shader/arithmetic_integer: Fix tracking issue in temporary

This temporary is not needed as we mark Rd.CC + IADD.X as unimplemented.
It caused issues when tracking global buffers.

.ci/templates/build-msvc.yml

@@ -4,7 +4,7 @@ parameters:
   version: ''
 
 steps:
-- script: mkdir build && cd build && cmake -G "Visual Studio 15 2017 Win64" --config Release -DYUZU_USE_BUNDLED_QT=1 -DYUZU_USE_BUNDLED_SDL2=1 -DYUZU_USE_BUNDLED_UNICORN=1 -DYUZU_USE_QT_WEB_ENGINE=ON -DENABLE_COMPATIBILITY_LIST_DOWNLOAD=ON -DYUZU_ENABLE_COMPATIBILITY_REPORTING=${COMPAT} -DUSE_DISCORD_PRESENCE=ON -DDISPLAY_VERSION=${{ parameters['version'] }} .. && cd ..
+- script: mkdir build && cd build && cmake -G "Visual Studio 16 2019" -A x64 --config Release -DYUZU_USE_BUNDLED_QT=1 -DYUZU_USE_BUNDLED_SDL2=1 -DYUZU_USE_BUNDLED_UNICORN=1 -DYUZU_USE_QT_WEB_ENGINE=ON -DENABLE_COMPATIBILITY_LIST_DOWNLOAD=ON -DYUZU_ENABLE_COMPATIBILITY_REPORTING=${COMPAT} -DUSE_DISCORD_PRESENCE=ON -DDISPLAY_VERSION=${{ parameters['version'] }} .. && cd ..
   displayName: 'Configure CMake'
 - task: MSBuild@1
   displayName: 'Build'

.ci/yuzu-mainline-step2.yml

@@ -45,7 +45,7 @@ stages:
   - job: build
     displayName: 'msvc'
     pool:
-      vmImage: vs2017-win2016
+      vmImage: windows-2019
     steps:
     - template: ./templates/sync-source.yml
       parameters:

.ci/yuzu-patreon-step2.yml

@@ -22,7 +22,7 @@ stages:
   - job: build
     displayName: 'windows-msvc'
     pool:
-      vmImage: vs2017-win2016
+      vmImage: windows-2019
     steps:
     - template: ./templates/sync-source.yml
       parameters:

CMakeLists.txt

@@ -164,7 +164,7 @@ if (ENABLE_SDL2)
             set(SDL2_LIBRARIES "SDL2::SDL2")
         endif()
 
-        include_directories(${SDL2_INCLUDE_DIRS})
+        include_directories(SYSTEM ${SDL2_INCLUDE_DIRS})
         add_library(SDL2 INTERFACE)
         target_link_libraries(SDL2 INTERFACE "${SDL2_LIBRARIES}")
     endif()

CMakeModules/CopyYuzuQt5Deps.cmake

@@ -6,9 +6,9 @@ function(copy_yuzu_Qt5_deps target_dir)
     set(Qt5_STYLES_DIR "${Qt5_DIR}/../../../plugins/styles/")
     set(Qt5_IMAGEFORMATS_DIR "${Qt5_DIR}/../../../plugins/imageformats/")
     set(Qt5_RESOURCES_DIR "${Qt5_DIR}/../../../resources/")
-    set(PLATFORMS ${DLL_DEST}platforms/)
-    set(STYLES ${DLL_DEST}styles/)
-    set(IMAGEFORMATS ${DLL_DEST}imageformats/)
+    set(PLATFORMS ${DLL_DEST}plugins/platforms/)
+    set(STYLES ${DLL_DEST}plugins/styles/)
+    set(IMAGEFORMATS ${DLL_DEST}plugins/imageformats/)
     windows_copy_files(${target_dir} ${Qt5_DLL_DIR} ${DLL_DEST}
         icudt*.dll
         icuin*.dll
@@ -42,11 +42,15 @@ function(copy_yuzu_Qt5_deps target_dir)
             icudtl.dat
         )
     endif ()
-
     windows_copy_files(yuzu ${Qt5_PLATFORMS_DIR} ${PLATFORMS} qwindows$<$<CONFIG:Debug>:d>.*)
     windows_copy_files(yuzu ${Qt5_STYLES_DIR} ${STYLES} qwindowsvistastyle$<$<CONFIG:Debug>:d>.*)
     windows_copy_files(yuzu ${Qt5_IMAGEFORMATS_DIR} ${IMAGEFORMATS}
         qjpeg$<$<CONFIG:Debug>:d>.*
         qgif$<$<CONFIG:Debug>:d>.*
         )
+    # Create an empty qt.conf file. Qt will detect that this file exists, and use the folder that its in as the root folder.
+    # This way it'll look for plugins in the root/plugins/ folder
+    add_custom_command(TARGET yuzu POST_BUILD
+        COMMAND ${CMAKE_COMMAND} -E touch ${DLL_DEST}qt.conf
+    )
 endfunction(copy_yuzu_Qt5_deps)

CMakeModules/GenerateSCMRev.cmake

@@ -57,8 +57,6 @@ set(HASH_FILES
     "${VIDEO_CORE}/renderer_opengl/gl_shader_decompiler.h"
     "${VIDEO_CORE}/renderer_opengl/gl_shader_disk_cache.cpp"
     "${VIDEO_CORE}/renderer_opengl/gl_shader_disk_cache.h"
-    "${VIDEO_CORE}/renderer_opengl/gl_shader_gen.cpp"
-    "${VIDEO_CORE}/renderer_opengl/gl_shader_gen.h"
     "${VIDEO_CORE}/shader/decode/arithmetic.cpp"
     "${VIDEO_CORE}/shader/decode/arithmetic_half.cpp"
     "${VIDEO_CORE}/shader/decode/arithmetic_half_immediate.cpp"
@@ -91,8 +89,6 @@ set(HASH_FILES
     "${VIDEO_CORE}/shader/ast.h"
     "${VIDEO_CORE}/shader/compiler_settings.cpp"
     "${VIDEO_CORE}/shader/compiler_settings.h"
-    "${VIDEO_CORE}/shader/const_buffer_locker.cpp"
-    "${VIDEO_CORE}/shader/const_buffer_locker.h"
     "${VIDEO_CORE}/shader/control_flow.cpp"
     "${VIDEO_CORE}/shader/control_flow.h"
     "${VIDEO_CORE}/shader/decode.cpp"
@@ -101,9 +97,13 @@ set(HASH_FILES
     "${VIDEO_CORE}/shader/node.h"
     "${VIDEO_CORE}/shader/node_helper.cpp"
     "${VIDEO_CORE}/shader/node_helper.h"
+    "${VIDEO_CORE}/shader/registry.cpp"
+    "${VIDEO_CORE}/shader/registry.h"
     "${VIDEO_CORE}/shader/shader_ir.cpp"
     "${VIDEO_CORE}/shader/shader_ir.h"
     "${VIDEO_CORE}/shader/track.cpp"
+    "${VIDEO_CORE}/shader/transform_feedback.cpp"
+    "${VIDEO_CORE}/shader/transform_feedback.h"
 )
 set(COMBINED "")
 foreach (F IN LISTS HASH_FILES)

README.md

@@ -1,7 +1,8 @@
 yuzu emulator
 =============
-[![Travis CI Build Status](https://travis-ci.org/yuzu-emu/yuzu.svg?branch=master)](https://travis-ci.org/yuzu-emu/yuzu)
+[![Travis CI Build Status](https://travis-ci.com/yuzu-emu/yuzu.svg?branch=master)](https://travis-ci.com/yuzu-emu/yuzu)
 [![Azure Mainline CI Build Status](https://dev.azure.com/yuzu-emu/yuzu/_apis/build/status/yuzu%20mainline?branchName=master)](https://dev.azure.com/yuzu-emu/yuzu/)
+[![Discord](https://img.shields.io/discord/398318088170242053?color=%237289DA&label=yuzu&logo=discord&logoColor=white)](https://discord.gg/XQV6dn9)
 
 yuzu is an experimental open-source emulator for the Nintendo Switch from the creators of [Citra](https://citra-emu.org/).
 
@@ -21,7 +22,7 @@ For development discussion, please join us on [Discord](https://discord.gg/XQV6d
 
 Most of the development happens on GitHub. It's also where [our central repository](https://github.com/yuzu-emu/yuzu) is hosted.
 
-If you want to contribute please take a look at the [Contributor's Guide](CONTRIBUTING.md) and [Developer Information](https://github.com/yuzu-emu/yuzu/wiki/Developer-Information). You should as well contact any of the developers on Discord in order to know about the current state of the emulator.
+If you want to contribute please take a look at the [Contributor's Guide](https://github.com/yuzu-emu/yuzu/wiki/Contributing) and [Developer Information](https://github.com/yuzu-emu/yuzu/wiki/Developer-Information). You should also contact any of the developers on Discord in order to know about the current state of the emulator.
 
 ### Building
 

dist/license.md

@@ -2,8 +2,8 @@ The icons in this folder and its subfolders have the following licenses:
 
 Icon Name | License | Origin/Author
 --- | --- | ---
-qt_themes/default/icons/16x16/checked.png | Free for non-commercial use
-qt_themes/default/icons/16x16/failed.png | Free for non-commercial use
+qt_themes/default/icons/16x16/checked.png | CC BY-ND 3.0 | https://icons8.com
+qt_themes/default/icons/16x16/failed.png | CC BY-ND 3.0 | https://icons8.com
 qt_themes/default/icons/16x16/lock.png | CC BY-ND 3.0 | https://icons8.com
 qt_themes/default/icons/256x256/plus_folder.png | CC BY-ND 3.0 | https://icons8.com
 qt_themes/default/icons/48x48/bad_folder.png | CC BY-ND 3.0 | https://icons8.com
@@ -11,8 +11,6 @@ qt_themes/default/icons/48x48/chip.png | CC BY-ND 3.0 | https://icons8.com
 qt_themes/default/icons/48x48/folder.png | CC BY-ND 3.0 | https://icons8.com
 qt_themes/default/icons/48x48/plus.png | CC0 1.0 | Designed by BreadFish64 from the Citra team
 qt_themes/default/icons/48x48/sd_card.png | CC BY-ND 3.0 | https://icons8.com
-qt_themes/qdarkstyle/icons/16x16/checked.png | Free for non-commercial use
-qt_themes/qdarkstyle/icons/16x16/failed.png | Free for non-commercial use
 qt_themes/qdarkstyle/icons/16x16/lock.png | CC BY-ND 3.0 | https://icons8.com
 qt_themes/qdarkstyle/icons/256x256/plus_folder.png | CC BY-ND 3.0 | https://icons8.com
 qt_themes/qdarkstyle/icons/48x48/bad_folder.png | CC BY-ND 3.0 | https://icons8.com

externals/CMakeLists.txt

@@ -3,13 +3,27 @@
 set(CMAKE_MODULE_PATH ${CMAKE_MODULE_PATH} ${PROJECT_SOURCE_DIR}/CMakeModules)
 include(DownloadExternals)
 
+# pkgconfig -- it is used to find shared libraries without cmake modules on linux systems
+find_package(PkgConfig)
+if (NOT PkgConfig_FOUND)
+    function(pkg_check_modules)
+        # STUB
+    endfunction()
+endif()
+
 # Catch
 add_library(catch-single-include INTERFACE)
 target_include_directories(catch-single-include INTERFACE catch/single_include)
 
 # libfmt
-add_subdirectory(fmt)
-add_library(fmt::fmt ALIAS fmt)
+pkg_check_modules(FMT IMPORTED_TARGET GLOBAL fmt>=6.2.0)
+if (FMT_FOUND)
+    add_library(fmt::fmt ALIAS PkgConfig::FMT)
+else()
+    message(STATUS "fmt 6.2.0 or newer not found, falling back to externals")
+    add_subdirectory(fmt)
+    add_library(fmt::fmt ALIAS fmt)
+endif()
 
 # Dynarmic
 if (ARCHITECTURE_x86_64)
@@ -30,9 +44,15 @@ add_subdirectory(glad)
 add_subdirectory(inih)
 
 # lz4
-set(LZ4_BUNDLED_MODE ON)
-add_subdirectory(lz4/contrib/cmake_unofficial EXCLUDE_FROM_ALL)
-target_include_directories(lz4_static INTERFACE ./lz4/lib)
+pkg_check_modules(LIBLZ4 IMPORTED_TARGET GLOBAL liblz4>=1.8.0)
+if (LIBLZ4_FOUND)
+    add_library(lz4_static ALIAS PkgConfig::LIBLZ4)
+else()
+    message(STATUS "liblz4 1.8.0 or newer not found, falling back to externals")
+    set(LZ4_BUNDLED_MODE ON)
+    add_subdirectory(lz4/contrib/cmake_unofficial EXCLUDE_FROM_ALL)
+    target_include_directories(lz4_static INTERFACE ./lz4/lib)
+endif()
 
 # mbedtls
 add_subdirectory(mbedtls EXCLUDE_FROM_ALL)
@@ -47,15 +67,27 @@ add_library(unicorn-headers INTERFACE)
 target_include_directories(unicorn-headers INTERFACE ./unicorn/include)
 
 # Zstandard
-add_subdirectory(zstd/build/cmake EXCLUDE_FROM_ALL)
-target_include_directories(libzstd_static INTERFACE ./zstd/lib)
+pkg_check_modules(LIBZSTD IMPORTED_TARGET GLOBAL libzstd>=1.3.8)
+if (LIBZSTD_FOUND)
+    add_library(libzstd_static ALIAS PkgConfig::LIBZSTD)
+else()
+    message(STATUS "libzstd 1.3.8 or newer not found, falling back to externals")
+    add_subdirectory(zstd/build/cmake EXCLUDE_FROM_ALL)
+    target_include_directories(libzstd_static INTERFACE ./zstd/lib)
+endif()
 
 # SoundTouch
 add_subdirectory(soundtouch)
 
 # Opus
-add_subdirectory(opus)
-target_include_directories(opus INTERFACE ./opus/include)
+pkg_check_modules(OPUS IMPORTED_TARGET GLOBAL opus>=1.3.1)
+if (OPUS_FOUND)
+    add_library(opus ALIAS PkgConfig::OPUS)
+else()
+    message(STATUS "opus 1.3.1 or newer not found, falling back to externals")
+    add_subdirectory(opus)
+    target_include_directories(opus INTERFACE ./opus/include)
+endif()
 
 # Cubeb
 if(ENABLE_CUBEB)
@@ -75,18 +107,35 @@ if (ENABLE_VULKAN)
 endif()
 
 # zlib
-add_subdirectory(zlib EXCLUDE_FROM_ALL)
-set(ZLIB_LIBRARIES z)
+find_package(ZLIB 1.2.11)
+if (NOT ZLIB_FOUND)
+    message(STATUS "zlib 1.2.11 or newer not found, falling back to externals")
+    add_subdirectory(zlib EXCLUDE_FROM_ALL)
+    set(ZLIB_LIBRARIES z)
+endif()
 
 # libzip
-add_subdirectory(libzip EXCLUDE_FROM_ALL)
+pkg_check_modules(LIBZIP IMPORTED_TARGET GLOBAL libzip>=1.5.3)
+if (LIBZIP_FOUND)
+    add_library(zip ALIAS PkgConfig::LIBZIP)
+else()
+    message(STATUS "libzip 1.5.3 or newer not found, falling back to externals")
+    add_subdirectory(libzip EXCLUDE_FROM_ALL)
+endif()
 
 if (ENABLE_WEB_SERVICE)
     # LibreSSL
-    set(LIBRESSL_SKIP_INSTALL ON CACHE BOOL "")
-    add_subdirectory(libressl EXCLUDE_FROM_ALL)
-    target_include_directories(ssl INTERFACE ./libressl/include)
-	target_compile_definitions(ssl PRIVATE -DHAVE_INET_NTOP)
+    find_package(OpenSSL COMPONENTS Crypto SSL)
+    if (NOT OpenSSL_FOUND)
+        message(STATUS "OpenSSL not found, falling back to externals")
+        set(LIBRESSL_SKIP_INSTALL ON CACHE BOOL "")
+        add_subdirectory(libressl EXCLUDE_FROM_ALL)
+        target_include_directories(ssl INTERFACE ./libressl/include)
+        target_compile_definitions(ssl PRIVATE -DHAVE_INET_NTOP)
+        get_directory_property(OPENSSL_LIBRARIES
+            DIRECTORY libressl
+            DEFINITION OPENSSL_LIBS)
+    endif()
 
     # lurlparser
     add_subdirectory(lurlparser EXCLUDE_FROM_ALL)
@@ -94,6 +143,8 @@ if (ENABLE_WEB_SERVICE)
     # httplib
     add_library(httplib INTERFACE)
     target_include_directories(httplib INTERFACE ./httplib)
+    target_compile_definitions(httplib INTERFACE -DCPPHTTPLIB_OPENSSL_SUPPORT)
+    target_link_libraries(httplib INTERFACE ${OPENSSL_LIBRARIES})
 
     # JSON
     add_library(json-headers INTERFACE)

externals/microprofile/microprofile.h

@@ -243,6 +243,7 @@ typedef uint32_t ThreadIdType;
 #define MICROPROFILE_DEFINE_GPU(var, name, color) MicroProfileToken g_mp_##var = MicroProfileGetToken("GPU", name, color, MicroProfileTokenTypeGpu)
 #define MICROPROFILE_TOKEN_PASTE0(a, b) a ## b
 #define MICROPROFILE_TOKEN_PASTE(a, b)  MICROPROFILE_TOKEN_PASTE0(a,b)
+#define MICROPROFILE_TOKEN(var) g_mp_##var
 #define MICROPROFILE_SCOPE(var) MicroProfileScopeHandler MICROPROFILE_TOKEN_PASTE(foo, __LINE__)(g_mp_##var)
 #define MICROPROFILE_SCOPE_TOKEN(token) MicroProfileScopeHandler MICROPROFILE_TOKEN_PASTE(foo, __LINE__)(token)
 #define MICROPROFILE_SCOPEI(group, name, color) static MicroProfileToken MICROPROFILE_TOKEN_PASTE(g_mp,__LINE__) = MicroProfileGetToken(group, name, color, MicroProfileTokenTypeCpu); MicroProfileScopeHandler MICROPROFILE_TOKEN_PASTE(foo,__LINE__)( MICROPROFILE_TOKEN_PASTE(g_mp,__LINE__))
@@ -827,7 +828,7 @@ inline MicroProfileLogEntry MicroProfileMakeLogIndex(uint64_t nBegin, MicroProfi
     MicroProfileLogEntry Entry =  (nBegin<<62) | ((0x3fff&nToken)<<48) | (MP_LOG_TICK_MASK&nTick);
     int t = MicroProfileLogType(Entry);
     uint64_t nTimerIndex = MicroProfileLogTimerIndex(Entry);
-    MP_ASSERT(t == nBegin);
+    MP_ASSERT((uint64_t)t == nBegin);
     MP_ASSERT(nTimerIndex == (nToken&0x3fff));
     return Entry;
 
@@ -909,14 +910,14 @@ typedef void* (*MicroProfileThreadFunc)(void*);
 
 #ifndef _WIN32
 typedef pthread_t MicroProfileThread;
-void MicroProfileThreadStart(MicroProfileThread* pThread, MicroProfileThreadFunc Func)
+inline void MicroProfileThreadStart(MicroProfileThread* pThread, MicroProfileThreadFunc Func)
 {
     pthread_attr_t Attr;
     int r  = pthread_attr_init(&Attr);
     MP_ASSERT(r == 0);
     pthread_create(pThread, &Attr, Func, 0);
 }
-void MicroProfileThreadJoin(MicroProfileThread* pThread)
+inline void MicroProfileThreadJoin(MicroProfileThread* pThread)
 {
     int r = pthread_join(*pThread, 0);
     MP_ASSERT(r == 0);
@@ -929,11 +930,11 @@ DWORD _stdcall ThreadTrampoline(void* pFunc)
     return (uint32_t)F(0);
 }
 
-void MicroProfileThreadStart(MicroProfileThread* pThread, MicroProfileThreadFunc Func)
+inline void MicroProfileThreadStart(MicroProfileThread* pThread, MicroProfileThreadFunc Func)
 {
     *pThread = CreateThread(0, 0, ThreadTrampoline, Func, 0, 0);
 }
-void MicroProfileThreadJoin(MicroProfileThread* pThread)
+inline void MicroProfileThreadJoin(MicroProfileThread* pThread)
 {
     WaitForSingleObject(*pThread, INFINITE);
     CloseHandle(*pThread);
@@ -1130,7 +1131,7 @@ inline void MicroProfileSetThreadLog(MicroProfileThreadLog* pLog)
     pthread_setspecific(g_MicroProfileThreadLogKey, pLog);
 }
 #else
-MicroProfileThreadLog* MicroProfileGetThreadLog()
+inline MicroProfileThreadLog* MicroProfileGetThreadLog()
 {
     return g_MicroProfileThreadLog;
 }
@@ -1246,7 +1247,7 @@ MicroProfileToken MicroProfileFindToken(const char* pGroup, const char* pName)
     return MICROPROFILE_INVALID_TOKEN;
 }
 
-uint16_t MicroProfileGetGroup(const char* pGroup, MicroProfileTokenType Type)
+inline uint16_t MicroProfileGetGroup(const char* pGroup, MicroProfileTokenType Type)
 {
     for(uint32_t i = 0; i < S.nGroupCount; ++i)
     {
@@ -1275,7 +1276,7 @@ uint16_t MicroProfileGetGroup(const char* pGroup, MicroProfileTokenType Type)
     return nGroupIndex;
 }
 
-void MicroProfileRegisterGroup(const char* pGroup, const char* pCategory, uint32_t nColor)
+inline void MicroProfileRegisterGroup(const char* pGroup, const char* pCategory, uint32_t nColor)
 {
     int nCategoryIndex = -1;
     for(uint32_t i = 0; i < S.nCategoryCount; ++i)
@@ -1441,7 +1442,7 @@ void MicroProfileGpuLeave(MicroProfileToken nToken_, uint64_t nTickStart)
     }
 }
 
-void MicroProfileContextSwitchPut(MicroProfileContextSwitch* pContextSwitch)
+inline void MicroProfileContextSwitchPut(MicroProfileContextSwitch* pContextSwitch)
 {
     if(S.nRunning || pContextSwitch->nTicks <= S.nPauseTicks)
     {
@@ -1555,10 +1556,10 @@ void MicroProfileFlip()
 
         pFramePut->nFrameStartCpu = MP_TICK();
         pFramePut->nFrameStartGpu = (uint32_t)MicroProfileGpuInsertTimeStamp();
-        if(pFrameNext->nFrameStartGpu != (uint64_t)-1)
+        if(pFrameNext->nFrameStartGpu != -1)
             pFrameNext->nFrameStartGpu = MicroProfileGpuGetTimeStamp((uint32_t)pFrameNext->nFrameStartGpu);
 
-        if(pFrameCurrent->nFrameStartGpu == (uint64_t)-1)
+        if(pFrameCurrent->nFrameStartGpu == -1)
             pFrameCurrent->nFrameStartGpu = pFrameNext->nFrameStartGpu + 1;
 
         uint64_t nFrameStartCpu = pFrameCurrent->nFrameStartCpu;
@@ -1893,7 +1894,7 @@ void MicroProfileSetEnableAllGroups(bool bEnableAllGroups)
     S.nAllGroupsWanted = bEnableAllGroups ? 1 : 0;
 }
 
-void MicroProfileEnableCategory(const char* pCategory, bool bEnabled)
+inline void MicroProfileEnableCategory(const char* pCategory, bool bEnabled)
 {
     int nCategoryIndex = -1;
     for(uint32_t i = 0; i < S.nCategoryCount; ++i)
@@ -2003,7 +2004,7 @@ void MicroProfileForceDisableGroup(const char* pGroup, MicroProfileTokenType Typ
 }
 
 
-void MicroProfileCalcAllTimers(float* pTimers, float* pAverage, float* pMax, float* pCallAverage, float* pExclusive, float* pAverageExclusive, float* pMaxExclusive, float* pTotal, uint32_t nSize)
+inline void MicroProfileCalcAllTimers(float* pTimers, float* pAverage, float* pMax, float* pCallAverage, float* pExclusive, float* pAverageExclusive, float* pMaxExclusive, float* pTotal, uint32_t nSize)
 {
     for(uint32_t i = 0; i < S.nTotalTimers && i < nSize; ++i)
     {

externals/microprofile/microprofileui.h

@@ -417,19 +417,19 @@ void MicroProfileToggleDisplayMode()
 }
 
 
-void MicroProfileStringArrayClear(MicroProfileStringArray* pArray)
+inline void MicroProfileStringArrayClear(MicroProfileStringArray* pArray)
 {
     pArray->nNumStrings = 0;
     pArray->pBufferPos = &pArray->Buffer[0];
 }
 
-void MicroProfileStringArrayAddLiteral(MicroProfileStringArray* pArray, const char* pLiteral)
+inline void MicroProfileStringArrayAddLiteral(MicroProfileStringArray* pArray, const char* pLiteral)
 {
     MP_ASSERT(pArray->nNumStrings < MICROPROFILE_TOOLTIP_MAX_STRINGS);
     pArray->ppStrings[pArray->nNumStrings++] = pLiteral;
 }
 
-void MicroProfileStringArrayFormat(MicroProfileStringArray* pArray, const char* fmt, ...)
+inline void MicroProfileStringArrayFormat(MicroProfileStringArray* pArray, const char* fmt, ...)
 {
     MP_ASSERT(pArray->nNumStrings < MICROPROFILE_TOOLTIP_MAX_STRINGS);
     pArray->ppStrings[pArray->nNumStrings++] = pArray->pBufferPos;
@@ -439,7 +439,7 @@ void MicroProfileStringArrayFormat(MicroProfileStringArray* pArray, const char*
     va_end(args);
     MP_ASSERT(pArray->pBufferPos < pArray->Buffer + MICROPROFILE_TOOLTIP_STRING_BUFFER_SIZE);
 }
-void MicroProfileStringArrayCopy(MicroProfileStringArray* pDest, MicroProfileStringArray* pSrc)
+inline void MicroProfileStringArrayCopy(MicroProfileStringArray* pDest, MicroProfileStringArray* pSrc)
 {
     memcpy(&pDest->ppStrings[0], &pSrc->ppStrings[0], sizeof(pDest->ppStrings));
     memcpy(&pDest->Buffer[0], &pSrc->Buffer[0], sizeof(pDest->Buffer));
@@ -456,7 +456,7 @@ void MicroProfileStringArrayCopy(MicroProfileStringArray* pDest, MicroProfileStr
     pDest->nNumStrings = pSrc->nNumStrings;
 }
 
-void MicroProfileFloatWindowSize(const char** ppStrings, uint32_t nNumStrings, uint32_t* pColors, uint32_t& nWidth, uint32_t& nHeight, uint32_t* pStringLengths = 0)
+inline void MicroProfileFloatWindowSize(const char** ppStrings, uint32_t nNumStrings, uint32_t* pColors, uint32_t& nWidth, uint32_t& nHeight, uint32_t* pStringLengths = 0)
 {
     uint32_t* nStringLengths = pStringLengths ? pStringLengths : (uint32_t*)alloca(nNumStrings * sizeof(uint32_t));
     uint32_t nTextCount = nNumStrings/2;
@@ -474,7 +474,7 @@ void MicroProfileFloatWindowSize(const char** ppStrings, uint32_t nNumStrings, u
     nHeight = (MICROPROFILE_TEXT_HEIGHT+1) * nTextCount + 2 * MICROPROFILE_BORDER_SIZE;
 }
 
-void MicroProfileDrawFloatWindow(uint32_t nX, uint32_t nY, const char** ppStrings, uint32_t nNumStrings, uint32_t nColor, uint32_t* pColors = 0)
+inline void MicroProfileDrawFloatWindow(uint32_t nX, uint32_t nY, const char** ppStrings, uint32_t nNumStrings, uint32_t nColor, uint32_t* pColors = 0)
 {
     uint32_t nWidth = 0, nHeight = 0;
     uint32_t* nStringLengths = (uint32_t*)alloca(nNumStrings * sizeof(uint32_t));
@@ -503,7 +503,7 @@ void MicroProfileDrawFloatWindow(uint32_t nX, uint32_t nY, const char** ppString
         nY += (MICROPROFILE_TEXT_HEIGHT+1);
     }
 }
-void MicroProfileDrawTextBox(uint32_t nX, uint32_t nY, const char** ppStrings, uint32_t nNumStrings, uint32_t nColor, uint32_t* pColors = 0)
+inline void MicroProfileDrawTextBox(uint32_t nX, uint32_t nY, const char** ppStrings, uint32_t nNumStrings, uint32_t nColor, uint32_t* pColors = 0)
 {
     uint32_t nWidth = 0, nHeight = 0;
     uint32_t* nStringLengths = (uint32_t*)alloca(nNumStrings * sizeof(uint32_t));
@@ -529,7 +529,7 @@ void MicroProfileDrawTextBox(uint32_t nX, uint32_t nY, const char** ppStrings, u
 
 
 
-void MicroProfileToolTipMeta(MicroProfileStringArray* pToolTip)
+inline void MicroProfileToolTipMeta(MicroProfileStringArray* pToolTip)
 {
     MicroProfile& S = *MicroProfileGet();
     if(UI.nRangeBeginIndex != UI.nRangeEndIndex && UI.pRangeLog)
@@ -608,7 +608,7 @@ void MicroProfileToolTipMeta(MicroProfileStringArray* pToolTip)
     }
 }
 
-void MicroProfileDrawFloatTooltip(uint32_t nX, uint32_t nY, uint32_t nToken, uint64_t nTime)
+inline void MicroProfileDrawFloatTooltip(uint32_t nX, uint32_t nY, uint32_t nToken, uint64_t nTime)
 {
     MicroProfile& S = *MicroProfileGet();
 
@@ -718,7 +718,7 @@ void MicroProfileDrawFloatTooltip(uint32_t nX, uint32_t nY, uint32_t nToken, uin
 }
 
 
-void MicroProfileZoomTo(int64_t nTickStart, int64_t nTickEnd)
+inline void MicroProfileZoomTo(int64_t nTickStart, int64_t nTickEnd)
 {
     MicroProfile& S = *MicroProfileGet();
 
@@ -728,7 +728,7 @@ void MicroProfileZoomTo(int64_t nTickStart, int64_t nTickEnd)
     UI.fDetailedRangeTarget = MicroProfileLogTickDifference(nTickStart, nTickEnd) * fToMs;
 }
 
-void MicroProfileCenter(int64_t nTickCenter)
+inline void MicroProfileCenter(int64_t nTickCenter)
 {
     MicroProfile& S = *MicroProfileGet();
     int64_t nStart = S.Frames[S.nFrameCurrent].nFrameStartCpu;
@@ -739,7 +739,7 @@ void MicroProfileCenter(int64_t nTickCenter)
 #ifdef MICROPROFILE_DEBUG
 uint64_t* g_pMicroProfileDumpStart = 0;
 uint64_t* g_pMicroProfileDumpEnd = 0;
-void MicroProfileDebugDumpRange()
+inline void MicroProfileDebugDumpRange()
 {
     MicroProfile& S = *MicroProfileGet();
     if(g_pMicroProfileDumpStart != g_pMicroProfileDumpEnd)
@@ -777,7 +777,7 @@ void MicroProfileDebugDumpRange()
 
 #define MICROPROFILE_HOVER_DIST 0.5f
 
-void MicroProfileDrawDetailedContextSwitchBars(uint32_t nY, uint32_t nThreadId, uint32_t nContextSwitchStart, uint32_t nContextSwitchEnd, int64_t nBaseTicks, uint32_t nBaseY)
+inline void MicroProfileDrawDetailedContextSwitchBars(uint32_t nY, uint32_t nThreadId, uint32_t nContextSwitchStart, uint32_t nContextSwitchEnd, int64_t nBaseTicks, uint32_t nBaseY)
 {
     MicroProfile& S = *MicroProfileGet();
     int64_t nTickIn = -1;
@@ -841,7 +841,7 @@ void MicroProfileDrawDetailedContextSwitchBars(uint32_t nY, uint32_t nThreadId,
     }
 }
 
-void MicroProfileDrawDetailedBars(uint32_t nWidth, uint32_t nHeight, int nBaseY, int nSelectedFrame)
+inline void MicroProfileDrawDetailedBars(uint32_t nWidth, uint32_t nHeight, int nBaseY, int nSelectedFrame)
 {
     MicroProfile& S = *MicroProfileGet();
     MP_DEBUG_DUMP_RANGE();
@@ -1325,7 +1325,7 @@ void MicroProfileDrawDetailedBars(uint32_t nWidth, uint32_t nHeight, int nBaseY,
 }
 
 
-void MicroProfileDrawDetailedFrameHistory(uint32_t nWidth, uint32_t nHeight, uint32_t nBaseY, uint32_t nSelectedFrame)
+inline void MicroProfileDrawDetailedFrameHistory(uint32_t nWidth, uint32_t nHeight, uint32_t nBaseY, uint32_t nSelectedFrame)
 {
     MicroProfile& S = *MicroProfileGet();
 
@@ -1379,7 +1379,7 @@ void MicroProfileDrawDetailedFrameHistory(uint32_t nWidth, uint32_t nHeight, uin
     }
     MicroProfileDrawBox(fSelectionStart, nBaseY, fSelectionEnd, nBaseY+MICROPROFILE_FRAME_HISTORY_HEIGHT, MICROPROFILE_FRAME_HISTORY_COLOR_HIGHTLIGHT, MicroProfileBoxTypeFlat);
 }
-void MicroProfileDrawDetailedView(uint32_t nWidth, uint32_t nHeight)
+inline void MicroProfileDrawDetailedView(uint32_t nWidth, uint32_t nHeight)
 {
     MicroProfile& S = *MicroProfileGet();
 
@@ -1416,11 +1416,11 @@ void MicroProfileDrawDetailedView(uint32_t nWidth, uint32_t nHeight)
     MicroProfileDrawDetailedFrameHistory(nWidth, nHeight, nBaseY, nSelectedFrame);
 }
 
-void MicroProfileDrawTextRight(uint32_t nX, uint32_t nY, uint32_t nColor, const char* pStr, uint32_t nStrLen)
+inline void MicroProfileDrawTextRight(uint32_t nX, uint32_t nY, uint32_t nColor, const char* pStr, uint32_t nStrLen)
 {
     MicroProfileDrawText(nX - nStrLen * (MICROPROFILE_TEXT_WIDTH+1), nY, nColor, pStr, nStrLen);
 }
-void MicroProfileDrawHeader(int32_t nX, uint32_t nWidth, const char* pName)
+inline void MicroProfileDrawHeader(int32_t nX, uint32_t nWidth, const char* pName)
 {
     if(pName)
     {
@@ -1432,7 +1432,7 @@ void MicroProfileDrawHeader(int32_t nX, uint32_t nWidth, const char* pName)
 
 typedef void (*MicroProfileLoopGroupCallback)(uint32_t nTimer, uint32_t nIdx, uint64_t nGroupMask, uint32_t nX, uint32_t nY, void* pData);
 
-void MicroProfileLoopActiveGroupsDraw(int32_t nX, int32_t nY, const char* pName, MicroProfileLoopGroupCallback CB, void* pData)
+inline void MicroProfileLoopActiveGroupsDraw(int32_t nX, int32_t nY, const char* pName, MicroProfileLoopGroupCallback CB, void* pData)
 {
     MicroProfile& S = *MicroProfileGet();
     nY += MICROPROFILE_TEXT_HEIGHT + 2;
@@ -1465,7 +1465,7 @@ void MicroProfileLoopActiveGroupsDraw(int32_t nX, int32_t nY, const char* pName,
 }
 
 
-void MicroProfileCalcTimers(float* pTimers, float* pAverage, float* pMax, float* pCallAverage, float* pExclusive, float* pAverageExclusive, float* pMaxExclusive, uint64_t nGroup, uint32_t nSize)
+inline void MicroProfileCalcTimers(float* pTimers, float* pAverage, float* pMax, float* pCallAverage, float* pExclusive, float* pAverageExclusive, float* pMaxExclusive, uint64_t nGroup, uint32_t nSize)
 {
     MicroProfile& S = *MicroProfileGet();
 
@@ -1527,7 +1527,7 @@ void MicroProfileCalcTimers(float* pTimers, float* pAverage, float* pMax, float*
 
 #define SBUF_MAX 32
 
-void MicroProfileDrawBarArrayCallback(uint32_t nTimer, uint32_t nIdx, uint64_t nGroupMask, uint32_t nX, uint32_t nY, void* pExtra)
+inline void MicroProfileDrawBarArrayCallback(uint32_t nTimer, uint32_t nIdx, uint64_t nGroupMask, uint32_t nX, uint32_t nY, void* pExtra)
 {
     const uint32_t nHeight = MICROPROFILE_TEXT_HEIGHT;
     const uint32_t nTextWidth = 6 * (1+MICROPROFILE_TEXT_WIDTH);
@@ -1547,7 +1547,7 @@ void MicroProfileDrawBarArrayCallback(uint32_t nTimer, uint32_t nIdx, uint64_t n
 }
 
 
-uint32_t MicroProfileDrawBarArray(int32_t nX, int32_t nY, float* pTimers, const char* pName, uint32_t nTotalHeight, float* pTimers2 = NULL)
+inline uint32_t MicroProfileDrawBarArray(int32_t nX, int32_t nY, float* pTimers, const char* pName, uint32_t nTotalHeight, float* pTimers2 = NULL)
 {
     const uint32_t nTextWidth = 6 * (1+MICROPROFILE_TEXT_WIDTH);
     const uint32_t nWidth = MICROPROFILE_BAR_WIDTH;
@@ -1559,7 +1559,7 @@ uint32_t MicroProfileDrawBarArray(int32_t nX, int32_t nY, float* pTimers, const
     return nWidth + 5 + nTextWidth;
 
 }
-void MicroProfileDrawBarCallCountCallback(uint32_t nTimer, uint32_t nIdx, uint64_t nGroupMask, uint32_t nX, uint32_t nY, void* pExtra)
+inline void MicroProfileDrawBarCallCountCallback(uint32_t nTimer, uint32_t nIdx, uint64_t nGroupMask, uint32_t nX, uint32_t nY, void* pExtra)
 {
     MicroProfile& S = *MicroProfileGet();
     char sBuffer[SBUF_MAX];
@@ -1567,7 +1567,7 @@ void MicroProfileDrawBarCallCountCallback(uint32_t nTimer, uint32_t nIdx, uint64
     MicroProfileDrawText(nX, nY, (uint32_t)-1, sBuffer, nLen);
 }
 
-uint32_t MicroProfileDrawBarCallCount(int32_t nX, int32_t nY, const char* pName)
+inline uint32_t MicroProfileDrawBarCallCount(int32_t nX, int32_t nY, const char* pName)
 {
     MicroProfileLoopActiveGroupsDraw(nX, nY, pName, MicroProfileDrawBarCallCountCallback, 0);
     const uint32_t nTextWidth = 6 * MICROPROFILE_TEXT_WIDTH;
@@ -1581,7 +1581,7 @@ struct MicroProfileMetaAverageArgs
     float fRcpFrames;
 };
 
-void MicroProfileDrawBarMetaAverageCallback(uint32_t nTimer, uint32_t nIdx, uint64_t nGroupMask, uint32_t nX, uint32_t nY, void* pExtra)
+inline void MicroProfileDrawBarMetaAverageCallback(uint32_t nTimer, uint32_t nIdx, uint64_t nGroupMask, uint32_t nX, uint32_t nY, void* pExtra)
 {
     MicroProfileMetaAverageArgs* pArgs = (MicroProfileMetaAverageArgs*)pExtra;
     uint64_t* pCounters = pArgs->pCounters;
@@ -1591,7 +1591,7 @@ void MicroProfileDrawBarMetaAverageCallback(uint32_t nTimer, uint32_t nIdx, uint
     MicroProfileDrawText(nX - nLen * (MICROPROFILE_TEXT_WIDTH+1), nY, (uint32_t)-1, sBuffer, nLen);
 }
 
-uint32_t MicroProfileDrawBarMetaAverage(int32_t nX, int32_t nY, uint64_t* pCounters, const char* pName, uint32_t nTotalHeight)
+inline uint32_t MicroProfileDrawBarMetaAverage(int32_t nX, int32_t nY, uint64_t* pCounters, const char* pName, uint32_t nTotalHeight)
 {
     if(!pName)
         return 0;
@@ -1605,7 +1605,7 @@ uint32_t MicroProfileDrawBarMetaAverage(int32_t nX, int32_t nY, uint64_t* pCount
 }
 
 
-void MicroProfileDrawBarMetaCountCallback(uint32_t nTimer, uint32_t nIdx, uint64_t nGroupMask, uint32_t nX, uint32_t nY, void* pExtra)
+inline void MicroProfileDrawBarMetaCountCallback(uint32_t nTimer, uint32_t nIdx, uint64_t nGroupMask, uint32_t nX, uint32_t nY, void* pExtra)
 {
     uint64_t* pCounters = (uint64_t*)pExtra;
     char sBuffer[SBUF_MAX];
@@ -1613,7 +1613,7 @@ void MicroProfileDrawBarMetaCountCallback(uint32_t nTimer, uint32_t nIdx, uint64
     MicroProfileDrawText(nX - nLen * (MICROPROFILE_TEXT_WIDTH+1), nY, (uint32_t)-1, sBuffer, nLen);
 }
 
-uint32_t MicroProfileDrawBarMetaCount(int32_t nX, int32_t nY, uint64_t* pCounters, const char* pName, uint32_t nTotalHeight)
+inline uint32_t MicroProfileDrawBarMetaCount(int32_t nX, int32_t nY, uint64_t* pCounters, const char* pName, uint32_t nTotalHeight)
 {
     if(!pName)
         return 0;
@@ -1625,7 +1625,7 @@ uint32_t MicroProfileDrawBarMetaCount(int32_t nX, int32_t nY, uint64_t* pCounter
     return 5 + nTextWidth;
 }
 
-void MicroProfileDrawBarLegendCallback(uint32_t nTimer, uint32_t nIdx, uint64_t nGroupMask, uint32_t nX, uint32_t nY, void* pExtra)
+inline void MicroProfileDrawBarLegendCallback(uint32_t nTimer, uint32_t nIdx, uint64_t nGroupMask, uint32_t nX, uint32_t nY, void* pExtra)
 {
     MicroProfile& S = *MicroProfileGet();
     if (S.TimerInfo[nTimer].bGraph)
@@ -1640,7 +1640,7 @@ void MicroProfileDrawBarLegendCallback(uint32_t nTimer, uint32_t nIdx, uint64_t
     }
 }
 
-uint32_t MicroProfileDrawBarLegend(int32_t nX, int32_t nY, uint32_t nTotalHeight, uint32_t nMaxWidth)
+inline uint32_t MicroProfileDrawBarLegend(int32_t nX, int32_t nY, uint32_t nTotalHeight, uint32_t nMaxWidth)
 {
     MicroProfileDrawLineVertical(nX-5, nY, nTotalHeight, UI.nOpacityBackground | g_nMicroProfileBackColors[0]|g_nMicroProfileBackColors[1]);
     MicroProfileLoopActiveGroupsDraw(nMaxWidth, nY, 0, MicroProfileDrawBarLegendCallback, 0);
@@ -1807,7 +1807,7 @@ void MicroProfileDumpTimers()
     }
 }
 
-void MicroProfileDrawBarView(uint32_t nScreenWidth, uint32_t nScreenHeight)
+inline void MicroProfileDrawBarView(uint32_t nScreenWidth, uint32_t nScreenHeight)
 {
     MicroProfile& S = *MicroProfileGet();
 
@@ -1951,7 +1951,7 @@ typedef const char* (*MicroProfileSubmenuCallback)(int, bool* bSelected);
 typedef void (*MicroProfileClickCallback)(int);
 
 
-const char* MicroProfileUIMenuMode(int nIndex, bool* bSelected)
+inline const char* MicroProfileUIMenuMode(int nIndex, bool* bSelected)
 {
     MicroProfile& S = *MicroProfileGet();
     switch(nIndex)
@@ -1979,7 +1979,7 @@ const char* MicroProfileUIMenuMode(int nIndex, bool* bSelected)
     }
 }
 
-const char* MicroProfileUIMenuGroups(int nIndex, bool* bSelected)
+inline const char* MicroProfileUIMenuGroups(int nIndex, bool* bSelected)
 {
     MicroProfile& S = *MicroProfileGet();
     *bSelected = false;
@@ -2012,7 +2012,7 @@ const char* MicroProfileUIMenuGroups(int nIndex, bool* bSelected)
     }
 }
 
-const char* MicroProfileUIMenuAggregate(int nIndex, bool* bSelected)
+inline const char* MicroProfileUIMenuAggregate(int nIndex, bool* bSelected)
 {
     MicroProfile& S = *MicroProfileGet();
     if(nIndex < sizeof(g_MicroProfileAggregatePresets)/sizeof(g_MicroProfileAggregatePresets[0]))
@@ -2032,7 +2032,7 @@ const char* MicroProfileUIMenuAggregate(int nIndex, bool* bSelected)
 
 }
 
-const char* MicroProfileUIMenuTimers(int nIndex, bool* bSelected)
+inline const char* MicroProfileUIMenuTimers(int nIndex, bool* bSelected)
 {
     MicroProfile& S = *MicroProfileGet();
     *bSelected = 0 != (S.nBars & (1 << nIndex));
@@ -2054,7 +2054,7 @@ const char* MicroProfileUIMenuTimers(int nIndex, bool* bSelected)
     return 0;
 }
 
-const char* MicroProfileUIMenuOptions(int nIndex, bool* bSelected)
+inline const char* MicroProfileUIMenuOptions(int nIndex, bool* bSelected)
 {
     MicroProfile& S = *MicroProfileGet();
     if(nIndex >= MICROPROFILE_OPTION_SIZE) return 0;
@@ -2094,7 +2094,7 @@ const char* MicroProfileUIMenuOptions(int nIndex, bool* bSelected)
     return UI.Options[nIndex].Text;
 }
 
-const char* MicroProfileUIMenuPreset(int nIndex, bool* bSelected)
+inline const char* MicroProfileUIMenuPreset(int nIndex, bool* bSelected)
 {
     static char buf[128];
     *bSelected = false;
@@ -2118,7 +2118,7 @@ const char* MicroProfileUIMenuPreset(int nIndex, bool* bSelected)
     }
 }
 
-const char* MicroProfileUIMenuCustom(int nIndex, bool* bSelected)
+inline const char* MicroProfileUIMenuCustom(int nIndex, bool* bSelected)
 {
     if((uint32_t)-1 == UI.nCustomActive)
     {
@@ -2145,13 +2145,13 @@ const char* MicroProfileUIMenuCustom(int nIndex, bool* bSelected)
     }
 }
 
-const char* MicroProfileUIMenuEmpty(int nIndex, bool* bSelected)
+inline const char* MicroProfileUIMenuEmpty(int nIndex, bool* bSelected)
 {
     return 0;
 }
 
 
-void MicroProfileUIClickMode(int nIndex)
+inline void MicroProfileUIClickMode(int nIndex)
 {
     MicroProfile& S = *MicroProfileGet();
     switch(nIndex)
@@ -2176,7 +2176,7 @@ void MicroProfileUIClickMode(int nIndex)
     }
 }
 
-void MicroProfileUIClickGroups(int nIndex)
+inline void MicroProfileUIClickGroups(int nIndex)
 {
     MicroProfile& S = *MicroProfileGet();
     if(nIndex == 0)
@@ -2208,7 +2208,7 @@ void MicroProfileUIClickGroups(int nIndex)
     }
 }
 
-void MicroProfileUIClickAggregate(int nIndex)
+inline void MicroProfileUIClickAggregate(int nIndex)
 {
     MicroProfile& S = *MicroProfileGet();
     S.nAggregateFlip = g_MicroProfileAggregatePresets[nIndex];
@@ -2218,13 +2218,13 @@ void MicroProfileUIClickAggregate(int nIndex)
     }
 }
 
-void MicroProfileUIClickTimers(int nIndex)
+inline void MicroProfileUIClickTimers(int nIndex)
 {
     MicroProfile& S = *MicroProfileGet();
     S.nBars ^= (1 << nIndex);
 }
 
-void MicroProfileUIClickOptions(int nIndex)
+inline void MicroProfileUIClickOptions(int nIndex)
 {
     MicroProfile& S = *MicroProfileGet();
     switch(UI.Options[nIndex].nSubType)
@@ -2271,7 +2271,7 @@ void MicroProfileUIClickOptions(int nIndex)
     }
 }
 
-void MicroProfileUIClickPreset(int nIndex)
+inline void MicroProfileUIClickPreset(int nIndex)
 {
     int nNumPresets = sizeof(g_MicroProfilePresetNames) / sizeof(g_MicroProfilePresetNames[0]);
     int nIndexSave = nIndex - nNumPresets - 1;
@@ -2285,7 +2285,7 @@ void MicroProfileUIClickPreset(int nIndex)
     }
 }
 
-void MicroProfileUIClickCustom(int nIndex)
+inline void MicroProfileUIClickCustom(int nIndex)
 {
     if(nIndex == 0)
     {
@@ -2298,13 +2298,13 @@ void MicroProfileUIClickCustom(int nIndex)
 
 }
 
-void MicroProfileUIClickEmpty(int nIndex)
+inline void MicroProfileUIClickEmpty(int nIndex)
 {
 
 }
 
 
-void MicroProfileDrawMenu(uint32_t nWidth, uint32_t nHeight)
+inline void MicroProfileDrawMenu(uint32_t nWidth, uint32_t nHeight)
 {
     MicroProfile& S = *MicroProfileGet();
 
@@ -2489,7 +2489,7 @@ void MicroProfileDrawMenu(uint32_t nWidth, uint32_t nHeight)
 }
 
 
-void MicroProfileMoveGraph()
+inline void MicroProfileMoveGraph()
 {
 
     int nZoom = UI.nMouseWheelDelta;
@@ -2536,7 +2536,7 @@ void MicroProfileMoveGraph()
         UI.nOffsetY = 0;
 }
 
-void MicroProfileDrawCustom(uint32_t nWidth, uint32_t nHeight)
+inline void MicroProfileDrawCustom(uint32_t nWidth, uint32_t nHeight)
 {
     if((uint32_t)-1 != UI.nCustomActive)
     {
@@ -2633,7 +2633,7 @@ void MicroProfileDrawCustom(uint32_t nWidth, uint32_t nHeight)
         }
     }
 }
-void MicroProfileDraw(uint32_t nWidth, uint32_t nHeight)
+inline void MicroProfileDraw(uint32_t nWidth, uint32_t nHeight)
 {
     MICROPROFILE_SCOPE(g_MicroProfileDraw);
     MicroProfile& S = *MicroProfileGet();
@@ -3226,7 +3226,7 @@ void MicroProfileLoadPreset(const char* pSuffix)
     }
 }
 
-uint32_t MicroProfileCustomGroupFind(const char* pCustomName)
+inline uint32_t MicroProfileCustomGroupFind(const char* pCustomName)
 {
     for(uint32_t i = 0; i < UI.nCustomCount; ++i)
     {
@@ -3238,7 +3238,7 @@ uint32_t MicroProfileCustomGroupFind(const char* pCustomName)
     return (uint32_t)-1;
 }
 
-uint32_t MicroProfileCustomGroup(const char* pCustomName)
+inline uint32_t MicroProfileCustomGroup(const char* pCustomName)
 {
     for(uint32_t i = 0; i < UI.nCustomCount; ++i)
     {
@@ -3271,7 +3271,7 @@ void MicroProfileCustomGroup(const char* pCustomName, uint32_t nMaxTimers, uint3
     UI.Custom[nIndex].nAggregateFlip = nAggregateFlip;
 }
 
-void MicroProfileCustomGroupEnable(uint32_t nIndex)
+inline void MicroProfileCustomGroupEnable(uint32_t nIndex)
 {
     if(nIndex < UI.nCustomCount)
     {

externals/opus/CMakeLists.txt

@@ -203,7 +203,11 @@ endif()
 target_compile_definitions(opus PRIVATE OPUS_BUILD ENABLE_HARDENING)
 
 if(NOT MSVC)
-    target_compile_definitions(opus PRIVATE _FORTIFY_SOURCE=2)
+    if(MINGW)
+        target_compile_definitions(opus PRIVATE _FORTIFY_SOURCE=0)
+    else()
+        target_compile_definitions(opus PRIVATE _FORTIFY_SOURCE=2)
+    endif()
 endif()
 
 # It is strongly recommended to uncomment one of these VAR_ARRAYS: Use C99

license.txt

@@ -343,8 +343,8 @@ The icons used in this project have the following licenses:
 
 Icon Name                                   | License       | Origin/Author
 ---                                         | ---           | ---
-checked.png                                 | Free for non-commercial use
-failed.png                                  | Free for non-commercial use
+checked.png                                 | CC BY-ND 3.0  | https://icons8.com
+failed.png                                  | CC BY-ND 3.0  | https://icons8.com
 lock.png                                    | CC BY-ND 3.0  | https://icons8.com
 plus_folder.png (Default, Dark)             | CC BY-ND 3.0  | https://icons8.com
 bad_folder.png (Default, Dark)              | CC BY-ND 3.0  | https://icons8.com

src/CMakeLists.txt

@@ -53,7 +53,13 @@ if (MSVC)
 else()
     add_compile_options(
         -Wall
+        -Werror=implicit-fallthrough
+        -Werror=missing-declarations
+        -Werror=reorder
+        -Wextra
+        -Wmissing-declarations
         -Wno-attributes
+        -Wno-unused-parameter
     )
 
     if (APPLE AND CMAKE_CXX_COMPILER_ID STREQUAL Clang)

src/audio_core/CMakeLists.txt

@@ -7,9 +7,12 @@ add_library(audio_core STATIC
     audio_out.h
     audio_renderer.cpp
     audio_renderer.h
+    behavior_info.cpp
+    behavior_info.h
     buffer.h
     codec.cpp
     codec.h
+    common.h
     null_sink.h
     sink.h
     sink_details.cpp

src/audio_core/algorithm/interpolate.cpp

@@ -5,16 +5,17 @@
 #define _USE_MATH_DEFINES
 
 #include <algorithm>
+#include <climits>
 #include <cmath>
-#include <limits>
 #include <vector>
+
 #include "audio_core/algorithm/interpolate.h"
 #include "common/common_types.h"
 #include "common/logging/log.h"
 
 namespace AudioCore {
 
-constexpr std::array<s16, 512> curve_lut0 = {
+constexpr std::array<s16, 512> curve_lut0{
     6600,  19426, 6722,  3,     6479,  19424, 6845,  9,     6359,  19419, 6968,  15,    6239,
     19412, 7093,  22,    6121,  19403, 7219,  28,    6004,  19391, 7345,  34,    5888,  19377,
     7472,  41,    5773,  19361, 7600,  48,    5659,  19342, 7728,  55,    5546,  19321, 7857,
@@ -56,7 +57,7 @@ constexpr std::array<s16, 512> curve_lut0 = {
     19403, 6121,  22,    7093,  19412, 6239,  15,    6968,  19419, 6359,  9,     6845,  19424,
     6479,  3,     6722,  19426, 6600};
 
-constexpr std::array<s16, 512> curve_lut1 = {
+constexpr std::array<s16, 512> curve_lut1{
     -68,   32639, 69,    -5,    -200,  32630, 212,   -15,   -328,  32613, 359,   -26,   -450,
     32586, 512,   -36,   -568,  32551, 669,   -47,   -680,  32507, 832,   -58,   -788,  32454,
     1000,  -69,   -891,  32393, 1174,  -80,   -990,  32323, 1352,  -92,   -1084, 32244, 1536,
@@ -98,7 +99,7 @@ constexpr std::array<s16, 512> curve_lut1 = {
     32551, -568,  -36,   512,   32586, -450,  -26,   359,   32613, -328,  -15,   212,   32630,
     -200,  -5,    69,    32639, -68};
 
-constexpr std::array<s16, 512> curve_lut2 = {
+constexpr std::array<s16, 512> curve_lut2{
     3195,  26287, 3329,  -32,   3064,  26281, 3467,  -34,   2936,  26270, 3608,  -38,   2811,
     26253, 3751,  -42,   2688,  26230, 3897,  -46,   2568,  26202, 4046,  -50,   2451,  26169,
     4199,  -54,   2338,  26130, 4354,  -58,   2227,  26085, 4512,  -63,   2120,  26035, 4673,
@@ -146,10 +147,10 @@ std::vector<s16> Interpolate(InterpolationState& state, std::vector<s16> input,
 
     if (ratio <= 0) {
         LOG_CRITICAL(Audio, "Nonsensical interpolation ratio {}", ratio);
-        ratio = 1.0;
+        return input;
     }
 
-    const int step = static_cast<int>(ratio * 0x8000);
+    const s32 step{static_cast<s32>(ratio * 0x8000)};
     const std::array<s16, 512>& lut = [step] {
         if (step > 0xaaaa) {
             return curve_lut0;
@@ -160,28 +161,37 @@ std::vector<s16> Interpolate(InterpolationState& state, std::vector<s16> input,
         return curve_lut2;
     }();
 
-    std::vector<s16> output(static_cast<std::size_t>(input.size() / ratio));
-    int in_offset = 0;
-    for (std::size_t out_offset = 0; out_offset < output.size(); out_offset += 2) {
-        const int lut_index = (state.fraction >> 8) * 4;
+    const std::size_t num_frames{input.size() / 2};
 
-        const int l = input[(in_offset + 0) * 2 + 0] * lut[lut_index + 0] +
-                      input[(in_offset + 1) * 2 + 0] * lut[lut_index + 1] +
-                      input[(in_offset + 2) * 2 + 0] * lut[lut_index + 2] +
-                      input[(in_offset + 3) * 2 + 0] * lut[lut_index + 3];
+    std::vector<s16> output;
+    output.reserve(static_cast<std::size_t>(input.size() / ratio + InterpolationState::taps));
 
-        const int r = input[(in_offset + 0) * 2 + 1] * lut[lut_index + 0] +
-                      input[(in_offset + 1) * 2 + 1] * lut[lut_index + 1] +
-                      input[(in_offset + 2) * 2 + 1] * lut[lut_index + 2] +
-                      input[(in_offset + 3) * 2 + 1] * lut[lut_index + 3];
+    for (std::size_t frame{}; frame < num_frames; ++frame) {
+        const std::size_t lut_index{(state.fraction >> 8) * InterpolationState::taps};
 
-        const int new_offset = state.fraction + step;
+        std::rotate(state.history.begin(), state.history.end() - 1, state.history.end());
+        state.history[0][0] = input[frame * 2 + 0];
+        state.history[0][1] = input[frame * 2 + 1];
 
-        in_offset += new_offset >> 15;
-        state.fraction = new_offset & 0x7fff;
+        while (state.position <= 1.0) {
+            const s32 left{state.history[0][0] * lut[lut_index + 0] +
+                           state.history[1][0] * lut[lut_index + 1] +
+                           state.history[2][0] * lut[lut_index + 2] +
+                           state.history[3][0] * lut[lut_index + 3]};
+            const s32 right{state.history[0][1] * lut[lut_index + 0] +
+                            state.history[1][1] * lut[lut_index + 1] +
+                            state.history[2][1] * lut[lut_index + 2] +
+                            state.history[3][1] * lut[lut_index + 3]};
+            const s32 new_offset{state.fraction + step};
 
-        output[out_offset + 0] = static_cast<s16>(std::clamp(l >> 15, SHRT_MIN, SHRT_MAX));
-        output[out_offset + 1] = static_cast<s16>(std::clamp(r >> 15, SHRT_MIN, SHRT_MAX));
+            state.fraction = new_offset & 0x7fff;
+
+            output.emplace_back(static_cast<s16>(std::clamp(left >> 15, SHRT_MIN, SHRT_MAX)));
+            output.emplace_back(static_cast<s16>(std::clamp(right >> 15, SHRT_MIN, SHRT_MAX)));
+
+            state.position += ratio;
+        }
+        state.position -= 1.0;
     }
 
     return output;

src/audio_core/algorithm/interpolate.h

@@ -6,12 +6,17 @@
 
 #include <array>
 #include <vector>
+
 #include "common/common_types.h"
 
 namespace AudioCore {
 
 struct InterpolationState {
-    int fraction = 0;
+    static constexpr std::size_t taps{4};
+    static constexpr std::size_t history_size{taps * 2 - 1};
+    std::array<std::array<s16, 2>, history_size> history{};
+    double position{};
+    s32 fraction{};
 };
 
 /// Interpolates input signal to produce output signal.

src/audio_core/audio_renderer.cpp

@@ -6,6 +6,7 @@
 #include "audio_core/audio_out.h"
 #include "audio_core/audio_renderer.h"
 #include "audio_core/codec.h"
+#include "audio_core/common.h"
 #include "common/assert.h"
 #include "common/logging/log.h"
 #include "core/core.h"
@@ -36,9 +37,9 @@ public:
     }
 
     void SetWaveIndex(std::size_t index);
-    std::vector<s16> DequeueSamples(std::size_t sample_count, Memory::Memory& memory);
+    std::vector<s16> DequeueSamples(std::size_t sample_count, Core::Memory::Memory& memory);
     void UpdateState();
-    void RefreshBuffer(Memory::Memory& memory);
+    void RefreshBuffer(Core::Memory::Memory& memory);
 
 private:
     bool is_in_use{};
@@ -66,19 +67,20 @@ public:
         return info;
     }
 
-    void UpdateState(Memory::Memory& memory);
+    void UpdateState(Core::Memory::Memory& memory);
 
 private:
     EffectOutStatus out_status{};
     EffectInStatus info{};
 };
-AudioRenderer::AudioRenderer(Core::Timing::CoreTiming& core_timing, Memory::Memory& memory_,
+
+AudioRenderer::AudioRenderer(Core::Timing::CoreTiming& core_timing, Core::Memory::Memory& memory_,
                              AudioRendererParameter params,
                              std::shared_ptr<Kernel::WritableEvent> buffer_event,
                              std::size_t instance_number)
     : worker_params{params}, buffer_event{buffer_event}, voices(params.voice_count),
       effects(params.effect_count), memory{memory_} {
-
+    behavior_info.SetUserRevision(params.revision);
     audio_out = std::make_unique<AudioCore::AudioOut>();
     stream = audio_out->OpenStream(core_timing, STREAM_SAMPLE_RATE, STREAM_NUM_CHANNELS,
                                    fmt::format("AudioRenderer-Instance{}", instance_number),
@@ -108,17 +110,17 @@ Stream::State AudioRenderer::GetStreamState() const {
     return stream->GetState();
 }
 
-static constexpr u32 VersionFromRevision(u32_le rev) {
-    // "REV7" -> 7
-    return ((rev >> 24) & 0xff) - 0x30;
-}
-
-std::vector<u8> AudioRenderer::UpdateAudioRenderer(const std::vector<u8>& input_params) {
+ResultVal<std::vector<u8>> AudioRenderer::UpdateAudioRenderer(const std::vector<u8>& input_params) {
     // Copy UpdateDataHeader struct
     UpdateDataHeader config{};
     std::memcpy(&config, input_params.data(), sizeof(UpdateDataHeader));
     u32 memory_pool_count = worker_params.effect_count + (worker_params.voice_count * 4);
 
+    if (!behavior_info.UpdateInput(input_params, sizeof(UpdateDataHeader))) {
+        LOG_ERROR(Audio, "Failed to update behavior info input parameters");
+        return Audren::ERR_INVALID_PARAMETERS;
+    }
+
     // Copy MemoryPoolInfo structs
     std::vector<MemoryPoolInfo> mem_pool_info(memory_pool_count);
     std::memcpy(mem_pool_info.data(),
@@ -172,8 +174,7 @@ std::vector<u8> AudioRenderer::UpdateAudioRenderer(const std::vector<u8>& input_
     // Copy output header
     UpdateDataHeader response_data{worker_params};
     std::vector<u8> output_params(response_data.total_size);
-    const auto audren_revision = VersionFromRevision(config.revision);
-    if (audren_revision >= 5) {
+    if (behavior_info.IsElapsedFrameCountSupported()) {
         response_data.frame_count = 0x10;
         response_data.total_size += 0x10;
     }
@@ -199,7 +200,19 @@ std::vector<u8> AudioRenderer::UpdateAudioRenderer(const std::vector<u8>& input_
                     sizeof(EffectOutStatus));
         effect_out_status_offset += sizeof(EffectOutStatus);
     }
-    return output_params;
+
+    // Update behavior info output
+    const std::size_t behavior_out_status_offset{
+        sizeof(UpdateDataHeader) + response_data.memory_pools_size + response_data.voices_size +
+        response_data.effects_size + response_data.sinks_size +
+        response_data.performance_manager_size};
+
+    if (!behavior_info.UpdateOutput(output_params, behavior_out_status_offset)) {
+        LOG_ERROR(Audio, "Failed to update behavior info output parameters");
+        return Audren::ERR_INVALID_PARAMETERS;
+    }
+
+    return MakeResult(output_params);
 }
 
 void AudioRenderer::VoiceState::SetWaveIndex(std::size_t index) {
@@ -208,7 +221,7 @@ void AudioRenderer::VoiceState::SetWaveIndex(std::size_t index) {
 }
 
 std::vector<s16> AudioRenderer::VoiceState::DequeueSamples(std::size_t sample_count,
-                                                           Memory::Memory& memory) {
+                                                           Core::Memory::Memory& memory) {
     if (!IsPlaying()) {
         return {};
     }
@@ -258,7 +271,7 @@ void AudioRenderer::VoiceState::UpdateState() {
     is_in_use = info.is_in_use;
 }
 
-void AudioRenderer::VoiceState::RefreshBuffer(Memory::Memory& memory) {
+void AudioRenderer::VoiceState::RefreshBuffer(Core::Memory::Memory& memory) {
     const auto wave_buffer_address = info.wave_buffer[wave_index].buffer_addr;
     const auto wave_buffer_size = info.wave_buffer[wave_index].buffer_sz;
     std::vector<s16> new_samples(wave_buffer_size / sizeof(s16));
@@ -310,7 +323,7 @@ void AudioRenderer::VoiceState::RefreshBuffer(Memory::Memory& memory) {
     is_refresh_pending = false;
 }
 
-void AudioRenderer::EffectState::UpdateState(Memory::Memory& memory) {
+void AudioRenderer::EffectState::UpdateState(Core::Memory::Memory& memory) {
     if (info.is_new) {
         out_status.state = EffectStatus::New;
     } else {

src/audio_core/audio_renderer.h

@@ -8,11 +8,13 @@
 #include <memory>
 #include <vector>
 
+#include "audio_core/behavior_info.h"
 #include "audio_core/stream.h"
 #include "common/common_funcs.h"
 #include "common/common_types.h"
 #include "common/swap.h"
 #include "core/hle/kernel/object.h"
+#include "core/hle/result.h"
 
 namespace Core::Timing {
 class CoreTiming;
@@ -22,7 +24,7 @@ namespace Kernel {
 class WritableEvent;
 }
 
-namespace Memory {
+namespace Core::Memory {
 class Memory;
 }
 
@@ -189,7 +191,7 @@ struct UpdateDataHeader {
     UpdateDataHeader() {}
 
     explicit UpdateDataHeader(const AudioRendererParameter& config) {
-        revision = Common::MakeMagic('R', 'E', 'V', '4'); // 5.1.0 Revision
+        revision = Common::MakeMagic('R', 'E', 'V', '8'); // 9.2.0 Revision
         behavior_size = 0xb0;
         memory_pools_size = (config.effect_count + (config.voice_count * 4)) * 0x10;
         voices_size = config.voice_count * 0x10;
@@ -221,12 +223,12 @@ static_assert(sizeof(UpdateDataHeader) == 0x40, "UpdateDataHeader has wrong size
 
 class AudioRenderer {
 public:
-    AudioRenderer(Core::Timing::CoreTiming& core_timing, Memory::Memory& memory_,
+    AudioRenderer(Core::Timing::CoreTiming& core_timing, Core::Memory::Memory& memory_,
                   AudioRendererParameter params,
                   std::shared_ptr<Kernel::WritableEvent> buffer_event, std::size_t instance_number);
     ~AudioRenderer();
 
-    std::vector<u8> UpdateAudioRenderer(const std::vector<u8>& input_params);
+    ResultVal<std::vector<u8>> UpdateAudioRenderer(const std::vector<u8>& input_params);
     void QueueMixedBuffer(Buffer::Tag tag);
     void ReleaseAndQueueBuffers();
     u32 GetSampleRate() const;
@@ -237,6 +239,7 @@ public:
 private:
     class EffectState;
     class VoiceState;
+    BehaviorInfo behavior_info{};
 
     AudioRendererParameter worker_params;
     std::shared_ptr<Kernel::WritableEvent> buffer_event;
@@ -244,7 +247,7 @@ private:
     std::vector<EffectState> effects;
     std::unique_ptr<AudioOut> audio_out;
     StreamPtr stream;
-    Memory::Memory& memory;
+    Core::Memory::Memory& memory;
 };
 
 } // namespace AudioCore

src/audio_core/behavior_info.cpp

@@ -0,0 +1,100 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <cstring>
+#include "audio_core/behavior_info.h"
+#include "audio_core/common.h"
+#include "common/logging/log.h"
+
+namespace AudioCore {
+
+BehaviorInfo::BehaviorInfo() : process_revision(CURRENT_PROCESS_REVISION) {}
+BehaviorInfo::~BehaviorInfo() = default;
+
+bool BehaviorInfo::UpdateInput(const std::vector<u8>& buffer, std::size_t offset) {
+    if (!CanConsumeBuffer(buffer.size(), offset, sizeof(InParams))) {
+        LOG_ERROR(Audio, "Buffer is an invalid size!");
+        return false;
+    }
+    InParams params{};
+    std::memcpy(&params, buffer.data() + offset, sizeof(InParams));
+
+    if (!IsValidRevision(params.revision)) {
+        LOG_ERROR(Audio, "Invalid input revision, revision=0x{:08X}", params.revision);
+        return false;
+    }
+
+    if (user_revision != params.revision) {
+        LOG_ERROR(Audio,
+                  "User revision differs from input revision, expecting 0x{:08X} but got 0x{:08X}",
+                  user_revision, params.revision);
+        return false;
+    }
+
+    ClearError();
+    UpdateFlags(params.flags);
+
+    // TODO(ogniK): Check input params size when InfoUpdater is used
+
+    return true;
+}
+
+bool BehaviorInfo::UpdateOutput(std::vector<u8>& buffer, std::size_t offset) {
+    if (!CanConsumeBuffer(buffer.size(), offset, sizeof(OutParams))) {
+        LOG_ERROR(Audio, "Buffer is an invalid size!");
+        return false;
+    }
+
+    OutParams params{};
+    std::memcpy(params.errors.data(), errors.data(), sizeof(ErrorInfo) * errors.size());
+    params.error_count = static_cast<u32_le>(error_count);
+    std::memcpy(buffer.data() + offset, &params, sizeof(OutParams));
+    return true;
+}
+
+void BehaviorInfo::ClearError() {
+    error_count = 0;
+}
+
+void BehaviorInfo::UpdateFlags(u64_le dest_flags) {
+    flags = dest_flags;
+}
+
+void BehaviorInfo::SetUserRevision(u32_le revision) {
+    user_revision = revision;
+}
+
+bool BehaviorInfo::IsAdpcmLoopContextBugFixed() const {
+    return IsRevisionSupported(2, user_revision);
+}
+
+bool BehaviorInfo::IsSplitterSupported() const {
+    return IsRevisionSupported(2, user_revision);
+}
+
+bool BehaviorInfo::IsLongSizePreDelaySupported() const {
+    return IsRevisionSupported(3, user_revision);
+}
+
+bool BehaviorInfo::IsAudioRenererProcessingTimeLimit80PercentSupported() const {
+    return IsRevisionSupported(5, user_revision);
+}
+
+bool BehaviorInfo::IsAudioRenererProcessingTimeLimit75PercentSupported() const {
+    return IsRevisionSupported(4, user_revision);
+}
+
+bool BehaviorInfo::IsAudioRenererProcessingTimeLimit70PercentSupported() const {
+    return IsRevisionSupported(1, user_revision);
+}
+
+bool BehaviorInfo::IsElapsedFrameCountSupported() const {
+    return IsRevisionSupported(5, user_revision);
+}
+
+bool BehaviorInfo::IsMemoryPoolForceMappingEnabled() const {
+    return (flags & 1) != 0;
+}
+
+} // namespace AudioCore

src/audio_core/behavior_info.h

@@ -0,0 +1,66 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <array>
+
+#include <vector>
+#include "common/common_funcs.h"
+#include "common/common_types.h"
+#include "common/swap.h"
+
+namespace AudioCore {
+class BehaviorInfo {
+public:
+    explicit BehaviorInfo();
+    ~BehaviorInfo();
+
+    bool UpdateInput(const std::vector<u8>& buffer, std::size_t offset);
+    bool UpdateOutput(std::vector<u8>& buffer, std::size_t offset);
+
+    void ClearError();
+    void UpdateFlags(u64_le dest_flags);
+    void SetUserRevision(u32_le revision);
+
+    bool IsAdpcmLoopContextBugFixed() const;
+    bool IsSplitterSupported() const;
+    bool IsLongSizePreDelaySupported() const;
+    bool IsAudioRenererProcessingTimeLimit80PercentSupported() const;
+    bool IsAudioRenererProcessingTimeLimit75PercentSupported() const;
+    bool IsAudioRenererProcessingTimeLimit70PercentSupported() const;
+    bool IsElapsedFrameCountSupported() const;
+    bool IsMemoryPoolForceMappingEnabled() const;
+
+private:
+    u32_le process_revision{};
+    u32_le user_revision{};
+    u64_le flags{};
+
+    struct ErrorInfo {
+        u32_le result{};
+        INSERT_PADDING_WORDS(1);
+        u64_le result_info{};
+    };
+    static_assert(sizeof(ErrorInfo) == 0x10, "ErrorInfo is an invalid size");
+
+    std::array<ErrorInfo, 10> errors{};
+    std::size_t error_count{};
+
+    struct InParams {
+        u32_le revision{};
+        u32_le padding{};
+        u64_le flags{};
+    };
+    static_assert(sizeof(InParams) == 0x10, "InParams is an invalid size");
+
+    struct OutParams {
+        std::array<ErrorInfo, 10> errors{};
+        u32_le error_count{};
+        INSERT_PADDING_BYTES(12);
+    };
+    static_assert(sizeof(OutParams) == 0xb0, "OutParams is an invalid size");
+};
+
+} // namespace AudioCore

src/audio_core/common.h

@@ -0,0 +1,47 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+#include "common/common_funcs.h"
+#include "common/common_types.h"
+#include "common/swap.h"
+#include "core/hle/result.h"
+
+namespace AudioCore {
+namespace Audren {
+constexpr ResultCode ERR_INVALID_PARAMETERS{ErrorModule::Audio, 41};
+}
+
+constexpr u32_le CURRENT_PROCESS_REVISION = Common::MakeMagic('R', 'E', 'V', '8');
+
+static constexpr u32 VersionFromRevision(u32_le rev) {
+    // "REV7" -> 7
+    return ((rev >> 24) & 0xff) - 0x30;
+}
+
+static constexpr bool IsRevisionSupported(u32 required, u32_le user_revision) {
+    const auto base = VersionFromRevision(user_revision);
+    return required <= base;
+}
+
+static constexpr bool IsValidRevision(u32_le revision) {
+    const auto base = VersionFromRevision(revision);
+    constexpr auto max_rev = VersionFromRevision(CURRENT_PROCESS_REVISION);
+    return base <= max_rev;
+}
+
+static constexpr bool CanConsumeBuffer(std::size_t size, std::size_t offset, std::size_t required) {
+    if (offset > size) {
+        return false;
+    }
+    if (size < required) {
+        return false;
+    }
+    if ((size - offset) < required) {
+        return false;
+    }
+    return true;
+}
+
+} // namespace AudioCore

src/audio_core/cubeb_sink.cpp

@@ -8,6 +8,7 @@
 #include "audio_core/cubeb_sink.h"
 #include "audio_core/stream.h"
 #include "audio_core/time_stretch.h"
+#include "common/assert.h"
 #include "common/logging/log.h"
 #include "common/ring_buffer.h"
 #include "core/settings.h"
@@ -65,12 +66,25 @@ public:
     void EnqueueSamples(u32 source_num_channels, const std::vector<s16>& samples) override {
         if (source_num_channels > num_channels) {
             // Downsample 6 channels to 2
+            ASSERT_MSG(source_num_channels == 6, "Channel count must be 6");
+
             std::vector<s16> buf;
             buf.reserve(samples.size() * num_channels / source_num_channels);
             for (std::size_t i = 0; i < samples.size(); i += source_num_channels) {
-                for (std::size_t ch = 0; ch < num_channels; ch++) {
-                    buf.push_back(samples[i + ch]);
-                }
+                // Downmixing implementation taken from the ATSC standard
+                const s16 left{samples[i + 0]};
+                const s16 right{samples[i + 1]};
+                const s16 center{samples[i + 2]};
+                const s16 surround_left{samples[i + 4]};
+                const s16 surround_right{samples[i + 5]};
+                // Not used in the ATSC reference implementation
+                [[maybe_unused]] const s16 low_frequency_effects { samples[i + 3] };
+
+                constexpr s32 clev{707}; // center mixing level coefficient
+                constexpr s32 slev{707}; // surround mixing level coefficient
+
+                buf.push_back(left + (clev * center / 1000) + (slev * surround_left / 1000));
+                buf.push_back(right + (clev * center / 1000) + (slev * surround_right / 1000));
             }
             queue.Push(buf);
             return;

src/common/CMakeLists.txt

@@ -38,8 +38,6 @@ add_custom_command(OUTPUT scm_rev.cpp
       "${VIDEO_CORE}/renderer_opengl/gl_shader_decompiler.h"
       "${VIDEO_CORE}/renderer_opengl/gl_shader_disk_cache.cpp"
       "${VIDEO_CORE}/renderer_opengl/gl_shader_disk_cache.h"
-      "${VIDEO_CORE}/renderer_opengl/gl_shader_gen.cpp"
-      "${VIDEO_CORE}/renderer_opengl/gl_shader_gen.h"
       "${VIDEO_CORE}/shader/decode/arithmetic.cpp"
       "${VIDEO_CORE}/shader/decode/arithmetic_half.cpp"
       "${VIDEO_CORE}/shader/decode/arithmetic_half_immediate.cpp"
@@ -72,8 +70,6 @@ add_custom_command(OUTPUT scm_rev.cpp
       "${VIDEO_CORE}/shader/ast.h"
       "${VIDEO_CORE}/shader/compiler_settings.cpp"
       "${VIDEO_CORE}/shader/compiler_settings.h"
-      "${VIDEO_CORE}/shader/const_buffer_locker.cpp"
-      "${VIDEO_CORE}/shader/const_buffer_locker.h"
       "${VIDEO_CORE}/shader/control_flow.cpp"
       "${VIDEO_CORE}/shader/control_flow.h"
       "${VIDEO_CORE}/shader/decode.cpp"
@@ -82,9 +78,13 @@ add_custom_command(OUTPUT scm_rev.cpp
       "${VIDEO_CORE}/shader/node.h"
       "${VIDEO_CORE}/shader/node_helper.cpp"
       "${VIDEO_CORE}/shader/node_helper.h"
+      "${VIDEO_CORE}/shader/registry.cpp"
+      "${VIDEO_CORE}/shader/registry.h"
       "${VIDEO_CORE}/shader/shader_ir.cpp"
       "${VIDEO_CORE}/shader/shader_ir.h"
       "${VIDEO_CORE}/shader/track.cpp"
+      "${VIDEO_CORE}/shader/transform_feedback.cpp"
+      "${VIDEO_CORE}/shader/transform_feedback.h"
       # and also check that the scm_rev files haven't changed
       "${CMAKE_CURRENT_SOURCE_DIR}/scm_rev.cpp.in"
       "${CMAKE_CURRENT_SOURCE_DIR}/scm_rev.h"
@@ -106,6 +106,8 @@ add_library(common STATIC
     common_funcs.h
     common_paths.h
     common_types.h
+    dynamic_library.cpp
+    dynamic_library.h
     file_util.cpp
     file_util.h
     hash.h
@@ -153,6 +155,8 @@ add_library(common STATIC
     uuid.cpp
     uuid.h
     vector_math.h
+    virtual_buffer.cpp
+    virtual_buffer.h
     web_result.h
     zstd_compression.cpp
     zstd_compression.h

src/common/alignment.h

@@ -38,6 +38,13 @@ constexpr bool IsWordAligned(T value) {
     return (value & 0b11) == 0;
 }
 
+template <typename T>
+constexpr bool IsAligned(T value, std::size_t alignment) {
+    using U = typename std::make_unsigned<T>::type;
+    const U mask = static_cast<U>(alignment - 1);
+    return (value & mask) == 0;
+}
+
 template <typename T, std::size_t Align = 16>
 class AlignmentAllocator {
 public:

src/common/bit_field.h

@@ -180,7 +180,7 @@ public:
     }
 
     constexpr void Assign(const T& value) {
-        storage = (static_cast<StorageType>(storage) & ~mask) | FormatValue(value);
+        storage = static_cast<StorageType>((storage & ~mask) | FormatValue(value));
     }
 
     constexpr T Value() const {

src/common/common_funcs.h

@@ -55,6 +55,38 @@ __declspec(dllimport) void __stdcall DebugBreak(void);
 // Defined in Misc.cpp.
 std::string GetLastErrorMsg();
 
+#define DECLARE_ENUM_FLAG_OPERATORS(type)                                                          \
+    constexpr type operator|(type a, type b) noexcept {                                            \
+        using T = std::underlying_type_t<type>;                                                    \
+        return static_cast<type>(static_cast<T>(a) | static_cast<T>(b));                           \
+    }                                                                                              \
+    constexpr type operator&(type a, type b) noexcept {                                            \
+        using T = std::underlying_type_t<type>;                                                    \
+        return static_cast<type>(static_cast<T>(a) & static_cast<T>(b));                           \
+    }                                                                                              \
+    constexpr type& operator|=(type& a, type b) noexcept {                                         \
+        using T = std::underlying_type_t<type>;                                                    \
+        a = static_cast<type>(static_cast<T>(a) | static_cast<T>(b));                              \
+        return a;                                                                                  \
+    }                                                                                              \
+    constexpr type& operator&=(type& a, type b) noexcept {                                         \
+        using T = std::underlying_type_t<type>;                                                    \
+        a = static_cast<type>(static_cast<T>(a) & static_cast<T>(b));                              \
+        return a;                                                                                  \
+    }                                                                                              \
+    constexpr type operator~(type key) noexcept {                                                  \
+        using T = std::underlying_type_t<type>;                                                    \
+        return static_cast<type>(~static_cast<T>(key));                                            \
+    }                                                                                              \
+    constexpr bool True(type key) noexcept {                                                       \
+        using T = std::underlying_type_t<type>;                                                    \
+        return static_cast<T>(key) != 0;                                                           \
+    }                                                                                              \
+    constexpr bool False(type key) noexcept {                                                      \
+        using T = std::underlying_type_t<type>;                                                    \
+        return static_cast<T>(key) == 0;                                                           \
+    }
+
 namespace Common {
 
 constexpr u32 MakeMagic(char a, char b, char c, char d) {

src/common/dynamic_library.cpp

@@ -0,0 +1,106 @@
+// Copyright 2019 Dolphin Emulator Project
+// Licensed under GPLv2+
+// Refer to the license.txt file included.
+
+#include <cstring>
+#include <string>
+#include <utility>
+
+#include <fmt/format.h>
+
+#include "common/dynamic_library.h"
+
+#ifdef _WIN32
+#include <windows.h>
+#else
+#include <dlfcn.h>
+#endif
+
+namespace Common {
+
+DynamicLibrary::DynamicLibrary() = default;
+
+DynamicLibrary::DynamicLibrary(const char* filename) {
+    Open(filename);
+}
+
+DynamicLibrary::DynamicLibrary(DynamicLibrary&& rhs) noexcept
+    : handle{std::exchange(rhs.handle, nullptr)} {}
+
+DynamicLibrary& DynamicLibrary::operator=(DynamicLibrary&& rhs) noexcept {
+    Close();
+    handle = std::exchange(rhs.handle, nullptr);
+    return *this;
+}
+
+DynamicLibrary::~DynamicLibrary() {
+    Close();
+}
+
+std::string DynamicLibrary::GetUnprefixedFilename(const char* filename) {
+#if defined(_WIN32)
+    return std::string(filename) + ".dll";
+#elif defined(__APPLE__)
+    return std::string(filename) + ".dylib";
+#else
+    return std::string(filename) + ".so";
+#endif
+}
+
+std::string DynamicLibrary::GetVersionedFilename(const char* libname, int major, int minor) {
+#if defined(_WIN32)
+    if (major >= 0 && minor >= 0)
+        return fmt::format("{}-{}-{}.dll", libname, major, minor);
+    else if (major >= 0)
+        return fmt::format("{}-{}.dll", libname, major);
+    else
+        return fmt::format("{}.dll", libname);
+#elif defined(__APPLE__)
+    const char* prefix = std::strncmp(libname, "lib", 3) ? "lib" : "";
+    if (major >= 0 && minor >= 0)
+        return fmt::format("{}{}.{}.{}.dylib", prefix, libname, major, minor);
+    else if (major >= 0)
+        return fmt::format("{}{}.{}.dylib", prefix, libname, major);
+    else
+        return fmt::format("{}{}.dylib", prefix, libname);
+#else
+    const char* prefix = std::strncmp(libname, "lib", 3) ? "lib" : "";
+    if (major >= 0 && minor >= 0)
+        return fmt::format("{}{}.so.{}.{}", prefix, libname, major, minor);
+    else if (major >= 0)
+        return fmt::format("{}{}.so.{}", prefix, libname, major);
+    else
+        return fmt::format("{}{}.so", prefix, libname);
+#endif
+}
+
+bool DynamicLibrary::Open(const char* filename) {
+#ifdef _WIN32
+    handle = reinterpret_cast<void*>(LoadLibraryA(filename));
+#else
+    handle = dlopen(filename, RTLD_NOW);
+#endif
+    return handle != nullptr;
+}
+
+void DynamicLibrary::Close() {
+    if (!IsOpen())
+        return;
+
+#ifdef _WIN32
+    FreeLibrary(reinterpret_cast<HMODULE>(handle));
+#else
+    dlclose(handle);
+#endif
+    handle = nullptr;
+}
+
+void* DynamicLibrary::GetSymbolAddress(const char* name) const {
+#ifdef _WIN32
+    return reinterpret_cast<void*>(GetProcAddress(reinterpret_cast<HMODULE>(handle), name));
+#else
+    return reinterpret_cast<void*>(dlsym(handle, name));
+#endif
+}
+
+} // namespace Common

src/common/dynamic_library.h

@@ -0,0 +1,75 @@
+// Copyright 2019 Dolphin Emulator Project
+// Licensed under GPLv2+
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <string>
+
+namespace Common {
+
+/**
+ * Provides a platform-independent interface for loading a dynamic library and retrieving symbols.
+ * The interface maintains an internal reference count to allow one handle to be shared between
+ * multiple users.
+ */
+class DynamicLibrary final {
+public:
+    /// Default constructor, does not load a library.
+    explicit DynamicLibrary();
+
+    /// Automatically loads the specified library. Call IsOpen() to check validity before use.
+    explicit DynamicLibrary(const char* filename);
+
+    /// Moves the library.
+    DynamicLibrary(DynamicLibrary&&) noexcept;
+    DynamicLibrary& operator=(DynamicLibrary&&) noexcept;
+
+    /// Delete copies, we can't copy a dynamic library.
+    DynamicLibrary(const DynamicLibrary&) = delete;
+    DynamicLibrary& operator=(const DynamicLibrary&) = delete;
+
+    /// Closes the library.
+    ~DynamicLibrary();
+
+    /// Returns the specified library name with the platform-specific suffix added.
+    static std::string GetUnprefixedFilename(const char* filename);
+
+    /// Returns the specified library name in platform-specific format.
+    /// Major/minor versions will not be included if set to -1.
+    /// If libname already contains the "lib" prefix, it will not be added again.
+    /// Windows: LIBNAME-MAJOR-MINOR.dll
+    /// Linux: libLIBNAME.so.MAJOR.MINOR
+    /// Mac: libLIBNAME.MAJOR.MINOR.dylib
+    static std::string GetVersionedFilename(const char* libname, int major = -1, int minor = -1);
+
+    /// Returns true if a module is loaded, otherwise false.
+    bool IsOpen() const {
+        return handle != nullptr;
+    }
+
+    /// Loads (or replaces) the handle with the specified library file name.
+    /// Returns true if the library was loaded and can be used.
+    bool Open(const char* filename);
+
+    /// Unloads the library, any function pointers from this library are no longer valid.
+    void Close();
+
+    /// Returns the address of the specified symbol (function or variable) as an untyped pointer.
+    /// If the specified symbol does not exist in this library, nullptr is returned.
+    void* GetSymbolAddress(const char* name) const;
+
+    /// Obtains the address of the specified symbol, automatically casting to the correct type.
+    /// Returns true if the symbol was found and assigned, otherwise false.
+    template <typename T>
+    bool GetSymbol(const char* name, T* ptr) const {
+        *ptr = reinterpret_cast<T>(GetSymbolAddress(name));
+        return *ptr != nullptr;
+    }
+
+private:
+    /// Platform-dependent data type representing a dynamic library handle.
+    void* handle = nullptr;
+};
+
+} // namespace Common

src/common/file_util.cpp

@@ -3,6 +3,7 @@
 // Refer to the license.txt file included.
 
 #include <array>
+#include <limits>
 #include <memory>
 #include <sstream>
 #include <unordered_map>
@@ -530,11 +531,11 @@ void CopyDir(const std::string& source_path, const std::string& dest_path) {
 std::optional<std::string> GetCurrentDir() {
 // Get the current working directory (getcwd uses malloc)
 #ifdef _WIN32
-    wchar_t* dir;
-    if (!(dir = _wgetcwd(nullptr, 0))) {
+    wchar_t* dir = _wgetcwd(nullptr, 0);
+    if (!dir) {
 #else
-    char* dir;
-    if (!(dir = getcwd(nullptr, 0))) {
+    char* dir = getcwd(nullptr, 0);
+    if (!dir) {
 #endif
         LOG_ERROR(Common_Filesystem, "GetCurrentDirectory failed: {}", GetLastErrorMsg());
         return {};
@@ -887,7 +888,14 @@ std::string SanitizePath(std::string_view path_, DirectorySeparator directory_se
     }
 
     std::replace(path.begin(), path.end(), type1, type2);
-    path.erase(std::unique(path.begin(), path.end(),
+
+    auto start = path.begin();
+#ifdef _WIN32
+    // allow network paths which start with a double backslash (e.g. \\server\share)
+    if (start != path.end())
+        ++start;
+#endif
+    path.erase(std::unique(start, path.end(),
                            [type2](char c1, char c2) { return c1 == type2 && c2 == type2; }),
                path.end());
     return std::string(RemoveTrailingSlash(path));
@@ -918,19 +926,22 @@ void IOFile::Swap(IOFile& other) noexcept {
 
 bool IOFile::Open(const std::string& filename, const char openmode[], int flags) {
     Close();
+    bool m_good;
 #ifdef _WIN32
     if (flags != 0) {
         m_file = _wfsopen(Common::UTF8ToUTF16W(filename).c_str(),
                           Common::UTF8ToUTF16W(openmode).c_str(), flags);
+        m_good = m_file != nullptr;
     } else {
-        _wfopen_s(&m_file, Common::UTF8ToUTF16W(filename).c_str(),
-                  Common::UTF8ToUTF16W(openmode).c_str());
+        m_good = _wfopen_s(&m_file, Common::UTF8ToUTF16W(filename).c_str(),
+                           Common::UTF8ToUTF16W(openmode).c_str()) == 0;
     }
 #else
-    m_file = fopen(filename.c_str(), openmode);
+    m_file = std::fopen(filename.c_str(), openmode);
+    m_good = m_file != nullptr;
 #endif
 
-    return IsOpen();
+    return m_good;
 }
 
 bool IOFile::Close() {
@@ -956,13 +967,41 @@ u64 IOFile::Tell() const {
     if (IsOpen())
         return ftello(m_file);
 
-    return -1;
+    return std::numeric_limits<u64>::max();
 }
 
 bool IOFile::Flush() {
     return IsOpen() && 0 == std::fflush(m_file);
 }
 
+std::size_t IOFile::ReadImpl(void* data, std::size_t length, std::size_t data_size) const {
+    if (!IsOpen()) {
+        return std::numeric_limits<std::size_t>::max();
+    }
+
+    if (length == 0) {
+        return 0;
+    }
+
+    DEBUG_ASSERT(data != nullptr);
+
+    return std::fread(data, data_size, length, m_file);
+}
+
+std::size_t IOFile::WriteImpl(const void* data, std::size_t length, std::size_t data_size) {
+    if (!IsOpen()) {
+        return std::numeric_limits<std::size_t>::max();
+    }
+
+    if (length == 0) {
+        return 0;
+    }
+
+    DEBUG_ASSERT(data != nullptr);
+
+    return std::fwrite(data, data_size, length, m_file);
+}
+
 bool IOFile::Resize(u64 size) {
     return IsOpen() && 0 ==
 #ifdef _WIN32

src/common/file_util.h

@@ -222,22 +222,15 @@ public:
         static_assert(std::is_trivially_copyable_v<T>,
                       "Given array does not consist of trivially copyable objects");
 
-        if (!IsOpen()) {
-            return std::numeric_limits<std::size_t>::max();
-        }
-
-        return std::fread(data, sizeof(T), length, m_file);
+        return ReadImpl(data, length, sizeof(T));
     }
 
     template <typename T>
     std::size_t WriteArray(const T* data, std::size_t length) {
         static_assert(std::is_trivially_copyable_v<T>,
                       "Given array does not consist of trivially copyable objects");
-        if (!IsOpen()) {
-            return std::numeric_limits<std::size_t>::max();
-        }
 
-        return std::fwrite(data, sizeof(T), length, m_file);
+        return WriteImpl(data, length, sizeof(T));
     }
 
     template <typename T>
@@ -278,6 +271,9 @@ public:
     }
 
 private:
+    std::size_t ReadImpl(void* data, std::size_t length, std::size_t data_size) const;
+    std::size_t WriteImpl(const void* data, std::size_t length, std::size_t data_size);
+
     std::FILE* m_file = nullptr;
 };
 

src/common/math_util.h

@@ -24,17 +24,29 @@ struct Rectangle {
         : left(left), top(top), right(right), bottom(bottom) {}
 
     T GetWidth() const {
-        return std::abs(static_cast<std::make_signed_t<T>>(right - left));
+        if constexpr (std::is_floating_point_v<T>) {
+            return std::abs(right - left);
+        } else {
+            return std::abs(static_cast<std::make_signed_t<T>>(right - left));
+        }
     }
+
     T GetHeight() const {
-        return std::abs(static_cast<std::make_signed_t<T>>(bottom - top));
+        if constexpr (std::is_floating_point_v<T>) {
+            return std::abs(bottom - top);
+        } else {
+            return std::abs(static_cast<std::make_signed_t<T>>(bottom - top));
+        }
     }
+
     Rectangle<T> TranslateX(const T x) const {
         return Rectangle{left + x, top, right + x, bottom};
     }
+
     Rectangle<T> TranslateY(const T y) const {
         return Rectangle{left, top + y, right, bottom + y};
     }
+
     Rectangle<T> Scale(const float s) const {
         return Rectangle{left, top, static_cast<T>(left + GetWidth() * s),
                          static_cast<T>(top + GetHeight() * s)};

src/common/page_table.cpp

@@ -6,26 +6,20 @@
 
 namespace Common {
 
-PageTable::PageTable(std::size_t page_size_in_bits) : page_size_in_bits{page_size_in_bits} {}
+PageTable::PageTable() = default;
 
 PageTable::~PageTable() = default;
 
-void PageTable::Resize(std::size_t address_space_width_in_bits) {
-    const std::size_t num_page_table_entries = 1ULL
-                                               << (address_space_width_in_bits - page_size_in_bits);
-
+void PageTable::Resize(std::size_t address_space_width_in_bits, std::size_t page_size_in_bits,
+                       bool has_attribute) {
+    const std::size_t num_page_table_entries{1ULL
+                                             << (address_space_width_in_bits - page_size_in_bits)};
     pointers.resize(num_page_table_entries);
-    attributes.resize(num_page_table_entries);
     backing_addr.resize(num_page_table_entries);
 
-    // The default is a 39-bit address space, which causes an initial 1GB allocation size. If the
-    // vector size is subsequently decreased (via resize), the vector might not automatically
-    // actually reallocate/resize its underlying allocation, which wastes up to ~800 MB for
-    // 36-bit titles. Call shrink_to_fit to reduce capacity to what's actually in use.
-
-    pointers.shrink_to_fit();
-    attributes.shrink_to_fit();
-    backing_addr.shrink_to_fit();
+    if (has_attribute) {
+        attributes.resize(num_page_table_entries);
+    }
 }
 
 } // namespace Common

src/common/page_table.h

@@ -5,9 +5,12 @@
 #pragma once
 
 #include <vector>
+
 #include <boost/icl/interval_map.hpp>
+
 #include "common/common_types.h"
 #include "common/memory_hook.h"
+#include "common/virtual_buffer.h"
 
 namespace Common {
 
@@ -47,7 +50,7 @@ struct SpecialRegion {
  * mimics the way a real CPU page table works.
  */
 struct PageTable {
-    explicit PageTable(std::size_t page_size_in_bits);
+    PageTable();
     ~PageTable();
 
     /**
@@ -56,29 +59,18 @@ struct PageTable {
      *
      * @param address_space_width_in_bits The address size width in bits.
      */
-    void Resize(std::size_t address_space_width_in_bits);
+    void Resize(std::size_t address_space_width_in_bits, std::size_t page_size_in_bits,
+                bool has_attribute);
 
     /**
      * Vector of memory pointers backing each page. An entry can only be non-null if the
      * corresponding entry in the `attributes` vector is of type `Memory`.
      */
-    std::vector<u8*> pointers;
+    VirtualBuffer<u8*> pointers;
 
-    /**
-     * Contains MMIO handlers that back memory regions whose entries in the `attribute` vector is
-     * of type `Special`.
-     */
-    boost::icl::interval_map<u64, std::set<SpecialRegion>> special_regions;
+    VirtualBuffer<u64> backing_addr;
 
-    /**
-     * Vector of fine grained page attributes. If it is set to any value other than `Memory`, then
-     * the corresponding entry in `pointers` MUST be set to null.
-     */
-    std::vector<PageType> attributes;
-
-    std::vector<u64> backing_addr;
-
-    const std::size_t page_size_in_bits{};
+    VirtualBuffer<PageType> attributes;
 };
 
 } // namespace Common

src/common/scope_exit.h

@@ -12,10 +12,17 @@ template <typename Func>
 struct ScopeExitHelper {
     explicit ScopeExitHelper(Func&& func) : func(std::move(func)) {}
     ~ScopeExitHelper() {
-        func();
+        if (active) {
+            func();
+        }
+    }
+
+    void Cancel() {
+        active = false;
     }
 
     Func func;
+    bool active{true};
 };
 
 template <typename Func>

src/common/thread.cpp

@@ -28,11 +28,8 @@ namespace Common {
 #ifdef _MSC_VER
 
 // Sets the debugger-visible name of the current thread.
-// Uses undocumented (actually, it is now documented) trick.
-// http://msdn.microsoft.com/library/default.asp?url=/library/en-us/vsdebug/html/vxtsksettingthreadname.asp
-
-// This is implemented much nicer in upcoming msvc++, see:
-// http://msdn.microsoft.com/en-us/library/xcb2z8hs(VS.100).aspx
+// Uses trick documented in:
+// https://docs.microsoft.com/en-us/visualstudio/debugger/how-to-set-a-thread-name-in-native-code
 void SetCurrentThreadName(const char* name) {
     static const DWORD MS_VC_EXCEPTION = 0x406D1388;
 
@@ -47,7 +44,7 @@ void SetCurrentThreadName(const char* name) {
 
     info.dwType = 0x1000;
     info.szName = name;
-    info.dwThreadID = -1; // dwThreadID;
+    info.dwThreadID = std::numeric_limits<DWORD>::max();
     info.dwFlags = 0;
 
     __try {

src/common/virtual_buffer.cpp

@@ -0,0 +1,52 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#ifdef _WIN32
+#include <windows.h>
+#else
+#include <stdio.h>
+#include <sys/mman.h>
+#include <sys/types.h>
+#if defined __APPLE__ || defined __FreeBSD__ || defined __OpenBSD__
+#include <sys/sysctl.h>
+#elif defined __HAIKU__
+#include <OS.h>
+#else
+#include <sys/sysinfo.h>
+#endif
+#endif
+
+#include "common/assert.h"
+#include "common/virtual_buffer.h"
+
+namespace Common {
+
+void* AllocateMemoryPages(std::size_t size) {
+#ifdef _WIN32
+    void* base{VirtualAlloc(nullptr, size, MEM_COMMIT, PAGE_READWRITE)};
+#else
+    void* base{mmap(nullptr, size, PROT_READ | PROT_WRITE, MAP_ANON | MAP_PRIVATE, -1, 0)};
+
+    if (base == MAP_FAILED) {
+        base = nullptr;
+    }
+#endif
+
+    ASSERT(base);
+
+    return base;
+}
+
+void FreeMemoryPages(void* base, std::size_t size) {
+    if (!base) {
+        return;
+    }
+#ifdef _WIN32
+    ASSERT(VirtualFree(base, 0, MEM_RELEASE));
+#else
+    ASSERT(munmap(base, size) == 0);
+#endif
+}
+
+} // namespace Common

src/common/virtual_buffer.h

@@ -0,0 +1,58 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "common/common_funcs.h"
+
+namespace Common {
+
+void* AllocateMemoryPages(std::size_t size);
+void FreeMemoryPages(void* base, std::size_t size);
+
+template <typename T>
+class VirtualBuffer final : NonCopyable {
+public:
+    constexpr VirtualBuffer() = default;
+    explicit VirtualBuffer(std::size_t count) : alloc_size{count * sizeof(T)} {
+        base_ptr = reinterpret_cast<T*>(AllocateMemoryPages(alloc_size));
+    }
+
+    ~VirtualBuffer() {
+        FreeMemoryPages(base_ptr, alloc_size);
+    }
+
+    void resize(std::size_t count) {
+        FreeMemoryPages(base_ptr, alloc_size);
+
+        alloc_size = count * sizeof(T);
+        base_ptr = reinterpret_cast<T*>(AllocateMemoryPages(alloc_size));
+    }
+
+    constexpr const T& operator[](std::size_t index) const {
+        return base_ptr[index];
+    }
+
+    constexpr T& operator[](std::size_t index) {
+        return base_ptr[index];
+    }
+
+    constexpr T* data() {
+        return base_ptr;
+    }
+
+    constexpr const T* data() const {
+        return base_ptr;
+    }
+
+    constexpr std::size_t size() const {
+        return alloc_size / sizeof(T);
+    }
+
+private:
+    std::size_t alloc_size{};
+    T* base_ptr{};
+};
+
+} // namespace Common

src/core/CMakeLists.txt

@@ -35,6 +35,8 @@ add_library(core STATIC
     crypto/ctr_encryption_layer.h
     crypto/xts_encryption_layer.cpp
     crypto/xts_encryption_layer.h
+    device_memory.cpp
+    device_memory.h
     file_sys/bis_factory.cpp
     file_sys/bis_factory.h
     file_sys/card_image.cpp
@@ -131,8 +133,6 @@ add_library(core STATIC
     frontend/framebuffer_layout.cpp
     frontend/framebuffer_layout.h
     frontend/input.h
-    frontend/scope_acquire_window_context.cpp
-    frontend/scope_acquire_window_context.h
     gdbstub/gdbstub.cpp
     gdbstub/gdbstub.h
     hardware_interrupt_manager.cpp
@@ -154,6 +154,23 @@ add_library(core STATIC
     hle/kernel/hle_ipc.h
     hle/kernel/kernel.cpp
     hle/kernel/kernel.h
+    hle/kernel/memory/address_space_info.cpp
+    hle/kernel/memory/address_space_info.h
+    hle/kernel/memory/memory_block.h
+    hle/kernel/memory/memory_block_manager.cpp
+    hle/kernel/memory/memory_block_manager.h
+    hle/kernel/memory/memory_layout.h
+    hle/kernel/memory/memory_manager.cpp
+    hle/kernel/memory/memory_manager.h
+    hle/kernel/memory/memory_types.h
+    hle/kernel/memory/page_linked_list.h
+    hle/kernel/memory/page_heap.cpp
+    hle/kernel/memory/page_heap.h
+    hle/kernel/memory/page_table.cpp
+    hle/kernel/memory/page_table.h
+    hle/kernel/memory/slab_heap.h
+    hle/kernel/memory/system_control.cpp
+    hle/kernel/memory/system_control.h
     hle/kernel/mutex.cpp
     hle/kernel/mutex.h
     hle/kernel/object.cpp
@@ -180,6 +197,7 @@ add_library(core STATIC
     hle/kernel/shared_memory.h
     hle/kernel/svc.cpp
     hle/kernel/svc.h
+    hle/kernel/svc_types.h
     hle/kernel/svc_wrap.h
     hle/kernel/synchronization_object.cpp
     hle/kernel/synchronization_object.h
@@ -187,10 +205,10 @@ add_library(core STATIC
     hle/kernel/synchronization.h
     hle/kernel/thread.cpp
     hle/kernel/thread.h
+    hle/kernel/time_manager.cpp
+    hle/kernel/time_manager.h
     hle/kernel/transfer_memory.cpp
     hle/kernel/transfer_memory.h
-    hle/kernel/vm_manager.cpp
-    hle/kernel/vm_manager.h
     hle/kernel/writable_event.cpp
     hle/kernel/writable_event.h
     hle/lock.cpp
@@ -285,6 +303,18 @@ add_library(core STATIC
     hle/service/btm/btm.h
     hle/service/caps/caps.cpp
     hle/service/caps/caps.h
+    hle/service/caps/caps_a.cpp
+    hle/service/caps/caps_a.h
+    hle/service/caps/caps_c.cpp
+    hle/service/caps/caps_c.h
+    hle/service/caps/caps_u.cpp
+    hle/service/caps/caps_u.h
+    hle/service/caps/caps_sc.cpp
+    hle/service/caps/caps_sc.h
+    hle/service/caps/caps_ss.cpp
+    hle/service/caps/caps_ss.h
+    hle/service/caps/caps_su.cpp
+    hle/service/caps/caps_su.h
     hle/service/erpt/erpt.cpp
     hle/service/erpt/erpt.h
     hle/service/es/es.cpp
@@ -579,11 +609,8 @@ target_link_libraries(core PUBLIC common PRIVATE audio_core video_core)
 target_link_libraries(core PUBLIC Boost::boost PRIVATE fmt json-headers mbedtls opus unicorn)
 
 if (YUZU_ENABLE_BOXCAT)
-    get_directory_property(OPENSSL_LIBS
-        DIRECTORY ${PROJECT_SOURCE_DIR}/externals/libressl
-        DEFINITION OPENSSL_LIBS)
-    target_compile_definitions(core PRIVATE -DCPPHTTPLIB_OPENSSL_SUPPORT -DYUZU_ENABLE_BOXCAT)
-    target_link_libraries(core PRIVATE httplib json-headers ${OPENSSL_LIBS} zip)
+    target_compile_definitions(core PRIVATE -DYUZU_ENABLE_BOXCAT)
+    target_link_libraries(core PRIVATE httplib json-headers zip)
 endif()
 
 if (ENABLE_WEB_SERVICE)
@@ -593,8 +620,12 @@ endif()
 
 if (ARCHITECTURE_x86_64)
     target_sources(core PRIVATE
-        arm/dynarmic/arm_dynarmic.cpp
-        arm/dynarmic/arm_dynarmic.h
+        arm/dynarmic/arm_dynarmic_32.cpp
+        arm/dynarmic/arm_dynarmic_32.h
+        arm/dynarmic/arm_dynarmic_64.cpp
+        arm/dynarmic/arm_dynarmic_64.h
+        arm/dynarmic/arm_dynarmic_cp15.cpp
+        arm/dynarmic/arm_dynarmic_cp15.h
     )
     target_link_libraries(core PRIVATE dynarmic)
 endif()

src/core/arm/arm_interface.cpp

@@ -60,7 +60,7 @@ static_assert(sizeof(ELFSymbol) == 0x18, "ELFSymbol has incorrect size.");
 
 using Symbols = std::vector<std::pair<ELFSymbol, std::string>>;
 
-Symbols GetSymbols(VAddr text_offset, Memory::Memory& memory) {
+Symbols GetSymbols(VAddr text_offset, Core::Memory::Memory& memory) {
     const auto mod_offset = text_offset + memory.Read32(text_offset + 4);
 
     if (mod_offset < text_offset || (mod_offset & 0b11) != 0 ||
@@ -123,7 +123,7 @@ Symbols GetSymbols(VAddr text_offset, Memory::Memory& memory) {
 std::optional<std::string> GetSymbolName(const Symbols& symbols, VAddr func_address) {
     const auto iter =
         std::find_if(symbols.begin(), symbols.end(), [func_address](const auto& pair) {
-            const auto& [symbol, name] = pair;
+            const auto& symbol = pair.first;
             const auto end_address = symbol.value + symbol.size;
             return func_address >= symbol.value && func_address < end_address;
         });
@@ -146,7 +146,7 @@ std::vector<ARM_Interface::BacktraceEntry> ARM_Interface::GetBacktrace() const {
     auto fp = GetReg(29);
     auto lr = GetReg(30);
     while (true) {
-        out.push_back({"", 0, lr, 0});
+        out.push_back({"", 0, lr, 0, ""});
         if (!fp) {
             break;
         }

src/core/arm/arm_interface.h

@@ -25,20 +25,33 @@ public:
     explicit ARM_Interface(System& system_) : system{system_} {}
     virtual ~ARM_Interface() = default;
 
-    struct ThreadContext {
-        std::array<u64, 31> cpu_registers;
-        u64 sp;
-        u64 pc;
-        u32 pstate;
-        std::array<u8, 4> padding;
-        std::array<u128, 32> vector_registers;
-        u32 fpcr;
-        u32 fpsr;
-        u64 tpidr;
+    struct ThreadContext32 {
+        std::array<u32, 16> cpu_registers{};
+        u32 cpsr{};
+        std::array<u8, 4> padding{};
+        std::array<u64, 32> fprs{};
+        u32 fpscr{};
+        u32 fpexc{};
+        u32 tpidr{};
+    };
+    // Internally within the kernel, it expects the AArch32 version of the
+    // thread context to be 344 bytes in size.
+    static_assert(sizeof(ThreadContext32) == 0x158);
+
+    struct ThreadContext64 {
+        std::array<u64, 31> cpu_registers{};
+        u64 sp{};
+        u64 pc{};
+        u32 pstate{};
+        std::array<u8, 4> padding{};
+        std::array<u128, 32> vector_registers{};
+        u32 fpcr{};
+        u32 fpsr{};
+        u64 tpidr{};
     };
     // Internally within the kernel, it expects the AArch64 version of the
     // thread context to be 800 bytes in size.
-    static_assert(sizeof(ThreadContext) == 0x320);
+    static_assert(sizeof(ThreadContext64) == 0x320);
 
     /// Runs the CPU until an event happens
     virtual void Run() = 0;
@@ -130,17 +143,10 @@ public:
      */
     virtual void SetTPIDR_EL0(u64 value) = 0;
 
-    /**
-     * Saves the current CPU context
-     * @param ctx Thread context to save
-     */
-    virtual void SaveContext(ThreadContext& ctx) = 0;
-
-    /**
-     * Loads a CPU context
-     * @param ctx Thread context to load
-     */
-    virtual void LoadContext(const ThreadContext& ctx) = 0;
+    virtual void SaveContext(ThreadContext32& ctx) = 0;
+    virtual void SaveContext(ThreadContext64& ctx) = 0;
+    virtual void LoadContext(const ThreadContext32& ctx) = 0;
+    virtual void LoadContext(const ThreadContext64& ctx) = 0;
 
     /// Clears the exclusive monitor's state.
     virtual void ClearExclusiveState() = 0;

src/core/arm/dynarmic/arm_dynarmic_32.cpp

@@ -0,0 +1,208 @@
+// Copyright 2020 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <cinttypes>
+#include <memory>
+#include <dynarmic/A32/a32.h>
+#include <dynarmic/A32/config.h>
+#include <dynarmic/A32/context.h>
+#include "common/microprofile.h"
+#include "core/arm/dynarmic/arm_dynarmic_32.h"
+#include "core/arm/dynarmic/arm_dynarmic_64.h"
+#include "core/arm/dynarmic/arm_dynarmic_cp15.h"
+#include "core/core.h"
+#include "core/core_manager.h"
+#include "core/core_timing.h"
+#include "core/hle/kernel/svc.h"
+#include "core/memory.h"
+
+namespace Core {
+
+class DynarmicCallbacks32 : public Dynarmic::A32::UserCallbacks {
+public:
+    explicit DynarmicCallbacks32(ARM_Dynarmic_32& parent) : parent(parent) {}
+
+    u8 MemoryRead8(u32 vaddr) override {
+        return parent.system.Memory().Read8(vaddr);
+    }
+    u16 MemoryRead16(u32 vaddr) override {
+        return parent.system.Memory().Read16(vaddr);
+    }
+    u32 MemoryRead32(u32 vaddr) override {
+        return parent.system.Memory().Read32(vaddr);
+    }
+    u64 MemoryRead64(u32 vaddr) override {
+        return parent.system.Memory().Read64(vaddr);
+    }
+
+    void MemoryWrite8(u32 vaddr, u8 value) override {
+        parent.system.Memory().Write8(vaddr, value);
+    }
+    void MemoryWrite16(u32 vaddr, u16 value) override {
+        parent.system.Memory().Write16(vaddr, value);
+    }
+    void MemoryWrite32(u32 vaddr, u32 value) override {
+        parent.system.Memory().Write32(vaddr, value);
+    }
+    void MemoryWrite64(u32 vaddr, u64 value) override {
+        parent.system.Memory().Write64(vaddr, value);
+    }
+
+    void InterpreterFallback(u32 pc, std::size_t num_instructions) override {
+        UNIMPLEMENTED();
+    }
+
+    void ExceptionRaised(u32 pc, Dynarmic::A32::Exception exception) override {
+        switch (exception) {
+        case Dynarmic::A32::Exception::UndefinedInstruction:
+        case Dynarmic::A32::Exception::UnpredictableInstruction:
+            break;
+        case Dynarmic::A32::Exception::Breakpoint:
+            break;
+        }
+        LOG_CRITICAL(HW_GPU, "ExceptionRaised(exception = {}, pc = {:08X}, code = {:08X})",
+                     static_cast<std::size_t>(exception), pc, MemoryReadCode(pc));
+        UNIMPLEMENTED();
+    }
+
+    void CallSVC(u32 swi) override {
+        Kernel::Svc::Call(parent.system, swi);
+    }
+
+    void AddTicks(u64 ticks) override {
+        // Divide the number of ticks by the amount of CPU cores. TODO(Subv): This yields only a
+        // rough approximation of the amount of executed ticks in the system, it may be thrown off
+        // if not all cores are doing a similar amount of work. Instead of doing this, we should
+        // device a way so that timing is consistent across all cores without increasing the ticks 4
+        // times.
+        u64 amortized_ticks = (ticks - num_interpreted_instructions) / Core::NUM_CPU_CORES;
+        // Always execute at least one tick.
+        amortized_ticks = std::max<u64>(amortized_ticks, 1);
+
+        parent.system.CoreTiming().AddTicks(amortized_ticks);
+        num_interpreted_instructions = 0;
+    }
+    u64 GetTicksRemaining() override {
+        return std::max(parent.system.CoreTiming().GetDowncount(), {});
+    }
+
+    ARM_Dynarmic_32& parent;
+    std::size_t num_interpreted_instructions{};
+    u64 tpidrro_el0{};
+    u64 tpidr_el0{};
+};
+
+std::shared_ptr<Dynarmic::A32::Jit> ARM_Dynarmic_32::MakeJit(Common::PageTable& page_table,
+                                                             std::size_t address_space_bits) const {
+    Dynarmic::A32::UserConfig config;
+    config.callbacks = cb.get();
+    // TODO(bunnei): Implement page table for 32-bit
+    // config.page_table = &page_table.pointers;
+    config.coprocessors[15] = std::make_shared<DynarmicCP15>((u32*)&CP15_regs[0]);
+    config.define_unpredictable_behaviour = true;
+    return std::make_unique<Dynarmic::A32::Jit>(config);
+}
+
+MICROPROFILE_DEFINE(ARM_Jit_Dynarmic_32, "ARM JIT", "Dynarmic", MP_RGB(255, 64, 64));
+
+void ARM_Dynarmic_32::Run() {
+    MICROPROFILE_SCOPE(ARM_Jit_Dynarmic_32);
+    jit->Run();
+}
+
+void ARM_Dynarmic_32::Step() {
+    cb->InterpreterFallback(jit->Regs()[15], 1);
+}
+
+ARM_Dynarmic_32::ARM_Dynarmic_32(System& system, ExclusiveMonitor& exclusive_monitor,
+                                 std::size_t core_index)
+    : ARM_Interface{system},
+      cb(std::make_unique<DynarmicCallbacks32>(*this)), core_index{core_index},
+      exclusive_monitor{dynamic_cast<DynarmicExclusiveMonitor&>(exclusive_monitor)} {}
+
+ARM_Dynarmic_32::~ARM_Dynarmic_32() = default;
+
+void ARM_Dynarmic_32::SetPC(u64 pc) {
+    jit->Regs()[15] = static_cast<u32>(pc);
+}
+
+u64 ARM_Dynarmic_32::GetPC() const {
+    return jit->Regs()[15];
+}
+
+u64 ARM_Dynarmic_32::GetReg(int index) const {
+    return jit->Regs()[index];
+}
+
+void ARM_Dynarmic_32::SetReg(int index, u64 value) {
+    jit->Regs()[index] = static_cast<u32>(value);
+}
+
+u128 ARM_Dynarmic_32::GetVectorReg(int index) const {
+    return {};
+}
+
+void ARM_Dynarmic_32::SetVectorReg(int index, u128 value) {}
+
+u32 ARM_Dynarmic_32::GetPSTATE() const {
+    return jit->Cpsr();
+}
+
+void ARM_Dynarmic_32::SetPSTATE(u32 cpsr) {
+    jit->SetCpsr(cpsr);
+}
+
+u64 ARM_Dynarmic_32::GetTlsAddress() const {
+    return CP15_regs[static_cast<std::size_t>(CP15Register::CP15_THREAD_URO)];
+}
+
+void ARM_Dynarmic_32::SetTlsAddress(VAddr address) {
+    CP15_regs[static_cast<std::size_t>(CP15Register::CP15_THREAD_URO)] = static_cast<u32>(address);
+}
+
+u64 ARM_Dynarmic_32::GetTPIDR_EL0() const {
+    return cb->tpidr_el0;
+}
+
+void ARM_Dynarmic_32::SetTPIDR_EL0(u64 value) {
+    cb->tpidr_el0 = value;
+}
+
+void ARM_Dynarmic_32::SaveContext(ThreadContext32& ctx) {
+    Dynarmic::A32::Context context;
+    jit->SaveContext(context);
+    ctx.cpu_registers = context.Regs();
+    ctx.cpsr = context.Cpsr();
+}
+
+void ARM_Dynarmic_32::LoadContext(const ThreadContext32& ctx) {
+    Dynarmic::A32::Context context;
+    context.Regs() = ctx.cpu_registers;
+    context.SetCpsr(ctx.cpsr);
+    jit->LoadContext(context);
+}
+
+void ARM_Dynarmic_32::PrepareReschedule() {
+    jit->HaltExecution();
+}
+
+void ARM_Dynarmic_32::ClearInstructionCache() {
+    jit->ClearCache();
+}
+
+void ARM_Dynarmic_32::ClearExclusiveState() {}
+
+void ARM_Dynarmic_32::PageTableChanged(Common::PageTable& page_table,
+                                       std::size_t new_address_space_size_in_bits) {
+    auto key = std::make_pair(&page_table, new_address_space_size_in_bits);
+    auto iter = jit_cache.find(key);
+    if (iter != jit_cache.end()) {
+        jit = iter->second;
+        return;
+    }
+    jit = MakeJit(page_table, new_address_space_size_in_bits);
+    jit_cache.emplace(key, jit);
+}
+
+} // namespace Core

src/core/arm/dynarmic/arm_dynarmic_32.h

@@ -0,0 +1,77 @@
+// Copyright 2020 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <memory>
+#include <unordered_map>
+
+#include <dynarmic/A32/a32.h>
+#include <dynarmic/A64/a64.h>
+#include <dynarmic/A64/exclusive_monitor.h>
+#include "common/common_types.h"
+#include "common/hash.h"
+#include "core/arm/arm_interface.h"
+#include "core/arm/exclusive_monitor.h"
+
+namespace Core::Memory {
+class Memory;
+}
+
+namespace Core {
+
+class DynarmicCallbacks32;
+class DynarmicExclusiveMonitor;
+class System;
+
+class ARM_Dynarmic_32 final : public ARM_Interface {
+public:
+    ARM_Dynarmic_32(System& system, ExclusiveMonitor& exclusive_monitor, std::size_t core_index);
+    ~ARM_Dynarmic_32() override;
+
+    void SetPC(u64 pc) override;
+    u64 GetPC() const override;
+    u64 GetReg(int index) const override;
+    void SetReg(int index, u64 value) override;
+    u128 GetVectorReg(int index) const override;
+    void SetVectorReg(int index, u128 value) override;
+    u32 GetPSTATE() const override;
+    void SetPSTATE(u32 pstate) override;
+    void Run() override;
+    void Step() override;
+    VAddr GetTlsAddress() const override;
+    void SetTlsAddress(VAddr address) override;
+    void SetTPIDR_EL0(u64 value) override;
+    u64 GetTPIDR_EL0() const override;
+
+    void SaveContext(ThreadContext32& ctx) override;
+    void SaveContext(ThreadContext64& ctx) override {}
+    void LoadContext(const ThreadContext32& ctx) override;
+    void LoadContext(const ThreadContext64& ctx) override {}
+
+    void PrepareReschedule() override;
+    void ClearExclusiveState() override;
+
+    void ClearInstructionCache() override;
+    void PageTableChanged(Common::PageTable& new_page_table,
+                          std::size_t new_address_space_size_in_bits) override;
+
+private:
+    std::shared_ptr<Dynarmic::A32::Jit> MakeJit(Common::PageTable& page_table,
+                                                std::size_t address_space_bits) const;
+
+    using JitCacheKey = std::pair<Common::PageTable*, std::size_t>;
+    using JitCacheType =
+        std::unordered_map<JitCacheKey, std::shared_ptr<Dynarmic::A32::Jit>, Common::PairHash>;
+
+    friend class DynarmicCallbacks32;
+    std::unique_ptr<DynarmicCallbacks32> cb;
+    JitCacheType jit_cache;
+    std::shared_ptr<Dynarmic::A32::Jit> jit;
+    std::size_t core_index;
+    DynarmicExclusiveMonitor& exclusive_monitor;
+    std::array<u32, 84> CP15_regs{};
+};
+
+} // namespace Core

src/core/arm/dynarmic/arm_dynarmic_64.cpp

@@ -8,7 +8,8 @@
 #include <dynarmic/A64/config.h>
 #include "common/logging/log.h"
 #include "common/microprofile.h"
-#include "core/arm/dynarmic/arm_dynarmic.h"
+#include "common/page_table.h"
+#include "core/arm/dynarmic/arm_dynarmic_64.h"
 #include "core/core.h"
 #include "core/core_manager.h"
 #include "core/core_timing.h"
@@ -18,16 +19,16 @@
 #include "core/hle/kernel/process.h"
 #include "core/hle/kernel/scheduler.h"
 #include "core/hle/kernel/svc.h"
-#include "core/hle/kernel/vm_manager.h"
 #include "core/memory.h"
+#include "core/settings.h"
 
 namespace Core {
 
 using Vector = Dynarmic::A64::Vector;
 
-class ARM_Dynarmic_Callbacks : public Dynarmic::A64::UserCallbacks {
+class DynarmicCallbacks64 : public Dynarmic::A64::UserCallbacks {
 public:
-    explicit ARM_Dynarmic_Callbacks(ARM_Dynarmic& parent) : parent(parent) {}
+    explicit DynarmicCallbacks64(ARM_Dynarmic_64& parent) : parent(parent) {}
 
     u8 MemoryRead8(u64 vaddr) override {
         return parent.system.Memory().Read8(vaddr);
@@ -68,7 +69,7 @@ public:
         LOG_INFO(Core_ARM, "Unicorn fallback @ 0x{:X} for {} instructions (instr = {:08X})", pc,
                  num_instructions, MemoryReadCode(pc));
 
-        ARM_Interface::ThreadContext ctx;
+        ARM_Interface::ThreadContext64 ctx;
         parent.SaveContext(ctx);
         parent.inner_unicorn.LoadContext(ctx);
         parent.inner_unicorn.ExecuteInstructions(num_instructions);
@@ -90,7 +91,7 @@ public:
                 parent.jit->HaltExecution();
                 parent.SetPC(pc);
                 Kernel::Thread* const thread = parent.system.CurrentScheduler().GetCurrentThread();
-                parent.SaveContext(thread->GetContext());
+                parent.SaveContext(thread->GetContext64());
                 GDBStub::Break();
                 GDBStub::SendTrap(thread, 5);
                 return;
@@ -103,7 +104,7 @@ public:
     }
 
     void CallSVC(u32 swi) override {
-        Kernel::CallSVC(parent.system, swi);
+        Kernel::Svc::Call(parent.system, swi);
     }
 
     void AddTicks(u64 ticks) override {
@@ -126,14 +127,14 @@ public:
         return Timing::CpuCyclesToClockCycles(parent.system.CoreTiming().GetTicks());
     }
 
-    ARM_Dynarmic& parent;
+    ARM_Dynarmic_64& parent;
     std::size_t num_interpreted_instructions = 0;
     u64 tpidrro_el0 = 0;
     u64 tpidr_el0 = 0;
 };
 
-std::unique_ptr<Dynarmic::A64::Jit> ARM_Dynarmic::MakeJit(Common::PageTable& page_table,
-                                                          std::size_t address_space_bits) const {
+std::shared_ptr<Dynarmic::A64::Jit> ARM_Dynarmic_64::MakeJit(Common::PageTable& page_table,
+                                                             std::size_t address_space_bits) const {
     Dynarmic::A64::UserConfig config;
 
     // Callbacks
@@ -144,6 +145,8 @@ std::unique_ptr<Dynarmic::A64::Jit> ARM_Dynarmic::MakeJit(Common::PageTable& pag
     config.page_table_address_space_bits = address_space_bits;
     config.silently_mirror_page_table = false;
     config.absolute_offset_page_table = true;
+    config.detect_misaligned_access_via_page_table = 16 | 32 | 64 | 128;
+    config.only_detect_misalignment_via_page_table_on_page_boundary = true;
 
     // Multi-process state
     config.processor_id = core_index;
@@ -159,79 +162,85 @@ std::unique_ptr<Dynarmic::A64::Jit> ARM_Dynarmic::MakeJit(Common::PageTable& pag
     // Unpredictable instructions
     config.define_unpredictable_behaviour = true;
 
-    return std::make_unique<Dynarmic::A64::Jit>(config);
+    // Optimizations
+    if (Settings::values.disable_cpu_opt) {
+        config.enable_optimizations = false;
+        config.enable_fast_dispatch = false;
+    }
+
+    return std::make_shared<Dynarmic::A64::Jit>(config);
 }
 
-MICROPROFILE_DEFINE(ARM_Jit_Dynarmic, "ARM JIT", "Dynarmic", MP_RGB(255, 64, 64));
+MICROPROFILE_DEFINE(ARM_Jit_Dynarmic_64, "ARM JIT", "Dynarmic", MP_RGB(255, 64, 64));
 
-void ARM_Dynarmic::Run() {
-    MICROPROFILE_SCOPE(ARM_Jit_Dynarmic);
+void ARM_Dynarmic_64::Run() {
+    MICROPROFILE_SCOPE(ARM_Jit_Dynarmic_64);
 
     jit->Run();
 }
 
-void ARM_Dynarmic::Step() {
+void ARM_Dynarmic_64::Step() {
     cb->InterpreterFallback(jit->GetPC(), 1);
 }
 
-ARM_Dynarmic::ARM_Dynarmic(System& system, ExclusiveMonitor& exclusive_monitor,
-                           std::size_t core_index)
+ARM_Dynarmic_64::ARM_Dynarmic_64(System& system, ExclusiveMonitor& exclusive_monitor,
+                                 std::size_t core_index)
     : ARM_Interface{system},
-      cb(std::make_unique<ARM_Dynarmic_Callbacks>(*this)), inner_unicorn{system},
+      cb(std::make_unique<DynarmicCallbacks64>(*this)), inner_unicorn{system},
       core_index{core_index}, exclusive_monitor{
                                   dynamic_cast<DynarmicExclusiveMonitor&>(exclusive_monitor)} {}
 
-ARM_Dynarmic::~ARM_Dynarmic() = default;
+ARM_Dynarmic_64::~ARM_Dynarmic_64() = default;
 
-void ARM_Dynarmic::SetPC(u64 pc) {
+void ARM_Dynarmic_64::SetPC(u64 pc) {
     jit->SetPC(pc);
 }
 
-u64 ARM_Dynarmic::GetPC() const {
+u64 ARM_Dynarmic_64::GetPC() const {
     return jit->GetPC();
 }
 
-u64 ARM_Dynarmic::GetReg(int index) const {
+u64 ARM_Dynarmic_64::GetReg(int index) const {
     return jit->GetRegister(index);
 }
 
-void ARM_Dynarmic::SetReg(int index, u64 value) {
+void ARM_Dynarmic_64::SetReg(int index, u64 value) {
     jit->SetRegister(index, value);
 }
 
-u128 ARM_Dynarmic::GetVectorReg(int index) const {
+u128 ARM_Dynarmic_64::GetVectorReg(int index) const {
     return jit->GetVector(index);
 }
 
-void ARM_Dynarmic::SetVectorReg(int index, u128 value) {
+void ARM_Dynarmic_64::SetVectorReg(int index, u128 value) {
     jit->SetVector(index, value);
 }
 
-u32 ARM_Dynarmic::GetPSTATE() const {
+u32 ARM_Dynarmic_64::GetPSTATE() const {
     return jit->GetPstate();
 }
 
-void ARM_Dynarmic::SetPSTATE(u32 pstate) {
+void ARM_Dynarmic_64::SetPSTATE(u32 pstate) {
     jit->SetPstate(pstate);
 }
 
-u64 ARM_Dynarmic::GetTlsAddress() const {
+u64 ARM_Dynarmic_64::GetTlsAddress() const {
     return cb->tpidrro_el0;
 }
 
-void ARM_Dynarmic::SetTlsAddress(VAddr address) {
+void ARM_Dynarmic_64::SetTlsAddress(VAddr address) {
     cb->tpidrro_el0 = address;
 }
 
-u64 ARM_Dynarmic::GetTPIDR_EL0() const {
+u64 ARM_Dynarmic_64::GetTPIDR_EL0() const {
     return cb->tpidr_el0;
 }
 
-void ARM_Dynarmic::SetTPIDR_EL0(u64 value) {
+void ARM_Dynarmic_64::SetTPIDR_EL0(u64 value) {
     cb->tpidr_el0 = value;
 }
 
-void ARM_Dynarmic::SaveContext(ThreadContext& ctx) {
+void ARM_Dynarmic_64::SaveContext(ThreadContext64& ctx) {
     ctx.cpu_registers = jit->GetRegisters();
     ctx.sp = jit->GetSP();
     ctx.pc = jit->GetPC();
@@ -242,7 +251,7 @@ void ARM_Dynarmic::SaveContext(ThreadContext& ctx) {
     ctx.tpidr = cb->tpidr_el0;
 }
 
-void ARM_Dynarmic::LoadContext(const ThreadContext& ctx) {
+void ARM_Dynarmic_64::LoadContext(const ThreadContext64& ctx) {
     jit->SetRegisters(ctx.cpu_registers);
     jit->SetSP(ctx.sp);
     jit->SetPC(ctx.pc);
@@ -253,25 +262,32 @@ void ARM_Dynarmic::LoadContext(const ThreadContext& ctx) {
     SetTPIDR_EL0(ctx.tpidr);
 }
 
-void ARM_Dynarmic::PrepareReschedule() {
+void ARM_Dynarmic_64::PrepareReschedule() {
     jit->HaltExecution();
 }
 
-void ARM_Dynarmic::ClearInstructionCache() {
+void ARM_Dynarmic_64::ClearInstructionCache() {
     jit->ClearCache();
 }
 
-void ARM_Dynarmic::ClearExclusiveState() {
+void ARM_Dynarmic_64::ClearExclusiveState() {
     jit->ClearExclusiveState();
 }
 
-void ARM_Dynarmic::PageTableChanged(Common::PageTable& page_table,
-                                    std::size_t new_address_space_size_in_bits) {
+void ARM_Dynarmic_64::PageTableChanged(Common::PageTable& page_table,
+                                       std::size_t new_address_space_size_in_bits) {
+    auto key = std::make_pair(&page_table, new_address_space_size_in_bits);
+    auto iter = jit_cache.find(key);
+    if (iter != jit_cache.end()) {
+        jit = iter->second;
+        return;
+    }
     jit = MakeJit(page_table, new_address_space_size_in_bits);
+    jit_cache.emplace(key, jit);
 }
 
-DynarmicExclusiveMonitor::DynarmicExclusiveMonitor(Memory::Memory& memory_, std::size_t core_count)
-    : monitor(core_count), memory{memory_} {}
+DynarmicExclusiveMonitor::DynarmicExclusiveMonitor(Memory::Memory& memory, std::size_t core_count)
+    : monitor(core_count), memory{memory} {}
 
 DynarmicExclusiveMonitor::~DynarmicExclusiveMonitor() = default;
 

src/core/arm/dynarmic/arm_dynarmic_64.h

@@ -5,27 +5,30 @@
 #pragma once
 
 #include <memory>
+#include <unordered_map>
+
 #include <dynarmic/A64/a64.h>
 #include <dynarmic/A64/exclusive_monitor.h>
 #include "common/common_types.h"
+#include "common/hash.h"
 #include "core/arm/arm_interface.h"
 #include "core/arm/exclusive_monitor.h"
 #include "core/arm/unicorn/arm_unicorn.h"
 
-namespace Memory {
+namespace Core::Memory {
 class Memory;
 }
 
 namespace Core {
 
-class ARM_Dynarmic_Callbacks;
+class DynarmicCallbacks64;
 class DynarmicExclusiveMonitor;
 class System;
 
-class ARM_Dynarmic final : public ARM_Interface {
+class ARM_Dynarmic_64 final : public ARM_Interface {
 public:
-    ARM_Dynarmic(System& system, ExclusiveMonitor& exclusive_monitor, std::size_t core_index);
-    ~ARM_Dynarmic() override;
+    ARM_Dynarmic_64(System& system, ExclusiveMonitor& exclusive_monitor, std::size_t core_index);
+    ~ARM_Dynarmic_64() override;
 
     void SetPC(u64 pc) override;
     u64 GetPC() const override;
@@ -42,8 +45,10 @@ public:
     void SetTPIDR_EL0(u64 value) override;
     u64 GetTPIDR_EL0() const override;
 
-    void SaveContext(ThreadContext& ctx) override;
-    void LoadContext(const ThreadContext& ctx) override;
+    void SaveContext(ThreadContext32& ctx) override {}
+    void SaveContext(ThreadContext64& ctx) override;
+    void LoadContext(const ThreadContext32& ctx) override {}
+    void LoadContext(const ThreadContext64& ctx) override;
 
     void PrepareReschedule() override;
     void ClearExclusiveState() override;
@@ -53,12 +58,17 @@ public:
                           std::size_t new_address_space_size_in_bits) override;
 
 private:
-    std::unique_ptr<Dynarmic::A64::Jit> MakeJit(Common::PageTable& page_table,
+    std::shared_ptr<Dynarmic::A64::Jit> MakeJit(Common::PageTable& page_table,
                                                 std::size_t address_space_bits) const;
 
-    friend class ARM_Dynarmic_Callbacks;
-    std::unique_ptr<ARM_Dynarmic_Callbacks> cb;
-    std::unique_ptr<Dynarmic::A64::Jit> jit;
+    using JitCacheKey = std::pair<Common::PageTable*, std::size_t>;
+    using JitCacheType =
+        std::unordered_map<JitCacheKey, std::shared_ptr<Dynarmic::A64::Jit>, Common::PairHash>;
+
+    friend class DynarmicCallbacks64;
+    std::unique_ptr<DynarmicCallbacks64> cb;
+    JitCacheType jit_cache;
+    std::shared_ptr<Dynarmic::A64::Jit> jit;
     ARM_Unicorn inner_unicorn;
 
     std::size_t core_index;
@@ -67,7 +77,7 @@ private:
 
 class DynarmicExclusiveMonitor final : public ExclusiveMonitor {
 public:
-    explicit DynarmicExclusiveMonitor(Memory::Memory& memory_, std::size_t core_count);
+    explicit DynarmicExclusiveMonitor(Memory::Memory& memory, std::size_t core_count);
     ~DynarmicExclusiveMonitor() override;
 
     void SetExclusive(std::size_t core_index, VAddr addr) override;
@@ -80,9 +90,9 @@ public:
     bool ExclusiveWrite128(std::size_t core_index, VAddr vaddr, u128 value) override;
 
 private:
-    friend class ARM_Dynarmic;
+    friend class ARM_Dynarmic_64;
     Dynarmic::A64::ExclusiveMonitor monitor;
-    Memory::Memory& memory;
+    Core::Memory::Memory& memory;
 };
 
 } // namespace Core

src/core/arm/dynarmic/arm_dynarmic_cp15.cpp

@@ -0,0 +1,80 @@
+// Copyright 2017 Citra Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "core/arm/dynarmic/arm_dynarmic_cp15.h"
+
+using Callback = Dynarmic::A32::Coprocessor::Callback;
+using CallbackOrAccessOneWord = Dynarmic::A32::Coprocessor::CallbackOrAccessOneWord;
+using CallbackOrAccessTwoWords = Dynarmic::A32::Coprocessor::CallbackOrAccessTwoWords;
+
+std::optional<Callback> DynarmicCP15::CompileInternalOperation(bool two, unsigned opc1,
+                                                               CoprocReg CRd, CoprocReg CRn,
+                                                               CoprocReg CRm, unsigned opc2) {
+    return {};
+}
+
+CallbackOrAccessOneWord DynarmicCP15::CompileSendOneWord(bool two, unsigned opc1, CoprocReg CRn,
+                                                         CoprocReg CRm, unsigned opc2) {
+    // TODO(merry): Privileged CP15 registers
+
+    if (!two && CRn == CoprocReg::C7 && opc1 == 0 && CRm == CoprocReg::C5 && opc2 == 4) {
+        // This is a dummy write, we ignore the value written here.
+        return &CP15[static_cast<std::size_t>(CP15Register::CP15_FLUSH_PREFETCH_BUFFER)];
+    }
+
+    if (!two && CRn == CoprocReg::C7 && opc1 == 0 && CRm == CoprocReg::C10) {
+        switch (opc2) {
+        case 4:
+            // This is a dummy write, we ignore the value written here.
+            return &CP15[static_cast<std::size_t>(CP15Register::CP15_DATA_SYNC_BARRIER)];
+        case 5:
+            // This is a dummy write, we ignore the value written here.
+            return &CP15[static_cast<std::size_t>(CP15Register::CP15_DATA_MEMORY_BARRIER)];
+        default:
+            return {};
+        }
+    }
+
+    if (!two && CRn == CoprocReg::C13 && opc1 == 0 && CRm == CoprocReg::C0 && opc2 == 2) {
+        return &CP15[static_cast<std::size_t>(CP15Register::CP15_THREAD_UPRW)];
+    }
+
+    return {};
+}
+
+CallbackOrAccessTwoWords DynarmicCP15::CompileSendTwoWords(bool two, unsigned opc, CoprocReg CRm) {
+    return {};
+}
+
+CallbackOrAccessOneWord DynarmicCP15::CompileGetOneWord(bool two, unsigned opc1, CoprocReg CRn,
+                                                        CoprocReg CRm, unsigned opc2) {
+    // TODO(merry): Privileged CP15 registers
+
+    if (!two && CRn == CoprocReg::C13 && opc1 == 0 && CRm == CoprocReg::C0) {
+        switch (opc2) {
+        case 2:
+            return &CP15[static_cast<std::size_t>(CP15Register::CP15_THREAD_UPRW)];
+        case 3:
+            return &CP15[static_cast<std::size_t>(CP15Register::CP15_THREAD_URO)];
+        default:
+            return {};
+        }
+    }
+
+    return {};
+}
+
+CallbackOrAccessTwoWords DynarmicCP15::CompileGetTwoWords(bool two, unsigned opc, CoprocReg CRm) {
+    return {};
+}
+
+std::optional<Callback> DynarmicCP15::CompileLoadWords(bool two, bool long_transfer, CoprocReg CRd,
+                                                       std::optional<u8> option) {
+    return {};
+}
+
+std::optional<Callback> DynarmicCP15::CompileStoreWords(bool two, bool long_transfer, CoprocReg CRd,
+                                                        std::optional<u8> option) {
+    return {};
+}

src/core/arm/dynarmic/arm_dynarmic_cp15.h

@@ -0,0 +1,152 @@
+// Copyright 2017 Citra Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <memory>
+#include <optional>
+
+#include <dynarmic/A32/coprocessor.h>
+#include "common/common_types.h"
+
+enum class CP15Register {
+    // c0 - Information registers
+    CP15_MAIN_ID,
+    CP15_CACHE_TYPE,
+    CP15_TCM_STATUS,
+    CP15_TLB_TYPE,
+    CP15_CPU_ID,
+    CP15_PROCESSOR_FEATURE_0,
+    CP15_PROCESSOR_FEATURE_1,
+    CP15_DEBUG_FEATURE_0,
+    CP15_AUXILIARY_FEATURE_0,
+    CP15_MEMORY_MODEL_FEATURE_0,
+    CP15_MEMORY_MODEL_FEATURE_1,
+    CP15_MEMORY_MODEL_FEATURE_2,
+    CP15_MEMORY_MODEL_FEATURE_3,
+    CP15_ISA_FEATURE_0,
+    CP15_ISA_FEATURE_1,
+    CP15_ISA_FEATURE_2,
+    CP15_ISA_FEATURE_3,
+    CP15_ISA_FEATURE_4,
+
+    // c1 - Control registers
+    CP15_CONTROL,
+    CP15_AUXILIARY_CONTROL,
+    CP15_COPROCESSOR_ACCESS_CONTROL,
+
+    // c2 - Translation table registers
+    CP15_TRANSLATION_BASE_TABLE_0,
+    CP15_TRANSLATION_BASE_TABLE_1,
+    CP15_TRANSLATION_BASE_CONTROL,
+    CP15_DOMAIN_ACCESS_CONTROL,
+    CP15_RESERVED,
+
+    // c5 - Fault status registers
+    CP15_FAULT_STATUS,
+    CP15_INSTR_FAULT_STATUS,
+    CP15_COMBINED_DATA_FSR = CP15_FAULT_STATUS,
+    CP15_INST_FSR,
+
+    // c6 - Fault Address registers
+    CP15_FAULT_ADDRESS,
+    CP15_COMBINED_DATA_FAR = CP15_FAULT_ADDRESS,
+    CP15_WFAR,
+    CP15_IFAR,
+
+    // c7 - Cache operation registers
+    CP15_WAIT_FOR_INTERRUPT,
+    CP15_PHYS_ADDRESS,
+    CP15_INVALIDATE_INSTR_CACHE,
+    CP15_INVALIDATE_INSTR_CACHE_USING_MVA,
+    CP15_INVALIDATE_INSTR_CACHE_USING_INDEX,
+    CP15_FLUSH_PREFETCH_BUFFER,
+    CP15_FLUSH_BRANCH_TARGET_CACHE,
+    CP15_FLUSH_BRANCH_TARGET_CACHE_ENTRY,
+    CP15_INVALIDATE_DATA_CACHE,
+    CP15_INVALIDATE_DATA_CACHE_LINE_USING_MVA,
+    CP15_INVALIDATE_DATA_CACHE_LINE_USING_INDEX,
+    CP15_INVALIDATE_DATA_AND_INSTR_CACHE,
+    CP15_CLEAN_DATA_CACHE,
+    CP15_CLEAN_DATA_CACHE_LINE_USING_MVA,
+    CP15_CLEAN_DATA_CACHE_LINE_USING_INDEX,
+    CP15_DATA_SYNC_BARRIER,
+    CP15_DATA_MEMORY_BARRIER,
+    CP15_CLEAN_AND_INVALIDATE_DATA_CACHE,
+    CP15_CLEAN_AND_INVALIDATE_DATA_CACHE_LINE_USING_MVA,
+    CP15_CLEAN_AND_INVALIDATE_DATA_CACHE_LINE_USING_INDEX,
+
+    // c8 - TLB operations
+    CP15_INVALIDATE_ITLB,
+    CP15_INVALIDATE_ITLB_SINGLE_ENTRY,
+    CP15_INVALIDATE_ITLB_ENTRY_ON_ASID_MATCH,
+    CP15_INVALIDATE_ITLB_ENTRY_ON_MVA,
+    CP15_INVALIDATE_DTLB,
+    CP15_INVALIDATE_DTLB_SINGLE_ENTRY,
+    CP15_INVALIDATE_DTLB_ENTRY_ON_ASID_MATCH,
+    CP15_INVALIDATE_DTLB_ENTRY_ON_MVA,
+    CP15_INVALIDATE_UTLB,
+    CP15_INVALIDATE_UTLB_SINGLE_ENTRY,
+    CP15_INVALIDATE_UTLB_ENTRY_ON_ASID_MATCH,
+    CP15_INVALIDATE_UTLB_ENTRY_ON_MVA,
+
+    // c9 - Data cache lockdown register
+    CP15_DATA_CACHE_LOCKDOWN,
+
+    // c10 - TLB/Memory map registers
+    CP15_TLB_LOCKDOWN,
+    CP15_PRIMARY_REGION_REMAP,
+    CP15_NORMAL_REGION_REMAP,
+
+    // c13 - Thread related registers
+    CP15_PID,
+    CP15_CONTEXT_ID,
+    CP15_THREAD_UPRW, // Thread ID register - User/Privileged Read/Write
+    CP15_THREAD_URO,  // Thread ID register - User Read Only (Privileged R/W)
+    CP15_THREAD_PRW,  // Thread ID register - Privileged R/W only.
+
+    // c15 - Performance and TLB lockdown registers
+    CP15_PERFORMANCE_MONITOR_CONTROL,
+    CP15_CYCLE_COUNTER,
+    CP15_COUNT_0,
+    CP15_COUNT_1,
+    CP15_READ_MAIN_TLB_LOCKDOWN_ENTRY,
+    CP15_WRITE_MAIN_TLB_LOCKDOWN_ENTRY,
+    CP15_MAIN_TLB_LOCKDOWN_VIRT_ADDRESS,
+    CP15_MAIN_TLB_LOCKDOWN_PHYS_ADDRESS,
+    CP15_MAIN_TLB_LOCKDOWN_ATTRIBUTE,
+    CP15_TLB_DEBUG_CONTROL,
+
+    // Skyeye defined
+    CP15_TLB_FAULT_ADDR,
+    CP15_TLB_FAULT_STATUS,
+
+    // Not an actual register.
+    // All registers should be defined above this.
+    CP15_REGISTER_COUNT,
+};
+
+class DynarmicCP15 final : public Dynarmic::A32::Coprocessor {
+public:
+    using CoprocReg = Dynarmic::A32::CoprocReg;
+
+    explicit DynarmicCP15(u32* cp15) : CP15(cp15){};
+
+    std::optional<Callback> CompileInternalOperation(bool two, unsigned opc1, CoprocReg CRd,
+                                                     CoprocReg CRn, CoprocReg CRm,
+                                                     unsigned opc2) override;
+    CallbackOrAccessOneWord CompileSendOneWord(bool two, unsigned opc1, CoprocReg CRn,
+                                               CoprocReg CRm, unsigned opc2) override;
+    CallbackOrAccessTwoWords CompileSendTwoWords(bool two, unsigned opc, CoprocReg CRm) override;
+    CallbackOrAccessOneWord CompileGetOneWord(bool two, unsigned opc1, CoprocReg CRn, CoprocReg CRm,
+                                              unsigned opc2) override;
+    CallbackOrAccessTwoWords CompileGetTwoWords(bool two, unsigned opc, CoprocReg CRm) override;
+    std::optional<Callback> CompileLoadWords(bool two, bool long_transfer, CoprocReg CRd,
+                                             std::optional<u8> option) override;
+    std::optional<Callback> CompileStoreWords(bool two, bool long_transfer, CoprocReg CRd,
+                                              std::optional<u8> option) override;
+
+private:
+    u32* CP15{};
+};

src/core/arm/exclusive_monitor.cpp

@@ -3,7 +3,7 @@
 // Refer to the license.txt file included.
 
 #ifdef ARCHITECTURE_x86_64
-#include "core/arm/dynarmic/arm_dynarmic.h"
+#include "core/arm/dynarmic/arm_dynarmic_64.h"
 #endif
 #include "core/arm/exclusive_monitor.h"
 #include "core/memory.h"

src/core/arm/exclusive_monitor.h

@@ -8,7 +8,7 @@
 
 #include "common/common_types.h"
 
-namespace Memory {
+namespace Core::Memory {
 class Memory;
 }
 

src/core/arm/unicorn/arm_unicorn.cpp

@@ -11,6 +11,7 @@
 #include "core/core_timing.h"
 #include "core/hle/kernel/scheduler.h"
 #include "core/hle/kernel/svc.h"
+#include "core/memory.h"
 
 namespace Core {
 
@@ -53,7 +54,7 @@ static bool UnmappedMemoryHook(uc_engine* uc, uc_mem_type type, u64 addr, int si
                                void* user_data) {
     auto* const system = static_cast<System*>(user_data);
 
-    ARM_Interface::ThreadContext ctx{};
+    ARM_Interface::ThreadContext64 ctx{};
     system->CurrentArmInterface().SaveContext(ctx);
     ASSERT_MSG(false, "Attempted to read from unmapped memory: 0x{:X}, pc=0x{:X}, lr=0x{:X}", addr,
                ctx.pc, ctx.cpu_registers[30]);
@@ -171,7 +172,17 @@ MICROPROFILE_DEFINE(ARM_Jit_Unicorn, "ARM JIT", "Unicorn", MP_RGB(255, 64, 64));
 
 void ARM_Unicorn::ExecuteInstructions(std::size_t num_instructions) {
     MICROPROFILE_SCOPE(ARM_Jit_Unicorn);
+
+    // Temporarily map the code page for Unicorn
+    u64 map_addr{GetPC() & ~Memory::PAGE_MASK};
+    std::vector<u8> page_buffer(Memory::PAGE_SIZE);
+    system.Memory().ReadBlock(map_addr, page_buffer.data(), page_buffer.size());
+
+    CHECKED(uc_mem_map_ptr(uc, map_addr, page_buffer.size(),
+                           UC_PROT_READ | UC_PROT_WRITE | UC_PROT_EXEC, page_buffer.data()));
     CHECKED(uc_emu_start(uc, GetPC(), 1ULL << 63, 0, num_instructions));
+    CHECKED(uc_mem_unmap(uc, map_addr, page_buffer.size()));
+
     system.CoreTiming().AddTicks(num_instructions);
     if (GDBStub::IsServerEnabled()) {
         if (last_bkpt_hit && last_bkpt.type == GDBStub::BreakpointType::Execute) {
@@ -179,7 +190,7 @@ void ARM_Unicorn::ExecuteInstructions(std::size_t num_instructions) {
         }
 
         Kernel::Thread* const thread = system.CurrentScheduler().GetCurrentThread();
-        SaveContext(thread->GetContext());
+        SaveContext(thread->GetContext64());
         if (last_bkpt_hit || GDBStub::IsMemoryBreak() || GDBStub::GetCpuStepFlag()) {
             last_bkpt_hit = false;
             GDBStub::Break();
@@ -188,7 +199,7 @@ void ARM_Unicorn::ExecuteInstructions(std::size_t num_instructions) {
     }
 }
 
-void ARM_Unicorn::SaveContext(ThreadContext& ctx) {
+void ARM_Unicorn::SaveContext(ThreadContext64& ctx) {
     int uregs[32];
     void* tregs[32];
 
@@ -215,7 +226,7 @@ void ARM_Unicorn::SaveContext(ThreadContext& ctx) {
     CHECKED(uc_reg_read_batch(uc, uregs, tregs, 32));
 }
 
-void ARM_Unicorn::LoadContext(const ThreadContext& ctx) {
+void ARM_Unicorn::LoadContext(const ThreadContext64& ctx) {
     int uregs[32];
     void* tregs[32];
 
@@ -266,7 +277,7 @@ void ARM_Unicorn::InterruptHook(uc_engine* uc, u32 int_no, void* user_data) {
 
     switch (ec) {
     case 0x15: // SVC
-        Kernel::CallSVC(arm_instance->system, iss);
+        Kernel::Svc::Call(arm_instance->system, iss);
         break;
     }
 }

src/core/arm/unicorn/arm_unicorn.h

@@ -30,8 +30,6 @@ public:
     void SetTlsAddress(VAddr address) override;
     void SetTPIDR_EL0(u64 value) override;
     u64 GetTPIDR_EL0() const override;
-    void SaveContext(ThreadContext& ctx) override;
-    void LoadContext(const ThreadContext& ctx) override;
     void PrepareReschedule() override;
     void ClearExclusiveState() override;
     void ExecuteInstructions(std::size_t num_instructions);
@@ -41,6 +39,11 @@ public:
     void PageTableChanged(Common::PageTable&, std::size_t) override {}
     void RecordBreak(GDBStub::BreakpointAddress bkpt);
 
+    void SaveContext(ThreadContext32& ctx) override {}
+    void SaveContext(ThreadContext64& ctx) override;
+    void LoadContext(const ThreadContext32& ctx) override {}
+    void LoadContext(const ThreadContext64& ctx) override;
+
 private:
     static void InterruptHook(uc_engine* uc, u32 int_no, void* user_data);
 

src/core/core.cpp

@@ -14,6 +14,7 @@
 #include "core/core_manager.h"
 #include "core/core_timing.h"
 #include "core/cpu_manager.h"
+#include "core/device_memory.h"
 #include "core/file_sys/bis_factory.h"
 #include "core/file_sys/card_image.h"
 #include "core/file_sys/mode.h"
@@ -140,6 +141,8 @@ struct System::Impl {
     ResultStatus Init(System& system, Frontend::EmuWindow& emu_window) {
         LOG_DEBUG(HW_Memory, "initialized OK");
 
+        device_memory = std::make_unique<Core::DeviceMemory>(system);
+
         core_timing.Initialize();
         kernel.Initialize();
         cpu_manager.Initialize();
@@ -165,14 +168,14 @@ struct System::Impl {
         service_manager = std::make_shared<Service::SM::ServiceManager>();
 
         Service::Init(service_manager, system);
-        GDBStub::Init();
+        GDBStub::DeferStart();
 
-        renderer = VideoCore::CreateRenderer(emu_window, system);
-        if (!renderer->Init()) {
+        interrupt_manager = std::make_unique<Core::Hardware::InterruptManager>(system);
+        gpu_core = VideoCore::CreateGPU(emu_window, system);
+        if (!gpu_core) {
             return ResultStatus::ErrorVideoCore;
         }
-        interrupt_manager = std::make_unique<Core::Hardware::InterruptManager>(system);
-        gpu_core = VideoCore::CreateGPU(system);
+        gpu_core->Renderer().Rasterizer().SetupDirtyFlags();
 
         is_powered_on = true;
         exit_lock = false;
@@ -212,10 +215,6 @@ struct System::Impl {
         AddGlueRegistrationForProcess(*app_loader, *main_process);
         kernel.MakeCurrentProcess(main_process.get());
 
-        // Main process has been loaded and been made current.
-        // Begin GPU and CPU execution.
-        gpu_core->Start();
-
         // Initialize cheat engine
         if (cheat_engine) {
             cheat_engine->Initialize();
@@ -273,7 +272,6 @@ struct System::Impl {
         }
 
         // Shutdown emulation session
-        renderer.reset();
         GDBStub::Shutdown();
         Service::Shutdown();
         service_manager.reset();
@@ -281,6 +279,7 @@ struct System::Impl {
         telemetry_session.reset();
         perf_stats.reset();
         gpu_core.reset();
+        device_memory.reset();
 
         // Close all CPU/threading state
         cpu_manager.Shutdown();
@@ -349,10 +348,10 @@ struct System::Impl {
     Service::FileSystem::FileSystemController fs_controller;
     /// AppLoader used to load the current executing application
     std::unique_ptr<Loader::AppLoader> app_loader;
-    std::unique_ptr<VideoCore::RendererBase> renderer;
     std::unique_ptr<Tegra::GPU> gpu_core;
     std::unique_ptr<Hardware::InterruptManager> interrupt_manager;
-    Memory::Memory memory;
+    std::unique_ptr<Core::DeviceMemory> device_memory;
+    Core::Memory::Memory memory;
     CpuManager cpu_manager;
     bool is_powered_on = false;
     bool exit_lock = false;
@@ -478,6 +477,14 @@ Kernel::Process* System::CurrentProcess() {
     return impl->kernel.CurrentProcess();
 }
 
+Core::DeviceMemory& System::DeviceMemory() {
+    return *impl->device_memory;
+}
+
+const Core::DeviceMemory& System::DeviceMemory() const {
+    return *impl->device_memory;
+}
+
 const Kernel::Process* System::CurrentProcess() const {
     return impl->kernel.CurrentProcess();
 }
@@ -511,7 +518,7 @@ Memory::Memory& System::Memory() {
     return impl->memory;
 }
 
-const Memory::Memory& System::Memory() const {
+const Core::Memory::Memory& System::Memory() const {
     return impl->memory;
 }
 
@@ -532,11 +539,11 @@ const Core::Hardware::InterruptManager& System::InterruptManager() const {
 }
 
 VideoCore::RendererBase& System::Renderer() {
-    return *impl->renderer;
+    return impl->gpu_core->Renderer();
 }
 
 const VideoCore::RendererBase& System::Renderer() const {
-    return *impl->renderer;
+    return impl->gpu_core->Renderer();
 }
 
 Kernel::KernelCore& System::Kernel() {
@@ -707,4 +714,12 @@ const Service::SM::ServiceManager& System::ServiceManager() const {
     return *impl->service_manager;
 }
 
+void System::RegisterCoreThread(std::size_t id) {
+    impl->kernel.RegisterCoreThread(id);
+}
+
+void System::RegisterHostThread() {
+    impl->kernel.RegisterHostThread();
+}
+
 } // namespace Core

src/core/core.h

@@ -36,9 +36,10 @@ class AppLoader;
 enum class ResultStatus : u16;
 } // namespace Loader
 
-namespace Memory {
+namespace Core::Memory {
 struct CheatEntry;
-} // namespace Memory
+class Memory;
+} // namespace Core::Memory
 
 namespace Service {
 
@@ -86,14 +87,11 @@ namespace Core::Hardware {
 class InterruptManager;
 }
 
-namespace Memory {
-class Memory;
-}
-
 namespace Core {
 
 class ARM_Interface;
 class CoreManager;
+class DeviceMemory;
 class ExclusiveMonitor;
 class FrameLimiter;
 class PerfStats;
@@ -230,10 +228,10 @@ public:
     const ExclusiveMonitor& Monitor() const;
 
     /// Gets a mutable reference to the system memory instance.
-    Memory::Memory& Memory();
+    Core::Memory::Memory& Memory();
 
     /// Gets a constant reference to the system memory instance.
-    const Memory::Memory& Memory() const;
+    const Core::Memory::Memory& Memory() const;
 
     /// Gets a mutable reference to the GPU interface
     Tegra::GPU& GPU();
@@ -259,6 +257,12 @@ public:
     /// Gets the global scheduler
     const Kernel::GlobalScheduler& GlobalScheduler() const;
 
+    /// Gets the manager for the guest device memory
+    Core::DeviceMemory& DeviceMemory();
+
+    /// Gets the manager for the guest device memory
+    const Core::DeviceMemory& DeviceMemory() const;
+
     /// Provides a pointer to the current process
     Kernel::Process* CurrentProcess();
 
@@ -360,6 +364,12 @@ public:
 
     const CurrentBuildProcessID& GetCurrentProcessBuildID() const;
 
+    /// Register a host thread as an emulated CPU Core.
+    void RegisterCoreThread(std::size_t id);
+
+    /// Register a host thread as an auxiliary thread.
+    void RegisterHostThread();
+
 private:
     System();
 

src/core/core_manager.cpp

@@ -6,9 +6,6 @@
 #include <mutex>
 
 #include "common/logging/log.h"
-#ifdef ARCHITECTURE_x86_64
-#include "core/arm/dynarmic/arm_dynarmic.h"
-#endif
 #include "core/arm/exclusive_monitor.h"
 #include "core/arm/unicorn/arm_unicorn.h"
 #include "core/core.h"

src/core/core_manager.h

@@ -22,7 +22,7 @@ namespace Core::Timing {
 class CoreTiming;
 }
 
-namespace Memory {
+namespace Core::Memory {
 class Memory;
 }
 

src/core/crypto/key_manager.cpp

@@ -1202,7 +1202,8 @@ const boost::container::flat_map<std::string, KeyIndex<S128KeyType>> KeyManager:
      {S128KeyType::Source, static_cast<u64>(SourceKeyType::KeyAreaKey),
       static_cast<u64>(KeyAreaKeyType::System)}},
     {"titlekek_source", {S128KeyType::Source, static_cast<u64>(SourceKeyType::Titlekek), 0}},
-    {"keyblob_mac_key_source", {S128KeyType::Source, static_cast<u64>(SourceKeyType::KeyblobMAC)}},
+    {"keyblob_mac_key_source",
+     {S128KeyType::Source, static_cast<u64>(SourceKeyType::KeyblobMAC), 0}},
     {"tsec_key", {S128KeyType::TSEC, 0, 0}},
     {"secure_boot_key", {S128KeyType::SecureBoot, 0, 0}},
     {"sd_seed", {S128KeyType::SDSeed, 0, 0}},

src/core/crypto/partition_data_manager.cpp

@@ -202,8 +202,8 @@ static std::array<Key128, 0x20> FindEncryptedMasterKeyFromHex(const std::vector<
     return out;
 }
 
-FileSys::VirtualFile FindFileInDirWithNames(const FileSys::VirtualDir& dir,
-                                            const std::string& name) {
+static FileSys::VirtualFile FindFileInDirWithNames(const FileSys::VirtualDir& dir,
+                                                   const std::string& name) {
     const auto upper = Common::ToUpper(name);
 
     for (const auto& fname : {name, name + ".bin", upper, upper + ".BIN"}) {
@@ -345,8 +345,7 @@ FileSys::VirtualFile PartitionDataManager::GetPackage2Raw(Package2Type type) con
     return package2.at(static_cast<size_t>(type));
 }
 
-bool AttemptDecrypt(const std::array<u8, 16>& key, Package2Header& header) {
-
+static bool AttemptDecrypt(const std::array<u8, 16>& key, Package2Header& header) {
     const std::vector<u8> iv(header.header_ctr.begin(), header.header_ctr.end());
     Package2Header temp = header;
     AESCipher<Key128> cipher(key, Mode::CTR);

src/core/device_memory.cpp

@@ -0,0 +1,15 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "core/core.h"
+#include "core/device_memory.h"
+#include "core/memory.h"
+
+namespace Core {
+
+DeviceMemory::DeviceMemory(System& system) : buffer{DramMemoryMap::Size}, system{system} {}
+
+DeviceMemory::~DeviceMemory() = default;
+
+} // namespace Core

src/core/device_memory.h

@@ -0,0 +1,51 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "common/assert.h"
+#include "common/common_funcs.h"
+#include "common/virtual_buffer.h"
+
+namespace Core {
+
+class System;
+
+namespace DramMemoryMap {
+enum : u64 {
+    Base = 0x80000000ULL,
+    Size = 0x100000000ULL,
+    End = Base + Size,
+    KernelReserveBase = Base + 0x60000,
+    SlabHeapBase = KernelReserveBase + 0x85000,
+    SlapHeapSize = 0xa21000,
+    SlabHeapEnd = SlabHeapBase + SlapHeapSize,
+};
+}; // namespace DramMemoryMap
+
+class DeviceMemory : NonCopyable {
+public:
+    explicit DeviceMemory(Core::System& system);
+    ~DeviceMemory();
+
+    template <typename T>
+    PAddr GetPhysicalAddr(const T* ptr) const {
+        return (reinterpret_cast<uintptr_t>(ptr) - reinterpret_cast<uintptr_t>(buffer.data())) +
+               DramMemoryMap::Base;
+    }
+
+    u8* GetPointer(PAddr addr) {
+        return buffer.data() + (addr - DramMemoryMap::Base);
+    }
+
+    const u8* GetPointer(PAddr addr) const {
+        return buffer.data() + (addr - DramMemoryMap::Base);
+    }
+
+private:
+    Common::VirtualBuffer<u8> buffer;
+    Core::System& system;
+};
+
+} // namespace Core

src/core/file_sys/patch_manager.cpp

@@ -249,7 +249,7 @@ bool PatchManager::HasNSOPatch(const std::array<u8, 32>& build_id_) const {
 }
 
 namespace {
-std::optional<std::vector<Memory::CheatEntry>> ReadCheatFileFromFolder(
+std::optional<std::vector<Core::Memory::CheatEntry>> ReadCheatFileFromFolder(
     const Core::System& system, u64 title_id, const std::array<u8, 0x20>& build_id_,
     const VirtualDir& base_path, bool upper) {
     const auto build_id_raw = Common::HexToString(build_id_, upper);
@@ -269,14 +269,14 @@ std::optional<std::vector<Memory::CheatEntry>> ReadCheatFileFromFolder(
         return std::nullopt;
     }
 
-    Memory::TextCheatParser parser;
+    Core::Memory::TextCheatParser parser;
     return parser.Parse(
         system, std::string_view(reinterpret_cast<const char* const>(data.data()), data.size()));
 }
 
 } // Anonymous namespace
 
-std::vector<Memory::CheatEntry> PatchManager::CreateCheatList(
+std::vector<Core::Memory::CheatEntry> PatchManager::CreateCheatList(
     const Core::System& system, const std::array<u8, 32>& build_id_) const {
     const auto load_dir = system.GetFileSystemController().GetModificationLoadRoot(title_id);
     if (load_dir == nullptr) {
@@ -289,7 +289,7 @@ std::vector<Memory::CheatEntry> PatchManager::CreateCheatList(
     std::sort(patch_dirs.begin(), patch_dirs.end(),
               [](const VirtualDir& l, const VirtualDir& r) { return l->GetName() < r->GetName(); });
 
-    std::vector<Memory::CheatEntry> out;
+    std::vector<Core::Memory::CheatEntry> out;
     for (const auto& subdir : patch_dirs) {
         if (std::find(disabled.cbegin(), disabled.cend(), subdir->GetName()) != disabled.cend()) {
             continue;
@@ -348,6 +348,12 @@ static void ApplyLayeredFS(VirtualFile& romfs, u64 title_id, ContentRecordType t
         if (ext_dir != nullptr)
             layers_ext.push_back(std::move(ext_dir));
     }
+
+    // When there are no layers to apply, return early as there is no need to rebuild the RomFS
+    if (layers.empty() && layers_ext.empty()) {
+        return;
+    }
+
     layers.push_back(std::move(extracted));
 
     auto layered = LayeredVfsDirectory::MakeLayeredDirectory(std::move(layers));
@@ -434,7 +440,8 @@ std::map<std::string, std::string, std::less<>> PatchManager::GetPatchVersionNam
     // Game Updates
     const auto update_tid = GetUpdateTitleID(title_id);
     PatchManager update{update_tid};
-    auto [nacp, discard_icon_file] = update.GetControlMetadata();
+    const auto metadata = update.GetControlMetadata();
+    const auto& nacp = metadata.first;
 
     const auto update_disabled =
         std::find(disabled.cbegin(), disabled.cend(), "Update") != disabled.cend();

src/core/file_sys/patch_manager.h

@@ -51,8 +51,8 @@ public:
     bool HasNSOPatch(const std::array<u8, 0x20>& build_id) const;
 
     // Creates a CheatList object with all
-    std::vector<Memory::CheatEntry> CreateCheatList(const Core::System& system,
-                                                    const std::array<u8, 0x20>& build_id) const;
+    std::vector<Core::Memory::CheatEntry> CreateCheatList(
+        const Core::System& system, const std::array<u8, 0x20>& build_id) const;
 
     // Currently tracked RomFS patches:
     // - Game Updates

src/core/file_sys/program_metadata.cpp

@@ -51,6 +51,17 @@ Loader::ResultStatus ProgramMetadata::Load(VirtualFile file) {
     return Loader::ResultStatus::Success;
 }
 
+/*static*/ ProgramMetadata ProgramMetadata::GetDefault() {
+    ProgramMetadata result;
+
+    result.LoadManual(
+        true /*is_64_bit*/, FileSys::ProgramAddressSpaceType::Is39Bit /*address_space*/,
+        0x2c /*main_thread_prio*/, 0 /*main_thread_core*/, 0x00100000 /*main_thread_stack_size*/,
+        {}, 0xFFFFFFFFFFFFFFFF /*filesystem_permissions*/, {} /*capabilities*/);
+
+    return result;
+}
+
 void ProgramMetadata::LoadManual(bool is_64_bit, ProgramAddressSpaceType address_space,
                                  s32 main_thread_prio, u32 main_thread_core,
                                  u32 main_thread_stack_size, u64 title_id,

src/core/file_sys/program_metadata.h

@@ -44,9 +44,13 @@ public:
     ProgramMetadata();
     ~ProgramMetadata();
 
+    /// Gets a default ProgramMetadata configuration, should only be used for homebrew formats where
+    /// we do not have an NPDM file
+    static ProgramMetadata GetDefault();
+
     Loader::ResultStatus Load(VirtualFile file);
 
-    // Load from parameters instead of NPDM file, used for KIP
+    /// Load from parameters instead of NPDM file, used for KIP
     void LoadManual(bool is_64_bit, ProgramAddressSpaceType address_space, s32 main_thread_prio,
                     u32 main_thread_core, u32 main_thread_stack_size, u64 title_id,
                     u64 filesystem_permissions, KernelCapabilityDescriptors capabilities);

src/core/file_sys/registered_cache.cpp

@@ -591,14 +591,18 @@ InstallResult RegisteredCache::InstallEntry(const NSP& nsp, bool overwrite_if_ex
 InstallResult RegisteredCache::InstallEntry(const NCA& nca, TitleType type,
                                             bool overwrite_if_exists, const VfsCopyFunction& copy) {
     CNMTHeader header{
-        nca.GetTitleId(), ///< Title ID
-        0,                ///< Ignore/Default title version
-        type,             ///< Type
-        {},               ///< Padding
-        0x10,             ///< Default table offset
-        1,                ///< 1 Content Entry
-        0,                ///< No Meta Entries
-        {},               ///< Padding
+        nca.GetTitleId(), // Title ID
+        0,                // Ignore/Default title version
+        type,             // Type
+        {},               // Padding
+        0x10,             // Default table offset
+        1,                // 1 Content Entry
+        0,                // No Meta Entries
+        {},               // Padding
+        {},               // Reserved 1
+        0,                // Is committed
+        0,                // Required download system version
+        {},               // Reserved 2
     };
     OptionalHeader opt_header{0, 0};
     ContentRecord c_rec{{}, {}, {}, GetCRTypeFromNCAType(nca.GetType()), {}};
@@ -848,7 +852,8 @@ VirtualFile ManualContentProvider::GetEntryUnparsed(u64 title_id, ContentRecordT
 VirtualFile ManualContentProvider::GetEntryRaw(u64 title_id, ContentRecordType type) const {
     const auto iter =
         std::find_if(entries.begin(), entries.end(), [title_id, type](const auto& entry) {
-            const auto [title_type, content_type, e_title_id] = entry.first;
+            const auto content_type = std::get<1>(entry.first);
+            const auto e_title_id = std::get<2>(entry.first);
             return content_type == type && e_title_id == title_id;
         });
     if (iter == entries.end())

src/core/file_sys/romfs.cpp

@@ -5,6 +5,7 @@
 #include <memory>
 
 #include "common/common_types.h"
+#include "common/string_util.h"
 #include "common/swap.h"
 #include "core/file_sys/fsmitm_romfsbuild.h"
 #include "core/file_sys/romfs.h"
@@ -126,7 +127,7 @@ VirtualDir ExtractRomFS(VirtualFile file, RomFSExtractionType type) {
         return out->GetSubdirectories().front();
 
     while (out->GetSubdirectories().size() == 1 && out->GetFiles().empty()) {
-        if (out->GetSubdirectories().front()->GetName() == "data" &&
+        if (Common::ToLower(out->GetSubdirectories().front()->GetName()) == "data" &&
             type == RomFSExtractionType::Truncated)
             break;
         out = out->GetSubdirectories().front();

src/core/file_sys/vfs_libzip.cpp

@@ -42,11 +42,11 @@ VirtualDir ExtractZIP(VirtualFile file) {
             continue;
 
         if (name.back() != '/') {
-            std::unique_ptr<zip_file_t, decltype(&zip_fclose)> file{
+            std::unique_ptr<zip_file_t, decltype(&zip_fclose)> file2{
                 zip_fopen_index(zip.get(), i, 0), zip_fclose};
 
             std::vector<u8> buf(stat.size);
-            if (zip_fread(file.get(), buf.data(), buf.size()) != buf.size())
+            if (zip_fread(file2.get(), buf.data(), buf.size()) != s64(buf.size()))
                 return nullptr;
 
             const auto parts = FileUtil::SplitPathComponents(stat.name);

src/core/frontend/emu_window.h

@@ -12,23 +12,49 @@
 
 namespace Core::Frontend {
 
+/// Information for the Graphics Backends signifying what type of screen pointer is in
+/// WindowInformation
+enum class WindowSystemType {
+    Headless,
+    Windows,
+    X11,
+    Wayland,
+};
+
 /**
- * Represents a graphics context that can be used for background computation or drawing. If the
- * graphics backend doesn't require the context, then the implementation of these methods can be
- * stubs
+ * Represents a drawing context that supports graphics operations.
  */
 class GraphicsContext {
 public:
     virtual ~GraphicsContext();
 
+    /// Inform the driver to swap the front/back buffers and present the current image
+    virtual void SwapBuffers() {}
+
     /// Makes the graphics context current for the caller thread
-    virtual void MakeCurrent() = 0;
+    virtual void MakeCurrent() {}
 
     /// Releases (dunno if this is the "right" word) the context from the caller thread
-    virtual void DoneCurrent() = 0;
+    virtual void DoneCurrent() {}
 
-    /// Swap buffers to display the next frame
-    virtual void SwapBuffers() = 0;
+    class Scoped {
+    public:
+        explicit Scoped(GraphicsContext& context_) : context(context_) {
+            context.MakeCurrent();
+        }
+        ~Scoped() {
+            context.DoneCurrent();
+        }
+
+    private:
+        GraphicsContext& context;
+    };
+
+    /// Calls MakeCurrent on the context and calls DoneCurrent when the scope for the returned value
+    /// ends
+    Scoped Acquire() {
+        return Scoped{*this};
+    }
 };
 
 /**
@@ -49,7 +75,7 @@ public:
  * - DO NOT TREAT THIS CLASS AS A GUI TOOLKIT ABSTRACTION LAYER. That's not what it is. Please
  *   re-read the upper points again and think about it if you don't see this.
  */
-class EmuWindow : public GraphicsContext {
+class EmuWindow {
 public:
     /// Data structure to store emuwindow configuration
     struct WindowConfig {
@@ -59,29 +85,34 @@ public:
         std::pair<unsigned, unsigned> min_client_area_size;
     };
 
+    /// Data describing host window system information
+    struct WindowSystemInfo {
+        // Window system type. Determines which GL context or Vulkan WSI is used.
+        WindowSystemType type = WindowSystemType::Headless;
+
+        // Connection to a display server. This is used on X11 and Wayland platforms.
+        void* display_connection = nullptr;
+
+        // Render surface. This is a pointer to the native window handle, which depends
+        // on the platform. e.g. HWND for Windows, Window for X11. If the surface is
+        // set to nullptr, the video backend will run in headless mode.
+        void* render_surface = nullptr;
+
+        // Scale of the render surface. For hidpi systems, this will be >1.
+        float render_surface_scale = 1.0f;
+    };
+
     /// Polls window events
     virtual void PollEvents() = 0;
 
     /**
-     * Returns a GraphicsContext that the frontend provides that is shared with the emu window. This
-     * context can be used from other threads for background graphics computation. If the frontend
-     * is using a graphics backend that doesn't need anything specific to run on a different thread,
-     * then it can use a stubbed implemenation for GraphicsContext.
-     *
-     * If the return value is null, then the core should assume that the frontend cannot provide a
-     * Shared Context
+     * Returns a GraphicsContext that the frontend provides to be used for rendering.
      */
-    virtual std::unique_ptr<GraphicsContext> CreateSharedContext() const {
-        return nullptr;
-    }
+    virtual std::unique_ptr<GraphicsContext> CreateSharedContext() const = 0;
 
     /// Returns if window is shown (not minimized)
     virtual bool IsShown() const = 0;
 
-    /// Retrieves Vulkan specific handlers from the window
-    virtual void RetrieveVulkanHandlers(void* get_instance_proc_addr, void* instance,
-                                        void* surface) const = 0;
-
     /**
      * Signal that a touch pressed event has occurred (e.g. mouse click pressed)
      * @param framebuffer_x Framebuffer x-coordinate that was pressed
@@ -118,6 +149,13 @@ public:
         config = val;
     }
 
+    /**
+     * Returns system information about the drawing area.
+     */
+    const WindowSystemInfo& GetWindowInfo() const {
+        return window_info;
+    }
+
     /**
      * Gets the framebuffer layout (width, height, and screen regions)
      * @note This method is thread-safe
@@ -133,7 +171,7 @@ public:
     void UpdateCurrentFramebufferLayout(unsigned width, unsigned height);
 
 protected:
-    EmuWindow();
+    explicit EmuWindow();
     virtual ~EmuWindow();
 
     /**
@@ -170,6 +208,8 @@ protected:
         client_area_height = size.second;
     }
 
+    WindowSystemInfo window_info;
+
 private:
     /**
      * Handler called when the minimal client area was requested to be changed via SetConfig.

src/core/frontend/framebuffer_layout.cpp

@@ -25,7 +25,7 @@ FramebufferLayout DefaultFrameLayout(u32 width, u32 height) {
     ASSERT(height > 0);
     // The drawing code needs at least somewhat valid values for both screens
     // so just calculate them both even if the other isn't showing.
-    FramebufferLayout res{width, height};
+    FramebufferLayout res{width, height, false, {}};
 
     const float window_aspect_ratio = static_cast<float>(height) / width;
     const float emulation_aspect_ratio = EmulationAspectRatio(
@@ -48,8 +48,8 @@ FramebufferLayout FrameLayoutFromResolutionScale(u32 res_scale) {
     u32 width, height;
 
     if (Settings::values.use_docked_mode) {
-        width = ScreenDocked::WidthDocked * res_scale;
-        height = ScreenDocked::HeightDocked * res_scale;
+        width = ScreenDocked::Width * res_scale;
+        height = ScreenDocked::Height * res_scale;
     } else {
         width = ScreenUndocked::Width * res_scale;
         height = ScreenUndocked::Height * res_scale;

src/core/frontend/framebuffer_layout.h

@@ -8,15 +8,15 @@
 
 namespace Layout {
 
-enum ScreenUndocked : u32 {
-    Width = 1280,
-    Height = 720,
-};
+namespace ScreenUndocked {
+constexpr u32 Width = 1280;
+constexpr u32 Height = 720;
+} // namespace ScreenUndocked
 
-enum ScreenDocked : u32 {
-    WidthDocked = 1920,
-    HeightDocked = 1080,
-};
+namespace ScreenDocked {
+constexpr u32 Width = 1920;
+constexpr u32 Height = 1080;
+} // namespace ScreenDocked
 
 enum class AspectRatio {
     Default,
@@ -29,6 +29,7 @@ enum class AspectRatio {
 struct FramebufferLayout {
     u32 width{ScreenUndocked::Width};
     u32 height{ScreenUndocked::Height};
+    bool is_srgb{};
 
     Common::Rectangle<u32> screen;
 

src/core/frontend/scope_acquire_window_context.cpp

@@ -1,18 +0,0 @@
-// Copyright 2019 yuzu Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#include "core/frontend/emu_window.h"
-#include "core/frontend/scope_acquire_window_context.h"
-
-namespace Core::Frontend {
-
-ScopeAcquireWindowContext::ScopeAcquireWindowContext(Core::Frontend::EmuWindow& emu_window_)
-    : emu_window{emu_window_} {
-    emu_window.MakeCurrent();
-}
-ScopeAcquireWindowContext::~ScopeAcquireWindowContext() {
-    emu_window.DoneCurrent();
-}
-
-} // namespace Core::Frontend

src/core/frontend/scope_acquire_window_context.h

@@ -1,23 +0,0 @@
-// Copyright 2019 yuzu Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#pragma once
-
-#include "common/common_types.h"
-
-namespace Core::Frontend {
-
-class EmuWindow;
-
-/// Helper class to acquire/release window context within a given scope
-class ScopeAcquireWindowContext : NonCopyable {
-public:
-    explicit ScopeAcquireWindowContext(Core::Frontend::EmuWindow& window);
-    ~ScopeAcquireWindowContext();
-
-private:
-    Core::Frontend::EmuWindow& emu_window;
-};
-
-} // namespace Core::Frontend

src/core/gdbstub/gdbstub.cpp

@@ -37,9 +37,9 @@
 #include "core/core.h"
 #include "core/core_manager.h"
 #include "core/gdbstub/gdbstub.h"
+#include "core/hle/kernel/memory/page_table.h"
 #include "core/hle/kernel/process.h"
 #include "core/hle/kernel/scheduler.h"
-#include "core/hle/kernel/vm_manager.h"
 #include "core/loader/loader.h"
 #include "core/memory.h"
 
@@ -141,6 +141,7 @@ constexpr char target_xml[] =
 )";
 
 int gdbserver_socket = -1;
+bool defer_start = false;
 
 u8 command_buffer[GDB_BUFFER_SIZE];
 u32 command_length;
@@ -217,7 +218,7 @@ static u64 RegRead(std::size_t id, Kernel::Thread* thread = nullptr) {
         return 0;
     }
 
-    const auto& thread_context = thread->GetContext();
+    const auto& thread_context = thread->GetContext64();
 
     if (id < SP_REGISTER) {
         return thread_context.cpu_registers[id];
@@ -239,7 +240,7 @@ static void RegWrite(std::size_t id, u64 val, Kernel::Thread* thread = nullptr)
         return;
     }
 
-    auto& thread_context = thread->GetContext();
+    auto& thread_context = thread->GetContext64();
 
     if (id < SP_REGISTER) {
         thread_context.cpu_registers[id] = val;
@@ -259,7 +260,7 @@ static u128 FpuRead(std::size_t id, Kernel::Thread* thread = nullptr) {
         return u128{0};
     }
 
-    auto& thread_context = thread->GetContext();
+    auto& thread_context = thread->GetContext64();
 
     if (id >= UC_ARM64_REG_Q0 && id < FPCR_REGISTER) {
         return thread_context.vector_registers[id - UC_ARM64_REG_Q0];
@@ -275,7 +276,7 @@ static void FpuWrite(std::size_t id, u128 val, Kernel::Thread* thread = nullptr)
         return;
     }
 
-    auto& thread_context = thread->GetContext();
+    auto& thread_context = thread->GetContext64();
 
     if (id >= UC_ARM64_REG_Q0 && id < FPCR_REGISTER) {
         thread_context.vector_registers[id - UC_ARM64_REG_Q0] = val;
@@ -642,7 +643,7 @@ static void HandleQuery() {
         SendReply(target_xml);
     } else if (strncmp(query, "Offsets", strlen("Offsets")) == 0) {
         const VAddr base_address =
-            Core::System::GetInstance().CurrentProcess()->VMManager().GetCodeRegionBaseAddress();
+            Core::System::GetInstance().CurrentProcess()->PageTable().GetCodeRegionStart();
         std::string buffer = fmt::format("TextSeg={:0x}", base_address);
         SendReply(buffer.c_str());
     } else if (strncmp(query, "fThreadInfo", strlen("fThreadInfo")) == 0) {
@@ -916,7 +917,7 @@ static void WriteRegister() {
     // Update ARM context, skipping scheduler - no running threads at this point
     Core::System::GetInstance()
         .ArmInterface(current_core)
-        .LoadContext(current_thread->GetContext());
+        .LoadContext(current_thread->GetContext64());
 
     SendReply("OK");
 }
@@ -947,7 +948,7 @@ static void WriteRegisters() {
     // Update ARM context, skipping scheduler - no running threads at this point
     Core::System::GetInstance()
         .ArmInterface(current_core)
-        .LoadContext(current_thread->GetContext());
+        .LoadContext(current_thread->GetContext64());
 
     SendReply("OK");
 }
@@ -1019,7 +1020,7 @@ static void Step() {
         // Update ARM context, skipping scheduler - no running threads at this point
         Core::System::GetInstance()
             .ArmInterface(current_core)
-            .LoadContext(current_thread->GetContext());
+            .LoadContext(current_thread->GetContext64());
     }
     step_loop = true;
     halt_loop = true;
@@ -1166,6 +1167,9 @@ static void RemoveBreakpoint() {
 
 void HandlePacket() {
     if (!IsConnected()) {
+        if (defer_start) {
+            ToggleServer(true);
+        }
         return;
     }
 
@@ -1256,6 +1260,10 @@ void ToggleServer(bool status) {
     }
 }
 
+void DeferStart() {
+    defer_start = true;
+}
+
 static void Init(u16 port) {
     if (!server_enabled) {
         // Set the halt loop to false in case the user enabled the gdbstub mid-execution.
@@ -1341,6 +1349,7 @@ void Shutdown() {
     if (!server_enabled) {
         return;
     }
+    defer_start = false;
 
     LOG_INFO(Debug_GDBStub, "Stopping GDB ...");
     if (gdbserver_socket != -1) {
@@ -1380,10 +1389,9 @@ void SendTrap(Kernel::Thread* thread, int trap) {
         return;
     }
 
-    if (!halt_loop || current_thread == thread) {
-        current_thread = thread;
-        SendSignal(thread, trap);
-    }
+    current_thread = thread;
+    SendSignal(thread, trap);
+
     halt_loop = true;
     send_trap = false;
 }

src/core/gdbstub/gdbstub.h

@@ -43,6 +43,13 @@ void ToggleServer(bool status);
 /// Start the gdbstub server.
 void Init();
 
+/**
+ * Defer initialization of the gdbstub to the first packet processing functions.
+ * This avoids a case where the gdbstub thread is frozen after initialization
+ * and fails to respond in time to packets.
+ */
+void DeferStart();
+
 /// Stop gdbstub server.
 void Shutdown();
 

src/core/hardware_properties.h

@@ -20,6 +20,8 @@ constexpr u32 NUM_CPU_CORES = 4;            // Number of CPU Cores
 
 } // namespace Hardware
 
+constexpr u32 INVALID_HOST_THREAD_ID = 0xFFFFFFFF;
+
 struct EmuThreadHandle {
     u32 host_handle;
     u32 guest_handle;

src/core/hle/kernel/client_session.cpp

@@ -47,7 +47,8 @@ ResultVal<std::shared_ptr<ClientSession>> ClientSession::Create(KernelCore& kern
     return MakeResult(std::move(client_session));
 }
 
-ResultCode ClientSession::SendSyncRequest(std::shared_ptr<Thread> thread, Memory::Memory& memory) {
+ResultCode ClientSession::SendSyncRequest(std::shared_ptr<Thread> thread,
+                                          Core::Memory::Memory& memory) {
     // Keep ServerSession alive until we're done working with it.
     if (!parent->Server()) {
         return ERR_SESSION_CLOSED_BY_REMOTE;

src/core/hle/kernel/client_session.h

@@ -12,7 +12,7 @@
 
 union ResultCode;
 
-namespace Memory {
+namespace Core::Memory {
 class Memory;
 }
 
@@ -42,7 +42,7 @@ public:
         return HANDLE_TYPE;
     }
 
-    ResultCode SendSyncRequest(std::shared_ptr<Thread> thread, Memory::Memory& memory);
+    ResultCode SendSyncRequest(std::shared_ptr<Thread> thread, Core::Memory::Memory& memory);
 
     bool ShouldWait(const Thread* thread) const override;
 

src/core/hle/kernel/errors.h

@@ -14,6 +14,7 @@ constexpr ResultCode ERR_MAX_CONNECTIONS_REACHED{ErrorModule::Kernel, 7};
 constexpr ResultCode ERR_INVALID_CAPABILITY_DESCRIPTOR{ErrorModule::Kernel, 14};
 constexpr ResultCode ERR_INVALID_SIZE{ErrorModule::Kernel, 101};
 constexpr ResultCode ERR_INVALID_ADDRESS{ErrorModule::Kernel, 102};
+constexpr ResultCode ERR_OUT_OF_RESOURCES{ErrorModule::Kernel, 103};
 constexpr ResultCode ERR_OUT_OF_MEMORY{ErrorModule::Kernel, 104};
 constexpr ResultCode ERR_HANDLE_TABLE_FULL{ErrorModule::Kernel, 105};
 constexpr ResultCode ERR_INVALID_ADDRESS_STATE{ErrorModule::Kernel, 106};

src/core/hle/kernel/hle_ipc.cpp

@@ -282,7 +282,7 @@ ResultCode HLERequestContext::WriteToOutgoingCommandBuffer(Thread& thread) {
     return RESULT_SUCCESS;
 }
 
-std::vector<u8> HLERequestContext::ReadBuffer(int buffer_index) const {
+std::vector<u8> HLERequestContext::ReadBuffer(std::size_t buffer_index) const {
     std::vector<u8> buffer;
     const bool is_buffer_a{BufferDescriptorA().size() > buffer_index &&
                            BufferDescriptorA()[buffer_index].Size()};
@@ -304,7 +304,7 @@ std::vector<u8> HLERequestContext::ReadBuffer(int buffer_index) const {
 }
 
 std::size_t HLERequestContext::WriteBuffer(const void* buffer, std::size_t size,
-                                           int buffer_index) const {
+                                           std::size_t buffer_index) const {
     if (size == 0) {
         LOG_WARNING(Core, "skip empty buffer write");
         return 0;
@@ -337,7 +337,7 @@ std::size_t HLERequestContext::WriteBuffer(const void* buffer, std::size_t size,
     return size;
 }
 
-std::size_t HLERequestContext::GetReadBufferSize(int buffer_index) const {
+std::size_t HLERequestContext::GetReadBufferSize(std::size_t buffer_index) const {
     const bool is_buffer_a{BufferDescriptorA().size() > buffer_index &&
                            BufferDescriptorA()[buffer_index].Size()};
     if (is_buffer_a) {
@@ -355,7 +355,7 @@ std::size_t HLERequestContext::GetReadBufferSize(int buffer_index) const {
     }
 }
 
-std::size_t HLERequestContext::GetWriteBufferSize(int buffer_index) const {
+std::size_t HLERequestContext::GetWriteBufferSize(std::size_t buffer_index) const {
     const bool is_buffer_b{BufferDescriptorB().size() > buffer_index &&
                            BufferDescriptorB()[buffer_index].Size()};
     if (is_buffer_b) {

src/core/hle/kernel/hle_ipc.h

@@ -179,10 +179,11 @@ public:
     }
 
     /// Helper function to read a buffer using the appropriate buffer descriptor
-    std::vector<u8> ReadBuffer(int buffer_index = 0) const;
+    std::vector<u8> ReadBuffer(std::size_t buffer_index = 0) const;
 
     /// Helper function to write a buffer using the appropriate buffer descriptor
-    std::size_t WriteBuffer(const void* buffer, std::size_t size, int buffer_index = 0) const;
+    std::size_t WriteBuffer(const void* buffer, std::size_t size,
+                            std::size_t buffer_index = 0) const;
 
     /* Helper function to write a buffer using the appropriate buffer descriptor
      *
@@ -194,7 +195,8 @@ public:
      */
     template <typename ContiguousContainer,
               typename = std::enable_if_t<!std::is_pointer_v<ContiguousContainer>>>
-    std::size_t WriteBuffer(const ContiguousContainer& container, int buffer_index = 0) const {
+    std::size_t WriteBuffer(const ContiguousContainer& container,
+                            std::size_t buffer_index = 0) const {
         using ContiguousType = typename ContiguousContainer::value_type;
 
         static_assert(std::is_trivially_copyable_v<ContiguousType>,
@@ -205,10 +207,10 @@ public:
     }
 
     /// Helper function to get the size of the input buffer
-    std::size_t GetReadBufferSize(int buffer_index = 0) const;
+    std::size_t GetReadBufferSize(std::size_t buffer_index = 0) const;
 
     /// Helper function to get the size of the output buffer
-    std::size_t GetWriteBufferSize(int buffer_index = 0) const;
+    std::size_t GetWriteBufferSize(std::size_t buffer_index = 0) const;
 
     template <typename T>
     std::shared_ptr<T> GetCopyObject(std::size_t index) {

src/core/hle/kernel/kernel.cpp

@@ -3,9 +3,12 @@
 // Refer to the license.txt file included.
 
 #include <atomic>
+#include <bitset>
 #include <functional>
 #include <memory>
 #include <mutex>
+#include <thread>
+#include <unordered_map>
 #include <utility>
 
 #include "common/assert.h"
@@ -15,16 +18,23 @@
 #include "core/core.h"
 #include "core/core_timing.h"
 #include "core/core_timing_util.h"
+#include "core/device_memory.h"
+#include "core/hardware_properties.h"
 #include "core/hle/kernel/client_port.h"
 #include "core/hle/kernel/errors.h"
 #include "core/hle/kernel/handle_table.h"
 #include "core/hle/kernel/kernel.h"
+#include "core/hle/kernel/memory/memory_layout.h"
+#include "core/hle/kernel/memory/memory_manager.h"
+#include "core/hle/kernel/memory/slab_heap.h"
 #include "core/hle/kernel/physical_core.h"
 #include "core/hle/kernel/process.h"
 #include "core/hle/kernel/resource_limit.h"
 #include "core/hle/kernel/scheduler.h"
+#include "core/hle/kernel/shared_memory.h"
 #include "core/hle/kernel/synchronization.h"
 #include "core/hle/kernel/thread.h"
+#include "core/hle/kernel/time_manager.h"
 #include "core/hle/lock.h"
 #include "core/hle/result.h"
 #include "core/memory.h"
@@ -44,7 +54,7 @@ static void ThreadWakeupCallback(u64 thread_handle, [[maybe_unused]] s64 cycles_
     std::lock_guard lock{HLE::g_hle_lock};
 
     std::shared_ptr<Thread> thread =
-        system.Kernel().RetrieveThreadFromWakeupCallbackHandleTable(proper_handle);
+        system.Kernel().RetrieveThreadFromGlobalHandleTable(proper_handle);
     if (thread == nullptr) {
         LOG_CRITICAL(Kernel, "Callback fired for invalid thread {:08X}", proper_handle);
         return;
@@ -97,14 +107,15 @@ static void ThreadWakeupCallback(u64 thread_handle, [[maybe_unused]] s64 cycles_
 }
 
 struct KernelCore::Impl {
-    explicit Impl(Core::System& system)
-        : system{system}, global_scheduler{system}, synchronization{system} {}
+    explicit Impl(Core::System& system, KernelCore& kernel)
+        : global_scheduler{kernel}, synchronization{system}, time_manager{system}, system{system} {}
 
     void Initialize(KernelCore& kernel) {
         Shutdown();
 
         InitializePhysicalCores();
         InitializeSystemResourceLimit(kernel);
+        InitializeMemoryLayout();
         InitializeThreads();
         InitializePreemption();
     }
@@ -120,7 +131,7 @@ struct KernelCore::Impl {
 
         system_resource_limit = nullptr;
 
-        thread_wakeup_callback_handle_table.Clear();
+        global_handle_table.Clear();
         thread_wakeup_event_type = nullptr;
         preemption_event = nullptr;
 
@@ -138,8 +149,8 @@ struct KernelCore::Impl {
 
     void InitializePhysicalCores() {
         exclusive_monitor =
-            Core::MakeExclusiveMonitor(system.Memory(), global_scheduler.CpuCoresCount());
-        for (std::size_t i = 0; i < global_scheduler.CpuCoresCount(); i++) {
+            Core::MakeExclusiveMonitor(system.Memory(), Core::Hardware::NUM_CPU_CORES);
+        for (std::size_t i = 0; i < Core::Hardware::NUM_CPU_CORES; i++) {
             cores.emplace_back(system, i, *exclusive_monitor);
         }
     }
@@ -149,12 +160,17 @@ struct KernelCore::Impl {
         system_resource_limit = ResourceLimit::Create(kernel);
 
         // If setting the default system values fails, then something seriously wrong has occurred.
-        ASSERT(system_resource_limit->SetLimitValue(ResourceType::PhysicalMemory, 0x200000000)
+        ASSERT(system_resource_limit->SetLimitValue(ResourceType::PhysicalMemory, 0x100000000)
                    .IsSuccess());
         ASSERT(system_resource_limit->SetLimitValue(ResourceType::Threads, 800).IsSuccess());
         ASSERT(system_resource_limit->SetLimitValue(ResourceType::Events, 700).IsSuccess());
         ASSERT(system_resource_limit->SetLimitValue(ResourceType::TransferMemory, 200).IsSuccess());
         ASSERT(system_resource_limit->SetLimitValue(ResourceType::Sessions, 900).IsSuccess());
+
+        if (!system_resource_limit->Reserve(ResourceType::PhysicalMemory, 0) ||
+            !system_resource_limit->Reserve(ResourceType::PhysicalMemory, 0x60000)) {
+            UNREACHABLE();
+        }
     }
 
     void InitializeThreads() {
@@ -181,9 +197,108 @@ struct KernelCore::Impl {
             return;
         }
 
+        for (auto& core : cores) {
+            core.SetIs64Bit(process->Is64BitProcess());
+        }
+
         system.Memory().SetCurrentPageTable(*process);
     }
 
+    void RegisterCoreThread(std::size_t core_id) {
+        std::unique_lock lock{register_thread_mutex};
+        const std::thread::id this_id = std::this_thread::get_id();
+        const auto it = host_thread_ids.find(this_id);
+        ASSERT(core_id < Core::Hardware::NUM_CPU_CORES);
+        ASSERT(it == host_thread_ids.end());
+        ASSERT(!registered_core_threads[core_id]);
+        host_thread_ids[this_id] = static_cast<u32>(core_id);
+        registered_core_threads.set(core_id);
+    }
+
+    void RegisterHostThread() {
+        std::unique_lock lock{register_thread_mutex};
+        const std::thread::id this_id = std::this_thread::get_id();
+        const auto it = host_thread_ids.find(this_id);
+        ASSERT(it == host_thread_ids.end());
+        host_thread_ids[this_id] = registered_thread_ids++;
+    }
+
+    u32 GetCurrentHostThreadID() const {
+        const std::thread::id this_id = std::this_thread::get_id();
+        const auto it = host_thread_ids.find(this_id);
+        if (it == host_thread_ids.end()) {
+            return Core::INVALID_HOST_THREAD_ID;
+        }
+        return it->second;
+    }
+
+    Core::EmuThreadHandle GetCurrentEmuThreadID() const {
+        Core::EmuThreadHandle result = Core::EmuThreadHandle::InvalidHandle();
+        result.host_handle = GetCurrentHostThreadID();
+        if (result.host_handle >= Core::Hardware::NUM_CPU_CORES) {
+            return result;
+        }
+        const Kernel::Scheduler& sched = cores[result.host_handle].Scheduler();
+        const Kernel::Thread* current = sched.GetCurrentThread();
+        if (current != nullptr) {
+            result.guest_handle = current->GetGlobalHandle();
+        } else {
+            result.guest_handle = InvalidHandle;
+        }
+        return result;
+    }
+
+    void InitializeMemoryLayout() {
+        // Initialize memory layout
+        constexpr Memory::MemoryLayout layout{Memory::MemoryLayout::GetDefaultLayout()};
+        constexpr std::size_t hid_size{0x40000};
+        constexpr std::size_t font_size{0x1100000};
+        constexpr std::size_t irs_size{0x8000};
+        constexpr std::size_t time_size{0x1000};
+        constexpr PAddr hid_addr{layout.System().StartAddress()};
+        constexpr PAddr font_pa{layout.System().StartAddress() + hid_size};
+        constexpr PAddr irs_addr{layout.System().StartAddress() + hid_size + font_size};
+        constexpr PAddr time_addr{layout.System().StartAddress() + hid_size + font_size + irs_size};
+
+        // Initialize memory manager
+        memory_manager = std::make_unique<Memory::MemoryManager>();
+        memory_manager->InitializeManager(Memory::MemoryManager::Pool::Application,
+                                          layout.Application().StartAddress(),
+                                          layout.Application().EndAddress());
+        memory_manager->InitializeManager(Memory::MemoryManager::Pool::Applet,
+                                          layout.Applet().StartAddress(),
+                                          layout.Applet().EndAddress());
+        memory_manager->InitializeManager(Memory::MemoryManager::Pool::System,
+                                          layout.System().StartAddress(),
+                                          layout.System().EndAddress());
+
+        hid_shared_mem = Kernel::SharedMemory::Create(
+            system.Kernel(), system.DeviceMemory(), nullptr,
+            {hid_addr, hid_size / Memory::PageSize}, Memory::MemoryPermission::None,
+            Memory::MemoryPermission::Read, hid_addr, hid_size, "HID:SharedMemory");
+        font_shared_mem = Kernel::SharedMemory::Create(
+            system.Kernel(), system.DeviceMemory(), nullptr,
+            {font_pa, font_size / Memory::PageSize}, Memory::MemoryPermission::None,
+            Memory::MemoryPermission::Read, font_pa, font_size, "Font:SharedMemory");
+        irs_shared_mem = Kernel::SharedMemory::Create(
+            system.Kernel(), system.DeviceMemory(), nullptr,
+            {irs_addr, irs_size / Memory::PageSize}, Memory::MemoryPermission::None,
+            Memory::MemoryPermission::Read, irs_addr, irs_size, "IRS:SharedMemory");
+        time_shared_mem = Kernel::SharedMemory::Create(
+            system.Kernel(), system.DeviceMemory(), nullptr,
+            {time_addr, time_size / Memory::PageSize}, Memory::MemoryPermission::None,
+            Memory::MemoryPermission::Read, time_addr, time_size, "Time:SharedMemory");
+
+        // Allocate slab heaps
+        user_slab_heap_pages = std::make_unique<Memory::SlabHeap<Memory::Page>>();
+
+        // Initialize slab heaps
+        constexpr u64 user_slab_heap_size{0x3de000};
+        user_slab_heap_pages->Initialize(
+            system.DeviceMemory().GetPointer(Core::DramMemoryMap::SlabHeapBase),
+            user_slab_heap_size);
+    }
+
     std::atomic<u32> next_object_id{0};
     std::atomic<u64> next_kernel_process_id{Process::InitialKIPIDMin};
     std::atomic<u64> next_user_process_id{Process::ProcessIDMin};
@@ -194,15 +309,16 @@ struct KernelCore::Impl {
     Process* current_process = nullptr;
     Kernel::GlobalScheduler global_scheduler;
     Kernel::Synchronization synchronization;
+    Kernel::TimeManager time_manager;
 
     std::shared_ptr<ResourceLimit> system_resource_limit;
 
     std::shared_ptr<Core::Timing::EventType> thread_wakeup_event_type;
     std::shared_ptr<Core::Timing::EventType> preemption_event;
 
-    // TODO(yuriks): This can be removed if Thread objects are explicitly pooled in the future,
-    // allowing us to simply use a pool index or similar.
-    Kernel::HandleTable thread_wakeup_callback_handle_table;
+    // This is the kernel's handle table or supervisor handle table which
+    // stores all the objects in place.
+    Kernel::HandleTable global_handle_table;
 
     /// Map of named ports managed by the kernel, which can be retrieved using
     /// the ConnectToPort SVC.
@@ -211,11 +327,27 @@ struct KernelCore::Impl {
     std::unique_ptr<Core::ExclusiveMonitor> exclusive_monitor;
     std::vector<Kernel::PhysicalCore> cores;
 
+    // 0-3 IDs represent core threads, >3 represent others
+    std::unordered_map<std::thread::id, u32> host_thread_ids;
+    u32 registered_thread_ids{Core::Hardware::NUM_CPU_CORES};
+    std::bitset<Core::Hardware::NUM_CPU_CORES> registered_core_threads;
+    std::mutex register_thread_mutex;
+
+    // Kernel memory management
+    std::unique_ptr<Memory::MemoryManager> memory_manager;
+    std::unique_ptr<Memory::SlabHeap<Memory::Page>> user_slab_heap_pages;
+
+    // Shared memory for services
+    std::shared_ptr<Kernel::SharedMemory> hid_shared_mem;
+    std::shared_ptr<Kernel::SharedMemory> font_shared_mem;
+    std::shared_ptr<Kernel::SharedMemory> irs_shared_mem;
+    std::shared_ptr<Kernel::SharedMemory> time_shared_mem;
+
     // System context
     Core::System& system;
 };
 
-KernelCore::KernelCore(Core::System& system) : impl{std::make_unique<Impl>(system)} {}
+KernelCore::KernelCore(Core::System& system) : impl{std::make_unique<Impl>(system, *this)} {}
 KernelCore::~KernelCore() {
     Shutdown();
 }
@@ -232,9 +364,8 @@ std::shared_ptr<ResourceLimit> KernelCore::GetSystemResourceLimit() const {
     return impl->system_resource_limit;
 }
 
-std::shared_ptr<Thread> KernelCore::RetrieveThreadFromWakeupCallbackHandleTable(
-    Handle handle) const {
-    return impl->thread_wakeup_callback_handle_table.Get<Thread>(handle);
+std::shared_ptr<Thread> KernelCore::RetrieveThreadFromGlobalHandleTable(Handle handle) const {
+    return impl->global_handle_table.Get<Thread>(handle);
 }
 
 void KernelCore::AppendNewProcess(std::shared_ptr<Process> process) {
@@ -265,6 +396,14 @@ const Kernel::GlobalScheduler& KernelCore::GlobalScheduler() const {
     return impl->global_scheduler;
 }
 
+Kernel::Scheduler& KernelCore::Scheduler(std::size_t id) {
+    return impl->cores[id].Scheduler();
+}
+
+const Kernel::Scheduler& KernelCore::Scheduler(std::size_t id) const {
+    return impl->cores[id].Scheduler();
+}
+
 Kernel::PhysicalCore& KernelCore::PhysicalCore(std::size_t id) {
     return impl->cores[id];
 }
@@ -281,6 +420,14 @@ const Kernel::Synchronization& KernelCore::Synchronization() const {
     return impl->synchronization;
 }
 
+Kernel::TimeManager& KernelCore::TimeManager() {
+    return impl->time_manager;
+}
+
+const Kernel::TimeManager& KernelCore::TimeManager() const {
+    return impl->time_manager;
+}
+
 Core::ExclusiveMonitor& KernelCore::GetExclusiveMonitor() {
     return *impl->exclusive_monitor;
 }
@@ -338,12 +485,76 @@ const std::shared_ptr<Core::Timing::EventType>& KernelCore::ThreadWakeupCallback
     return impl->thread_wakeup_event_type;
 }
 
-Kernel::HandleTable& KernelCore::ThreadWakeupCallbackHandleTable() {
-    return impl->thread_wakeup_callback_handle_table;
+Kernel::HandleTable& KernelCore::GlobalHandleTable() {
+    return impl->global_handle_table;
 }
 
-const Kernel::HandleTable& KernelCore::ThreadWakeupCallbackHandleTable() const {
-    return impl->thread_wakeup_callback_handle_table;
+const Kernel::HandleTable& KernelCore::GlobalHandleTable() const {
+    return impl->global_handle_table;
+}
+
+void KernelCore::RegisterCoreThread(std::size_t core_id) {
+    impl->RegisterCoreThread(core_id);
+}
+
+void KernelCore::RegisterHostThread() {
+    impl->RegisterHostThread();
+}
+
+u32 KernelCore::GetCurrentHostThreadID() const {
+    return impl->GetCurrentHostThreadID();
+}
+
+Core::EmuThreadHandle KernelCore::GetCurrentEmuThreadID() const {
+    return impl->GetCurrentEmuThreadID();
+}
+
+Memory::MemoryManager& KernelCore::MemoryManager() {
+    return *impl->memory_manager;
+}
+
+const Memory::MemoryManager& KernelCore::MemoryManager() const {
+    return *impl->memory_manager;
+}
+
+Memory::SlabHeap<Memory::Page>& KernelCore::GetUserSlabHeapPages() {
+    return *impl->user_slab_heap_pages;
+}
+
+const Memory::SlabHeap<Memory::Page>& KernelCore::GetUserSlabHeapPages() const {
+    return *impl->user_slab_heap_pages;
+}
+
+Kernel::SharedMemory& KernelCore::GetHidSharedMem() {
+    return *impl->hid_shared_mem;
+}
+
+const Kernel::SharedMemory& KernelCore::GetHidSharedMem() const {
+    return *impl->hid_shared_mem;
+}
+
+Kernel::SharedMemory& KernelCore::GetFontSharedMem() {
+    return *impl->font_shared_mem;
+}
+
+const Kernel::SharedMemory& KernelCore::GetFontSharedMem() const {
+    return *impl->font_shared_mem;
+}
+
+Kernel::SharedMemory& KernelCore::GetIrsSharedMem() {
+    return *impl->irs_shared_mem;
+}
+
+const Kernel::SharedMemory& KernelCore::GetIrsSharedMem() const {
+    return *impl->irs_shared_mem;
+}
+
+Kernel::SharedMemory& KernelCore::GetTimeSharedMem() {
+    return *impl->time_shared_mem;
+}
+
+const Kernel::SharedMemory& KernelCore::GetTimeSharedMem() const {
+    return *impl->time_shared_mem;
 }
 
 } // namespace Kernel

src/core/hle/kernel/kernel.h

@@ -8,9 +8,11 @@
 #include <string>
 #include <unordered_map>
 #include <vector>
+#include "core/hle/kernel/memory/memory_types.h"
 #include "core/hle/kernel/object.h"
 
 namespace Core {
+struct EmuThreadHandle;
 class ExclusiveMonitor;
 class System;
 } // namespace Core
@@ -22,6 +24,12 @@ struct EventType;
 
 namespace Kernel {
 
+namespace Memory {
+class MemoryManager;
+template <typename T>
+class SlabHeap;
+} // namespace Memory
+
 class AddressArbiter;
 class ClientPort;
 class GlobalScheduler;
@@ -29,8 +37,11 @@ class HandleTable;
 class PhysicalCore;
 class Process;
 class ResourceLimit;
+class Scheduler;
+class SharedMemory;
 class Synchronization;
 class Thread;
+class TimeManager;
 
 /// Represents a single instance of the kernel.
 class KernelCore {
@@ -64,7 +75,7 @@ public:
     std::shared_ptr<ResourceLimit> GetSystemResourceLimit() const;
 
     /// Retrieves a shared pointer to a Thread instance within the thread wakeup handle table.
-    std::shared_ptr<Thread> RetrieveThreadFromWakeupCallbackHandleTable(Handle handle) const;
+    std::shared_ptr<Thread> RetrieveThreadFromGlobalHandleTable(Handle handle) const;
 
     /// Adds the given shared pointer to an internal list of active processes.
     void AppendNewProcess(std::shared_ptr<Process> process);
@@ -87,6 +98,12 @@ public:
     /// Gets the sole instance of the global scheduler
     const Kernel::GlobalScheduler& GlobalScheduler() const;
 
+    /// Gets the sole instance of the Scheduler assoviated with cpu core 'id'
+    Kernel::Scheduler& Scheduler(std::size_t id);
+
+    /// Gets the sole instance of the Scheduler assoviated with cpu core 'id'
+    const Kernel::Scheduler& Scheduler(std::size_t id) const;
+
     /// Gets the an instance of the respective physical CPU core.
     Kernel::PhysicalCore& PhysicalCore(std::size_t id);
 
@@ -99,6 +116,12 @@ public:
     /// Gets the an instance of the Synchronization Interface.
     const Kernel::Synchronization& Synchronization() const;
 
+    /// Gets the an instance of the TimeManager Interface.
+    Kernel::TimeManager& TimeManager();
+
+    /// Gets the an instance of the TimeManager Interface.
+    const Kernel::TimeManager& TimeManager() const;
+
     /// Stops execution of 'id' core, in order to reschedule a new thread.
     void PrepareReschedule(std::size_t id);
 
@@ -120,6 +143,54 @@ public:
     /// Determines whether or not the given port is a valid named port.
     bool IsValidNamedPort(NamedPortTable::const_iterator port) const;
 
+    /// Gets the current host_thread/guest_thread handle.
+    Core::EmuThreadHandle GetCurrentEmuThreadID() const;
+
+    /// Gets the current host_thread handle.
+    u32 GetCurrentHostThreadID() const;
+
+    /// Register the current thread as a CPU Core Thread.
+    void RegisterCoreThread(std::size_t core_id);
+
+    /// Register the current thread as a non CPU core thread.
+    void RegisterHostThread();
+
+    /// Gets the virtual memory manager for the kernel.
+    Memory::MemoryManager& MemoryManager();
+
+    /// Gets the virtual memory manager for the kernel.
+    const Memory::MemoryManager& MemoryManager() const;
+
+    /// Gets the slab heap allocated for user space pages.
+    Memory::SlabHeap<Memory::Page>& GetUserSlabHeapPages();
+
+    /// Gets the slab heap allocated for user space pages.
+    const Memory::SlabHeap<Memory::Page>& GetUserSlabHeapPages() const;
+
+    /// Gets the shared memory object for HID services.
+    Kernel::SharedMemory& GetHidSharedMem();
+
+    /// Gets the shared memory object for HID services.
+    const Kernel::SharedMemory& GetHidSharedMem() const;
+
+    /// Gets the shared memory object for font services.
+    Kernel::SharedMemory& GetFontSharedMem();
+
+    /// Gets the shared memory object for font services.
+    const Kernel::SharedMemory& GetFontSharedMem() const;
+
+    /// Gets the shared memory object for IRS services.
+    Kernel::SharedMemory& GetIrsSharedMem();
+
+    /// Gets the shared memory object for IRS services.
+    const Kernel::SharedMemory& GetIrsSharedMem() const;
+
+    /// Gets the shared memory object for Time services.
+    Kernel::SharedMemory& GetTimeSharedMem();
+
+    /// Gets the shared memory object for Time services.
+    const Kernel::SharedMemory& GetTimeSharedMem() const;
+
 private:
     friend class Object;
     friend class Process;
@@ -140,11 +211,11 @@ private:
     /// Retrieves the event type used for thread wakeup callbacks.
     const std::shared_ptr<Core::Timing::EventType>& ThreadWakeupCallbackEventType() const;
 
-    /// Provides a reference to the thread wakeup callback handle table.
-    Kernel::HandleTable& ThreadWakeupCallbackHandleTable();
+    /// Provides a reference to the global handle table.
+    Kernel::HandleTable& GlobalHandleTable();
 
-    /// Provides a const reference to the thread wakeup callback handle table.
-    const Kernel::HandleTable& ThreadWakeupCallbackHandleTable() const;
+    /// Provides a const reference to the global handle table.
+    const Kernel::HandleTable& GlobalHandleTable() const;
 
     struct Impl;
     std::unique_ptr<Impl> impl;

src/core/hle/kernel/memory/address_space_info.cpp

@@ -0,0 +1,118 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+// This file references various implementation details from Atmosph<70>re, an open-source firmware for
+// the Nintendo Switch. Copyright 2018-2020 Atmosph<70>re-NX.
+
+#include <array>
+
+#include "common/assert.h"
+#include "core/hle/kernel/memory/address_space_info.h"
+
+namespace Kernel::Memory {
+
+namespace {
+
+enum : u64 {
+    Size_1_MB = 0x100000,
+    Size_2_MB = 2 * Size_1_MB,
+    Size_128_MB = 128 * Size_1_MB,
+    Size_1_GB = 0x40000000,
+    Size_2_GB = 2 * Size_1_GB,
+    Size_4_GB = 4 * Size_1_GB,
+    Size_6_GB = 6 * Size_1_GB,
+    Size_64_GB = 64 * Size_1_GB,
+    Size_512_GB = 512 * Size_1_GB,
+    Invalid = std::numeric_limits<u64>::max(),
+};
+
+// clang-format off
+constexpr std::array<AddressSpaceInfo, 13> AddressSpaceInfos{{
+   { 32 /*bit_width*/, Size_2_MB   /*addr*/, Size_1_GB   - Size_2_MB   /*size*/, AddressSpaceInfo::Type::Is32Bit,    },
+   { 32 /*bit_width*/, Size_1_GB   /*addr*/, Size_4_GB   - Size_1_GB   /*size*/, AddressSpaceInfo::Type::Small64Bit, },
+   { 32 /*bit_width*/, Invalid     /*addr*/, Size_1_GB                 /*size*/, AddressSpaceInfo::Type::Heap,       },
+   { 32 /*bit_width*/, Invalid     /*addr*/, Size_1_GB                 /*size*/, AddressSpaceInfo::Type::Alias,      },
+   { 36 /*bit_width*/, Size_128_MB /*addr*/, Size_2_GB   - Size_128_MB /*size*/, AddressSpaceInfo::Type::Is32Bit,    },
+   { 36 /*bit_width*/, Size_2_GB   /*addr*/, Size_64_GB  - Size_2_GB   /*size*/, AddressSpaceInfo::Type::Small64Bit, },
+   { 36 /*bit_width*/, Invalid     /*addr*/, Size_6_GB                 /*size*/, AddressSpaceInfo::Type::Heap,       },
+   { 36 /*bit_width*/, Invalid     /*addr*/, Size_6_GB                 /*size*/, AddressSpaceInfo::Type::Alias,      },
+   { 39 /*bit_width*/, Size_128_MB /*addr*/, Size_512_GB - Size_128_MB /*size*/, AddressSpaceInfo::Type::Large64Bit, },
+   { 39 /*bit_width*/, Invalid     /*addr*/, Size_64_GB                /*size*/, AddressSpaceInfo::Type::Is32Bit     },
+   { 39 /*bit_width*/, Invalid     /*addr*/, Size_6_GB                 /*size*/, AddressSpaceInfo::Type::Heap,       },
+   { 39 /*bit_width*/, Invalid     /*addr*/, Size_64_GB                /*size*/, AddressSpaceInfo::Type::Alias,      },
+   { 39 /*bit_width*/, Invalid     /*addr*/, Size_2_GB                 /*size*/, AddressSpaceInfo::Type::Stack,      },
+}};
+// clang-format on
+
+constexpr bool IsAllowedIndexForAddress(std::size_t index) {
+    return index < std::size(AddressSpaceInfos) && AddressSpaceInfos[index].GetAddress() != Invalid;
+}
+
+constexpr std::size_t
+    AddressSpaceIndices32Bit[static_cast<std::size_t>(AddressSpaceInfo::Type::Count)]{
+        0, 1, 0, 2, 0, 3,
+    };
+
+constexpr std::size_t
+    AddressSpaceIndices36Bit[static_cast<std::size_t>(AddressSpaceInfo::Type::Count)]{
+        4, 5, 4, 6, 4, 7,
+    };
+
+constexpr std::size_t
+    AddressSpaceIndices39Bit[static_cast<std::size_t>(AddressSpaceInfo::Type::Count)]{
+        9, 8, 8, 10, 12, 11,
+    };
+
+constexpr bool IsAllowed32BitType(AddressSpaceInfo::Type type) {
+    return type < AddressSpaceInfo::Type::Count && type != AddressSpaceInfo::Type::Large64Bit &&
+           type != AddressSpaceInfo::Type::Stack;
+}
+
+constexpr bool IsAllowed36BitType(AddressSpaceInfo::Type type) {
+    return type < AddressSpaceInfo::Type::Count && type != AddressSpaceInfo::Type::Large64Bit &&
+           type != AddressSpaceInfo::Type::Stack;
+}
+
+constexpr bool IsAllowed39BitType(AddressSpaceInfo::Type type) {
+    return type < AddressSpaceInfo::Type::Count && type != AddressSpaceInfo::Type::Small64Bit;
+}
+
+} // namespace
+
+u64 AddressSpaceInfo::GetAddressSpaceStart(std::size_t width, AddressSpaceInfo::Type type) {
+    const std::size_t index{static_cast<std::size_t>(type)};
+    switch (width) {
+    case 32:
+        ASSERT(IsAllowed32BitType(type));
+        ASSERT(IsAllowedIndexForAddress(AddressSpaceIndices32Bit[index]));
+        return AddressSpaceInfos[AddressSpaceIndices32Bit[index]].GetAddress();
+    case 36:
+        ASSERT(IsAllowed36BitType(type));
+        ASSERT(IsAllowedIndexForAddress(AddressSpaceIndices36Bit[index]));
+        return AddressSpaceInfos[AddressSpaceIndices36Bit[index]].GetAddress();
+    case 39:
+        ASSERT(IsAllowed39BitType(type));
+        ASSERT(IsAllowedIndexForAddress(AddressSpaceIndices39Bit[index]));
+        return AddressSpaceInfos[AddressSpaceIndices39Bit[index]].GetAddress();
+    }
+    UNREACHABLE();
+}
+
+std::size_t AddressSpaceInfo::GetAddressSpaceSize(std::size_t width, AddressSpaceInfo::Type type) {
+    const std::size_t index{static_cast<std::size_t>(type)};
+    switch (width) {
+    case 32:
+        ASSERT(IsAllowed32BitType(type));
+        return AddressSpaceInfos[AddressSpaceIndices32Bit[index]].GetSize();
+    case 36:
+        ASSERT(IsAllowed36BitType(type));
+        return AddressSpaceInfos[AddressSpaceIndices36Bit[index]].GetSize();
+    case 39:
+        ASSERT(IsAllowed39BitType(type));
+        return AddressSpaceInfos[AddressSpaceIndices39Bit[index]].GetSize();
+    }
+    UNREACHABLE();
+}
+
+} // namespace Kernel::Memory

src/core/hle/kernel/memory/address_space_info.h

@@ -0,0 +1,54 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+// This file references various implementation details from Atmosph<70>re, an open-source firmware for
+// the Nintendo Switch. Copyright 2018-2020 Atmosph<70>re-NX.
+
+#pragma once
+
+#include "common/common_funcs.h"
+#include "common/common_types.h"
+
+namespace Kernel::Memory {
+
+class AddressSpaceInfo final : NonCopyable {
+public:
+    enum class Type : u32 {
+        Is32Bit = 0,
+        Small64Bit = 1,
+        Large64Bit = 2,
+        Heap = 3,
+        Stack = 4,
+        Alias = 5,
+        Count,
+    };
+
+private:
+    std::size_t bit_width{};
+    std::size_t addr{};
+    std::size_t size{};
+    Type type{};
+
+public:
+    static u64 GetAddressSpaceStart(std::size_t width, Type type);
+    static std::size_t GetAddressSpaceSize(std::size_t width, Type type);
+
+    constexpr AddressSpaceInfo(std::size_t bit_width, std::size_t addr, std::size_t size, Type type)
+        : bit_width{bit_width}, addr{addr}, size{size}, type{type} {}
+
+    constexpr std::size_t GetWidth() const {
+        return bit_width;
+    }
+    constexpr std::size_t GetAddress() const {
+        return addr;
+    }
+    constexpr std::size_t GetSize() const {
+        return size;
+    }
+    constexpr Type GetType() const {
+        return type;
+    }
+};
+
+} // namespace Kernel::Memory

src/core/hle/kernel/memory/memory_block.h

@@ -0,0 +1,335 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+// This file references various implementation details from Atmosph<70>re, an open-source firmware for
+// the Nintendo Switch. Copyright 2018-2020 Atmosph<70>re-NX.
+
+#pragma once
+
+#include "common/alignment.h"
+#include "common/assert.h"
+#include "common/common_types.h"
+#include "core/hle/kernel/memory/memory_types.h"
+#include "core/hle/kernel/svc_types.h"
+
+namespace Kernel::Memory {
+
+enum class MemoryState : u32 {
+    None = 0,
+    Mask = 0xFF,
+    All = ~None,
+
+    FlagCanReprotect = (1 << 8),
+    FlagCanDebug = (1 << 9),
+    FlagCanUseIpc = (1 << 10),
+    FlagCanUseNonDeviceIpc = (1 << 11),
+    FlagCanUseNonSecureIpc = (1 << 12),
+    FlagMapped = (1 << 13),
+    FlagCode = (1 << 14),
+    FlagCanAlias = (1 << 15),
+    FlagCanCodeAlias = (1 << 16),
+    FlagCanTransfer = (1 << 17),
+    FlagCanQueryPhysical = (1 << 18),
+    FlagCanDeviceMap = (1 << 19),
+    FlagCanAlignedDeviceMap = (1 << 20),
+    FlagCanIpcUserBuffer = (1 << 21),
+    FlagReferenceCounted = (1 << 22),
+    FlagCanMapProcess = (1 << 23),
+    FlagCanChangeAttribute = (1 << 24),
+    FlagCanCodeMemory = (1 << 25),
+
+    FlagsData = FlagCanReprotect | FlagCanUseIpc | FlagCanUseNonDeviceIpc | FlagCanUseNonSecureIpc |
+                FlagMapped | FlagCanAlias | FlagCanTransfer | FlagCanQueryPhysical |
+                FlagCanDeviceMap | FlagCanAlignedDeviceMap | FlagCanIpcUserBuffer |
+                FlagReferenceCounted | FlagCanChangeAttribute,
+
+    FlagsCode = FlagCanDebug | FlagCanUseIpc | FlagCanUseNonDeviceIpc | FlagCanUseNonSecureIpc |
+                FlagMapped | FlagCode | FlagCanQueryPhysical | FlagCanDeviceMap |
+                FlagCanAlignedDeviceMap | FlagReferenceCounted,
+
+    FlagsMisc = FlagMapped | FlagReferenceCounted | FlagCanQueryPhysical | FlagCanDeviceMap,
+
+    Free = static_cast<u32>(Svc::MemoryState::Free),
+    Io = static_cast<u32>(Svc::MemoryState::Io) | FlagMapped,
+    Static = static_cast<u32>(Svc::MemoryState::Static) | FlagMapped | FlagCanQueryPhysical,
+    Code = static_cast<u32>(Svc::MemoryState::Code) | FlagsCode | FlagCanMapProcess,
+    CodeData = static_cast<u32>(Svc::MemoryState::CodeData) | FlagsData | FlagCanMapProcess |
+               FlagCanCodeMemory,
+    Shared = static_cast<u32>(Svc::MemoryState::Shared) | FlagMapped | FlagReferenceCounted,
+    Normal = static_cast<u32>(Svc::MemoryState::Normal) | FlagsData | FlagCanCodeMemory,
+
+    AliasCode = static_cast<u32>(Svc::MemoryState::AliasCode) | FlagsCode | FlagCanMapProcess |
+                FlagCanCodeAlias,
+    AliasCodeData = static_cast<u32>(Svc::MemoryState::AliasCodeData) | FlagsData |
+                    FlagCanMapProcess | FlagCanCodeAlias | FlagCanCodeMemory,
+
+    Ipc = static_cast<u32>(Svc::MemoryState::Ipc) | FlagsMisc | FlagCanAlignedDeviceMap |
+          FlagCanUseIpc | FlagCanUseNonSecureIpc | FlagCanUseNonDeviceIpc,
+
+    Stack = static_cast<u32>(Svc::MemoryState::Stack) | FlagsMisc | FlagCanAlignedDeviceMap |
+            FlagCanUseIpc | FlagCanUseNonSecureIpc | FlagCanUseNonDeviceIpc,
+
+    ThreadLocal =
+        static_cast<u32>(Svc::MemoryState::ThreadLocal) | FlagMapped | FlagReferenceCounted,
+
+    Transfered = static_cast<u32>(Svc::MemoryState::Transfered) | FlagsMisc |
+                 FlagCanAlignedDeviceMap | FlagCanChangeAttribute | FlagCanUseIpc |
+                 FlagCanUseNonSecureIpc | FlagCanUseNonDeviceIpc,
+
+    SharedTransfered = static_cast<u32>(Svc::MemoryState::SharedTransfered) | FlagsMisc |
+                       FlagCanAlignedDeviceMap | FlagCanUseNonSecureIpc | FlagCanUseNonDeviceIpc,
+
+    SharedCode = static_cast<u32>(Svc::MemoryState::SharedCode) | FlagMapped |
+                 FlagReferenceCounted | FlagCanUseNonSecureIpc | FlagCanUseNonDeviceIpc,
+
+    Inaccessible = static_cast<u32>(Svc::MemoryState::Inaccessible),
+
+    NonSecureIpc = static_cast<u32>(Svc::MemoryState::NonSecureIpc) | FlagsMisc |
+                   FlagCanAlignedDeviceMap | FlagCanUseNonSecureIpc | FlagCanUseNonDeviceIpc,
+
+    NonDeviceIpc =
+        static_cast<u32>(Svc::MemoryState::NonDeviceIpc) | FlagsMisc | FlagCanUseNonDeviceIpc,
+
+    Kernel = static_cast<u32>(Svc::MemoryState::Kernel) | FlagMapped,
+
+    GeneratedCode = static_cast<u32>(Svc::MemoryState::GeneratedCode) | FlagMapped |
+                    FlagReferenceCounted | FlagCanDebug,
+    CodeOut = static_cast<u32>(Svc::MemoryState::CodeOut) | FlagMapped | FlagReferenceCounted,
+};
+DECLARE_ENUM_FLAG_OPERATORS(MemoryState);
+
+static_assert(static_cast<u32>(MemoryState::Free) == 0x00000000);
+static_assert(static_cast<u32>(MemoryState::Io) == 0x00002001);
+static_assert(static_cast<u32>(MemoryState::Static) == 0x00042002);
+static_assert(static_cast<u32>(MemoryState::Code) == 0x00DC7E03);
+static_assert(static_cast<u32>(MemoryState::CodeData) == 0x03FEBD04);
+static_assert(static_cast<u32>(MemoryState::Normal) == 0x037EBD05);
+static_assert(static_cast<u32>(MemoryState::Shared) == 0x00402006);
+static_assert(static_cast<u32>(MemoryState::AliasCode) == 0x00DD7E08);
+static_assert(static_cast<u32>(MemoryState::AliasCodeData) == 0x03FFBD09);
+static_assert(static_cast<u32>(MemoryState::Ipc) == 0x005C3C0A);
+static_assert(static_cast<u32>(MemoryState::Stack) == 0x005C3C0B);
+static_assert(static_cast<u32>(MemoryState::ThreadLocal) == 0x0040200C);
+static_assert(static_cast<u32>(MemoryState::Transfered) == 0x015C3C0D);
+static_assert(static_cast<u32>(MemoryState::SharedTransfered) == 0x005C380E);
+static_assert(static_cast<u32>(MemoryState::SharedCode) == 0x0040380F);
+static_assert(static_cast<u32>(MemoryState::Inaccessible) == 0x00000010);
+static_assert(static_cast<u32>(MemoryState::NonSecureIpc) == 0x005C3811);
+static_assert(static_cast<u32>(MemoryState::NonDeviceIpc) == 0x004C2812);
+static_assert(static_cast<u32>(MemoryState::Kernel) == 0x00002013);
+static_assert(static_cast<u32>(MemoryState::GeneratedCode) == 0x00402214);
+static_assert(static_cast<u32>(MemoryState::CodeOut) == 0x00402015);
+
+enum class MemoryPermission : u8 {
+    None = 0,
+    Mask = static_cast<u8>(~None),
+
+    Read = 1 << 0,
+    Write = 1 << 1,
+    Execute = 1 << 2,
+
+    ReadAndWrite = Read | Write,
+    ReadAndExecute = Read | Execute,
+
+    UserMask = static_cast<u8>(Svc::MemoryPermission::Read | Svc::MemoryPermission::Write |
+                               Svc::MemoryPermission::Execute),
+};
+DECLARE_ENUM_FLAG_OPERATORS(MemoryPermission);
+
+enum class MemoryAttribute : u8 {
+    None = 0x00,
+    Mask = 0x7F,
+    All = Mask,
+    DontCareMask = 0x80,
+
+    Locked = static_cast<u8>(Svc::MemoryAttribute::Locked),
+    IpcLocked = static_cast<u8>(Svc::MemoryAttribute::IpcLocked),
+    DeviceShared = static_cast<u8>(Svc::MemoryAttribute::DeviceShared),
+    Uncached = static_cast<u8>(Svc::MemoryAttribute::Uncached),
+
+    IpcAndDeviceMapped = IpcLocked | DeviceShared,
+    LockedAndIpcLocked = Locked | IpcLocked,
+    DeviceSharedAndUncached = DeviceShared | Uncached
+};
+DECLARE_ENUM_FLAG_OPERATORS(MemoryAttribute);
+
+static_assert((static_cast<u8>(MemoryAttribute::Mask) &
+               static_cast<u8>(MemoryAttribute::DontCareMask)) == 0);
+
+struct MemoryInfo {
+    VAddr addr{};
+    std::size_t size{};
+    MemoryState state{};
+    MemoryPermission perm{};
+    MemoryAttribute attribute{};
+    MemoryPermission original_perm{};
+    u16 ipc_lock_count{};
+    u16 device_use_count{};
+
+    constexpr Svc::MemoryInfo GetSvcMemoryInfo() const {
+        return {
+            addr,
+            size,
+            static_cast<Svc::MemoryState>(state & MemoryState::Mask),
+            static_cast<Svc::MemoryAttribute>(attribute & MemoryAttribute::Mask),
+            static_cast<Svc::MemoryPermission>(perm & MemoryPermission::UserMask),
+            ipc_lock_count,
+            device_use_count,
+        };
+    }
+
+    constexpr VAddr GetAddress() const {
+        return addr;
+    }
+    constexpr std::size_t GetSize() const {
+        return size;
+    }
+    constexpr std::size_t GetNumPages() const {
+        return GetSize() / PageSize;
+    }
+    constexpr VAddr GetEndAddress() const {
+        return GetAddress() + GetSize();
+    }
+    constexpr VAddr GetLastAddress() const {
+        return GetEndAddress() - 1;
+    }
+};
+
+class MemoryBlock final {
+    friend class MemoryBlockManager;
+
+private:
+    VAddr addr{};
+    std::size_t num_pages{};
+    MemoryState state{MemoryState::None};
+    u16 ipc_lock_count{};
+    u16 device_use_count{};
+    MemoryPermission perm{MemoryPermission::None};
+    MemoryPermission original_perm{MemoryPermission::None};
+    MemoryAttribute attribute{MemoryAttribute::None};
+
+public:
+    static constexpr int Compare(const MemoryBlock& lhs, const MemoryBlock& rhs) {
+        if (lhs.GetAddress() < rhs.GetAddress()) {
+            return -1;
+        } else if (lhs.GetAddress() <= rhs.GetLastAddress()) {
+            return 0;
+        } else {
+            return 1;
+        }
+    }
+
+public:
+    constexpr MemoryBlock() = default;
+    constexpr MemoryBlock(VAddr addr, std::size_t num_pages, MemoryState state,
+                          MemoryPermission perm, MemoryAttribute attribute)
+        : addr{addr}, num_pages(num_pages), state{state}, perm{perm}, attribute{attribute} {}
+
+    constexpr VAddr GetAddress() const {
+        return addr;
+    }
+
+    constexpr std::size_t GetNumPages() const {
+        return num_pages;
+    }
+
+    constexpr std::size_t GetSize() const {
+        return GetNumPages() * PageSize;
+    }
+
+    constexpr VAddr GetEndAddress() const {
+        return GetAddress() + GetSize();
+    }
+
+    constexpr VAddr GetLastAddress() const {
+        return GetEndAddress() - 1;
+    }
+
+    constexpr MemoryInfo GetMemoryInfo() const {
+        return {
+            GetAddress(), GetSize(),     state,          perm,
+            attribute,    original_perm, ipc_lock_count, device_use_count,
+        };
+    }
+
+    void ShareToDevice(MemoryPermission /*new_perm*/) {
+        ASSERT((attribute & MemoryAttribute::DeviceShared) == MemoryAttribute::DeviceShared ||
+               device_use_count == 0);
+        attribute |= MemoryAttribute::DeviceShared;
+        const u16 new_use_count{++device_use_count};
+        ASSERT(new_use_count > 0);
+    }
+
+    void UnshareToDevice(MemoryPermission /*new_perm*/) {
+        ASSERT((attribute & MemoryAttribute::DeviceShared) == MemoryAttribute::DeviceShared);
+        const u16 prev_use_count{device_use_count--};
+        ASSERT(prev_use_count > 0);
+        if (prev_use_count == 1) {
+            attribute &= ~MemoryAttribute::DeviceShared;
+        }
+    }
+
+private:
+    constexpr bool HasProperties(MemoryState s, MemoryPermission p, MemoryAttribute a) const {
+        constexpr MemoryAttribute AttributeIgnoreMask{MemoryAttribute::DontCareMask |
+                                                      MemoryAttribute::IpcLocked |
+                                                      MemoryAttribute::DeviceShared};
+        return state == s && perm == p &&
+               (attribute | AttributeIgnoreMask) == (a | AttributeIgnoreMask);
+    }
+
+    constexpr bool HasSameProperties(const MemoryBlock& rhs) const {
+        return state == rhs.state && perm == rhs.perm && original_perm == rhs.original_perm &&
+               attribute == rhs.attribute && ipc_lock_count == rhs.ipc_lock_count &&
+               device_use_count == rhs.device_use_count;
+    }
+
+    constexpr bool Contains(VAddr start) const {
+        return GetAddress() <= start && start <= GetEndAddress();
+    }
+
+    constexpr void Add(std::size_t count) {
+        ASSERT(count > 0);
+        ASSERT(GetAddress() + count * PageSize - 1 < GetEndAddress() + count * PageSize - 1);
+
+        num_pages += count;
+    }
+
+    constexpr void Update(MemoryState new_state, MemoryPermission new_perm,
+                          MemoryAttribute new_attribute) {
+        ASSERT(original_perm == MemoryPermission::None);
+        ASSERT((attribute & MemoryAttribute::IpcLocked) == MemoryAttribute::None);
+
+        state = new_state;
+        perm = new_perm;
+
+        attribute = static_cast<MemoryAttribute>(
+            new_attribute |
+            (attribute & (MemoryAttribute::IpcLocked | MemoryAttribute::DeviceShared)));
+    }
+
+    constexpr MemoryBlock Split(VAddr split_addr) {
+        ASSERT(GetAddress() < split_addr);
+        ASSERT(Contains(split_addr));
+        ASSERT(Common::IsAligned(split_addr, PageSize));
+
+        MemoryBlock block;
+        block.addr = addr;
+        block.num_pages = (split_addr - GetAddress()) / PageSize;
+        block.state = state;
+        block.ipc_lock_count = ipc_lock_count;
+        block.device_use_count = device_use_count;
+        block.perm = perm;
+        block.original_perm = original_perm;
+        block.attribute = attribute;
+
+        addr = split_addr;
+        num_pages -= block.num_pages;
+
+        return block;
+    }
+};
+static_assert(std::is_trivially_destructible<MemoryBlock>::value);
+
+} // namespace Kernel::Memory

src/core/hle/kernel/memory/memory_block_manager.cpp

@@ -0,0 +1,226 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "core/hle/kernel/memory/memory_block_manager.h"
+#include "core/hle/kernel/memory/memory_types.h"
+
+namespace Kernel::Memory {
+
+MemoryBlockManager::MemoryBlockManager(VAddr start_addr, VAddr end_addr)
+    : start_addr{start_addr}, end_addr{end_addr} {
+    const u64 num_pages{(end_addr - start_addr) / PageSize};
+    memory_block_tree.emplace_back(start_addr, num_pages, MemoryState::Free, MemoryPermission::None,
+                                   MemoryAttribute::None);
+}
+
+MemoryBlockManager::iterator MemoryBlockManager::FindIterator(VAddr addr) {
+    auto node{memory_block_tree.begin()};
+    while (node != end()) {
+        const VAddr end_addr{node->GetNumPages() * PageSize + node->GetAddress()};
+        if (node->GetAddress() <= addr && end_addr - 1 >= addr) {
+            return node;
+        }
+        node = std::next(node);
+    }
+    return end();
+}
+
+VAddr MemoryBlockManager::FindFreeArea(VAddr region_start, std::size_t region_num_pages,
+                                       std::size_t num_pages, std::size_t align, std::size_t offset,
+                                       std::size_t guard_pages) {
+    if (num_pages == 0) {
+        return {};
+    }
+
+    const VAddr region_end{region_start + region_num_pages * PageSize};
+    const VAddr region_last{region_end - 1};
+    for (auto it{FindIterator(region_start)}; it != memory_block_tree.cend(); it++) {
+        const auto info{it->GetMemoryInfo()};
+        if (region_last < info.GetAddress()) {
+            break;
+        }
+
+        if (info.state != MemoryState::Free) {
+            continue;
+        }
+
+        VAddr area{(info.GetAddress() <= region_start) ? region_start : info.GetAddress()};
+        area += guard_pages * PageSize;
+
+        const VAddr offset_area{Common::AlignDown(area, align) + offset};
+        area = (area <= offset_area) ? offset_area : offset_area + align;
+
+        const VAddr area_end{area + num_pages * PageSize + guard_pages * PageSize};
+        const VAddr area_last{area_end - 1};
+
+        if (info.GetAddress() <= area && area < area_last && area_last <= region_last &&
+            area_last <= info.GetLastAddress()) {
+            return area;
+        }
+    }
+
+    return {};
+}
+
+void MemoryBlockManager::Update(VAddr addr, std::size_t num_pages, MemoryState prev_state,
+                                MemoryPermission prev_perm, MemoryAttribute prev_attribute,
+                                MemoryState state, MemoryPermission perm,
+                                MemoryAttribute attribute) {
+    const std::size_t prev_count{memory_block_tree.size()};
+    const VAddr end_addr{addr + num_pages * PageSize};
+    iterator node{memory_block_tree.begin()};
+
+    prev_attribute |= MemoryAttribute::IpcAndDeviceMapped;
+
+    while (node != memory_block_tree.end()) {
+        MemoryBlock* block{&(*node)};
+        iterator next_node{std::next(node)};
+        const VAddr cur_addr{block->GetAddress()};
+        const VAddr cur_end_addr{block->GetNumPages() * PageSize + cur_addr};
+
+        if (addr < cur_end_addr && cur_addr < end_addr) {
+            if (!block->HasProperties(prev_state, prev_perm, prev_attribute)) {
+                node = next_node;
+                continue;
+            }
+
+            iterator new_node{node};
+            if (addr > cur_addr) {
+                memory_block_tree.insert(node, block->Split(addr));
+            }
+
+            if (end_addr < cur_end_addr) {
+                new_node = memory_block_tree.insert(node, block->Split(end_addr));
+            }
+
+            new_node->Update(state, perm, attribute);
+
+            MergeAdjacent(new_node, next_node);
+        }
+
+        if (cur_end_addr - 1 >= end_addr - 1) {
+            break;
+        }
+
+        node = next_node;
+    }
+}
+
+void MemoryBlockManager::Update(VAddr addr, std::size_t num_pages, MemoryState state,
+                                MemoryPermission perm, MemoryAttribute attribute) {
+    const std::size_t prev_count{memory_block_tree.size()};
+    const VAddr end_addr{addr + num_pages * PageSize};
+    iterator node{memory_block_tree.begin()};
+
+    while (node != memory_block_tree.end()) {
+        MemoryBlock* block{&(*node)};
+        iterator next_node{std::next(node)};
+        const VAddr cur_addr{block->GetAddress()};
+        const VAddr cur_end_addr{block->GetNumPages() * PageSize + cur_addr};
+
+        if (addr < cur_end_addr && cur_addr < end_addr) {
+            iterator new_node{node};
+
+            if (addr > cur_addr) {
+                memory_block_tree.insert(node, block->Split(addr));
+            }
+
+            if (end_addr < cur_end_addr) {
+                new_node = memory_block_tree.insert(node, block->Split(end_addr));
+            }
+
+            new_node->Update(state, perm, attribute);
+
+            MergeAdjacent(new_node, next_node);
+        }
+
+        if (cur_end_addr - 1 >= end_addr - 1) {
+            break;
+        }
+
+        node = next_node;
+    }
+}
+
+void MemoryBlockManager::UpdateLock(VAddr addr, std::size_t num_pages, LockFunc&& lock_func,
+                                    MemoryPermission perm) {
+    const std::size_t prev_count{memory_block_tree.size()};
+    const VAddr end_addr{addr + num_pages * PageSize};
+    iterator node{memory_block_tree.begin()};
+
+    while (node != memory_block_tree.end()) {
+        MemoryBlock* block{&(*node)};
+        iterator next_node{std::next(node)};
+        const VAddr cur_addr{block->GetAddress()};
+        const VAddr cur_end_addr{block->GetNumPages() * PageSize + cur_addr};
+
+        if (addr < cur_end_addr && cur_addr < end_addr) {
+            iterator new_node{node};
+
+            if (addr > cur_addr) {
+                memory_block_tree.insert(node, block->Split(addr));
+            }
+
+            if (end_addr < cur_end_addr) {
+                new_node = memory_block_tree.insert(node, block->Split(end_addr));
+            }
+
+            lock_func(new_node, perm);
+
+            MergeAdjacent(new_node, next_node);
+        }
+
+        if (cur_end_addr - 1 >= end_addr - 1) {
+            break;
+        }
+
+        node = next_node;
+    }
+}
+
+void MemoryBlockManager::IterateForRange(VAddr start, VAddr end, IterateFunc&& func) {
+    const_iterator it{FindIterator(start)};
+    MemoryInfo info{};
+    do {
+        info = it->GetMemoryInfo();
+        func(info);
+        it = std::next(it);
+    } while (info.addr + info.size - 1 < end - 1 && it != cend());
+}
+
+void MemoryBlockManager::MergeAdjacent(iterator it, iterator& next_it) {
+    MemoryBlock* block{&(*it)};
+
+    auto EraseIt = [&](const iterator it_to_erase) {
+        if (next_it == it_to_erase) {
+            next_it = std::next(next_it);
+        }
+        memory_block_tree.erase(it_to_erase);
+    };
+
+    if (it != memory_block_tree.begin()) {
+        MemoryBlock* prev{&(*std::prev(it))};
+
+        if (block->HasSameProperties(*prev)) {
+            const iterator prev_it{std::prev(it)};
+
+            prev->Add(block->GetNumPages());
+            EraseIt(it);
+
+            it = prev_it;
+            block = prev;
+        }
+    }
+
+    if (it != cend()) {
+        const MemoryBlock* const next{&(*std::next(it))};
+
+        if (block->HasSameProperties(*next)) {
+            block->Add(next->GetNumPages());
+            EraseIt(std::next(it));
+        }
+    }
+}
+
+} // namespace Kernel::Memory

src/core/hle/kernel/memory/memory_block_manager.h

@@ -0,0 +1,67 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <functional>
+#include <list>
+#include <memory>
+
+#include "common/common_types.h"
+#include "core/hle/kernel/memory/memory_block.h"
+
+namespace Kernel::Memory {
+
+class MemoryBlockManager final {
+public:
+    using MemoryBlockTree = std::list<MemoryBlock>;
+    using iterator = MemoryBlockTree::iterator;
+    using const_iterator = MemoryBlockTree::const_iterator;
+
+public:
+    MemoryBlockManager(VAddr start_addr, VAddr end_addr);
+
+    iterator end() {
+        return memory_block_tree.end();
+    }
+    const_iterator end() const {
+        return memory_block_tree.end();
+    }
+    const_iterator cend() const {
+        return memory_block_tree.cend();
+    }
+
+    iterator FindIterator(VAddr addr);
+
+    VAddr FindFreeArea(VAddr region_start, std::size_t region_num_pages, std::size_t num_pages,
+                       std::size_t align, std::size_t offset, std::size_t guard_pages);
+
+    void Update(VAddr addr, std::size_t num_pages, MemoryState prev_state,
+                MemoryPermission prev_perm, MemoryAttribute prev_attribute, MemoryState state,
+                MemoryPermission perm, MemoryAttribute attribute);
+
+    void Update(VAddr addr, std::size_t num_pages, MemoryState state,
+                MemoryPermission perm = MemoryPermission::None,
+                MemoryAttribute attribute = MemoryAttribute::None);
+
+    using LockFunc = std::function<void(iterator, MemoryPermission)>;
+    void UpdateLock(VAddr addr, std::size_t num_pages, LockFunc&& lock_func, MemoryPermission perm);
+
+    using IterateFunc = std::function<void(const MemoryInfo&)>;
+    void IterateForRange(VAddr start, VAddr end, IterateFunc&& func);
+
+    MemoryBlock& FindBlock(VAddr addr) {
+        return *FindIterator(addr);
+    }
+
+private:
+    void MergeAdjacent(iterator it, iterator& next_it);
+
+    const VAddr start_addr;
+    const VAddr end_addr;
+
+    MemoryBlockTree memory_block_tree;
+};
+
+} // namespace Kernel::Memory

src/core/hle/kernel/memory/memory_layout.h

@@ -0,0 +1,73 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "common/common_types.h"
+
+namespace Kernel::Memory {
+
+class MemoryRegion final {
+    friend class MemoryLayout;
+
+public:
+    constexpr PAddr StartAddress() const {
+        return start_address;
+    }
+
+    constexpr PAddr EndAddress() const {
+        return end_address;
+    }
+
+private:
+    constexpr MemoryRegion() = default;
+    constexpr MemoryRegion(PAddr start_address, PAddr end_address)
+        : start_address{start_address}, end_address{end_address} {}
+
+    const PAddr start_address{};
+    const PAddr end_address{};
+};
+
+class MemoryLayout final {
+public:
+    constexpr const MemoryRegion& Application() const {
+        return application;
+    }
+
+    constexpr const MemoryRegion& Applet() const {
+        return applet;
+    }
+
+    constexpr const MemoryRegion& System() const {
+        return system;
+    }
+
+    static constexpr MemoryLayout GetDefaultLayout() {
+        constexpr std::size_t application_size{0xcd500000};
+        constexpr std::size_t applet_size{0x1fb00000};
+        constexpr PAddr application_start_address{Core::DramMemoryMap::End - application_size};
+        constexpr PAddr application_end_address{Core::DramMemoryMap::End};
+        constexpr PAddr applet_start_address{application_start_address - applet_size};
+        constexpr PAddr applet_end_address{applet_start_address + applet_size};
+        constexpr PAddr system_start_address{Core::DramMemoryMap::SlabHeapEnd};
+        constexpr PAddr system_end_address{applet_start_address};
+        return {application_start_address, application_end_address, applet_start_address,
+                applet_end_address,        system_start_address,    system_end_address};
+    }
+
+private:
+    constexpr MemoryLayout(PAddr application_start_address, std::size_t application_size,
+                           PAddr applet_start_address, std::size_t applet_size,
+                           PAddr system_start_address, std::size_t system_size)
+        : application{application_start_address, application_size},
+          applet{applet_start_address, applet_size}, system{system_start_address, system_size} {}
+
+    const MemoryRegion application;
+    const MemoryRegion applet;
+    const MemoryRegion system;
+
+    const PAddr start_address{};
+};
+
+} // namespace Kernel::Memory

src/core/hle/kernel/memory/memory_manager.cpp

@@ -0,0 +1,176 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <algorithm>
+
+#include "common/alignment.h"
+#include "common/assert.h"
+#include "common/common_types.h"
+#include "common/scope_exit.h"
+#include "core/hle/kernel/errors.h"
+#include "core/hle/kernel/memory/memory_manager.h"
+#include "core/hle/kernel/memory/page_linked_list.h"
+
+namespace Kernel::Memory {
+
+std::size_t MemoryManager::Impl::Initialize(Pool new_pool, u64 start_address, u64 end_address) {
+    const auto size{end_address - start_address};
+
+    // Calculate metadata sizes
+    const auto ref_count_size{(size / PageSize) * sizeof(u16)};
+    const auto optimize_map_size{(Common::AlignUp((size / PageSize), 64) / 64) * sizeof(u64)};
+    const auto manager_size{Common::AlignUp(optimize_map_size + ref_count_size, PageSize)};
+    const auto page_heap_size{PageHeap::CalculateMetadataOverheadSize(size)};
+    const auto total_metadata_size{manager_size + page_heap_size};
+    ASSERT(manager_size <= total_metadata_size);
+    ASSERT(Common::IsAligned(total_metadata_size, PageSize));
+
+    // Setup region
+    pool = new_pool;
+
+    // Initialize the manager's KPageHeap
+    heap.Initialize(start_address, size, page_heap_size);
+
+    // Free the memory to the heap
+    heap.Free(start_address, size / PageSize);
+
+    // Update the heap's used size
+    heap.UpdateUsedSize();
+
+    return total_metadata_size;
+}
+
+void MemoryManager::InitializeManager(Pool pool, u64 start_address, u64 end_address) {
+    ASSERT(pool < Pool::Count);
+    managers[static_cast<std::size_t>(pool)].Initialize(pool, start_address, end_address);
+}
+
+VAddr MemoryManager::AllocateContinuous(std::size_t num_pages, std::size_t align_pages, Pool pool,
+                                        Direction dir) {
+    // Early return if we're allocating no pages
+    if (num_pages == 0) {
+        return {};
+    }
+
+    // Lock the pool that we're allocating from
+    const auto pool_index{static_cast<std::size_t>(pool)};
+    std::lock_guard lock{pool_locks[pool_index]};
+
+    // Choose a heap based on our page size request
+    const s32 heap_index{PageHeap::GetAlignedBlockIndex(num_pages, align_pages)};
+
+    // Loop, trying to iterate from each block
+    // TODO (bunnei): Support multiple managers
+    Impl& chosen_manager{managers[pool_index]};
+    VAddr allocated_block{chosen_manager.AllocateBlock(heap_index)};
+
+    // If we failed to allocate, quit now
+    if (!allocated_block) {
+        return {};
+    }
+
+    // If we allocated more than we need, free some
+    const auto allocated_pages{PageHeap::GetBlockNumPages(heap_index)};
+    if (allocated_pages > num_pages) {
+        chosen_manager.Free(allocated_block + num_pages * PageSize, allocated_pages - num_pages);
+    }
+
+    return allocated_block;
+}
+
+ResultCode MemoryManager::Allocate(PageLinkedList& page_list, std::size_t num_pages, Pool pool,
+                                   Direction dir) {
+    ASSERT(page_list.GetNumPages() == 0);
+
+    // Early return if we're allocating no pages
+    if (num_pages == 0) {
+        return RESULT_SUCCESS;
+    }
+
+    // Lock the pool that we're allocating from
+    const auto pool_index{static_cast<std::size_t>(pool)};
+    std::lock_guard lock{pool_locks[pool_index]};
+
+    // Choose a heap based on our page size request
+    const s32 heap_index{PageHeap::GetBlockIndex(num_pages)};
+    if (heap_index < 0) {
+        return ERR_OUT_OF_MEMORY;
+    }
+
+    // TODO (bunnei): Support multiple managers
+    Impl& chosen_manager{managers[pool_index]};
+
+    // Ensure that we don't leave anything un-freed
+    auto group_guard = detail::ScopeExit([&] {
+        for (const auto& it : page_list.Nodes()) {
+            const auto num_pages{std::min(
+                it.GetNumPages(), (chosen_manager.GetEndAddress() - it.GetAddress()) / PageSize)};
+            chosen_manager.Free(it.GetAddress(), num_pages);
+        }
+    });
+
+    // Keep allocating until we've allocated all our pages
+    for (s32 index{heap_index}; index >= 0 && num_pages > 0; index--) {
+        const auto pages_per_alloc{PageHeap::GetBlockNumPages(index)};
+
+        while (num_pages >= pages_per_alloc) {
+            // Allocate a block
+            VAddr allocated_block{chosen_manager.AllocateBlock(index)};
+            if (!allocated_block) {
+                break;
+            }
+
+            // Safely add it to our group
+            {
+                auto block_guard = detail::ScopeExit(
+                    [&] { chosen_manager.Free(allocated_block, pages_per_alloc); });
+
+                if (const ResultCode result{page_list.AddBlock(allocated_block, pages_per_alloc)};
+                    result.IsError()) {
+                    return result;
+                }
+
+                block_guard.Cancel();
+            }
+
+            num_pages -= pages_per_alloc;
+        }
+    }
+
+    // Only succeed if we allocated as many pages as we wanted
+    ASSERT(num_pages >= 0);
+    if (num_pages) {
+        return ERR_OUT_OF_MEMORY;
+    }
+
+    // We succeeded!
+    group_guard.Cancel();
+    return RESULT_SUCCESS;
+}
+
+ResultCode MemoryManager::Free(PageLinkedList& page_list, std::size_t num_pages, Pool pool,
+                               Direction dir) {
+    // Early return if we're freeing no pages
+    if (!num_pages) {
+        return RESULT_SUCCESS;
+    }
+
+    // Lock the pool that we're freeing from
+    const auto pool_index{static_cast<std::size_t>(pool)};
+    std::lock_guard lock{pool_locks[pool_index]};
+
+    // TODO (bunnei): Support multiple managers
+    Impl& chosen_manager{managers[pool_index]};
+
+    // Free all of the pages
+    for (const auto& it : page_list.Nodes()) {
+        const auto num_pages{std::min(
+            it.GetNumPages(), (chosen_manager.GetEndAddress() - it.GetAddress()) / PageSize)};
+        chosen_manager.Free(it.GetAddress(), num_pages);
+    }
+
+    return RESULT_SUCCESS;
+}
+
+} // namespace Kernel::Memory

src/core/hle/kernel/memory/memory_manager.h

@@ -0,0 +1,97 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <array>
+#include <mutex>
+
+#include "common/common_funcs.h"
+#include "common/common_types.h"
+#include "core/hle/kernel/memory/page_heap.h"
+#include "core/hle/result.h"
+
+namespace Kernel::Memory {
+
+class PageLinkedList;
+
+class MemoryManager final : NonCopyable {
+public:
+    enum class Pool : u32 {
+        Application = 0,
+        Applet = 1,
+        System = 2,
+        SystemNonSecure = 3,
+
+        Count,
+
+        Shift = 4,
+        Mask = (0xF << Shift),
+    };
+
+    enum class Direction : u32 {
+        FromFront = 0,
+        FromBack = 1,
+
+        Shift = 0,
+        Mask = (0xF << Shift),
+    };
+
+    MemoryManager() = default;
+
+    constexpr std::size_t GetSize(Pool pool) const {
+        return managers[static_cast<std::size_t>(pool)].GetSize();
+    }
+
+    void InitializeManager(Pool pool, u64 start_address, u64 end_address);
+    VAddr AllocateContinuous(std::size_t num_pages, std::size_t align_pages, Pool pool,
+                             Direction dir = Direction::FromFront);
+    ResultCode Allocate(PageLinkedList& page_list, std::size_t num_pages, Pool pool,
+                        Direction dir = Direction::FromFront);
+    ResultCode Free(PageLinkedList& page_list, std::size_t num_pages, Pool pool,
+                    Direction dir = Direction::FromFront);
+
+    static constexpr std::size_t MaxManagerCount = 10;
+
+private:
+    class Impl final : NonCopyable {
+    private:
+        using RefCount = u16;
+
+    private:
+        PageHeap heap;
+        Pool pool{};
+
+    public:
+        Impl() = default;
+
+        std::size_t Initialize(Pool new_pool, u64 start_address, u64 end_address);
+
+        VAddr AllocateBlock(s32 index) {
+            return heap.AllocateBlock(index);
+        }
+
+        void Free(VAddr addr, std::size_t num_pages) {
+            heap.Free(addr, num_pages);
+        }
+
+        constexpr std::size_t GetSize() const {
+            return heap.GetSize();
+        }
+
+        constexpr VAddr GetAddress() const {
+            return heap.GetAddress();
+        }
+
+        constexpr VAddr GetEndAddress() const {
+            return heap.GetEndAddress();
+        }
+    };
+
+private:
+    std::array<std::mutex, static_cast<std::size_t>(Pool::Count)> pool_locks;
+    std::array<Impl, MaxManagerCount> managers;
+};
+
+} // namespace Kernel::Memory

src/core/hle/kernel/memory/memory_types.h

@@ -0,0 +1,18 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <array>
+
+#include "common/common_types.h"
+
+namespace Kernel::Memory {
+
+constexpr std::size_t PageBits{12};
+constexpr std::size_t PageSize{1 << PageBits};
+
+using Page = std::array<u8, PageSize>;
+
+} // namespace Kernel::Memory

src/core/hle/kernel/memory/page_heap.cpp

@@ -0,0 +1,119 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+// This file references various implementation details from Atmosph<70>re, an open-source firmware for
+// the Nintendo Switch. Copyright 2018-2020 Atmosph<70>re-NX.
+
+#include "core/core.h"
+#include "core/hle/kernel/memory/page_heap.h"
+#include "core/memory.h"
+
+namespace Kernel::Memory {
+
+void PageHeap::Initialize(VAddr address, std::size_t size, std::size_t metadata_size) {
+    // Check our assumptions
+    ASSERT(Common::IsAligned((address), PageSize));
+    ASSERT(Common::IsAligned(size, PageSize));
+
+    // Set our members
+    heap_address = address;
+    heap_size = size;
+
+    // Setup bitmaps
+    metadata.resize(metadata_size / sizeof(u64));
+    u64* cur_bitmap_storage{metadata.data()};
+    for (std::size_t i = 0; i < MemoryBlockPageShifts.size(); i++) {
+        const std::size_t cur_block_shift{MemoryBlockPageShifts[i]};
+        const std::size_t next_block_shift{
+            (i != MemoryBlockPageShifts.size() - 1) ? MemoryBlockPageShifts[i + 1] : 0};
+        cur_bitmap_storage = blocks[i].Initialize(heap_address, heap_size, cur_block_shift,
+                                                  next_block_shift, cur_bitmap_storage);
+    }
+}
+
+VAddr PageHeap::AllocateBlock(s32 index) {
+    const std::size_t needed_size{blocks[index].GetSize()};
+
+    for (s32 i{index}; i < static_cast<s32>(MemoryBlockPageShifts.size()); i++) {
+        if (const VAddr addr{blocks[i].PopBlock()}; addr) {
+            if (const std::size_t allocated_size{blocks[i].GetSize()};
+                allocated_size > needed_size) {
+                Free(addr + needed_size, (allocated_size - needed_size) / PageSize);
+            }
+            return addr;
+        }
+    }
+
+    return 0;
+}
+
+void PageHeap::FreeBlock(VAddr block, s32 index) {
+    do {
+        block = blocks[index++].PushBlock(block);
+    } while (block != 0);
+}
+
+void PageHeap::Free(VAddr addr, std::size_t num_pages) {
+    // Freeing no pages is a no-op
+    if (num_pages == 0) {
+        return;
+    }
+
+    // Find the largest block size that we can free, and free as many as possible
+    s32 big_index{static_cast<s32>(MemoryBlockPageShifts.size()) - 1};
+    const VAddr start{addr};
+    const VAddr end{(num_pages * PageSize) + addr};
+    VAddr before_start{start};
+    VAddr before_end{start};
+    VAddr after_start{end};
+    VAddr after_end{end};
+    while (big_index >= 0) {
+        const std::size_t block_size{blocks[big_index].GetSize()};
+        const VAddr big_start{Common::AlignUp((start), block_size)};
+        const VAddr big_end{Common::AlignDown((end), block_size)};
+        if (big_start < big_end) {
+            // Free as many big blocks as we can
+            for (auto block{big_start}; block < big_end; block += block_size) {
+                FreeBlock(block, big_index);
+            }
+            before_end = big_start;
+            after_start = big_end;
+            break;
+        }
+        big_index--;
+    }
+    ASSERT(big_index >= 0);
+
+    // Free space before the big blocks
+    for (s32 i{big_index - 1}; i >= 0; i--) {
+        const std::size_t block_size{blocks[i].GetSize()};
+        while (before_start + block_size <= before_end) {
+            before_end -= block_size;
+            FreeBlock(before_end, i);
+        }
+    }
+
+    // Free space after the big blocks
+    for (s32 i{big_index - 1}; i >= 0; i--) {
+        const std::size_t block_size{blocks[i].GetSize()};
+        while (after_start + block_size <= after_end) {
+            FreeBlock(after_start, i);
+            after_start += block_size;
+        }
+    }
+}
+
+std::size_t PageHeap::CalculateMetadataOverheadSize(std::size_t region_size) {
+    std::size_t overhead_size = 0;
+    for (std::size_t i = 0; i < MemoryBlockPageShifts.size(); i++) {
+        const std::size_t cur_block_shift{MemoryBlockPageShifts[i]};
+        const std::size_t next_block_shift{
+            (i != MemoryBlockPageShifts.size() - 1) ? MemoryBlockPageShifts[i + 1] : 0};
+        overhead_size += PageHeap::Block::CalculateMetadataOverheadSize(
+            region_size, cur_block_shift, next_block_shift);
+    }
+    return Common::AlignUp(overhead_size, PageSize);
+}
+
+} // namespace Kernel::Memory

src/core/hle/kernel/memory/page_heap.h

@@ -0,0 +1,370 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+// This file references various implementation details from Atmosph<70>re, an open-source firmware for
+// the Nintendo Switch. Copyright 2018-2020 Atmosph<70>re-NX.
+
+#pragma once
+
+#include <array>
+#include <vector>
+
+#include "common/alignment.h"
+#include "common/assert.h"
+#include "common/bit_util.h"
+#include "common/common_funcs.h"
+#include "common/common_types.h"
+#include "core/hle/kernel/memory/memory_types.h"
+
+namespace Kernel::Memory {
+
+class PageHeap final : NonCopyable {
+public:
+    static constexpr s32 GetAlignedBlockIndex(std::size_t num_pages, std::size_t align_pages) {
+        const auto target_pages{std::max(num_pages, align_pages)};
+        for (std::size_t i = 0; i < NumMemoryBlockPageShifts; i++) {
+            if (target_pages <=
+                (static_cast<std::size_t>(1) << MemoryBlockPageShifts[i]) / PageSize) {
+                return static_cast<s32>(i);
+            }
+        }
+        return -1;
+    }
+
+    static constexpr s32 GetBlockIndex(std::size_t num_pages) {
+        for (s32 i{static_cast<s32>(NumMemoryBlockPageShifts) - 1}; i >= 0; i--) {
+            if (num_pages >= (static_cast<std::size_t>(1) << MemoryBlockPageShifts[i]) / PageSize) {
+                return i;
+            }
+        }
+        return -1;
+    }
+
+    static constexpr std::size_t GetBlockSize(std::size_t index) {
+        return static_cast<std::size_t>(1) << MemoryBlockPageShifts[index];
+    }
+
+    static constexpr std::size_t GetBlockNumPages(std::size_t index) {
+        return GetBlockSize(index) / PageSize;
+    }
+
+private:
+    static constexpr std::size_t NumMemoryBlockPageShifts{7};
+    static constexpr std::array<std::size_t, NumMemoryBlockPageShifts> MemoryBlockPageShifts{
+        0xC, 0x10, 0x15, 0x16, 0x19, 0x1D, 0x1E,
+    };
+
+    class Block final : NonCopyable {
+    private:
+        class Bitmap final : NonCopyable {
+        public:
+            static constexpr std::size_t MaxDepth{4};
+
+        private:
+            std::array<u64*, MaxDepth> bit_storages{};
+            std::size_t num_bits{};
+            std::size_t used_depths{};
+
+        public:
+            constexpr Bitmap() = default;
+
+            constexpr std::size_t GetNumBits() const {
+                return num_bits;
+            }
+            constexpr s32 GetHighestDepthIndex() const {
+                return static_cast<s32>(used_depths) - 1;
+            }
+
+            constexpr u64* Initialize(u64* storage, std::size_t size) {
+                //* Initially, everything is un-set
+                num_bits = 0;
+
+                // Calculate the needed bitmap depth
+                used_depths = static_cast<std::size_t>(GetRequiredDepth(size));
+                ASSERT(used_depths <= MaxDepth);
+
+                // Set the bitmap pointers
+                for (s32 depth{GetHighestDepthIndex()}; depth >= 0; depth--) {
+                    bit_storages[depth] = storage;
+                    size = Common::AlignUp(size, 64) / 64;
+                    storage += size;
+                }
+
+                return storage;
+            }
+
+            s64 FindFreeBlock() const {
+                uintptr_t offset{};
+                s32 depth{};
+
+                do {
+                    const u64 v{bit_storages[depth][offset]};
+                    if (v == 0) {
+                        // Non-zero depth indicates that a previous level had a free block
+                        ASSERT(depth == 0);
+                        return -1;
+                    }
+                    offset = offset * 64 + Common::CountTrailingZeroes64(v);
+                    ++depth;
+                } while (depth < static_cast<s32>(used_depths));
+
+                return static_cast<s64>(offset);
+            }
+
+            constexpr void SetBit(std::size_t offset) {
+                SetBit(GetHighestDepthIndex(), offset);
+                num_bits++;
+            }
+
+            constexpr void ClearBit(std::size_t offset) {
+                ClearBit(GetHighestDepthIndex(), offset);
+                num_bits--;
+            }
+
+            constexpr bool ClearRange(std::size_t offset, std::size_t count) {
+                const s32 depth{GetHighestDepthIndex()};
+                const auto bit_ind{offset / 64};
+                u64* bits{bit_storages[depth]};
+                if (count < 64) {
+                    const auto shift{offset % 64};
+                    ASSERT(shift + count <= 64);
+                    // Check that all the bits are set
+                    const u64 mask{((1ULL << count) - 1) << shift};
+                    u64 v{bits[bit_ind]};
+                    if ((v & mask) != mask) {
+                        return false;
+                    }
+
+                    // Clear the bits
+                    v &= ~mask;
+                    bits[bit_ind] = v;
+                    if (v == 0) {
+                        ClearBit(depth - 1, bit_ind);
+                    }
+                } else {
+                    ASSERT(offset % 64 == 0);
+                    ASSERT(count % 64 == 0);
+                    // Check that all the bits are set
+                    std::size_t remaining{count};
+                    std::size_t i = 0;
+                    do {
+                        if (bits[bit_ind + i++] != ~u64(0)) {
+                            return false;
+                        }
+                        remaining -= 64;
+                    } while (remaining > 0);
+
+                    // Clear the bits
+                    remaining = count;
+                    i = 0;
+                    do {
+                        bits[bit_ind + i] = 0;
+                        ClearBit(depth - 1, bit_ind + i);
+                        i++;
+                        remaining -= 64;
+                    } while (remaining > 0);
+                }
+
+                num_bits -= count;
+                return true;
+            }
+
+        private:
+            constexpr void SetBit(s32 depth, std::size_t offset) {
+                while (depth >= 0) {
+                    const auto ind{offset / 64};
+                    const auto which{offset % 64};
+                    const u64 mask{1ULL << which};
+
+                    u64* bit{std::addressof(bit_storages[depth][ind])};
+                    const u64 v{*bit};
+                    ASSERT((v & mask) == 0);
+                    *bit = v | mask;
+                    if (v) {
+                        break;
+                    }
+                    offset = ind;
+                    depth--;
+                }
+            }
+
+            constexpr void ClearBit(s32 depth, std::size_t offset) {
+                while (depth >= 0) {
+                    const auto ind{offset / 64};
+                    const auto which{offset % 64};
+                    const u64 mask{1ULL << which};
+
+                    u64* bit{std::addressof(bit_storages[depth][ind])};
+                    u64 v{*bit};
+                    ASSERT((v & mask) != 0);
+                    v &= ~mask;
+                    *bit = v;
+                    if (v) {
+                        break;
+                    }
+                    offset = ind;
+                    depth--;
+                }
+            }
+
+        private:
+            static constexpr s32 GetRequiredDepth(std::size_t region_size) {
+                s32 depth = 0;
+                while (true) {
+                    region_size /= 64;
+                    depth++;
+                    if (region_size == 0) {
+                        return depth;
+                    }
+                }
+            }
+
+        public:
+            static constexpr std::size_t CalculateMetadataOverheadSize(std::size_t region_size) {
+                std::size_t overhead_bits = 0;
+                for (s32 depth{GetRequiredDepth(region_size) - 1}; depth >= 0; depth--) {
+                    region_size = Common::AlignUp(region_size, 64) / 64;
+                    overhead_bits += region_size;
+                }
+                return overhead_bits * sizeof(u64);
+            }
+        };
+
+    private:
+        Bitmap bitmap;
+        VAddr heap_address{};
+        uintptr_t end_offset{};
+        std::size_t block_shift{};
+        std::size_t next_block_shift{};
+
+    public:
+        constexpr Block() = default;
+
+        constexpr std::size_t GetShift() const {
+            return block_shift;
+        }
+        constexpr std::size_t GetNextShift() const {
+            return next_block_shift;
+        }
+        constexpr std::size_t GetSize() const {
+            return static_cast<std::size_t>(1) << GetShift();
+        }
+        constexpr std::size_t GetNumPages() const {
+            return GetSize() / PageSize;
+        }
+        constexpr std::size_t GetNumFreeBlocks() const {
+            return bitmap.GetNumBits();
+        }
+        constexpr std::size_t GetNumFreePages() const {
+            return GetNumFreeBlocks() * GetNumPages();
+        }
+
+        constexpr u64* Initialize(VAddr addr, std::size_t size, std::size_t bs, std::size_t nbs,
+                                  u64* bit_storage) {
+            // Set shifts
+            block_shift = bs;
+            next_block_shift = nbs;
+
+            // Align up the address
+            VAddr end{addr + size};
+            const auto align{(next_block_shift != 0) ? (1ULL << next_block_shift)
+                                                     : (1ULL << block_shift)};
+            addr = Common::AlignDown((addr), align);
+            end = Common::AlignUp((end), align);
+
+            heap_address = addr;
+            end_offset = (end - addr) / (1ULL << block_shift);
+            return bitmap.Initialize(bit_storage, end_offset);
+        }
+
+        constexpr VAddr PushBlock(VAddr address) {
+            // Set the bit for the free block
+            std::size_t offset{(address - heap_address) >> GetShift()};
+            bitmap.SetBit(offset);
+
+            // If we have a next shift, try to clear the blocks below and return the address
+            if (GetNextShift()) {
+                const auto diff{1ULL << (GetNextShift() - GetShift())};
+                offset = Common::AlignDown(offset, diff);
+                if (bitmap.ClearRange(offset, diff)) {
+                    return heap_address + (offset << GetShift());
+                }
+            }
+
+            // We couldn't coalesce, or we're already as big as possible
+            return 0;
+        }
+
+        VAddr PopBlock() {
+            // Find a free block
+            const s64 soffset{bitmap.FindFreeBlock()};
+            if (soffset < 0) {
+                return 0;
+            }
+            const auto offset{static_cast<std::size_t>(soffset)};
+
+            // Update our tracking and return it
+            bitmap.ClearBit(offset);
+            return heap_address + (offset << GetShift());
+        }
+
+    public:
+        static constexpr std::size_t CalculateMetadataOverheadSize(std::size_t region_size,
+                                                                   std::size_t cur_block_shift,
+                                                                   std::size_t next_block_shift) {
+            const auto cur_block_size{(1ULL << cur_block_shift)};
+            const auto next_block_size{(1ULL << next_block_shift)};
+            const auto align{(next_block_shift != 0) ? next_block_size : cur_block_size};
+            return Bitmap::CalculateMetadataOverheadSize(
+                (align * 2 + Common::AlignUp(region_size, align)) / cur_block_size);
+        }
+    };
+
+public:
+    PageHeap() = default;
+
+    constexpr VAddr GetAddress() const {
+        return heap_address;
+    }
+    constexpr std::size_t GetSize() const {
+        return heap_size;
+    }
+    constexpr VAddr GetEndAddress() const {
+        return GetAddress() + GetSize();
+    }
+    constexpr std::size_t GetPageOffset(VAddr block) const {
+        return (block - GetAddress()) / PageSize;
+    }
+
+    void Initialize(VAddr heap_address, std::size_t heap_size, std::size_t metadata_size);
+    VAddr AllocateBlock(s32 index);
+    void Free(VAddr addr, std::size_t num_pages);
+
+    void UpdateUsedSize() {
+        used_size = heap_size - (GetNumFreePages() * PageSize);
+    }
+
+    static std::size_t CalculateMetadataOverheadSize(std::size_t region_size);
+
+private:
+    constexpr std::size_t GetNumFreePages() const {
+        std::size_t num_free{};
+
+        for (const auto& block : blocks) {
+            num_free += block.GetNumFreePages();
+        }
+
+        return num_free;
+    }
+
+    void FreeBlock(VAddr block, s32 index);
+
+    VAddr heap_address{};
+    std::size_t heap_size{};
+    std::size_t used_size{};
+    std::array<Block, NumMemoryBlockPageShifts> blocks{};
+    std::vector<u64> metadata;
+};
+
+} // namespace Kernel::Memory

src/core/hle/kernel/memory/page_linked_list.h

@@ -0,0 +1,93 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <list>
+
+#include "common/assert.h"
+#include "common/common_funcs.h"
+#include "common/common_types.h"
+#include "core/hle/kernel/memory/memory_types.h"
+#include "core/hle/result.h"
+
+namespace Kernel::Memory {
+
+class PageLinkedList final {
+public:
+    class Node final {
+    public:
+        constexpr Node(u64 addr, std::size_t num_pages) : addr{addr}, num_pages{num_pages} {}
+
+        constexpr u64 GetAddress() const {
+            return addr;
+        }
+
+        constexpr std::size_t GetNumPages() const {
+            return num_pages;
+        }
+
+    private:
+        u64 addr{};
+        std::size_t num_pages{};
+    };
+
+public:
+    PageLinkedList() = default;
+    PageLinkedList(u64 address, u64 num_pages) {
+        ASSERT(AddBlock(address, num_pages).IsSuccess());
+    }
+
+    constexpr std::list<Node>& Nodes() {
+        return nodes;
+    }
+
+    constexpr const std::list<Node>& Nodes() const {
+        return nodes;
+    }
+
+    std::size_t GetNumPages() const {
+        std::size_t num_pages = 0;
+        for (const Node& node : nodes) {
+            num_pages += node.GetNumPages();
+        }
+        return num_pages;
+    }
+
+    bool IsEqual(PageLinkedList& other) const {
+        auto this_node = nodes.begin();
+        auto other_node = other.nodes.begin();
+        while (this_node != nodes.end() && other_node != other.nodes.end()) {
+            if (this_node->GetAddress() != other_node->GetAddress() ||
+                this_node->GetNumPages() != other_node->GetNumPages()) {
+                return false;
+            }
+            this_node = std::next(this_node);
+            other_node = std::next(other_node);
+        }
+
+        return this_node == nodes.end() && other_node == other.nodes.end();
+    }
+
+    ResultCode AddBlock(u64 address, u64 num_pages) {
+        if (!num_pages) {
+            return RESULT_SUCCESS;
+        }
+        if (!nodes.empty()) {
+            const auto node = nodes.back();
+            if (node.GetAddress() + node.GetNumPages() * PageSize == address) {
+                address = node.GetAddress();
+                num_pages += node.GetNumPages();
+                nodes.pop_back();
+            }
+        }
+        nodes.push_back({address, num_pages});
+        return RESULT_SUCCESS;
+    }
+
+private:
+    std::list<Node> nodes;
+};
+
+} // namespace Kernel::Memory