const SceNpCommunicationId& communicationId);
 
 	void string_to_npid(std::string_view str, SceNpId& npid);