Compare commits

..

14 Commits

Author SHA1 Message Date
lizzie b4309dfa4f nullptr check on multistream 2026-09-03 07:43:44 +00:00
lizzie 234086ba8d reset to nil 2026-09-03 06:55:12 +00:00
crueter d8b3a57322 a
Signed-off-by: crueter <crueter@eden-emu.dev>
2026-09-03 06:32:44 +00:00
crueter 51d4906738 Fix license headers 2026-09-03 06:32:44 +00:00
crueter ebe7bf7d9b Cleanup
Signed-off-by: crueter <crueter@eden-emu.dev>
2026-09-03 06:32:44 +00:00
crueter 3c133db2a0 Improve padding
Signed-off-by: crueter <crueter@eden-emu.dev>
2026-09-03 06:32:44 +00:00
crueter dc75c629ca Rebase
Signed-off-by: crueter <crueter@eden-emu.dev>
2026-09-03 06:32:44 +00:00
lizzie a514f74ffb adjust IsValidStreamCounts 2026-09-03 06:32:44 +00:00
lizzie a276b29fa7 fix debian 2026-09-03 06:32:44 +00:00
lizzie 68cbf12283 Trigger Build 2026-09-03 06:32:44 +00:00
lizzie ee708f56f9 fx 2026-09-03 06:32:44 +00:00
lizzie 6fb2892e17 brain no work 2026-09-03 06:32:44 +00:00
lizzie 7b0804cc15 implodes 2026-09-03 06:32:44 +00:00
lizzie 1a4d6fc199 [audio_core/dsp] nuke libopus, use ffmpeg for opus decoding
Signed-off-by: lizzie <lizzie@eden-emu.dev>
2026-09-03 06:32:44 +00:00
76 changed files with 840 additions and 2215 deletions
@@ -1,28 +0,0 @@
From cc15da16e533b2a801934eab2dfeaf3c3949a1dc Mon Sep 17 00:00:00 2001
From: crueter <crueter@eden-emu.dev>
Date: Mon, 8 Sep 2025 12:28:55 -0400
Subject: [PATCH] [cmake] disable NEON runtime check on clang-cl
When enabling runtime NEON checking for clang-cl, the linker would error out with `undefined symbol: __emit`, since clang doesn't actually implement this instruction. Therefore it makes sense to disable the runtime check by default on this platform, until either this is fixed or a clang-cl compatible intrinsic check is added (I don't have enough knowledge of MSVC to do this)
---
cmake/OpusConfig.cmake | 7 ++++++-
1 file changed, 6 insertions(+), 1 deletion(-)
diff --git a/cmake/OpusConfig.cmake b/cmake/OpusConfig.cmake
index e9319fbad..d0f459e88 100644
--- a/cmake/OpusConfig.cmake
+++ b/cmake/OpusConfig.cmake
@@ -71,7 +71,12 @@ elseif(OPUS_CPU_ARM AND NOT OPUS_DISABLE_INTRINSICS)
opus_detect_neon(COMPILER_SUPPORT_NEON)
if(COMPILER_SUPPORT_NEON)
option(OPUS_USE_NEON "Option to enable NEON" ON)
- option(OPUS_MAY_HAVE_NEON "Does runtime check for neon support" ON)
+ if (MSVC AND CMAKE_CXX_COMPILER_ID STREQUAL "Clang")
+ set(NEON_RUNTIME_CHECK_DEFAULT OFF)
+ else()
+ set(NEON_RUNTIME_CHECK_DEFAULT ON)
+ endif()
+ option(OPUS_MAY_HAVE_NEON "Does runtime check for neon support" ${NEON_RUNTIME_CHECK_DEFAULT})
option(OPUS_PRESUME_NEON "Assume target CPU has NEON support" OFF)
if(CMAKE_SYSTEM_PROCESSOR MATCHES "aarch64")
set(OPUS_PRESUME_NEON ON)
-153
View File
@@ -1,153 +0,0 @@
From bf455b67b4eaa446ffae5d25410b141b7b1b1082 Mon Sep 17 00:00:00 2001
From: crueter <crueter@eden-emu.dev>
Date: Mon, 8 Sep 2025 12:08:20 -0400
Subject: [PATCH] [cmake] `OPUS_INSTALL` option; only default install if root
project
Signed-off-by: crueter <crueter@eden-emu.dev>
---
CMakeLists.txt | 112 ++++++++++++++++++++++++++++---------------------
1 file changed, 64 insertions(+), 48 deletions(-)
diff --git a/CMakeLists.txt b/CMakeLists.txt
index fcf034b19..08b5e16f8 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -4,6 +4,13 @@ list(APPEND CMAKE_MODULE_PATH "${CMAKE_CURRENT_SOURCE_DIR}/cmake")
include(OpusPackageVersion)
get_package_version(PACKAGE_VERSION PROJECT_VERSION)
+# root project detection
+if(DEFINED PROJECT_NAME)
+ set(root_project OFF)
+else()
+ set(root_project ON)
+endif()
+
project(Opus LANGUAGES C VERSION ${PROJECT_VERSION})
include(OpusFunctions)
@@ -83,12 +90,16 @@ set(OPUS_DNN_FLOAT_DEBUG_HELP_STR "Run DNN computations as float for debugging p
option(OPUS_DNN_FLOAT_DEBUG ${OPUS_DNN_FLOAT_DEBUG_HELP_STR} OFF)
add_feature_info(OPUS_DNN_FLOAT_DEBUG OPUS_DNN_FLOAT_DEBUG ${OPUS_DNN_FLOAT_DEBUG_HELP_STR})
+set(OPUS_INSTALL_HELP_STR "Install Opus targets")
+option(OPUS_INSTALL ${OPUS_INSTALL_HELP_STR} ${root_project})
+add_feature_info(OPUS_INSTALL OPUS_INSTALL ${OPUS_INSTALL_HELP_STR})
+
set(OPUS_INSTALL_PKG_CONFIG_MODULE_HELP_STR "install pkg-config module.")
-option(OPUS_INSTALL_PKG_CONFIG_MODULE ${OPUS_INSTALL_PKG_CONFIG_MODULE_HELP_STR} ON)
+option(OPUS_INSTALL_PKG_CONFIG_MODULE ${OPUS_INSTALL_PKG_CONFIG_MODULE_HELP_STR} ${OPUS_INSTALL})
add_feature_info(OPUS_INSTALL_PKG_CONFIG_MODULE OPUS_INSTALL_PKG_CONFIG_MODULE ${OPUS_INSTALL_PKG_CONFIG_MODULE_HELP_STR})
set(OPUS_INSTALL_CMAKE_CONFIG_MODULE_HELP_STR "install CMake package config module.")
-option(OPUS_INSTALL_CMAKE_CONFIG_MODULE ${OPUS_INSTALL_CMAKE_CONFIG_MODULE_HELP_STR} ON)
+option(OPUS_INSTALL_CMAKE_CONFIG_MODULE ${OPUS_INSTALL_CMAKE_CONFIG_MODULE_HELP_STR} ${OPUS_INSTALL})
add_feature_info(OPUS_INSTALL_CMAKE_CONFIG_MODULE OPUS_INSTALL_CMAKE_CONFIG_MODULE ${OPUS_INSTALL_CMAKE_CONFIG_MODULE_HELP_STR})
set(OPUS_DRED_HELP_STR "enable DRED.")
@@ -613,53 +624,58 @@ if(OPUS_BUILD_FRAMEWORK)
OUTPUT_NAME Opus)
endif()
-install(TARGETS opus
- EXPORT OpusTargets
- ARCHIVE DESTINATION ${CMAKE_INSTALL_LIBDIR}
- LIBRARY DESTINATION ${CMAKE_INSTALL_LIBDIR}
- RUNTIME DESTINATION ${CMAKE_INSTALL_BINDIR}
- FRAMEWORK DESTINATION ${CMAKE_INSTALL_PREFIX}
- PUBLIC_HEADER DESTINATION ${CMAKE_INSTALL_INCLUDEDIR}/opus)
-
-if(OPUS_INSTALL_PKG_CONFIG_MODULE)
- set(prefix ${CMAKE_INSTALL_PREFIX})
- set(exec_prefix ${CMAKE_INSTALL_PREFIX})
- set(libdir ${CMAKE_INSTALL_FULL_LIBDIR})
- set(includedir ${CMAKE_INSTALL_FULL_INCLUDEDIR})
- set(VERSION ${PACKAGE_VERSION})
- if(HAVE_LIBM)
- set(LIBM "-lm")
+if (OPUS_INSTALL)
+ install(TARGETS opus
+ EXPORT OpusTargets
+ ARCHIVE DESTINATION ${CMAKE_INSTALL_LIBDIR}
+ LIBRARY DESTINATION ${CMAKE_INSTALL_LIBDIR}
+ RUNTIME DESTINATION ${CMAKE_INSTALL_BINDIR}
+ FRAMEWORK DESTINATION ${CMAKE_INSTALL_PREFIX}
+ PUBLIC_HEADER DESTINATION ${CMAKE_INSTALL_INCLUDEDIR}/opus)
+
+ if(OPUS_INSTALL_PKG_CONFIG_MODULE)
+ set(prefix ${CMAKE_INSTALL_PREFIX})
+ set(exec_prefix ${CMAKE_INSTALL_PREFIX})
+ set(libdir ${CMAKE_INSTALL_FULL_LIBDIR})
+ set(includedir ${CMAKE_INSTALL_FULL_INCLUDEDIR})
+ set(VERSION ${PACKAGE_VERSION})
+ if(HAVE_LIBM)
+ set(LIBM "-lm")
+ endif()
+ configure_file(opus.pc.in opus.pc)
+ install(FILES ${CMAKE_CURRENT_BINARY_DIR}/opus.pc
+ DESTINATION ${CMAKE_INSTALL_LIBDIR}/pkgconfig)
+ endif()
+
+ if(OPUS_INSTALL_CMAKE_CONFIG_MODULE)
+ set(CPACK_GENERATOR TGZ)
+ include(CPack)
+ set(CMAKE_INSTALL_PACKAGEDIR ${CMAKE_INSTALL_LIBDIR}/cmake/${PROJECT_NAME})
+ install(EXPORT OpusTargets
+ NAMESPACE Opus::
+ DESTINATION ${CMAKE_INSTALL_PACKAGEDIR})
+
+ include(CMakePackageConfigHelpers)
+
+ set(INCLUDE_INSTALL_DIR ${CMAKE_INSTALL_INCLUDEDIR})
+ configure_package_config_file(
+ ${PROJECT_SOURCE_DIR}/cmake/OpusConfig.cmake.in
+ OpusConfig.cmake
+ INSTALL_DESTINATION
+ ${CMAKE_INSTALL_PACKAGEDIR}
+ PATH_VARS
+ INCLUDE_INSTALL_DIR
+ INSTALL_PREFIX
+ ${CMAKE_INSTALL_PREFIX})
+
+ write_basic_package_version_file(OpusConfigVersion.cmake
+ VERSION ${PROJECT_VERSION}
+ COMPATIBILITY SameMajorVersion)
+
+ install(FILES ${CMAKE_CURRENT_BINARY_DIR}/OpusConfig.cmake
+ ${CMAKE_CURRENT_BINARY_DIR}/OpusConfigVersion.cmake
+ DESTINATION ${CMAKE_INSTALL_PACKAGEDIR})
endif()
- configure_file(opus.pc.in opus.pc)
- install(FILES ${CMAKE_CURRENT_BINARY_DIR}/opus.pc
- DESTINATION ${CMAKE_INSTALL_LIBDIR}/pkgconfig)
-endif()
-
-if(OPUS_INSTALL_CMAKE_CONFIG_MODULE)
- set(CPACK_GENERATOR TGZ)
- include(CPack)
- set(CMAKE_INSTALL_PACKAGEDIR ${CMAKE_INSTALL_LIBDIR}/cmake/${PROJECT_NAME})
- install(EXPORT OpusTargets
- NAMESPACE Opus::
- DESTINATION ${CMAKE_INSTALL_PACKAGEDIR})
-
- include(CMakePackageConfigHelpers)
-
- set(INCLUDE_INSTALL_DIR ${CMAKE_INSTALL_INCLUDEDIR})
- configure_package_config_file(${PROJECT_SOURCE_DIR}/cmake/OpusConfig.cmake.in
- OpusConfig.cmake
- INSTALL_DESTINATION
- ${CMAKE_INSTALL_PACKAGEDIR}
- PATH_VARS
- INCLUDE_INSTALL_DIR
- INSTALL_PREFIX
- ${CMAKE_INSTALL_PREFIX})
- write_basic_package_version_file(OpusConfigVersion.cmake
- VERSION ${PROJECT_VERSION}
- COMPATIBILITY SameMajorVersion)
- install(FILES ${CMAKE_CURRENT_BINARY_DIR}/OpusConfig.cmake
- ${CMAKE_CURRENT_BINARY_DIR}/OpusConfigVersion.cmake
- DESTINATION ${CMAKE_INSTALL_PACKAGEDIR})
endif()
if(OPUS_BUILD_PROGRAMS)
@@ -1,112 +0,0 @@
diff --git a/include/vk_mem_alloc.h b/include/vk_mem_alloc.h
index 8df0364..4856064 100644
--- a/include/vk_mem_alloc.h
+++ b/include/vk_mem_alloc.h
@@ -3017,7 +3017,7 @@ remove them if not needed.
#if defined(__ANDROID_API__) && (__ANDROID_API__ < 16)
#include <cstdlib>
-static void* vma_aligned_alloc(size_t alignment, size_t size)
+static inline void* vma_aligned_alloc(size_t alignment, size_t size)
{
// alignment must be >= sizeof(void*)
if(alignment < sizeof(void*))
@@ -3860,7 +3860,7 @@ Returned value is the found element, if present in the collection or place where
new element with value (key) should be inserted.
*/
template <typename CmpLess, typename IterT, typename KeyT>
-static IterT VmaBinaryFindFirstNotLess(IterT beg, IterT end, const KeyT& key, const CmpLess& cmp)
+static inline IterT VmaBinaryFindFirstNotLess(IterT beg, IterT end, const KeyT& key, const CmpLess& cmp)
{
size_t down = 0;
size_t up = size_t(end - beg);
@@ -3898,7 +3898,7 @@ Warning! O(n^2) complexity. Use only inside VMA_HEAVY_ASSERT.
T must be pointer type, e.g. VmaAllocation, VmaPool.
*/
template<typename T>
-static bool VmaValidatePointerArray(uint32_t count, const T* arr)
+static inline bool VmaValidatePointerArray(uint32_t count, const T* arr)
{
for (uint32_t i = 0; i < count; ++i)
{
@@ -4188,13 +4188,13 @@ static void VmaFree(const VkAllocationCallbacks* pAllocationCallbacks, void* ptr
}
template<typename T>
-static T* VmaAllocate(const VkAllocationCallbacks* pAllocationCallbacks)
+static inline T* VmaAllocate(const VkAllocationCallbacks* pAllocationCallbacks)
{
return (T*)VmaMalloc(pAllocationCallbacks, sizeof(T), VMA_ALIGN_OF(T));
}
template<typename T>
-static T* VmaAllocateArray(const VkAllocationCallbacks* pAllocationCallbacks, size_t count)
+static inline T* VmaAllocateArray(const VkAllocationCallbacks* pAllocationCallbacks, size_t count)
{
return (T*)VmaMalloc(pAllocationCallbacks, sizeof(T) * count, VMA_ALIGN_OF(T));
}
@@ -4204,14 +4204,14 @@ static T* VmaAllocateArray(const VkAllocationCallbacks* pAllocationCallbacks, si
#define vma_new_array(allocator, type, count) new(VmaAllocateArray<type>((allocator), (count)))(type)
template<typename T>
-static void vma_delete(const VkAllocationCallbacks* pAllocationCallbacks, T* ptr)
+static inline void vma_delete(const VkAllocationCallbacks* pAllocationCallbacks, T* ptr)
{
ptr->~T();
VmaFree(pAllocationCallbacks, ptr);
}
template<typename T>
-static void vma_delete_array(const VkAllocationCallbacks* pAllocationCallbacks, T* ptr, size_t count)
+static inline void vma_delete_array(const VkAllocationCallbacks* pAllocationCallbacks, T* ptr, size_t count)
{
if (ptr != VMA_NULL)
{
@@ -4658,13 +4658,13 @@ void VmaVector<T, AllocatorT>::remove(size_t index)
#endif // _VMA_VECTOR_FUNCTIONS
template<typename T, typename allocatorT>
-static void VmaVectorInsert(VmaVector<T, allocatorT>& vec, size_t index, const T& item)
+static inline void VmaVectorInsert(VmaVector<T, allocatorT>& vec, size_t index, const T& item)
{
vec.insert(index, item);
}
template<typename T, typename allocatorT>
-static void VmaVectorRemove(VmaVector<T, allocatorT>& vec, size_t index)
+static inline void VmaVectorRemove(VmaVector<T, allocatorT>& vec, size_t index)
{
vec.remove(index);
}
@@ -10620,19 +10620,19 @@ static void VmaFree(VmaAllocator hAllocator, void* ptr)
}
template<typename T>
-static T* VmaAllocate(VmaAllocator hAllocator)
+static inline T* VmaAllocate(VmaAllocator hAllocator)
{
return (T*)VmaMalloc(hAllocator, sizeof(T), VMA_ALIGN_OF(T));
}
template<typename T>
-static T* VmaAllocateArray(VmaAllocator hAllocator, size_t count)
+static inline T* VmaAllocateArray(VmaAllocator hAllocator, size_t count)
{
return (T*)VmaMalloc(hAllocator, sizeof(T) * count, VMA_ALIGN_OF(T));
}
template<typename T>
-static void vma_delete(VmaAllocator hAllocator, T* ptr)
+static inline void vma_delete(VmaAllocator hAllocator, T* ptr)
{
if(ptr != VMA_NULL)
{
@@ -10642,7 +10642,7 @@ static void vma_delete(VmaAllocator hAllocator, T* ptr)
}
template<typename T>
-static void vma_delete_array(VmaAllocator hAllocator, T* ptr, size_t count)
+static inline void vma_delete_array(VmaAllocator hAllocator, T* ptr, size_t count)
{
if(ptr != VMA_NULL)
{
+1 -17
View File
@@ -3,7 +3,6 @@
cmake_minimum_required(VERSION 3.31)
set(CMAKE_OSX_DEPLOYMENT_TARGET "15.0" CACHE STRING "macOS deployment target")
project(yuzu)
list(APPEND CMAKE_MODULE_PATH "${CMAKE_CURRENT_SOURCE_DIR}/CMakeModules")
@@ -462,21 +461,6 @@ if (NOT YUZU_STATIC_ROOM)
if (ZLIB_ADDED)
add_library(ZLIB::ZLIB ALIAS zlibstatic)
endif()
# Opus
AddJsonPackage(opus)
if (Opus_ADDED)
if (MSVC AND CXX_CLANG)
target_compile_options(opus PRIVATE
$<$<COMPILE_LANGUAGE:C,CXX>:-Wno-implicit-function-declaration>
)
endif()
endif()
if (NOT TARGET Opus::opus)
add_library(Opus::opus ALIAS opus)
endif()
endif()
if(NOT TARGET Boost::headers)
@@ -513,7 +497,7 @@ endfunction()
# =============================================
if (APPLE)
foreach(fw Carbon Metal Cocoa IOKit CoreVideo CoreMedia Security UniformTypeIdentifiers Foundation)
foreach(fw Carbon Metal Cocoa IOKit CoreVideo CoreMedia Security UniformTypeIdentifiers)
find_library(${fw}_LIBRARY ${fw} REQUIRED)
list(APPEND PLATFORM_LIBRARIES ${${fw}_LIBRARY})
endforeach()
-22
View File
@@ -1,22 +0,0 @@
# SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
# SPDX-License-Identifier: GPL-3.0-or-later
# SPDX-FileCopyrightText: 2022 yuzu Emulator Project
# SPDX-License-Identifier: GPL-2.0-or-later
find_package(PkgConfig QUIET)
pkg_search_module(OPUS QUIET IMPORTED_TARGET opus)
include(FindPackageHandleStandardArgs)
find_package_handle_standard_args(Opus
REQUIRED_VARS OPUS_LINK_LIBRARIES
VERSION_VAR OPUS_VERSION
)
if (MSYS2)
FixMsysPath(PkgConfig::OPUS)
endif()
if (Opus_FOUND AND NOT TARGET Opus::opus)
add_library(Opus::opus ALIAS PkgConfig::OPUS)
endif()
+11 -21
View File
@@ -1,4 +1,12 @@
{
"": {
"ci": true,
"hash": "9f50d993c39529e022ad456163de91ac5934e16faf8fc348f305355fc0a643c534f87ded707e11bd03bcbc0a1760bd20852b6533a67ac0558a078385181cb184",
"name": "SDL3",
"package": "SDL3",
"repo": "crueter-ci/SDL3",
"version": "3.4.14-1788231389-147a8ee32d"
},
"biscuit": {
"hash": "1229f345b014f7ca544dedb4edb3311e41ba736f9aa9a67f88b5f26f3c983288c6bb6cdedcfb0b8a02c63088a37e6a0d7ba97d9c2a4d721b213916327cffe28a",
"min_version": "0.9.1",
@@ -60,10 +68,10 @@
},
"discord-rpc": {
"find_args": "MODULE",
"hash": "8d680b3a16d6f6bf292ad823cf8635595ff986f2a49f758e3009f505b540a9d75194a8cf44cddea75736a8c3e3b5160166e716a0939ce3f18cc463cf648753fe",
"hash": "8213c43dcb0f7d479f5861091d111ed12fbdec1e62e6d729d65a4bc181d82f48a35d5fd3cd5c291f2393ac7c9681eabc6b76609755f55376284c8a8d67e148f3",
"package": "DiscordRPC",
"repo": "eden-emulator/discord-rpc",
"version": "76616d8675"
"version": "0d8b2d6a37"
},
"enet": {
"find_args": "MODULE",
@@ -82,7 +90,7 @@
"name": "ffmpeg",
"package": "FFmpeg",
"repo": "crueter-ci/FFmpeg",
"version": "9.0.1-1788120736-bf1b838f2a"
"version": "9.0.1-1788303113-bf1b838f2a"
},
"fmt": {
"hash": "f0da82c545b01692e9fd30fdfb613dbb8dd9716983dcd0ff19ac2a8d36f74beb5540ef38072fdecc1e34191b3682a8542ecbf3a61ef287dbba0a2679d4e023f2",
@@ -210,21 +218,6 @@
"repo": "jimmy-park/openssl-cmake",
"version": "3.6.2"
},
"opus": {
"find_args": "MODULE",
"hash": "9506147b0de35befda8633ff272981cc2575c860874791bd455b752f797fd7dbd1079f0ba42ccdd7bb1fe6773fa5e84b3d75667c2883dd1fb2d0e4a5fa4f8387",
"min_version": "1.3",
"options": [
"OPUS_PRESUME_NEON ON"
],
"package": "Opus",
"patches": [
"0001-disable-clang-runtime-neon.patch",
"0002-no-install.patch"
],
"repo": "xiph/opus",
"version": "a3f0ec02b3"
},
"quazip": {
"hash": "609c240c7f029ac26a37d8fbab51bc16284e05e128b78b9b9c0e95d083538c36047a67d682759ac990e4adb0eeb90f04f1ea7fe2253bbda7e7e3bcce32e53dd8",
"min_version": "1.3",
@@ -303,9 +296,6 @@
"find_args": "CONFIG",
"hash": "deb5902ef8db0e329fbd5f3f4385eb0e26bdd9f14f3a2334823fb3fe18f36bc5d235d620d6e5f6fe3551ec3ea7038638899db8778c09f6d5c278f5ff95c3344b",
"package": "VulkanMemoryAllocator",
"patches": [
"0001-macos-clang.patch"
],
"repo": "GPUOpen-LibrariesAndSDKs/VulkanMemoryAllocator",
"version": "v3.3.0"
},
+9 -10
View File
@@ -60,7 +60,6 @@ All other dependencies will be downloaded and built by [CPM](https://github.com/
* [ZLIB](https://www.zlib.net/) 1.2+
* [zstd](https://facebook.github.io/zstd/) 1.5+
* [enet](http://enet.bespin.org/) 1.3+
* [Opus](https://opus-codec.org/) 1.3+
Vulkan 1.3.274+ is also needed:
@@ -121,7 +120,7 @@ sudo emerge -a \
dev-libs/boost dev-libs/openssl dev-libs/discord-rpc \
dev-util/spirv-tools dev-util/spirv-headers dev-util/vulkan-headers \
dev-util/vulkan-utility-libraries dev-util/glslang \
media-gfx/renderdoc media-libs/libva media-libs/opus media-video/ffmpeg \
media-gfx/renderdoc media-libs/libva media-video/ffmpeg \
media-libs/VulkanMemoryAllocator media-libs/libsdl3 media-libs/cubeb \
net-libs/enet \
sys-libs/zlib \
@@ -153,7 +152,7 @@ Required USE flags:
<summary>Arch Linux</summary>
```sh
sudo pacman -Syu --needed base-devel boost catch2 cmake enet ffmpeg fmt git glslang libzip lz4 ninja nlohmann-json openssl opus qt6-base qt6-multimedia qt6-charts sdl3 zlib zstd zip unzip vulkan-headers vulkan-utility-libraries libusb spirv-tools spirv-headers
sudo pacman -Syu --needed base-devel boost catch2 cmake enet ffmpeg fmt git glslang libzip lz4 ninja nlohmann-json openssl qt6-base qt6-multimedia qt6-charts sdl3 zlib zstd zip unzip vulkan-headers vulkan-utility-libraries libusb spirv-tools spirv-headers
```
* Building with QT Web Engine requires `qt6-webengine` as well.
@@ -166,7 +165,7 @@ sudo pacman -Syu --needed base-devel boost catch2 cmake enet ffmpeg fmt git glsl
<summary>Ubuntu, Debian, Mint Linux</summary>
```sh
sudo apt-get install autoconf cmake g++ gcc git glslang-tools libglu1-mesa-dev libhidapi-dev libpulse-dev libtool libudev-dev libxcb-icccm4 libxcb-image0 libxcb-keysyms1 libxcb-render-util0 libxcb-xinerama0 libxcb-xkb1 libxext-dev libxkbcommon-x11-0 mesa-common-dev nasm ninja-build qt6-base-private-dev catch2 libfmt-dev liblz4-dev nlohmann-json3-dev libzstd-dev libssl-dev libavfilter-dev libavcodec-dev libswscale-dev pkg-config zlib1g-dev libva-dev libvdpau-dev qt6-tools-dev qt6-charts-dev libvulkan-dev spirv-tools spirv-headers libusb-1.0-0-dev libxbyak-dev libboost-dev libboost-fiber-dev libboost-context-dev libsdl3-dev libopus-dev libasound2t64 vulkan-utility-libraries-dev
sudo apt-get install autoconf cmake g++ gcc git glslang-tools libglu1-mesa-dev libhidapi-dev libpulse-dev libtool libudev-dev libxcb-icccm4 libxcb-image0 libxcb-keysyms1 libxcb-render-util0 libxcb-xinerama0 libxcb-xkb1 libxext-dev libxkbcommon-x11-0 mesa-common-dev nasm ninja-build qt6-base-private-dev catch2 libfmt-dev liblz4-dev nlohmann-json3-dev libzstd-dev libssl-dev libavfilter-dev libavcodec-dev libswscale-dev pkg-config zlib1g-dev libva-dev libvdpau-dev qt6-tools-dev qt6-charts-dev libvulkan-dev spirv-tools spirv-headers libusb-1.0-0-dev libxbyak-dev libboost-dev libboost-fiber-dev libboost-context-dev libsdl3-dev libasound2t64 vulkan-utility-libraries-dev
```
* Ubuntu 26.04, Linux Mint 22.3, or Debian 13 or later is required.
@@ -213,7 +212,7 @@ First, enable the community repository; [see here](https://wiki.alpinelinux.org/
# Enable the community repository
setup-apkrepos -c
# Install
apk add g++ git cmake make mesa-dev qt6-qtbase-dev qt6-qtbase-private-dev libquazip1-qt6 ffmpeg-dev qt6-charts-dev libusb-dev libtool boost-dev sdl3-dev zstd-dev vulkan-utility-libraries spirv-tools-dev openssl-dev nlohmann-json lz4-dev opus-dev jq patch
apk add g++ git cmake make mesa-dev qt6-qtbase-dev qt6-qtbase-private-dev libquazip1-qt6 ffmpeg-dev qt6-charts-dev libusb-dev libtool boost-dev sdl3-dev zstd-dev vulkan-utility-libraries spirv-tools-dev openssl-dev nlohmann-json lz4-dev jq patch
```
</details>
@@ -261,7 +260,7 @@ brew install molten-vk
As root run:
```sh
pkg install devel/cmake devel/sdl3 devel/boost-libs devel/catch2 devel/libfmt devel/nlohmann-json devel/ninja devel/nasm devel/autoconf devel/pkgconf devel/qt6-base x11-toolkits/qt6-charts devel/simpleini net/enet multimedia/ffnvcodec-headers multimedia/ffmpeg audio/opus archivers/liblz4 lang/gcc12 graphics/glslang graphics/vulkan-utility-libraries graphics/spirv-tools www/cpp-httplib graphics/vulkan-utility-libraries graphics/vulkan-headers graphics/spirv-headers quazip-qt6
pkg install devel/cmake devel/sdl3 devel/boost-libs devel/catch2 devel/libfmt devel/nlohmann-json devel/ninja devel/nasm devel/autoconf devel/pkgconf devel/qt6-base x11-toolkits/qt6-charts devel/simpleini net/enet multimedia/ffnvcodec-headers multimedia/ffmpeg archivers/liblz4 lang/gcc12 graphics/glslang graphics/vulkan-utility-libraries graphics/spirv-tools www/cpp-httplib graphics/vulkan-utility-libraries graphics/vulkan-headers graphics/spirv-headers quazip-qt6
```
If using FreeBSD 12 or prior, use `devel/pkg-config` instead.
@@ -275,7 +274,7 @@ If using FreeBSD 12 or prior, use `devel/pkg-config` instead.
For NetBSD +10.1:
```sh
pkgin install git cmake boost fmtlib SDL3 catch2 libjwt spirv-headers spirv-tools ffmpeg7 libva nlohmann-json jq libopus qt6-qtbase qt6-qtcharts qt6-qtmultimedia qt6-qttools cpp-httplib lz4 vulkan-headers nasm autoconf enet pkg-config libusb1 libcxx frozen
pkgin install git cmake boost fmtlib SDL3 catch2 libjwt spirv-headers spirv-tools ffmpeg7 libva nlohmann-json jq qt6-qtbase qt6-qtcharts qt6-qtmultimedia qt6-qttools cpp-httplib lz4 vulkan-headers nasm autoconf enet pkg-config libusb1 libcxx frozen
```
[Caveats](./Caveats.md#netbsd).
@@ -306,7 +305,7 @@ pkg install gcc14 git cmake unzip nasm autoconf bash pkgconf ffmpeg glslang gmak
<summary>OpenIndiana</summary>
```sh
sudo pkg install git cmake qt6 boost glslang libzip library/lz4 libusb-1 nlohmann-json openssl opus sdl3 zlib compress/zstd unzip pkg-config nasm autoconf mesa library/libdrm header-drm developer/fmt
sudo pkg install git cmake qt6 boost glslang libzip library/lz4 libusb-1 nlohmann-json openssl sdl3 zlib compress/zstd unzip pkg-config nasm autoconf mesa library/libdrm header-drm developer/fmt
```
[Caveats](./Caveats.md#openindiana).
@@ -330,7 +329,7 @@ sudo pkgin install git cmake autoconf build-essential libusb-1 nasm gcc13
```sh
BASE="git make autoconf libtool automake-wrapper jq patch"
MINGW="qt6-base qt6-charts qt6-tools qt6-translations qt6-svg cmake toolchain clang python-pip openssl vulkan-memory-allocator vulkan-devel glslang boost fmt lz4 nlohmann-json zlib zstd enet opus libusb openssl SDL3"
MINGW="qt6-base qt6-charts qt6-tools qt6-translations qt6-svg cmake toolchain clang python-pip openssl vulkan-memory-allocator vulkan-devel glslang boost fmt lz4 nlohmann-json zlib zstd enet libusb openssl SDL3"
# Either x86_64 or clang-aarch64 (Windows on ARM)
packages="$BASE"
for pkg in $MINGW; do
@@ -356,7 +355,7 @@ pacman -Syuu --needed --noconfirm $packages
<summary>HaikuOS</summary>
```sh
pkgman install git cmake patch libfmt_devel nlohmann_json lz4_devel opus_devel boost1.90_devel vulkan_devel qt6_base_devel qt6_declarative_devel libsdl3_devel ffmpeg7_devel libx11_devel enet_devel catch2_devel quazip1_qt5_devel qt6_5compat_devel glslang qt6_devel qt6_charts_devel cubeb_devel simpleini quazip_qt6_devel
pkgman install git cmake patch libfmt_devel nlohmann_json lz4_devel boost1.90_devel vulkan_devel qt6_base_devel qt6_declarative_devel libsdl3_devel ffmpeg7_devel libx11_devel enet_devel catch2_devel quazip1_qt5_devel qt6_5compat_devel glslang qt6_devel qt6_charts_devel cubeb_devel simpleini quazip_qt6_devel
```
[Caveats](./Caveats.md#haikuos).
-1
View File
@@ -11,7 +11,6 @@
#include <limits>
#include <span>
#include <array>
#include <algorithm>
#include <time.h>
namespace Tz {
+1 -1
View File
@@ -12,7 +12,7 @@ pkgs.mkShellNoCC {
git cmake clang gnumake patch jq pkg-config
# libraries
openssl boost fmt nlohmann_json lz4 zlib zstd
enet libopus vulkan-headers vulkan-utility-libraries
enet vulkan-headers vulkan-utility-libraries
spirv-tools spirv-headers vulkan-loader unzip
glslang python3 httplib cpp-jwt ffmpeg-headless
libusb1 cubeb
@@ -79,13 +79,6 @@ class LicensesFragment : Fragment() {
R.string.license_ffmpeg_copyright,
R.string.license_ffmpeg_text
),
License(
R.string.license_opus,
R.string.license_opus_description,
R.string.license_opus_link,
R.string.license_opus_copyright,
R.string.license_opus_text
),
License(
R.string.license_sirit,
R.string.license_sirit_description,
@@ -1762,51 +1762,6 @@ RENDERED INACCURATE OR LOSSES SUSTAINED BY YOU OR THIRD PARTIES OR A
FAILURE OF THE LIBRARY TO OPERATE WITH ANY OTHER SOFTWARE), EVEN IF
SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH
DAMAGES.
</string>
<string name="license_opus" translatable="false">Opus</string>
<string name="license_opus_description" translatable="false">Modern audio compression for the internet</string>
<string name="license_opus_link" translatable="false">https://github.com/xiph/opus</string>
<string name="license_opus_copyright" translatable="false">Copyright 20012011 Xiph.Org, Skype Limited, Octasic, Jean-Marc Valin, Timothy B. Terriberry, CSIRO, Gregory Maxwell, Mark Borgerding, Erik de Castro Lopo</string>
<string name="license_opus_text" translatable="false">
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions
are met:\n\n
- Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.\n\n
- Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.\n\n
- Neither the name of Internet Society, IETF or IETF Trust, nor the
names of specific contributors, may be used to endorse or promote
products derived from this software without specific prior written
permission.\n\n
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
``AS IS\'\' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER
OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.\n\n
Opus is subject to the royalty-free patent licenses which are
specified at:\n\n
Xiph.Org Foundation:
https://datatracker.ietf.org/ipr/1524/ \n\n
Microsoft Corporation:
https://datatracker.ietf.org/ipr/1914/ \n\n
Broadcom Corporation:
https://datatracker.ietf.org/ipr/1526/
</string>
<string name="license_sirit" translatable="false">Sirit</string>
<string name="license_sirit_description" translatable="false">A runtime SPIR-V assembler</string>
+8 -2
View File
@@ -226,8 +226,14 @@ else()
$<$<COMPILE_LANGUAGE:C,CXX>:-Wno-sign-conversion>)
endif()
target_include_directories(audio_core PRIVATE ${OPUS_INCLUDE_DIRS})
target_link_libraries(audio_core PUBLIC common core Opus::opus)
if (YUZU_USE_EXTERNAL_FFMPEG)
add_dependencies(audio_core ffmpeg-build)
endif()
target_include_directories(audio_core PUBLIC ${FFmpeg_INCLUDE_DIR})
target_link_libraries(audio_core PRIVATE ${FFmpeg_LIBRARIES})
target_link_options(audio_core PRIVATE ${FFmpeg_LDFLAGS})
target_link_libraries(audio_core PUBLIC common core)
if (ENABLE_CUBEB)
target_sources(audio_core PRIVATE
@@ -1,107 +1,72 @@
// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
// SPDX-FileCopyrightText: Copyright 2023 yuzu Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
extern "C" {
#include <libavcodec/avcodec.h>
#include <libavcodec/codec.h>
#include <libavcodec/packet.h>
#include <libavutil/channel_layout.h>
#include <libavutil/frame.h>
}
#include "audio_core/adsp/apps/opus/opus_decode_object.h"
#include "common/assert.h"
#include "core/hle/service/audio/errors.h"
namespace AudioCore::ADSP::OpusDecoder {
namespace {
bool IsValidChannelCount(u32 channel_count) {
return channel_count == 1 || channel_count == 2;
}
} // namespace
u32 OpusDecodeObject::GetWorkBufferSize(u32 channel_count) {
if (!IsValidChannelCount(channel_count)) {
return 0;
}
return static_cast<u32>(sizeof(OpusDecodeObject)) + opus_decoder_get_size(channel_count);
if (channel_count == 1 || channel_count == 2)
return u32(sizeof(OpusDecodeObject)) + 16 * channel_count;
return 0;
}
OpusDecodeObject& OpusDecodeObject::Initialize(u64 buffer, u64 buffer2) {
auto* new_decoder = reinterpret_cast<OpusDecodeObject*>(buffer);
auto* comparison = reinterpret_cast<OpusDecodeObject*>(buffer2);
if (new_decoder->magic == DecodeObjectMagic) {
if (!new_decoder->initialized ||
(new_decoder->initialized && new_decoder->self == comparison)) {
new_decoder->state_valid = true;
}
} else {
new_decoder->initialized = false;
new_decoder->state_valid = true;
}
return *new_decoder;
Result OpusDecodeObject::InitializeDecoder(u32 sample_rate, u32 channel_count) {
if ((codec = avcodec_find_decoder(AV_CODEC_ID_OPUS)) == nullptr)
return Service::Audio::ResultLibOpusInvalidState;
if ((avctx = avcodec_alloc_context3(codec)) == nullptr)
return Service::Audio::ResultLibOpusInvalidState;
avctx->sample_rate = sample_rate;
av_channel_layout_default(&avctx->ch_layout, channel_count);
return ResultSuccess;
}
s32 OpusDecodeObject::InitializeDecoder(u32 sample_rate, u32 channel_count) {
if (!state_valid) {
return OPUS_INVALID_STATE;
Result OpusDecodeObject::Shutdown() {
if (avctx) {
avcodec_free_context(&avctx);
avctx = nullptr;
}
if (initialized) {
return OPUS_OK;
}
// Unfortunately libopus does not expose the OpusDecoder struct publicly, so we can't include
// it in this class. Nintendo does not allocate memory, which is why we have a workbuffer
// provided.
// We could use _create and have libopus allocate it for us, but then we have to separately
// track which decoder is being used between this and multistream in order to call the correct
// destroy from the host side.
// This is a bit cringe, but is safe as these objects are only ever initialized inside the given
// workbuffer, and GetWorkBufferSize will guarantee there's enough space to follow.
decoder = (LibOpusDecoder*)(this + 1);
s32 ret = opus_decoder_init(decoder, sample_rate, channel_count);
if (ret == OPUS_OK) {
magic = DecodeObjectMagic;
initialized = true;
state_valid = true;
self = this;
final_range = 0;
}
return ret;
return ResultSuccess;
}
s32 OpusDecodeObject::Shutdown() {
if (!state_valid) {
return OPUS_INVALID_STATE;
Result OpusDecodeObject::ResetDecoder() {
if (avctx) {
avcodec_flush_buffers(avctx);
return ResultSuccess;
}
if (initialized) {
magic = 0x0;
initialized = false;
state_valid = false;
self = nullptr;
final_range = 0;
decoder = nullptr;
}
return OPUS_OK;
return Service::Audio::ResultLibOpusInvalidState;
}
s32 OpusDecodeObject::ResetDecoder() {
return opus_decoder_ctl(decoder, OPUS_RESET_STATE);
}
s32 OpusDecodeObject::Decode(u32& out_sample_count, u64 output_data, u64 output_data_size,
u64 input_data, u64 input_data_size) {
ASSERT(initialized);
Result OpusDecodeObject::Decode(u32& out_sample_count, u64 output_data, u64 output_data_size, u64 input_data, u64 input_data_size) {
out_sample_count = 0;
if (avctx) {
AVPacket* avpkt = av_packet_alloc();
av_new_packet(avpkt, int(input_data_size));
std::memcpy(avpkt->data, reinterpret_cast<const u8*>(input_data), input_data_size);
avcodec_send_packet(avctx, avpkt);
if (!state_valid) {
return OPUS_INVALID_STATE;
AVFrame* frame = av_frame_alloc();
avcodec_receive_frame(avctx, frame);
std::memcpy(reinterpret_cast<u16*>(output_data), frame->data, output_data_size);
av_frame_free(&frame);
av_packet_free(&avpkt);
out_sample_count = frame->nb_samples;
return ResultSuccess;
}
auto ret_code_or_samples = opus_decode(
decoder, reinterpret_cast<const u8*>(input_data), static_cast<opus_int32>(input_data_size),
reinterpret_cast<opus_int16*>(output_data), static_cast<opus_int32>(output_data_size), 0);
if (ret_code_or_samples < OPUS_OK) {
return ret_code_or_samples;
}
out_sample_count = ret_code_or_samples;
return opus_decoder_ctl(decoder, OPUS_GET_FINAL_RANGE_REQUEST, &final_range);
return Service::Audio::ResultLibOpusInvalidState;
}
} // namespace AudioCore::ADSP::OpusDecoder
@@ -1,38 +1,30 @@
// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
// SPDX-FileCopyrightText: Copyright 2023 yuzu Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
#pragma once
#include <opus.h>
#include "common/common_types.h"
#include "core/hle/result.h"
struct AVCodec;
struct AVCodecContext;
namespace AudioCore::ADSP::OpusDecoder {
using LibOpusDecoder = ::OpusDecoder;
static constexpr u32 DecodeObjectMagic = 0xDEADBEEF;
class OpusDecodeObject {
public:
static u32 GetWorkBufferSize(u32 channel_count);
static OpusDecodeObject& Initialize(u64 buffer, u64 buffer2);
s32 InitializeDecoder(u32 sample_rate, u32 channel_count);
s32 Shutdown();
s32 ResetDecoder();
s32 Decode(u32& out_sample_count, u64 output_data, u64 output_data_size, u64 input_data,
u64 input_data_size);
u32 GetFinalRange() const noexcept {
return final_range;
}
Result InitializeDecoder(u32 sample_rate, u32 channel_count);
Result Shutdown();
Result ResetDecoder();
Result Decode(u32& out_sample_count, u64 output_data, u64 output_data_size, u64 input_data, u64 input_data_size);
private:
u32 magic;
bool initialized;
bool state_valid;
OpusDecodeObject* self;
u32 final_range;
LibOpusDecoder* decoder;
AVCodec const* codec = nullptr;
AVCodecContext* avctx = nullptr;
};
static_assert(std::is_trivially_constructible_v<OpusDecodeObject>);
} // namespace AudioCore::ADSP::OpusDecoder
+193 -196
View File
@@ -5,17 +5,16 @@
// SPDX-License-Identifier: GPL-2.0-or-later
#include <array>
#include <chrono>
#include "audio_core/adsp/apps/opus/opus_decode_object.h"
#include "audio_core/adsp/apps/opus/opus_multistream_decode_object.h"
#include "audio_core/adsp/apps/opus/shared_memory.h"
#include "audio_core/audio_core.h"
#include "audio_core/common/common.h"
#include "common/logging.h"
#include "common/thread.h"
#include "core/core.h"
#include "core/core_timing.h"
#include "core/hle/service/audio/errors.h"
namespace AudioCore::ADSP::OpusDecoder {
@@ -49,11 +48,16 @@ OpusDecoder::~OpusDecoder() {
// Shutdown the thread
Send(Direction::DSP, Message::Shutdown);
auto msg = Receive(Direction::Host);
ASSERT_MSG(msg == Message::ShutdownOK, "Expected Opus shutdown code {}, got {}",
Message::ShutdownOK, msg);
ASSERT_MSG(msg == Message::ShutdownOK, "Expected Opus shutdown code {}, got {}", Message::ShutdownOK, msg);
main_thread.request_stop();
main_thread.join();
running = false;
// Must shutdown as there are AV allocations which are manual
for (auto& e : decode_objects)
e.second.Shutdown();
for (auto& e : ms_decode_objects)
e.second.Shutdown();
}
void OpusDecoder::Send(Direction dir, u32 message) {
@@ -68,202 +72,195 @@ void OpusDecoder::Init(std::stop_token stop_token) {
Common::SetCurrentThreadName("DSP_OpusDecoder_Init");
if (Receive(Direction::DSP, stop_token) != Message::Start) {
LOG_ERROR(Service_Audio,
"DSP OpusDecoder failed to receive Start message. Opus initialization failed.");
LOG_ERROR(Service_Audio, "DSP OpusDecoder failed to receive Start message. Opus initialization failed.");
return;
}
main_thread = std::jthread([this](std::stop_token st) { Main(st); });
// Main OpusDecoder thread, responsible for processing the incoming Opus packets.
main_thread = std::jthread([this](std::stop_token thread_stop_token) {
Common::SetCurrentThreadName("DSP_OpusDecoder_Main");
while (!thread_stop_token.stop_requested()) {
auto msg = Receive(Direction::DSP, thread_stop_token);
switch (msg) {
case Shutdown:
Send(Direction::Host, Message::ShutdownOK);
return;
case GetWorkBufferSize: {
auto channel_count = s32(shared_memory->host_send_data[0]);
ASSERT(IsValidChannelCount(channel_count));
shared_memory->dsp_return_data[0] = OpusDecodeObject::GetWorkBufferSize(channel_count);
Send(Direction::Host, Message::GetWorkBufferSizeOK);
break;
}
case InitializeDecodeObject: {
auto buffer = shared_memory->host_send_data[0];
auto buffer_size = shared_memory->host_send_data[1];
auto sample_rate = s32(shared_memory->host_send_data[2]);
auto channel_count = s32(shared_memory->host_send_data[3]);
ASSERT(sample_rate >= 0);
ASSERT(IsValidChannelCount(channel_count));
ASSERT(buffer_size >= OpusDecodeObject::GetWorkBufferSize(channel_count));
if (auto const it = decode_objects.find(buffer); it != decode_objects.end()) {
it->second.Shutdown();
shared_memory->dsp_return_data[0] = it->second.InitializeDecoder(sample_rate, channel_count).raw;
} else {
OpusDecodeObject obj{};
shared_memory->dsp_return_data[0] = obj.InitializeDecoder(sample_rate, channel_count).raw;
decode_objects.insert_or_assign(buffer, obj);
}
Send(Direction::Host, Message::InitializeDecodeObjectOK);
break;
}
case ShutdownDecodeObject: {
auto buffer = shared_memory->host_send_data[0];
//[[maybe_unused]] auto buffer_size = shared_memory->host_send_data[1];
if (auto const it = decode_objects.find(buffer); it != decode_objects.end()) {
shared_memory->dsp_return_data[0] = it->second.Shutdown().raw;
} else {
LOG_ERROR(Audio_DSP, "operating unregistered buffer {}", buffer);
shared_memory->dsp_return_data[0] = Service::Audio::ResultLibOpusInvalidState.raw;
}
Send(Direction::Host, Message::ShutdownDecodeObjectOK);
break;
}
case DecodeInterleaved: {
auto start_time = system.CoreTiming().GetGlobalTimeUs();
auto buffer = shared_memory->host_send_data[0];
auto input_data = shared_memory->host_send_data[1];
auto input_data_size = shared_memory->host_send_data[2];
auto output_data = shared_memory->host_send_data[3];
auto output_data_size = shared_memory->host_send_data[4];
//auto final_range = static_cast<u32>(shared_memory->host_send_data[5]);
auto reset_requested = shared_memory->host_send_data[6];
u32 decoded_samples{0};
if (auto const it = decode_objects.find(buffer); it != decode_objects.end()) {
auto res = ResultSuccess;
if (reset_requested)
res = it->second.ResetDecoder();
if (res == ResultSuccess)
res = it->second.Decode(decoded_samples, output_data, output_data_size, input_data, input_data_size);
auto end_time = system.CoreTiming().GetGlobalTimeUs();
shared_memory->dsp_return_data[0] = res.raw;
shared_memory->dsp_return_data[1] = decoded_samples;
shared_memory->dsp_return_data[2] = (end_time - start_time).count();
} else {
LOG_ERROR(Audio_DSP, "operating unregistered buffer {}", buffer);
shared_memory->dsp_return_data[0] = Service::Audio::ResultLibOpusInvalidState.raw;
}
Send(Direction::Host, Message::DecodeInterleavedOK);
break;
}
case MapMemory: {
[[maybe_unused]] auto buffer = shared_memory->host_send_data[0];
[[maybe_unused]] auto buffer_size = shared_memory->host_send_data[1];
Send(Direction::Host, Message::MapMemoryOK);
break;
}
case UnmapMemory: {
[[maybe_unused]] auto buffer = shared_memory->host_send_data[0];
[[maybe_unused]] auto buffer_size = shared_memory->host_send_data[1];
Send(Direction::Host, Message::UnmapMemoryOK);
break;
}
case GetWorkBufferSizeForMultiStream: {
auto total_stream_count = s32(shared_memory->host_send_data[0]);
auto stereo_stream_count = s32(shared_memory->host_send_data[1]);
ASSERT(IsValidMultiStreamStreamCounts(total_stream_count, stereo_stream_count));
shared_memory->dsp_return_data[0] = OpusMultiStreamDecodeObject::GetWorkBufferSize(
total_stream_count, stereo_stream_count);
Send(Direction::Host, Message::GetWorkBufferSizeForMultiStreamOK);
break;
}
case InitializeMultiStreamDecodeObject: {
auto buffer = shared_memory->host_send_data[0];
auto buffer_size = shared_memory->host_send_data[1];
auto sample_rate = s32(shared_memory->host_send_data[2]);
auto channel_count = s32(shared_memory->host_send_data[3]);
auto total_stream_count = s32(shared_memory->host_send_data[4]);
auto stereo_stream_count = s32(shared_memory->host_send_data[5]);
// Nintendo seem to have a bug here, they try to use &host_send_data[6] for the channel
// mappings, but [6] is never set, and there is not enough room in the argument data for
// more than 40 channels, when 255 are possible.
// It also means the mapping values are undefined, though likely always 0,
// and the mappings given by the game are ignored. The mappings are copied to this
// dedicated buffer host side, so let's do as intended.
auto mappings = shared_memory->channel_mapping.data();
ASSERT(IsValidMultiStreamStreamCounts(total_stream_count, stereo_stream_count));
ASSERT(sample_rate >= 0);
ASSERT(buffer_size >= OpusMultiStreamDecodeObject::GetWorkBufferSize(total_stream_count, stereo_stream_count));
if (auto const it = ms_decode_objects.find(buffer); it != ms_decode_objects.end()) {
it->second.Shutdown();
shared_memory->dsp_return_data[0] = it->second.InitializeDecoder(sample_rate, total_stream_count, channel_count, stereo_stream_count, mappings).raw;
} else {
OpusMultiStreamDecodeObject obj{};
shared_memory->dsp_return_data[0] = obj.InitializeDecoder(sample_rate, total_stream_count, channel_count, stereo_stream_count, mappings).raw;
ms_decode_objects.insert_or_assign(buffer, obj);
}
Send(Direction::Host, Message::InitializeMultiStreamDecodeObjectOK);
break;
}
case ShutdownMultiStreamDecodeObject: {
auto buffer = shared_memory->host_send_data[0];
//[[maybe_unused]] auto buffer_size = shared_memory->host_send_data[1];
if (auto const it = ms_decode_objects.find(buffer); it != ms_decode_objects.end()) {
shared_memory->dsp_return_data[0] = it->second.Shutdown().raw;
} else {
LOG_ERROR(Audio_DSP, "operating unregistered buffer {}", buffer);
shared_memory->dsp_return_data[0] = Service::Audio::ResultLibOpusInvalidState.raw;
}
Send(Direction::Host, Message::ShutdownMultiStreamDecodeObjectOK);
break;
}
case DecodeInterleavedForMultiStream: {
auto start_time = system.CoreTiming().GetGlobalTimeUs();
auto buffer = shared_memory->host_send_data[0];
auto input_data = shared_memory->host_send_data[1];
auto input_data_size = shared_memory->host_send_data[2];
auto output_data = shared_memory->host_send_data[3];
auto output_data_size = shared_memory->host_send_data[4];
//auto final_range = static_cast<u32>(shared_memory->host_send_data[5]);
auto reset_requested = shared_memory->host_send_data[6];
u32 decoded_samples{0};
if (auto const it = ms_decode_objects.find(buffer); it != ms_decode_objects.end()) {
auto res = ResultSuccess;
if (reset_requested)
res = it->second.ResetDecoder();
if (res == ResultSuccess)
res = it->second.Decode(decoded_samples, output_data, output_data_size, input_data, input_data_size);
auto end_time = system.CoreTiming().GetGlobalTimeUs();
shared_memory->dsp_return_data[0] = res.raw;
shared_memory->dsp_return_data[1] = decoded_samples;
shared_memory->dsp_return_data[2] = (end_time - start_time).count();
} else {
LOG_ERROR(Audio_DSP, "operating unregistered buffer {}", buffer);
shared_memory->dsp_return_data[0] = Service::Audio::ResultLibOpusInvalidState.raw;
}
Send(Direction::Host, Message::DecodeInterleavedForMultiStreamOK);
break;
}
default:
LOG_ERROR(Audio_DSP, "Invalid OpusDecoder command {}", msg);
continue;
}
}
});
running = true;
Send(Direction::Host, Message::StartOK);
}
void OpusDecoder::Main(std::stop_token stop_token) {
Common::SetCurrentThreadName("DSP_OpusDecoder_Main");
while (!stop_token.stop_requested()) {
auto msg = Receive(Direction::DSP, stop_token);
switch (msg) {
case Shutdown:
Send(Direction::Host, Message::ShutdownOK);
return;
case GetWorkBufferSize: {
auto channel_count = static_cast<s32>(shared_memory->host_send_data[0]);
ASSERT(IsValidChannelCount(channel_count));
shared_memory->dsp_return_data[0] = OpusDecodeObject::GetWorkBufferSize(channel_count);
Send(Direction::Host, Message::GetWorkBufferSizeOK);
} break;
case InitializeDecodeObject: {
auto buffer = shared_memory->host_send_data[0];
auto buffer_size = shared_memory->host_send_data[1];
auto sample_rate = static_cast<s32>(shared_memory->host_send_data[2]);
auto channel_count = static_cast<s32>(shared_memory->host_send_data[3]);
ASSERT(sample_rate >= 0);
ASSERT(IsValidChannelCount(channel_count));
ASSERT(buffer_size >= OpusDecodeObject::GetWorkBufferSize(channel_count));
auto& decoder_object = OpusDecodeObject::Initialize(buffer, buffer);
shared_memory->dsp_return_data[0] =
decoder_object.InitializeDecoder(sample_rate, channel_count);
Send(Direction::Host, Message::InitializeDecodeObjectOK);
} break;
case ShutdownDecodeObject: {
auto buffer = shared_memory->host_send_data[0];
[[maybe_unused]] auto buffer_size = shared_memory->host_send_data[1];
auto& decoder_object = OpusDecodeObject::Initialize(buffer, buffer);
shared_memory->dsp_return_data[0] = decoder_object.Shutdown();
Send(Direction::Host, Message::ShutdownDecodeObjectOK);
} break;
case DecodeInterleaved: {
auto start_time = system.CoreTiming().GetGlobalTimeUs();
auto buffer = shared_memory->host_send_data[0];
auto input_data = shared_memory->host_send_data[1];
auto input_data_size = shared_memory->host_send_data[2];
auto output_data = shared_memory->host_send_data[3];
auto output_data_size = shared_memory->host_send_data[4];
auto final_range = static_cast<u32>(shared_memory->host_send_data[5]);
auto reset_requested = shared_memory->host_send_data[6];
u32 decoded_samples{0};
auto& decoder_object = OpusDecodeObject::Initialize(buffer, buffer);
s32 error_code{OPUS_OK};
if (reset_requested) {
error_code = decoder_object.ResetDecoder();
}
if (error_code == OPUS_OK) {
error_code = decoder_object.Decode(decoded_samples, output_data, output_data_size,
input_data, input_data_size);
}
if (error_code == OPUS_OK) {
if (final_range && decoder_object.GetFinalRange() != final_range) {
error_code = OPUS_INVALID_PACKET;
}
}
auto end_time = system.CoreTiming().GetGlobalTimeUs();
shared_memory->dsp_return_data[0] = error_code;
shared_memory->dsp_return_data[1] = decoded_samples;
shared_memory->dsp_return_data[2] = (end_time - start_time).count();
Send(Direction::Host, Message::DecodeInterleavedOK);
} break;
case MapMemory: {
[[maybe_unused]] auto buffer = shared_memory->host_send_data[0];
[[maybe_unused]] auto buffer_size = shared_memory->host_send_data[1];
Send(Direction::Host, Message::MapMemoryOK);
} break;
case UnmapMemory: {
[[maybe_unused]] auto buffer = shared_memory->host_send_data[0];
[[maybe_unused]] auto buffer_size = shared_memory->host_send_data[1];
Send(Direction::Host, Message::UnmapMemoryOK);
} break;
case GetWorkBufferSizeForMultiStream: {
auto total_stream_count = static_cast<s32>(shared_memory->host_send_data[0]);
auto stereo_stream_count = static_cast<s32>(shared_memory->host_send_data[1]);
ASSERT(IsValidMultiStreamStreamCounts(total_stream_count, stereo_stream_count));
shared_memory->dsp_return_data[0] = OpusMultiStreamDecodeObject::GetWorkBufferSize(
total_stream_count, stereo_stream_count);
Send(Direction::Host, Message::GetWorkBufferSizeForMultiStreamOK);
} break;
case InitializeMultiStreamDecodeObject: {
auto buffer = shared_memory->host_send_data[0];
auto buffer_size = shared_memory->host_send_data[1];
auto sample_rate = static_cast<s32>(shared_memory->host_send_data[2]);
auto channel_count = static_cast<s32>(shared_memory->host_send_data[3]);
auto total_stream_count = static_cast<s32>(shared_memory->host_send_data[4]);
auto stereo_stream_count = static_cast<s32>(shared_memory->host_send_data[5]);
// Nintendo seem to have a bug here, they try to use &host_send_data[6] for the channel
// mappings, but [6] is never set, and there is not enough room in the argument data for
// more than 40 channels, when 255 are possible.
// It also means the mapping values are undefined, though likely always 0,
// and the mappings given by the game are ignored. The mappings are copied to this
// dedicated buffer host side, so let's do as intended.
auto mappings = shared_memory->channel_mapping.data();
ASSERT(IsValidMultiStreamStreamCounts(total_stream_count, stereo_stream_count));
ASSERT(sample_rate >= 0);
ASSERT(buffer_size >= OpusMultiStreamDecodeObject::GetWorkBufferSize(
total_stream_count, stereo_stream_count));
auto& decoder_object = OpusMultiStreamDecodeObject::Initialize(buffer, buffer);
shared_memory->dsp_return_data[0] = decoder_object.InitializeDecoder(
sample_rate, total_stream_count, channel_count, stereo_stream_count, mappings);
Send(Direction::Host, Message::InitializeMultiStreamDecodeObjectOK);
} break;
case ShutdownMultiStreamDecodeObject: {
auto buffer = shared_memory->host_send_data[0];
[[maybe_unused]] auto buffer_size = shared_memory->host_send_data[1];
auto& decoder_object = OpusMultiStreamDecodeObject::Initialize(buffer, buffer);
shared_memory->dsp_return_data[0] = decoder_object.Shutdown();
Send(Direction::Host, Message::ShutdownMultiStreamDecodeObjectOK);
} break;
case DecodeInterleavedForMultiStream: {
auto start_time = system.CoreTiming().GetGlobalTimeUs();
auto buffer = shared_memory->host_send_data[0];
auto input_data = shared_memory->host_send_data[1];
auto input_data_size = shared_memory->host_send_data[2];
auto output_data = shared_memory->host_send_data[3];
auto output_data_size = shared_memory->host_send_data[4];
auto final_range = static_cast<u32>(shared_memory->host_send_data[5]);
auto reset_requested = shared_memory->host_send_data[6];
u32 decoded_samples{0};
auto& decoder_object = OpusMultiStreamDecodeObject::Initialize(buffer, buffer);
s32 error_code{OPUS_OK};
if (reset_requested) {
error_code = decoder_object.ResetDecoder();
}
if (error_code == OPUS_OK) {
error_code = decoder_object.Decode(decoded_samples, output_data, output_data_size,
input_data, input_data_size);
}
if (error_code == OPUS_OK) {
if (final_range && decoder_object.GetFinalRange() != final_range) {
error_code = OPUS_INVALID_PACKET;
}
}
auto end_time = system.CoreTiming().GetGlobalTimeUs();
shared_memory->dsp_return_data[0] = error_code;
shared_memory->dsp_return_data[1] = decoded_samples;
shared_memory->dsp_return_data[2] = (end_time - start_time).count();
Send(Direction::Host, Message::DecodeInterleavedForMultiStreamOK);
} break;
default:
LOG_ERROR(Service_Audio, "Invalid OpusDecoder command {}", msg);
continue;
}
}
}
} // namespace AudioCore::ADSP::OpusDecoder
+13 -9
View File
@@ -6,9 +6,11 @@
#pragma once
#include <memory>
#include <thread>
#include "common/container/unordered_map.h"
#include "audio_core/adsp/apps/opus/opus_decode_object.h"
#include "audio_core/adsp/apps/opus/opus_multistream_decode_object.h"
#include "audio_core/adsp/apps/opus/shared_memory.h"
#include "audio_core/adsp/mailbox.h"
#include "common/common_types.h"
@@ -68,28 +70,30 @@ public:
}
private:
/**
* Initializing thread, launched at audio_core boot to avoid blocking the main emu boot thread.
*/
/// @brief Initializing thread, launched at audio_core boot to avoid blocking the main emu boot thread.
void Init(std::stop_token stop_token);
/**
* Main OpusDecoder thread, responsible for processing the incoming Opus packets.
*/
void Main(std::stop_token stop_token);
/// Core system
Core::System& system;
/// Mailbox to communicate messages with the host, drives the main thread
Mailbox mailbox;
/// Core system
Core::System& system;
/// Structure shared with the host, input data set by the host before sending a mailbox message,
/// and the responses are written back by the OpusDecoder.
SharedMemory* shared_memory{};
/// Init thread
std::jthread init_thread{};
/// Main thread
std::jthread main_thread{};
::Common::unordered_map<u64, OpusDecodeObject> decode_objects;
::Common::unordered_map<u64, OpusMultiStreamDecodeObject> ms_decode_objects;
/// The current state
bool running{};
/// Structure shared with the host, input data set by the host before sending a mailbox message,
/// and the responses are written back by the OpusDecoder.
SharedMemory* shared_memory{};
};
} // namespace AudioCore::ADSP::OpusDecoder
@@ -4,8 +4,17 @@
// SPDX-FileCopyrightText: Copyright 2023 yuzu Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
extern "C" {
#include <libavcodec/avcodec.h>
#include <libavcodec/codec.h>
#include <libavcodec/packet.h>
#include <libavutil/channel_layout.h>
#include <libavutil/frame.h>
}
#include "audio_core/adsp/apps/opus/opus_multistream_decode_object.h"
#include "common/assert.h"
#include "core/hle/result.h"
#include "core/hle/service/audio/errors.h"
namespace AudioCore::ADSP::OpusDecoder {
@@ -13,101 +22,61 @@ namespace {
constexpr u32 OpusStreamCountMax = 255;
bool IsValidStreamCounts(u32 total_stream_count, u32 stereo_stream_count) {
return total_stream_count > 0 && total_stream_count <= OpusStreamCountMax &&
static_cast<s32>(stereo_stream_count) >= 0 &&
stereo_stream_count <= total_stream_count;
return total_stream_count > 0 && total_stream_count <= OpusStreamCountMax
&& s32(stereo_stream_count) >= 0 && stereo_stream_count <= total_stream_count;
}
} // namespace
u32 OpusMultiStreamDecodeObject::GetWorkBufferSize(u32 total_stream_count,
u32 stereo_stream_count) {
if (IsValidStreamCounts(total_stream_count, stereo_stream_count)) {
return static_cast<u32>(sizeof(OpusMultiStreamDecodeObject)) +
opus_multistream_decoder_get_size(total_stream_count, stereo_stream_count);
}
u32 OpusMultiStreamDecodeObject::GetWorkBufferSize(u32 total_stream_count, u32 stereo_stream_count) {
if (IsValidStreamCounts(total_stream_count, stereo_stream_count))
return u32(sizeof(OpusMultiStreamDecodeObject)) + 2556 * (total_stream_count * stereo_stream_count);
return 0;
}
OpusMultiStreamDecodeObject& OpusMultiStreamDecodeObject::Initialize(u64 buffer, u64 buffer2) {
auto* new_decoder = reinterpret_cast<OpusMultiStreamDecodeObject*>(buffer);
auto* comparison = reinterpret_cast<OpusMultiStreamDecodeObject*>(buffer2);
if (new_decoder->magic == DecodeMultiStreamObjectMagic) {
if (!new_decoder->initialized ||
(new_decoder->initialized && new_decoder->self == comparison)) {
new_decoder->state_valid = true;
}
} else {
new_decoder->initialized = false;
new_decoder->state_valid = true;
}
return *new_decoder;
Result OpusMultiStreamDecodeObject::InitializeDecoder(u32 sample_rate, u32 total_stream_count, u32 channel_count, u32 stereo_stream_count, u8* mappings) {
if ((codec = avcodec_find_decoder(AV_CODEC_ID_OPUS)) == nullptr)
return Service::Audio::ResultLibOpusInvalidState;
if ((avctx = avcodec_alloc_context3(codec)) == nullptr)
return Service::Audio::ResultLibOpusInvalidState;
avctx->sample_rate = sample_rate;
av_channel_layout_default(&avctx->ch_layout, channel_count);
return ResultSuccess;
}
s32 OpusMultiStreamDecodeObject::InitializeDecoder(u32 sample_rate, u32 total_stream_count,
u32 channel_count, u32 stereo_stream_count,
u8* mappings) {
if (!state_valid) {
return OPUS_INVALID_STATE;
Result OpusMultiStreamDecodeObject::Shutdown() {
if (avctx) {
avcodec_free_context(&avctx);
avctx = nullptr;
}
if (initialized) {
return OPUS_OK;
}
// See OpusDecodeObject::InitializeDecoder for an explanation of this
decoder = (LibOpusMSDecoder*)(this + 1);
s32 ret = opus_multistream_decoder_init(decoder, sample_rate, channel_count, total_stream_count,
stereo_stream_count, mappings);
if (ret == OPUS_OK) {
magic = DecodeMultiStreamObjectMagic;
initialized = true;
state_valid = true;
self = this;
final_range = 0;
}
return ret;
return ResultSuccess;
}
s32 OpusMultiStreamDecodeObject::Shutdown() {
if (!state_valid) {
return OPUS_INVALID_STATE;
Result OpusMultiStreamDecodeObject::ResetDecoder() {
if (avctx) {
avcodec_flush_buffers(avctx);
return ResultSuccess;
}
if (initialized) {
magic = 0x0;
initialized = false;
state_valid = false;
self = nullptr;
final_range = 0;
decoder = nullptr;
}
return OPUS_OK;
return Service::Audio::ResultLibOpusInvalidState;
}
s32 OpusMultiStreamDecodeObject::ResetDecoder() {
return opus_multistream_decoder_ctl(decoder, OPUS_RESET_STATE);
}
s32 OpusMultiStreamDecodeObject::Decode(u32& out_sample_count, u64 output_data,
u64 output_data_size, u64 input_data, u64 input_data_size) {
ASSERT(initialized);
Result OpusMultiStreamDecodeObject::Decode(u32& out_sample_count, u64 output_data, u64 output_data_size, u64 input_data, u64 input_data_size) {
out_sample_count = 0;
if (avctx) {
AVPacket* avpkt = av_packet_alloc();
av_new_packet(avpkt, int(input_data_size));
std::memcpy(avpkt->data, reinterpret_cast<const u8*>(input_data), input_data_size);
avcodec_send_packet(avctx, avpkt);
if (!state_valid) {
return OPUS_INVALID_STATE;
AVFrame* frame = av_frame_alloc();
avcodec_receive_frame(avctx, frame);
std::memcpy(reinterpret_cast<u16*>(output_data), frame->data, output_data_size);
av_frame_free(&frame);
av_packet_free(&avpkt);
out_sample_count = frame->nb_samples;
return ResultSuccess;
}
auto ret_code_or_samples = opus_multistream_decode(
decoder, reinterpret_cast<const u8*>(input_data), static_cast<opus_int32>(input_data_size),
reinterpret_cast<opus_int16*>(output_data), static_cast<opus_int32>(output_data_size), 0);
if (ret_code_or_samples < OPUS_OK) {
return ret_code_or_samples;
}
out_sample_count = ret_code_or_samples;
return opus_multistream_decoder_ctl(decoder, OPUS_GET_FINAL_RANGE_REQUEST, &final_range);
return Service::Audio::ResultLibOpusInvalidState;
}
} // namespace AudioCore::ADSP::OpusDecoder
@@ -1,39 +1,30 @@
// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
// SPDX-FileCopyrightText: Copyright 2023 yuzu Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
#pragma once
#include <opus_multistream.h>
#include "common/common_types.h"
#include "core/hle/result.h"
struct AVCodec;
struct AVCodecContext;
namespace AudioCore::ADSP::OpusDecoder {
using LibOpusMSDecoder = ::OpusMSDecoder;
static constexpr u32 DecodeMultiStreamObjectMagic = 0xDEADBEEF;
class OpusMultiStreamDecodeObject {
public:
static u32 GetWorkBufferSize(u32 total_stream_count, u32 stereo_stream_count);
static OpusMultiStreamDecodeObject& Initialize(u64 buffer, u64 buffer2);
s32 InitializeDecoder(u32 sample_rate, u32 total_stream_count, u32 channel_count,
u32 stereo_stream_count, u8* mappings);
s32 Shutdown();
s32 ResetDecoder();
s32 Decode(u32& out_sample_count, u64 output_data, u64 output_data_size, u64 input_data,
u64 input_data_size);
u32 GetFinalRange() const noexcept {
return final_range;
}
Result InitializeDecoder(u32 sample_rate, u32 total_stream_count, u32 channel_count, u32 stereo_stream_count, u8* mappings);
Result Shutdown();
Result ResetDecoder();
Result Decode(u32& out_sample_count, u64 output_data, u64 output_data_size, u64 input_data, u64 input_data_size);
private:
u32 magic;
bool initialized;
bool state_valid;
OpusMultiStreamDecodeObject* self;
u32 final_range;
LibOpusMSDecoder* decoder;
AVCodec const* codec = nullptr;
AVCodecContext* avctx = nullptr;
};
static_assert(std::is_trivially_constructible_v<OpusMultiStreamDecodeObject>);
} // namespace AudioCore::ADSP::OpusDecoder
@@ -0,0 +1,32 @@
// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
#pragma once
#include "common/common_types.h"
namespace AudioCore::ADSP {
static constexpr u32 DECODE_OBJECT_MAGIC = 0xDEADBEEF;
struct LibOpusDecoder {
u32 magic;
bool initialized;
bool state_valid;
LibOpusDecoder* self;
u32 final_range;
void* decoder;
};
static_assert(sizeof(LibOpusDecoder) == 32);
static constexpr u32 DECODE_MULTISTREAM_OBJECT_MAGIC = 0xDEADBEEF;
struct LibOpusMultistreamDecoder {
u32 magic;
bool initialized;
bool state_valid;
LibOpusMultistreamDecoder* self;
u32 final_range;
void* decoder;
};
static_assert(sizeof(LibOpusMultistreamDecoder) == 32);
}
@@ -1,9 +1,11 @@
// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
// SPDX-FileCopyrightText: Copyright 2023 yuzu Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
#pragma once
#include "common/common_funcs.h"
#include "common/common_types.h"
namespace AudioCore::ADSP::OpusDecoder {
+23 -48
View File
@@ -9,39 +9,18 @@
#include "audio_core/audio_core.h"
#include "audio_core/opus/hardware_opus.h"
#include "core/core.h"
#include "core/hle/result.h"
namespace AudioCore::OpusDecoder {
namespace {
using namespace Service::Audio;
static constexpr Result ResultCodeFromLibOpusErrorCode(u64 error_code) {
s32 error{static_cast<s32>(error_code)};
ASSERT(error <= OPUS_OK);
switch (error) {
case OPUS_ALLOC_FAIL:
R_THROW(ResultLibOpusAllocFail);
case OPUS_INVALID_STATE:
R_THROW(ResultLibOpusInvalidState);
case OPUS_UNIMPLEMENTED:
R_THROW(ResultLibOpusUnimplemented);
case OPUS_INVALID_PACKET:
R_THROW(ResultLibOpusInvalidPacket);
case OPUS_INTERNAL_ERROR:
R_THROW(ResultLibOpusInternalError);
case OPUS_BUFFER_TOO_SMALL:
R_THROW(ResultBufferTooSmall);
case OPUS_BAD_ARG:
R_THROW(ResultLibOpusBadArg);
case OPUS_OK:
R_RETURN(ResultSuccess);
}
UNREACHABLE();
}
} // namespace
HardwareOpus::HardwareOpus(Core::System& system_)
: system{system_}, opus_decoder{system.AudioCore().ADSP().OpusDecoder()} {
: system{system_}
, opus_decoder{system.AudioCore().ADSP().OpusDecoder()}
{
opus_decoder.SetSharedMemory(shared_memory);
}
@@ -92,7 +71,7 @@ Result HardwareOpus::InitializeDecodeObject(u32 sample_rate, u32 channel_count,
R_THROW(ResultInvalidOpusDSPReturnCode);
}
R_RETURN(ResultCodeFromLibOpusErrorCode(shared_memory.dsp_return_data[0]));
R_RETURN(Result(u32(shared_memory.dsp_return_data[0])));
}
Result HardwareOpus::InitializeMultiStreamDecodeObject(u32 sample_rate, u32 channel_count,
@@ -120,7 +99,7 @@ Result HardwareOpus::InitializeMultiStreamDecodeObject(u32 sample_rate, u32 chan
R_THROW(ResultInvalidOpusDSPReturnCode);
}
R_RETURN(ResultCodeFromLibOpusErrorCode(shared_memory.dsp_return_data[0]));
R_RETURN(Result(u32(shared_memory.dsp_return_data[0])));
}
Result HardwareOpus::ShutdownDecodeObject(void* buffer, u64 buffer_size) {
@@ -134,7 +113,7 @@ Result HardwareOpus::ShutdownDecodeObject(void* buffer, u64 buffer_size) {
"Expected Opus shutdown code {}, got {}",
ADSP::OpusDecoder::Message::ShutdownDecodeObjectOK, msg);
R_RETURN(ResultCodeFromLibOpusErrorCode(shared_memory.dsp_return_data[0]));
R_RETURN(Result(u32(shared_memory.dsp_return_data[0])));
}
Result HardwareOpus::ShutdownMultiStreamDecodeObject(void* buffer, u64 buffer_size) {
@@ -149,7 +128,7 @@ Result HardwareOpus::ShutdownMultiStreamDecodeObject(void* buffer, u64 buffer_si
"Expected Opus shutdown code {}, got {}",
ADSP::OpusDecoder::Message::ShutdownMultiStreamDecodeObjectOK, msg);
R_RETURN(ResultCodeFromLibOpusErrorCode(shared_memory.dsp_return_data[0]));
R_RETURN(Result(u32(shared_memory.dsp_return_data[0])));
}
Result HardwareOpus::DecodeInterleaved(u32& out_sample_count, void* output_data,
@@ -173,12 +152,12 @@ Result HardwareOpus::DecodeInterleaved(u32& out_sample_count, void* output_data,
R_THROW(ResultInvalidOpusDSPReturnCode);
}
auto error_code{static_cast<s32>(shared_memory.dsp_return_data[0])};
if (error_code == OPUS_OK) {
out_sample_count = static_cast<u32>(shared_memory.dsp_return_data[1]);
auto error_code = s32(shared_memory.dsp_return_data[0]);
if (error_code == ResultSuccess.raw) {
out_sample_count = u32(shared_memory.dsp_return_data[1]);
out_time_taken = 1000 * shared_memory.dsp_return_data[2];
}
R_RETURN(ResultCodeFromLibOpusErrorCode(error_code));
R_RETURN(Result(u32(error_code)));
}
Result HardwareOpus::DecodeInterleavedForMultiStream(u32& out_sample_count, void* output_data,
@@ -187,29 +166,27 @@ Result HardwareOpus::DecodeInterleavedForMultiStream(u32& out_sample_count, void
void* buffer, u64& out_time_taken,
bool reset) {
std::scoped_lock l{mutex};
shared_memory.host_send_data[0] = (u64)buffer;
shared_memory.host_send_data[1] = (u64)input_data;
shared_memory.host_send_data[0] = u64(buffer);
shared_memory.host_send_data[1] = u64(input_data);
shared_memory.host_send_data[2] = input_data_size;
shared_memory.host_send_data[3] = (u64)output_data;
shared_memory.host_send_data[3] = u64(output_data);
shared_memory.host_send_data[4] = output_data_size;
shared_memory.host_send_data[5] = 0;
shared_memory.host_send_data[6] = reset;
opus_decoder.Send(ADSP::Direction::DSP,
ADSP::OpusDecoder::Message::DecodeInterleavedForMultiStream);
opus_decoder.Send(ADSP::Direction::DSP, ADSP::OpusDecoder::Message::DecodeInterleavedForMultiStream);
auto msg = opus_decoder.Receive(ADSP::Direction::Host);
if (msg != ADSP::OpusDecoder::Message::DecodeInterleavedForMultiStreamOK) {
LOG_ERROR(Service_Audio, "OpusDecoder returned invalid message. Expected {} got {}",
ADSP::OpusDecoder::Message::DecodeInterleavedForMultiStreamOK, msg);
LOG_ERROR(Service_Audio, "OpusDecoder returned invalid message. Expected {} got {}", ADSP::OpusDecoder::Message::DecodeInterleavedForMultiStreamOK, msg);
R_THROW(ResultInvalidOpusDSPReturnCode);
}
auto error_code{static_cast<s32>(shared_memory.dsp_return_data[0])};
if (error_code == OPUS_OK) {
auto const error_code = shared_memory.dsp_return_data[0];
if (error_code == ResultSuccess.raw) {
out_sample_count = static_cast<u32>(shared_memory.dsp_return_data[1]);
out_time_taken = 1000 * shared_memory.dsp_return_data[2];
}
R_RETURN(ResultCodeFromLibOpusErrorCode(error_code));
R_RETURN(Result(u32(shared_memory.dsp_return_data[0])));
}
Result HardwareOpus::MapMemory(void* buffer, u64 buffer_size) {
@@ -220,8 +197,7 @@ Result HardwareOpus::MapMemory(void* buffer, u64 buffer_size) {
opus_decoder.Send(ADSP::Direction::DSP, ADSP::OpusDecoder::Message::MapMemory);
auto msg = opus_decoder.Receive(ADSP::Direction::Host);
if (msg != ADSP::OpusDecoder::Message::MapMemoryOK) {
LOG_ERROR(Service_Audio, "OpusDecoder returned invalid message. Expected {} got {}",
ADSP::OpusDecoder::Message::MapMemoryOK, msg);
LOG_ERROR(Service_Audio, "OpusDecoder returned invalid message. Expected {} got {}", ADSP::OpusDecoder::Message::MapMemoryOK, msg);
R_THROW(ResultInvalidOpusDSPReturnCode);
}
R_SUCCEED();
@@ -235,8 +211,7 @@ Result HardwareOpus::UnmapMemory(void* buffer, u64 buffer_size) {
opus_decoder.Send(ADSP::Direction::DSP, ADSP::OpusDecoder::Message::UnmapMemory);
auto msg = opus_decoder.Receive(ADSP::Direction::Host);
if (msg != ADSP::OpusDecoder::Message::UnmapMemoryOK) {
LOG_ERROR(Service_Audio, "OpusDecoder returned invalid message. Expected {} got {}",
ADSP::OpusDecoder::Message::UnmapMemoryOK, msg);
LOG_ERROR(Service_Audio, "OpusDecoder returned invalid message. Expected {} got {}", ADSP::OpusDecoder::Message::UnmapMemoryOK, msg);
R_THROW(ResultInvalidOpusDSPReturnCode);
}
R_SUCCEED();
+3 -2
View File
@@ -1,11 +1,12 @@
// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
// SPDX-FileCopyrightText: Copyright 2023 yuzu Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
#pragma once
#include <mutex>
#include <opus.h>
#include "audio_core/adsp/apps/opus/opus_decoder.h"
#include "audio_core/adsp/apps/opus/shared_memory.h"
#include "audio_core/adsp/mailbox.h"
-1
View File
@@ -5,7 +5,6 @@
// SPDX-License-Identifier: GPL-2.0-or-later
#include <string>
#include <cstdlib>
#include <string_view>
#ifdef _WIN32
#include <llvm/Demangle/Demangle.h>
+1 -5
View File
@@ -1,13 +1,9 @@
// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
#pragma once
#include <iterator>
#include <cstring>
#include "common/make_unique_for_overwrite.h"
@@ -65,7 +61,7 @@ public:
void resize(size_type size) {
if (size > buffer_capacity) {
auto new_buffer = Common::make_unique_for_overwrite<T[]>(size);
std::memcpy(new_buffer.get(), buffer.get(), buffer_capacity * sizeof(T));
std::move(buffer.get(), buffer.get() + buffer_capacity, new_buffer.get());
buffer = std::move(new_buffer);
buffer_capacity = size;
}
-1
View File
@@ -10,7 +10,6 @@
#include <functional>
#include <span>
#include <string>
#include <type_traits>
#include "common/common_types.h"
-1
View File
@@ -1205,7 +1205,6 @@ else()
endif()
endif()
target_include_directories(core PRIVATE ${OPUS_INCLUDE_DIRS})
target_link_libraries(core PUBLIC common PRIVATE audio_core hid_core network video_core nx_tzdb tz)
if (BOOST_NO_HEADERS)
+2 -2
View File
@@ -6,13 +6,13 @@
#include <mutex>
#include <utility>
#include <type_traits>
#include <boost/asio.hpp>
#include <boost/version.hpp>
#if BOOST_VERSION > 108400 && (!defined(_WINDOWS) && !defined(__ANDROID__)) || defined(YUZU_BOOST_v1)
#define USE_BOOST_v1
#endif
#ifdef USE_BOOST_v1
#include <boost/process/v1/async_pipe.hpp>
#else
+1 -2
View File
@@ -1,4 +1,4 @@
// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
// SPDX-FileCopyrightText: Copyright 2025 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
// SPDX-FileCopyrightText: Copyright 2023 yuzu Emulator Project
@@ -6,7 +6,6 @@
#pragma once
#include <type_traits>
#include "common/common_funcs.h"
namespace FileSys {
@@ -7,7 +7,6 @@
#pragma once
#include <optional>
#include <type_traits>
#include "common/literals.h"
#include "core/file_sys/fssystem/fs_i_storage.h"
@@ -4,7 +4,6 @@
// SPDX-FileCopyrightText: Copyright 2023 yuzu Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
#include <type_traits>
#include "core/file_sys/errors.h"
#include "core/file_sys/fssystem/fssystem_bucket_tree.h"
#include "core/file_sys/fssystem/fssystem_bucket_tree_utils.h"
@@ -1,13 +1,9 @@
// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
// SPDX-FileCopyrightText: Copyright 2023 yuzu Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
#pragma once
#include <mutex>
#include <type_traits>
#include "common/alignment.h"
#include "common/common_funcs.h"
@@ -1,4 +1,4 @@
// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
// SPDX-FileCopyrightText: Copyright 2025 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
// SPDX-FileCopyrightText: Copyright 2023 yuzu Emulator Project
@@ -6,7 +6,6 @@
#pragma once
#include <type_traits>
#include "core/file_sys/errors.h"
#include "core/file_sys/fssystem/fssystem_bucket_tree.h"
#include "core/file_sys/fssystem/fssystem_bucket_tree_utils.h"
@@ -6,8 +6,6 @@
#pragma once
#include <type_traits>
#include <cstddef>
#include "common/literals.h"
#include "core/file_sys/errors.h"
@@ -1,12 +1,8 @@
// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
// SPDX-FileCopyrightText: Copyright 2023 yuzu Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
#pragma once
#include <type_traits>
#include "common/alignment.h"
#include "core/file_sys/fssystem/fs_i_storage.h"
#include "core/file_sys/fssystem/fs_types.h"
@@ -6,9 +6,6 @@
#pragma once
#include <type_traits>
#include <array>
#include <cstddef>
#include "core/file_sys/errors.h"
#include "core/file_sys/fssystem/fs_i_storage.h"
#include "core/file_sys/fssystem/fssystem_bucket_tree.h"
@@ -1,15 +1,9 @@
// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
// SPDX-FileCopyrightText: Copyright 2023 yuzu Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
#pragma once
#include <optional>
#include <array>
#include <cstddef>
#include <type_traits>
#include "core/file_sys/fssystem/fs_i_storage.h"
#include "core/file_sys/fssystem/fs_types.h"
@@ -1,4 +1,4 @@
// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
// SPDX-FileCopyrightText: Copyright 2025 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
// SPDX-FileCopyrightText: Copyright 2023 yuzu Emulator Project
@@ -6,8 +6,6 @@
#pragma once
#include <type_traits>
#include <cstddef>
#include "core/file_sys/fssystem/fssystem_compression_common.h"
#include "core/file_sys/fssystem/fssystem_nca_header.h"
#include "core/file_sys/vfs/vfs.h"
@@ -1,14 +1,8 @@
// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
// SPDX-FileCopyrightText: Copyright 2023 yuzu Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
#pragma once
#include <type_traits>
#include <array>
#include <cstddef>
#include "common/common_funcs.h"
#include "common/common_types.h"
#include "common/literals.h"
-1
View File
@@ -7,7 +7,6 @@
#pragma once
#include <memory>
#include <type_traits>
#include "common/common_funcs.h"
#include "common/page_table.h"
-1
View File
@@ -6,7 +6,6 @@
#pragma once
#include <type_traits>
#include "common/assert.h"
#include "common/bit_field.h"
#include "common/common_funcs.h"
-4
View File
@@ -1,13 +1,9 @@
// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
// SPDX-FileCopyrightText: Copyright 2024 yuzu Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
#pragma once
#include <array>
#include <type_traits>
#include <functional>
#include "common/common_funcs.h"
+26 -8
View File
@@ -175,10 +175,19 @@ Result AlbumManager::LoadAlbumScreenShotImage(LoadAlbumScreenShotImageOutput& ou
return ResultIsNotMounted;
}
out_image_output = {};
out_image_output.width = 1280;
out_image_output.height = 720;
out_image_output.attribute.orientation = AlbumImageOrientation::None;
out_image_output = {
.width = 1280,
.height = 720,
.attribute =
{
.unknown_0{},
.orientation = AlbumImageOrientation::None,
.unknown_1{},
.unknown_2{},
.pad163{},
},
.pad179{},
};
std::filesystem::path path;
const auto result = GetFile(path, file_id);
@@ -202,10 +211,19 @@ Result AlbumManager::LoadAlbumScreenShotThumbnail(
return ResultIsNotMounted;
}
out_image_output = {};
out_image_output.width = 320;
out_image_output.height = 180;
out_image_output.attribute.orientation = AlbumImageOrientation::None;
out_image_output = {
.width = 320,
.height = 180,
.attribute =
{
.unknown_0{},
.orientation = AlbumImageOrientation::None,
.unknown_1{},
.unknown_2{},
.pad163{},
},
.pad179{},
};
std::filesystem::path path;
const auto result = GetFile(path, file_id);
+7 -2
View File
@@ -73,8 +73,13 @@ void IScreenShotApplicationService::CaptureAndSaveScreenshot(AlbumReportOption r
Layout::FramebufferLayout layout =
Layout::DefaultFrameLayout(screenshot_width, screenshot_height);
Capture::ScreenShotAttribute attribute{};
attribute.orientation = Capture::AlbumImageOrientation::None;
const Capture::ScreenShotAttribute attribute{
.unknown_0{},
.orientation = Capture::AlbumImageOrientation::None,
.unknown_1{},
.unknown_2{},
.pad163{},
};
renderer.RequestScreenshot(
image_data.data(),
-4
View File
@@ -1,12 +1,8 @@
// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
// SPDX-FileCopyrightText: Copyright 2018 yuzu Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
#pragma once
#include <type_traits>
#include "common/common_funcs.h"
#include "common/common_types.h"
+1 -2
View File
@@ -1,4 +1,4 @@
// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
// SPDX-FileCopyrightText: Copyright 2025 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
@@ -6,7 +6,6 @@
#pragma once
#include <type_traits>
#include <fmt/ranges.h>
#include "common/common_funcs.h"
-1
View File
@@ -8,7 +8,6 @@
#include <array>
#include <chrono>
#include <type_traits>
#include <fmt/ranges.h>
#include "common/common_types.h"
@@ -1,13 +1,10 @@
// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
// SPDX-FileCopyrightText: Copyright 2023 yuzu Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
#pragma once
#include <array>
#include <type_traits>
#include "common/common_types.h"
#include "core/hle/service/psc/time/common.h"
@@ -7,7 +7,6 @@
#pragma once
#include <array>
#include <type_traits>
#include "common/bit_field.h"
#include "common/common_funcs.h"
-7
View File
@@ -1,15 +1,8 @@
// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
// SPDX-FileCopyrightText: Copyright 2021 yuzu Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
#pragma once
#include <cstddef>
#include <array>
#include <type_traits>
#include "common/bit_field.h"
#include "common/common_funcs.h"
#include "common/common_types.h"
+51 -53
View File
@@ -6,6 +6,55 @@
add_library(shader_recompiler STATIC
backend/bindings.h
backend/glasm/emit_glasm.cpp
backend/glasm/emit_glasm.h
backend/glasm/emit_glasm_barriers.cpp
backend/glasm/emit_glasm_bitwise_conversion.cpp
backend/glasm/emit_glasm_composite.cpp
backend/glasm/emit_glasm_context_get_set.cpp
backend/glasm/emit_glasm_control_flow.cpp
backend/glasm/emit_glasm_convert.cpp
backend/glasm/emit_glasm_floating_point.cpp
backend/glasm/emit_glasm_image.cpp
backend/glasm/emit_glasm_instructions.h
backend/glasm/emit_glasm_integer.cpp
backend/glasm/emit_glasm_logical.cpp
backend/glasm/emit_glasm_memory.cpp
backend/glasm/emit_glasm_not_implemented.cpp
backend/glasm/emit_glasm_select.cpp
backend/glasm/emit_glasm_shared_memory.cpp
backend/glasm/emit_glasm_special.cpp
backend/glasm/emit_glasm_undefined.cpp
backend/glasm/emit_glasm_warp.cpp
backend/glasm/glasm_emit_context.cpp
backend/glasm/glasm_emit_context.h
backend/glasm/reg_alloc.cpp
backend/glasm/reg_alloc.h
backend/glsl/emit_glsl.cpp
backend/glsl/emit_glsl.h
backend/glsl/emit_glsl_atomic.cpp
backend/glsl/emit_glsl_barriers.cpp
backend/glsl/emit_glsl_bitwise_conversion.cpp
backend/glsl/emit_glsl_composite.cpp
backend/glsl/emit_glsl_context_get_set.cpp
backend/glsl/emit_glsl_control_flow.cpp
backend/glsl/emit_glsl_convert.cpp
backend/glsl/emit_glsl_floating_point.cpp
backend/glsl/emit_glsl_image.cpp
backend/glsl/emit_glsl_instructions.h
backend/glsl/emit_glsl_integer.cpp
backend/glsl/emit_glsl_logical.cpp
backend/glsl/emit_glsl_memory.cpp
backend/glsl/emit_glsl_not_implemented.cpp
backend/glsl/emit_glsl_select.cpp
backend/glsl/emit_glsl_shared_memory.cpp
backend/glsl/emit_glsl_special.cpp
backend/glsl/emit_glsl_undefined.cpp
backend/glsl/emit_glsl_warp.cpp
backend/glsl/glsl_emit_context.cpp
backend/glsl/glsl_emit_context.h
backend/glsl/var_alloc.cpp
backend/glsl/var_alloc.h
backend/spirv/emit_spirv.cpp
backend/spirv/emit_spirv.h
backend/spirv/emit_spirv_atomic.cpp
@@ -190,60 +239,9 @@ add_library(shader_recompiler STATIC
program_header.h
runtime_info.h
shader_info.h
varying_state.h)
varying_state.h
if (ENABLE_OPENGL)
target_sources(shader_recompiler PRIVATE
backend/glasm/emit_glasm.cpp
backend/glasm/emit_glasm.h
backend/glasm/emit_glasm_barriers.cpp
backend/glasm/emit_glasm_bitwise_conversion.cpp
backend/glasm/emit_glasm_composite.cpp
backend/glasm/emit_glasm_context_get_set.cpp
backend/glasm/emit_glasm_control_flow.cpp
backend/glasm/emit_glasm_convert.cpp
backend/glasm/emit_glasm_floating_point.cpp
backend/glasm/emit_glasm_image.cpp
backend/glasm/emit_glasm_instructions.h
backend/glasm/emit_glasm_integer.cpp
backend/glasm/emit_glasm_logical.cpp
backend/glasm/emit_glasm_memory.cpp
backend/glasm/emit_glasm_not_implemented.cpp
backend/glasm/emit_glasm_select.cpp
backend/glasm/emit_glasm_shared_memory.cpp
backend/glasm/emit_glasm_special.cpp
backend/glasm/emit_glasm_undefined.cpp
backend/glasm/emit_glasm_warp.cpp
backend/glasm/glasm_emit_context.cpp
backend/glasm/glasm_emit_context.h
backend/glasm/reg_alloc.cpp
backend/glasm/reg_alloc.h
backend/glsl/emit_glsl.cpp
backend/glsl/emit_glsl.h
backend/glsl/emit_glsl_atomic.cpp
backend/glsl/emit_glsl_barriers.cpp
backend/glsl/emit_glsl_bitwise_conversion.cpp
backend/glsl/emit_glsl_composite.cpp
backend/glsl/emit_glsl_context_get_set.cpp
backend/glsl/emit_glsl_control_flow.cpp
backend/glsl/emit_glsl_convert.cpp
backend/glsl/emit_glsl_floating_point.cpp
backend/glsl/emit_glsl_image.cpp
backend/glsl/emit_glsl_instructions.h
backend/glsl/emit_glsl_integer.cpp
backend/glsl/emit_glsl_logical.cpp
backend/glsl/emit_glsl_memory.cpp
backend/glsl/emit_glsl_not_implemented.cpp
backend/glsl/emit_glsl_select.cpp
backend/glsl/emit_glsl_shared_memory.cpp
backend/glsl/emit_glsl_special.cpp
backend/glsl/emit_glsl_undefined.cpp
backend/glsl/emit_glsl_warp.cpp
backend/glsl/glsl_emit_context.cpp
backend/glsl/glsl_emit_context.h
backend/glsl/var_alloc.cpp
backend/glsl/var_alloc.h)
endif()
)
target_link_libraries(shader_recompiler PUBLIC common fmt::fmt sirit::sirit)
@@ -553,6 +553,17 @@ void GlobalMemoryToStorageBufferPass(IR::Program& program, const HostTranslateIn
}
}
template <typename Descriptors, typename Descriptor, typename Func>
static u32 Add(Descriptors& descriptors, const Descriptor& desc, Func&& pred) {
// TODO: Handle arrays
const auto it{std::ranges::find_if(descriptors, pred)};
if (it != descriptors.end()) {
return static_cast<u32>(std::distance(descriptors.begin(), it));
}
descriptors.push_back(desc);
return static_cast<u32>(descriptors.size()) - 1;
}
void JoinStorageInfo(Info& base, Info& source) {
auto& descriptors = base.storage_buffers_descriptors;
for (auto& desc : source.storage_buffers_descriptors) {
+3 -59
View File
@@ -312,66 +312,10 @@ static inline std::optional<ConstBufferAddr> TrackCached(const IR::Value& v, Env
std::optional<ConstBufferAddr> TryGetConstBuffer(const IR::Inst* inst, Environment& env, const HostTranslateInfo& host_info);
bool IsSameConstBufferAddr(const ConstBufferAddr& lhs, const ConstBufferAddr& rhs) {
return lhs.index == rhs.index && lhs.offset == rhs.offset &&
lhs.shift_left == rhs.shift_left && lhs.secondary_index == rhs.secondary_index &&
lhs.secondary_offset == rhs.secondary_offset &&
lhs.secondary_shift_left == rhs.secondary_shift_left && lhs.count == rhs.count &&
lhs.has_secondary == rhs.has_secondary && lhs.dynamic_offset == rhs.dynamic_offset;
}
std::optional<ConstBufferAddr> TrackPhi(const IR::Inst* phi, Environment& env,
const HostTranslateInfo& host_info, bool& ambiguous) {
std::optional<ConstBufferAddr> agreed;
const size_t num_args{phi->NumArgs()};
for (size_t index = 0; index < num_args; ++index) {
const IR::Value arg{phi->Arg(index).Resolve()};
if (arg.IsImmediate()) {
ambiguous = true;
return std::nullopt;
}
const IR::Inst* arg_inst{arg.InstRecursive()};
if (arg_inst == phi) {
continue;
}
if (arg_inst->GetOpcode() == IR::Opcode::Phi) {
ambiguous = true;
return std::nullopt;
}
const std::optional<ConstBufferAddr> operand{TrackCached(arg, env, host_info)};
if (!operand) {
ambiguous = true;
return std::nullopt;
}
if (!agreed) {
agreed = operand;
continue;
}
if (!IsSameConstBufferAddr(*agreed, *operand)) {
ambiguous = true;
return std::nullopt;
}
}
if (!agreed) {
ambiguous = true;
}
return agreed;
}
std::optional<ConstBufferAddr> Track(const IR::Value& value, Environment& env, const HostTranslateInfo& host_info) {
bool ambiguous = false;
const std::optional<ConstBufferAddr> result{IR::BreadthFirstSearch(
value, [&env, &host_info, &ambiguous](const IR::Inst* inst)
-> std::optional<ConstBufferAddr> {
if (inst->GetOpcode() == IR::Opcode::Phi) {
return TrackPhi(inst, env, host_info, ambiguous);
}
return TryGetConstBuffer(inst, env, host_info);
})};
if (ambiguous) {
return std::nullopt;
}
return result;
return IR::BreadthFirstSearch(value, [&env, &host_info](const IR::Inst* inst) {
return TryGetConstBuffer(inst, env, host_info);
});
}
std::optional<u32> TryGetConstant(IR::Value& value, Environment& env) {
-5
View File
@@ -20,7 +20,6 @@ add_library(video_core STATIC
buffer_cache/buffer_cache.h
buffer_cache/memory_tracker_base.h
buffer_cache/usage_tracker.h
buffer_cache/virtual_range_cache.h
buffer_cache/word_manager.h
cache_types.h
capture.h
@@ -167,8 +166,6 @@ add_library(video_core STATIC
renderer_vulkan/vk_fence_manager.h
renderer_vulkan/vk_graphics_pipeline.cpp
renderer_vulkan/vk_graphics_pipeline.h
renderer_vulkan/vk_multi_range_buffer.cpp
renderer_vulkan/vk_multi_range_buffer.h
renderer_vulkan/vk_master_semaphore.cpp
renderer_vulkan/vk_master_semaphore.h
renderer_vulkan/vk_pipeline_cache.cpp
@@ -370,9 +367,7 @@ endif()
if (YUZU_USE_EXTERNAL_FFMPEG)
add_dependencies(video_core ffmpeg-build)
endif()
target_include_directories(video_core PUBLIC ${FFmpeg_INCLUDE_DIR})
target_link_libraries(video_core PRIVATE ${FFmpeg_LIBRARIES})
target_link_options(video_core PRIVATE ${FFmpeg_LDFLAGS})
+9 -76
View File
@@ -112,11 +112,6 @@ void BufferCache<P>::TickFrame() {
async_buffers_death_ring.clear();
}
template <class P>
void BufferCache<P>::UnmapGPUMemory(size_t as_id, GPUVAddr gpu_addr, size_t size) {
virtual_ranges.Unmap(as_id, gpu_addr, size);
}
template <class P>
void BufferCache<P>::WriteMemory(DAddr device_addr, u64 size) {
if (memory_tracker.IsRegionGpuModified(device_addr, size)) {
@@ -1003,56 +998,11 @@ void BufferCache<P>::BindHostGraphicsUniformBuffer(size_t stage, u32 index, u32
channel_state->fast_bound_uniform_buffers[stage] &= ~(1u << binding_index);
}
template <class P>
bool BufferCache<P>::BindMultiRangeStorage(const Binding& binding, bool is_written) {
if constexpr (requires { runtime.BindMultiRangeStorageBuffer(u64{}); }) {
if (binding.gpu_addr == 0 || binding.size == 0) {
return false;
}
if (is_written && !runtime.PrefersSparseSources()) {
return false;
}
const VirtualSegments* segments =
virtual_ranges.Query(*gpu_memory, binding.gpu_addr, binding.size);
if (!segments || segments->size() < 2) {
return false;
}
const u64 key = (static_cast<u64>(gpu_memory->GetID()) << 48) ^ binding.gpu_addr;
const bool prefer_sparse = runtime.PrefersSparseSources();
runtime.ResetMultiRange();
for (const VirtualSegment& segment : *segments) {
const BufferId buffer_id =
FindBuffer(segment.device_addr, segment.size, prefer_sparse);
if (!buffer_id) {
return false;
}
Buffer& buffer = slot_buffers[buffer_id];
TouchBuffer(buffer, buffer_id);
if (SynchronizeBuffer(buffer, segment.device_addr, segment.size)) {
runtime.InvalidateMultiRange(key);
}
const u32 offset = buffer.Offset(segment.device_addr);
buffer.MarkUsage(offset, segment.size);
if (is_written) {
MarkWrittenBuffer(buffer_id, segment.device_addr, segment.size);
}
runtime.PushMultiRangeSource(buffer, offset, segment.size);
}
return runtime.BindMultiRangeStorageBuffer(key);
} else {
return false;
}
}
template <class P>
void BufferCache<P>::BindHostGraphicsStorageBuffers(size_t stage) {
u32 binding_index = 0;
ForEachEnabledBit(channel_state->enabled_storage_buffers[stage], [&](u32 index) {
const Binding& binding = channel_state->storage_buffers[stage][index];
const bool is_written = ((channel_state->written_storage_buffers[stage] >> index) & 1) != 0;
if (BindMultiRangeStorage(binding, is_written)) {
return;
}
Buffer& buffer = slot_buffers[binding.buffer_id];
TouchBuffer(buffer, binding.buffer_id);
const u32 size = binding.size;
@@ -1060,6 +1010,7 @@ void BufferCache<P>::BindHostGraphicsStorageBuffers(size_t stage) {
const u32 offset = buffer.Offset(binding.device_addr);
buffer.MarkUsage(offset, size);
const bool is_written = ((channel_state->written_storage_buffers[stage] >> index) & 1) != 0;
if (is_written) {
MarkWrittenBuffer(binding.buffer_id, binding.device_addr, size);
@@ -1188,11 +1139,6 @@ void BufferCache<P>::BindHostComputeStorageBuffers() {
u32 binding_index = 0;
ForEachEnabledBit(channel_state->enabled_compute_storage_buffers, [&](u32 index) {
const Binding& binding = channel_state->compute_storage_buffers[index];
const bool is_written =
((channel_state->written_compute_storage_buffers >> index) & 1) != 0;
if (BindMultiRangeStorage(binding, is_written)) {
return;
}
Buffer& buffer = slot_buffers[binding.buffer_id];
TouchBuffer(buffer, binding.buffer_id);
const u32 size = binding.size;
@@ -1200,6 +1146,8 @@ void BufferCache<P>::BindHostComputeStorageBuffers() {
const u32 offset = buffer.Offset(binding.device_addr);
buffer.MarkUsage(offset, size);
const bool is_written =
((channel_state->written_compute_storage_buffers >> index) & 1) != 0;
if (is_written) {
MarkWrittenBuffer(binding.buffer_id, binding.device_addr, size);
@@ -1492,7 +1440,7 @@ void BufferCache<P>::MarkWrittenBuffer(BufferId buffer_id, DAddr device_addr, u3
}
template <class P>
BufferId BufferCache<P>::FindBuffer(DAddr device_addr, u32 size, bool sparse_compatible) {
BufferId BufferCache<P>::FindBuffer(DAddr device_addr, u32 size) {
if (device_addr == 0) {
return NULL_BUFFER_ID;
}
@@ -1502,18 +1450,10 @@ BufferId BufferCache<P>::FindBuffer(DAddr device_addr, u32 size, bool sparse_com
Buffer& buffer = slot_buffers[buffer_id];
WaitForGpuFenceIfNeeded(buffer);
if (buffer.IsInBounds(device_addr, size)) {
bool usable = true;
if constexpr (requires { buffer.IsSparseCompatible(); }) {
if (sparse_compatible && !buffer.IsSparseCompatible()) {
usable = false;
}
}
if (usable) {
return buffer_id;
}
return buffer_id;
}
}
return CreateBuffer(device_addr, size, sparse_compatible);
return CreateBuffer(device_addr, size);
}
template <class P>
@@ -1635,15 +1575,13 @@ void BufferCache<P>::JoinOverlap(BufferId new_buffer_id, BufferId overlap_id,
}
template <class P>
BufferId BufferCache<P>::CreateBuffer(DAddr device_addr, u32 wanted_size,
bool sparse_compatible) {
BufferId BufferCache<P>::CreateBuffer(DAddr device_addr, u32 wanted_size) {
DAddr device_addr_end = Common::AlignUp(device_addr + wanted_size, CACHING_PAGESIZE);
device_addr = Common::AlignDown(device_addr, CACHING_PAGESIZE);
wanted_size = static_cast<u32>(device_addr_end - device_addr);
const OverlapResult overlap = ResolveOverlaps(device_addr, wanted_size);
const u32 size = static_cast<u32>(overlap.end - overlap.begin);
const BufferId new_buffer_id =
slot_buffers.insert(runtime, overlap.begin, size, sparse_compatible);
const BufferId new_buffer_id = slot_buffers.insert(runtime, overlap.begin, size);
auto& new_buffer = slot_buffers[new_buffer_id];
const size_t size_bytes = new_buffer.SizeBytes();
runtime.ClearBuffer(new_buffer, 0, size_bytes, 0);
@@ -1996,15 +1934,10 @@ Binding BufferCache<P>::StorageBufferBinding(GPUVAddr ssbo_addr, u32 cbuf_index,
// The end address used for size calculation does not need to be aligned
const DAddr cpu_end = Common::AlignUp(*device_addr + size, Core::DEVICE_PAGESIZE);
u32 binding_size = static_cast<u32>(cpu_end - *aligned_device_addr);
if (is_written) {
binding_size = aligned_size;
}
const Binding binding{
.device_addr = *aligned_device_addr,
.size = binding_size,
.size = is_written ? aligned_size : static_cast<u32>(cpu_end - *aligned_device_addr),
.buffer_id = BufferId{},
.gpu_addr = aligned_gpu_addr,
};
return binding;
}
@@ -29,7 +29,6 @@
#include "common/settings.h"
#include "common/slot_vector.h"
#include "video_core/buffer_cache/buffer_base.h"
#include "video_core/buffer_cache/virtual_range_cache.h"
#include "video_core/control/channel_state_cache.h"
#include "video_core/delayed_destruction_ring.h"
#include "video_core/dirty_flags.h"
@@ -84,7 +83,6 @@ struct Binding {
DAddr device_addr{};
u32 size{};
BufferId buffer_id;
GPUVAddr gpu_addr{};
};
struct TextureBufferBinding : Binding {
@@ -217,10 +215,6 @@ public:
void TickFrame();
bool BindMultiRangeStorage(const Binding& binding, bool is_written);
void UnmapGPUMemory(size_t as_id, GPUVAddr gpu_addr, size_t size);
void WriteMemory(DAddr device_addr, u64 size);
void CachedWriteMemory(DAddr device_addr, u64 size);
@@ -420,8 +414,7 @@ private:
void MarkWrittenBuffer(BufferId buffer_id, DAddr device_addr, u32 size);
[[nodiscard]] BufferId FindBuffer(DAddr device_addr, u32 size,
bool sparse_compatible = false);
[[nodiscard]] BufferId FindBuffer(DAddr device_addr, u32 size);
void WaitForGpuFenceIfNeeded(Buffer& buffer);
@@ -429,8 +422,7 @@ private:
void JoinOverlap(BufferId new_buffer_id, BufferId overlap_id, bool accumulate_stream_score);
[[nodiscard]] BufferId CreateBuffer(DAddr device_addr, u32 wanted_size,
bool sparse_compatible = false);
[[nodiscard]] BufferId CreateBuffer(DAddr device_addr, u32 wanted_size);
void Register(BufferId buffer_id);
@@ -522,7 +514,6 @@ private:
};
Common::LeastRecentlyUsedCache<LRUItemParams> lru_cache;
u64 frame_tick = 0;
VirtualRangeCache virtual_ranges;
u64 total_used_memory = 0;
u64 minimum_memory = 0;
u64 critical_memory = 0;
@@ -1,152 +0,0 @@
// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
#pragma once
#include <atomic>
#include <limits>
#include <mutex>
#include <optional>
#include <unordered_map>
#include <vector>
#include <boost/container/small_vector.hpp>
#include "common/common_types.h"
#include "video_core/memory_manager.h"
namespace VideoCommon {
struct VirtualSegment {
GPUVAddr gpu_addr;
DAddr device_addr;
u32 size;
};
using VirtualSegments = boost::container::small_vector<VirtualSegment, 8>;
class VirtualRangeCache {
public:
const VirtualSegments* Query(Tegra::MemoryManager& memory, GPUVAddr gpu_addr, u32 size) {
if (has_deferred.load(std::memory_order_acquire)) {
ApplyDeferred();
}
const size_t as_id = memory.GetID();
const u64 key = MakeKey(as_id, gpu_addr);
const auto it = entries.find(key);
if (it != entries.end() && it->second.as_id == as_id &&
it->second.gpu_addr == gpu_addr && it->second.size == size) {
return &it->second.segments;
}
Entry entry;
entry.as_id = as_id;
entry.gpu_addr = gpu_addr;
entry.size = size;
const auto ranges = memory.GetSubmappedRange(gpu_addr, size);
for (const auto& [range_addr, range_size] : ranges) {
const std::optional<DAddr> device_addr = memory.GpuToCpuAddress(range_addr);
if (!device_addr) {
break;
}
u32 segment_size = (std::numeric_limits<u32>::max)();
if (range_size < static_cast<size_t>(segment_size)) {
segment_size = static_cast<u32>(range_size);
}
entry.segments.push_back(VirtualSegment{
.gpu_addr = range_addr,
.device_addr = *device_addr,
.size = segment_size,
});
}
const auto result = entries.insert_or_assign(key, std::move(entry));
return &result.first->second.segments;
}
void Unmap(size_t as_id, GPUVAddr gpu_addr, u64 size) {
if (size == 0) {
return;
}
{
std::scoped_lock lock{deferred_mutex};
if (!deferred.empty()) {
DeferredUnmap& last = deferred.back();
if (last.as_id == as_id && last.gpu_addr + last.size == gpu_addr) {
last.size += size;
has_deferred.store(true, std::memory_order_release);
return;
}
}
deferred.push_back(DeferredUnmap{
.as_id = as_id,
.gpu_addr = gpu_addr,
.size = size,
});
}
has_deferred.store(true, std::memory_order_release);
}
void Clear() {
{
std::scoped_lock lock{deferred_mutex};
deferred.clear();
}
has_deferred.store(false, std::memory_order_release);
entries.clear();
}
private:
struct Entry {
size_t as_id{};
GPUVAddr gpu_addr{};
u32 size{};
VirtualSegments segments;
};
struct DeferredUnmap {
size_t as_id;
GPUVAddr gpu_addr;
u64 size;
};
static u64 MakeKey(size_t as_id, GPUVAddr gpu_addr) {
return (static_cast<u64>(as_id) << 48) ^ gpu_addr;
}
void ApplyDeferred() {
std::vector<DeferredUnmap> pending;
{
std::scoped_lock lock{deferred_mutex};
has_deferred.store(false, std::memory_order_release);
pending.swap(deferred);
}
if (pending.empty() || entries.empty()) {
return;
}
for (auto it = entries.begin(); it != entries.end();) {
const Entry& entry = it->second;
const GPUVAddr entry_end = entry.gpu_addr + entry.size;
bool overlaps = false;
for (const DeferredUnmap& unmap : pending) {
if (unmap.as_id != entry.as_id) {
continue;
}
if (entry.gpu_addr < unmap.gpu_addr + unmap.size && unmap.gpu_addr < entry_end) {
overlaps = true;
break;
}
}
if (overlaps) {
it = entries.erase(it);
} else {
++it;
}
}
}
std::unordered_map<u64, Entry> entries;
std::vector<DeferredUnmap> deferred;
std::mutex deferred_mutex;
std::atomic<bool> has_deferred{false};
};
} // namespace VideoCommon
-1
View File
@@ -8,7 +8,6 @@
#include <array>
#include <vector>
#include <type_traits>
#include "common/bit_field.h"
#include "common/common_funcs.h"
-7
View File
@@ -1,15 +1,8 @@
// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
// SPDX-FileCopyrightText: Copyright 2020 yuzu Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
#pragma once
#include <array>
#include <cstddef>
#include <type_traits>
#include "common/bit_field.h"
#include "common/common_funcs.h"
#include "common/common_types.h"
-1
View File
@@ -7,7 +7,6 @@
#pragma once
#include <memory>
#include <type_traits>
#include "common/common_types.h"
#include "common/scratch_buffer.h"
@@ -52,7 +52,7 @@ constexpr std::array PROGRAM_LUT{
Buffer::Buffer(BufferCacheRuntime&, VideoCommon::NullBufferParams null_params)
: VideoCommon::BufferBase(null_params) {}
Buffer::Buffer(BufferCacheRuntime& runtime, DAddr cpu_addr_, u64 size_bytes_, bool)
Buffer::Buffer(BufferCacheRuntime& runtime, DAddr cpu_addr_, u64 size_bytes_)
: VideoCommon::BufferBase(cpu_addr_, size_bytes_) {
buffer.Create();
if (runtime.device.HasDebuggingToolAttached()) {
@@ -23,8 +23,7 @@ class BufferCacheRuntime;
class Buffer : public VideoCommon::BufferBase {
public:
explicit Buffer(BufferCacheRuntime&, DAddr cpu_addr, u64 size_bytes,
bool sparse_compatible = false);
explicit Buffer(BufferCacheRuntime&, DAddr cpu_addr, u64 size_bytes);
explicit Buffer(BufferCacheRuntime&, VideoCommon::NullBufferParams);
void ImmediateUpload(size_t offset, std::span<const u8> data) noexcept;
@@ -56,8 +56,7 @@ size_t BytesPerIndex(VkIndexType index_type) {
}
}
vk::Buffer CreateBuffer(const Device& device, const MemoryAllocator& memory_allocator, u64 size,
VkDeviceSize sparse_alignment) {
vk::Buffer CreateBuffer(const Device& device, const MemoryAllocator& memory_allocator, u64 size) {
VkBufferUsageFlags flags =
VK_BUFFER_USAGE_TRANSFER_SRC_BIT | VK_BUFFER_USAGE_TRANSFER_DST_BIT |
VK_BUFFER_USAGE_UNIFORM_TEXEL_BUFFER_BIT | VK_BUFFER_USAGE_STORAGE_TEXEL_BUFFER_BIT |
@@ -83,9 +82,6 @@ vk::Buffer CreateBuffer(const Device& device, const MemoryAllocator& memory_allo
.queueFamilyIndexCount = 0,
.pQueueFamilyIndices = nullptr,
};
if (sparse_alignment > 1) {
return memory_allocator.CreateBuffer(buffer_ci, MemoryUsage::DeviceLocal, sparse_alignment);
}
return memory_allocator.CreateBuffer(buffer_ci, MemoryUsage::DeviceLocal);
}
} // Anonymous namespace
@@ -103,14 +99,10 @@ Buffer::Buffer(BufferCacheRuntime& runtime, VideoCommon::NullBufferParams null_p
}
}
Buffer::Buffer(BufferCacheRuntime& runtime, DAddr cpu_addr_, u64 size_bytes_,
bool sparse_compatible_)
Buffer::Buffer(BufferCacheRuntime& runtime, DAddr cpu_addr_, u64 size_bytes_)
: VideoCommon::BufferBase(cpu_addr_, size_bytes_), device{&runtime.device},
scheduler{&runtime.scheduler},
buffer{CreateBuffer(*device, runtime.memory_allocator, SizeBytes(),
runtime.SparseAlignmentFor(sparse_compatible_))},
tracker{SizeBytes()} {
sparse_compatible = sparse_compatible_;
buffer{CreateBuffer(*device, runtime.memory_allocator, SizeBytes())}, tracker{SizeBytes()} {
if (runtime.device.HasDebuggingToolAttached()) {
buffer.SetObjectNameEXT(fmt::format("Buffer {:#x}", CpuAddr()).c_str());
}
@@ -356,8 +348,7 @@ BufferCacheRuntime::BufferCacheRuntime(const Device& device_, MemoryAllocator& m
: device{device_}, memory_allocator{memory_allocator_}, scheduler{scheduler_},
staging_pool{staging_pool_}, guest_descriptor_queue{guest_descriptor_queue_},
quad_index_pass(device, scheduler, descriptor_pool, staging_pool,
compute_pass_descriptor_queue),
multi_range_buffers(device_, memory_allocator_, scheduler_) {
compute_pass_descriptor_queue) {
const VkDriverIdKHR driver_id = device.GetDriverID();
limit_dynamic_storage_buffers = driver_id == VK_DRIVER_ID_QUALCOMM_PROPRIETARY ||
driver_id == VK_DRIVER_ID_ARM_PROPRIETARY;
@@ -545,33 +536,6 @@ void BufferCacheRuntime::ClearBuffer(VkBuffer dest_buffer, u32 offset, size_t si
});
}
bool BufferCacheRuntime::BindMultiRangeStorageBuffer(u64 key) {
if (multi_range_sources.empty() || multi_range_total == 0) {
return false;
}
const MultiRangeRef ref = multi_range_buffers.Get(key, multi_range_sources, multi_range_total);
if (ref.handle == VK_NULL_HANDLE) {
return false;
}
if (ref.needs_gather) {
PreCopyBarrier();
VkDeviceSize dst_offset = 0;
for (const MultiRangeSource& source : multi_range_sources) {
const std::array<VideoCommon::BufferCopy, 1> copy{VideoCommon::BufferCopy{
.src_offset = static_cast<u64>(source.offset),
.dst_offset = static_cast<u64>(dst_offset),
.size = static_cast<size_t>(source.size),
}};
CopyBuffer(ref.handle, source.handle, copy, false);
dst_offset += source.size;
}
PostCopyBarrier();
multi_range_buffers.MarkGathered(key);
}
guest_descriptor_queue.AddBuffer(ref.handle, ref.address, 0, ref.size);
return true;
}
void BufferCacheRuntime::BindIndexBuffer(PrimitiveTopology topology, IndexFormat index_format,
u32 base_vertex, u32 num_indices, VkBuffer buffer,
u32 offset, [[maybe_unused]] u32 size) {
@@ -8,14 +8,11 @@
#include <limits>
#include <boost/container/small_vector.hpp>
#include "video_core/buffer_cache/buffer_cache_base.h"
#include "video_core/buffer_cache/memory_tracker_base.h"
#include "video_core/buffer_cache/usage_tracker.h"
#include "video_core/engines/maxwell_3d.h"
#include "video_core/renderer_vulkan/vk_compute_pass.h"
#include "video_core/renderer_vulkan/vk_multi_range_buffer.h"
#include "video_core/renderer_vulkan/vk_staging_buffer_pool.h"
#include "video_core/renderer_vulkan/vk_update_descriptor.h"
#include "video_core/surface.h"
@@ -34,8 +31,7 @@ class BufferCacheRuntime;
class Buffer : public VideoCommon::BufferBase {
public:
explicit Buffer(BufferCacheRuntime&, VideoCommon::NullBufferParams null_params);
explicit Buffer(BufferCacheRuntime& runtime, VAddr cpu_addr_, u64 size_bytes_,
bool sparse_compatible_ = false);
explicit Buffer(BufferCacheRuntime& runtime, VAddr cpu_addr_, u64 size_bytes_);
[[nodiscard]] VkBufferView View(u32 offset, u32 size, VideoCore::Surface::PixelFormat format);
@@ -47,14 +43,6 @@ public:
return device_address;
}
[[nodiscard]] bool IsSparseCompatible() const noexcept {
return sparse_compatible;
}
[[nodiscard]] vk::MemoryLocation Location() const noexcept {
return buffer.Location();
}
[[nodiscard]] bool IsRegionUsed(u64 offset, u64 size) const noexcept {
return tracker.IsUsed(offset, size);
}
@@ -89,7 +77,6 @@ private:
VkDeviceAddress device_address{};
u64 last_usage_tick{};
bool is_null{};
bool sparse_compatible{};
};
class QuadArrayIndexBuffer;
@@ -138,7 +125,7 @@ public:
void PreCopyBarrier();
void CopyBuffer(VkBuffer dst_buffer, VkBuffer src_buffer,
void CopyBuffer(VkBuffer src_buffer, VkBuffer dst_buffer,
std::span<const VideoCommon::BufferCopy> copies, bool barrier,
bool can_reorder_upload = false);
@@ -168,40 +155,6 @@ public:
return ref.mapped_span;
}
[[nodiscard]] VkDeviceSize SparseAlignmentFor(bool sparse_compatible) const noexcept {
if (!sparse_compatible || !multi_range_buffers.UsesSparse()) {
return 0;
}
return multi_range_buffers.BlockSize();
}
[[nodiscard]] bool PrefersSparseSources() const noexcept {
return multi_range_buffers.UsesSparse();
}
void ResetMultiRange() noexcept {
multi_range_sources.clear();
multi_range_total = 0;
}
void PushMultiRangeSource(const Buffer& buffer, u32 offset, u32 size) {
const vk::MemoryLocation location = buffer.Location();
multi_range_sources.push_back(MultiRangeSource{
.handle = buffer.Handle(),
.memory = location.memory,
.memory_offset = location.offset,
.offset = offset,
.size = size,
});
multi_range_total += size;
}
bool BindMultiRangeStorageBuffer(u64 key);
void InvalidateMultiRange(u64 key) {
multi_range_buffers.Invalidate(key);
}
void BindUniformBuffer(const Buffer& buffer, u32 offset, u32 size) {
BindBuffer(buffer, offset, size);
}
@@ -255,10 +208,6 @@ private:
std::unique_ptr<Uint8Pass> uint8_pass;
QuadIndexedPass quad_index_pass;
MultiRangeBufferCache multi_range_buffers;
boost::container::small_vector<MultiRangeSource, 16> multi_range_sources;
VkDeviceSize multi_range_total{};
bool limit_dynamic_storage_buffers = false;
u32 max_dynamic_storage_buffers = (std::numeric_limits<u32>::max)();
};
@@ -1,304 +0,0 @@
// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
#include <mutex>
#include "video_core/renderer_vulkan/vk_multi_range_buffer.h"
#include "video_core/renderer_vulkan/vk_scheduler.h"
#include "video_core/vulkan_common/vulkan_device.h"
namespace Vulkan {
MultiRangeBufferCache::MultiRangeBufferCache(const Device& device_,
MemoryAllocator& memory_allocator_,
Scheduler& scheduler_)
: device{device_}, memory_allocator{memory_allocator_}, scheduler{scheduler_} {
sparse_usage = VK_BUFFER_USAGE_TRANSFER_SRC_BIT | VK_BUFFER_USAGE_TRANSFER_DST_BIT |
VK_BUFFER_USAGE_STORAGE_BUFFER_BIT;
if (device.IsBufferDeviceAddressSupported()) {
sparse_usage |= VK_BUFFER_USAGE_SHADER_DEVICE_ADDRESS_BIT;
}
if (!device.IsSparseBindingSupported()) {
return;
}
const VkDeviceSize queried = QueryBlockSize();
if (queried == 0) {
return;
}
block_size = queried;
use_sparse = true;
}
MultiRangeBufferCache::~MultiRangeBufferCache() {
const VkDevice logical = *device.GetLogical();
const auto& dld = device.GetDispatchLoader();
for (auto& [key, entry] : entries) {
if (entry.sparse_handle != VK_NULL_HANDLE) {
dld.vkDestroyBuffer(logical, entry.sparse_handle, nullptr);
}
}
entries.clear();
for (const Retired& item : retired) {
dld.vkDestroyBuffer(logical, item.handle, nullptr);
}
retired.clear();
}
VkDeviceSize MultiRangeBufferCache::QueryBlockSize() const {
const VkDevice logical = *device.GetLogical();
const auto& dld = device.GetDispatchLoader();
const VkBufferCreateInfo probe_ci{
.sType = VK_STRUCTURE_TYPE_BUFFER_CREATE_INFO,
.pNext = nullptr,
.flags = VK_BUFFER_CREATE_SPARSE_BINDING_BIT | VK_BUFFER_CREATE_SPARSE_ALIASED_BIT,
.size = DEFAULT_BLOCK_SIZE,
.usage = sparse_usage,
.sharingMode = VK_SHARING_MODE_EXCLUSIVE,
.queueFamilyIndexCount = 0,
.pQueueFamilyIndices = nullptr,
};
VkBuffer probe{};
if (dld.vkCreateBuffer(logical, &probe_ci, nullptr, &probe) != VK_SUCCESS) {
return 0;
}
const VkBufferMemoryRequirementsInfo2 reqs_info{
.sType = VK_STRUCTURE_TYPE_BUFFER_MEMORY_REQUIREMENTS_INFO_2,
.pNext = nullptr,
.buffer = probe,
};
VkMemoryRequirements2 reqs2{
.sType = VK_STRUCTURE_TYPE_MEMORY_REQUIREMENTS_2,
.pNext = nullptr,
.memoryRequirements = {},
};
dld.vkGetBufferMemoryRequirements2(logical, &reqs_info, &reqs2);
dld.vkDestroyBuffer(logical, probe, nullptr);
return reqs2.memoryRequirements.alignment;
}
u64 MultiRangeBufferCache::HashSources(std::span<const MultiRangeSource> sources) const {
u64 hash = 0xcbf29ce484222325ULL;
const auto mix = [&hash](u64 value) {
hash ^= value;
hash *= 0x100000001b3ULL;
};
for (const MultiRangeSource& source : sources) {
mix(reinterpret_cast<u64>(source.handle));
mix(static_cast<u64>(source.offset));
mix(static_cast<u64>(source.size));
}
return hash;
}
bool MultiRangeBufferCache::CanBindSparse(std::span<const MultiRangeSource> sources) const {
if (!use_sparse) {
return false;
}
for (const MultiRangeSource& source : sources) {
if (source.memory == VK_NULL_HANDLE) {
return false;
}
const VkDeviceSize memory_offset = source.memory_offset + source.offset;
if ((memory_offset % block_size) != 0) {
return false;
}
if ((source.size % block_size) != 0) {
return false;
}
}
return true;
}
VkBuffer MultiRangeBufferCache::CreateSparse(std::span<const MultiRangeSource> sources,
VkDeviceSize total) {
const VkDevice logical = *device.GetLogical();
const auto& dld = device.GetDispatchLoader();
const VkBufferCreateInfo buffer_ci{
.sType = VK_STRUCTURE_TYPE_BUFFER_CREATE_INFO,
.pNext = nullptr,
.flags = VK_BUFFER_CREATE_SPARSE_BINDING_BIT | VK_BUFFER_CREATE_SPARSE_ALIASED_BIT,
.size = total,
.usage = sparse_usage,
.sharingMode = VK_SHARING_MODE_EXCLUSIVE,
.queueFamilyIndexCount = 0,
.pQueueFamilyIndices = nullptr,
};
VkBuffer handle{};
if (dld.vkCreateBuffer(logical, &buffer_ci, nullptr, &handle) != VK_SUCCESS) {
return VK_NULL_HANDLE;
}
std::vector<VkSparseMemoryBind> binds;
binds.reserve(sources.size());
VkDeviceSize resource_offset = 0;
for (const MultiRangeSource& source : sources) {
binds.push_back(VkSparseMemoryBind{
.resourceOffset = resource_offset,
.size = source.size,
.memory = source.memory,
.memoryOffset = source.memory_offset + source.offset,
.flags = 0,
});
resource_offset += source.size;
}
const VkSparseBufferMemoryBindInfo buffer_bind{
.buffer = handle,
.bindCount = static_cast<u32>(binds.size()),
.pBinds = binds.data(),
};
const VkBindSparseInfo bind_info{
.sType = VK_STRUCTURE_TYPE_BIND_SPARSE_INFO,
.pNext = nullptr,
.waitSemaphoreCount = 0,
.pWaitSemaphores = nullptr,
.bufferBindCount = 1,
.pBufferBinds = &buffer_bind,
.imageOpaqueBindCount = 0,
.pImageOpaqueBinds = nullptr,
.imageBindCount = 0,
.pImageBinds = nullptr,
.signalSemaphoreCount = 0,
.pSignalSemaphores = nullptr,
};
const VkFenceCreateInfo fence_ci{
.sType = VK_STRUCTURE_TYPE_FENCE_CREATE_INFO,
.pNext = nullptr,
.flags = 0,
};
vk::Fence fence = device.GetLogical().CreateFence(fence_ci);
VkResult bind_result = VK_ERROR_UNKNOWN;
{
std::scoped_lock lock{scheduler.submit_mutex};
bind_result = device.GetGraphicsQueue().BindSparse(bind_info, *fence);
}
if (bind_result != VK_SUCCESS) {
dld.vkDestroyBuffer(logical, handle, nullptr);
return VK_NULL_HANDLE;
}
fence.Wait();
return handle;
}
void MultiRangeBufferCache::DestroySparse(VkBuffer handle) {
if (handle == VK_NULL_HANDLE) {
return;
}
retired.push_back(Retired{
.handle = handle,
.tick = scheduler.CurrentTick(),
});
}
void MultiRangeBufferCache::DrainRetired() {
const VkDevice logical = *device.GetLogical();
const auto& dld = device.GetDispatchLoader();
size_t index = 0;
while (index < retired.size()) {
if (scheduler.IsFree(retired[index].tick)) {
dld.vkDestroyBuffer(logical, retired[index].handle, nullptr);
retired[index] = retired.back();
retired.pop_back();
} else {
++index;
}
}
}
MultiRangeRef MultiRangeBufferCache::Get(u64 key, std::span<const MultiRangeSource> sources,
VkDeviceSize total) {
if (sources.empty() || total == 0) {
return MultiRangeRef{};
}
if (!retired.empty()) {
DrainRetired();
}
const u64 geometry = HashSources(sources);
const auto it = entries.find(key);
if (it != entries.end() && it->second.geometry == geometry && it->second.size == total) {
Entry& entry = it->second;
MultiRangeRef ref{
.handle = entry.sparse_handle,
.address = entry.address,
.size = entry.size,
.needs_gather = false,
};
if (entry.sparse_handle == VK_NULL_HANDLE) {
ref.handle = *entry.gathered;
ref.needs_gather = entry.dirty;
}
return ref;
}
if (it != entries.end()) {
DestroySparse(it->second.sparse_handle);
entries.erase(it);
}
Entry entry;
entry.geometry = geometry;
entry.size = total;
if (CanBindSparse(sources)) {
entry.sparse_handle = CreateSparse(sources, total);
}
if (entry.sparse_handle == VK_NULL_HANDLE) {
VkBufferUsageFlags flags = VK_BUFFER_USAGE_TRANSFER_SRC_BIT |
VK_BUFFER_USAGE_TRANSFER_DST_BIT |
VK_BUFFER_USAGE_STORAGE_BUFFER_BIT;
if (device.IsBufferDeviceAddressSupported()) {
flags |= VK_BUFFER_USAGE_SHADER_DEVICE_ADDRESS_BIT;
}
const VkBufferCreateInfo gather_ci{
.sType = VK_STRUCTURE_TYPE_BUFFER_CREATE_INFO,
.pNext = nullptr,
.flags = 0,
.size = total,
.usage = flags,
.sharingMode = VK_SHARING_MODE_EXCLUSIVE,
.queueFamilyIndexCount = 0,
.pQueueFamilyIndices = nullptr,
};
entry.gathered = memory_allocator.CreateBuffer(gather_ci, MemoryUsage::DeviceLocal);
entry.dirty = true;
}
if (device.IsBufferDeviceAddressSupported()) {
VkBuffer address_handle = entry.sparse_handle;
if (address_handle == VK_NULL_HANDLE) {
address_handle = *entry.gathered;
}
entry.address = device.GetLogical().GetBufferDeviceAddress(address_handle);
}
MultiRangeRef ref{
.handle = entry.sparse_handle,
.address = entry.address,
.size = entry.size,
.needs_gather = false,
};
if (entry.sparse_handle == VK_NULL_HANDLE) {
ref.handle = *entry.gathered;
ref.needs_gather = true;
}
entries.emplace(key, std::move(entry));
return ref;
}
void MultiRangeBufferCache::MarkGathered(u64 key) {
const auto it = entries.find(key);
if (it != entries.end()) {
it->second.dirty = false;
}
}
void MultiRangeBufferCache::Invalidate(u64 key) {
const auto it = entries.find(key);
if (it != entries.end()) {
it->second.dirty = true;
}
}
void MultiRangeBufferCache::Clear() {
for (auto& [key, entry] : entries) {
DestroySparse(entry.sparse_handle);
}
entries.clear();
}
} // namespace Vulkan
@@ -1,100 +0,0 @@
// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
#pragma once
#include <span>
#include <unordered_map>
#include <vector>
#include "common/common_types.h"
#include "video_core/vulkan_common/vulkan_memory_allocator.h"
#include "video_core/vulkan_common/vulkan_wrapper.h"
namespace Vulkan {
class Device;
class Scheduler;
struct MultiRangeSource {
VkBuffer handle{};
VkDeviceMemory memory{};
VkDeviceSize memory_offset{};
VkDeviceSize offset{};
VkDeviceSize size{};
};
struct MultiRangeRef {
VkBuffer handle{};
VkDeviceAddress address{};
VkDeviceSize size{};
bool needs_gather{};
};
class MultiRangeBufferCache final {
public:
static constexpr VkDeviceSize DEFAULT_BLOCK_SIZE = 64 * 1024;
explicit MultiRangeBufferCache(const Device& device_, MemoryAllocator& memory_allocator_,
Scheduler& scheduler_);
~MultiRangeBufferCache();
MultiRangeBufferCache(const MultiRangeBufferCache&) = delete;
MultiRangeBufferCache& operator=(const MultiRangeBufferCache&) = delete;
[[nodiscard]] bool UsesSparse() const noexcept {
return use_sparse;
}
[[nodiscard]] VkDeviceSize BlockSize() const noexcept {
return block_size;
}
[[nodiscard]] MultiRangeRef Get(u64 key, std::span<const MultiRangeSource> sources,
VkDeviceSize total);
void MarkGathered(u64 key);
void Invalidate(u64 key);
void Clear();
private:
struct Retired {
VkBuffer handle{};
u64 tick{};
};
struct Entry {
vk::Buffer gathered;
VkBuffer sparse_handle{};
VkDeviceAddress address{};
VkDeviceSize size{};
u64 geometry{};
bool dirty{true};
};
[[nodiscard]] u64 HashSources(std::span<const MultiRangeSource> sources) const;
[[nodiscard]] bool CanBindSparse(std::span<const MultiRangeSource> sources) const;
[[nodiscard]] VkBuffer CreateSparse(std::span<const MultiRangeSource> sources,
VkDeviceSize total);
[[nodiscard]] VkDeviceSize QueryBlockSize() const;
void DestroySparse(VkBuffer handle);
void DrainRetired();
const Device& device;
MemoryAllocator& memory_allocator;
Scheduler& scheduler;
bool use_sparse{};
VkDeviceSize block_size{DEFAULT_BLOCK_SIZE};
VkBufferUsageFlags sparse_usage{};
std::unordered_map<u64, Entry> entries;
std::vector<Retired> retired;
};
} // namespace Vulkan
@@ -338,7 +338,7 @@ void PresentManager::PresentThread(std::stop_token token) {
// By exchanging the lock ownership we take the swapchain lock
// before the queue lock goes out of scope. This way the swapchain
// lock in WaitPresent is guaranteed to occur after here.
void(std::exchange(lock, std::unique_lock{swapchain_mutex}));
std::exchange(lock, std::unique_lock{swapchain_mutex});
CopyToSwapchain(frame);
// Free the frame for reuse
@@ -819,7 +819,6 @@ void RasterizerVulkan::ModifyGPUMemory(size_t as_id, GPUVAddr addr, u64 size) {
std::scoped_lock lock{texture_cache.mutex};
texture_cache.UnmapGPUMemory(as_id, addr, size);
}
buffer_cache.UnmapGPUMemory(as_id, addr, size);
}
void RasterizerVulkan::SignalFence(std::function<void()>&& func) {
@@ -300,7 +300,7 @@ void Scheduler::WorkerThread(std::stop_token stop_token) {
// Exchange lock ownership so that we take the execution lock before
// the queue lock goes out of scope. This allows us to force execution
// to complete in the next step.
void(std::exchange(lk, std::unique_lock{execution_mutex}));
std::exchange(lk, std::unique_lock{execution_mutex});
// Perform the work, tracking whether the chunk was a submission
// before executing.
@@ -1570,8 +1570,6 @@ void Device::SetupFamilies(VkSurfaceKHR surface) {
}
if (graphics) {
graphics_family = *graphics;
graphics_family_sparse_binding =
(queue_family_properties[*graphics].queueFlags & VK_QUEUE_SPARSE_BINDING_BIT) != 0;
}
if (present) {
present_family = *present;
@@ -317,10 +317,6 @@ public:
return properties.driver.driverID;
}
bool IsSparseBindingSupported() const {
return features.features.sparseBinding && graphics_family_sparse_binding;
}
/// Returns true for tile-based deferred renderers.
bool IsTiler() const {
switch (GetDriverID()) {
@@ -1150,7 +1146,6 @@ private:
bool owns_static_pipeline_cache{};
u32 instance_version{}; ///< Vulkan instance version.
u32 graphics_family{}; ///< Main graphics queue family index.
bool graphics_family_sparse_binding{};
u32 present_family{}; ///< Main present queue family index.
struct Extensions {
@@ -26,408 +26,350 @@
#include "common/settings.h"
namespace Vulkan {
namespace {
namespace {
// Helpers translating MemoryUsage to flags/usage
[[maybe_unused]] VkMemoryPropertyFlags MemoryUsagePropertyFlags(MemoryUsage usage) {
switch (usage) {
case MemoryUsage::DeviceLocal:
return VK_MEMORY_PROPERTY_DEVICE_LOCAL_BIT;
case MemoryUsage::Upload:
return VK_MEMORY_PROPERTY_HOST_VISIBLE_BIT |
VK_MEMORY_PROPERTY_HOST_COHERENT_BIT;
case MemoryUsage::Download:
return VK_MEMORY_PROPERTY_HOST_VISIBLE_BIT |
VK_MEMORY_PROPERTY_HOST_COHERENT_BIT |
VK_MEMORY_PROPERTY_HOST_CACHED_BIT;
case MemoryUsage::Stream:
return VK_MEMORY_PROPERTY_DEVICE_LOCAL_BIT |
VK_MEMORY_PROPERTY_HOST_VISIBLE_BIT |
VK_MEMORY_PROPERTY_HOST_COHERENT_BIT;
[[maybe_unused]] VkMemoryPropertyFlags MemoryUsagePropertyFlags(MemoryUsage usage) {
switch (usage) {
case MemoryUsage::DeviceLocal:
return VK_MEMORY_PROPERTY_DEVICE_LOCAL_BIT;
case MemoryUsage::Upload:
return VK_MEMORY_PROPERTY_HOST_VISIBLE_BIT |
VK_MEMORY_PROPERTY_HOST_COHERENT_BIT;
case MemoryUsage::Download:
return VK_MEMORY_PROPERTY_HOST_VISIBLE_BIT |
VK_MEMORY_PROPERTY_HOST_COHERENT_BIT |
VK_MEMORY_PROPERTY_HOST_CACHED_BIT;
case MemoryUsage::Stream:
return VK_MEMORY_PROPERTY_DEVICE_LOCAL_BIT |
VK_MEMORY_PROPERTY_HOST_VISIBLE_BIT |
VK_MEMORY_PROPERTY_HOST_COHERENT_BIT;
}
ASSERT_MSG(false, "Invalid memory usage={}", usage);
return VK_MEMORY_PROPERTY_HOST_VISIBLE_BIT | VK_MEMORY_PROPERTY_HOST_COHERENT_BIT;
}
ASSERT_MSG(false, "Invalid memory usage={}", usage);
return VK_MEMORY_PROPERTY_HOST_VISIBLE_BIT | VK_MEMORY_PROPERTY_HOST_COHERENT_BIT;
}
[[nodiscard]] VkMemoryPropertyFlags MemoryUsagePreferredVmaFlags(MemoryUsage usage) {
if (usage == MemoryUsage::Download) {
return VK_MEMORY_PROPERTY_HOST_CACHED_BIT | VK_MEMORY_PROPERTY_HOST_COHERENT_BIT;
[[nodiscard]] VkMemoryPropertyFlags MemoryUsagePreferredVmaFlags(MemoryUsage usage) {
if (usage == MemoryUsage::Download) {
return VK_MEMORY_PROPERTY_HOST_CACHED_BIT | VK_MEMORY_PROPERTY_HOST_COHERENT_BIT;
}
return usage != MemoryUsage::DeviceLocal ? VK_MEMORY_PROPERTY_HOST_COHERENT_BIT
: VkMemoryPropertyFlagBits{};
}
return usage != MemoryUsage::DeviceLocal ? VK_MEMORY_PROPERTY_HOST_COHERENT_BIT
: VkMemoryPropertyFlagBits{};
}
[[nodiscard]] VmaAllocationCreateFlags MemoryUsageVmaFlags(MemoryUsage usage) {
switch (usage) {
case MemoryUsage::Upload:
case MemoryUsage::Stream:
return VMA_ALLOCATION_CREATE_MAPPED_BIT |
VMA_ALLOCATION_CREATE_HOST_ACCESS_SEQUENTIAL_WRITE_BIT;
case MemoryUsage::Download:
return VMA_ALLOCATION_CREATE_MAPPED_BIT |
VMA_ALLOCATION_CREATE_HOST_ACCESS_RANDOM_BIT;
case MemoryUsage::DeviceLocal:
return {};
[[nodiscard]] VmaAllocationCreateFlags MemoryUsageVmaFlags(MemoryUsage usage) {
switch (usage) {
case MemoryUsage::Upload:
case MemoryUsage::Stream:
return VMA_ALLOCATION_CREATE_MAPPED_BIT |
VMA_ALLOCATION_CREATE_HOST_ACCESS_SEQUENTIAL_WRITE_BIT;
case MemoryUsage::Download:
return VMA_ALLOCATION_CREATE_MAPPED_BIT |
VMA_ALLOCATION_CREATE_HOST_ACCESS_RANDOM_BIT;
case MemoryUsage::DeviceLocal:
return {};
}
return {};
}
return {};
}
[[nodiscard]] VmaMemoryUsage MemoryUsageVma(MemoryUsage usage) {
switch (usage) {
case MemoryUsage::DeviceLocal:
case MemoryUsage::Stream:
return VMA_MEMORY_USAGE_AUTO_PREFER_DEVICE;
case MemoryUsage::Upload:
case MemoryUsage::Download:
return VMA_MEMORY_USAGE_AUTO_PREFER_HOST;
[[nodiscard]] VmaMemoryUsage MemoryUsageVma(MemoryUsage usage) {
switch (usage) {
case MemoryUsage::DeviceLocal:
case MemoryUsage::Stream:
return VMA_MEMORY_USAGE_AUTO_PREFER_DEVICE;
case MemoryUsage::Upload:
case MemoryUsage::Download:
return VMA_MEMORY_USAGE_AUTO_PREFER_HOST;
}
return VMA_MEMORY_USAGE_AUTO_PREFER_DEVICE;
}
return VMA_MEMORY_USAGE_AUTO_PREFER_DEVICE;
}
} // namespace
// This avoids calling vkGetBufferMemoryRequirements* directly.
template<typename T>
static VkBuffer GetVkHandleFromBuffer(const T &buf) {
if constexpr (requires { static_cast<VkBuffer>(buf); }) {
return static_cast<VkBuffer>(buf);
} else if constexpr (requires {{ buf.GetHandle() } -> std::convertible_to<VkBuffer>; }) {
return buf.GetHandle();
} else if constexpr (requires {{ buf.Handle() } -> std::convertible_to<VkBuffer>; }) {
return buf.Handle();
} else if constexpr (requires {{ buf.vk_handle() } -> std::convertible_to<VkBuffer>; }) {
return buf.vk_handle();
} else {
static_assert(sizeof(T) == 0, "Cannot extract VkBuffer handle from vk::Buffer");
return VK_NULL_HANDLE;
}
}
} // namespace
//MemoryCommit is now VMA-backed
MemoryCommit::MemoryCommit(VmaAllocator alloc, VmaAllocation a,
const VmaAllocationInfo &info) noexcept
: allocator{alloc}, allocation{a}, memory{info.deviceMemory},
offset{info.offset}, size{info.size}, mapped_ptr{info.pMappedData} {
// Log GPU memory allocation
if (GPU::Logging::IsActive() &&
Settings::values.gpu_log_memory_tracking.GetValue()) {
GPU::Logging::GPULogger::GetInstance().LogMemoryAllocation(
reinterpret_cast<uintptr_t>(memory),
static_cast<u64>(size),
0 // Memory property flags (not easily available from VMA)
);
}
}
MemoryCommit::~MemoryCommit() { Release(); }
MemoryCommit::MemoryCommit(MemoryCommit &&rhs) noexcept
: allocator{std::exchange(rhs.allocator, nullptr)},
allocation{std::exchange(rhs.allocation, nullptr)},
memory{std::exchange(rhs.memory, VK_NULL_HANDLE)},
offset{std::exchange(rhs.offset, 0)},
size{std::exchange(rhs.size, 0)},
mapped_ptr{std::exchange(rhs.mapped_ptr, nullptr)} {}
MemoryCommit &MemoryCommit::operator=(MemoryCommit &&rhs) noexcept {
if (this != &rhs) {
Release();
allocator = std::exchange(rhs.allocator, nullptr);
allocation = std::exchange(rhs.allocation, nullptr);
memory = std::exchange(rhs.memory, VK_NULL_HANDLE);
offset = std::exchange(rhs.offset, 0);
size = std::exchange(rhs.size, 0);
mapped_ptr = std::exchange(rhs.mapped_ptr, nullptr);
}
return *this;
}
std::span<u8> MemoryCommit::Map()
{
if (!allocation) return {};
if (!mapped_ptr) {
if (vmaMapMemory(allocator, allocation, &mapped_ptr) != VK_SUCCESS) return {};
}
const size_t n = static_cast<size_t>(std::min<VkDeviceSize>(size,
(std::numeric_limits<size_t>::max)()));
return std::span<u8>{static_cast<u8 *>(mapped_ptr), n};
}
std::span<const u8> MemoryCommit::Map() const
{
if (!allocation) return {};
if (!mapped_ptr) {
void *p = nullptr;
if (vmaMapMemory(allocator, allocation, &p) != VK_SUCCESS) return {};
const_cast<MemoryCommit *>(this)->mapped_ptr = p;
}
const size_t n = static_cast<size_t>(std::min<VkDeviceSize>(size,
(std::numeric_limits<size_t>::max)()));
return std::span<const u8>{static_cast<const u8 *>(mapped_ptr), n};
}
void MemoryCommit::Unmap()
{
if (allocation && mapped_ptr) {
vmaUnmapMemory(allocator, allocation);
mapped_ptr = nullptr;
}
}
void MemoryCommit::Release() {
if (allocation && allocator) {
// Log GPU memory deallocation
MemoryCommit::MemoryCommit(VmaAllocator alloc, VmaAllocation a,
const VmaAllocationInfo &info) noexcept
: allocator{alloc}, allocation{a}, memory{info.deviceMemory},
offset{info.offset}, size{info.size}, mapped_ptr{info.pMappedData} {
// Log GPU memory allocation
if (GPU::Logging::IsActive() &&
Settings::values.gpu_log_memory_tracking.GetValue() &&
memory != VK_NULL_HANDLE) {
GPU::Logging::GPULogger::GetInstance().LogMemoryDeallocation(
reinterpret_cast<uintptr_t>(memory)
Settings::values.gpu_log_memory_tracking.GetValue()) {
GPU::Logging::GPULogger::GetInstance().LogMemoryAllocation(
reinterpret_cast<uintptr_t>(memory),
static_cast<u64>(size),
0 // Memory property flags (not easily available from VMA)
);
}
}
if (mapped_ptr) {
MemoryCommit::~MemoryCommit() { Release(); }
MemoryCommit::MemoryCommit(MemoryCommit &&rhs) noexcept
: allocator{std::exchange(rhs.allocator, nullptr)},
allocation{std::exchange(rhs.allocation, nullptr)},
memory{std::exchange(rhs.memory, VK_NULL_HANDLE)},
offset{std::exchange(rhs.offset, 0)},
size{std::exchange(rhs.size, 0)},
mapped_ptr{std::exchange(rhs.mapped_ptr, nullptr)} {}
MemoryCommit &MemoryCommit::operator=(MemoryCommit &&rhs) noexcept {
if (this != &rhs) {
Release();
allocator = std::exchange(rhs.allocator, nullptr);
allocation = std::exchange(rhs.allocation, nullptr);
memory = std::exchange(rhs.memory, VK_NULL_HANDLE);
offset = std::exchange(rhs.offset, 0);
size = std::exchange(rhs.size, 0);
mapped_ptr = std::exchange(rhs.mapped_ptr, nullptr);
}
return *this;
}
std::span<u8> MemoryCommit::Map()
{
if (!allocation) return {};
if (!mapped_ptr) {
if (vmaMapMemory(allocator, allocation, &mapped_ptr) != VK_SUCCESS) return {};
}
const size_t n = static_cast<size_t>(std::min<VkDeviceSize>(size,
(std::numeric_limits<size_t>::max)()));
return std::span<u8>{static_cast<u8 *>(mapped_ptr), n};
}
std::span<const u8> MemoryCommit::Map() const
{
if (!allocation) return {};
if (!mapped_ptr) {
void *p = nullptr;
if (vmaMapMemory(allocator, allocation, &p) != VK_SUCCESS) return {};
const_cast<MemoryCommit *>(this)->mapped_ptr = p;
}
const size_t n = static_cast<size_t>(std::min<VkDeviceSize>(size,
(std::numeric_limits<size_t>::max)()));
return std::span<const u8>{static_cast<const u8 *>(mapped_ptr), n};
}
void MemoryCommit::Unmap()
{
if (allocation && mapped_ptr) {
vmaUnmapMemory(allocator, allocation);
mapped_ptr = nullptr;
}
vmaFreeMemory(allocator, allocation);
}
allocation = nullptr;
allocator = nullptr;
memory = VK_NULL_HANDLE;
offset = 0;
size = 0;
}
MemoryAllocator::MemoryAllocator(const Device &device_)
: device{device_}, allocator{device.GetAllocator()},
properties{device_.GetPhysical().GetMemoryProperties().memoryProperties},
buffer_image_granularity{
device_.GetPhysical().GetProperties().limits.bufferImageGranularity} {
void MemoryCommit::Release() {
if (allocation && allocator) {
// Log GPU memory deallocation
if (GPU::Logging::IsActive() &&
Settings::values.gpu_log_memory_tracking.GetValue() &&
memory != VK_NULL_HANDLE) {
GPU::Logging::GPULogger::GetInstance().LogMemoryDeallocation(
reinterpret_cast<uintptr_t>(memory)
);
}
// Preserve the previous "RenderDoc small heap" trimming behavior that we had in original vma minus the heap bug
if (device.HasDebuggingToolAttached())
{
using namespace Common::Literals;
ForEachDeviceLocalHostVisibleHeap(device, [this](size_t heap_idx, VkMemoryHeap &heap) {
if (heap.size <= 256_MiB) {
for (u32 t = 0; t < properties.memoryTypeCount; ++t) {
if (properties.memoryTypes[t].heapIndex == heap_idx) {
valid_memory_types &= ~(1u << t);
if (mapped_ptr) {
vmaUnmapMemory(allocator, allocation);
mapped_ptr = nullptr;
}
vmaFreeMemory(allocator, allocation);
}
allocation = nullptr;
allocator = nullptr;
memory = VK_NULL_HANDLE;
offset = 0;
size = 0;
}
MemoryAllocator::MemoryAllocator(const Device &device_)
: device{device_}, allocator{device.GetAllocator()},
properties{device_.GetPhysical().GetMemoryProperties().memoryProperties},
buffer_image_granularity{
device_.GetPhysical().GetProperties().limits.bufferImageGranularity} {
// Preserve the previous "RenderDoc small heap" trimming behavior that we had in original vma minus the heap bug
if (device.HasDebuggingToolAttached())
{
using namespace Common::Literals;
ForEachDeviceLocalHostVisibleHeap(device, [this](size_t heap_idx, VkMemoryHeap &heap) {
if (heap.size <= 256_MiB) {
for (u32 t = 0; t < properties.memoryTypeCount; ++t) {
if (properties.memoryTypes[t].heapIndex == heap_idx) {
valid_memory_types &= ~(1u << t);
}
}
}
}
});
});
}
}
}
MemoryAllocator::~MemoryAllocator() = default;
MemoryAllocator::~MemoryAllocator() = default;
vk::Image MemoryAllocator::CreateImage(const VkImageCreateInfo &ci) const
{
const VmaAllocationCreateInfo alloc_ci = {
.flags = VMA_ALLOCATION_CREATE_WITHIN_BUDGET_BIT,
.usage = VMA_MEMORY_USAGE_AUTO_PREFER_DEVICE,
vk::Image MemoryAllocator::CreateImage(const VkImageCreateInfo &ci) const
{
const VmaAllocationCreateInfo alloc_ci = {
.flags = VMA_ALLOCATION_CREATE_WITHIN_BUDGET_BIT,
.usage = VMA_MEMORY_USAGE_AUTO_PREFER_DEVICE,
.requiredFlags = 0,
.preferredFlags = VK_MEMORY_PROPERTY_DEVICE_LOCAL_BIT,
.memoryTypeBits = 0,
.pool = VK_NULL_HANDLE,
.pUserData = nullptr,
.priority = 0.f,
};
VkImage handle{};
VmaAllocation allocation{};
VmaAllocationInfo alloc_info{};
vk::Check(vmaCreateImage(allocator, &ci, &alloc_ci, &handle, &allocation, &alloc_info));
// Log GPU memory allocation for images
if (GPU::Logging::IsActive() &&
Settings::values.gpu_log_memory_tracking.GetValue()) {
GPU::Logging::GPULogger::GetInstance().LogMemoryAllocation(
reinterpret_cast<uintptr_t>(alloc_info.deviceMemory),
static_cast<u64>(alloc_info.size),
VK_MEMORY_PROPERTY_DEVICE_LOCAL_BIT
);
}
return vk::Image(handle, ci.usage, *device.GetLogical(), allocator, allocation,
device.GetDispatchLoader());
}
vk::Buffer MemoryAllocator::CreateBuffer(const VkBufferCreateInfo &ci, MemoryUsage usage) const {
// MESA will do memcpy() if not marked as host cached, so just force mark it for most buffers
auto const anv_flags = (usage == MemoryUsage::Stream
&& device.GetDriverID() == VK_DRIVER_ID_INTEL_OPEN_SOURCE_MESA)
? VK_MEMORY_PROPERTY_HOST_CACHED_BIT : 0;
const VmaAllocationCreateInfo alloc_ci = {
.flags = VMA_ALLOCATION_CREATE_WITHIN_BUDGET_BIT | MemoryUsageVmaFlags(usage),
.usage = MemoryUsageVma(usage),
.requiredFlags = 0,
.preferredFlags = VK_MEMORY_PROPERTY_DEVICE_LOCAL_BIT,
.memoryTypeBits = 0,
.preferredFlags = MemoryUsagePreferredVmaFlags(usage) | anv_flags,
.memoryTypeBits = usage == MemoryUsage::Stream ? 0u : valid_memory_types,
.pool = VK_NULL_HANDLE,
.pUserData = nullptr,
.priority = 0.f,
};
};
VkImage handle{};
VmaAllocation allocation{};
VmaAllocationInfo alloc_info{};
vk::Check(vmaCreateImage(allocator, &ci, &alloc_ci, &handle, &allocation, &alloc_info));
VkBuffer handle{};
VmaAllocationInfo alloc_info{};
VmaAllocation allocation{};
VkMemoryPropertyFlags property_flags{};
// Log GPU memory allocation for images
if (GPU::Logging::IsActive() &&
Settings::values.gpu_log_memory_tracking.GetValue()) {
GPU::Logging::GPULogger::GetInstance().LogMemoryAllocation(
reinterpret_cast<uintptr_t>(alloc_info.deviceMemory),
static_cast<u64>(alloc_info.size),
VK_MEMORY_PROPERTY_DEVICE_LOCAL_BIT
);
vk::Check(vmaCreateBuffer(allocator, &ci, &alloc_ci, &handle, &allocation, &alloc_info));
vmaGetAllocationMemoryProperties(allocator, allocation, &property_flags);
// Log GPU memory allocation for buffers
if (GPU::Logging::IsActive() &&
Settings::values.gpu_log_memory_tracking.GetValue()) {
GPU::Logging::GPULogger::GetInstance().LogMemoryAllocation(
reinterpret_cast<uintptr_t>(alloc_info.deviceMemory),
static_cast<u64>(alloc_info.size),
property_flags
);
}
u8 *data = reinterpret_cast<u8 *>(alloc_info.pMappedData);
const std::span<u8> mapped_data = data ? std::span<u8>{data, ci.size} : std::span<u8>{};
const bool is_coherent = (property_flags & VK_MEMORY_PROPERTY_HOST_COHERENT_BIT) != 0;
return vk::Buffer(handle, *device.GetLogical(), allocator, allocation, mapped_data,
is_coherent,
device.GetDispatchLoader());
}
return vk::Image(handle, ci.usage, *device.GetLogical(), allocator, allocation,
device.GetDispatchLoader());
}
MemoryCommit MemoryAllocator::Commit(const VkMemoryRequirements &reqs, MemoryUsage usage)
{
const auto vma_usage = MemoryUsageVma(usage);
VmaAllocationCreateInfo ci{};
ci.flags = VMA_ALLOCATION_CREATE_WITHIN_BUDGET_BIT | MemoryUsageVmaFlags(usage);
ci.usage = vma_usage;
ci.memoryTypeBits = reqs.memoryTypeBits & valid_memory_types;
ci.requiredFlags = 0;
ci.preferredFlags = MemoryUsagePreferredVmaFlags(usage);
vk::Buffer MemoryAllocator::CreateBuffer(const VkBufferCreateInfo &ci, MemoryUsage usage) const {
// MESA will do memcpy() if not marked as host cached, so just force mark it for most buffers
auto const anv_flags = (usage == MemoryUsage::Stream
&& device.GetDriverID() == VK_DRIVER_ID_INTEL_OPEN_SOURCE_MESA)
? VK_MEMORY_PROPERTY_HOST_CACHED_BIT : 0;
const VmaAllocationCreateInfo alloc_ci = {
.flags = VMA_ALLOCATION_CREATE_WITHIN_BUDGET_BIT | MemoryUsageVmaFlags(usage),
.usage = MemoryUsageVma(usage),
.requiredFlags = 0,
.preferredFlags = MemoryUsagePreferredVmaFlags(usage) | anv_flags,
.memoryTypeBits = usage == MemoryUsage::Stream ? 0u : valid_memory_types,
.pool = VK_NULL_HANDLE,
.pUserData = nullptr,
.priority = 0.f,
};
VmaAllocation a{};
VmaAllocationInfo info{};
VkBuffer handle{};
VmaAllocationInfo alloc_info{};
VmaAllocation allocation{};
VkMemoryPropertyFlags property_flags{};
VkResult res = vmaAllocateMemory(allocator, &reqs, &ci, &a, &info);
vk::Check(vmaCreateBuffer(allocator, &ci, &alloc_ci, &handle, &allocation, &alloc_info));
vmaGetAllocationMemoryProperties(allocator, allocation, &property_flags);
if (res != VK_SUCCESS) {
// Relax 1: drop budget constraint
auto ci2 = ci;
ci2.flags &= ~VMA_ALLOCATION_CREATE_WITHIN_BUDGET_BIT;
res = vmaAllocateMemory(allocator, &reqs, &ci2, &a, &info);
// Log GPU memory allocation for buffers
if (GPU::Logging::IsActive() &&
Settings::values.gpu_log_memory_tracking.GetValue()) {
GPU::Logging::GPULogger::GetInstance().LogMemoryAllocation(
reinterpret_cast<uintptr_t>(alloc_info.deviceMemory),
static_cast<u64>(alloc_info.size),
property_flags
);
}
// Relax 2: if we preferred DEVICE_LOCAL, drop that preference
if (res != VK_SUCCESS && (ci.preferredFlags & VK_MEMORY_PROPERTY_DEVICE_LOCAL_BIT)) {
auto ci3 = ci2;
ci3.preferredFlags &= ~VK_MEMORY_PROPERTY_DEVICE_LOCAL_BIT;
res = vmaAllocateMemory(allocator, &reqs, &ci3, &a, &info);
}
}
u8 *data = reinterpret_cast<u8 *>(alloc_info.pMappedData);
const std::span<u8> mapped_data = data ? std::span<u8>{data, ci.size} : std::span<u8>{};
const bool is_coherent = (property_flags & VK_MEMORY_PROPERTY_HOST_COHERENT_BIT) != 0;
return vk::Buffer(handle, *device.GetLogical(), allocator, allocation, mapped_data,
is_coherent,
device.GetDispatchLoader());
}
vk::Buffer MemoryAllocator::CreateBuffer(const VkBufferCreateInfo &ci, MemoryUsage usage,
VkDeviceSize min_alignment) const {
if (min_alignment <= 1) {
return CreateBuffer(ci, usage);
}
VkMemoryPropertyFlags anv_flags = 0;
if (usage == MemoryUsage::Stream &&
device.GetDriverID() == VK_DRIVER_ID_INTEL_OPEN_SOURCE_MESA) {
anv_flags = VK_MEMORY_PROPERTY_HOST_CACHED_BIT;
}
u32 memory_type_bits = valid_memory_types;
if (usage == MemoryUsage::Stream) {
memory_type_bits = 0u;
}
const VmaAllocationCreateInfo alloc_ci = {
.flags = VMA_ALLOCATION_CREATE_WITHIN_BUDGET_BIT | MemoryUsageVmaFlags(usage),
.usage = MemoryUsageVma(usage),
.requiredFlags = 0,
.preferredFlags = MemoryUsagePreferredVmaFlags(usage) | anv_flags,
.memoryTypeBits = memory_type_bits,
.pool = VK_NULL_HANDLE,
.pUserData = nullptr,
.priority = 0.f,
};
const VkDevice logical = *device.GetLogical();
const auto &dld = device.GetDispatchLoader();
VkBuffer handle{};
vk::Check(dld.vkCreateBuffer(logical, &ci, nullptr, &handle));
const VkBufferMemoryRequirementsInfo2 reqs_info{
.sType = VK_STRUCTURE_TYPE_BUFFER_MEMORY_REQUIREMENTS_INFO_2,
.pNext = nullptr,
.buffer = handle,
};
VkMemoryRequirements2 reqs2{
.sType = VK_STRUCTURE_TYPE_MEMORY_REQUIREMENTS_2,
.pNext = nullptr,
.memoryRequirements = {},
};
dld.vkGetBufferMemoryRequirements2(logical, &reqs_info, &reqs2);
VkMemoryRequirements reqs = reqs2.memoryRequirements;
reqs.alignment = (std::max)(reqs.alignment, min_alignment);
reqs.memoryTypeBits &= alloc_ci.memoryTypeBits;
VmaAllocation allocation{};
VmaAllocationInfo alloc_info{};
VkResult res = vmaAllocateMemory(allocator, &reqs, &alloc_ci, &allocation, &alloc_info);
if (res != VK_SUCCESS) {
auto relaxed = alloc_ci;
relaxed.flags &= ~VMA_ALLOCATION_CREATE_WITHIN_BUDGET_BIT;
res = vmaAllocateMemory(allocator, &reqs, &relaxed, &allocation, &alloc_info);
}
if (res != VK_SUCCESS) {
dld.vkDestroyBuffer(logical, handle, nullptr);
vk::Check(res);
}
const VkResult bind_res = vmaBindBufferMemory(allocator, allocation, handle);
if (bind_res != VK_SUCCESS) {
vmaFreeMemory(allocator, allocation);
dld.vkDestroyBuffer(logical, handle, nullptr);
vk::Check(bind_res);
return MemoryCommit(allocator, a, info);
}
VkMemoryPropertyFlags property_flags{};
vmaGetAllocationMemoryProperties(allocator, allocation, &property_flags);
MemoryCommit MemoryAllocator::Commit(const vk::Buffer &buffer, MemoryUsage usage) {
// Allocate memory appropriate for this buffer automatically
const auto vma_usage = MemoryUsageVma(usage);
u8 *data = reinterpret_cast<u8 *>(alloc_info.pMappedData);
std::span<u8> mapped_data{};
if (data) {
mapped_data = std::span<u8>{data, ci.size};
}
const bool is_coherent = (property_flags & VK_MEMORY_PROPERTY_HOST_COHERENT_BIT) != 0;
VmaAllocationCreateInfo ci{};
ci.flags = VMA_ALLOCATION_CREATE_WITHIN_BUDGET_BIT | MemoryUsageVmaFlags(usage);
ci.usage = vma_usage;
ci.requiredFlags = 0;
ci.preferredFlags = MemoryUsagePreferredVmaFlags(usage);
ci.pool = VK_NULL_HANDLE;
ci.pUserData = nullptr;
ci.priority = 0.0f;
return vk::Buffer(handle, logical, allocator, allocation, mapped_data, is_coherent, dld);
}
const VkBuffer raw = *buffer;
MemoryCommit MemoryAllocator::Commit(const VkMemoryRequirements &reqs, MemoryUsage usage)
{
const auto vma_usage = MemoryUsageVma(usage);
VmaAllocationCreateInfo ci{};
ci.flags = VMA_ALLOCATION_CREATE_WITHIN_BUDGET_BIT | MemoryUsageVmaFlags(usage);
ci.usage = vma_usage;
ci.memoryTypeBits = reqs.memoryTypeBits & valid_memory_types;
ci.requiredFlags = 0;
ci.preferredFlags = MemoryUsagePreferredVmaFlags(usage);
VmaAllocation a{};
VmaAllocationInfo info{};
VmaAllocation a{};
VmaAllocationInfo info{};
// Let VMA infer memory requirements from the buffer
VkResult res = vmaAllocateMemoryForBuffer(allocator, raw, &ci, &a, &info);
VkResult res = vmaAllocateMemory(allocator, &reqs, &ci, &a, &info);
if (res != VK_SUCCESS) {
auto ci2 = ci;
ci2.flags &= ~VMA_ALLOCATION_CREATE_WITHIN_BUDGET_BIT;
res = vmaAllocateMemoryForBuffer(allocator, raw, &ci2, &a, &info);
if (res != VK_SUCCESS) {
// Relax 1: drop budget constraint
auto ci2 = ci;
ci2.flags &= ~VMA_ALLOCATION_CREATE_WITHIN_BUDGET_BIT;
res = vmaAllocateMemory(allocator, &reqs, &ci2, &a, &info);
// Relax 2: if we preferred DEVICE_LOCAL, drop that preference
if (res != VK_SUCCESS && (ci.preferredFlags & VK_MEMORY_PROPERTY_DEVICE_LOCAL_BIT)) {
auto ci3 = ci2;
ci3.preferredFlags &= ~VK_MEMORY_PROPERTY_DEVICE_LOCAL_BIT;
res = vmaAllocateMemory(allocator, &reqs, &ci3, &a, &info);
if (res != VK_SUCCESS && (ci.preferredFlags & VK_MEMORY_PROPERTY_DEVICE_LOCAL_BIT)) {
auto ci3 = ci2;
ci3.preferredFlags &= ~VK_MEMORY_PROPERTY_DEVICE_LOCAL_BIT;
res = vmaAllocateMemoryForBuffer(allocator, raw, &ci3, &a, &info);
}
}
vk::Check(res);
vk::Check(vmaBindBufferMemory2(allocator, a, 0, raw, nullptr));
return MemoryCommit(allocator, a, info);
}
vk::Check(res);
return MemoryCommit(allocator, a, info);
}
MemoryCommit MemoryAllocator::Commit(const vk::Buffer &buffer, MemoryUsage usage) {
// Allocate memory appropriate for this buffer automatically
const auto vma_usage = MemoryUsageVma(usage);
VmaAllocationCreateInfo ci{};
ci.flags = VMA_ALLOCATION_CREATE_WITHIN_BUDGET_BIT | MemoryUsageVmaFlags(usage);
ci.usage = vma_usage;
ci.requiredFlags = 0;
ci.preferredFlags = MemoryUsagePreferredVmaFlags(usage);
ci.pool = VK_NULL_HANDLE;
ci.pUserData = nullptr;
ci.priority = 0.0f;
const VkBuffer raw = *buffer;
VmaAllocation a{};
VmaAllocationInfo info{};
// Let VMA infer memory requirements from the buffer
VkResult res = vmaAllocateMemoryForBuffer(allocator, raw, &ci, &a, &info);
if (res != VK_SUCCESS) {
auto ci2 = ci;
ci2.flags &= ~VMA_ALLOCATION_CREATE_WITHIN_BUDGET_BIT;
res = vmaAllocateMemoryForBuffer(allocator, raw, &ci2, &a, &info);
if (res != VK_SUCCESS && (ci.preferredFlags & VK_MEMORY_PROPERTY_DEVICE_LOCAL_BIT)) {
auto ci3 = ci2;
ci3.preferredFlags &= ~VK_MEMORY_PROPERTY_DEVICE_LOCAL_BIT;
res = vmaAllocateMemoryForBuffer(allocator, raw, &ci3, &a, &info);
}
}
vk::Check(res);
vk::Check(vmaBindBufferMemory2(allocator, a, 0, raw, nullptr));
return MemoryCommit(allocator, a, info);
}
} // namespace Vulkan
@@ -107,9 +107,6 @@ namespace Vulkan {
vk::Buffer CreateBuffer(const VkBufferCreateInfo &ci, MemoryUsage usage) const;
vk::Buffer CreateBuffer(const VkBufferCreateInfo &ci, MemoryUsage usage,
VkDeviceSize min_alignment) const;
/**
* Commits a memory with the specified requirements.
*
@@ -229,7 +229,6 @@ void Load(VkDevice device, DeviceDispatch& dld) noexcept {
X(vkGetPipelineExecutableStatisticsKHR);
X(vkGetSemaphoreCounterValue);
X(vkMapMemory);
X(vkQueueBindSparse);
X(vkQueueSubmit);
X(vkQueueSubmit2);
X(vkResetFences);
@@ -540,18 +539,6 @@ void Buffer::SetObjectNameEXT(const char* name) const {
SetObjectName(dld, owner, handle, VK_OBJECT_TYPE_BUFFER, name);
}
MemoryLocation Buffer::Location() const noexcept {
if (!allocation) {
return MemoryLocation{};
}
VmaAllocationInfo info{};
vmaGetAllocationInfo(allocator, allocation, &info);
return MemoryLocation{
.memory = info.deviceMemory,
.offset = info.offset,
};
}
void Buffer::Release() const noexcept {
if (handle) {
vmaDestroyBuffer(allocator, handle, allocation);
@@ -345,7 +345,6 @@ struct DeviceDispatch : InstanceDispatch {
PFN_vkGetQueryPoolResults vkGetQueryPoolResults{};
PFN_vkGetSemaphoreCounterValue vkGetSemaphoreCounterValue{};
PFN_vkMapMemory vkMapMemory{};
PFN_vkQueueBindSparse vkQueueBindSparse{};
PFN_vkQueueSubmit vkQueueSubmit{};
PFN_vkQueueSubmit2 vkQueueSubmit2{};
PFN_vkResetFences vkResetFences{};
@@ -741,11 +740,6 @@ private:
const DeviceDispatch* dld = nullptr;
};
struct MemoryLocation {
VkDeviceMemory memory{};
VkDeviceSize offset{};
};
class Buffer {
public:
explicit Buffer(VkBuffer handle_, VkDevice owner_, VmaAllocator allocator_,
@@ -817,8 +811,6 @@ public:
void SetObjectNameEXT(const char* name) const;
MemoryLocation Location() const noexcept;
private:
void Release() const noexcept;
@@ -851,11 +843,6 @@ public:
return dld->vkQueueSubmit2(queue, submit_infos.size(), submit_infos.data(), fence);
}
VkResult BindSparse(Span<VkBindSparseInfo> bind_infos,
VkFence fence = VK_NULL_HANDLE) const noexcept {
return dld->vkQueueBindSparse(queue, bind_infos.size(), bind_infos.data(), fence);
}
VkResult Present(const VkPresentInfoKHR& present_info) const noexcept {
return dld->vkQueuePresentKHR(queue, &present_info);
}
+1 -1
View File
@@ -4796,6 +4796,6 @@ void VolumeButton::ResetMultiplier() {
#endif
#if !defined(QT_STATICPLUGIN) || defined(__APPLE__)
#define VMA_IMPLEMENTATION 1
#define VMA_IMPLEMENTATION
#include "video_core/vulkan_common/vma.h"
#endif
+1 -1
View File
@@ -65,7 +65,7 @@ else()
endif()
# update cached cpmfile content
string(JSON cpmfile SET "${cpmfile}" "${KEY}" "${new_object}")
string(JSON cpmfile SET "${cpmfile}" "${key}" "${new_object}")
# write cached cpmfile
get_cpmfile_path(file)