mii: Handle logging of unknown database source

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

@@ -1,15 +0,0 @@
-#!/bin/bash -ex
-
-# Copy documentation
-cp license.txt "$REV_NAME"
-cp README.md "$REV_NAME"
-
-tar $COMPRESSION_FLAGS "$ARCHIVE_NAME" "$REV_NAME"
-
-mv "$REV_NAME" $RELEASE_NAME
-
-7z a "$REV_NAME.7z" $RELEASE_NAME
-
-# move the compiled archive into the artifacts directory to be uploaded by travis releases
-mv "$ARCHIVE_NAME" artifacts/
-mv "$REV_NAME.7z" artifacts/

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

@@ -1,6 +0,0 @@
-#!/bin/bash -ex
-
-GITDATE="`git show -s --date=short --format='%ad' | sed 's/-//g'`"
-GITREV="`git show -s --format='%h'`"
-
-mkdir -p artifacts

.ci/scripts/format/docker.sh

@@ -1,6 +0,0 @@
-#!/bin/bash -ex
-
-# Run clang-format
-cd /yuzu
-chmod a+x ./.ci/scripts/format/script.sh
-./.ci/scripts/format/script.sh

.ci/scripts/format/exec.sh

@@ -1,4 +0,0 @@
-#!/bin/bash -ex
-
-chmod a+x ./.ci/scripts/format/docker.sh
-docker run -v $(pwd):/yuzu yuzuemu/build-environments:linux-clang-format /bin/bash -ex /yuzu/.ci/scripts/format/docker.sh

.ci/scripts/format/script.sh

@@ -1,37 +0,0 @@
-#!/bin/bash -ex
-
-if grep -nrI '\s$' src *.yml *.txt *.md Doxyfile .gitignore .gitmodules .ci* dist/*.desktop \
-                 dist/*.svg dist/*.xml; then
-    echo Trailing whitespace found, aborting
-    exit 1
-fi
-
-# Default clang-format points to default 3.5 version one
-CLANG_FORMAT=clang-format-6.0
-$CLANG_FORMAT --version
-
-if [ "$TRAVIS_EVENT_TYPE" = "pull_request" ]; then
-    # Get list of every file modified in this pull request
-    files_to_lint="$(git diff --name-only --diff-filter=ACMRTUXB $TRAVIS_COMMIT_RANGE | grep '^src/[^.]*[.]\(cpp\|h\)$' || true)"
-else
-    # Check everything for branch pushes
-    files_to_lint="$(find src/ -name '*.cpp' -or -name '*.h')"
-fi
-
-# Turn off tracing for this because it's too verbose
-set +x
-
-for f in $files_to_lint; do
-    d=$(diff -u "$f" <($CLANG_FORMAT "$f") || true)
-    if ! [ -z "$d" ]; then
-        echo "!!! $f not compliant to coding style, here is the fix:"
-        echo "$d"
-        fail=1
-    fi
-done
-
-set -x
-
-if [ "$fail" = 1 ]; then
-    exit 1
-fi

.ci/scripts/linux/docker.sh

@@ -1,14 +0,0 @@
-#!/bin/bash -ex
-
-cd /yuzu
-
-ccache -s
-
-mkdir build || true && cd build
-cmake .. -G Ninja -DYUZU_USE_BUNDLED_UNICORN=ON -DYUZU_USE_QT_WEB_ENGINE=ON -DCMAKE_BUILD_TYPE=Release -DCMAKE_C_COMPILER=/usr/lib/ccache/gcc -DCMAKE_CXX_COMPILER=/usr/lib/ccache/g++ -DYUZU_ENABLE_COMPATIBILITY_REPORTING=${ENABLE_COMPATIBILITY_REPORTING:-"OFF"} -DENABLE_COMPATIBILITY_LIST_DOWNLOAD=ON -DUSE_DISCORD_PRESENCE=ON
-
-ninja
-
-ccache -s
-
-ctest -VV -C Release

.ci/scripts/linux/exec.sh

@@ -1,5 +0,0 @@
-#!/bin/bash -ex
-
-mkdir -p "ccache"  || true
-chmod a+x ./.ci/scripts/linux/docker.sh
-docker run -e ENABLE_COMPATIBILITY_REPORTING -e CCACHE_DIR=/yuzu/ccache -v $(pwd):/yuzu yuzuemu/build-environments:linux-fresh /bin/bash /yuzu/.ci/scripts/linux/docker.sh

.ci/scripts/linux/upload.sh

@@ -1,14 +0,0 @@
-#!/bin/bash -ex
-
-. .ci/scripts/common/pre-upload.sh
-
-REV_NAME="yuzu-linux-${GITDATE}-${GITREV}"
-ARCHIVE_NAME="${REV_NAME}.tar.xz"
-COMPRESSION_FLAGS="-cJvf"
-
-mkdir "$REV_NAME"
-
-cp build/bin/yuzu-cmd "$REV_NAME"
-cp build/bin/yuzu "$REV_NAME"
-
-. .ci/scripts/common/post-upload.sh

.ci/scripts/merge/apply-patches-by-label.py

@@ -1,28 +0,0 @@
-# Download all pull requests as patches that match a specific label
-# Usage: python download-patches-by-label.py <Label to Match> <Root Path Folder to DL to>
-
-import requests, sys, json, urllib3.request, shutil, subprocess
-
-http = urllib3.PoolManager()
-dl_list = {}
-
-def check_individual(labels):
-    for label in labels:
-        if (label["name"] == sys.argv[1]):
-            return True
-    return False
-
-try:
-    url = 'https://api.github.com/repos/yuzu-emu/yuzu/pulls'
-    response = requests.get(url)
-    if (response.ok):
-        j = json.loads(response.content)
-        for pr in j:
-            if (check_individual(pr["labels"])):
-                pn = pr["number"]
-                print("Matched PR# %s" % pn)
-                print(subprocess.check_output(["git", "fetch", "https://github.com/yuzu-emu/yuzu.git", "pull/%s/head:pr-%s" % (pn, pn), "-f"]))
-                print(subprocess.check_output(["git", "merge", "--squash", "pr-%s" % pn]))
-                print(subprocess.check_output(["git", "commit", "-m\"Merge PR %s\"" % pn]))
-except:
-    sys.exit(-1)

.ci/scripts/merge/check-label-presence.py

@@ -1,18 +0,0 @@
-# Checks to see if the specified pull request # has the specified tag
-# Usage: python check-label-presence.py <Pull Request ID> <Name of Label>
-
-import requests, json, sys
-
-try:
-    url = 'https://api.github.com/repos/yuzu-emu/yuzu/issues/%s' % sys.argv[1]
-    response = requests.get(url)
-    if (response.ok):
-        j = json.loads(response.content)
-        for label in j["labels"]:
-            if label["name"] == sys.argv[2]:
-                print('##vso[task.setvariable variable=enabletesting;]true')
-                sys.exit()
-except:
-    sys.exit(-1)
-
-print('##vso[task.setvariable variable=enabletesting;]false')

.ci/scripts/merge/yuzubot-git-config.sh

@@ -1,2 +0,0 @@
-git config --global user.email "yuzu@yuzu-emu.org"
-git config --global user.name "yuzubot"

.ci/scripts/windows/docker.sh

@@ -1,50 +0,0 @@
-#!/bin/bash -ex
-
-cd /yuzu
-
-ccache -s
-
-# Dirty hack to trick unicorn makefile into believing we are in a MINGW system
-mv /bin/uname /bin/uname1 && echo -e '#!/bin/sh\necho MINGW64' >> /bin/uname
-chmod +x /bin/uname
-
-# Dirty hack to trick unicorn makefile into believing we have cmd
-echo '' >> /bin/cmd
-chmod +x /bin/cmd
-
-mkdir build || true && cd build
-cmake .. -G Ninja -DCMAKE_TOOLCHAIN_FILE="$(pwd)/../CMakeModules/MinGWCross.cmake" -DUSE_CCACHE=ON -DYUZU_USE_BUNDLED_UNICORN=ON -DENABLE_COMPATIBILITY_LIST_DOWNLOAD=ON -DCMAKE_BUILD_TYPE=Release
-ninja
-
-# Clean up the dirty hacks
-rm /bin/uname && mv /bin/uname1 /bin/uname
-rm /bin/cmd
-
-ccache -s
-
-echo "Tests skipped"
-#ctest -VV -C Release
-
-echo 'Prepare binaries...'
-cd ..
-mkdir package
-
-QT_PLATFORM_DLL_PATH='/usr/x86_64-w64-mingw32/lib/qt5/plugins/platforms/'
-find build/ -name "yuzu*.exe" -exec cp {} 'package' \;
-
-# copy Qt plugins
-mkdir package/platforms
-cp "${QT_PLATFORM_DLL_PATH}/qwindows.dll" package/platforms/
-cp -rv "${QT_PLATFORM_DLL_PATH}/../mediaservice/" package/
-cp -rv "${QT_PLATFORM_DLL_PATH}/../imageformats/" package/
-rm -f package/mediaservice/*d.dll
-
-for i in package/*.exe; do
-  # we need to process pdb here, however, cv2pdb
-  # does not work here, so we just simply strip all the debug symbols
-  x86_64-w64-mingw32-strip "${i}"
-done
-
-pip3 install pefile
-python3 .ci/scripts/windows/scan_dll.py package/*.exe "package/"
-python3 .ci/scripts/windows/scan_dll.py package/imageformats/*.dll "package/"

.ci/scripts/windows/exec.sh

@@ -1,5 +0,0 @@
-#!/bin/bash -ex
-
-mkdir -p "ccache" || true
-chmod a+x ./.ci/scripts/windows/docker.sh
-docker run -e CCACHE_DIR=/yuzu/ccache -v $(pwd):/yuzu yuzuemu/build-environments:linux-mingw /bin/bash -ex /yuzu/.ci/scripts/windows/docker.sh

.ci/scripts/windows/scan_dll.py

@@ -1,106 +0,0 @@
-import pefile
-import sys
-import re
-import os
-import queue
-import shutil
-
-# constant definitions
-KNOWN_SYS_DLLS = ['WINMM.DLL', 'MSVCRT.DLL', 'VERSION.DLL', 'MPR.DLL',
-                  'DWMAPI.DLL', 'UXTHEME.DLL', 'DNSAPI.DLL', 'IPHLPAPI.DLL']
-# below is for Ubuntu 18.04 with specified PPA enabled, if you are using
-# other distro or different repositories, change the following accordingly
-DLL_PATH = [
-    '/usr/x86_64-w64-mingw32/bin/',
-    '/usr/x86_64-w64-mingw32/lib/',
-    '/usr/lib/gcc/x86_64-w64-mingw32/7.3-posix/'
-]
-
-missing = []
-
-
-def parse_imports(file_name):
-    results = []
-    pe = pefile.PE(file_name, fast_load=True)
-    pe.parse_data_directories()
-
-    for entry in pe.DIRECTORY_ENTRY_IMPORT:
-        current = entry.dll.decode()
-        current_u = current.upper()  # b/c Windows is often case insensitive
-        # here we filter out system dlls
-        # dll w/ names like *32.dll are likely to be system dlls
-        if current_u.upper() not in KNOWN_SYS_DLLS and not re.match(string=current_u, pattern=r'.*32\.DLL'):
-            results.append(current)
-
-    return results
-
-
-def parse_imports_recursive(file_name, path_list=[]):
-    q = queue.Queue()  # create a FIFO queue
-    # file_name can be a string or a list for the convience
-    if isinstance(file_name, str):
-        q.put(file_name)
-    elif isinstance(file_name, list):
-        for i in file_name:
-            q.put(i)
-    full_list = []
-    while q.qsize():
-        current = q.get_nowait()
-        print('> %s' % current)
-        deps = parse_imports(current)
-        # if this dll does not have any import, ignore it
-        if not deps:
-            continue
-        for dep in deps:
-            # the dependency already included in the list, skip
-            if dep in full_list:
-                continue
-            # find the requested dll in the provided paths
-            full_path = find_dll(dep)
-            if not full_path:
-                missing.append(dep)
-                continue
-            full_list.append(dep)
-            q.put(full_path)
-            path_list.append(full_path)
-    return full_list
-
-
-def find_dll(name):
-    for path in DLL_PATH:
-        for root, _, files in os.walk(path):
-            for f in files:
-                if name.lower() == f.lower():
-                    return os.path.join(root, f)
-
-
-def deploy(name, dst, dry_run=False):
-    dlls_path = []
-    parse_imports_recursive(name, dlls_path)
-    for dll_entry in dlls_path:
-        if not dry_run:
-            shutil.copy(dll_entry, dst)
-        else:
-            print('[Dry-Run] Copy %s to %s' % (dll_entry, dst))
-    print('Deploy completed.')
-    return dlls_path
-
-
-def main():
-    if len(sys.argv) < 3:
-        print('Usage: %s [files to examine ...] [target deploy directory]')
-        return 1
-    to_deploy = sys.argv[1:-1]
-    tgt_dir = sys.argv[-1]
-    if not os.path.isdir(tgt_dir):
-        print('%s is not a directory.' % tgt_dir)
-        return 1
-    print('Scanning dependencies...')
-    deploy(to_deploy, tgt_dir)
-    if missing:
-        print('Following DLLs are not found: %s' % ('\n'.join(missing)))
-    return 0
-
-
-if __name__ == '__main__':
-    main()

.ci/scripts/windows/upload.sh

@@ -1,13 +0,0 @@
-#!/bin/bash -ex
-
-. .ci/scripts/common/pre-upload.sh
-
-REV_NAME="yuzu-windows-mingw-${GITDATE}-${GITREV}"
-ARCHIVE_NAME="${REV_NAME}.tar.gz"
-COMPRESSION_FLAGS="-czvf"
-
-mkdir "$REV_NAME"
-# get around the permission issues
-cp -r package/* "$REV_NAME"
-
-. .ci/scripts/common/post-upload.sh

.ci/templates/build-single.yml

@@ -1,21 +0,0 @@
-parameters:
-  artifactSource: 'true'
-
-steps:
-- task: DockerInstaller@0
-  displayName: 'Prepare Environment'
-  inputs:
-    dockerVersion: '17.09.0-ce'
-- task: CacheBeta@0
-  displayName: 'Cache Build System'
-  inputs:
-    key: yuzu-v1-$(BuildName)-$(BuildSuffix)-$(CacheSuffix)
-    path: $(System.DefaultWorkingDirectory)/ccache
-    cacheHitVar: CACHE_RESTORED
-- script: chmod a+x ./.ci/scripts/$(ScriptFolder)/exec.sh && ./.ci/scripts/$(ScriptFolder)/exec.sh
-  displayName: 'Build'
-- script: chmod a+x ./.ci/scripts/$(ScriptFolder)/upload.sh && ./.ci/scripts/$(ScriptFolder)/upload.sh
-  displayName: 'Package Artifacts'
-- publish: artifacts
-  artifact: 'yuzu-$(BuildName)-$(BuildSuffix)'
-  displayName: 'Upload Artifacts'

.ci/templates/build-standard.yml

@@ -1,22 +0,0 @@
-jobs:
-- job: build
-  displayName: 'standard'
-  pool:
-    vmImage: ubuntu-latest
-  strategy:
-    maxParallel: 10
-    matrix:
-      windows:
-        BuildSuffix: 'windows-mingw'
-        ScriptFolder: 'windows'
-      linux:
-        BuildSuffix: 'linux'
-        ScriptFolder: 'linux'
-  steps:
-  - template: ./sync-source.yml
-    parameters:
-      artifactSource: $(parameters.artifactSource)
-      needSubmodules: 'true'
-  - template: ./build-single.yml
-    parameters:
-      artifactSource: 'false'

.ci/templates/build-testing.yml

@@ -1,30 +0,0 @@
-jobs:
-- job: build_test
-  displayName: 'testing'
-  pool:
-    vmImage: ubuntu-latest
-  strategy:
-    maxParallel: 10
-    matrix:
-      windows:
-        BuildSuffix: 'windows-testing'
-        ScriptFolder: 'windows'
-  steps:
-  - task: PythonScript@0
-    condition: eq(variables['Build.Reason'], 'PullRequest')
-    displayName: 'Determine Testing Status'
-    inputs:
-      scriptSource: 'filePath'
-      scriptPath: '../scripts/merge/check-label-presence.py'
-      arguments: '$(System.PullRequest.PullRequestNumber) create-testing-build'
-  - ${{ if eq(variables.enabletesting, 'true') }}:
-    - template: ./sync-source.yml
-      parameters:
-        artifactSource: $(parameters.artifactSource)
-        needSubmodules: 'true'
-    - template: ./mergebot.yml
-      parameters:
-        matchLabel: 'testing-merge'
-    - template: ./build-single.yml
-      parameters:
-        artifactSource: 'false'

.ci/templates/format-check.yml

@@ -1,14 +0,0 @@
-parameters:
-  artifactSource: 'true'
-
-steps:
-- template: ./sync-source.yml
-  parameters:
-    artifactSource: $(parameters.artifactSource)
-    needSubmodules: 'false'
-- task: DockerInstaller@0
-  displayName: 'Prepare Environment'
-  inputs:
-    dockerVersion: '17.09.0-ce'
-- script: chmod a+x ./.ci/scripts/format/exec.sh && ./.ci/scripts/format/exec.sh
-  displayName: 'Verify Formatting'

.ci/templates/merge.yml

@@ -1,46 +0,0 @@
-jobs:
-- job: merge
-  displayName: 'pull requests'
-  steps:
-  - checkout: self
-    submodules: recursive
-  - template: ./mergebot.yml
-    parameters:
-      matchLabel: '$(BuildName)-merge'
-  - task: ArchiveFiles@2
-    displayName: 'Package Source'
-    inputs:
-      rootFolderOrFile: '$(System.DefaultWorkingDirectory)'
-      includeRootFolder: false
-      archiveType: '7z'
-      archiveFile: '$(Build.ArtifactStagingDirectory)/yuzu-$(BuildName)-source.7z'
-  - task: PublishPipelineArtifact@1
-    displayName: 'Upload Artifacts'
-    inputs:
-      targetPath: '$(Build.ArtifactStagingDirectory)/yuzu-$(BuildName)-source.7z'
-      artifact: 'yuzu-$(BuildName)-source'
-      replaceExistingArchive: true
-- job: upload_source
-  displayName: 'upload'
-  dependsOn: merge
-  steps:
-  - template: ./sync-source.yml
-    parameters:
-      artifactSource: 'true'
-      needSubmodules: 'true'
-  - script: chmod a+x $(System.DefaultWorkingDirectory)/.ci/scripts/merge/yuzubot-git-config.sh && $(System.DefaultWorkingDirectory)/.ci/scripts/merge/yuzubot-git-config.sh
-    displayName: 'Apply Git Configuration'
-  - script: git tag -a $(BuildName)-$(Build.BuildId) -m "yuzu $(BuildName) $(Build.BuildNumber) $(Build.DefinitionName)"
-    displayName: 'Tag Source'
-  - script: git remote add other $(GitRepoPushChangesURL)
-    displayName: 'Register Repository'
-  - script: git push --follow-tags --force other HEAD:$(GitPushBranch)
-    displayName: 'Update Code'
-  - script: git rev-list -n 1 $(BuildName)-$(Build.BuildId) > $(Build.ArtifactStagingDirectory)/tag-commit.sha
-    displayName: 'Calculate Release Point'
-  - task: PublishPipelineArtifact@1
-    displayName: 'Upload Release Point'
-    inputs:
-      targetPath: '$(Build.ArtifactStagingDirectory)/tag-commit.sha'
-      artifact: 'yuzu-$(BuildName)-release-point'
-      replaceExistingArchive: true

.ci/templates/mergebot.yml

@@ -1,15 +0,0 @@
-parameters:
-  matchLabel: 'dummy-merge'
-
-steps:
-  - script: mkdir $(System.DefaultWorkingDirectory)/patches && pip install requests urllib3
-    displayName: 'Prepare Environment'
-  - script: chmod a+x $(System.DefaultWorkingDirectory)/.ci/scripts/merge/yuzubot-git-config.sh && $(System.DefaultWorkingDirectory)/.ci/scripts/merge/yuzubot-git-config.sh
-    displayName: 'Apply Git Configuration'
-  - task: PythonScript@0
-    displayName: 'Discover, Download, and Apply Patches'
-    inputs:
-      scriptSource: 'filePath'
-      scriptPath: '.ci/scripts/merge/apply-patches-by-label.py'
-      arguments: '${{ parameters.matchLabel }} patches'
-      workingDirectory: '$(System.DefaultWorkingDirectory)'

.ci/templates/retrieve-artifact-source.yml

@@ -1,16 +0,0 @@
-steps:
-- checkout: none
-- task: DownloadPipelineArtifact@2
-  displayName: 'Download Source'
-  inputs:
-    artifactName: 'yuzu-$(BuildName)-source'
-    buildType: 'current'
-    targetPath: '$(Build.ArtifactStagingDirectory)'
-- script: rm -rf $(System.DefaultWorkingDirectory) && mkdir $(System.DefaultWorkingDirectory)
-  displayName: 'Clean Working Directory'
-- task: ExtractFiles@1
-  displayName: 'Prepare Source'
-  inputs:
-    archiveFilePatterns: '$(Build.ArtifactStagingDirectory)/*.7z'
-    destinationFolder: '$(System.DefaultWorkingDirectory)'
-    cleanDestinationFolder: false

.ci/templates/retrieve-master-source.yml

@@ -1,11 +0,0 @@
-parameters:
-  needSubmodules: 'true'
-
-steps:
-- checkout: self
-  displayName: 'Checkout Recursive'
-  submodules: recursive
-#  condition: eq(parameters.needSubmodules, 'true')
-#- checkout: self
-#  displayName: 'Checkout Fast'
-#  condition: ne(parameters.needSubmodules, 'true')

.ci/templates/sync-source.yml

@@ -1,7 +0,0 @@
-steps:
-- ${{ if eq(parameters.artifactSource, 'true') }}:
-  - template: ./retrieve-artifact-source.yml
-- ${{ if ne(parameters.artifactSource, 'true') }}:
-  - template: ./retrieve-master-source.yml
-    parameters:
-      needSubmodules: $(parameters.needSubmodules)

.ci/yuzu-mainline.yml

@@ -1,23 +0,0 @@
-trigger:
-- master
-
-stages:
-- stage: merge
-  displayName: 'merge'
-  jobs:
-  - template: ./templates/merge.yml
-- stage: format
-  dependsOn: merge
-  displayName: 'format'
-  jobs:
-  - job: format
-    displayName: 'clang'
-    pool:
-      vmImage: ubuntu-latest
-    steps:
-    - template: ./templates/format-check.yml
-- stage: build
-  displayName: 'build'
-  dependsOn: format
-  jobs:
-  - template: ./templates/build-standard.yml

.ci/yuzu-repo-sync.yml

@@ -1,19 +0,0 @@
-trigger:
-- master
-
-jobs:
-- job: copy
-  displayName: 'Sync Repository'
-  pool:
-    vmImage: 'ubuntu-latest'
-  steps:
-  - script: echo 'https://$(GitUsername):$(GitAccessToken)@dev.azure.com' > $HOME/.git-credentials
-    displayName: 'Load Credentials'
-  - script: git config --global credential.helper store
-    displayName: 'Register Credential Helper'
-  - script: git remote add other $(GitRepoPushChangesURL)
-    displayName: 'Register Repository'
-  - script: git push --force other HEAD:$(GitPushBranch)
-    displayName: 'Update Code'
-  - script: rm -rf $HOME/.git-credentials
-    displayName: 'Clear Cached Credentials'

.ci/yuzu-verify.yml

@@ -1,18 +0,0 @@
-stages:
-- stage: format
-  displayName: 'format'
-  jobs:
-  - job: format
-    displayName: 'clang'
-    pool:
-      vmImage: ubuntu-latest
-    steps:
-    - template: ./templates/format-check.yml
-      parameters:
-        artifactSource: 'false'
-- stage: build
-  displayName: 'build'
-  dependsOn: format
-  jobs:
-  - template: ./templates/build-standard.yml
-  - template: ./templates/build-testing.yml

CMakeModules/GenerateSCMRev.cmake

@@ -82,8 +82,6 @@ set(HASH_FILES
     "${VIDEO_CORE}/shader/decode/shift.cpp"
     "${VIDEO_CORE}/shader/decode/video.cpp"
     "${VIDEO_CORE}/shader/decode/xmad.cpp"
-    "${VIDEO_CORE}/shader/control_flow.cpp"
-    "${VIDEO_CORE}/shader/control_flow.h"
     "${VIDEO_CORE}/shader/decode.cpp"
     "${VIDEO_CORE}/shader/node.h"
     "${VIDEO_CORE}/shader/node_helper.cpp"

README.md

@@ -2,7 +2,6 @@ yuzu emulator
 =============
 [![Travis CI Build Status](https://travis-ci.org/yuzu-emu/yuzu.svg?branch=master)](https://travis-ci.org/yuzu-emu/yuzu)
 [![AppVeyor CI Build Status](https://ci.appveyor.com/api/projects/status/77k97svb2usreu68?svg=true)](https://ci.appveyor.com/project/bunnei/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/)
 
 yuzu is an experimental open-source emulator for the Nintendo Switch from the creators of [Citra](https://citra-emu.org/).
 

src/audio_core/audio_renderer.cpp

@@ -73,15 +73,13 @@ private:
     EffectInStatus info{};
 };
 AudioRenderer::AudioRenderer(Core::Timing::CoreTiming& core_timing, AudioRendererParameter params,
-                             Kernel::SharedPtr<Kernel::WritableEvent> buffer_event,
-                             std::size_t instance_number)
+                             Kernel::SharedPtr<Kernel::WritableEvent> buffer_event)
     : worker_params{params}, buffer_event{buffer_event}, voices(params.voice_count),
       effects(params.effect_count) {
 
     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),
-                                   [=]() { buffer_event->Signal(); });
+                                   "AudioRenderer", [=]() { buffer_event->Signal(); });
     audio_out->StartStream(stream);
 
     QueueMixedBuffer(0);

src/audio_core/audio_renderer.h

@@ -215,8 +215,7 @@ static_assert(sizeof(UpdateDataHeader) == 0x40, "UpdateDataHeader has wrong size
 class AudioRenderer {
 public:
     AudioRenderer(Core::Timing::CoreTiming& core_timing, AudioRendererParameter params,
-                  Kernel::SharedPtr<Kernel::WritableEvent> buffer_event,
-                  std::size_t instance_number);
+                  Kernel::SharedPtr<Kernel::WritableEvent> buffer_event);
     ~AudioRenderer();
 
     std::vector<u8> UpdateAudioRenderer(const std::vector<u8>& input_params);

src/common/CMakeLists.txt

@@ -56,8 +56,6 @@ add_custom_command(OUTPUT scm_rev.cpp
       "${VIDEO_CORE}/shader/decode/shift.cpp"
       "${VIDEO_CORE}/shader/decode/video.cpp"
       "${VIDEO_CORE}/shader/decode/xmad.cpp"
-      "${VIDEO_CORE}/shader/control_flow.cpp"
-      "${VIDEO_CORE}/shader/control_flow.h"
       "${VIDEO_CORE}/shader/decode.cpp"
       "${VIDEO_CORE}/shader/node.h"
       "${VIDEO_CORE}/shader/node_helper.cpp"

src/core/arm/arm_interface.h

@@ -44,6 +44,13 @@ public:
     /// Step CPU by one instruction
     virtual void Step() = 0;
 
+    /// Maps a backing memory region for the CPU
+    virtual void MapBackingMemory(VAddr address, std::size_t size, u8* memory,
+                                  Kernel::VMAPermission perms) = 0;
+
+    /// Unmaps a region of memory that was previously mapped using MapBackingMemory
+    virtual void UnmapMemory(VAddr address, std::size_t size) = 0;
+
     /// Clear all instruction cache
     virtual void ClearInstructionCache() = 0;
 

src/core/arm/dynarmic/arm_dynarmic.cpp

@@ -177,6 +177,15 @@ ARM_Dynarmic::ARM_Dynarmic(System& system, ExclusiveMonitor& exclusive_monitor,
 
 ARM_Dynarmic::~ARM_Dynarmic() = default;
 
+void ARM_Dynarmic::MapBackingMemory(u64 address, std::size_t size, u8* memory,
+                                    Kernel::VMAPermission perms) {
+    inner_unicorn.MapBackingMemory(address, size, memory, perms);
+}
+
+void ARM_Dynarmic::UnmapMemory(u64 address, std::size_t size) {
+    inner_unicorn.UnmapMemory(address, size);
+}
+
 void ARM_Dynarmic::SetPC(u64 pc) {
     jit->SetPC(pc);
 }

src/core/arm/dynarmic/arm_dynarmic.h

@@ -23,6 +23,9 @@ public:
     ARM_Dynarmic(System& system, ExclusiveMonitor& exclusive_monitor, std::size_t core_index);
     ~ARM_Dynarmic() override;
 
+    void MapBackingMemory(VAddr address, std::size_t size, u8* memory,
+                          Kernel::VMAPermission perms) override;
+    void UnmapMemory(u64 address, std::size_t size) override;
     void SetPC(u64 pc) override;
     u64 GetPC() const override;
     u64 GetReg(int index) const override;

src/core/arm/unicorn/arm_unicorn.cpp

@@ -50,14 +50,11 @@ static void CodeHook(uc_engine* uc, uint64_t address, uint32_t size, void* user_
 
 static bool UnmappedMemoryHook(uc_engine* uc, uc_mem_type type, u64 addr, int size, u64 value,
                                void* user_data) {
-    auto* const system = static_cast<System*>(user_data);
-
     ARM_Interface::ThreadContext ctx{};
-    system->CurrentArmInterface().SaveContext(ctx);
+    Core::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]);
-
-    return false;
+    return {};
 }
 
 ARM_Unicorn::ARM_Unicorn(System& system) : system{system} {
@@ -68,7 +65,7 @@ ARM_Unicorn::ARM_Unicorn(System& system) : system{system} {
 
     uc_hook hook{};
     CHECKED(uc_hook_add(uc, &hook, UC_HOOK_INTR, (void*)InterruptHook, this, 0, -1));
-    CHECKED(uc_hook_add(uc, &hook, UC_HOOK_MEM_INVALID, (void*)UnmappedMemoryHook, &system, 0, -1));
+    CHECKED(uc_hook_add(uc, &hook, UC_HOOK_MEM_INVALID, (void*)UnmappedMemoryHook, this, 0, -1));
     if (GDBStub::IsServerEnabled()) {
         CHECKED(uc_hook_add(uc, &hook, UC_HOOK_CODE, (void*)CodeHook, this, 0, -1));
         last_bkpt_hit = false;
@@ -79,6 +76,15 @@ ARM_Unicorn::~ARM_Unicorn() {
     CHECKED(uc_close(uc));
 }
 
+void ARM_Unicorn::MapBackingMemory(VAddr address, std::size_t size, u8* memory,
+                                   Kernel::VMAPermission perms) {
+    CHECKED(uc_mem_map_ptr(uc, address, size, static_cast<u32>(perms), memory));
+}
+
+void ARM_Unicorn::UnmapMemory(VAddr address, std::size_t size) {
+    CHECKED(uc_mem_unmap(uc, address, size));
+}
+
 void ARM_Unicorn::SetPC(u64 pc) {
     CHECKED(uc_reg_write(uc, UC_ARM64_REG_PC, &pc));
 }

src/core/arm/unicorn/arm_unicorn.h

@@ -18,6 +18,9 @@ public:
     explicit ARM_Unicorn(System& system);
     ~ARM_Unicorn() override;
 
+    void MapBackingMemory(VAddr address, std::size_t size, u8* memory,
+                          Kernel::VMAPermission perms) override;
+    void UnmapMemory(VAddr address, std::size_t size) override;
     void SetPC(u64 pc) override;
     u64 GetPC() const override;
     u64 GetReg(int index) const override;

src/core/core.h

@@ -327,6 +327,10 @@ private:
     static System s_instance;
 };
 
+inline ARM_Interface& CurrentArmInterface() {
+    return System::GetInstance().CurrentArmInterface();
+}
+
 inline Kernel::Process* CurrentProcess() {
     return System::GetInstance().CurrentProcess();
 }

src/core/core_cpu.cpp

@@ -53,12 +53,16 @@ bool CpuBarrier::Rendezvous() {
 Cpu::Cpu(System& system, ExclusiveMonitor& exclusive_monitor, CpuBarrier& cpu_barrier,
          std::size_t core_index)
     : cpu_barrier{cpu_barrier}, core_timing{system.CoreTiming()}, core_index{core_index} {
+    if (Settings::values.cpu_jit_enabled) {
 #ifdef ARCHITECTURE_x86_64
-    arm_interface = std::make_unique<ARM_Dynarmic>(system, exclusive_monitor, core_index);
+        arm_interface = std::make_unique<ARM_Dynarmic>(system, exclusive_monitor, core_index);
 #else
-    arm_interface = std::make_unique<ARM_Unicorn>(system);
-    LOG_WARNING(Core, "CPU JIT requested, but Dynarmic not available");
+        arm_interface = std::make_unique<ARM_Unicorn>(system);
+        LOG_WARNING(Core, "CPU JIT requested, but Dynarmic not available");
 #endif
+    } else {
+        arm_interface = std::make_unique<ARM_Unicorn>(system);
+    }
 
     scheduler = std::make_unique<Kernel::Scheduler>(system, *arm_interface);
 }
@@ -66,12 +70,15 @@ Cpu::Cpu(System& system, ExclusiveMonitor& exclusive_monitor, CpuBarrier& cpu_ba
 Cpu::~Cpu() = default;
 
 std::unique_ptr<ExclusiveMonitor> Cpu::MakeExclusiveMonitor(std::size_t num_cores) {
+    if (Settings::values.cpu_jit_enabled) {
 #ifdef ARCHITECTURE_x86_64
-    return std::make_unique<DynarmicExclusiveMonitor>(num_cores);
+        return std::make_unique<DynarmicExclusiveMonitor>(num_cores);
 #else
-    // TODO(merry): Passthrough exclusive monitor
-    return nullptr;
+        return nullptr; // TODO(merry): Passthrough exclusive monitor
 #endif
+    } else {
+        return nullptr; // TODO(merry): Passthrough exclusive monitor
+    }
 }
 
 void Cpu::RunLoop(bool tight_loop) {

src/core/file_sys/program_metadata.cpp

@@ -94,10 +94,6 @@ u64 ProgramMetadata::GetFilesystemPermissions() const {
     return aci_file_access.permissions;
 }
 
-u32 ProgramMetadata::GetSystemResourceSize() const {
-    return npdm_header.system_resource_size;
-}
-
 const ProgramMetadata::KernelCapabilityDescriptors& ProgramMetadata::GetKernelCapabilities() const {
     return aci_kernel_capabilities;
 }

src/core/file_sys/program_metadata.h

@@ -58,7 +58,6 @@ public:
     u32 GetMainThreadStackSize() const;
     u64 GetTitleID() const;
     u64 GetFilesystemPermissions() const;
-    u32 GetSystemResourceSize() const;
     const KernelCapabilityDescriptors& GetKernelCapabilities() const;
 
     void Print() const;
@@ -77,8 +76,7 @@ private:
         u8 reserved_3;
         u8 main_thread_priority;
         u8 main_thread_cpu;
-        std::array<u8, 4> reserved_4;
-        u32_le system_resource_size;
+        std::array<u8, 8> reserved_4;
         u32_le process_category;
         u32_le main_stack_size;
         std::array<u8, 0x10> application_name;

src/core/hle/kernel/process.cpp

@@ -129,17 +129,20 @@ u64 Process::GetTotalPhysicalMemoryAvailable() const {
     return vm_manager.GetTotalPhysicalMemoryAvailable();
 }
 
-u64 Process::GetTotalPhysicalMemoryAvailableWithoutSystemResource() const {
-    return GetTotalPhysicalMemoryAvailable() - GetSystemResourceSize();
+u64 Process::GetTotalPhysicalMemoryAvailableWithoutMmHeap() const {
+    // TODO: Subtract the personal heap size from this when the
+    //       personal heap is implemented.
+    return GetTotalPhysicalMemoryAvailable();
 }
 
 u64 Process::GetTotalPhysicalMemoryUsed() const {
-    return vm_manager.GetCurrentHeapSize() + main_thread_stack_size + code_memory_size +
-           GetSystemResourceUsage();
+    return vm_manager.GetCurrentHeapSize() + main_thread_stack_size + code_memory_size;
 }
 
-u64 Process::GetTotalPhysicalMemoryUsedWithoutSystemResource() const {
-    return GetTotalPhysicalMemoryUsed() - GetSystemResourceUsage();
+u64 Process::GetTotalPhysicalMemoryUsedWithoutMmHeap() const {
+    // TODO: Subtract the personal heap size from this when the
+    //       personal heap is implemented.
+    return GetTotalPhysicalMemoryUsed();
 }
 
 void Process::RegisterThread(const Thread* thread) {
@@ -169,7 +172,6 @@ ResultCode Process::LoadFromMetadata(const FileSys::ProgramMetadata& metadata) {
     program_id = metadata.GetTitleID();
     ideal_core = metadata.GetMainThreadCore();
     is_64bit_process = metadata.Is64BitProgram();
-    system_resource_size = metadata.GetSystemResourceSize();
 
     vm_manager.Reset(metadata.GetAddressSpaceType());
 
@@ -184,11 +186,19 @@ ResultCode Process::LoadFromMetadata(const FileSys::ProgramMetadata& metadata) {
 }
 
 void Process::Run(s32 main_thread_priority, u64 stack_size) {
-    AllocateMainThreadStack(stack_size);
-    tls_region_address = CreateTLSRegion();
+    // The kernel always ensures that the given stack size is page aligned.
+    main_thread_stack_size = Common::AlignUp(stack_size, Memory::PAGE_SIZE);
+
+    // Allocate and map the main thread stack
+    // TODO(bunnei): This is heap area that should be allocated by the kernel and not mapped as part
+    // of the user address space.
+    const VAddr mapping_address = vm_manager.GetTLSIORegionEndAddress() - main_thread_stack_size;
+    vm_manager
+        .MapMemoryBlock(mapping_address, std::make_shared<std::vector<u8>>(main_thread_stack_size),
+                        0, main_thread_stack_size, MemoryState::Stack)
+        .Unwrap();
 
     vm_manager.LogLayout();
-
     ChangeStatus(ProcessStatus::Running);
 
     SetupMainThread(*this, kernel, main_thread_priority);
@@ -218,9 +228,6 @@ void Process::PrepareForTermination() {
     stop_threads(system.Scheduler(2).GetThreadList());
     stop_threads(system.Scheduler(3).GetThreadList());
 
-    FreeTLSRegion(tls_region_address);
-    tls_region_address = 0;
-
     ChangeStatus(ProcessStatus::Exited);
 }
 
@@ -320,16 +327,4 @@ void Process::ChangeStatus(ProcessStatus new_status) {
     WakeupAllWaitingThreads();
 }
 
-void Process::AllocateMainThreadStack(u64 stack_size) {
-    // The kernel always ensures that the given stack size is page aligned.
-    main_thread_stack_size = Common::AlignUp(stack_size, Memory::PAGE_SIZE);
-
-    // Allocate and map the main thread stack
-    const VAddr mapping_address = vm_manager.GetTLSIORegionEndAddress() - main_thread_stack_size;
-    vm_manager
-        .MapMemoryBlock(mapping_address, std::make_shared<std::vector<u8>>(main_thread_stack_size),
-                        0, main_thread_stack_size, MemoryState::Stack)
-        .Unwrap();
-}
-
 } // namespace Kernel

src/core/hle/kernel/process.h

@@ -135,11 +135,6 @@ public:
         return mutex;
     }
 
-    /// Gets the address to the process' dedicated TLS region.
-    VAddr GetTLSRegionAddress() const {
-        return tls_region_address;
-    }
-
     /// Gets the current status of the process
     ProcessStatus GetStatus() const {
         return status;
@@ -173,24 +168,8 @@ public:
         return capabilities.GetPriorityMask();
     }
 
-    /// Gets the amount of secure memory to allocate for memory management.
-    u32 GetSystemResourceSize() const {
-        return system_resource_size;
-    }
-
-    /// Gets the amount of secure memory currently in use for memory management.
-    u32 GetSystemResourceUsage() const {
-        // On hardware, this returns the amount of system resource memory that has
-        // been used by the kernel. This is problematic for Yuzu to emulate, because
-        // system resource memory is used for page tables -- and yuzu doesn't really
-        // have a way to calculate how much memory is required for page tables for
-        // the current process at any given time.
-        // TODO: Is this even worth implementing? Games may retrieve this value via
-        // an SDK function that gets used + available system resource size for debug
-        // or diagnostic purposes. However, it seems unlikely that a game would make
-        // decisions based on how much system memory is dedicated to its page tables.
-        // Is returning a value other than zero wise?
-        return 0;
+    u32 IsVirtualMemoryEnabled() const {
+        return is_virtual_address_memory_enabled;
     }
 
     /// Whether this process is an AArch64 or AArch32 process.
@@ -217,15 +196,15 @@ public:
     u64 GetTotalPhysicalMemoryAvailable() const;
 
     /// Retrieves the total physical memory available to this process in bytes,
-    /// without the size of the personal system resource heap added to it.
-    u64 GetTotalPhysicalMemoryAvailableWithoutSystemResource() const;
+    /// without the size of the personal heap added to it.
+    u64 GetTotalPhysicalMemoryAvailableWithoutMmHeap() const;
 
     /// Retrieves the total physical memory used by this process in bytes.
     u64 GetTotalPhysicalMemoryUsed() const;
 
     /// Retrieves the total physical memory used by this process in bytes,
-    /// without the size of the personal system resource heap added to it.
-    u64 GetTotalPhysicalMemoryUsedWithoutSystemResource() const;
+    /// without the size of the personal heap added to it.
+    u64 GetTotalPhysicalMemoryUsedWithoutMmHeap() const;
 
     /// Gets the list of all threads created with this process as their owner.
     const std::list<const Thread*>& GetThreadList() const {
@@ -301,9 +280,6 @@ private:
     /// a process signal.
     void ChangeStatus(ProcessStatus new_status);
 
-    /// Allocates the main thread stack for the process, given the stack size in bytes.
-    void AllocateMainThreadStack(u64 stack_size);
-
     /// Memory manager for this process.
     Kernel::VMManager vm_manager;
 
@@ -322,16 +298,12 @@ private:
     /// Title ID corresponding to the process
     u64 program_id = 0;
 
-    /// Specifies additional memory to be reserved for the process's memory management by the
-    /// system. When this is non-zero, secure memory is allocated and used for page table allocation
-    /// instead of using the normal global page tables/memory block management.
-    u32 system_resource_size = 0;
-
     /// Resource limit descriptor for this process
     SharedPtr<ResourceLimit> resource_limit;
 
     /// The ideal CPU core for this process, threads are scheduled on this core by default.
     u8 ideal_core = 0;
+    u32 is_virtual_address_memory_enabled = 0;
 
     /// The Thread Local Storage area is allocated as processes create threads,
     /// each TLS area is 0x200 bytes, so one page (0x1000) is split up in 8 parts, and each part
@@ -366,9 +338,6 @@ private:
     /// variable related facilities.
     Mutex mutex;
 
-    /// Address indicating the location of the process' dedicated TLS region.
-    VAddr tls_region_address = 0;
-
     /// Random values for svcGetInfo RandomEntropy
     std::array<u64, RANDOM_ENTROPY_SIZE> random_entropy{};
 

src/core/hle/kernel/svc.cpp

@@ -318,14 +318,7 @@ static ResultCode UnmapMemory(Core::System& system, VAddr dst_addr, VAddr src_ad
         return result;
     }
 
-    const auto unmap_res = vm_manager.UnmapRange(dst_addr, size);
-
-    // Reprotect the source mapping on success
-    if (unmap_res.IsSuccess()) {
-        ASSERT(vm_manager.ReprotectRange(src_addr, size, VMAPermission::ReadWrite).IsSuccess());
-    }
-
-    return unmap_res;
+    return vm_manager.UnmapRange(dst_addr, size);
 }
 
 /// Connect to an OS service given the port name, returns the handle to the port to out
@@ -736,16 +729,16 @@ static ResultCode GetInfo(Core::System& system, u64* result, u64 info_id, u64 ha
         StackRegionBaseAddr = 14,
         StackRegionSize = 15,
         // 3.0.0+
-        SystemResourceSize = 16,
-        SystemResourceUsage = 17,
+        IsVirtualAddressMemoryEnabled = 16,
+        PersonalMmHeapUsage = 17,
         TitleId = 18,
         // 4.0.0+
         PrivilegedProcessId = 19,
         // 5.0.0+
         UserExceptionContextAddr = 20,
         // 6.0.0+
-        TotalPhysicalMemoryAvailableWithoutSystemResource = 21,
-        TotalPhysicalMemoryUsedWithoutSystemResource = 22,
+        TotalPhysicalMemoryAvailableWithoutMmHeap = 21,
+        TotalPhysicalMemoryUsedWithoutMmHeap = 22,
     };
 
     const auto info_id_type = static_cast<GetInfoType>(info_id);
@@ -763,12 +756,12 @@ static ResultCode GetInfo(Core::System& system, u64* result, u64 info_id, u64 ha
     case GetInfoType::StackRegionSize:
     case GetInfoType::TotalPhysicalMemoryAvailable:
     case GetInfoType::TotalPhysicalMemoryUsed:
-    case GetInfoType::SystemResourceSize:
-    case GetInfoType::SystemResourceUsage:
+    case GetInfoType::IsVirtualAddressMemoryEnabled:
+    case GetInfoType::PersonalMmHeapUsage:
     case GetInfoType::TitleId:
     case GetInfoType::UserExceptionContextAddr:
-    case GetInfoType::TotalPhysicalMemoryAvailableWithoutSystemResource:
-    case GetInfoType::TotalPhysicalMemoryUsedWithoutSystemResource: {
+    case GetInfoType::TotalPhysicalMemoryAvailableWithoutMmHeap:
+    case GetInfoType::TotalPhysicalMemoryUsedWithoutMmHeap: {
         if (info_sub_id != 0) {
             return ERR_INVALID_ENUM_VALUE;
         }
@@ -829,13 +822,8 @@ static ResultCode GetInfo(Core::System& system, u64* result, u64 info_id, u64 ha
             *result = process->GetTotalPhysicalMemoryUsed();
             return RESULT_SUCCESS;
 
-        case GetInfoType::SystemResourceSize:
-            *result = process->GetSystemResourceSize();
-            return RESULT_SUCCESS;
-
-        case GetInfoType::SystemResourceUsage:
-            LOG_WARNING(Kernel_SVC, "(STUBBED) Attempted to query system resource usage");
-            *result = process->GetSystemResourceUsage();
+        case GetInfoType::IsVirtualAddressMemoryEnabled:
+            *result = process->IsVirtualMemoryEnabled();
             return RESULT_SUCCESS;
 
         case GetInfoType::TitleId:
@@ -843,15 +831,17 @@ static ResultCode GetInfo(Core::System& system, u64* result, u64 info_id, u64 ha
             return RESULT_SUCCESS;
 
         case GetInfoType::UserExceptionContextAddr:
-            *result = process->GetTLSRegionAddress();
+            LOG_WARNING(Kernel_SVC,
+                        "(STUBBED) Attempted to query user exception context address, returned 0");
+            *result = 0;
             return RESULT_SUCCESS;
 
-        case GetInfoType::TotalPhysicalMemoryAvailableWithoutSystemResource:
-            *result = process->GetTotalPhysicalMemoryAvailableWithoutSystemResource();
+        case GetInfoType::TotalPhysicalMemoryAvailableWithoutMmHeap:
+            *result = process->GetTotalPhysicalMemoryAvailable();
             return RESULT_SUCCESS;
 
-        case GetInfoType::TotalPhysicalMemoryUsedWithoutSystemResource:
-            *result = process->GetTotalPhysicalMemoryUsedWithoutSystemResource();
+        case GetInfoType::TotalPhysicalMemoryUsedWithoutMmHeap:
+            *result = process->GetTotalPhysicalMemoryUsedWithoutMmHeap();
             return RESULT_SUCCESS;
 
         default:
@@ -956,86 +946,6 @@ static ResultCode GetInfo(Core::System& system, u64* result, u64 info_id, u64 ha
     }
 }
 
-/// Maps memory at a desired address
-static ResultCode MapPhysicalMemory(Core::System& system, VAddr addr, u64 size) {
-    LOG_DEBUG(Kernel_SVC, "called, addr=0x{:016X}, size=0x{:X}", addr, size);
-
-    if (!Common::Is4KBAligned(addr)) {
-        LOG_ERROR(Kernel_SVC, "Address is not aligned to 4KB, 0x{:016X}", addr);
-        return ERR_INVALID_ADDRESS;
-    }
-
-    if (!Common::Is4KBAligned(size)) {
-        LOG_ERROR(Kernel_SVC, "Size is not aligned to 4KB, 0x{:X}", size);
-        return ERR_INVALID_SIZE;
-    }
-
-    if (size == 0) {
-        LOG_ERROR(Kernel_SVC, "Size is zero");
-        return ERR_INVALID_SIZE;
-    }
-
-    if (!(addr < addr + size)) {
-        LOG_ERROR(Kernel_SVC, "Size causes 64-bit overflow of address");
-        return ERR_INVALID_MEMORY_RANGE;
-    }
-
-    Process* const current_process = system.Kernel().CurrentProcess();
-    auto& vm_manager = current_process->VMManager();
-
-    if (current_process->GetSystemResourceSize() == 0) {
-        LOG_ERROR(Kernel_SVC, "System Resource Size is zero");
-        return ERR_INVALID_STATE;
-    }
-
-    if (!vm_manager.IsWithinMapRegion(addr, size)) {
-        LOG_ERROR(Kernel_SVC, "Range not within map region");
-        return ERR_INVALID_MEMORY_RANGE;
-    }
-
-    return vm_manager.MapPhysicalMemory(addr, size);
-}
-
-/// Unmaps memory previously mapped via MapPhysicalMemory
-static ResultCode UnmapPhysicalMemory(Core::System& system, VAddr addr, u64 size) {
-    LOG_DEBUG(Kernel_SVC, "called, addr=0x{:016X}, size=0x{:X}", addr, size);
-
-    if (!Common::Is4KBAligned(addr)) {
-        LOG_ERROR(Kernel_SVC, "Address is not aligned to 4KB, 0x{:016X}", addr);
-        return ERR_INVALID_ADDRESS;
-    }
-
-    if (!Common::Is4KBAligned(size)) {
-        LOG_ERROR(Kernel_SVC, "Size is not aligned to 4KB, 0x{:X}", size);
-        return ERR_INVALID_SIZE;
-    }
-
-    if (size == 0) {
-        LOG_ERROR(Kernel_SVC, "Size is zero");
-        return ERR_INVALID_SIZE;
-    }
-
-    if (!(addr < addr + size)) {
-        LOG_ERROR(Kernel_SVC, "Size causes 64-bit overflow of address");
-        return ERR_INVALID_MEMORY_RANGE;
-    }
-
-    Process* const current_process = system.Kernel().CurrentProcess();
-    auto& vm_manager = current_process->VMManager();
-
-    if (current_process->GetSystemResourceSize() == 0) {
-        LOG_ERROR(Kernel_SVC, "System Resource Size is zero");
-        return ERR_INVALID_STATE;
-    }
-
-    if (!vm_manager.IsWithinMapRegion(addr, size)) {
-        LOG_ERROR(Kernel_SVC, "Range not within map region");
-        return ERR_INVALID_MEMORY_RANGE;
-    }
-
-    return vm_manager.UnmapPhysicalMemory(addr, size);
-}
-
 /// Sets the thread activity
 static ResultCode SetThreadActivity(Core::System& system, Handle handle, u32 activity) {
     LOG_DEBUG(Kernel_SVC, "called, handle=0x{:08X}, activity=0x{:08X}", handle, activity);
@@ -1737,8 +1647,8 @@ static ResultCode SignalProcessWideKey(Core::System& system, VAddr condition_var
 // Wait for an address (via Address Arbiter)
 static ResultCode WaitForAddress(Core::System& system, VAddr address, u32 type, s32 value,
                                  s64 timeout) {
-    LOG_TRACE(Kernel_SVC, "called, address=0x{:X}, type=0x{:X}, value=0x{:X}, timeout={}", address,
-              type, value, timeout);
+    LOG_WARNING(Kernel_SVC, "called, address=0x{:X}, type=0x{:X}, value=0x{:X}, timeout={}",
+                address, type, value, timeout);
 
     // If the passed address is a kernel virtual address, return invalid memory state.
     if (Memory::IsKernelVirtualAddress(address)) {
@@ -1760,8 +1670,8 @@ static ResultCode WaitForAddress(Core::System& system, VAddr address, u32 type,
 // Signals to an address (via Address Arbiter)
 static ResultCode SignalToAddress(Core::System& system, VAddr address, u32 type, s32 value,
                                   s32 num_to_wake) {
-    LOG_TRACE(Kernel_SVC, "called, address=0x{:X}, type=0x{:X}, value=0x{:X}, num_to_wake=0x{:X}",
-              address, type, value, num_to_wake);
+    LOG_WARNING(Kernel_SVC, "called, address=0x{:X}, type=0x{:X}, value=0x{:X}, num_to_wake=0x{:X}",
+                address, type, value, num_to_wake);
 
     // If the passed address is a kernel virtual address, return invalid memory state.
     if (Memory::IsKernelVirtualAddress(address)) {
@@ -2393,8 +2303,8 @@ static const FunctionDef SVC_Table[] = {
     {0x29, SvcWrap<GetInfo>, "GetInfo"},
     {0x2A, nullptr, "FlushEntireDataCache"},
     {0x2B, nullptr, "FlushDataCache"},
-    {0x2C, SvcWrap<MapPhysicalMemory>, "MapPhysicalMemory"},
-    {0x2D, SvcWrap<UnmapPhysicalMemory>, "UnmapPhysicalMemory"},
+    {0x2C, nullptr, "MapPhysicalMemory"},
+    {0x2D, nullptr, "UnmapPhysicalMemory"},
     {0x2E, nullptr, "GetFutureThreadInfo"},
     {0x2F, nullptr, "GetLastThreadInfo"},
     {0x30, SvcWrap<GetResourceLimitLimitValue>, "GetResourceLimitLimitValue"},

src/core/hle/kernel/svc_wrap.h

@@ -32,11 +32,6 @@ void SvcWrap(Core::System& system) {
     FuncReturn(system, func(system, Param(system, 0)).raw);
 }
 
-template <ResultCode func(Core::System&, u64, u64)>
-void SvcWrap(Core::System& system) {
-    FuncReturn(system, func(system, Param(system, 0), Param(system, 1)).raw);
-}
-
 template <ResultCode func(Core::System&, u32)>
 void SvcWrap(Core::System& system) {
     FuncReturn(system, func(system, static_cast<u32>(Param(system, 0))).raw);

src/core/hle/kernel/vm_manager.cpp

@@ -8,11 +8,10 @@
 #include "common/assert.h"
 #include "common/logging/log.h"
 #include "common/memory_hook.h"
+#include "core/arm/arm_interface.h"
 #include "core/core.h"
 #include "core/file_sys/program_metadata.h"
 #include "core/hle/kernel/errors.h"
-#include "core/hle/kernel/process.h"
-#include "core/hle/kernel/resource_limit.h"
 #include "core/hle/kernel/vm_manager.h"
 #include "core/memory.h"
 #include "core/memory_setup.h"
@@ -50,14 +49,10 @@ bool VirtualMemoryArea::CanBeMergedWith(const VirtualMemoryArea& next) const {
         type != next.type) {
         return false;
     }
-    if ((attribute & MemoryAttribute::DeviceMapped) == MemoryAttribute::DeviceMapped) {
-        // TODO: Can device mapped memory be merged sanely?
-        // Not merging it may cause inaccuracies versus hardware when memory layout is queried.
+    if (type == VMAType::AllocatedMemoryBlock &&
+        (backing_block != next.backing_block || offset + size != next.offset)) {
         return false;
     }
-    if (type == VMAType::AllocatedMemoryBlock) {
-        return true;
-    }
     if (type == VMAType::BackingMemory && backing_memory + size != next.backing_memory) {
         return false;
     }
@@ -105,7 +100,7 @@ bool VMManager::IsValidHandle(VMAHandle handle) const {
 ResultVal<VMManager::VMAHandle> VMManager::MapMemoryBlock(VAddr target,
                                                           std::shared_ptr<std::vector<u8>> block,
                                                           std::size_t offset, u64 size,
-                                                          MemoryState state, VMAPermission perm) {
+                                                          MemoryState state) {
     ASSERT(block != nullptr);
     ASSERT(offset + size <= block->size());
 
@@ -114,8 +109,17 @@ ResultVal<VMManager::VMAHandle> VMManager::MapMemoryBlock(VAddr target,
     VirtualMemoryArea& final_vma = vma_handle->second;
     ASSERT(final_vma.size == size);
 
+    system.ArmInterface(0).MapBackingMemory(target, size, block->data() + offset,
+                                            VMAPermission::ReadWriteExecute);
+    system.ArmInterface(1).MapBackingMemory(target, size, block->data() + offset,
+                                            VMAPermission::ReadWriteExecute);
+    system.ArmInterface(2).MapBackingMemory(target, size, block->data() + offset,
+                                            VMAPermission::ReadWriteExecute);
+    system.ArmInterface(3).MapBackingMemory(target, size, block->data() + offset,
+                                            VMAPermission::ReadWriteExecute);
+
     final_vma.type = VMAType::AllocatedMemoryBlock;
-    final_vma.permissions = perm;
+    final_vma.permissions = VMAPermission::ReadWrite;
     final_vma.state = state;
     final_vma.backing_block = std::move(block);
     final_vma.offset = offset;
@@ -133,6 +137,11 @@ ResultVal<VMManager::VMAHandle> VMManager::MapBackingMemory(VAddr target, u8* me
     VirtualMemoryArea& final_vma = vma_handle->second;
     ASSERT(final_vma.size == size);
 
+    system.ArmInterface(0).MapBackingMemory(target, size, memory, VMAPermission::ReadWriteExecute);
+    system.ArmInterface(1).MapBackingMemory(target, size, memory, VMAPermission::ReadWriteExecute);
+    system.ArmInterface(2).MapBackingMemory(target, size, memory, VMAPermission::ReadWriteExecute);
+    system.ArmInterface(3).MapBackingMemory(target, size, memory, VMAPermission::ReadWriteExecute);
+
     final_vma.type = VMAType::BackingMemory;
     final_vma.permissions = VMAPermission::ReadWrite;
     final_vma.state = state;
@@ -221,6 +230,11 @@ ResultCode VMManager::UnmapRange(VAddr target, u64 size) {
 
     ASSERT(FindVMA(target)->second.size >= size);
 
+    system.ArmInterface(0).UnmapMemory(target, size);
+    system.ArmInterface(1).UnmapMemory(target, size);
+    system.ArmInterface(2).UnmapMemory(target, size);
+    system.ArmInterface(3).UnmapMemory(target, size);
+
     return RESULT_SUCCESS;
 }
 
@@ -294,166 +308,6 @@ ResultVal<VAddr> VMManager::SetHeapSize(u64 size) {
     return MakeResult<VAddr>(heap_region_base);
 }
 
-ResultCode VMManager::MapPhysicalMemory(VAddr target, u64 size) {
-    const auto end_addr = target + size;
-    const auto last_addr = end_addr - 1;
-    VAddr cur_addr = target;
-
-    ResultCode result = RESULT_SUCCESS;
-
-    // Check how much memory we've already mapped.
-    const auto mapped_size_result = SizeOfAllocatedVMAsInRange(target, size);
-    if (mapped_size_result.Failed()) {
-        return mapped_size_result.Code();
-    }
-
-    // If we've already mapped the desired amount, return early.
-    const std::size_t mapped_size = *mapped_size_result;
-    if (mapped_size == size) {
-        return RESULT_SUCCESS;
-    }
-
-    // Check that we can map the memory we want.
-    const auto res_limit = system.CurrentProcess()->GetResourceLimit();
-    const u64 physmem_remaining = res_limit->GetMaxResourceValue(ResourceType::PhysicalMemory) -
-                                  res_limit->GetCurrentResourceValue(ResourceType::PhysicalMemory);
-    if (physmem_remaining < (size - mapped_size)) {
-        return ERR_RESOURCE_LIMIT_EXCEEDED;
-    }
-
-    // Keep track of the memory regions we unmap.
-    std::vector<std::pair<u64, u64>> mapped_regions;
-
-    // Iterate, trying to map memory.
-    {
-        cur_addr = target;
-
-        auto iter = FindVMA(target);
-        ASSERT_MSG(iter != vma_map.end(), "MapPhysicalMemory iter != end");
-
-        while (true) {
-            const auto& vma = iter->second;
-            const auto vma_start = vma.base;
-            const auto vma_end = vma_start + vma.size;
-            const auto vma_last = vma_end - 1;
-
-            // Map the memory block
-            const auto map_size = std::min(end_addr - cur_addr, vma_end - cur_addr);
-            if (vma.state == MemoryState::Unmapped) {
-                const auto map_res =
-                    MapMemoryBlock(cur_addr, std::make_shared<std::vector<u8>>(map_size, 0), 0,
-                                   map_size, MemoryState::Heap, VMAPermission::ReadWrite);
-                result = map_res.Code();
-                if (result.IsError()) {
-                    break;
-                }
-
-                mapped_regions.emplace_back(cur_addr, map_size);
-            }
-
-            // Break once we hit the end of the range.
-            if (last_addr <= vma_last) {
-                break;
-            }
-
-            // Advance to the next block.
-            cur_addr = vma_end;
-            iter = FindVMA(cur_addr);
-            ASSERT_MSG(iter != vma_map.end(), "MapPhysicalMemory iter != end");
-        }
-    }
-
-    // If we failed, unmap memory.
-    if (result.IsError()) {
-        for (const auto [unmap_address, unmap_size] : mapped_regions) {
-            ASSERT_MSG(UnmapRange(unmap_address, unmap_size).IsSuccess(),
-                       "MapPhysicalMemory un-map on error");
-        }
-
-        return result;
-    }
-
-    // Update amount of mapped physical memory.
-    physical_memory_mapped += size - mapped_size;
-
-    return RESULT_SUCCESS;
-}
-
-ResultCode VMManager::UnmapPhysicalMemory(VAddr target, u64 size) {
-    const auto end_addr = target + size;
-    const auto last_addr = end_addr - 1;
-    VAddr cur_addr = target;
-
-    ResultCode result = RESULT_SUCCESS;
-
-    // Check how much memory is currently mapped.
-    const auto mapped_size_result = SizeOfUnmappablePhysicalMemoryInRange(target, size);
-    if (mapped_size_result.Failed()) {
-        return mapped_size_result.Code();
-    }
-
-    // If we've already unmapped all the memory, return early.
-    const std::size_t mapped_size = *mapped_size_result;
-    if (mapped_size == 0) {
-        return RESULT_SUCCESS;
-    }
-
-    // Keep track of the memory regions we unmap.
-    std::vector<std::pair<u64, u64>> unmapped_regions;
-
-    // Try to unmap regions.
-    {
-        cur_addr = target;
-
-        auto iter = FindVMA(target);
-        ASSERT_MSG(iter != vma_map.end(), "UnmapPhysicalMemory iter != end");
-
-        while (true) {
-            const auto& vma = iter->second;
-            const auto vma_start = vma.base;
-            const auto vma_end = vma_start + vma.size;
-            const auto vma_last = vma_end - 1;
-
-            // Unmap the memory block
-            const auto unmap_size = std::min(end_addr - cur_addr, vma_end - cur_addr);
-            if (vma.state == MemoryState::Heap) {
-                result = UnmapRange(cur_addr, unmap_size);
-                if (result.IsError()) {
-                    break;
-                }
-
-                unmapped_regions.emplace_back(cur_addr, unmap_size);
-            }
-
-            // Break once we hit the end of the range.
-            if (last_addr <= vma_last) {
-                break;
-            }
-
-            // Advance to the next block.
-            cur_addr = vma_end;
-            iter = FindVMA(cur_addr);
-            ASSERT_MSG(iter != vma_map.end(), "UnmapPhysicalMemory iter != end");
-        }
-    }
-
-    // If we failed, re-map regions.
-    // TODO: Preserve memory contents?
-    if (result.IsError()) {
-        for (const auto [map_address, map_size] : unmapped_regions) {
-            const auto remap_res =
-                MapMemoryBlock(map_address, std::make_shared<std::vector<u8>>(map_size, 0), 0,
-                               map_size, MemoryState::Heap, VMAPermission::None);
-            ASSERT_MSG(remap_res.Succeeded(), "UnmapPhysicalMemory re-map on error");
-        }
-    }
-
-    // Update mapped amount
-    physical_memory_mapped -= mapped_size;
-
-    return RESULT_SUCCESS;
-}
-
 ResultCode VMManager::MapCodeMemory(VAddr dst_address, VAddr src_address, u64 size) {
     constexpr auto ignore_attribute = MemoryAttribute::LockedForIPC | MemoryAttribute::DeviceMapped;
     const auto src_check_result = CheckRangeState(
@@ -601,7 +455,7 @@ ResultCode VMManager::MirrorMemory(VAddr dst_addr, VAddr src_addr, u64 size, Mem
     // Protect mirror with permissions from old region
     Reprotect(new_vma, vma->second.permissions);
     // Remove permissions from old region
-    ReprotectRange(src_addr, size, VMAPermission::None);
+    Reprotect(vma, VMAPermission::None);
 
     return RESULT_SUCCESS;
 }
@@ -734,14 +588,14 @@ VMManager::VMAIter VMManager::SplitVMA(VMAIter vma_handle, u64 offset_in_vma) {
 VMManager::VMAIter VMManager::MergeAdjacent(VMAIter iter) {
     const VMAIter next_vma = std::next(iter);
     if (next_vma != vma_map.end() && iter->second.CanBeMergedWith(next_vma->second)) {
-        MergeAdjacentVMA(iter->second, next_vma->second);
+        iter->second.size += next_vma->second.size;
         vma_map.erase(next_vma);
     }
 
     if (iter != vma_map.begin()) {
         VMAIter prev_vma = std::prev(iter);
         if (prev_vma->second.CanBeMergedWith(iter->second)) {
-            MergeAdjacentVMA(prev_vma->second, iter->second);
+            prev_vma->second.size += iter->second.size;
             vma_map.erase(iter);
             iter = prev_vma;
         }
@@ -750,38 +604,6 @@ VMManager::VMAIter VMManager::MergeAdjacent(VMAIter iter) {
     return iter;
 }
 
-void VMManager::MergeAdjacentVMA(VirtualMemoryArea& left, const VirtualMemoryArea& right) {
-    ASSERT(left.CanBeMergedWith(right));
-
-    // Always merge allocated memory blocks, even when they don't share the same backing block.
-    if (left.type == VMAType::AllocatedMemoryBlock &&
-        (left.backing_block != right.backing_block || left.offset + left.size != right.offset)) {
-        // Check if we can save work.
-        if (left.offset == 0 && left.size == left.backing_block->size()) {
-            // Fast case: left is an entire backing block.
-            left.backing_block->insert(left.backing_block->end(),
-                                       right.backing_block->begin() + right.offset,
-                                       right.backing_block->begin() + right.offset + right.size);
-        } else {
-            // Slow case: make a new memory block for left and right.
-            auto new_memory = std::make_shared<std::vector<u8>>();
-            new_memory->insert(new_memory->end(), left.backing_block->begin() + left.offset,
-                               left.backing_block->begin() + left.offset + left.size);
-            new_memory->insert(new_memory->end(), right.backing_block->begin() + right.offset,
-                               right.backing_block->begin() + right.offset + right.size);
-            left.backing_block = new_memory;
-            left.offset = 0;
-        }
-
-        // Page table update is needed, because backing memory changed.
-        left.size += right.size;
-        UpdatePageTableForVMA(left);
-    } else {
-        // Just update the size.
-        left.size += right.size;
-    }
-}
-
 void VMManager::UpdatePageTableForVMA(const VirtualMemoryArea& vma) {
     switch (vma.type) {
     case VMAType::Free:
@@ -956,84 +778,6 @@ VMManager::CheckResults VMManager::CheckRangeState(VAddr address, u64 size, Memo
         std::make_tuple(initial_state, initial_permissions, initial_attributes & ~ignore_mask));
 }
 
-ResultVal<std::size_t> VMManager::SizeOfAllocatedVMAsInRange(VAddr address,
-                                                             std::size_t size) const {
-    const VAddr end_addr = address + size;
-    const VAddr last_addr = end_addr - 1;
-    std::size_t mapped_size = 0;
-
-    VAddr cur_addr = address;
-    auto iter = FindVMA(cur_addr);
-    ASSERT_MSG(iter != vma_map.end(), "SizeOfAllocatedVMAsInRange iter != end");
-
-    while (true) {
-        const auto& vma = iter->second;
-        const VAddr vma_start = vma.base;
-        const VAddr vma_end = vma_start + vma.size;
-        const VAddr vma_last = vma_end - 1;
-
-        // Add size if relevant.
-        if (vma.state != MemoryState::Unmapped) {
-            mapped_size += std::min(end_addr - cur_addr, vma_end - cur_addr);
-        }
-
-        // Break once we hit the end of the range.
-        if (last_addr <= vma_last) {
-            break;
-        }
-
-        // Advance to the next block.
-        cur_addr = vma_end;
-        iter = std::next(iter);
-        ASSERT_MSG(iter != vma_map.end(), "SizeOfAllocatedVMAsInRange iter != end");
-    }
-
-    return MakeResult(mapped_size);
-}
-
-ResultVal<std::size_t> VMManager::SizeOfUnmappablePhysicalMemoryInRange(VAddr address,
-                                                                        std::size_t size) const {
-    const VAddr end_addr = address + size;
-    const VAddr last_addr = end_addr - 1;
-    std::size_t mapped_size = 0;
-
-    VAddr cur_addr = address;
-    auto iter = FindVMA(cur_addr);
-    ASSERT_MSG(iter != vma_map.end(), "SizeOfUnmappablePhysicalMemoryInRange iter != end");
-
-    while (true) {
-        const auto& vma = iter->second;
-        const auto vma_start = vma.base;
-        const auto vma_end = vma_start + vma.size;
-        const auto vma_last = vma_end - 1;
-        const auto state = vma.state;
-        const auto attr = vma.attribute;
-
-        // Memory within region must be free or mapped heap.
-        if (!((state == MemoryState::Heap && attr == MemoryAttribute::None) ||
-              (state == MemoryState::Unmapped))) {
-            return ERR_INVALID_ADDRESS_STATE;
-        }
-
-        // Add size if relevant.
-        if (state != MemoryState::Unmapped) {
-            mapped_size += std::min(end_addr - cur_addr, vma_end - cur_addr);
-        }
-
-        // Break once we hit the end of the range.
-        if (last_addr <= vma_last) {
-            break;
-        }
-
-        // Advance to the next block.
-        cur_addr = vma_end;
-        iter = std::next(iter);
-        ASSERT_MSG(iter != vma_map.end(), "SizeOfUnmappablePhysicalMemoryInRange iter != end");
-    }
-
-    return MakeResult(mapped_size);
-}
-
 u64 VMManager::GetTotalPhysicalMemoryAvailable() const {
     LOG_WARNING(Kernel, "(STUBBED) called");
     return 0xF8000000;

src/core/hle/kernel/vm_manager.h

@@ -349,8 +349,7 @@ public:
      * @param state MemoryState tag to attach to the VMA.
      */
     ResultVal<VMAHandle> MapMemoryBlock(VAddr target, std::shared_ptr<std::vector<u8>> block,
-                                        std::size_t offset, u64 size, MemoryState state,
-                                        VMAPermission perm = VMAPermission::ReadWrite);
+                                        std::size_t offset, u64 size, MemoryState state);
 
     /**
      * Maps an unmanaged host memory pointer at a given address.
@@ -451,34 +450,6 @@ public:
     ///
     ResultVal<VAddr> SetHeapSize(u64 size);
 
-    /// Maps memory at a given address.
-    ///
-    /// @param addr The virtual address to map memory at.
-    /// @param size The amount of memory to map.
-    ///
-    /// @note The destination address must lie within the Map region.
-    ///
-    /// @note This function requires that SystemResourceSize be non-zero,
-    ///       however, this is just because if it were not then the
-    ///       resulting page tables could be exploited on hardware by
-    ///       a malicious program. SystemResource usage does not need
-    ///       to be explicitly checked or updated here.
-    ResultCode MapPhysicalMemory(VAddr target, u64 size);
-
-    /// Unmaps memory at a given address.
-    ///
-    /// @param addr The virtual address to unmap memory at.
-    /// @param size The amount of memory to unmap.
-    ///
-    /// @note The destination address must lie within the Map region.
-    ///
-    /// @note This function requires that SystemResourceSize be non-zero,
-    ///       however, this is just because if it were not then the
-    ///       resulting page tables could be exploited on hardware by
-    ///       a malicious program. SystemResource usage does not need
-    ///       to be explicitly checked or updated here.
-    ResultCode UnmapPhysicalMemory(VAddr target, u64 size);
-
     /// Maps a region of memory as code memory.
     ///
     /// @param dst_address The base address of the region to create the aliasing memory region.
@@ -686,11 +657,6 @@ private:
      */
     VMAIter MergeAdjacent(VMAIter vma);
 
-    /**
-     * Merges two adjacent VMAs.
-     */
-    void MergeAdjacentVMA(VirtualMemoryArea& left, const VirtualMemoryArea& right);
-
     /// Updates the pages corresponding to this VMA so they match the VMA's attributes.
     void UpdatePageTableForVMA(const VirtualMemoryArea& vma);
 
@@ -735,13 +701,6 @@ private:
                                  MemoryAttribute attribute_mask, MemoryAttribute attribute,
                                  MemoryAttribute ignore_mask) const;
 
-    /// Gets the amount of memory currently mapped (state != Unmapped) in a range.
-    ResultVal<std::size_t> SizeOfAllocatedVMAsInRange(VAddr address, std::size_t size) const;
-
-    /// Gets the amount of memory unmappable by UnmapPhysicalMemory in a range.
-    ResultVal<std::size_t> SizeOfUnmappablePhysicalMemoryInRange(VAddr address,
-                                                                 std::size_t size) const;
-
     /**
      * A map covering the entirety of the managed address space, keyed by the `base` field of each
      * VMA. It must always be modified by splitting or merging VMAs, so that the invariant
@@ -783,11 +742,6 @@ private:
     // end of the range. This is essentially 'base_address + current_size'.
     VAddr heap_end = 0;
 
-    // The current amount of memory mapped via MapPhysicalMemory.
-    // This is used here (and in Nintendo's kernel) only for debugging, and does not impact
-    // any behavior.
-    u64 physical_memory_mapped = 0;
-
     Core::System& system;
 };
 } // namespace Kernel

src/core/hle/service/am/am.cpp

@@ -266,8 +266,8 @@ ISelfController::ISelfController(std::shared_ptr<NVFlinger::NVFlinger> nvflinger
         {65, nullptr, "ReportUserIsActive"},
         {66, nullptr, "GetCurrentIlluminance"},
         {67, nullptr, "IsIlluminanceAvailable"},
-        {68, &ISelfController::SetAutoSleepDisabled, "SetAutoSleepDisabled"},
-        {69, &ISelfController::IsAutoSleepDisabled, "IsAutoSleepDisabled"},
+        {68, nullptr, "SetAutoSleepDisabled"},
+        {69, nullptr, "IsAutoSleepDisabled"},
         {70, nullptr, "ReportMultimediaError"},
         {71, nullptr, "GetCurrentIlluminanceEx"},
         {80, nullptr, "SetWirelessPriorityMode"},
@@ -454,34 +454,6 @@ void ISelfController::GetIdleTimeDetectionExtension(Kernel::HLERequestContext& c
     rb.Push<u32>(idle_time_detection_extension);
 }
 
-void ISelfController::SetAutoSleepDisabled(Kernel::HLERequestContext& ctx) {
-    IPC::RequestParser rp{ctx};
-    is_auto_sleep_disabled = rp.Pop<bool>();
-
-    // On the system itself, if the previous state of is_auto_sleep_disabled
-    // differed from the current value passed in, it'd signify the internal
-    // window manager to update (and also increment some statistics like update counts)
-    //
-    // It'd also indicate this change to an idle handling context.
-    //
-    // However, given we're emulating this behavior, most of this can be ignored
-    // and it's sufficient to simply set the member variable for querying via
-    // IsAutoSleepDisabled().
-
-    LOG_DEBUG(Service_AM, "called. is_auto_sleep_disabled={}", is_auto_sleep_disabled);
-
-    IPC::ResponseBuilder rb{ctx, 2};
-    rb.Push(RESULT_SUCCESS);
-}
-
-void ISelfController::IsAutoSleepDisabled(Kernel::HLERequestContext& ctx) {
-    LOG_DEBUG(Service_AM, "called.");
-
-    IPC::ResponseBuilder rb{ctx, 3};
-    rb.Push(RESULT_SUCCESS);
-    rb.Push(is_auto_sleep_disabled);
-}
-
 void ISelfController::GetAccumulatedSuspendedTickValue(Kernel::HLERequestContext& ctx) {
     LOG_DEBUG(Service_AM, "called.");
 

src/core/hle/service/am/am.h

@@ -133,8 +133,6 @@ private:
     void SetHandlesRequestToDisplay(Kernel::HLERequestContext& ctx);
     void SetIdleTimeDetectionExtension(Kernel::HLERequestContext& ctx);
     void GetIdleTimeDetectionExtension(Kernel::HLERequestContext& ctx);
-    void SetAutoSleepDisabled(Kernel::HLERequestContext& ctx);
-    void IsAutoSleepDisabled(Kernel::HLERequestContext& ctx);
     void GetAccumulatedSuspendedTickValue(Kernel::HLERequestContext& ctx);
     void GetAccumulatedSuspendedTickChangedEvent(Kernel::HLERequestContext& ctx);
 
@@ -144,7 +142,6 @@ private:
 
     u32 idle_time_detection_extension = 0;
     u64 num_fatal_sections_entered = 0;
-    bool is_auto_sleep_disabled = false;
 };
 
 class ICommonStateGetter final : public ServiceFramework<ICommonStateGetter> {

src/core/hle/service/audio/audren_u.cpp

@@ -25,8 +25,7 @@ namespace Service::Audio {
 
 class IAudioRenderer final : public ServiceFramework<IAudioRenderer> {
 public:
-    explicit IAudioRenderer(AudioCore::AudioRendererParameter audren_params,
-                            const std::size_t instance_number)
+    explicit IAudioRenderer(AudioCore::AudioRendererParameter audren_params)
         : ServiceFramework("IAudioRenderer") {
         // clang-format off
         static const FunctionInfo functions[] = {
@@ -49,8 +48,8 @@ public:
         auto& system = Core::System::GetInstance();
         system_event = Kernel::WritableEvent::CreateEventPair(
             system.Kernel(), Kernel::ResetType::Manual, "IAudioRenderer:SystemEvent");
-        renderer = std::make_unique<AudioCore::AudioRenderer>(
-            system.CoreTiming(), audren_params, system_event.writable, instance_number);
+        renderer = std::make_unique<AudioCore::AudioRenderer>(system.CoreTiming(), audren_params,
+                                                              system_event.writable);
     }
 
 private:
@@ -608,7 +607,7 @@ void AudRenU::OpenAudioRendererImpl(Kernel::HLERequestContext& ctx) {
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
 
     rb.Push(RESULT_SUCCESS);
-    rb.PushIpcInterface<IAudioRenderer>(params, audren_instance_count++);
+    rb.PushIpcInterface<IAudioRenderer>(params);
 }
 
 bool AudRenU::IsFeatureSupported(AudioFeatures feature, u32_le revision) const {

src/core/hle/service/audio/audren_u.h

@@ -33,7 +33,6 @@ private:
     };
 
     bool IsFeatureSupported(AudioFeatures feature, u32_le revision) const;
-    std::size_t audren_instance_count = 0;
 };
 
 } // namespace Service::Audio

src/core/hle/service/friend/friend.cpp

@@ -22,7 +22,7 @@ public:
             {0, nullptr, "GetCompletionEvent"},
             {1, nullptr, "Cancel"},
             {10100, nullptr, "GetFriendListIds"},
-            {10101, &IFriendService::GetFriendList, "GetFriendList"},
+            {10101, nullptr, "GetFriendList"},
             {10102, nullptr, "UpdateFriendInfo"},
             {10110, nullptr, "GetFriendProfileImage"},
             {10200, nullptr, "SendFriendRequestForApplication"},
@@ -99,23 +99,6 @@ public:
     }
 
 private:
-    enum class PresenceFilter : u32 {
-        None = 0,
-        Online = 1,
-        OnlinePlay = 2,
-        OnlineOrOnlinePlay = 3,
-    };
-
-    struct SizedFriendFilter {
-        PresenceFilter presence;
-        u8 is_favorite;
-        u8 same_app;
-        u8 same_app_played;
-        u8 arbitary_app_played;
-        u64 group_id;
-    };
-    static_assert(sizeof(SizedFriendFilter) == 0x10, "SizedFriendFilter is an invalid size");
-
     void DeclareCloseOnlinePlaySession(Kernel::HLERequestContext& ctx) {
         // Stub used by Splatoon 2
         LOG_WARNING(Service_ACC, "(STUBBED) called");
@@ -129,22 +112,6 @@ private:
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
     }
-
-    void GetFriendList(Kernel::HLERequestContext& ctx) {
-        IPC::RequestParser rp{ctx};
-        const auto friend_offset = rp.Pop<u32>();
-        const auto uuid = rp.PopRaw<Common::UUID>();
-        [[maybe_unused]] const auto filter = rp.PopRaw<SizedFriendFilter>();
-        const auto pid = rp.Pop<u64>();
-        LOG_WARNING(Service_ACC, "(STUBBED) called, offset={}, uuid={}, pid={}", friend_offset,
-                    uuid.Format(), pid);
-
-        IPC::ResponseBuilder rb{ctx, 3};
-        rb.Push(RESULT_SUCCESS);
-
-        rb.Push<u32>(0); // Friend count
-        // TODO(ogniK): Return a buffer of u64s which are the "NetworkServiceAccountId"
-    }
 };
 
 class INotificationService final : public ServiceFramework<INotificationService> {

src/core/hle/service/ldr/ldr.cpp

@@ -345,16 +345,14 @@ public:
             vm_manager
                 .MirrorMemory(*map_address, nro_address, nro_size, Kernel::MemoryState::ModuleCode)
                 .IsSuccess());
-        ASSERT(vm_manager.ReprotectRange(nro_address, nro_size, Kernel::VMAPermission::None)
-                   .IsSuccess());
+        ASSERT(vm_manager.UnmapRange(nro_address, nro_size).IsSuccess());
 
         if (bss_size > 0) {
             ASSERT(vm_manager
                        .MirrorMemory(*map_address + nro_size, bss_address, bss_size,
                                      Kernel::MemoryState::ModuleCode)
                        .IsSuccess());
-            ASSERT(vm_manager.ReprotectRange(bss_address, bss_size, Kernel::VMAPermission::None)
-                       .IsSuccess());
+            ASSERT(vm_manager.UnmapRange(bss_address, bss_size).IsSuccess());
         }
 
         vm_manager.ReprotectRange(*map_address, header.text_size,
@@ -366,8 +364,7 @@ public:
 
         Core::System::GetInstance().InvalidateCpuInstructionCaches();
 
-        nro.insert_or_assign(*map_address,
-                             NROInfo{hash, nro_address, nro_size, bss_address, bss_size});
+        nro.insert_or_assign(*map_address, NROInfo{hash, nro_size + bss_size});
 
         IPC::ResponseBuilder rb{ctx, 4};
         rb.Push(RESULT_SUCCESS);
@@ -412,23 +409,9 @@ public:
         }
 
         auto& vm_manager = Core::CurrentProcess()->VMManager();
-        const auto& nro_info = iter->second;
+        const auto& nro_size = iter->second.size;
 
-        // Unmap the mirrored memory
-        ASSERT(
-            vm_manager.UnmapRange(nro_address, nro_info.nro_size + nro_info.bss_size).IsSuccess());
-
-        // Reprotect the source memory
-        ASSERT(vm_manager
-                   .ReprotectRange(nro_info.nro_address, nro_info.nro_size,
-                                   Kernel::VMAPermission::ReadWrite)
-                   .IsSuccess());
-        if (nro_info.bss_size > 0) {
-            ASSERT(vm_manager
-                       .ReprotectRange(nro_info.bss_address, nro_info.bss_size,
-                                       Kernel::VMAPermission::ReadWrite)
-                       .IsSuccess());
-        }
+        ASSERT(vm_manager.UnmapRange(nro_address, nro_size).IsSuccess());
 
         Core::System::GetInstance().InvalidateCpuInstructionCaches();
 
@@ -490,10 +473,7 @@ private:
 
     struct NROInfo {
         SHA256Hash hash;
-        VAddr nro_address;
-        u64 nro_size;
-        VAddr bss_address;
-        u64 bss_size;
+        u64 size;
     };
 
     bool initialized = false;

src/core/hle/service/mii/mii.cpp

@@ -48,7 +48,7 @@ public:
             {19, nullptr, "Export"},
             {20, nullptr, "IsBrokenDatabaseWithClearFlag"},
             {21, &IDatabaseService::GetIndex, "GetIndex"},
-            {22, &IDatabaseService::SetInterfaceVersion, "SetInterfaceVersion"},
+            {22, nullptr, "SetInterfaceVersion"},
             {23, nullptr, "Convert"},
         };
         // clang-format on
@@ -350,22 +350,8 @@ private:
         rb.Push(index);
     }
 
-    void SetInterfaceVersion(Kernel::HLERequestContext& ctx) {
-        IPC::RequestParser rp{ctx};
-        current_interface_version = rp.PopRaw<u32>();
-
-        LOG_DEBUG(Service_Mii, "called, interface_version={:08X}", current_interface_version);
-
-        UNIMPLEMENTED_IF(current_interface_version != 1);
-
-        IPC::ResponseBuilder rb{ctx, 2};
-        rb.Push(RESULT_SUCCESS);
-    }
-
     MiiManager db;
 
-    u32 current_interface_version = 0;
-
     // Last read offsets of Get functions
     std::array<u32, 4> offsets{};
 };

src/core/hle/service/mii/mii_manager.cpp

@@ -175,6 +175,10 @@ MiiStoreData ConvertInfoToStoreData(const MiiInfo& info) {
 } // namespace
 
 std::ostream& operator<<(std::ostream& os, Source source) {
+    if (static_cast<std::size_t>(source) >= SOURCE_NAMES.size()) {
+        return os << "[UNKNOWN SOURCE]";
+    }
+
     os << SOURCE_NAMES.at(static_cast<std::size_t>(source));
     return os;
 }

src/core/hle/service/pm/pm.cpp

@@ -3,44 +3,11 @@
 // Refer to the license.txt file included.
 
 #include "core/hle/ipc_helpers.h"
-#include "core/hle/kernel/kernel.h"
-#include "core/hle/kernel/process.h"
 #include "core/hle/service/pm/pm.h"
 #include "core/hle/service/service.h"
 
 namespace Service::PM {
 
-namespace {
-
-constexpr ResultCode ERROR_PROCESS_NOT_FOUND{ErrorModule::PM, 1};
-
-constexpr u64 NO_PROCESS_FOUND_PID{0};
-
-std::optional<Kernel::SharedPtr<Kernel::Process>> SearchProcessList(
-    const std::vector<Kernel::SharedPtr<Kernel::Process>>& process_list,
-    std::function<bool(const Kernel::SharedPtr<Kernel::Process>&)> predicate) {
-    const auto iter = std::find_if(process_list.begin(), process_list.end(), predicate);
-
-    if (iter == process_list.end()) {
-        return std::nullopt;
-    }
-
-    return *iter;
-}
-
-void GetApplicationPidGeneric(Kernel::HLERequestContext& ctx,
-                              const std::vector<Kernel::SharedPtr<Kernel::Process>>& process_list) {
-    const auto process = SearchProcessList(process_list, [](const auto& process) {
-        return process->GetProcessID() == Kernel::Process::ProcessIDMin;
-    });
-
-    IPC::ResponseBuilder rb{ctx, 4};
-    rb.Push(RESULT_SUCCESS);
-    rb.Push(process.has_value() ? (*process)->GetProcessID() : NO_PROCESS_FOUND_PID);
-}
-
-} // Anonymous namespace
-
 class BootMode final : public ServiceFramework<BootMode> {
 public:
     explicit BootMode() : ServiceFramework{"pm:bm"} {
@@ -74,15 +41,14 @@ private:
 
 class DebugMonitor final : public ServiceFramework<DebugMonitor> {
 public:
-    explicit DebugMonitor(const Kernel::KernelCore& kernel)
-        : ServiceFramework{"pm:dmnt"}, kernel(kernel) {
+    explicit DebugMonitor() : ServiceFramework{"pm:dmnt"} {
         // clang-format off
         static const FunctionInfo functions[] = {
             {0, nullptr, "GetDebugProcesses"},
             {1, nullptr, "StartDebugProcess"},
-            {2, &DebugMonitor::GetTitlePid, "GetTitlePid"},
+            {2, nullptr, "GetTitlePid"},
             {3, nullptr, "EnableDebugForTitleId"},
-            {4, &DebugMonitor::GetApplicationPid, "GetApplicationPid"},
+            {4, nullptr, "GetApplicationPid"},
             {5, nullptr, "EnableDebugForApplication"},
             {6, nullptr, "DisableDebug"},
         };
@@ -90,77 +56,21 @@ public:
 
         RegisterHandlers(functions);
     }
-
-private:
-    void GetTitlePid(Kernel::HLERequestContext& ctx) {
-        IPC::RequestParser rp{ctx};
-        const auto title_id = rp.PopRaw<u64>();
-
-        LOG_DEBUG(Service_PM, "called, title_id={:016X}", title_id);
-
-        const auto process =
-            SearchProcessList(kernel.GetProcessList(), [title_id](const auto& process) {
-                return process->GetTitleID() == title_id;
-            });
-
-        if (!process.has_value()) {
-            IPC::ResponseBuilder rb{ctx, 2};
-            rb.Push(ERROR_PROCESS_NOT_FOUND);
-            return;
-        }
-
-        IPC::ResponseBuilder rb{ctx, 4};
-        rb.Push(RESULT_SUCCESS);
-        rb.Push((*process)->GetProcessID());
-    }
-
-    void GetApplicationPid(Kernel::HLERequestContext& ctx) {
-        LOG_DEBUG(Service_PM, "called");
-        GetApplicationPidGeneric(ctx, kernel.GetProcessList());
-    }
-
-    const Kernel::KernelCore& kernel;
 };
 
 class Info final : public ServiceFramework<Info> {
 public:
-    explicit Info(const std::vector<Kernel::SharedPtr<Kernel::Process>>& process_list)
-        : ServiceFramework{"pm:info"}, process_list(process_list) {
+    explicit Info() : ServiceFramework{"pm:info"} {
         static const FunctionInfo functions[] = {
-            {0, &Info::GetTitleId, "GetTitleId"},
+            {0, nullptr, "GetTitleId"},
         };
         RegisterHandlers(functions);
     }
-
-private:
-    void GetTitleId(Kernel::HLERequestContext& ctx) {
-        IPC::RequestParser rp{ctx};
-        const auto process_id = rp.PopRaw<u64>();
-
-        LOG_DEBUG(Service_PM, "called, process_id={:016X}", process_id);
-
-        const auto process = SearchProcessList(process_list, [process_id](const auto& process) {
-            return process->GetProcessID() == process_id;
-        });
-
-        if (!process.has_value()) {
-            IPC::ResponseBuilder rb{ctx, 2};
-            rb.Push(ERROR_PROCESS_NOT_FOUND);
-            return;
-        }
-
-        IPC::ResponseBuilder rb{ctx, 4};
-        rb.Push(RESULT_SUCCESS);
-        rb.Push((*process)->GetTitleID());
-    }
-
-    const std::vector<Kernel::SharedPtr<Kernel::Process>>& process_list;
 };
 
 class Shell final : public ServiceFramework<Shell> {
 public:
-    explicit Shell(const Kernel::KernelCore& kernel)
-        : ServiceFramework{"pm:shell"}, kernel(kernel) {
+    explicit Shell() : ServiceFramework{"pm:shell"} {
         // clang-format off
         static const FunctionInfo functions[] = {
             {0, nullptr, "LaunchProcess"},
@@ -169,31 +79,21 @@ public:
             {3, nullptr, "GetProcessEventWaiter"},
             {4, nullptr, "GetProcessEventType"},
             {5, nullptr, "NotifyBootFinished"},
-            {6, &Shell::GetApplicationPid, "GetApplicationPid"},
+            {6, nullptr, "GetApplicationPid"},
             {7, nullptr, "BoostSystemMemoryResourceLimit"},
             {8, nullptr, "EnableAdditionalSystemThreads"},
-            {9, nullptr, "GetUnimplementedEventHandle"},
         };
         // clang-format on
 
         RegisterHandlers(functions);
     }
-
-private:
-    void GetApplicationPid(Kernel::HLERequestContext& ctx) {
-        LOG_DEBUG(Service_PM, "called");
-        GetApplicationPidGeneric(ctx, kernel.GetProcessList());
-    }
-
-    const Kernel::KernelCore& kernel;
 };
 
-void InstallInterfaces(Core::System& system) {
-    std::make_shared<BootMode>()->InstallAsService(system.ServiceManager());
-    std::make_shared<DebugMonitor>(system.Kernel())->InstallAsService(system.ServiceManager());
-    std::make_shared<Info>(system.Kernel().GetProcessList())
-        ->InstallAsService(system.ServiceManager());
-    std::make_shared<Shell>(system.Kernel())->InstallAsService(system.ServiceManager());
+void InstallInterfaces(SM::ServiceManager& sm) {
+    std::make_shared<BootMode>()->InstallAsService(sm);
+    std::make_shared<DebugMonitor>()->InstallAsService(sm);
+    std::make_shared<Info>()->InstallAsService(sm);
+    std::make_shared<Shell>()->InstallAsService(sm);
 }
 
 } // namespace Service::PM

src/core/hle/service/pm/pm.h

@@ -4,8 +4,8 @@
 
 #pragma once
 
-namespace Core {
-class System;
+namespace Service::SM {
+class ServiceManager;
 }
 
 namespace Service::PM {
@@ -16,6 +16,6 @@ enum class SystemBootMode {
 };
 
 /// Registers all PM services with the specified service manager.
-void InstallInterfaces(Core::System& system);
+void InstallInterfaces(SM::ServiceManager& service_manager);
 
 } // namespace Service::PM

src/core/hle/service/service.cpp

@@ -241,7 +241,7 @@ void Init(std::shared_ptr<SM::ServiceManager>& sm, Core::System& system) {
     PCTL::InstallInterfaces(*sm);
     PCV::InstallInterfaces(*sm);
     PlayReport::InstallInterfaces(*sm);
-    PM::InstallInterfaces(system);
+    PM::InstallInterfaces(*sm);
     PSC::InstallInterfaces(*sm);
     PSM::InstallInterfaces(*sm);
     Set::InstallInterfaces(*sm);

src/core/settings.cpp

@@ -85,6 +85,7 @@ void LogSettings() {
     LogSetting("System_RngSeed", Settings::values.rng_seed.value_or(0));
     LogSetting("System_CurrentUser", Settings::values.current_user);
     LogSetting("System_LanguageIndex", Settings::values.language_index);
+    LogSetting("Core_CpuJitEnabled", Settings::values.cpu_jit_enabled);
     LogSetting("Core_UseMultiCore", Settings::values.use_multi_core);
     LogSetting("Renderer_UseResolutionFactor", Settings::values.resolution_factor);
     LogSetting("Renderer_UseFrameLimit", Settings::values.use_frame_limit);

src/core/settings.h

@@ -378,6 +378,7 @@ struct Values {
     std::atomic_bool is_device_reload_pending{true};
 
     // Core
+    bool cpu_jit_enabled;
     bool use_multi_core;
 
     // Data Storage

src/core/telemetry_session.cpp

@@ -168,6 +168,7 @@ void TelemetrySession::AddInitialInfo(Loader::AppLoader& app_loader) {
     AddField(Telemetry::FieldType::UserConfig, "Audio_SinkId", Settings::values.sink_id);
     AddField(Telemetry::FieldType::UserConfig, "Audio_EnableAudioStretching",
              Settings::values.enable_audio_stretching);
+    AddField(Telemetry::FieldType::UserConfig, "Core_UseCpuJit", Settings::values.cpu_jit_enabled);
     AddField(Telemetry::FieldType::UserConfig, "Core_UseMultiCore",
              Settings::values.use_multi_core);
     AddField(Telemetry::FieldType::UserConfig, "Renderer_ResolutionFactor",

src/video_core/CMakeLists.txt

@@ -1,5 +1,4 @@
 add_library(video_core STATIC
-    buffer_cache.h
     dma_pusher.cpp
     dma_pusher.h
     debug_utils/debug_utils.cpp
@@ -44,6 +43,8 @@ add_library(video_core STATIC
     renderer_opengl/gl_device.h
     renderer_opengl/gl_framebuffer_cache.cpp
     renderer_opengl/gl_framebuffer_cache.h
+    renderer_opengl/gl_global_cache.cpp
+    renderer_opengl/gl_global_cache.h
     renderer_opengl/gl_rasterizer.cpp
     renderer_opengl/gl_rasterizer.h
     renderer_opengl/gl_resource_manager.cpp
@@ -102,8 +103,6 @@ add_library(video_core STATIC
     shader/decode/video.cpp
     shader/decode/xmad.cpp
     shader/decode/other.cpp
-    shader/control_flow.cpp
-    shader/control_flow.h
     shader/decode.cpp
     shader/node_helper.cpp
     shader/node_helper.h

src/video_core/buffer_cache.h

@@ -1,299 +0,0 @@
-// Copyright 2019 yuzu Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#pragma once
-
-#include <array>
-#include <memory>
-#include <mutex>
-#include <unordered_map>
-#include <unordered_set>
-#include <utility>
-#include <vector>
-
-#include "common/alignment.h"
-#include "common/common_types.h"
-#include "core/core.h"
-#include "video_core/memory_manager.h"
-#include "video_core/rasterizer_cache.h"
-
-namespace VideoCore {
-class RasterizerInterface;
-}
-
-namespace VideoCommon {
-
-template <typename BufferStorageType>
-class CachedBuffer final : public RasterizerCacheObject {
-public:
-    explicit CachedBuffer(VAddr cpu_addr, u8* host_ptr)
-        : RasterizerCacheObject{host_ptr}, host_ptr{host_ptr}, cpu_addr{cpu_addr} {}
-    ~CachedBuffer() override = default;
-
-    VAddr GetCpuAddr() const override {
-        return cpu_addr;
-    }
-
-    std::size_t GetSizeInBytes() const override {
-        return size;
-    }
-
-    u8* GetWritableHostPtr() const {
-        return host_ptr;
-    }
-
-    std::size_t GetSize() const {
-        return size;
-    }
-
-    std::size_t GetCapacity() const {
-        return capacity;
-    }
-
-    bool IsInternalized() const {
-        return is_internal;
-    }
-
-    const BufferStorageType& GetBuffer() const {
-        return buffer;
-    }
-
-    void SetSize(std::size_t new_size) {
-        size = new_size;
-    }
-
-    void SetInternalState(bool is_internal_) {
-        is_internal = is_internal_;
-    }
-
-    BufferStorageType ExchangeBuffer(BufferStorageType buffer_, std::size_t new_capacity) {
-        capacity = new_capacity;
-        std::swap(buffer, buffer_);
-        return buffer_;
-    }
-
-private:
-    u8* host_ptr{};
-    VAddr cpu_addr{};
-    std::size_t size{};
-    std::size_t capacity{};
-    bool is_internal{};
-    BufferStorageType buffer;
-};
-
-template <typename BufferStorageType, typename BufferType, typename StreamBuffer>
-class BufferCache : public RasterizerCache<std::shared_ptr<CachedBuffer<BufferStorageType>>> {
-public:
-    using Buffer = std::shared_ptr<CachedBuffer<BufferStorageType>>;
-    using BufferInfo = std::pair<const BufferType*, u64>;
-
-    explicit BufferCache(VideoCore::RasterizerInterface& rasterizer, Core::System& system,
-                         std::unique_ptr<StreamBuffer> stream_buffer)
-        : RasterizerCache<Buffer>{rasterizer}, system{system},
-          stream_buffer{std::move(stream_buffer)}, stream_buffer_handle{
-                                                       this->stream_buffer->GetHandle()} {}
-    ~BufferCache() = default;
-
-    void Unregister(const Buffer& entry) override {
-        std::lock_guard lock{RasterizerCache<Buffer>::mutex};
-        if (entry->IsInternalized()) {
-            internalized_entries.erase(entry->GetCacheAddr());
-        }
-        ReserveBuffer(entry);
-        RasterizerCache<Buffer>::Unregister(entry);
-    }
-
-    void TickFrame() {
-        marked_for_destruction_index =
-            (marked_for_destruction_index + 1) % marked_for_destruction_ring_buffer.size();
-        MarkedForDestruction().clear();
-    }
-
-    BufferInfo UploadMemory(GPUVAddr gpu_addr, std::size_t size, std::size_t alignment = 4,
-                            bool internalize = false, bool is_written = false) {
-        std::lock_guard lock{RasterizerCache<Buffer>::mutex};
-
-        auto& memory_manager = system.GPU().MemoryManager();
-        const auto host_ptr = memory_manager.GetPointer(gpu_addr);
-        if (!host_ptr) {
-            return {GetEmptyBuffer(size), 0};
-        }
-        const auto cache_addr = ToCacheAddr(host_ptr);
-
-        // Cache management is a big overhead, so only cache entries with a given size.
-        // TODO: Figure out which size is the best for given games.
-        constexpr std::size_t max_stream_size = 0x800;
-        if (!internalize && size < max_stream_size &&
-            internalized_entries.find(cache_addr) == internalized_entries.end()) {
-            return StreamBufferUpload(host_ptr, size, alignment);
-        }
-
-        auto entry = RasterizerCache<Buffer>::TryGet(cache_addr);
-        if (!entry) {
-            return FixedBufferUpload(gpu_addr, host_ptr, size, internalize, is_written);
-        }
-
-        if (entry->GetSize() < size) {
-            IncreaseBufferSize(entry, size);
-        }
-        if (is_written) {
-            entry->MarkAsModified(true, *this);
-        }
-        return {ToHandle(entry->GetBuffer()), 0};
-    }
-
-    /// Uploads from a host memory. Returns the OpenGL buffer where it's located and its offset.
-    BufferInfo UploadHostMemory(const void* raw_pointer, std::size_t size,
-                                std::size_t alignment = 4) {
-        std::lock_guard lock{RasterizerCache<Buffer>::mutex};
-        return StreamBufferUpload(raw_pointer, size, alignment);
-    }
-
-    void Map(std::size_t max_size) {
-        std::tie(buffer_ptr, buffer_offset_base, invalidated) = stream_buffer->Map(max_size, 4);
-        buffer_offset = buffer_offset_base;
-    }
-
-    /// Finishes the upload stream, returns true on bindings invalidation.
-    bool Unmap() {
-        stream_buffer->Unmap(buffer_offset - buffer_offset_base);
-        return std::exchange(invalidated, false);
-    }
-
-    virtual const BufferType* GetEmptyBuffer(std::size_t size) = 0;
-
-protected:
-    void FlushObjectInner(const Buffer& entry) override {
-        DownloadBufferData(entry->GetBuffer(), 0, entry->GetSize(), entry->GetWritableHostPtr());
-    }
-
-    virtual BufferStorageType CreateBuffer(std::size_t size) = 0;
-
-    virtual const BufferType* ToHandle(const BufferStorageType& storage) = 0;
-
-    virtual void UploadBufferData(const BufferStorageType& buffer, std::size_t offset,
-                                  std::size_t size, const u8* data) = 0;
-
-    virtual void DownloadBufferData(const BufferStorageType& buffer, std::size_t offset,
-                                    std::size_t size, u8* data) = 0;
-
-    virtual void CopyBufferData(const BufferStorageType& src, const BufferStorageType& dst,
-                                std::size_t src_offset, std::size_t dst_offset,
-                                std::size_t size) = 0;
-
-private:
-    BufferInfo StreamBufferUpload(const void* raw_pointer, std::size_t size,
-                                  std::size_t alignment) {
-        AlignBuffer(alignment);
-        const std::size_t uploaded_offset = buffer_offset;
-        std::memcpy(buffer_ptr, raw_pointer, size);
-
-        buffer_ptr += size;
-        buffer_offset += size;
-        return {&stream_buffer_handle, uploaded_offset};
-    }
-
-    BufferInfo FixedBufferUpload(GPUVAddr gpu_addr, u8* host_ptr, std::size_t size,
-                                 bool internalize, bool is_written) {
-        auto& memory_manager = Core::System::GetInstance().GPU().MemoryManager();
-        const auto cpu_addr = memory_manager.GpuToCpuAddress(gpu_addr);
-        ASSERT(cpu_addr);
-
-        auto entry = GetUncachedBuffer(*cpu_addr, host_ptr);
-        entry->SetSize(size);
-        entry->SetInternalState(internalize);
-        RasterizerCache<Buffer>::Register(entry);
-
-        if (internalize) {
-            internalized_entries.emplace(ToCacheAddr(host_ptr));
-        }
-        if (is_written) {
-            entry->MarkAsModified(true, *this);
-        }
-
-        if (entry->GetCapacity() < size) {
-            MarkedForDestruction().push_back(entry->ExchangeBuffer(CreateBuffer(size), size));
-        }
-
-        UploadBufferData(entry->GetBuffer(), 0, size, host_ptr);
-        return {ToHandle(entry->GetBuffer()), 0};
-    }
-
-    void IncreaseBufferSize(Buffer& entry, std::size_t new_size) {
-        const std::size_t old_size = entry->GetSize();
-        if (entry->GetCapacity() < new_size) {
-            const auto& old_buffer = entry->GetBuffer();
-            auto new_buffer = CreateBuffer(new_size);
-
-            // Copy bits from the old buffer to the new buffer.
-            CopyBufferData(old_buffer, new_buffer, 0, 0, old_size);
-            MarkedForDestruction().push_back(
-                entry->ExchangeBuffer(std::move(new_buffer), new_size));
-
-            // This buffer could have been used
-            invalidated = true;
-        }
-        // Upload the new bits.
-        const std::size_t size_diff = new_size - old_size;
-        UploadBufferData(entry->GetBuffer(), old_size, size_diff, entry->GetHostPtr() + old_size);
-
-        // Update entry's size in the object and in the cache.
-        Unregister(entry);
-
-        entry->SetSize(new_size);
-        RasterizerCache<Buffer>::Register(entry);
-    }
-
-    Buffer GetUncachedBuffer(VAddr cpu_addr, u8* host_ptr) {
-        if (auto entry = TryGetReservedBuffer(host_ptr)) {
-            return entry;
-        }
-        return std::make_shared<CachedBuffer<BufferStorageType>>(cpu_addr, host_ptr);
-    }
-
-    Buffer TryGetReservedBuffer(u8* host_ptr) {
-        const auto it = buffer_reserve.find(ToCacheAddr(host_ptr));
-        if (it == buffer_reserve.end()) {
-            return {};
-        }
-        auto& reserve = it->second;
-        auto entry = reserve.back();
-        reserve.pop_back();
-        return entry;
-    }
-
-    void ReserveBuffer(Buffer entry) {
-        buffer_reserve[entry->GetCacheAddr()].push_back(std::move(entry));
-    }
-
-    void AlignBuffer(std::size_t alignment) {
-        // Align the offset, not the mapped pointer
-        const std::size_t offset_aligned = Common::AlignUp(buffer_offset, alignment);
-        buffer_ptr += offset_aligned - buffer_offset;
-        buffer_offset = offset_aligned;
-    }
-
-    std::vector<BufferStorageType>& MarkedForDestruction() {
-        return marked_for_destruction_ring_buffer[marked_for_destruction_index];
-    }
-
-    Core::System& system;
-
-    std::unique_ptr<StreamBuffer> stream_buffer;
-    BufferType stream_buffer_handle{};
-
-    bool invalidated = false;
-
-    u8* buffer_ptr = nullptr;
-    u64 buffer_offset = 0;
-    u64 buffer_offset_base = 0;
-
-    std::size_t marked_for_destruction_index = 0;
-    std::array<std::vector<BufferStorageType>, 4> marked_for_destruction_ring_buffer;
-
-    std::unordered_set<CacheAddr> internalized_entries;
-    std::unordered_map<CacheAddr, std::vector<Buffer>> buffer_reserve;
-};
-
-} // namespace VideoCommon

src/video_core/engines/maxwell_3d.h

@@ -67,7 +67,6 @@ public:
         static constexpr std::size_t MaxShaderStage = 5;
         // Maximum number of const buffers per shader stage.
         static constexpr std::size_t MaxConstBuffers = 18;
-        static constexpr std::size_t MaxConstBufferSize = 0x10000;
 
         enum class QueryMode : u32 {
             Write = 0,

src/video_core/engines/shader_bytecode.h

@@ -78,7 +78,7 @@ union Attribute {
     constexpr explicit Attribute(u64 value) : value(value) {}
 
     enum class Index : u64 {
-        LayerViewportPointSize = 6,
+        PointSize = 6,
         Position = 7,
         Attribute_0 = 8,
         Attribute_31 = 39,
@@ -931,6 +931,8 @@ union Instruction {
     } csetp;
 
     union {
+        BitField<35, 4, PredCondition> cond;
+        BitField<49, 1, u64> h_and;
         BitField<6, 1, u64> ftz;
         BitField<45, 2, PredOperation> op;
         BitField<3, 3, u64> pred3;
@@ -938,21 +940,9 @@ union Instruction {
         BitField<43, 1, u64> negate_a;
         BitField<44, 1, u64> abs_a;
         BitField<47, 2, HalfType> type_a;
-        union {
-            BitField<35, 4, PredCondition> cond;
-            BitField<49, 1, u64> h_and;
-            BitField<31, 1, u64> negate_b;
-            BitField<30, 1, u64> abs_b;
-            BitField<28, 2, HalfType> type_b;
-        } reg;
-        union {
-            BitField<56, 1, u64> negate_b;
-            BitField<54, 1, u64> abs_b;
-        } cbuf;
-        union {
-            BitField<49, 4, PredCondition> cond;
-            BitField<53, 1, u64> h_and;
-        } cbuf_and_imm;
+        BitField<31, 1, u64> negate_b;
+        BitField<30, 1, u64> abs_b;
+        BitField<28, 2, HalfType> type_b;
         BitField<42, 1, u64> neg_pred;
         BitField<39, 3, u64> pred39;
     } hsetp2;
@@ -1018,6 +1008,8 @@ union Instruction {
         } f2i;
 
         union {
+            BitField<8, 2, Register::Size> src_size;
+            BitField<10, 2, Register::Size> dst_size;
             BitField<39, 4, u64> rounding;
             // H0, H1 extract for F16 missing
             BitField<41, 1, u64> selector; // Guessed as some games set it, TODO: reverse this value
@@ -1286,7 +1278,6 @@ union Instruction {
     union {
         BitField<49, 1, u64> nodep_flag;
         BitField<53, 4, u64> texture_info;
-        BitField<59, 1, u64> fp32_flag;
 
         TextureType GetTextureType() const {
             // The TLDS instruction has a weird encoding for the texture type.
@@ -1376,20 +1367,6 @@ union Instruction {
         }
     } bra;
 
-    union {
-        BitField<20, 24, u64> target;
-        BitField<5, 1, u64> constant_buffer;
-
-        s32 GetBranchExtend() const {
-            // Sign extend the branch target offset
-            u32 mask = 1U << (24 - 1);
-            u32 value = static_cast<u32>(target);
-            // The branch offset is relative to the next instruction and is stored in bytes, so
-            // divide it by the size of an instruction and add 1 to it.
-            return static_cast<s32>((value ^ mask) - mask) / sizeof(Instruction) + 1;
-        }
-    } brx;
-
     union {
         BitField<39, 1, u64> emit; // EmitVertex
         BitField<40, 1, u64> cut;  // EndPrimitive
@@ -1487,7 +1464,6 @@ public:
         BFE_IMM,
         BFI_IMM_R,
         BRA,
-        BRX,
         PBK,
         LD_A,
         LD_L,
@@ -1556,9 +1532,7 @@ public:
         HFMA2_RC,
         HFMA2_RR,
         HFMA2_IMM_R,
-        HSETP2_C,
         HSETP2_R,
-        HSETP2_IMM,
         HSET2_R,
         POPC_C,
         POPC_R,
@@ -1764,7 +1738,6 @@ private:
             INST("111000101001----", Id::SSY, Type::Flow, "SSY"),
             INST("111000101010----", Id::PBK, Type::Flow, "PBK"),
             INST("111000100100----", Id::BRA, Type::Flow, "BRA"),
-            INST("111000100101----", Id::BRX, Type::Flow, "BRX"),
             INST("1111000011111---", Id::SYNC, Type::Flow, "SYNC"),
             INST("111000110100---", Id::BRK, Type::Flow, "BRK"),
             INST("111000110000----", Id::EXIT, Type::Flow, "EXIT"),
@@ -1787,7 +1760,7 @@ private:
             INST("1101111101010---", Id::TXQ_B, Type::Texture, "TXQ_B"),
             INST("1101-00---------", Id::TEXS, Type::Texture, "TEXS"),
             INST("11011100--11----", Id::TLD, Type::Texture, "TLD"),
-            INST("1101-01---------", Id::TLDS, Type::Texture, "TLDS"),
+            INST("1101101---------", Id::TLDS, Type::Texture, "TLDS"),
             INST("110010----111---", Id::TLD4, Type::Texture, "TLD4"),
             INST("1101111100------", Id::TLD4S, Type::Texture, "TLD4S"),
             INST("110111110110----", Id::TMML_B, Type::Texture, "TMML_B"),
@@ -1841,9 +1814,7 @@ private:
             INST("01100---1-------", Id::HFMA2_RC, Type::Hfma2, "HFMA2_RC"),
             INST("0101110100000---", Id::HFMA2_RR, Type::Hfma2, "HFMA2_RR"),
             INST("01110---0-------", Id::HFMA2_IMM_R, Type::Hfma2, "HFMA2_R_IMM"),
-            INST("0111111-1-------", Id::HSETP2_C, Type::HalfSetPredicate, "HSETP2_C"),
-            INST("0101110100100---", Id::HSETP2_R, Type::HalfSetPredicate, "HSETP2_R"),
-            INST("0111111-0-------", Id::HSETP2_IMM, Type::HalfSetPredicate, "HSETP2_IMM"),
+            INST("0101110100100---", Id::HSETP2_R, Type::HalfSetPredicate, "HSETP_R"),
             INST("0101110100011---", Id::HSET2_R, Type::HalfSet, "HSET2_R"),
             INST("0101000010000---", Id::MUFU, Type::Arithmetic, "MUFU"),
             INST("0100110010010---", Id::RRO_C, Type::Arithmetic, "RRO_C"),

src/video_core/gpu.cpp

@@ -31,7 +31,7 @@ u32 FramebufferConfig::BytesPerPixel(PixelFormat format) {
 
 GPU::GPU(Core::System& system, VideoCore::RendererBase& renderer) : renderer{renderer} {
     auto& rasterizer{renderer.Rasterizer()};
-    memory_manager = std::make_unique<Tegra::MemoryManager>(system, rasterizer);
+    memory_manager = std::make_unique<Tegra::MemoryManager>(rasterizer);
     dma_pusher = std::make_unique<Tegra::DmaPusher>(*this);
     maxwell_3d = std::make_unique<Engines::Maxwell3D>(system, rasterizer, *memory_manager);
     fermi_2d = std::make_unique<Engines::Fermi2D>(rasterizer, *memory_manager);

src/video_core/macro_interpreter.cpp

@@ -4,18 +4,14 @@
 
 #include "common/assert.h"
 #include "common/logging/log.h"
-#include "common/microprofile.h"
 #include "video_core/engines/maxwell_3d.h"
 #include "video_core/macro_interpreter.h"
 
-MICROPROFILE_DEFINE(MacroInterp, "GPU", "Execute macro interpreter", MP_RGB(128, 128, 192));
-
 namespace Tegra {
 
 MacroInterpreter::MacroInterpreter(Engines::Maxwell3D& maxwell3d) : maxwell3d(maxwell3d) {}
 
 void MacroInterpreter::Execute(u32 offset, std::vector<u32> parameters) {
-    MICROPROFILE_SCOPE(MacroInterp);
     Reset();
     registers[1] = parameters[0];
     this->parameters = std::move(parameters);

src/video_core/memory_manager.cpp

@@ -5,17 +5,13 @@
 #include "common/alignment.h"
 #include "common/assert.h"
 #include "common/logging/log.h"
-#include "core/core.h"
-#include "core/hle/kernel/process.h"
-#include "core/hle/kernel/vm_manager.h"
 #include "core/memory.h"
 #include "video_core/memory_manager.h"
 #include "video_core/rasterizer_interface.h"
 
 namespace Tegra {
 
-MemoryManager::MemoryManager(Core::System& system, VideoCore::RasterizerInterface& rasterizer)
-    : rasterizer{rasterizer}, system{system} {
+MemoryManager::MemoryManager(VideoCore::RasterizerInterface& rasterizer) : rasterizer{rasterizer} {
     std::fill(page_table.pointers.begin(), page_table.pointers.end(), nullptr);
     std::fill(page_table.attributes.begin(), page_table.attributes.end(),
               Common::PageType::Unmapped);
@@ -53,11 +49,6 @@ GPUVAddr MemoryManager::MapBufferEx(VAddr cpu_addr, u64 size) {
     const GPUVAddr gpu_addr{FindFreeRegion(address_space_base, aligned_size)};
 
     MapBackingMemory(gpu_addr, Memory::GetPointer(cpu_addr), aligned_size, cpu_addr);
-    ASSERT(system.CurrentProcess()
-               ->VMManager()
-               .SetMemoryAttribute(cpu_addr, size, Kernel::MemoryAttribute::DeviceMapped,
-                                   Kernel::MemoryAttribute::DeviceMapped)
-               .IsSuccess());
 
     return gpu_addr;
 }
@@ -68,11 +59,7 @@ GPUVAddr MemoryManager::MapBufferEx(VAddr cpu_addr, GPUVAddr gpu_addr, u64 size)
     const u64 aligned_size{Common::AlignUp(size, page_size)};
 
     MapBackingMemory(gpu_addr, Memory::GetPointer(cpu_addr), aligned_size, cpu_addr);
-    ASSERT(system.CurrentProcess()
-               ->VMManager()
-               .SetMemoryAttribute(cpu_addr, size, Kernel::MemoryAttribute::DeviceMapped,
-                                   Kernel::MemoryAttribute::DeviceMapped)
-               .IsSuccess());
+
     return gpu_addr;
 }
 
@@ -81,16 +68,9 @@ GPUVAddr MemoryManager::UnmapBuffer(GPUVAddr gpu_addr, u64 size) {
 
     const u64 aligned_size{Common::AlignUp(size, page_size)};
     const CacheAddr cache_addr{ToCacheAddr(GetPointer(gpu_addr))};
-    const auto cpu_addr = GpuToCpuAddress(gpu_addr);
-    ASSERT(cpu_addr);
 
     rasterizer.FlushAndInvalidateRegion(cache_addr, aligned_size);
     UnmapRange(gpu_addr, aligned_size);
-    ASSERT(system.CurrentProcess()
-               ->VMManager()
-               .SetMemoryAttribute(cpu_addr.value(), size, Kernel::MemoryAttribute::DeviceMapped,
-                                   Kernel::MemoryAttribute::None)
-               .IsSuccess());
 
     return gpu_addr;
 }

src/video_core/memory_manager.h

@@ -14,10 +14,6 @@ namespace VideoCore {
 class RasterizerInterface;
 }
 
-namespace Core {
-class System;
-}
-
 namespace Tegra {
 
 /**
@@ -51,7 +47,7 @@ struct VirtualMemoryArea {
 
 class MemoryManager final {
 public:
-    explicit MemoryManager(Core::System& system, VideoCore::RasterizerInterface& rasterizer);
+    explicit MemoryManager(VideoCore::RasterizerInterface& rasterizer);
     ~MemoryManager();
 
     GPUVAddr AllocateSpace(u64 size, u64 align);
@@ -177,8 +173,6 @@ private:
     Common::PageTable page_table{page_bits};
     VMAMap vma_map;
     VideoCore::RasterizerInterface& rasterizer;
-
-    Core::System& system;
 };
 
 } // namespace Tegra

src/video_core/rasterizer_interface.h

@@ -47,9 +47,6 @@ public:
     /// and invalidated
     virtual void FlushAndInvalidateRegion(CacheAddr addr, u64 size) = 0;
 
-    /// Notify rasterizer that a frame is about to finish
-    virtual void TickFrame() = 0;
-
     /// Attempt to use a faster method to perform a surface copy
     virtual bool AccelerateSurfaceCopy(const Tegra::Engines::Fermi2D::Regs::Surface& src,
                                        const Tegra::Engines::Fermi2D::Regs::Surface& dst,

src/video_core/renderer_opengl/gl_buffer_cache.cpp

@@ -2,57 +2,103 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include <cstring>
 #include <memory>
 
-#include <glad/glad.h>
-
-#include "common/assert.h"
+#include "common/alignment.h"
+#include "core/core.h"
+#include "video_core/memory_manager.h"
 #include "video_core/renderer_opengl/gl_buffer_cache.h"
 #include "video_core/renderer_opengl/gl_rasterizer.h"
-#include "video_core/renderer_opengl/gl_resource_manager.h"
 
 namespace OpenGL {
 
-OGLBufferCache::OGLBufferCache(RasterizerOpenGL& rasterizer, Core::System& system,
-                               std::size_t stream_size)
-    : VideoCommon::BufferCache<OGLBuffer, GLuint, OGLStreamBuffer>{
-          rasterizer, system, std::make_unique<OGLStreamBuffer>(stream_size, true)} {}
+CachedBufferEntry::CachedBufferEntry(VAddr cpu_addr, std::size_t size, GLintptr offset,
+                                     std::size_t alignment, u8* host_ptr)
+    : RasterizerCacheObject{host_ptr}, cpu_addr{cpu_addr}, size{size}, offset{offset},
+      alignment{alignment} {}
 
-OGLBufferCache::~OGLBufferCache() = default;
+OGLBufferCache::OGLBufferCache(RasterizerOpenGL& rasterizer, std::size_t size)
+    : RasterizerCache{rasterizer}, stream_buffer(size, true) {}
 
-OGLBuffer OGLBufferCache::CreateBuffer(std::size_t size) {
-    OGLBuffer buffer;
-    buffer.Create();
-    glNamedBufferData(buffer.handle, static_cast<GLsizeiptr>(size), nullptr, GL_DYNAMIC_DRAW);
-    return buffer;
+GLintptr OGLBufferCache::UploadMemory(GPUVAddr gpu_addr, std::size_t size, std::size_t alignment,
+                                      bool cache) {
+    std::lock_guard lock{mutex};
+    auto& memory_manager = Core::System::GetInstance().GPU().MemoryManager();
+
+    // Cache management is a big overhead, so only cache entries with a given size.
+    // TODO: Figure out which size is the best for given games.
+    cache &= size >= 2048;
+
+    const auto& host_ptr{memory_manager.GetPointer(gpu_addr)};
+    if (cache) {
+        auto entry = TryGet(host_ptr);
+        if (entry) {
+            if (entry->GetSize() >= size && entry->GetAlignment() == alignment) {
+                return entry->GetOffset();
+            }
+            Unregister(entry);
+        }
+    }
+
+    AlignBuffer(alignment);
+    const GLintptr uploaded_offset = buffer_offset;
+
+    if (!host_ptr) {
+        return uploaded_offset;
+    }
+
+    std::memcpy(buffer_ptr, host_ptr, size);
+    buffer_ptr += size;
+    buffer_offset += size;
+
+    if (cache) {
+        auto entry = std::make_shared<CachedBufferEntry>(
+            *memory_manager.GpuToCpuAddress(gpu_addr), size, uploaded_offset, alignment, host_ptr);
+        Register(entry);
+    }
+
+    return uploaded_offset;
 }
 
-const GLuint* OGLBufferCache::ToHandle(const OGLBuffer& buffer) {
-    return &buffer.handle;
+GLintptr OGLBufferCache::UploadHostMemory(const void* raw_pointer, std::size_t size,
+                                          std::size_t alignment) {
+    std::lock_guard lock{mutex};
+    AlignBuffer(alignment);
+    std::memcpy(buffer_ptr, raw_pointer, size);
+    const GLintptr uploaded_offset = buffer_offset;
+
+    buffer_ptr += size;
+    buffer_offset += size;
+    return uploaded_offset;
 }
 
-const GLuint* OGLBufferCache::GetEmptyBuffer(std::size_t) {
-    static const GLuint null_buffer = 0;
-    return &null_buffer;
+bool OGLBufferCache::Map(std::size_t max_size) {
+    bool invalidate;
+    std::tie(buffer_ptr, buffer_offset_base, invalidate) =
+        stream_buffer.Map(static_cast<GLsizeiptr>(max_size), 4);
+    buffer_offset = buffer_offset_base;
+
+    if (invalidate) {
+        InvalidateAll();
+    }
+    return invalidate;
 }
 
-void OGLBufferCache::UploadBufferData(const OGLBuffer& buffer, std::size_t offset, std::size_t size,
-                                      const u8* data) {
-    glNamedBufferSubData(buffer.handle, static_cast<GLintptr>(offset),
-                         static_cast<GLsizeiptr>(size), data);
+void OGLBufferCache::Unmap() {
+    stream_buffer.Unmap(buffer_offset - buffer_offset_base);
 }
 
-void OGLBufferCache::DownloadBufferData(const OGLBuffer& buffer, std::size_t offset,
-                                        std::size_t size, u8* data) {
-    glGetNamedBufferSubData(buffer.handle, static_cast<GLintptr>(offset),
-                            static_cast<GLsizeiptr>(size), data);
+GLuint OGLBufferCache::GetHandle() const {
+    return stream_buffer.GetHandle();
 }
 
-void OGLBufferCache::CopyBufferData(const OGLBuffer& src, const OGLBuffer& dst,
-                                    std::size_t src_offset, std::size_t dst_offset,
-                                    std::size_t size) {
-    glCopyNamedBufferSubData(src.handle, dst.handle, static_cast<GLintptr>(src_offset),
-                             static_cast<GLintptr>(dst_offset), static_cast<GLsizeiptr>(size));
+void OGLBufferCache::AlignBuffer(std::size_t alignment) {
+    // Align the offset, not the mapped pointer
+    const GLintptr offset_aligned =
+        static_cast<GLintptr>(Common::AlignUp(static_cast<std::size_t>(buffer_offset), alignment));
+    buffer_ptr += offset_aligned - buffer_offset;
+    buffer_offset = offset_aligned;
 }
 
 } // namespace OpenGL

src/video_core/renderer_opengl/gl_buffer_cache.h

@@ -4,44 +4,80 @@
 
 #pragma once
 
+#include <cstddef>
 #include <memory>
+#include <tuple>
 
 #include "common/common_types.h"
-#include "video_core/buffer_cache.h"
 #include "video_core/rasterizer_cache.h"
 #include "video_core/renderer_opengl/gl_resource_manager.h"
 #include "video_core/renderer_opengl/gl_stream_buffer.h"
 
-namespace Core {
-class System;
-}
-
 namespace OpenGL {
 
-class OGLStreamBuffer;
 class RasterizerOpenGL;
 
-class OGLBufferCache final : public VideoCommon::BufferCache<OGLBuffer, GLuint, OGLStreamBuffer> {
+class CachedBufferEntry final : public RasterizerCacheObject {
 public:
-    explicit OGLBufferCache(RasterizerOpenGL& rasterizer, Core::System& system,
-                            std::size_t stream_size);
-    ~OGLBufferCache();
+    explicit CachedBufferEntry(VAddr cpu_addr, std::size_t size, GLintptr offset,
+                               std::size_t alignment, u8* host_ptr);
 
-    const GLuint* GetEmptyBuffer(std::size_t) override;
+    VAddr GetCpuAddr() const override {
+        return cpu_addr;
+    }
+
+    std::size_t GetSizeInBytes() const override {
+        return size;
+    }
+
+    std::size_t GetSize() const {
+        return size;
+    }
+
+    GLintptr GetOffset() const {
+        return offset;
+    }
+
+    std::size_t GetAlignment() const {
+        return alignment;
+    }
+
+private:
+    VAddr cpu_addr{};
+    std::size_t size{};
+    GLintptr offset{};
+    std::size_t alignment{};
+};
+
+class OGLBufferCache final : public RasterizerCache<std::shared_ptr<CachedBufferEntry>> {
+public:
+    explicit OGLBufferCache(RasterizerOpenGL& rasterizer, std::size_t size);
+
+    /// Uploads data from a guest GPU address. Returns host's buffer offset where it's been
+    /// allocated.
+    GLintptr UploadMemory(GPUVAddr gpu_addr, std::size_t size, std::size_t alignment = 4,
+                          bool cache = true);
+
+    /// Uploads from a host memory. Returns host's buffer offset where it's been allocated.
+    GLintptr UploadHostMemory(const void* raw_pointer, std::size_t size, std::size_t alignment = 4);
+
+    bool Map(std::size_t max_size);
+    void Unmap();
+
+    GLuint GetHandle() const;
 
 protected:
-    OGLBuffer CreateBuffer(std::size_t size) override;
+    void AlignBuffer(std::size_t alignment);
 
-    const GLuint* ToHandle(const OGLBuffer& buffer) override;
+    // We do not have to flush this cache as things in it are never modified by us.
+    void FlushObjectInner(const std::shared_ptr<CachedBufferEntry>& object) override {}
 
-    void UploadBufferData(const OGLBuffer& buffer, std::size_t offset, std::size_t size,
-                          const u8* data) override;
+private:
+    OGLStreamBuffer stream_buffer;
 
-    void DownloadBufferData(const OGLBuffer& buffer, std::size_t offset, std::size_t size,
-                            u8* data) override;
-
-    void CopyBufferData(const OGLBuffer& src, const OGLBuffer& dst, std::size_t src_offset,
-                        std::size_t dst_offset, std::size_t size) override;
+    u8* buffer_ptr = nullptr;
+    GLintptr buffer_offset = 0;
+    GLintptr buffer_offset_base = 0;
 };
 
 } // namespace OpenGL

src/video_core/renderer_opengl/gl_device.cpp

@@ -24,10 +24,8 @@ T GetInteger(GLenum pname) {
 
 Device::Device() {
     uniform_buffer_alignment = GetInteger<std::size_t>(GL_UNIFORM_BUFFER_OFFSET_ALIGNMENT);
-    shader_storage_alignment = GetInteger<std::size_t>(GL_SHADER_STORAGE_BUFFER_OFFSET_ALIGNMENT);
     max_vertex_attributes = GetInteger<u32>(GL_MAX_VERTEX_ATTRIBS);
     max_varyings = GetInteger<u32>(GL_MAX_VARYING_VECTORS);
-    has_vertex_viewport_layer = GLAD_GL_ARB_shader_viewport_layer_array;
     has_variable_aoffi = TestVariableAoffi();
     has_component_indexing_bug = TestComponentIndexingBug();
 }
@@ -36,7 +34,6 @@ Device::Device(std::nullptr_t) {
     uniform_buffer_alignment = 0;
     max_vertex_attributes = 16;
     max_varyings = 15;
-    has_vertex_viewport_layer = true;
     has_variable_aoffi = true;
     has_component_indexing_bug = false;
 }

src/video_core/renderer_opengl/gl_device.h

@@ -18,10 +18,6 @@ public:
         return uniform_buffer_alignment;
     }
 
-    std::size_t GetShaderStorageBufferAlignment() const {
-        return shader_storage_alignment;
-    }
-
     u32 GetMaxVertexAttributes() const {
         return max_vertex_attributes;
     }
@@ -30,10 +26,6 @@ public:
         return max_varyings;
     }
 
-    bool HasVertexViewportLayer() const {
-        return has_vertex_viewport_layer;
-    }
-
     bool HasVariableAoffi() const {
         return has_variable_aoffi;
     }
@@ -47,10 +39,8 @@ private:
     static bool TestComponentIndexingBug();
 
     std::size_t uniform_buffer_alignment{};
-    std::size_t shader_storage_alignment{};
     u32 max_vertex_attributes{};
     u32 max_varyings{};
-    bool has_vertex_viewport_layer{};
     bool has_variable_aoffi{};
     bool has_component_indexing_bug{};
 };

src/video_core/renderer_opengl/gl_global_cache.cpp

@@ -0,0 +1,102 @@
+// Copyright 2018 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <glad/glad.h>
+
+#include "common/logging/log.h"
+#include "core/core.h"
+#include "video_core/memory_manager.h"
+#include "video_core/renderer_opengl/gl_global_cache.h"
+#include "video_core/renderer_opengl/gl_rasterizer.h"
+#include "video_core/renderer_opengl/gl_shader_decompiler.h"
+#include "video_core/renderer_opengl/utils.h"
+
+namespace OpenGL {
+
+CachedGlobalRegion::CachedGlobalRegion(VAddr cpu_addr, u8* host_ptr, u32 size, u32 max_size)
+    : RasterizerCacheObject{host_ptr}, cpu_addr{cpu_addr}, host_ptr{host_ptr}, size{size},
+      max_size{max_size} {
+    buffer.Create();
+    LabelGLObject(GL_BUFFER, buffer.handle, cpu_addr, "GlobalMemory");
+}
+
+CachedGlobalRegion::~CachedGlobalRegion() = default;
+
+void CachedGlobalRegion::Reload(u32 size_) {
+    size = size_;
+    if (size > max_size) {
+        size = max_size;
+        LOG_CRITICAL(HW_GPU, "Global region size {} exceeded the supported size {}!", size_,
+                     max_size);
+    }
+    glNamedBufferData(buffer.handle, size, host_ptr, GL_STREAM_DRAW);
+}
+
+void CachedGlobalRegion::Flush() {
+    LOG_DEBUG(Render_OpenGL, "Flushing {} bytes to CPU memory address 0x{:16}", size, cpu_addr);
+    glGetNamedBufferSubData(buffer.handle, 0, static_cast<GLsizeiptr>(size), host_ptr);
+}
+
+GlobalRegion GlobalRegionCacheOpenGL::TryGetReservedGlobalRegion(CacheAddr addr, u32 size) const {
+    const auto search{reserve.find(addr)};
+    if (search == reserve.end()) {
+        return {};
+    }
+    return search->second;
+}
+
+GlobalRegion GlobalRegionCacheOpenGL::GetUncachedGlobalRegion(GPUVAddr addr, u8* host_ptr,
+                                                              u32 size) {
+    GlobalRegion region{TryGetReservedGlobalRegion(ToCacheAddr(host_ptr), size)};
+    if (!region) {
+        // No reserved surface available, create a new one and reserve it
+        auto& memory_manager{Core::System::GetInstance().GPU().MemoryManager()};
+        const auto cpu_addr{memory_manager.GpuToCpuAddress(addr)};
+        ASSERT(cpu_addr);
+
+        region = std::make_shared<CachedGlobalRegion>(*cpu_addr, host_ptr, size, max_ssbo_size);
+        ReserveGlobalRegion(region);
+    }
+    region->Reload(size);
+    return region;
+}
+
+void GlobalRegionCacheOpenGL::ReserveGlobalRegion(GlobalRegion region) {
+    reserve.insert_or_assign(region->GetCacheAddr(), std::move(region));
+}
+
+GlobalRegionCacheOpenGL::GlobalRegionCacheOpenGL(RasterizerOpenGL& rasterizer)
+    : RasterizerCache{rasterizer} {
+    GLint max_ssbo_size_;
+    glGetIntegerv(GL_MAX_SHADER_STORAGE_BLOCK_SIZE, &max_ssbo_size_);
+    max_ssbo_size = static_cast<u32>(max_ssbo_size_);
+}
+
+GlobalRegion GlobalRegionCacheOpenGL::GetGlobalRegion(
+    const GLShader::GlobalMemoryEntry& global_region,
+    Tegra::Engines::Maxwell3D::Regs::ShaderStage stage) {
+    std::lock_guard lock{mutex};
+
+    auto& gpu{Core::System::GetInstance().GPU()};
+    auto& memory_manager{gpu.MemoryManager()};
+    const auto cbufs{gpu.Maxwell3D().state.shader_stages[static_cast<std::size_t>(stage)]};
+    const auto addr{cbufs.const_buffers[global_region.GetCbufIndex()].address +
+                    global_region.GetCbufOffset()};
+    const auto actual_addr{memory_manager.Read<u64>(addr)};
+    const auto size{memory_manager.Read<u32>(addr + 8)};
+
+    // Look up global region in the cache based on address
+    const auto& host_ptr{memory_manager.GetPointer(actual_addr)};
+    GlobalRegion region{TryGet(host_ptr)};
+
+    if (!region) {
+        // No global region found - create a new one
+        region = GetUncachedGlobalRegion(actual_addr, host_ptr, size);
+        Register(region);
+    }
+
+    return region;
+}
+
+} // namespace OpenGL

src/video_core/renderer_opengl/gl_global_cache.h

@@ -0,0 +1,82 @@
+// Copyright 2018 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <memory>
+#include <unordered_map>
+
+#include <glad/glad.h>
+
+#include "common/assert.h"
+#include "common/common_types.h"
+#include "video_core/engines/maxwell_3d.h"
+#include "video_core/rasterizer_cache.h"
+#include "video_core/renderer_opengl/gl_resource_manager.h"
+
+namespace OpenGL {
+
+namespace GLShader {
+class GlobalMemoryEntry;
+}
+
+class RasterizerOpenGL;
+class CachedGlobalRegion;
+using GlobalRegion = std::shared_ptr<CachedGlobalRegion>;
+
+class CachedGlobalRegion final : public RasterizerCacheObject {
+public:
+    explicit CachedGlobalRegion(VAddr cpu_addr, u8* host_ptr, u32 size, u32 max_size);
+    ~CachedGlobalRegion();
+
+    VAddr GetCpuAddr() const override {
+        return cpu_addr;
+    }
+
+    std::size_t GetSizeInBytes() const override {
+        return size;
+    }
+
+    /// Gets the GL program handle for the buffer
+    GLuint GetBufferHandle() const {
+        return buffer.handle;
+    }
+
+    /// Reloads the global region from guest memory
+    void Reload(u32 size_);
+
+    void Flush();
+
+private:
+    VAddr cpu_addr{};
+    u8* host_ptr{};
+    u32 size{};
+    u32 max_size{};
+
+    OGLBuffer buffer;
+};
+
+class GlobalRegionCacheOpenGL final : public RasterizerCache<GlobalRegion> {
+public:
+    explicit GlobalRegionCacheOpenGL(RasterizerOpenGL& rasterizer);
+
+    /// Gets the current specified shader stage program
+    GlobalRegion GetGlobalRegion(const GLShader::GlobalMemoryEntry& descriptor,
+                                 Tegra::Engines::Maxwell3D::Regs::ShaderStage stage);
+
+protected:
+    void FlushObjectInner(const GlobalRegion& object) override {
+        object->Flush();
+    }
+
+private:
+    GlobalRegion TryGetReservedGlobalRegion(CacheAddr addr, u32 size) const;
+    GlobalRegion GetUncachedGlobalRegion(GPUVAddr addr, u8* host_ptr, u32 size);
+    void ReserveGlobalRegion(GlobalRegion region);
+
+    std::unordered_map<CacheAddr, GlobalRegion> reserve;
+    u32 max_ssbo_size{};
+};
+
+} // namespace OpenGL

src/video_core/renderer_opengl/gl_rasterizer.cpp

@@ -20,7 +20,6 @@
 #include "core/hle/kernel/process.h"
 #include "core/settings.h"
 #include "video_core/engines/maxwell_3d.h"
-#include "video_core/memory_manager.h"
 #include "video_core/renderer_opengl/gl_rasterizer.h"
 #include "video_core/renderer_opengl/gl_shader_cache.h"
 #include "video_core/renderer_opengl/gl_shader_gen.h"
@@ -81,25 +80,11 @@ struct DrawParameters {
     }
 };
 
-static std::size_t GetConstBufferSize(const Tegra::Engines::ConstBufferInfo& buffer,
-                                      const GLShader::ConstBufferEntry& entry) {
-    if (!entry.IsIndirect()) {
-        return entry.GetSize();
-    }
-
-    if (buffer.size > Maxwell::MaxConstBufferSize) {
-        LOG_WARNING(Render_OpenGL, "Indirect constbuffer size {} exceeds maximum {}", buffer.size,
-                    Maxwell::MaxConstBufferSize);
-        return Maxwell::MaxConstBufferSize;
-    }
-
-    return buffer.size;
-}
-
 RasterizerOpenGL::RasterizerOpenGL(Core::System& system, Core::Frontend::EmuWindow& emu_window,
                                    ScreenInfo& info)
     : texture_cache{system, *this, device}, shader_cache{*this, system, emu_window, device},
-      system{system}, screen_info{info}, buffer_cache{*this, system, STREAM_BUFFER_SIZE} {
+      global_cache{*this}, system{system}, screen_info{info},
+      buffer_cache(*this, STREAM_BUFFER_SIZE) {
     OpenGLState::ApplyDefaultState();
 
     shader_program_manager = std::make_unique<GLShader::ProgramManager>();
@@ -144,6 +129,8 @@ GLuint RasterizerOpenGL::SetupVertexFormat() {
         state.draw.vertex_array = vao;
         state.ApplyVertexArrayState();
 
+        glVertexArrayElementBuffer(vao, buffer_cache.GetHandle());
+
         // Use the vertex array as-is, assumes that the data is formatted correctly for OpenGL.
         // Enables the first 16 vertex attributes always, as we don't know which ones are actually
         // used until shader time. Note, Tegra technically supports 32, but we're capping this to 16
@@ -210,11 +197,11 @@ void RasterizerOpenGL::SetupVertexBuffer(GLuint vao) {
 
         ASSERT(end > start);
         const u64 size = end - start + 1;
-        const auto [vertex_buffer, vertex_buffer_offset] = buffer_cache.UploadMemory(start, size);
+        const GLintptr vertex_buffer_offset = buffer_cache.UploadMemory(start, size);
 
         // Bind the vertex array to the buffer at the current offset.
-        vertex_array_pushbuffer.SetVertexBuffer(index, vertex_buffer, vertex_buffer_offset,
-                                                vertex_array.stride);
+        glVertexArrayVertexBuffer(vao, index, buffer_cache.GetHandle(), vertex_buffer_offset,
+                                  vertex_array.stride);
 
         if (regs.instanced_arrays.IsInstancingEnabled(index) && vertex_array.divisor != 0) {
             // Enable vertex buffer instancing with the specified divisor.
@@ -228,19 +215,7 @@ void RasterizerOpenGL::SetupVertexBuffer(GLuint vao) {
     gpu.dirty_flags.vertex_array.reset();
 }
 
-GLintptr RasterizerOpenGL::SetupIndexBuffer() {
-    if (accelerate_draw != AccelDraw::Indexed) {
-        return 0;
-    }
-    MICROPROFILE_SCOPE(OpenGL_Index);
-    const auto& regs = system.GPU().Maxwell3D().regs;
-    const std::size_t size = CalculateIndexBufferSize();
-    const auto [buffer, offset] = buffer_cache.UploadMemory(regs.index_array.IndexStart(), size);
-    vertex_array_pushbuffer.SetIndexBuffer(buffer);
-    return offset;
-}
-
-DrawParameters RasterizerOpenGL::SetupDraw(GLintptr index_buffer_offset) {
+DrawParameters RasterizerOpenGL::SetupDraw() {
     const auto& gpu = system.GPU().Maxwell3D();
     const auto& regs = gpu.regs;
     const bool is_indexed = accelerate_draw == AccelDraw::Indexed;
@@ -252,9 +227,11 @@ DrawParameters RasterizerOpenGL::SetupDraw(GLintptr index_buffer_offset) {
     params.primitive_mode = MaxwellToGL::PrimitiveTopology(regs.draw.topology);
 
     if (is_indexed) {
+        MICROPROFILE_SCOPE(OpenGL_Index);
         params.index_format = MaxwellToGL::IndexFormat(regs.index_array.format);
         params.count = regs.index_array.count;
-        params.index_buffer_offset = index_buffer_offset;
+        params.index_buffer_offset =
+            buffer_cache.UploadMemory(regs.index_array.IndexStart(), CalculateIndexBufferSize());
         params.base_vertex = static_cast<GLint>(regs.vb_element_base);
     } else {
         params.count = regs.vertex_buffer.count;
@@ -270,6 +247,10 @@ void RasterizerOpenGL::SetupShaders(GLenum primitive_mode) {
     BaseBindings base_bindings;
     std::array<bool, Maxwell::NumClipDistances> clip_distances{};
 
+    // Prepare packed bindings
+    bind_ubo_pushbuffer.Setup(base_bindings.cbuf);
+    bind_ssbo_pushbuffer.Setup(base_bindings.gmem);
+
     for (std::size_t index = 0; index < Maxwell::MaxShaderProgram; ++index) {
         const auto& shader_config = gpu.regs.shader_config[index];
         const Maxwell::ShaderProgram program{static_cast<Maxwell::ShaderProgram>(index)};
@@ -290,11 +271,12 @@ void RasterizerOpenGL::SetupShaders(GLenum primitive_mode) {
 
         GLShader::MaxwellUniformData ubo{};
         ubo.SetFromRegs(gpu, stage);
-        const auto [buffer, offset] =
+        const GLintptr offset =
             buffer_cache.UploadHostMemory(&ubo, sizeof(ubo), device.GetUniformBufferAlignment());
 
         // Bind the emulation info buffer
-        bind_ubo_pushbuffer.Push(buffer, offset, static_cast<GLsizeiptr>(sizeof(ubo)));
+        bind_ubo_pushbuffer.Push(buffer_cache.GetHandle(), offset,
+                                 static_cast<GLsizeiptr>(sizeof(ubo)));
 
         Shader shader{shader_cache.GetStageProgram(program)};
 
@@ -339,6 +321,9 @@ void RasterizerOpenGL::SetupShaders(GLenum primitive_mode) {
         base_bindings = next_bindings;
     }
 
+    bind_ubo_pushbuffer.Bind();
+    bind_ssbo_pushbuffer.Bind();
+
     SyncClipEnabled(clip_distances);
 
     gpu.dirty_flags.shaders = false;
@@ -649,46 +634,26 @@ void RasterizerOpenGL::DrawArrays() {
                       Maxwell::MaxShaderStage;
 
     // Add space for at least 18 constant buffers
-    buffer_size += Maxwell::MaxConstBuffers *
-                   (Maxwell::MaxConstBufferSize + device.GetUniformBufferAlignment());
+    buffer_size +=
+        Maxwell::MaxConstBuffers * (MaxConstbufferSize + device.GetUniformBufferAlignment());
 
-    // Prepare the vertex array.
-    buffer_cache.Map(buffer_size);
+    const bool invalidate = buffer_cache.Map(buffer_size);
+    if (invalidate) {
+        // As all cached buffers are invalidated, we need to recheck their state.
+        gpu.dirty_flags.vertex_array.set();
+    }
 
-    // Prepare vertex array format.
     const GLuint vao = SetupVertexFormat();
-    vertex_array_pushbuffer.Setup(vao);
-
-    // Upload vertex and index data.
     SetupVertexBuffer(vao);
-    const GLintptr index_buffer_offset = SetupIndexBuffer();
 
-    // Setup draw parameters. It will automatically choose what glDraw* method to use.
-    const DrawParameters params = SetupDraw(index_buffer_offset);
-
-    // Prepare packed bindings.
-    bind_ubo_pushbuffer.Setup(0);
-    bind_ssbo_pushbuffer.Setup(0);
-
-    // Setup shaders and their used resources.
+    DrawParameters params = SetupDraw();
     texture_cache.GuardSamplers(true);
     SetupShaders(params.primitive_mode);
     texture_cache.GuardSamplers(false);
 
     ConfigureFramebuffers(state);
 
-    // Signal the buffer cache that we are not going to upload more things.
-    const bool invalidate = buffer_cache.Unmap();
-
-    // Now that we are no longer uploading data, we can safely bind the buffers to OpenGL.
-    vertex_array_pushbuffer.Bind();
-    bind_ubo_pushbuffer.Bind();
-    bind_ssbo_pushbuffer.Bind();
-
-    if (invalidate) {
-        // As all cached buffers are invalidated, we need to recheck their state.
-        gpu.dirty_flags.vertex_array.set();
-    }
+    buffer_cache.Unmap();
 
     shader_program_manager->ApplyTo(state);
     state.Apply();
@@ -710,7 +675,7 @@ void RasterizerOpenGL::FlushRegion(CacheAddr addr, u64 size) {
         return;
     }
     texture_cache.FlushRegion(addr, size);
-    buffer_cache.FlushRegion(addr, size);
+    global_cache.FlushRegion(addr, size);
 }
 
 void RasterizerOpenGL::InvalidateRegion(CacheAddr addr, u64 size) {
@@ -720,6 +685,7 @@ void RasterizerOpenGL::InvalidateRegion(CacheAddr addr, u64 size) {
     }
     texture_cache.InvalidateRegion(addr, size);
     shader_cache.InvalidateRegion(addr, size);
+    global_cache.InvalidateRegion(addr, size);
     buffer_cache.InvalidateRegion(addr, size);
 }
 
@@ -730,10 +696,6 @@ void RasterizerOpenGL::FlushAndInvalidateRegion(CacheAddr addr, u64 size) {
     InvalidateRegion(addr, size);
 }
 
-void RasterizerOpenGL::TickFrame() {
-    buffer_cache.TickFrame();
-}
-
 bool RasterizerOpenGL::AccelerateSurfaceCopy(const Tegra::Engines::Fermi2D::Regs::Surface& src,
                                              const Tegra::Engines::Fermi2D::Regs::Surface& dst,
                                              const Tegra::Engines::Fermi2D::Config& copy_config) {
@@ -777,9 +739,11 @@ void RasterizerOpenGL::SetupDrawConstBuffers(Tegra::Engines::Maxwell3D::Regs::Sh
     MICROPROFILE_SCOPE(OpenGL_UBO);
     const auto stage_index = static_cast<std::size_t>(stage);
     const auto& shader_stage = system.GPU().Maxwell3D().state.shader_stages[stage_index];
+    const auto& entries = shader->GetShaderEntries().const_buffers;
 
     // Upload only the enabled buffers from the 16 constbuffers of each shader stage
-    for (const auto& entry : shader->GetShaderEntries().const_buffers) {
+    for (u32 bindpoint = 0; bindpoint < entries.size(); ++bindpoint) {
+        const auto& entry = entries[bindpoint];
         SetupConstBuffer(shader_stage.const_buffers[entry.GetIndex()], entry);
     }
 }
@@ -788,34 +752,46 @@ void RasterizerOpenGL::SetupConstBuffer(const Tegra::Engines::ConstBufferInfo& b
                                         const GLShader::ConstBufferEntry& entry) {
     if (!buffer.enabled) {
         // Set values to zero to unbind buffers
-        bind_ubo_pushbuffer.Push(buffer_cache.GetEmptyBuffer(sizeof(float)), 0, sizeof(float));
+        bind_ubo_pushbuffer.Push(0, 0, 0);
         return;
     }
 
+    std::size_t size;
+    if (entry.IsIndirect()) {
+        // Buffer is accessed indirectly, so upload the entire thing
+        size = buffer.size;
+
+        if (size > MaxConstbufferSize) {
+            LOG_WARNING(Render_OpenGL, "Indirect constbuffer size {} exceeds maximum {}", size,
+                        MaxConstbufferSize);
+            size = MaxConstbufferSize;
+        }
+    } else {
+        // Buffer is accessed directly, upload just what we use
+        size = entry.GetSize();
+    }
+
     // Align the actual size so it ends up being a multiple of vec4 to meet the OpenGL std140
     // UBO alignment requirements.
-    const std::size_t size = Common::AlignUp(GetConstBufferSize(buffer, entry), sizeof(GLvec4));
+    size = Common::AlignUp(size, sizeof(GLvec4));
+    ASSERT_MSG(size <= MaxConstbufferSize, "Constant buffer is too big");
 
-    const auto alignment = device.GetUniformBufferAlignment();
-    const auto [cbuf, offset] = buffer_cache.UploadMemory(buffer.address, size, alignment);
-    bind_ubo_pushbuffer.Push(cbuf, offset, size);
+    const std::size_t alignment = device.GetUniformBufferAlignment();
+    const GLintptr offset = buffer_cache.UploadMemory(buffer.address, size, alignment);
+    bind_ubo_pushbuffer.Push(buffer_cache.GetHandle(), offset, size);
 }
 
 void RasterizerOpenGL::SetupGlobalRegions(Tegra::Engines::Maxwell3D::Regs::ShaderStage stage,
                                           const Shader& shader) {
-    auto& gpu{system.GPU()};
-    auto& memory_manager{gpu.MemoryManager()};
-    const auto cbufs{gpu.Maxwell3D().state.shader_stages[static_cast<std::size_t>(stage)]};
-    const auto alignment{device.GetShaderStorageBufferAlignment()};
-
-    for (const auto& entry : shader->GetShaderEntries().global_memory_entries) {
-        const auto addr{cbufs.const_buffers[entry.GetCbufIndex()].address + entry.GetCbufOffset()};
-        const auto actual_addr{memory_manager.Read<u64>(addr)};
-        const auto size{memory_manager.Read<u32>(addr + 8)};
-
-        const auto [ssbo, buffer_offset] =
-            buffer_cache.UploadMemory(actual_addr, size, alignment, true, entry.IsWritten());
-        bind_ssbo_pushbuffer.Push(ssbo, buffer_offset, static_cast<GLsizeiptr>(size));
+    const auto& entries = shader->GetShaderEntries().global_memory_entries;
+    for (std::size_t bindpoint = 0; bindpoint < entries.size(); ++bindpoint) {
+        const auto& entry{entries[bindpoint]};
+        const auto& region{global_cache.GetGlobalRegion(entry, stage)};
+        if (entry.IsWritten()) {
+            region->MarkAsModified(true, global_cache);
+        }
+        bind_ssbo_pushbuffer.Push(region->GetBufferHandle(), 0,
+                                  static_cast<GLsizeiptr>(region->GetSizeInBytes()));
     }
 }
 

src/video_core/renderer_opengl/gl_rasterizer.h

@@ -24,6 +24,7 @@
 #include "video_core/renderer_opengl/gl_buffer_cache.h"
 #include "video_core/renderer_opengl/gl_device.h"
 #include "video_core/renderer_opengl/gl_framebuffer_cache.h"
+#include "video_core/renderer_opengl/gl_global_cache.h"
 #include "video_core/renderer_opengl/gl_resource_manager.h"
 #include "video_core/renderer_opengl/gl_sampler_cache.h"
 #include "video_core/renderer_opengl/gl_shader_cache.h"
@@ -62,7 +63,6 @@ public:
     void FlushRegion(CacheAddr addr, u64 size) override;
     void InvalidateRegion(CacheAddr addr, u64 size) override;
     void FlushAndInvalidateRegion(CacheAddr addr, u64 size) override;
-    void TickFrame() override;
     bool AccelerateSurfaceCopy(const Tegra::Engines::Fermi2D::Regs::Surface& src,
                                const Tegra::Engines::Fermi2D::Regs::Surface& dst,
                                const Tegra::Engines::Fermi2D::Config& copy_config) override;
@@ -73,6 +73,11 @@ public:
     void LoadDiskResources(const std::atomic_bool& stop_loading,
                            const VideoCore::DiskResourceLoadCallback& callback) override;
 
+    /// Maximum supported size that a constbuffer can have in bytes.
+    static constexpr std::size_t MaxConstbufferSize = 0x10000;
+    static_assert(MaxConstbufferSize % sizeof(GLvec4) == 0,
+                  "The maximum size of a constbuffer must be a multiple of the size of GLvec4");
+
 private:
     struct FramebufferConfigState {
         bool using_color_fb{};
@@ -93,19 +98,17 @@ private:
 
     /**
      * Configures the color and depth framebuffer states.
-     *
-     * @param current_state       The current OpenGL state.
-     * @param using_color_fb      If true, configure color framebuffers.
-     * @param using_depth_fb      If true, configure the depth/stencil framebuffer.
-     * @param preserve_contents   If true, tries to preserve data from a previously used
-     *                            framebuffer.
+     * @param must_reconfigure If true, tells the framebuffer to skip the cache and reconfigure
+     * again. Used by the texture cache to solve texception conflicts
+     * @param use_color_fb If true, configure color framebuffers.
+     * @param using_depth_fb If true, configure the depth/stencil framebuffer.
+     * @param preserve_contents If true, tries to preserve data from a previously used framebuffer.
      * @param single_color_target Specifies if a single color buffer target should be used.
-     *
      * @returns If depth (first) or stencil (second) are being stored in the bound zeta texture
-     *          (requires using_depth_fb to be true)
+     * (requires using_depth_fb to be true)
      */
     std::pair<bool, bool> ConfigureFramebuffers(
-        OpenGLState& current_state, bool using_color_fb = true, bool using_depth_fb = true,
+        OpenGLState& current_state, bool use_color_fb = true, bool using_depth_fb = true,
         bool preserve_contents = true, std::optional<std::size_t> single_color_target = {});
 
     /// Configures the current constbuffers to use for the draw command.
@@ -186,6 +189,7 @@ private:
 
     TextureCacheOpenGL texture_cache;
     ShaderCacheOpenGL shader_cache;
+    GlobalRegionCacheOpenGL global_cache;
     SamplerCacheOpenGL sampler_cache;
     FramebufferCacheOpenGL framebuffer_cache;
 
@@ -204,7 +208,6 @@ private:
     static constexpr std::size_t STREAM_BUFFER_SIZE = 128 * 1024 * 1024;
     OGLBufferCache buffer_cache;
 
-    VertexArrayPushBuffer vertex_array_pushbuffer;
     BindBuffersRangePushBuffer bind_ubo_pushbuffer{GL_UNIFORM_BUFFER};
     BindBuffersRangePushBuffer bind_ssbo_pushbuffer{GL_SHADER_STORAGE_BUFFER};
 
@@ -217,9 +220,7 @@ private:
 
     void SetupVertexBuffer(GLuint vao);
 
-    GLintptr SetupIndexBuffer();
-
-    DrawParameters SetupDraw(GLintptr index_buffer_offset);
+    DrawParameters SetupDraw();
 
     void SetupShaders(GLenum primitive_mode);
 

src/video_core/renderer_opengl/gl_sampler_cache.h

@@ -17,9 +17,9 @@ public:
     ~SamplerCacheOpenGL();
 
 protected:
-    OGLSampler CreateSampler(const Tegra::Texture::TSCEntry& tsc) const override;
+    OGLSampler CreateSampler(const Tegra::Texture::TSCEntry& tsc) const;
 
-    GLuint ToSamplerType(const OGLSampler& sampler) const override;
+    GLuint ToSamplerType(const OGLSampler& sampler) const;
 };
 
 } // namespace OpenGL

src/video_core/renderer_opengl/gl_shader_cache.cpp

@@ -129,11 +129,9 @@ std::size_t CalculateProgramSize(const GLShader::ProgramCode& program) {
 
 /// Hashes one (or two) program streams
 u64 GetUniqueIdentifier(Maxwell::ShaderProgram program_type, const ProgramCode& code,
-                        const ProgramCode& code_b, std::size_t size_a = 0, std::size_t size_b = 0) {
-    if (size_a == 0) {
-        size_a = CalculateProgramSize(code);
-    }
-    u64 unique_identifier = Common::CityHash64(reinterpret_cast<const char*>(code.data()), size_a);
+                        const ProgramCode& code_b) {
+    u64 unique_identifier =
+        Common::CityHash64(reinterpret_cast<const char*>(code.data()), CalculateProgramSize(code));
     if (program_type != Maxwell::ShaderProgram::VertexA) {
         return unique_identifier;
     }
@@ -142,11 +140,8 @@ u64 GetUniqueIdentifier(Maxwell::ShaderProgram program_type, const ProgramCode&
     std::size_t seed = 0;
     boost::hash_combine(seed, unique_identifier);
 
-    if (size_b == 0) {
-        size_b = CalculateProgramSize(code_b);
-    }
-    const u64 identifier_b =
-        Common::CityHash64(reinterpret_cast<const char*>(code_b.data()), size_b);
+    const u64 identifier_b = Common::CityHash64(reinterpret_cast<const char*>(code_b.data()),
+                                                CalculateProgramSize(code_b));
     boost::hash_combine(seed, identifier_b);
     return static_cast<u64>(seed);
 }
@@ -155,17 +150,14 @@ u64 GetUniqueIdentifier(Maxwell::ShaderProgram program_type, const ProgramCode&
 GLShader::ProgramResult CreateProgram(const Device& device, Maxwell::ShaderProgram program_type,
                                       ProgramCode program_code, ProgramCode program_code_b) {
     GLShader::ShaderSetup setup(program_code);
-    setup.program.size_a = CalculateProgramSize(program_code);
-    setup.program.size_b = 0;
     if (program_type == Maxwell::ShaderProgram::VertexA) {
         // VertexB is always enabled, so when VertexA is enabled, we have two vertex shaders.
         // Conventional HW does not support this, so we combine VertexA and VertexB into one
         // stage here.
         setup.SetProgramB(program_code_b);
-        setup.program.size_b = CalculateProgramSize(program_code_b);
     }
-    setup.program.unique_identifier = GetUniqueIdentifier(
-        program_type, program_code, program_code_b, setup.program.size_a, setup.program.size_b);
+    setup.program.unique_identifier =
+        GetUniqueIdentifier(program_type, program_code, program_code_b);
 
     switch (program_type) {
     case Maxwell::ShaderProgram::VertexA:
@@ -190,11 +182,8 @@ CachedProgram SpecializeShader(const std::string& code, const GLShader::ShaderEn
     const auto texture_buffer_usage{variant.texture_buffer_usage};
 
     std::string source = "#version 430 core\n"
-                         "#extension GL_ARB_separate_shader_objects : enable\n";
-    if (entries.shader_viewport_layer_array) {
-        source += "#extension GL_ARB_shader_viewport_layer_array : enable\n";
-    }
-    source += fmt::format("\n#define EMULATION_UBO_BINDING {}\n", base_bindings.cbuf++);
+                         "#extension GL_ARB_separate_shader_objects : enable\n\n";
+    source += fmt::format("#define EMULATION_UBO_BINDING {}\n", base_bindings.cbuf++);
 
     for (const auto& cbuf : entries.const_buffers) {
         source +=

src/video_core/renderer_opengl/gl_shader_decompiler.cpp

@@ -14,7 +14,6 @@
 #include "common/alignment.h"
 #include "common/assert.h"
 #include "common/common_types.h"
-#include "common/logging/log.h"
 #include "video_core/engines/maxwell_3d.h"
 #include "video_core/renderer_opengl/gl_device.h"
 #include "video_core/renderer_opengl/gl_rasterizer.h"
@@ -47,7 +46,7 @@ using TextureArgument = std::pair<Type, Node>;
 using TextureIR = std::variant<TextureAoffi, TextureArgument>;
 
 constexpr u32 MAX_CONSTBUFFER_ELEMENTS =
-    static_cast<u32>(Maxwell::MaxConstBufferSize) / (4 * sizeof(float));
+    static_cast<u32>(RasterizerOpenGL::MaxConstbufferSize) / (4 * sizeof(float));
 
 class ShaderWriter {
 public:
@@ -192,12 +191,10 @@ public:
 
         // TODO(Subv): Figure out the actual depth of the flow stack, for now it seems
         // unlikely that shaders will use 20 nested SSYs and PBKs.
-        if (!ir.IsFlowStackDisabled()) {
-            constexpr u32 FLOW_STACK_SIZE = 20;
-            for (const auto stack : std::array{MetaStackClass::Ssy, MetaStackClass::Pbk}) {
-                code.AddLine("uint {}[{}];", FlowStackName(stack), FLOW_STACK_SIZE);
-                code.AddLine("uint {} = 0u;", FlowStackTopName(stack));
-            }
+        constexpr u32 FLOW_STACK_SIZE = 20;
+        for (const auto stack : std::array{MetaStackClass::Ssy, MetaStackClass::Pbk}) {
+            code.AddLine("uint {}[{}];", FlowStackName(stack), FLOW_STACK_SIZE);
+            code.AddLine("uint {} = 0u;", FlowStackTopName(stack));
         }
 
         code.AddLine("while (true) {{");
@@ -247,13 +244,15 @@ public:
                                                        usage.is_read, usage.is_written);
         }
         entries.clip_distances = ir.GetClipDistances();
-        entries.shader_viewport_layer_array =
-            stage == ShaderStage::Vertex && (ir.UsesLayer() || ir.UsesViewportIndex());
         entries.shader_length = ir.GetLength();
         return entries;
     }
 
 private:
+    using OperationDecompilerFn = std::string (GLSLDecompiler::*)(Operation);
+    using OperationDecompilersArray =
+        std::array<OperationDecompilerFn, static_cast<std::size_t>(OperationCode::Amount)>;
+
     void DeclareVertex() {
         if (stage != ShaderStage::Vertex)
             return;
@@ -281,35 +280,22 @@ private:
     }
 
     void DeclareVertexRedeclarations() {
+        bool clip_distances_declared = false;
+
         code.AddLine("out gl_PerVertex {{");
         ++code.scope;
 
         code.AddLine("vec4 gl_Position;");
 
-        for (const auto attribute : ir.GetOutputAttributes()) {
-            if (attribute == Attribute::Index::ClipDistances0123 ||
-                attribute == Attribute::Index::ClipDistances4567) {
+        for (const auto o : ir.GetOutputAttributes()) {
+            if (o == Attribute::Index::PointSize)
+                code.AddLine("float gl_PointSize;");
+            if (!clip_distances_declared && (o == Attribute::Index::ClipDistances0123 ||
+                                             o == Attribute::Index::ClipDistances4567)) {
                 code.AddLine("float gl_ClipDistance[];");
-                break;
+                clip_distances_declared = true;
             }
         }
-        if (stage != ShaderStage::Vertex || device.HasVertexViewportLayer()) {
-            if (ir.UsesLayer()) {
-                code.AddLine("int gl_Layer;");
-            }
-            if (ir.UsesViewportIndex()) {
-                code.AddLine("int gl_ViewportIndex;");
-            }
-        } else if ((ir.UsesLayer() || ir.UsesViewportIndex()) && stage == ShaderStage::Vertex &&
-                   !device.HasVertexViewportLayer()) {
-            LOG_ERROR(
-                Render_OpenGL,
-                "GL_ARB_shader_viewport_layer_array is not available and its required by a shader");
-        }
-
-        if (ir.UsesPointSize()) {
-            code.AddLine("float gl_PointSize;");
-        }
 
         --code.scope;
         code.AddLine("}};");
@@ -817,45 +803,6 @@ private:
         return CastOperand(VisitOperand(operation, operand_index), type);
     }
 
-    std::optional<std::pair<std::string, bool>> GetOutputAttribute(const AbufNode* abuf) {
-        switch (const auto attribute = abuf->GetIndex()) {
-        case Attribute::Index::Position:
-            return std::make_pair("gl_Position"s + GetSwizzle(abuf->GetElement()), false);
-        case Attribute::Index::LayerViewportPointSize:
-            switch (abuf->GetElement()) {
-            case 0:
-                UNIMPLEMENTED();
-                return {};
-            case 1:
-                if (stage == ShaderStage::Vertex && !device.HasVertexViewportLayer()) {
-                    return {};
-                }
-                return std::make_pair("gl_Layer", true);
-            case 2:
-                if (stage == ShaderStage::Vertex && !device.HasVertexViewportLayer()) {
-                    return {};
-                }
-                return std::make_pair("gl_ViewportIndex", true);
-            case 3:
-                UNIMPLEMENTED_MSG("Requires some state changes for gl_PointSize to work in shader");
-                return std::make_pair("gl_PointSize", false);
-            }
-            return {};
-        case Attribute::Index::ClipDistances0123:
-            return std::make_pair(fmt::format("gl_ClipDistance[{}]", abuf->GetElement()), false);
-        case Attribute::Index::ClipDistances4567:
-            return std::make_pair(fmt::format("gl_ClipDistance[{}]", abuf->GetElement() + 4),
-                                  false);
-        default:
-            if (IsGenericAttribute(attribute)) {
-                return std::make_pair(
-                    GetOutputAttribute(attribute) + GetSwizzle(abuf->GetElement()), false);
-            }
-            UNIMPLEMENTED_MSG("Unhandled output attribute: {}", static_cast<u32>(attribute));
-            return {};
-        }
-    }
-
     std::string CastOperand(const std::string& value, Type type) const {
         switch (type) {
         case Type::Bool:
@@ -1052,8 +999,6 @@ private:
         const Node& src = operation[1];
 
         std::string target;
-        bool is_integer = false;
-
         if (const auto gpr = std::get_if<GprNode>(&*dest)) {
             if (gpr->GetIndex() == Register::ZeroIndex) {
                 // Writing to Register::ZeroIndex is a no op
@@ -1062,12 +1007,26 @@ private:
             target = GetRegister(gpr->GetIndex());
         } else if (const auto abuf = std::get_if<AbufNode>(&*dest)) {
             UNIMPLEMENTED_IF(abuf->IsPhysicalBuffer());
-            const auto result = GetOutputAttribute(abuf);
-            if (!result) {
-                return {};
-            }
-            target = result->first;
-            is_integer = result->second;
+
+            target = [&]() -> std::string {
+                switch (const auto attribute = abuf->GetIndex(); abuf->GetIndex()) {
+                case Attribute::Index::Position:
+                    return "gl_Position"s + GetSwizzle(abuf->GetElement());
+                case Attribute::Index::PointSize:
+                    return "gl_PointSize";
+                case Attribute::Index::ClipDistances0123:
+                    return fmt::format("gl_ClipDistance[{}]", abuf->GetElement());
+                case Attribute::Index::ClipDistances4567:
+                    return fmt::format("gl_ClipDistance[{}]", abuf->GetElement() + 4);
+                default:
+                    if (IsGenericAttribute(attribute)) {
+                        return GetOutputAttribute(attribute) + GetSwizzle(abuf->GetElement());
+                    }
+                    UNIMPLEMENTED_MSG("Unhandled output attribute: {}",
+                                      static_cast<u32>(attribute));
+                    return "0";
+                }
+            }();
         } else if (const auto lmem = std::get_if<LmemNode>(&*dest)) {
             target = fmt::format("{}[ftou({}) / 4]", GetLocalMemory(), Visit(lmem->GetAddress()));
         } else if (const auto gmem = std::get_if<GmemNode>(&*dest)) {
@@ -1079,11 +1038,7 @@ private:
             UNREACHABLE_MSG("Assign called without a proper target");
         }
 
-        if (is_integer) {
-            code.AddLine("{} = ftoi({});", target, Visit(src));
-        } else {
-            code.AddLine("{} = {};", target, Visit(src));
-        }
+        code.AddLine("{} = {};", target, Visit(src));
         return {};
     }
 
@@ -1122,16 +1077,6 @@ private:
                                Type::Float);
     }
 
-    std::string FCastHalf0(Operation operation) {
-        const std::string op_a = VisitOperand(operation, 0, Type::HalfFloat);
-        return fmt::format("({})[0]", op_a);
-    }
-
-    std::string FCastHalf1(Operation operation) {
-        const std::string op_a = VisitOperand(operation, 0, Type::HalfFloat);
-        return fmt::format("({})[1]", op_a);
-    }
-
     template <Type type>
     std::string Min(Operation operation) {
         return GenerateBinaryCall(operation, "min", type, type, type);
@@ -1288,11 +1233,6 @@ private:
         return ApplyPrecise(operation, BitwiseCastResult(clamped, Type::HalfFloat));
     }
 
-    std::string HCastFloat(Operation operation) {
-        const std::string op_a = VisitOperand(operation, 0, Type::Float);
-        return fmt::format("fromHalf2(vec2({}, 0.0f))", op_a);
-    }
-
     std::string HUnpack(Operation operation) {
         const std::string operand{VisitOperand(operation, 0, Type::HalfFloat)};
         const auto value = [&]() -> std::string {
@@ -1411,10 +1351,14 @@ private:
         return fmt::format("{}[{}]", pair, VisitOperand(operation, 1, Type::Uint));
     }
 
-    std::string LogicalAnd2(Operation operation) {
+    std::string LogicalAll2(Operation operation) {
         return GenerateUnary(operation, "all", Type::Bool, Type::Bool2);
     }
 
+    std::string LogicalAny2(Operation operation) {
+        return GenerateUnary(operation, "any", Type::Bool, Type::Bool2);
+    }
+
     template <bool with_nan>
     std::string GenerateHalfComparison(Operation operation, const std::string& compare_op) {
         const std::string comparison{GenerateBinaryCall(operation, compare_op, Type::Bool2,
@@ -1611,14 +1555,6 @@ private:
         return {};
     }
 
-    std::string BranchIndirect(Operation operation) {
-        const std::string op_a = VisitOperand(operation, 0, Type::Uint);
-
-        code.AddLine("jmp_to = {};", op_a);
-        code.AddLine("break;");
-        return {};
-    }
-
     std::string PushFlowStack(Operation operation) {
         const auto stack = std::get<MetaStackClass>(operation.GetMeta());
         const auto target = std::get_if<ImmediateNode>(&*operation[0]);
@@ -1721,7 +1657,7 @@ private:
         return "utof(gl_WorkGroupID"s + GetSwizzle(element) + ')';
     }
 
-    static constexpr std::array operation_decompilers = {
+    static constexpr OperationDecompilersArray operation_decompilers = {
         &GLSLDecompiler::Assign,
 
         &GLSLDecompiler::Select,
@@ -1733,8 +1669,6 @@ private:
         &GLSLDecompiler::Negate<Type::Float>,
         &GLSLDecompiler::Absolute<Type::Float>,
         &GLSLDecompiler::FClamp,
-        &GLSLDecompiler::FCastHalf0,
-        &GLSLDecompiler::FCastHalf1,
         &GLSLDecompiler::Min<Type::Float>,
         &GLSLDecompiler::Max<Type::Float>,
         &GLSLDecompiler::FCos,
@@ -1795,7 +1729,6 @@ private:
         &GLSLDecompiler::Absolute<Type::HalfFloat>,
         &GLSLDecompiler::HNegate,
         &GLSLDecompiler::HClamp,
-        &GLSLDecompiler::HCastFloat,
         &GLSLDecompiler::HUnpack,
         &GLSLDecompiler::HMergeF32,
         &GLSLDecompiler::HMergeH0,
@@ -1808,7 +1741,8 @@ private:
         &GLSLDecompiler::LogicalXor,
         &GLSLDecompiler::LogicalNegate,
         &GLSLDecompiler::LogicalPick2,
-        &GLSLDecompiler::LogicalAnd2,
+        &GLSLDecompiler::LogicalAll2,
+        &GLSLDecompiler::LogicalAny2,
 
         &GLSLDecompiler::LogicalLessThan<Type::Float>,
         &GLSLDecompiler::LogicalEqual<Type::Float>,
@@ -1855,7 +1789,6 @@ private:
         &GLSLDecompiler::ImageStore,
 
         &GLSLDecompiler::Branch,
-        &GLSLDecompiler::BranchIndirect,
         &GLSLDecompiler::PushFlowStack,
         &GLSLDecompiler::PopFlowStack,
         &GLSLDecompiler::Exit,
@@ -1872,7 +1805,6 @@ private:
         &GLSLDecompiler::WorkGroupId<1>,
         &GLSLDecompiler::WorkGroupId<2>,
     };
-    static_assert(operation_decompilers.size() == static_cast<std::size_t>(OperationCode::Amount));
 
     std::string GetRegister(u32 index) const {
         return GetDeclarationWithSuffix(index, "gpr");

src/video_core/renderer_opengl/gl_shader_decompiler.h

@@ -78,7 +78,6 @@ struct ShaderEntries {
     std::vector<ImageEntry> images;
     std::vector<GlobalMemoryEntry> global_memory_entries;
     std::array<bool, Maxwell::NumClipDistances> clip_distances{};
-    bool shader_viewport_layer_array{};
     std::size_t shader_length{};
 };
 

src/video_core/renderer_opengl/gl_shader_disk_cache.cpp

@@ -373,12 +373,6 @@ std::optional<ShaderDiskCacheDecompiled> ShaderDiskCacheOpenGL::LoadDecompiledEn
         }
     }
 
-    bool shader_viewport_layer_array{};
-    if (!LoadObjectFromPrecompiled(shader_viewport_layer_array)) {
-        return {};
-    }
-    entry.entries.shader_viewport_layer_array = shader_viewport_layer_array;
-
     u64 shader_length{};
     if (!LoadObjectFromPrecompiled(shader_length)) {
         return {};
@@ -451,10 +445,6 @@ bool ShaderDiskCacheOpenGL::SaveDecompiledFile(u64 unique_identifier, const std:
         }
     }
 
-    if (!SaveObjectToPrecompiled(entries.shader_viewport_layer_array)) {
-        return false;
-    }
-
     if (!SaveObjectToPrecompiled(static_cast<u64>(entries.shader_length))) {
         return false;
     }

src/video_core/renderer_opengl/gl_shader_gen.cpp

@@ -29,14 +29,14 @@ layout (std140, binding = EMULATION_UBO_BINDING) uniform vs_config {
 };
 
 )";
-    const ShaderIR program_ir(setup.program.code, PROGRAM_OFFSET, setup.program.size_a);
+    const ShaderIR program_ir(setup.program.code, PROGRAM_OFFSET);
     ProgramResult program =
         Decompile(device, program_ir, Maxwell3D::Regs::ShaderStage::Vertex, "vertex");
 
     out += program.first;
 
     if (setup.IsDualProgram()) {
-        const ShaderIR program_ir_b(setup.program.code_b, PROGRAM_OFFSET, setup.program.size_b);
+        const ShaderIR program_ir_b(setup.program.code_b, PROGRAM_OFFSET);
         ProgramResult program_b =
             Decompile(device, program_ir_b, Maxwell3D::Regs::ShaderStage::Vertex, "vertex_b");
 
@@ -80,7 +80,7 @@ layout (std140, binding = EMULATION_UBO_BINDING) uniform gs_config {
 };
 
 )";
-    const ShaderIR program_ir(setup.program.code, PROGRAM_OFFSET, setup.program.size_a);
+    const ShaderIR program_ir(setup.program.code, PROGRAM_OFFSET);
     ProgramResult program =
         Decompile(device, program_ir, Maxwell3D::Regs::ShaderStage::Geometry, "geometry");
     out += program.first;
@@ -115,7 +115,7 @@ layout (std140, binding = EMULATION_UBO_BINDING) uniform fs_config {
 };
 
 )";
-    const ShaderIR program_ir(setup.program.code, PROGRAM_OFFSET, setup.program.size_a);
+    const ShaderIR program_ir(setup.program.code, PROGRAM_OFFSET);
     ProgramResult program =
         Decompile(device, program_ir, Maxwell3D::Regs::ShaderStage::Fragment, "fragment");
 

src/video_core/renderer_opengl/gl_shader_gen.h

@@ -27,8 +27,6 @@ struct ShaderSetup {
         ProgramCode code;
         ProgramCode code_b; // Used for dual vertex shaders
         u64 unique_identifier;
-        std::size_t size_a;
-        std::size_t size_b;
     } program;
 
     /// Used in scenarios where we have a dual vertex shaders

src/video_core/renderer_opengl/gl_state.cpp

@@ -6,11 +6,8 @@
 #include <glad/glad.h>
 #include "common/assert.h"
 #include "common/logging/log.h"
-#include "common/microprofile.h"
 #include "video_core/renderer_opengl/gl_state.h"
 
-MICROPROFILE_DEFINE(OpenGL_State, "OpenGL", "State Change", MP_RGB(192, 128, 128));
-
 namespace OpenGL {
 
 using Maxwell = Tegra::Engines::Maxwell3D::Regs;
@@ -527,7 +524,6 @@ void OpenGLState::ApplySamplers() const {
 }
 
 void OpenGLState::Apply() const {
-    MICROPROFILE_SCOPE(OpenGL_State);
     ApplyFramebufferState();
     ApplyVertexArrayState();
     ApplyShaderProgram();

src/video_core/renderer_opengl/gl_texture_cache.cpp

@@ -31,8 +31,6 @@ using VideoCore::Surface::SurfaceType;
 
 MICROPROFILE_DEFINE(OpenGL_Texture_Upload, "OpenGL", "Texture Upload", MP_RGB(128, 192, 128));
 MICROPROFILE_DEFINE(OpenGL_Texture_Download, "OpenGL", "Texture Download", MP_RGB(128, 192, 128));
-MICROPROFILE_DEFINE(OpenGL_Texture_Buffer_Copy, "OpenGL", "Texture Buffer Copy",
-                    MP_RGB(128, 192, 128));
 
 namespace {
 
@@ -537,7 +535,6 @@ void TextureCacheOpenGL::ImageBlit(View& src_view, View& dst_view,
 }
 
 void TextureCacheOpenGL::BufferCopy(Surface& src_surface, Surface& dst_surface) {
-    MICROPROFILE_SCOPE(OpenGL_Texture_Buffer_Copy);
     const auto& src_params = src_surface->GetSurfaceParams();
     const auto& dst_params = dst_surface->GetSurfaceParams();
     UNIMPLEMENTED_IF(src_params.num_levels > 1 || dst_params.num_levels > 1);

src/video_core/renderer_opengl/renderer_opengl.cpp

@@ -101,6 +101,7 @@ RendererOpenGL::RendererOpenGL(Core::Frontend::EmuWindow& emu_window, Core::Syst
 
 RendererOpenGL::~RendererOpenGL() = default;
 
+/// Swap buffers (render frame)
 void RendererOpenGL::SwapBuffers(
     std::optional<std::reference_wrapper<const Tegra::FramebufferConfig>> framebuffer) {
 
@@ -129,8 +130,6 @@ void RendererOpenGL::SwapBuffers(
 
         DrawScreen(render_window.GetFramebufferLayout());
 
-        rasterizer->TickFrame();
-
         render_window.SwapBuffers();
     }
 
@@ -263,6 +262,7 @@ void RendererOpenGL::CreateRasterizer() {
     if (rasterizer) {
         return;
     }
+    // Initialize sRGB Usage
     OpenGLState::ClearsRGBUsed();
     rasterizer = std::make_unique<RasterizerOpenGL>(system, emu_window, screen_info);
 }

src/video_core/renderer_opengl/utils.cpp

@@ -13,67 +13,29 @@
 
 namespace OpenGL {
 
-VertexArrayPushBuffer::VertexArrayPushBuffer() = default;
-
-VertexArrayPushBuffer::~VertexArrayPushBuffer() = default;
-
-void VertexArrayPushBuffer::Setup(GLuint vao_) {
-    vao = vao_;
-    index_buffer = nullptr;
-    vertex_buffers.clear();
-}
-
-void VertexArrayPushBuffer::SetIndexBuffer(const GLuint* buffer) {
-    index_buffer = buffer;
-}
-
-void VertexArrayPushBuffer::SetVertexBuffer(GLuint binding_index, const GLuint* buffer,
-                                            GLintptr offset, GLsizei stride) {
-    vertex_buffers.push_back(Entry{binding_index, buffer, offset, stride});
-}
-
-void VertexArrayPushBuffer::Bind() {
-    if (index_buffer) {
-        glVertexArrayElementBuffer(vao, *index_buffer);
-    }
-
-    // TODO(Rodrigo): Find a way to ARB_multi_bind this
-    for (const auto& entry : vertex_buffers) {
-        glVertexArrayVertexBuffer(vao, entry.binding_index, *entry.buffer, entry.offset,
-                                  entry.stride);
-    }
-}
-
 BindBuffersRangePushBuffer::BindBuffersRangePushBuffer(GLenum target) : target{target} {}
 
 BindBuffersRangePushBuffer::~BindBuffersRangePushBuffer() = default;
 
 void BindBuffersRangePushBuffer::Setup(GLuint first_) {
     first = first_;
-    buffer_pointers.clear();
+    buffers.clear();
     offsets.clear();
     sizes.clear();
 }
 
-void BindBuffersRangePushBuffer::Push(const GLuint* buffer, GLintptr offset, GLsizeiptr size) {
-    buffer_pointers.push_back(buffer);
+void BindBuffersRangePushBuffer::Push(GLuint buffer, GLintptr offset, GLsizeiptr size) {
+    buffers.push_back(buffer);
     offsets.push_back(offset);
     sizes.push_back(size);
 }
 
-void BindBuffersRangePushBuffer::Bind() {
-    // Ensure sizes are valid.
-    const std::size_t count{buffer_pointers.size()};
+void BindBuffersRangePushBuffer::Bind() const {
+    const std::size_t count{buffers.size()};
     DEBUG_ASSERT(count == offsets.size() && count == sizes.size());
     if (count == 0) {
         return;
     }
-
-    // Dereference buffers.
-    buffers.resize(count);
-    std::transform(buffer_pointers.begin(), buffer_pointers.end(), buffers.begin(),
-                   [](const GLuint* pointer) { return *pointer; });
-
     glBindBuffersRange(target, first, static_cast<GLsizei>(count), buffers.data(), offsets.data(),
                        sizes.data());
 }

src/video_core/renderer_opengl/utils.h

@@ -11,49 +11,20 @@
 
 namespace OpenGL {
 
-class VertexArrayPushBuffer final {
+class BindBuffersRangePushBuffer {
 public:
-    explicit VertexArrayPushBuffer();
-    ~VertexArrayPushBuffer();
-
-    void Setup(GLuint vao_);
-
-    void SetIndexBuffer(const GLuint* buffer);
-
-    void SetVertexBuffer(GLuint binding_index, const GLuint* buffer, GLintptr offset,
-                         GLsizei stride);
-
-    void Bind();
-
-private:
-    struct Entry {
-        GLuint binding_index{};
-        const GLuint* buffer{};
-        GLintptr offset{};
-        GLsizei stride{};
-    };
-
-    GLuint vao{};
-    const GLuint* index_buffer{};
-    std::vector<Entry> vertex_buffers;
-};
-
-class BindBuffersRangePushBuffer final {
-public:
-    explicit BindBuffersRangePushBuffer(GLenum target);
+    BindBuffersRangePushBuffer(GLenum target);
     ~BindBuffersRangePushBuffer();
 
     void Setup(GLuint first_);
 
-    void Push(const GLuint* buffer, GLintptr offset, GLsizeiptr size);
+    void Push(GLuint buffer, GLintptr offset, GLsizeiptr size);
 
-    void Bind();
+    void Bind() const;
 
 private:
-    GLenum target{};
-    GLuint first{};
-    std::vector<const GLuint*> buffer_pointers;
-
+    GLenum target;
+    GLuint first;
     std::vector<GLuint> buffers;
     std::vector<GLintptr> offsets;
     std::vector<GLsizeiptr> sizes;

src/video_core/renderer_vulkan/vk_buffer_cache.cpp

@@ -109,8 +109,8 @@ void VKBufferCache::Reserve(std::size_t max_size) {
     }
 }
 
-void VKBufferCache::Send() {
-    stream_buffer->Send(buffer_offset - buffer_offset_base);
+VKExecutionContext VKBufferCache::Send(VKExecutionContext exctx) {
+    return stream_buffer->Send(exctx, buffer_offset - buffer_offset_base);
 }
 
 void VKBufferCache::AlignBuffer(std::size_t alignment) {

src/video_core/renderer_vulkan/vk_buffer_cache.h

@@ -77,7 +77,7 @@ public:
     void Reserve(std::size_t max_size);
 
     /// Ensures that the set data is sent to the device.
-    void Send();
+    [[nodiscard]] VKExecutionContext Send(VKExecutionContext exctx);
 
     /// Returns the buffer cache handle.
     vk::Buffer GetBuffer() const {

src/video_core/renderer_vulkan/vk_sampler_cache.h

@@ -4,6 +4,9 @@
 
 #pragma once
 
+#include <unordered_map>
+
+#include "common/common_types.h"
 #include "video_core/renderer_vulkan/declarations.h"
 #include "video_core/sampler_cache.h"
 #include "video_core/textures/texture.h"
@@ -18,9 +21,9 @@ public:
     ~VKSamplerCache();
 
 protected:
-    UniqueSampler CreateSampler(const Tegra::Texture::TSCEntry& tsc) const override;
+    UniqueSampler CreateSampler(const Tegra::Texture::TSCEntry& tsc) const;
 
-    vk::Sampler ToSamplerType(const UniqueSampler& sampler) const override;
+    vk::Sampler ToSamplerType(const UniqueSampler& sampler) const;
 
 private:
     const VKDevice& device;

src/video_core/renderer_vulkan/vk_scheduler.cpp

@@ -19,19 +19,23 @@ VKScheduler::VKScheduler(const VKDevice& device, VKResourceManager& resource_man
 
 VKScheduler::~VKScheduler() = default;
 
-void VKScheduler::Flush(bool release_fence, vk::Semaphore semaphore) {
-    SubmitExecution(semaphore);
-    if (release_fence)
-        current_fence->Release();
-    AllocateNewContext();
+VKExecutionContext VKScheduler::GetExecutionContext() const {
+    return VKExecutionContext(current_fence, current_cmdbuf);
 }
 
-void VKScheduler::Finish(bool release_fence, vk::Semaphore semaphore) {
+VKExecutionContext VKScheduler::Flush(vk::Semaphore semaphore) {
+    SubmitExecution(semaphore);
+    current_fence->Release();
+    AllocateNewContext();
+    return GetExecutionContext();
+}
+
+VKExecutionContext VKScheduler::Finish(vk::Semaphore semaphore) {
     SubmitExecution(semaphore);
     current_fence->Wait();
-    if (release_fence)
-        current_fence->Release();
+    current_fence->Release();
     AllocateNewContext();
+    return GetExecutionContext();
 }
 
 void VKScheduler::SubmitExecution(vk::Semaphore semaphore) {

src/video_core/renderer_vulkan/vk_scheduler.h

@@ -10,43 +10,10 @@
 namespace Vulkan {
 
 class VKDevice;
+class VKExecutionContext;
 class VKFence;
 class VKResourceManager;
 
-class VKFenceView {
-public:
-    VKFenceView() = default;
-    VKFenceView(VKFence* const& fence) : fence{fence} {}
-
-    VKFence* operator->() const noexcept {
-        return fence;
-    }
-
-    operator VKFence&() const noexcept {
-        return *fence;
-    }
-
-private:
-    VKFence* const& fence;
-};
-
-class VKCommandBufferView {
-public:
-    VKCommandBufferView() = default;
-    VKCommandBufferView(const vk::CommandBuffer& cmdbuf) : cmdbuf{cmdbuf} {}
-
-    const vk::CommandBuffer* operator->() const noexcept {
-        return &cmdbuf;
-    }
-
-    operator vk::CommandBuffer() const noexcept {
-        return cmdbuf;
-    }
-
-private:
-    const vk::CommandBuffer& cmdbuf;
-};
-
 /// The scheduler abstracts command buffer and fence management with an interface that's able to do
 /// OpenGL-like operations on Vulkan command buffers.
 class VKScheduler {
@@ -54,21 +21,16 @@ public:
     explicit VKScheduler(const VKDevice& device, VKResourceManager& resource_manager);
     ~VKScheduler();
 
-    /// Gets a reference to the current fence.
-    VKFenceView GetFence() const {
-        return current_fence;
-    }
+    /// Gets the current execution context.
+    [[nodiscard]] VKExecutionContext GetExecutionContext() const;
 
-    /// Gets a reference to the current command buffer.
-    VKCommandBufferView GetCommandBuffer() const {
-        return current_cmdbuf;
-    }
+    /// Sends the current execution context to the GPU. It invalidates the current execution context
+    /// and returns a new one.
+    VKExecutionContext Flush(vk::Semaphore semaphore = nullptr);
 
-    /// Sends the current execution context to the GPU.
-    void Flush(bool release_fence = true, vk::Semaphore semaphore = nullptr);
-
-    /// Sends the current execution context to the GPU and waits for it to complete.
-    void Finish(bool release_fence = true, vk::Semaphore semaphore = nullptr);
+    /// Sends the current execution context to the GPU and waits for it to complete. It invalidates
+    /// the current execution context and returns a new one.
+    VKExecutionContext Finish(vk::Semaphore semaphore = nullptr);
 
 private:
     void SubmitExecution(vk::Semaphore semaphore);
@@ -82,4 +44,26 @@ private:
     VKFence* next_fence = nullptr;
 };
 
+class VKExecutionContext {
+    friend class VKScheduler;
+
+public:
+    VKExecutionContext() = default;
+
+    VKFence& GetFence() const {
+        return *fence;
+    }
+
+    vk::CommandBuffer GetCommandBuffer() const {
+        return cmdbuf;
+    }
+
+private:
+    explicit VKExecutionContext(VKFence* fence, vk::CommandBuffer cmdbuf)
+        : fence{fence}, cmdbuf{cmdbuf} {}
+
+    VKFence* fence{};
+    vk::CommandBuffer cmdbuf;
+};
+
 } // namespace Vulkan

src/video_core/renderer_vulkan/vk_shader_decompiler.cpp

@@ -205,6 +205,10 @@ public:
     }
 
 private:
+    using OperationDecompilerFn = Id (SPIRVDecompiler::*)(Operation);
+    using OperationDecompilersArray =
+        std::array<OperationDecompilerFn, static_cast<std::size_t>(OperationCode::Amount)>;
+
     static constexpr auto INTERNAL_FLAGS_COUNT = static_cast<std::size_t>(InternalFlag::Amount);
 
     void AllocateBindings() {
@@ -426,17 +430,20 @@ private:
         instance_index = DeclareBuiltIn(spv::BuiltIn::InstanceIndex, spv::StorageClass::Input,
                                         t_in_uint, "instance_index");
 
+        bool is_point_size_declared = false;
         bool is_clip_distances_declared = false;
         for (const auto index : ir.GetOutputAttributes()) {
-            if (index == Attribute::Index::ClipDistances0123 ||
-                index == Attribute::Index::ClipDistances4567) {
+            if (index == Attribute::Index::PointSize) {
+                is_point_size_declared = true;
+            } else if (index == Attribute::Index::ClipDistances0123 ||
+                       index == Attribute::Index::ClipDistances4567) {
                 is_clip_distances_declared = true;
             }
         }
 
         std::vector<Id> members;
         members.push_back(t_float4);
-        if (ir.UsesPointSize()) {
+        if (is_point_size_declared) {
             members.push_back(t_float);
         }
         if (is_clip_distances_declared) {
@@ -459,7 +466,7 @@ private:
 
         position_index = MemberDecorateBuiltIn(spv::BuiltIn::Position, "position", true);
         point_size_index =
-            MemberDecorateBuiltIn(spv::BuiltIn::PointSize, "point_size", ir.UsesPointSize());
+            MemberDecorateBuiltIn(spv::BuiltIn::PointSize, "point_size", is_point_size_declared);
         clip_distances_index = MemberDecorateBuiltIn(spv::BuiltIn::ClipDistance, "clip_distances",
                                                      is_clip_distances_declared);
 
@@ -705,8 +712,7 @@ private:
                 case Attribute::Index::Position:
                     return AccessElement(t_out_float, per_vertex, position_index,
                                          abuf->GetElement());
-                case Attribute::Index::LayerViewportPointSize:
-                    UNIMPLEMENTED_IF(abuf->GetElement() != 3);
+                case Attribute::Index::PointSize:
                     return AccessElement(t_out_float, per_vertex, point_size_index);
                 case Attribute::Index::ClipDistances0123:
                     return AccessElement(t_out_float, per_vertex, clip_distances_index,
@@ -735,16 +741,6 @@ private:
         return {};
     }
 
-    Id FCastHalf0(Operation operation) {
-        UNIMPLEMENTED();
-        return {};
-    }
-
-    Id FCastHalf1(Operation operation) {
-        UNIMPLEMENTED();
-        return {};
-    }
-
     Id HNegate(Operation operation) {
         UNIMPLEMENTED();
         return {};
@@ -755,11 +751,6 @@ private:
         return {};
     }
 
-    Id HCastFloat(Operation operation) {
-        UNIMPLEMENTED();
-        return {};
-    }
-
     Id HUnpack(Operation operation) {
         UNIMPLEMENTED();
         return {};
@@ -815,7 +806,12 @@ private:
         return {};
     }
 
-    Id LogicalAnd2(Operation operation) {
+    Id LogicalAll2(Operation operation) {
+        UNIMPLEMENTED();
+        return {};
+    }
+
+    Id LogicalAny2(Operation operation) {
         UNIMPLEMENTED();
         return {};
     }
@@ -953,14 +949,6 @@ private:
         return {};
     }
 
-    Id BranchIndirect(Operation operation) {
-        const Id op_a = VisitOperand<Type::Uint>(operation, 0);
-
-        Emit(OpStore(jmp_to, op_a));
-        BranchingOp([&]() { Emit(OpBranch(continue_label)); });
-        return {};
-    }
-
     Id PushFlowStack(Operation operation) {
         const auto target = std::get_if<ImmediateNode>(&*operation[0]);
         ASSERT(target);
@@ -1212,7 +1200,7 @@ private:
         return {};
     }
 
-    static constexpr std::array operation_decompilers = {
+    static constexpr OperationDecompilersArray operation_decompilers = {
         &SPIRVDecompiler::Assign,
 
         &SPIRVDecompiler::Ternary<&Module::OpSelect, Type::Float, Type::Bool, Type::Float,
@@ -1225,8 +1213,6 @@ private:
         &SPIRVDecompiler::Unary<&Module::OpFNegate, Type::Float>,
         &SPIRVDecompiler::Unary<&Module::OpFAbs, Type::Float>,
         &SPIRVDecompiler::Ternary<&Module::OpFClamp, Type::Float>,
-        &SPIRVDecompiler::FCastHalf0,
-        &SPIRVDecompiler::FCastHalf1,
         &SPIRVDecompiler::Binary<&Module::OpFMin, Type::Float>,
         &SPIRVDecompiler::Binary<&Module::OpFMax, Type::Float>,
         &SPIRVDecompiler::Unary<&Module::OpCos, Type::Float>,
@@ -1287,7 +1273,6 @@ private:
         &SPIRVDecompiler::Unary<&Module::OpFAbs, Type::HalfFloat>,
         &SPIRVDecompiler::HNegate,
         &SPIRVDecompiler::HClamp,
-        &SPIRVDecompiler::HCastFloat,
         &SPIRVDecompiler::HUnpack,
         &SPIRVDecompiler::HMergeF32,
         &SPIRVDecompiler::HMergeH0,
@@ -1300,7 +1285,8 @@ private:
         &SPIRVDecompiler::Binary<&Module::OpLogicalNotEqual, Type::Bool>,
         &SPIRVDecompiler::Unary<&Module::OpLogicalNot, Type::Bool>,
         &SPIRVDecompiler::LogicalPick2,
-        &SPIRVDecompiler::LogicalAnd2,
+        &SPIRVDecompiler::LogicalAll2,
+        &SPIRVDecompiler::LogicalAny2,
 
         &SPIRVDecompiler::Binary<&Module::OpFOrdLessThan, Type::Bool, Type::Float>,
         &SPIRVDecompiler::Binary<&Module::OpFOrdEqual, Type::Bool, Type::Float>,
@@ -1348,7 +1334,6 @@ private:
         &SPIRVDecompiler::ImageStore,
 
         &SPIRVDecompiler::Branch,
-        &SPIRVDecompiler::BranchIndirect,
         &SPIRVDecompiler::PushFlowStack,
         &SPIRVDecompiler::PopFlowStack,
         &SPIRVDecompiler::Exit,
@@ -1365,7 +1350,6 @@ private:
         &SPIRVDecompiler::WorkGroupId<1>,
         &SPIRVDecompiler::WorkGroupId<2>,
     };
-    static_assert(operation_decompilers.size() == static_cast<std::size_t>(OperationCode::Amount));
 
     const VKDevice& device;
     const ShaderIR& ir;

src/video_core/renderer_vulkan/vk_stream_buffer.cpp

@@ -46,12 +46,12 @@ std::tuple<u8*, u64, bool> VKStreamBuffer::Reserve(u64 size) {
     return {mapped_pointer + offset, offset, invalidation_mark.has_value()};
 }
 
-void VKStreamBuffer::Send(u64 size) {
+VKExecutionContext VKStreamBuffer::Send(VKExecutionContext exctx, u64 size) {
     ASSERT_MSG(size <= mapped_size, "Reserved size is too small");
 
     if (invalidation_mark) {
         // TODO(Rodrigo): Find a better way to invalidate than waiting for all watches to finish.
-        scheduler.Flush();
+        exctx = scheduler.Flush();
         std::for_each(watches.begin(), watches.begin() + *invalidation_mark,
                       [&](auto& resource) { resource->Wait(); });
         invalidation_mark = std::nullopt;
@@ -62,9 +62,11 @@ void VKStreamBuffer::Send(u64 size) {
         ReserveWatches(WATCHES_RESERVE_CHUNK);
     }
     // Add a watch for this allocation.
-    watches[used_watches++]->Watch(scheduler.GetFence());
+    watches[used_watches++]->Watch(exctx.GetFence());
 
     offset += size;
+
+    return exctx;
 }
 
 void VKStreamBuffer::CreateBuffers(VKMemoryManager& memory_manager, vk::BufferUsageFlags usage) {

src/video_core/renderer_vulkan/vk_stream_buffer.h

@@ -37,7 +37,7 @@ public:
     std::tuple<u8*, u64, bool> Reserve(u64 size);
 
     /// Ensures that "size" bytes of memory are available to the GPU, potentially recording a copy.
-    void Send(u64 size);
+    [[nodiscard]] VKExecutionContext Send(VKExecutionContext exctx, u64 size);
 
     vk::Buffer GetBuffer() const {
         return *buffer;

src/video_core/shader/control_flow.cpp

@@ -1,476 +0,0 @@
-// Copyright 2019 yuzu Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#include <list>
-#include <map>
-#include <stack>
-#include <unordered_map>
-#include <unordered_set>
-#include <vector>
-
-#include "common/assert.h"
-#include "common/common_types.h"
-#include "video_core/shader/control_flow.h"
-#include "video_core/shader/shader_ir.h"
-
-namespace VideoCommon::Shader {
-
-using Tegra::Shader::Instruction;
-using Tegra::Shader::OpCode;
-
-constexpr s32 unassigned_branch = -2;
-
-struct Query {
-    u32 address{};
-    std::stack<u32> ssy_stack{};
-    std::stack<u32> pbk_stack{};
-};
-
-struct BlockStack {
-    BlockStack() = default;
-    BlockStack(const BlockStack& b) = default;
-    BlockStack(const Query& q) : ssy_stack{q.ssy_stack}, pbk_stack{q.pbk_stack} {}
-    std::stack<u32> ssy_stack{};
-    std::stack<u32> pbk_stack{};
-};
-
-struct BlockBranchInfo {
-    Condition condition{};
-    s32 address{exit_branch};
-    bool kill{};
-    bool is_sync{};
-    bool is_brk{};
-    bool ignore{};
-};
-
-struct BlockInfo {
-    u32 start{};
-    u32 end{};
-    bool visited{};
-    BlockBranchInfo branch{};
-
-    bool IsInside(const u32 address) const {
-        return start <= address && address <= end;
-    }
-};
-
-struct CFGRebuildState {
-    explicit CFGRebuildState(const ProgramCode& program_code, const std::size_t program_size,
-                             const u32 start)
-        : program_code{program_code}, program_size{program_size}, start{start} {}
-
-    u32 start{};
-    std::vector<BlockInfo> block_info{};
-    std::list<u32> inspect_queries{};
-    std::list<Query> queries{};
-    std::unordered_map<u32, u32> registered{};
-    std::unordered_set<u32> labels{};
-    std::map<u32, u32> ssy_labels{};
-    std::map<u32, u32> pbk_labels{};
-    std::unordered_map<u32, BlockStack> stacks{};
-    const ProgramCode& program_code;
-    const std::size_t program_size;
-};
-
-enum class BlockCollision : u32 { None, Found, Inside };
-
-std::pair<BlockCollision, u32> TryGetBlock(CFGRebuildState& state, u32 address) {
-    const auto& blocks = state.block_info;
-    for (u32 index = 0; index < blocks.size(); index++) {
-        if (blocks[index].start == address) {
-            return {BlockCollision::Found, index};
-        }
-        if (blocks[index].IsInside(address)) {
-            return {BlockCollision::Inside, index};
-        }
-    }
-    return {BlockCollision::None, -1};
-}
-
-struct ParseInfo {
-    BlockBranchInfo branch_info{};
-    u32 end_address{};
-};
-
-BlockInfo& CreateBlockInfo(CFGRebuildState& state, u32 start, u32 end) {
-    auto& it = state.block_info.emplace_back();
-    it.start = start;
-    it.end = end;
-    const u32 index = static_cast<u32>(state.block_info.size() - 1);
-    state.registered.insert({start, index});
-    return it;
-}
-
-Pred GetPredicate(u32 index, bool negated) {
-    return static_cast<Pred>(index + (negated ? 8 : 0));
-}
-
-/**
- * Returns whether the instruction at the specified offset is a 'sched' instruction.
- * Sched instructions always appear before a sequence of 3 instructions.
- */
-constexpr bool IsSchedInstruction(u32 offset, u32 main_offset) {
-    constexpr u32 SchedPeriod = 4;
-    u32 absolute_offset = offset - main_offset;
-
-    return (absolute_offset % SchedPeriod) == 0;
-}
-
-enum class ParseResult : u32 {
-    ControlCaught,
-    BlockEnd,
-    AbnormalFlow,
-};
-
-std::pair<ParseResult, ParseInfo> ParseCode(CFGRebuildState& state, u32 address) {
-    u32 offset = static_cast<u32>(address);
-    const u32 end_address = static_cast<u32>(state.program_size / sizeof(Instruction));
-    ParseInfo parse_info{};
-
-    const auto insert_label = [](CFGRebuildState& state, u32 address) {
-        const auto pair = state.labels.emplace(address);
-        if (pair.second) {
-            state.inspect_queries.push_back(address);
-        }
-    };
-
-    while (true) {
-        if (offset >= end_address) {
-            // ASSERT_OR_EXECUTE can't be used, as it ignores the break
-            ASSERT_MSG(false, "Shader passed the current limit!");
-            parse_info.branch_info.address = exit_branch;
-            parse_info.branch_info.ignore = false;
-            break;
-        }
-        if (state.registered.count(offset) != 0) {
-            parse_info.branch_info.address = offset;
-            parse_info.branch_info.ignore = true;
-            break;
-        }
-        if (IsSchedInstruction(offset, state.start)) {
-            offset++;
-            continue;
-        }
-        const Instruction instr = {state.program_code[offset]};
-        const auto opcode = OpCode::Decode(instr);
-        if (!opcode || opcode->get().GetType() != OpCode::Type::Flow) {
-            offset++;
-            continue;
-        }
-
-        switch (opcode->get().GetId()) {
-        case OpCode::Id::EXIT: {
-            const auto pred_index = static_cast<u32>(instr.pred.pred_index);
-            parse_info.branch_info.condition.predicate =
-                GetPredicate(pred_index, instr.negate_pred != 0);
-            if (parse_info.branch_info.condition.predicate == Pred::NeverExecute) {
-                offset++;
-                continue;
-            }
-            const ConditionCode cc = instr.flow_condition_code;
-            parse_info.branch_info.condition.cc = cc;
-            if (cc == ConditionCode::F) {
-                offset++;
-                continue;
-            }
-            parse_info.branch_info.address = exit_branch;
-            parse_info.branch_info.kill = false;
-            parse_info.branch_info.is_sync = false;
-            parse_info.branch_info.is_brk = false;
-            parse_info.branch_info.ignore = false;
-            parse_info.end_address = offset;
-
-            return {ParseResult::ControlCaught, parse_info};
-        }
-        case OpCode::Id::BRA: {
-            if (instr.bra.constant_buffer != 0) {
-                return {ParseResult::AbnormalFlow, parse_info};
-            }
-            const auto pred_index = static_cast<u32>(instr.pred.pred_index);
-            parse_info.branch_info.condition.predicate =
-                GetPredicate(pred_index, instr.negate_pred != 0);
-            if (parse_info.branch_info.condition.predicate == Pred::NeverExecute) {
-                offset++;
-                continue;
-            }
-            const ConditionCode cc = instr.flow_condition_code;
-            parse_info.branch_info.condition.cc = cc;
-            if (cc == ConditionCode::F) {
-                offset++;
-                continue;
-            }
-            const u32 branch_offset = offset + instr.bra.GetBranchTarget();
-            if (branch_offset == 0) {
-                parse_info.branch_info.address = exit_branch;
-            } else {
-                parse_info.branch_info.address = branch_offset;
-            }
-            insert_label(state, branch_offset);
-            parse_info.branch_info.kill = false;
-            parse_info.branch_info.is_sync = false;
-            parse_info.branch_info.is_brk = false;
-            parse_info.branch_info.ignore = false;
-            parse_info.end_address = offset;
-
-            return {ParseResult::ControlCaught, parse_info};
-        }
-        case OpCode::Id::SYNC: {
-            const auto pred_index = static_cast<u32>(instr.pred.pred_index);
-            parse_info.branch_info.condition.predicate =
-                GetPredicate(pred_index, instr.negate_pred != 0);
-            if (parse_info.branch_info.condition.predicate == Pred::NeverExecute) {
-                offset++;
-                continue;
-            }
-            const ConditionCode cc = instr.flow_condition_code;
-            parse_info.branch_info.condition.cc = cc;
-            if (cc == ConditionCode::F) {
-                offset++;
-                continue;
-            }
-            parse_info.branch_info.address = unassigned_branch;
-            parse_info.branch_info.kill = false;
-            parse_info.branch_info.is_sync = true;
-            parse_info.branch_info.is_brk = false;
-            parse_info.branch_info.ignore = false;
-            parse_info.end_address = offset;
-
-            return {ParseResult::ControlCaught, parse_info};
-        }
-        case OpCode::Id::BRK: {
-            const auto pred_index = static_cast<u32>(instr.pred.pred_index);
-            parse_info.branch_info.condition.predicate =
-                GetPredicate(pred_index, instr.negate_pred != 0);
-            if (parse_info.branch_info.condition.predicate == Pred::NeverExecute) {
-                offset++;
-                continue;
-            }
-            const ConditionCode cc = instr.flow_condition_code;
-            parse_info.branch_info.condition.cc = cc;
-            if (cc == ConditionCode::F) {
-                offset++;
-                continue;
-            }
-            parse_info.branch_info.address = unassigned_branch;
-            parse_info.branch_info.kill = false;
-            parse_info.branch_info.is_sync = false;
-            parse_info.branch_info.is_brk = true;
-            parse_info.branch_info.ignore = false;
-            parse_info.end_address = offset;
-
-            return {ParseResult::ControlCaught, parse_info};
-        }
-        case OpCode::Id::KIL: {
-            const auto pred_index = static_cast<u32>(instr.pred.pred_index);
-            parse_info.branch_info.condition.predicate =
-                GetPredicate(pred_index, instr.negate_pred != 0);
-            if (parse_info.branch_info.condition.predicate == Pred::NeverExecute) {
-                offset++;
-                continue;
-            }
-            const ConditionCode cc = instr.flow_condition_code;
-            parse_info.branch_info.condition.cc = cc;
-            if (cc == ConditionCode::F) {
-                offset++;
-                continue;
-            }
-            parse_info.branch_info.address = exit_branch;
-            parse_info.branch_info.kill = true;
-            parse_info.branch_info.is_sync = false;
-            parse_info.branch_info.is_brk = false;
-            parse_info.branch_info.ignore = false;
-            parse_info.end_address = offset;
-
-            return {ParseResult::ControlCaught, parse_info};
-        }
-        case OpCode::Id::SSY: {
-            const u32 target = offset + instr.bra.GetBranchTarget();
-            insert_label(state, target);
-            state.ssy_labels.emplace(offset, target);
-            break;
-        }
-        case OpCode::Id::PBK: {
-            const u32 target = offset + instr.bra.GetBranchTarget();
-            insert_label(state, target);
-            state.pbk_labels.emplace(offset, target);
-            break;
-        }
-        case OpCode::Id::BRX: {
-            return {ParseResult::AbnormalFlow, parse_info};
-        }
-        default:
-            break;
-        }
-
-        offset++;
-    }
-    parse_info.branch_info.kill = false;
-    parse_info.branch_info.is_sync = false;
-    parse_info.branch_info.is_brk = false;
-    parse_info.end_address = offset - 1;
-    return {ParseResult::BlockEnd, parse_info};
-}
-
-bool TryInspectAddress(CFGRebuildState& state) {
-    if (state.inspect_queries.empty()) {
-        return false;
-    }
-
-    const u32 address = state.inspect_queries.front();
-    state.inspect_queries.pop_front();
-    const auto [result, block_index] = TryGetBlock(state, address);
-    switch (result) {
-    case BlockCollision::Found: {
-        return true;
-    }
-    case BlockCollision::Inside: {
-        // This case is the tricky one:
-        // We need to Split the block in 2 sepparate blocks
-        const u32 end = state.block_info[block_index].end;
-        BlockInfo& new_block = CreateBlockInfo(state, address, end);
-        BlockInfo& current_block = state.block_info[block_index];
-        current_block.end = address - 1;
-        new_block.branch = current_block.branch;
-        BlockBranchInfo forward_branch{};
-        forward_branch.address = address;
-        forward_branch.ignore = true;
-        current_block.branch = forward_branch;
-        return true;
-    }
-    default:
-        break;
-    }
-    const auto [parse_result, parse_info] = ParseCode(state, address);
-    if (parse_result == ParseResult::AbnormalFlow) {
-        // if it's AbnormalFlow, we end it as false, ending the CFG reconstruction
-        return false;
-    }
-
-    BlockInfo& block_info = CreateBlockInfo(state, address, parse_info.end_address);
-    block_info.branch = parse_info.branch_info;
-    if (parse_info.branch_info.condition.IsUnconditional()) {
-        return true;
-    }
-
-    const u32 fallthrough_address = parse_info.end_address + 1;
-    state.inspect_queries.push_front(fallthrough_address);
-    return true;
-}
-
-bool TryQuery(CFGRebuildState& state) {
-    const auto gather_labels = [](std::stack<u32>& cc, std::map<u32, u32>& labels,
-                                  BlockInfo& block) {
-        auto gather_start = labels.lower_bound(block.start);
-        const auto gather_end = labels.upper_bound(block.end);
-        while (gather_start != gather_end) {
-            cc.push(gather_start->second);
-            gather_start++;
-        }
-    };
-    if (state.queries.empty()) {
-        return false;
-    }
-    Query& q = state.queries.front();
-    const u32 block_index = state.registered[q.address];
-    BlockInfo& block = state.block_info[block_index];
-    // If the block is visted, check if the stacks match, else gather the ssy/pbk
-    // labels into the current stack and look if the branch at the end of the block
-    // consumes a label. Schedule new queries accordingly
-    if (block.visited) {
-        BlockStack& stack = state.stacks[q.address];
-        const bool all_okay = (stack.ssy_stack.size() == 0 || q.ssy_stack == stack.ssy_stack) &&
-                              (stack.pbk_stack.size() == 0 || q.pbk_stack == stack.pbk_stack);
-        state.queries.pop_front();
-        return all_okay;
-    }
-    block.visited = true;
-    state.stacks[q.address] = BlockStack{q};
-    Query q2(q);
-    state.queries.pop_front();
-    gather_labels(q2.ssy_stack, state.ssy_labels, block);
-    gather_labels(q2.pbk_stack, state.pbk_labels, block);
-    if (!block.branch.condition.IsUnconditional()) {
-        q2.address = block.end + 1;
-        state.queries.push_back(q2);
-    }
-    Query conditional_query{q2};
-    if (block.branch.is_sync) {
-        if (block.branch.address == unassigned_branch) {
-            block.branch.address = conditional_query.ssy_stack.top();
-        }
-        conditional_query.ssy_stack.pop();
-    }
-    if (block.branch.is_brk) {
-        if (block.branch.address == unassigned_branch) {
-            block.branch.address = conditional_query.pbk_stack.top();
-        }
-        conditional_query.pbk_stack.pop();
-    }
-    conditional_query.address = block.branch.address;
-    state.queries.push_back(conditional_query);
-    return true;
-}
-
-std::optional<ShaderCharacteristics> ScanFlow(const ProgramCode& program_code, u32 program_size,
-                                              u32 start_address) {
-    CFGRebuildState state{program_code, program_size, start_address};
-    // Inspect Code and generate blocks
-    state.labels.clear();
-    state.labels.emplace(start_address);
-    state.inspect_queries.push_back(state.start);
-    while (!state.inspect_queries.empty()) {
-        if (!TryInspectAddress(state)) {
-            return {};
-        }
-    }
-    // Decompile Stacks
-    Query start_query{};
-    start_query.address = state.start;
-    state.queries.push_back(start_query);
-    bool decompiled = true;
-    while (!state.queries.empty()) {
-        if (!TryQuery(state)) {
-            decompiled = false;
-            break;
-        }
-    }
-    // Sort and organize results
-    std::sort(state.block_info.begin(), state.block_info.end(),
-              [](const BlockInfo& a, const BlockInfo& b) -> bool { return a.start < b.start; });
-    ShaderCharacteristics result_out{};
-    result_out.decompilable = decompiled;
-    result_out.start = start_address;
-    result_out.end = start_address;
-    for (auto& block : state.block_info) {
-        ShaderBlock new_block{};
-        new_block.start = block.start;
-        new_block.end = block.end;
-        new_block.ignore_branch = block.branch.ignore;
-        if (!new_block.ignore_branch) {
-            new_block.branch.cond = block.branch.condition;
-            new_block.branch.kills = block.branch.kill;
-            new_block.branch.address = block.branch.address;
-        }
-        result_out.end = std::max(result_out.end, block.end);
-        result_out.blocks.push_back(new_block);
-    }
-    if (result_out.decompilable) {
-        result_out.labels = std::move(state.labels);
-        return {result_out};
-    }
-    // If it's not decompilable, merge the unlabelled blocks together
-    auto back = result_out.blocks.begin();
-    auto next = std::next(back);
-    while (next != result_out.blocks.end()) {
-        if (state.labels.count(next->start) == 0 && next->start == back->end + 1) {
-            back->end = next->end;
-            next = result_out.blocks.erase(next);
-            continue;
-        }
-        back = next;
-        next++;
-    }
-    return {result_out};
-}
-} // namespace VideoCommon::Shader