summaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
-rw-r--r--.ci/scripts/common/post-upload.sh15
-rw-r--r--.ci/scripts/common/pre-upload.sh6
-rw-r--r--.ci/scripts/format/docker.sh6
-rw-r--r--.ci/scripts/format/exec.sh4
-rw-r--r--.ci/scripts/format/script.sh37
-rw-r--r--.ci/scripts/linux/docker.sh14
-rw-r--r--.ci/scripts/linux/exec.sh5
-rw-r--r--.ci/scripts/linux/upload.sh14
-rw-r--r--.ci/scripts/merge/apply-patches-by-label.py28
-rw-r--r--.ci/scripts/merge/check-label-presence.py18
-rw-r--r--.ci/scripts/merge/yuzubot-git-config.sh2
-rw-r--r--.ci/scripts/windows/docker.sh50
-rw-r--r--.ci/scripts/windows/exec.sh5
-rw-r--r--.ci/scripts/windows/scan_dll.py106
-rw-r--r--.ci/scripts/windows/upload.sh13
-rw-r--r--.ci/templates/build-single.yml21
-rw-r--r--.ci/templates/build-standard.yml22
-rw-r--r--.ci/templates/build-testing.yml30
-rw-r--r--.ci/templates/format-check.yml14
-rw-r--r--.ci/templates/merge.yml46
-rw-r--r--.ci/templates/mergebot.yml15
-rw-r--r--.ci/templates/release.yml29
-rw-r--r--.ci/templates/retrieve-artifact-source.yml16
-rw-r--r--.ci/templates/retrieve-master-source.yml11
-rw-r--r--.ci/templates/sync-source.yml7
-rw-r--r--.ci/yuzu-mainline.yml23
-rw-r--r--.ci/yuzu-patreon.yml19
-rw-r--r--.ci/yuzu-repo-sync.yml19
-rw-r--r--.ci/yuzu-verify.yml18
-rw-r--r--CMakeModules/GenerateSCMRev.cmake2
-rw-r--r--src/audio_core/audio_renderer.cpp14
-rw-r--r--src/audio_core/audio_renderer.h3
-rw-r--r--src/common/CMakeLists.txt2
-rw-r--r--src/core/CMakeLists.txt3
-rw-r--r--src/core/arm/arm_interface.h7
-rw-r--r--src/core/arm/dynarmic/arm_dynarmic.cpp9
-rw-r--r--src/core/arm/dynarmic/arm_dynarmic.h3
-rw-r--r--src/core/arm/unicorn/arm_unicorn.cpp9
-rw-r--r--src/core/arm/unicorn/arm_unicorn.h3
-rw-r--r--src/core/core.cpp14
-rw-r--r--src/core/core.h8
-rw-r--r--src/core/core_cpu.cpp19
-rw-r--r--src/core/file_sys/program_metadata.cpp4
-rw-r--r--src/core/file_sys/program_metadata.h4
-rw-r--r--src/core/hle/kernel/process.cpp16
-rw-r--r--src/core/hle/kernel/process.h34
-rw-r--r--src/core/hle/kernel/svc.cpp126
-rw-r--r--src/core/hle/kernel/svc_wrap.h5
-rw-r--r--src/core/hle/kernel/vm_manager.cpp310
-rw-r--r--src/core/hle/kernel/vm_manager.h48
-rw-r--r--src/core/hle/service/am/am.cpp58
-rw-r--r--src/core/hle/service/am/am.h8
-rw-r--r--src/core/hle/service/am/applet_ae.cpp4
-rw-r--r--src/core/hle/service/am/applet_oe.cpp2
-rw-r--r--src/core/hle/service/apm/apm.cpp13
-rw-r--r--src/core/hle/service/apm/apm.h7
-rw-r--r--src/core/hle/service/apm/controller.cpp68
-rw-r--r--src/core/hle/service/apm/controller.h70
-rw-r--r--src/core/hle/service/apm/interface.cpp82
-rw-r--r--src/core/hle/service/apm/interface.h14
-rw-r--r--src/core/hle/service/audio/audren_u.cpp9
-rw-r--r--src/core/hle/service/audio/audren_u.h1
-rw-r--r--src/core/hle/service/filesystem/filesystem.cpp10
-rw-r--r--src/core/hle/service/filesystem/filesystem.h2
-rw-r--r--src/core/hle/service/filesystem/fsp_srv.cpp54
-rw-r--r--src/core/hle/service/filesystem/fsp_srv.h26
-rw-r--r--src/core/hle/service/friend/friend.cpp35
-rw-r--r--src/core/hle/service/hid/controllers/npad.cpp31
-rw-r--r--src/core/hle/service/hid/controllers/npad.h5
-rw-r--r--src/core/hle/service/hid/errors.h13
-rw-r--r--src/core/hle/service/hid/hid.cpp50
-rw-r--r--src/core/hle/service/hid/hid.h3
-rw-r--r--src/core/hle/service/ldr/ldr.cpp32
-rw-r--r--src/core/hle/service/mii/mii.cpp16
-rw-r--r--src/core/hle/service/pm/pm.cpp124
-rw-r--r--src/core/hle/service/pm/pm.h6
-rw-r--r--src/core/hle/service/service.cpp9
-rw-r--r--src/core/hle/service/service.h3
-rw-r--r--src/core/reporter.cpp18
-rw-r--r--src/core/reporter.h7
-rw-r--r--src/core/settings.cpp1
-rw-r--r--src/core/settings.h1
-rw-r--r--src/core/telemetry_session.cpp1
-rw-r--r--src/video_core/CMakeLists.txt5
-rw-r--r--src/video_core/buffer_cache.h299
-rw-r--r--src/video_core/engines/maxwell_3d.h1
-rw-r--r--src/video_core/engines/shader_bytecode.h19
-rw-r--r--src/video_core/gpu.cpp2
-rw-r--r--src/video_core/macro_interpreter.cpp4
-rw-r--r--src/video_core/memory_manager.cpp24
-rw-r--r--src/video_core/memory_manager.h8
-rw-r--r--src/video_core/rasterizer_interface.h3
-rw-r--r--src/video_core/renderer_opengl/gl_buffer_cache.cpp110
-rw-r--r--src/video_core/renderer_opengl/gl_buffer_cache.h76
-rw-r--r--src/video_core/renderer_opengl/gl_device.cpp1
-rw-r--r--src/video_core/renderer_opengl/gl_device.h5
-rw-r--r--src/video_core/renderer_opengl/gl_global_cache.cpp102
-rw-r--r--src/video_core/renderer_opengl/gl_global_cache.h82
-rw-r--r--src/video_core/renderer_opengl/gl_rasterizer.cpp154
-rw-r--r--src/video_core/renderer_opengl/gl_rasterizer.h29
-rw-r--r--src/video_core/renderer_opengl/gl_sampler_cache.h4
-rw-r--r--src/video_core/renderer_opengl/gl_shader_cache.cpp22
-rw-r--r--src/video_core/renderer_opengl/gl_shader_decompiler.cpp21
-rw-r--r--src/video_core/renderer_opengl/gl_shader_gen.cpp8
-rw-r--r--src/video_core/renderer_opengl/gl_shader_gen.h2
-rw-r--r--src/video_core/renderer_opengl/gl_state.cpp4
-rw-r--r--src/video_core/renderer_opengl/gl_texture_cache.cpp3
-rw-r--r--src/video_core/renderer_opengl/renderer_opengl.cpp4
-rw-r--r--src/video_core/renderer_opengl/utils.cpp48
-rw-r--r--src/video_core/renderer_opengl/utils.h41
-rw-r--r--src/video_core/renderer_vulkan/vk_buffer_cache.cpp4
-rw-r--r--src/video_core/renderer_vulkan/vk_buffer_cache.h2
-rw-r--r--src/video_core/renderer_vulkan/vk_sampler_cache.h7
-rw-r--r--src/video_core/renderer_vulkan/vk_scheduler.cpp16
-rw-r--r--src/video_core/renderer_vulkan/vk_scheduler.h78
-rw-r--r--src/video_core/renderer_vulkan/vk_shader_decompiler.cpp9
-rw-r--r--src/video_core/renderer_vulkan/vk_stream_buffer.cpp8
-rw-r--r--src/video_core/renderer_vulkan/vk_stream_buffer.h2
-rw-r--r--src/video_core/shader/control_flow.cpp476
-rw-r--r--src/video_core/shader/control_flow.h63
-rw-r--r--src/video_core/shader/decode.cpp177
-rw-r--r--src/video_core/shader/decode/other.cpp58
-rw-r--r--src/video_core/shader/decode/texture.cpp8
-rw-r--r--src/video_core/shader/node.h11
-rw-r--r--src/video_core/shader/shader_ir.cpp4
-rw-r--r--src/video_core/shader/shader_ir.h27
-rw-r--r--src/video_core/texture_cache/surface_base.cpp9
-rw-r--r--src/video_core/texture_cache/surface_base.h10
-rw-r--r--src/video_core/texture_cache/surface_params.cpp13
-rw-r--r--src/video_core/texture_cache/texture_cache.h37
-rw-r--r--src/yuzu/configuration/config.cpp3
-rw-r--r--src/yuzu_cmd/config.cpp1
-rw-r--r--src/yuzu_cmd/default_ini.h4
-rw-r--r--src/yuzu_tester/config.cpp1
-rw-r--r--src/yuzu_tester/default_ini.h4
135 files changed, 3285 insertions, 867 deletions
diff --git a/.ci/scripts/common/post-upload.sh b/.ci/scripts/common/post-upload.sh
new file mode 100644
index 000000000..bb4e9d328
--- /dev/null
+++ b/.ci/scripts/common/post-upload.sh
@@ -0,0 +1,15 @@
+#!/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/
diff --git a/.ci/scripts/common/pre-upload.sh b/.ci/scripts/common/pre-upload.sh
new file mode 100644
index 000000000..3c2fc79a2
--- /dev/null
+++ b/.ci/scripts/common/pre-upload.sh
@@ -0,0 +1,6 @@
+#!/bin/bash -ex
+
+GITDATE="`git show -s --date=short --format='%ad' | sed 's/-//g'`"
+GITREV="`git show -s --format='%h'`"
+
+mkdir -p artifacts
diff --git a/.ci/scripts/format/docker.sh b/.ci/scripts/format/docker.sh
new file mode 100644
index 000000000..778411e4a
--- /dev/null
+++ b/.ci/scripts/format/docker.sh
@@ -0,0 +1,6 @@
+#!/bin/bash -ex
+
+# Run clang-format
+cd /yuzu
+chmod a+x ./.ci/scripts/format/script.sh
+./.ci/scripts/format/script.sh
diff --git a/.ci/scripts/format/exec.sh b/.ci/scripts/format/exec.sh
new file mode 100644
index 000000000..5d6393b38
--- /dev/null
+++ b/.ci/scripts/format/exec.sh
@@ -0,0 +1,4 @@
+#!/bin/bash -ex
+
+chmod a+x ./.ci/scripts/format/docker.sh
+docker run -v $(pwd):/yuzu yuzuemu/build-environments:linux-clang-format /bin/bash -ex /yuzu/.ci/scripts/format/docker.sh
diff --git a/.ci/scripts/format/script.sh b/.ci/scripts/format/script.sh
new file mode 100644
index 000000000..5ab828d5e
--- /dev/null
+++ b/.ci/scripts/format/script.sh
@@ -0,0 +1,37 @@
+#!/bin/bash -ex
+
+if grep -nrI '\s$' 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 [ "$fail" = 1 ]; then
+ exit 1
+fi
diff --git a/.ci/scripts/linux/docker.sh b/.ci/scripts/linux/docker.sh
new file mode 100644
index 000000000..f538a4081
--- /dev/null
+++ b/.ci/scripts/linux/docker.sh
@@ -0,0 +1,14 @@
+#!/bin/bash -ex
+
+cd /yuzu
+
+ccache -s
+
+mkdir build || true && cd build
+cmake .. -G Ninja -DYUZU_USE_BUNDLED_UNICORN=ON -DYUZU_USE_QT_WEB_ENGINE=ON -DCMAKE_BUILD_TYPE=Release -DCMAKE_C_COMPILER=/usr/lib/ccache/gcc -DCMAKE_CXX_COMPILER=/usr/lib/ccache/g++ -DYUZU_ENABLE_COMPATIBILITY_REPORTING=${ENABLE_COMPATIBILITY_REPORTING:-"OFF"} -DENABLE_COMPATIBILITY_LIST_DOWNLOAD=ON -DUSE_DISCORD_PRESENCE=ON
+
+ninja
+
+ccache -s
+
+ctest -VV -C Release
diff --git a/.ci/scripts/linux/exec.sh b/.ci/scripts/linux/exec.sh
new file mode 100644
index 000000000..a5a6c34b9
--- /dev/null
+++ b/.ci/scripts/linux/exec.sh
@@ -0,0 +1,5 @@
+#!/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 $(pwd):/yuzu yuzuemu/build-environments:linux-fresh /bin/bash /yuzu/.ci/scripts/linux/docker.sh
diff --git a/.ci/scripts/linux/upload.sh b/.ci/scripts/linux/upload.sh
new file mode 100644
index 000000000..0d131d1dd
--- /dev/null
+++ b/.ci/scripts/linux/upload.sh
@@ -0,0 +1,14 @@
+#!/bin/bash -ex
+
+. .ci/scripts/common/pre-upload.sh
+
+REV_NAME="yuzu-linux-${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
diff --git a/.ci/scripts/merge/apply-patches-by-label.py b/.ci/scripts/merge/apply-patches-by-label.py
new file mode 100644
index 000000000..b346001a5
--- /dev/null
+++ b/.ci/scripts/merge/apply-patches-by-label.py
@@ -0,0 +1,28 @@
+# 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)
diff --git a/.ci/scripts/merge/check-label-presence.py b/.ci/scripts/merge/check-label-presence.py
new file mode 100644
index 000000000..048466d7e
--- /dev/null
+++ b/.ci/scripts/merge/check-label-presence.py
@@ -0,0 +1,18 @@
+# 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')
diff --git a/.ci/scripts/merge/yuzubot-git-config.sh b/.ci/scripts/merge/yuzubot-git-config.sh
new file mode 100644
index 000000000..d9d595bbc
--- /dev/null
+++ b/.ci/scripts/merge/yuzubot-git-config.sh
@@ -0,0 +1,2 @@
+git config --global user.email "yuzu@yuzu-emu.org"
+git config --global user.name "yuzubot" \ No newline at end of file
diff --git a/.ci/scripts/windows/docker.sh b/.ci/scripts/windows/docker.sh
new file mode 100644
index 000000000..f7093363b
--- /dev/null
+++ b/.ci/scripts/windows/docker.sh
@@ -0,0 +1,50 @@
+#!/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/"
diff --git a/.ci/scripts/windows/exec.sh b/.ci/scripts/windows/exec.sh
new file mode 100644
index 000000000..d6a994856
--- /dev/null
+++ b/.ci/scripts/windows/exec.sh
@@ -0,0 +1,5 @@
+#!/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
diff --git a/.ci/scripts/windows/scan_dll.py b/.ci/scripts/windows/scan_dll.py
new file mode 100644
index 000000000..163183f2e
--- /dev/null
+++ b/.ci/scripts/windows/scan_dll.py
@@ -0,0 +1,106 @@
+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()
diff --git a/.ci/scripts/windows/upload.sh b/.ci/scripts/windows/upload.sh
new file mode 100644
index 000000000..de73d3541
--- /dev/null
+++ b/.ci/scripts/windows/upload.sh
@@ -0,0 +1,13 @@
+#!/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
diff --git a/.ci/templates/build-single.yml b/.ci/templates/build-single.yml
new file mode 100644
index 000000000..77eeb96b5
--- /dev/null
+++ b/.ci/templates/build-single.yml
@@ -0,0 +1,21 @@
+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'
diff --git a/.ci/templates/build-standard.yml b/.ci/templates/build-standard.yml
new file mode 100644
index 000000000..9975f5c49
--- /dev/null
+++ b/.ci/templates/build-standard.yml
@@ -0,0 +1,22 @@
+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: $(parameters.artifactSource)
+ needSubmodules: 'true'
+ - template: ./build-single.yml
+ parameters:
+ artifactSource: 'false' \ No newline at end of file
diff --git a/.ci/templates/build-testing.yml b/.ci/templates/build-testing.yml
new file mode 100644
index 000000000..101e52996
--- /dev/null
+++ b/.ci/templates/build-testing.yml
@@ -0,0 +1,30 @@
+jobs:
+- job: build_test
+ displayName: 'testing'
+ pool:
+ vmImage: ubuntu-latest
+ strategy:
+ maxParallel: 10
+ matrix:
+ windows:
+ BuildSuffix: 'windows-testing'
+ ScriptFolder: 'windows'
+ steps:
+ - task: PythonScript@0
+ condition: eq(variables['Build.Reason'], 'PullRequest')
+ displayName: 'Determine Testing Status'
+ inputs:
+ scriptSource: 'filePath'
+ scriptPath: '../scripts/merge/check-label-presence.py'
+ arguments: '$(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' \ No newline at end of file
diff --git a/.ci/templates/format-check.yml b/.ci/templates/format-check.yml
new file mode 100644
index 000000000..5061f1cb8
--- /dev/null
+++ b/.ci/templates/format-check.yml
@@ -0,0 +1,14 @@
+parameters:
+ artifactSource: 'true'
+
+steps:
+- template: ./sync-source.yml
+ parameters:
+ artifactSource: $(parameters.artifactSource)
+ needSubmodules: 'false'
+- task: DockerInstaller@0
+ displayName: 'Prepare Environment'
+ inputs:
+ dockerVersion: '17.09.0-ce'
+- script: chmod a+x ./.ci/scripts/format/exec.sh && ./.ci/scripts/format/exec.sh
+ displayName: 'Verify Formatting'
diff --git a/.ci/templates/merge.yml b/.ci/templates/merge.yml
new file mode 100644
index 000000000..efc82778a
--- /dev/null
+++ b/.ci/templates/merge.yml
@@ -0,0 +1,46 @@
+jobs:
+- job: merge
+ displayName: 'pull requests'
+ steps:
+ - checkout: self
+ submodules: recursive
+ - template: ./mergebot.yml
+ parameters:
+ matchLabel: '$(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 \ No newline at end of file
diff --git a/.ci/templates/mergebot.yml b/.ci/templates/mergebot.yml
new file mode 100644
index 000000000..5211efcc6
--- /dev/null
+++ b/.ci/templates/mergebot.yml
@@ -0,0 +1,15 @@
+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: '$(System.DefaultWorkingDirectory)'
diff --git a/.ci/templates/release.yml b/.ci/templates/release.yml
new file mode 100644
index 000000000..60bebd2aa
--- /dev/null
+++ b/.ci/templates/release.yml
@@ -0,0 +1,29 @@
+steps:
+ - task: DownloadPipelineArtifact@2
+ displayName: 'Download Windows Release'
+ inputs:
+ artifactName: 'yuzu-$(BuildName)-windows-mingw'
+ buildType: 'current'
+ targetPath: '$(Build.ArtifactStagingDirectory)'
+ - task: DownloadPipelineArtifact@2
+ displayName: 'Download Linux Release'
+ inputs:
+ artifactName: 'yuzu-$(BuildName)-linux'
+ buildType: 'current'
+ targetPath: '$(Build.ArtifactStagingDirectory)'
+ - task: DownloadPipelineArtifact@2
+ displayName: 'Download Release Point'
+ inputs:
+ artifactName: 'yuzu-$(BuildName)-release-point'
+ buildType: 'current'
+ targetPath: '$(Build.ArtifactStagingDirectory)'
+ - script: echo '##vso[task.setvariable variable=tagcommit]' && cat $(Build.ArtifactStagingDirectory)/tag-commit.sha
+ displayName: 'Calculate Release Point'
+ - task: GitHubRelease@0
+ inputs:
+ gitHubConnection: $(GitHubReleaseConnectionName)
+ repositoryName: '$(GitHubReleaseRepoName)'
+ action: 'create'
+ target: $(variables.tagcommit)
+ title: 'yuzu $(BuildName) #$(Build.BuildId)'
+ assets: '$(Build.ArtifactStagingDirectory)/*'
diff --git a/.ci/templates/retrieve-artifact-source.yml b/.ci/templates/retrieve-artifact-source.yml
new file mode 100644
index 000000000..47d217e7b
--- /dev/null
+++ b/.ci/templates/retrieve-artifact-source.yml
@@ -0,0 +1,16 @@
+steps:
+- checkout: none
+- task: DownloadPipelineArtifact@2
+ displayName: 'Download Source'
+ inputs:
+ artifactName: 'yuzu-$(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 \ No newline at end of file
diff --git a/.ci/templates/retrieve-master-source.yml b/.ci/templates/retrieve-master-source.yml
new file mode 100644
index 000000000..a08a3f926
--- /dev/null
+++ b/.ci/templates/retrieve-master-source.yml
@@ -0,0 +1,11 @@
+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')
diff --git a/.ci/templates/sync-source.yml b/.ci/templates/sync-source.yml
new file mode 100644
index 000000000..409e1cd83
--- /dev/null
+++ b/.ci/templates/sync-source.yml
@@ -0,0 +1,7 @@
+steps:
+- ${{ if eq(parameters.artifactSource, 'true') }}:
+ - template: ./retrieve-artifact-source.yml
+- ${{ if ne(parameters.artifactSource, 'true') }}:
+ - template: ./retrieve-master-source.yml
+ parameters:
+ needSubmodules: $(parameters.needSubmodules) \ No newline at end of file
diff --git a/.ci/yuzu-mainline.yml b/.ci/yuzu-mainline.yml
new file mode 100644
index 000000000..164bcb165
--- /dev/null
+++ b/.ci/yuzu-mainline.yml
@@ -0,0 +1,23 @@
+trigger:
+- master
+
+stages:
+- stage: merge
+ displayName: 'merge'
+ jobs:
+ - template: ./templates/merge.yml
+- stage: format
+ dependsOn: merge
+ displayName: 'format'
+ jobs:
+ - job: format
+ displayName: 'clang'
+ pool:
+ vmImage: ubuntu-latest
+ steps:
+ - template: ./templates/format-check.yml
+- stage: build
+ displayName: 'build'
+ dependsOn: format
+ jobs:
+ - template: ./templates/build-standard.yml
diff --git a/.ci/yuzu-patreon.yml b/.ci/yuzu-patreon.yml
new file mode 100644
index 000000000..aa912913d
--- /dev/null
+++ b/.ci/yuzu-patreon.yml
@@ -0,0 +1,19 @@
+# Starter pipeline
+# Start with a minimal pipeline that you can customize to build and deploy your code.
+# Add steps that build, run tests, deploy, and more:
+# https://aka.ms/yaml
+
+trigger:
+- master
+
+pool:
+ vmImage: 'ubuntu-latest'
+
+steps:
+- script: echo Hello, world!
+ displayName: 'Run a one-line script'
+
+- script: |
+ echo Add other tasks to build, test, and deploy your project.
+ echo See https://aka.ms/yaml
+ displayName: 'Run a multi-line script'
diff --git a/.ci/yuzu-repo-sync.yml b/.ci/yuzu-repo-sync.yml
new file mode 100644
index 000000000..602e298a6
--- /dev/null
+++ b/.ci/yuzu-repo-sync.yml
@@ -0,0 +1,19 @@
+trigger:
+- master
+
+jobs:
+- job: copy
+ displayName: 'Sync Repository'
+ pool:
+ vmImage: 'ubuntu-latest'
+ steps:
+ - script: echo 'https://$(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 $HOME/.git-credentials
+ displayName: 'Clear Cached Credentials'
diff --git a/.ci/yuzu-verify.yml b/.ci/yuzu-verify.yml
new file mode 100644
index 000000000..d01c1feed
--- /dev/null
+++ b/.ci/yuzu-verify.yml
@@ -0,0 +1,18 @@
+stages:
+- stage: format
+ displayName: 'format'
+ jobs:
+ - job: format
+ displayName: 'clang'
+ pool:
+ vmImage: ubuntu-latest
+ steps:
+ - template: ./templates/format-check.yml
+ parameters:
+ artifactSource: 'false'
+- stage: build
+ displayName: 'build'
+ dependsOn: format
+ jobs:
+ - template: ./templates/build-standard.yml
+ - template: ./templates/build-testing.yml \ No newline at end of file
diff --git a/CMakeModules/GenerateSCMRev.cmake b/CMakeModules/GenerateSCMRev.cmake
index dd65cfe42..abdc74428 100644
--- a/CMakeModules/GenerateSCMRev.cmake
+++ b/CMakeModules/GenerateSCMRev.cmake
@@ -82,6 +82,8 @@ set(HASH_FILES
"${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"
diff --git a/src/audio_core/audio_renderer.cpp b/src/audio_core/audio_renderer.cpp
index 9a0939883..da50a0bbc 100644
--- a/src/audio_core/audio_renderer.cpp
+++ b/src/audio_core/audio_renderer.cpp
@@ -73,13 +73,15 @@ private:
EffectInStatus info{};
};
AudioRenderer::AudioRenderer(Core::Timing::CoreTiming& core_timing, AudioRendererParameter params,
- Kernel::SharedPtr<Kernel::WritableEvent> buffer_event)
+ Kernel::SharedPtr<Kernel::WritableEvent> buffer_event,
+ std::size_t instance_number)
: 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,
- "AudioRenderer", [=]() { buffer_event->Signal(); });
+ fmt::format("AudioRenderer-Instance{}", instance_number),
+ [=]() { buffer_event->Signal(); });
audio_out->StartStream(stream);
QueueMixedBuffer(0);
@@ -217,13 +219,15 @@ std::vector<s16> AudioRenderer::VoiceState::DequeueSamples(std::size_t sample_co
if (offset == samples.size()) {
offset = 0;
- if (!wave_buffer.is_looping) {
+ if (!wave_buffer.is_looping && wave_buffer.buffer_sz) {
SetWaveIndex(wave_index + 1);
}
- out_status.wave_buffer_consumed++;
+ if (wave_buffer.buffer_sz) {
+ out_status.wave_buffer_consumed++;
+ }
- if (wave_buffer.end_of_stream) {
+ if (wave_buffer.end_of_stream || wave_buffer.buffer_sz == 0) {
info.play_state = PlayState::Paused;
}
}
diff --git a/src/audio_core/audio_renderer.h b/src/audio_core/audio_renderer.h
index b2e5d336c..45afbe759 100644
--- a/src/audio_core/audio_renderer.h
+++ b/src/audio_core/audio_renderer.h
@@ -215,7 +215,8 @@ 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);
+ Kernel::SharedPtr<Kernel::WritableEvent> buffer_event,
+ std::size_t instance_number);
~AudioRenderer();
std::vector<u8> UpdateAudioRenderer(const std::vector<u8>& input_params);
diff --git a/src/common/CMakeLists.txt b/src/common/CMakeLists.txt
index 2554add28..2b4266f29 100644
--- a/src/common/CMakeLists.txt
+++ b/src/common/CMakeLists.txt
@@ -56,6 +56,8 @@ 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"
diff --git a/src/core/CMakeLists.txt b/src/core/CMakeLists.txt
index 30eb9d82e..f4325f0f8 100644
--- a/src/core/CMakeLists.txt
+++ b/src/core/CMakeLists.txt
@@ -208,6 +208,8 @@ add_library(core STATIC
hle/service/aoc/aoc_u.h
hle/service/apm/apm.cpp
hle/service/apm/apm.h
+ hle/service/apm/controller.cpp
+ hle/service/apm/controller.h
hle/service/apm/interface.cpp
hle/service/apm/interface.h
hle/service/audio/audctl.cpp
@@ -293,6 +295,7 @@ add_library(core STATIC
hle/service/hid/irs.h
hle/service/hid/xcd.cpp
hle/service/hid/xcd.h
+ hle/service/hid/errors.h
hle/service/hid/controllers/controller_base.cpp
hle/service/hid/controllers/controller_base.h
hle/service/hid/controllers/debug_pad.cpp
diff --git a/src/core/arm/arm_interface.h b/src/core/arm/arm_interface.h
index c6691a8e1..45e94e625 100644
--- a/src/core/arm/arm_interface.h
+++ b/src/core/arm/arm_interface.h
@@ -44,13 +44,6 @@ 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;
diff --git a/src/core/arm/dynarmic/arm_dynarmic.cpp b/src/core/arm/dynarmic/arm_dynarmic.cpp
index 44307fa19..f1506b372 100644
--- a/src/core/arm/dynarmic/arm_dynarmic.cpp
+++ b/src/core/arm/dynarmic/arm_dynarmic.cpp
@@ -177,15 +177,6 @@ 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);
}
diff --git a/src/core/arm/dynarmic/arm_dynarmic.h b/src/core/arm/dynarmic/arm_dynarmic.h
index b701e97a3..504d46c68 100644
--- a/src/core/arm/dynarmic/arm_dynarmic.h
+++ b/src/core/arm/dynarmic/arm_dynarmic.h
@@ -23,9 +23,6 @@ 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;
diff --git a/src/core/arm/unicorn/arm_unicorn.cpp b/src/core/arm/unicorn/arm_unicorn.cpp
index 4e07fe8b5..b0ee7821a 100644
--- a/src/core/arm/unicorn/arm_unicorn.cpp
+++ b/src/core/arm/unicorn/arm_unicorn.cpp
@@ -76,15 +76,6 @@ 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));
}
diff --git a/src/core/arm/unicorn/arm_unicorn.h b/src/core/arm/unicorn/arm_unicorn.h
index 34e974b4d..fe2ffd70c 100644
--- a/src/core/arm/unicorn/arm_unicorn.h
+++ b/src/core/arm/unicorn/arm_unicorn.h
@@ -18,9 +18,6 @@ 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;
diff --git a/src/core/core.cpp b/src/core/core.cpp
index 262411db8..4aceee785 100644
--- a/src/core/core.cpp
+++ b/src/core/core.cpp
@@ -25,6 +25,7 @@
#include "core/hle/kernel/scheduler.h"
#include "core/hle/kernel/thread.h"
#include "core/hle/service/am/applets/applets.h"
+#include "core/hle/service/apm/controller.h"
#include "core/hle/service/glue/manager.h"
#include "core/hle/service/service.h"
#include "core/hle/service/sm/sm.h"
@@ -143,7 +144,7 @@ struct System::Impl {
telemetry_session = std::make_unique<Core::TelemetrySession>();
service_manager = std::make_shared<Service::SM::ServiceManager>();
- Service::Init(service_manager, system, *virtual_filesystem);
+ Service::Init(service_manager, system);
GDBStub::Init();
renderer = VideoCore::CreateRenderer(emu_window, system);
@@ -306,6 +307,9 @@ struct System::Impl {
/// Frontend applets
Service::AM::Applets::AppletManager applet_manager;
+ /// APM (Performance) services
+ Service::APM::Controller apm_controller{core_timing};
+
/// Glue services
Service::Glue::ARPManager arp_manager;
@@ -568,6 +572,14 @@ const Service::Glue::ARPManager& System::GetARPManager() const {
return impl->arp_manager;
}
+Service::APM::Controller& System::GetAPMController() {
+ return impl->apm_controller;
+}
+
+const Service::APM::Controller& System::GetAPMController() const {
+ return impl->apm_controller;
+}
+
System::ResultStatus System::Init(Frontend::EmuWindow& emu_window) {
return impl->Init(*this, emu_window);
}
diff --git a/src/core/core.h b/src/core/core.h
index 70adb7af9..11e73278e 100644
--- a/src/core/core.h
+++ b/src/core/core.h
@@ -43,6 +43,10 @@ struct AppletFrontendSet;
class AppletManager;
} // namespace AM::Applets
+namespace APM {
+class Controller;
+}
+
namespace Glue {
class ARPManager;
}
@@ -296,6 +300,10 @@ public:
const Service::Glue::ARPManager& GetARPManager() const;
+ Service::APM::Controller& GetAPMController();
+
+ const Service::APM::Controller& GetAPMController() const;
+
private:
System();
diff --git a/src/core/core_cpu.cpp b/src/core/core_cpu.cpp
index 99b7d387d..21c410e34 100644
--- a/src/core/core_cpu.cpp
+++ b/src/core/core_cpu.cpp
@@ -53,16 +53,12 @@ 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);
}
@@ -70,15 +66,12 @@ 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
- return nullptr; // TODO(merry): Passthrough exclusive monitor
+ // TODO(merry): Passthrough exclusive monitor
+ return nullptr;
#endif
- } else {
- return nullptr; // TODO(merry): Passthrough exclusive monitor
- }
}
void Cpu::RunLoop(bool tight_loop) {
diff --git a/src/core/file_sys/program_metadata.cpp b/src/core/file_sys/program_metadata.cpp
index eb76174c5..7310b3602 100644
--- a/src/core/file_sys/program_metadata.cpp
+++ b/src/core/file_sys/program_metadata.cpp
@@ -94,6 +94,10 @@ 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;
}
diff --git a/src/core/file_sys/program_metadata.h b/src/core/file_sys/program_metadata.h
index 43bf2820a..88ec97d85 100644
--- a/src/core/file_sys/program_metadata.h
+++ b/src/core/file_sys/program_metadata.h
@@ -58,6 +58,7 @@ public:
u32 GetMainThreadStackSize() const;
u64 GetTitleID() const;
u64 GetFilesystemPermissions() const;
+ u32 GetSystemResourceSize() const;
const KernelCapabilityDescriptors& GetKernelCapabilities() const;
void Print() const;
@@ -76,7 +77,8 @@ private:
u8 reserved_3;
u8 main_thread_priority;
u8 main_thread_cpu;
- std::array<u8, 8> reserved_4;
+ std::array<u8, 4> reserved_4;
+ u32_le system_resource_size;
u32_le process_category;
u32_le main_stack_size;
std::array<u8, 0x10> application_name;
diff --git a/src/core/hle/kernel/process.cpp b/src/core/hle/kernel/process.cpp
index f45ef05f6..db3ab14ce 100644
--- a/src/core/hle/kernel/process.cpp
+++ b/src/core/hle/kernel/process.cpp
@@ -129,20 +129,17 @@ u64 Process::GetTotalPhysicalMemoryAvailable() const {
return vm_manager.GetTotalPhysicalMemoryAvailable();
}
-u64 Process::GetTotalPhysicalMemoryAvailableWithoutMmHeap() const {
- // TODO: Subtract the personal heap size from this when the
- // personal heap is implemented.
- return GetTotalPhysicalMemoryAvailable();
+u64 Process::GetTotalPhysicalMemoryAvailableWithoutSystemResource() const {
+ return GetTotalPhysicalMemoryAvailable() - GetSystemResourceSize();
}
u64 Process::GetTotalPhysicalMemoryUsed() const {
- return vm_manager.GetCurrentHeapSize() + main_thread_stack_size + code_memory_size;
+ return vm_manager.GetCurrentHeapSize() + main_thread_stack_size + code_memory_size +
+ GetSystemResourceUsage();
}
-u64 Process::GetTotalPhysicalMemoryUsedWithoutMmHeap() const {
- // TODO: Subtract the personal heap size from this when the
- // personal heap is implemented.
- return GetTotalPhysicalMemoryUsed();
+u64 Process::GetTotalPhysicalMemoryUsedWithoutSystemResource() const {
+ return GetTotalPhysicalMemoryUsed() - GetSystemResourceUsage();
}
void Process::RegisterThread(const Thread* thread) {
@@ -172,6 +169,7 @@ 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());
diff --git a/src/core/hle/kernel/process.h b/src/core/hle/kernel/process.h
index 83ea02bee..3196014da 100644
--- a/src/core/hle/kernel/process.h
+++ b/src/core/hle/kernel/process.h
@@ -168,8 +168,24 @@ public:
return capabilities.GetPriorityMask();
}
- u32 IsVirtualMemoryEnabled() const {
- return is_virtual_address_memory_enabled;
+ /// 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;
}
/// Whether this process is an AArch64 or AArch32 process.
@@ -196,15 +212,15 @@ public:
u64 GetTotalPhysicalMemoryAvailable() const;
/// Retrieves the total physical memory available to this process in bytes,
- /// without the size of the personal heap added to it.
- u64 GetTotalPhysicalMemoryAvailableWithoutMmHeap() const;
+ /// without the size of the personal system resource heap added to it.
+ u64 GetTotalPhysicalMemoryAvailableWithoutSystemResource() 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 heap added to it.
- u64 GetTotalPhysicalMemoryUsedWithoutMmHeap() const;
+ /// without the size of the personal system resource heap added to it.
+ u64 GetTotalPhysicalMemoryUsedWithoutSystemResource() const;
/// Gets the list of all threads created with this process as their owner.
const std::list<const Thread*>& GetThreadList() const {
@@ -298,12 +314,16 @@ 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
diff --git a/src/core/hle/kernel/svc.cpp b/src/core/hle/kernel/svc.cpp
index 332573a95..a46eed3da 100644
--- a/src/core/hle/kernel/svc.cpp
+++ b/src/core/hle/kernel/svc.cpp
@@ -318,7 +318,14 @@ static ResultCode UnmapMemory(Core::System& system, VAddr dst_addr, VAddr src_ad
return result;
}
- return vm_manager.UnmapRange(dst_addr, size);
+ 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;
}
/// Connect to an OS service given the port name, returns the handle to the port to out
@@ -729,16 +736,16 @@ static ResultCode GetInfo(Core::System& system, u64* result, u64 info_id, u64 ha
StackRegionBaseAddr = 14,
StackRegionSize = 15,
// 3.0.0+
- IsVirtualAddressMemoryEnabled = 16,
- PersonalMmHeapUsage = 17,
+ SystemResourceSize = 16,
+ SystemResourceUsage = 17,
TitleId = 18,
// 4.0.0+
PrivilegedProcessId = 19,
// 5.0.0+
UserExceptionContextAddr = 20,
// 6.0.0+
- TotalPhysicalMemoryAvailableWithoutMmHeap = 21,
- TotalPhysicalMemoryUsedWithoutMmHeap = 22,
+ TotalPhysicalMemoryAvailableWithoutSystemResource = 21,
+ TotalPhysicalMemoryUsedWithoutSystemResource = 22,
};
const auto info_id_type = static_cast<GetInfoType>(info_id);
@@ -756,12 +763,12 @@ static ResultCode GetInfo(Core::System& system, u64* result, u64 info_id, u64 ha
case GetInfoType::StackRegionSize:
case GetInfoType::TotalPhysicalMemoryAvailable:
case GetInfoType::TotalPhysicalMemoryUsed:
- case GetInfoType::IsVirtualAddressMemoryEnabled:
- case GetInfoType::PersonalMmHeapUsage:
+ case GetInfoType::SystemResourceSize:
+ case GetInfoType::SystemResourceUsage:
case GetInfoType::TitleId:
case GetInfoType::UserExceptionContextAddr:
- case GetInfoType::TotalPhysicalMemoryAvailableWithoutMmHeap:
- case GetInfoType::TotalPhysicalMemoryUsedWithoutMmHeap: {
+ case GetInfoType::TotalPhysicalMemoryAvailableWithoutSystemResource:
+ case GetInfoType::TotalPhysicalMemoryUsedWithoutSystemResource: {
if (info_sub_id != 0) {
return ERR_INVALID_ENUM_VALUE;
}
@@ -822,8 +829,13 @@ static ResultCode GetInfo(Core::System& system, u64* result, u64 info_id, u64 ha
*result = process->GetTotalPhysicalMemoryUsed();
return RESULT_SUCCESS;
- case GetInfoType::IsVirtualAddressMemoryEnabled:
- *result = process->IsVirtualMemoryEnabled();
+ 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();
return RESULT_SUCCESS;
case GetInfoType::TitleId:
@@ -836,12 +848,12 @@ static ResultCode GetInfo(Core::System& system, u64* result, u64 info_id, u64 ha
*result = 0;
return RESULT_SUCCESS;
- case GetInfoType::TotalPhysicalMemoryAvailableWithoutMmHeap:
- *result = process->GetTotalPhysicalMemoryAvailable();
+ case GetInfoType::TotalPhysicalMemoryAvailableWithoutSystemResource:
+ *result = process->GetTotalPhysicalMemoryAvailableWithoutSystemResource();
return RESULT_SUCCESS;
- case GetInfoType::TotalPhysicalMemoryUsedWithoutMmHeap:
- *result = process->GetTotalPhysicalMemoryUsedWithoutMmHeap();
+ case GetInfoType::TotalPhysicalMemoryUsedWithoutSystemResource:
+ *result = process->GetTotalPhysicalMemoryUsedWithoutSystemResource();
return RESULT_SUCCESS;
default:
@@ -946,6 +958,86 @@ 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);
@@ -2303,8 +2395,8 @@ static const FunctionDef SVC_Table[] = {
{0x29, SvcWrap<GetInfo>, "GetInfo"},
{0x2A, nullptr, "FlushEntireDataCache"},
{0x2B, nullptr, "FlushDataCache"},
- {0x2C, nullptr, "MapPhysicalMemory"},
- {0x2D, nullptr, "UnmapPhysicalMemory"},
+ {0x2C, SvcWrap<MapPhysicalMemory>, "MapPhysicalMemory"},
+ {0x2D, SvcWrap<UnmapPhysicalMemory>, "UnmapPhysicalMemory"},
{0x2E, nullptr, "GetFutureThreadInfo"},
{0x2F, nullptr, "GetLastThreadInfo"},
{0x30, SvcWrap<GetResourceLimitLimitValue>, "GetResourceLimitLimitValue"},
diff --git a/src/core/hle/kernel/svc_wrap.h b/src/core/hle/kernel/svc_wrap.h
index 865473c6f..c2d8d0dc3 100644
--- a/src/core/hle/kernel/svc_wrap.h
+++ b/src/core/hle/kernel/svc_wrap.h
@@ -32,6 +32,11 @@ 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);
diff --git a/src/core/hle/kernel/vm_manager.cpp b/src/core/hle/kernel/vm_manager.cpp
index 501544090..4f45fb03b 100644
--- a/src/core/hle/kernel/vm_manager.cpp
+++ b/src/core/hle/kernel/vm_manager.cpp
@@ -8,10 +8,11 @@
#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"
@@ -49,10 +50,14 @@ bool VirtualMemoryArea::CanBeMergedWith(const VirtualMemoryArea& next) const {
type != next.type) {
return false;
}
- if (type == VMAType::AllocatedMemoryBlock &&
- (backing_block != next.backing_block || offset + size != next.offset)) {
+ 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.
return false;
}
+ if (type == VMAType::AllocatedMemoryBlock) {
+ return true;
+ }
if (type == VMAType::BackingMemory && backing_memory + size != next.backing_memory) {
return false;
}
@@ -100,7 +105,7 @@ 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) {
+ MemoryState state, VMAPermission perm) {
ASSERT(block != nullptr);
ASSERT(offset + size <= block->size());
@@ -109,17 +114,8 @@ 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 = VMAPermission::ReadWrite;
+ final_vma.permissions = perm;
final_vma.state = state;
final_vma.backing_block = std::move(block);
final_vma.offset = offset;
@@ -137,11 +133,6 @@ 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;
@@ -230,11 +221,6 @@ 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;
}
@@ -308,6 +294,166 @@ 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(
@@ -455,7 +601,7 @@ 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
- Reprotect(vma, VMAPermission::None);
+ ReprotectRange(src_addr, size, VMAPermission::None);
return RESULT_SUCCESS;
}
@@ -588,14 +734,14 @@ 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)) {
- iter->second.size += next_vma->second.size;
+ MergeAdjacentVMA(iter->second, next_vma->second);
vma_map.erase(next_vma);
}
if (iter != vma_map.begin()) {
VMAIter prev_vma = std::prev(iter);
if (prev_vma->second.CanBeMergedWith(iter->second)) {
- prev_vma->second.size += iter->second.size;
+ MergeAdjacentVMA(prev_vma->second, iter->second);
vma_map.erase(iter);
iter = prev_vma;
}
@@ -604,6 +750,38 @@ 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:
@@ -778,6 +956,84 @@ 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;
diff --git a/src/core/hle/kernel/vm_manager.h b/src/core/hle/kernel/vm_manager.h
index 9fe6ac3f4..0aecb7499 100644
--- a/src/core/hle/kernel/vm_manager.h
+++ b/src/core/hle/kernel/vm_manager.h
@@ -349,7 +349,8 @@ 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);
+ std::size_t offset, u64 size, MemoryState state,
+ VMAPermission perm = VMAPermission::ReadWrite);
/**
* Maps an unmanaged host memory pointer at a given address.
@@ -450,6 +451,34 @@ 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.
@@ -657,6 +686,11 @@ 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);
@@ -701,6 +735,13 @@ 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
@@ -742,6 +783,11 @@ 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
diff --git a/src/core/hle/service/am/am.cpp b/src/core/hle/service/am/am.cpp
index eced38001..a192a1f5f 100644
--- a/src/core/hle/service/am/am.cpp
+++ b/src/core/hle/service/am/am.cpp
@@ -29,7 +29,8 @@
#include "core/hle/service/am/omm.h"
#include "core/hle/service/am/spsm.h"
#include "core/hle/service/am/tcap.h"
-#include "core/hle/service/apm/apm.h"
+#include "core/hle/service/apm/controller.h"
+#include "core/hle/service/apm/interface.h"
#include "core/hle/service/filesystem/filesystem.h"
#include "core/hle/service/ns/ns.h"
#include "core/hle/service/nvflinger/nvflinger.h"
@@ -265,8 +266,8 @@ ISelfController::ISelfController(std::shared_ptr<NVFlinger::NVFlinger> nvflinger
{65, nullptr, "ReportUserIsActive"},
{66, nullptr, "GetCurrentIlluminance"},
{67, nullptr, "IsIlluminanceAvailable"},
- {68, nullptr, "SetAutoSleepDisabled"},
- {69, nullptr, "IsAutoSleepDisabled"},
+ {68, &ISelfController::SetAutoSleepDisabled, "SetAutoSleepDisabled"},
+ {69, &ISelfController::IsAutoSleepDisabled, "IsAutoSleepDisabled"},
{70, nullptr, "ReportMultimediaError"},
{71, nullptr, "GetCurrentIlluminanceEx"},
{80, nullptr, "SetWirelessPriorityMode"},
@@ -453,6 +454,34 @@ 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.");
@@ -520,8 +549,9 @@ void AppletMessageQueue::OperationModeChanged() {
on_operation_mode_changed.writable->Signal();
}
-ICommonStateGetter::ICommonStateGetter(std::shared_ptr<AppletMessageQueue> msg_queue)
- : ServiceFramework("ICommonStateGetter"), msg_queue(std::move(msg_queue)) {
+ICommonStateGetter::ICommonStateGetter(Core::System& system,
+ std::shared_ptr<AppletMessageQueue> msg_queue)
+ : ServiceFramework("ICommonStateGetter"), system(system), msg_queue(std::move(msg_queue)) {
// clang-format off
static const FunctionInfo functions[] = {
{0, &ICommonStateGetter::GetEventHandle, "GetEventHandle"},
@@ -554,7 +584,7 @@ ICommonStateGetter::ICommonStateGetter(std::shared_ptr<AppletMessageQueue> msg_q
{63, nullptr, "GetHdcpAuthenticationStateChangeEvent"},
{64, nullptr, "SetTvPowerStateMatchingMode"},
{65, nullptr, "GetApplicationIdByContentActionName"},
- {66, nullptr, "SetCpuBoostMode"},
+ {66, &ICommonStateGetter::SetCpuBoostMode, "SetCpuBoostMode"},
{80, nullptr, "PerformSystemButtonPressingIfInFocus"},
{90, nullptr, "SetPerformanceConfigurationChangedNotification"},
{91, nullptr, "GetCurrentPerformanceConfiguration"},
@@ -635,6 +665,16 @@ void ICommonStateGetter::GetDefaultDisplayResolution(Kernel::HLERequestContext&
}
}
+void ICommonStateGetter::SetCpuBoostMode(Kernel::HLERequestContext& ctx) {
+ LOG_DEBUG(Service_AM, "called, forwarding to APM:SYS");
+
+ const auto& sm = system.ServiceManager();
+ const auto apm_sys = sm.GetService<APM::APM_Sys>("apm:sys");
+ ASSERT(apm_sys != nullptr);
+
+ apm_sys->SetCpuBoostMode(ctx);
+}
+
IStorage::IStorage(std::vector<u8> buffer)
: ServiceFramework("IStorage"), buffer(std::move(buffer)) {
// clang-format off
@@ -663,13 +703,11 @@ void ICommonStateGetter::GetOperationMode(Kernel::HLERequestContext& ctx) {
}
void ICommonStateGetter::GetPerformanceMode(Kernel::HLERequestContext& ctx) {
- const bool use_docked_mode{Settings::values.use_docked_mode};
- LOG_DEBUG(Service_AM, "called, use_docked_mode={}", use_docked_mode);
+ LOG_DEBUG(Service_AM, "called");
IPC::ResponseBuilder rb{ctx, 3};
rb.Push(RESULT_SUCCESS);
- rb.Push(static_cast<u32>(use_docked_mode ? APM::PerformanceMode::Docked
- : APM::PerformanceMode::Handheld));
+ rb.PushEnum(system.GetAPMController().GetCurrentPerformanceMode());
}
class ILibraryAppletAccessor final : public ServiceFramework<ILibraryAppletAccessor> {
diff --git a/src/core/hle/service/am/am.h b/src/core/hle/service/am/am.h
index 49ff20959..6cb582483 100644
--- a/src/core/hle/service/am/am.h
+++ b/src/core/hle/service/am/am.h
@@ -133,6 +133,8 @@ 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);
@@ -142,11 +144,13 @@ 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> {
public:
- explicit ICommonStateGetter(std::shared_ptr<AppletMessageQueue> msg_queue);
+ explicit ICommonStateGetter(Core::System& system,
+ std::shared_ptr<AppletMessageQueue> msg_queue);
~ICommonStateGetter() override;
private:
@@ -168,7 +172,9 @@ private:
void GetPerformanceMode(Kernel::HLERequestContext& ctx);
void GetBootMode(Kernel::HLERequestContext& ctx);
void GetDefaultDisplayResolution(Kernel::HLERequestContext& ctx);
+ void SetCpuBoostMode(Kernel::HLERequestContext& ctx);
+ Core::System& system;
std::shared_ptr<AppletMessageQueue> msg_queue;
};
diff --git a/src/core/hle/service/am/applet_ae.cpp b/src/core/hle/service/am/applet_ae.cpp
index fe5beb8f9..a34368c8b 100644
--- a/src/core/hle/service/am/applet_ae.cpp
+++ b/src/core/hle/service/am/applet_ae.cpp
@@ -42,7 +42,7 @@ private:
IPC::ResponseBuilder rb{ctx, 2, 0, 1};
rb.Push(RESULT_SUCCESS);
- rb.PushIpcInterface<ICommonStateGetter>(msg_queue);
+ rb.PushIpcInterface<ICommonStateGetter>(system, msg_queue);
}
void GetSelfController(Kernel::HLERequestContext& ctx) {
@@ -146,7 +146,7 @@ private:
IPC::ResponseBuilder rb{ctx, 2, 0, 1};
rb.Push(RESULT_SUCCESS);
- rb.PushIpcInterface<ICommonStateGetter>(msg_queue);
+ rb.PushIpcInterface<ICommonStateGetter>(system, msg_queue);
}
void GetSelfController(Kernel::HLERequestContext& ctx) {
diff --git a/src/core/hle/service/am/applet_oe.cpp b/src/core/hle/service/am/applet_oe.cpp
index 6e255fe95..5d53ef113 100644
--- a/src/core/hle/service/am/applet_oe.cpp
+++ b/src/core/hle/service/am/applet_oe.cpp
@@ -80,7 +80,7 @@ private:
IPC::ResponseBuilder rb{ctx, 2, 0, 1};
rb.Push(RESULT_SUCCESS);
- rb.PushIpcInterface<ICommonStateGetter>(msg_queue);
+ rb.PushIpcInterface<ICommonStateGetter>(system, msg_queue);
}
void GetLibraryAppletCreator(Kernel::HLERequestContext& ctx) {
diff --git a/src/core/hle/service/apm/apm.cpp b/src/core/hle/service/apm/apm.cpp
index f3c09bbb1..85bbf5988 100644
--- a/src/core/hle/service/apm/apm.cpp
+++ b/src/core/hle/service/apm/apm.cpp
@@ -2,7 +2,6 @@
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
-#include "common/logging/log.h"
#include "core/hle/ipc_helpers.h"
#include "core/hle/service/apm/apm.h"
#include "core/hle/service/apm/interface.h"
@@ -12,11 +11,15 @@ namespace Service::APM {
Module::Module() = default;
Module::~Module() = default;
-void InstallInterfaces(SM::ServiceManager& service_manager) {
+void InstallInterfaces(Core::System& system) {
auto module_ = std::make_shared<Module>();
- std::make_shared<APM>(module_, "apm")->InstallAsService(service_manager);
- std::make_shared<APM>(module_, "apm:p")->InstallAsService(service_manager);
- std::make_shared<APM_Sys>()->InstallAsService(service_manager);
+ std::make_shared<APM>(module_, system.GetAPMController(), "apm")
+ ->InstallAsService(system.ServiceManager());
+ std::make_shared<APM>(module_, system.GetAPMController(), "apm:p")
+ ->InstallAsService(system.ServiceManager());
+ std::make_shared<APM>(module_, system.GetAPMController(), "apm:am")
+ ->InstallAsService(system.ServiceManager());
+ std::make_shared<APM_Sys>(system.GetAPMController())->InstallAsService(system.ServiceManager());
}
} // namespace Service::APM
diff --git a/src/core/hle/service/apm/apm.h b/src/core/hle/service/apm/apm.h
index 4d7d5bb7c..cf4c2bb11 100644
--- a/src/core/hle/service/apm/apm.h
+++ b/src/core/hle/service/apm/apm.h
@@ -8,11 +8,6 @@
namespace Service::APM {
-enum class PerformanceMode : u8 {
- Handheld = 0,
- Docked = 1,
-};
-
class Module final {
public:
Module();
@@ -20,6 +15,6 @@ public:
};
/// Registers all AM services with the specified service manager.
-void InstallInterfaces(SM::ServiceManager& service_manager);
+void InstallInterfaces(Core::System& system);
} // namespace Service::APM
diff --git a/src/core/hle/service/apm/controller.cpp b/src/core/hle/service/apm/controller.cpp
new file mode 100644
index 000000000..4376612eb
--- /dev/null
+++ b/src/core/hle/service/apm/controller.cpp
@@ -0,0 +1,68 @@
+// Copyright 2019 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "common/logging/log.h"
+#include "core/core_timing.h"
+#include "core/hle/service/apm/controller.h"
+#include "core/settings.h"
+
+namespace Service::APM {
+
+constexpr PerformanceConfiguration DEFAULT_PERFORMANCE_CONFIGURATION =
+ PerformanceConfiguration::Config7;
+
+Controller::Controller(Core::Timing::CoreTiming& core_timing)
+ : core_timing(core_timing), configs{
+ {PerformanceMode::Handheld, DEFAULT_PERFORMANCE_CONFIGURATION},
+ {PerformanceMode::Docked, DEFAULT_PERFORMANCE_CONFIGURATION},
+ } {}
+
+Controller::~Controller() = default;
+
+void Controller::SetPerformanceConfiguration(PerformanceMode mode,
+ PerformanceConfiguration config) {
+ static const std::map<PerformanceConfiguration, u32> PCONFIG_TO_SPEED_MAP{
+ {PerformanceConfiguration::Config1, 1020}, {PerformanceConfiguration::Config2, 1020},
+ {PerformanceConfiguration::Config3, 1224}, {PerformanceConfiguration::Config4, 1020},
+ {PerformanceConfiguration::Config5, 1020}, {PerformanceConfiguration::Config6, 1224},
+ {PerformanceConfiguration::Config7, 1020}, {PerformanceConfiguration::Config8, 1020},
+ {PerformanceConfiguration::Config9, 1020}, {PerformanceConfiguration::Config10, 1020},
+ {PerformanceConfiguration::Config11, 1020}, {PerformanceConfiguration::Config12, 1020},
+ {PerformanceConfiguration::Config13, 1785}, {PerformanceConfiguration::Config14, 1785},
+ {PerformanceConfiguration::Config15, 1020}, {PerformanceConfiguration::Config16, 1020},
+ };
+
+ SetClockSpeed(PCONFIG_TO_SPEED_MAP.find(config)->second);
+ configs.insert_or_assign(mode, config);
+}
+
+void Controller::SetFromCpuBoostMode(CpuBoostMode mode) {
+ constexpr std::array<PerformanceConfiguration, 3> BOOST_MODE_TO_CONFIG_MAP{{
+ PerformanceConfiguration::Config7,
+ PerformanceConfiguration::Config13,
+ PerformanceConfiguration::Config15,
+ }};
+
+ SetPerformanceConfiguration(PerformanceMode::Docked,
+ BOOST_MODE_TO_CONFIG_MAP.at(static_cast<u32>(mode)));
+}
+
+PerformanceMode Controller::GetCurrentPerformanceMode() {
+ return Settings::values.use_docked_mode ? PerformanceMode::Docked : PerformanceMode::Handheld;
+}
+
+PerformanceConfiguration Controller::GetCurrentPerformanceConfiguration(PerformanceMode mode) {
+ if (configs.find(mode) == configs.end()) {
+ configs.insert_or_assign(mode, DEFAULT_PERFORMANCE_CONFIGURATION);
+ }
+
+ return configs[mode];
+}
+
+void Controller::SetClockSpeed(u32 mhz) {
+ LOG_INFO(Service_APM, "called, mhz={:08X}", mhz);
+ // TODO(DarkLordZach): Actually signal core_timing to change clock speed.
+}
+
+} // namespace Service::APM
diff --git a/src/core/hle/service/apm/controller.h b/src/core/hle/service/apm/controller.h
new file mode 100644
index 000000000..8ac80eaea
--- /dev/null
+++ b/src/core/hle/service/apm/controller.h
@@ -0,0 +1,70 @@
+// Copyright 2019 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <map>
+#include "common/common_types.h"
+
+namespace Core::Timing {
+class CoreTiming;
+}
+
+namespace Service::APM {
+
+enum class PerformanceConfiguration : u32 {
+ Config1 = 0x00010000,
+ Config2 = 0x00010001,
+ Config3 = 0x00010002,
+ Config4 = 0x00020000,
+ Config5 = 0x00020001,
+ Config6 = 0x00020002,
+ Config7 = 0x00020003,
+ Config8 = 0x00020004,
+ Config9 = 0x00020005,
+ Config10 = 0x00020006,
+ Config11 = 0x92220007,
+ Config12 = 0x92220008,
+ Config13 = 0x92220009,
+ Config14 = 0x9222000A,
+ Config15 = 0x9222000B,
+ Config16 = 0x9222000C,
+};
+
+enum class CpuBoostMode : u32 {
+ Disabled = 0,
+ Full = 1, // CPU + GPU -> Config 13, 14, 15, or 16
+ Partial = 2, // GPU Only -> Config 15 or 16
+};
+
+enum class PerformanceMode : u8 {
+ Handheld = 0,
+ Docked = 1,
+};
+
+// Class to manage the state and change of the emulated system performance.
+// Specifically, this deals with PerformanceMode, which corresponds to the system being docked or
+// undocked, and PerformanceConfig which specifies the exact CPU, GPU, and Memory clocks to operate
+// at. Additionally, this manages 'Boost Mode', which allows games to temporarily overclock the
+// system during times of high load -- this simply maps to different PerformanceConfigs to use.
+class Controller {
+public:
+ Controller(Core::Timing::CoreTiming& core_timing);
+ ~Controller();
+
+ void SetPerformanceConfiguration(PerformanceMode mode, PerformanceConfiguration config);
+ void SetFromCpuBoostMode(CpuBoostMode mode);
+
+ PerformanceMode GetCurrentPerformanceMode();
+ PerformanceConfiguration GetCurrentPerformanceConfiguration(PerformanceMode mode);
+
+private:
+ void SetClockSpeed(u32 mhz);
+
+ std::map<PerformanceMode, PerformanceConfiguration> configs;
+
+ Core::Timing::CoreTiming& core_timing;
+};
+
+} // namespace Service::APM
diff --git a/src/core/hle/service/apm/interface.cpp b/src/core/hle/service/apm/interface.cpp
index d058c0245..06f0f8edd 100644
--- a/src/core/hle/service/apm/interface.cpp
+++ b/src/core/hle/service/apm/interface.cpp
@@ -5,43 +5,32 @@
#include "common/logging/log.h"
#include "core/hle/ipc_helpers.h"
#include "core/hle/service/apm/apm.h"
+#include "core/hle/service/apm/controller.h"
#include "core/hle/service/apm/interface.h"
namespace Service::APM {
class ISession final : public ServiceFramework<ISession> {
public:
- ISession() : ServiceFramework("ISession") {
+ ISession(Controller& controller) : ServiceFramework("ISession"), controller(controller) {
static const FunctionInfo functions[] = {
{0, &ISession::SetPerformanceConfiguration, "SetPerformanceConfiguration"},
{1, &ISession::GetPerformanceConfiguration, "GetPerformanceConfiguration"},
+ {2, nullptr, "SetCpuOverclockEnabled"},
};
RegisterHandlers(functions);
}
private:
- enum class PerformanceConfiguration : u32 {
- Config1 = 0x00010000,
- Config2 = 0x00010001,
- Config3 = 0x00010002,
- Config4 = 0x00020000,
- Config5 = 0x00020001,
- Config6 = 0x00020002,
- Config7 = 0x00020003,
- Config8 = 0x00020004,
- Config9 = 0x00020005,
- Config10 = 0x00020006,
- Config11 = 0x92220007,
- Config12 = 0x92220008,
- };
-
void SetPerformanceConfiguration(Kernel::HLERequestContext& ctx) {
IPC::RequestParser rp{ctx};
- auto mode = static_cast<PerformanceMode>(rp.Pop<u32>());
- u32 config = rp.Pop<u32>();
- LOG_WARNING(Service_APM, "(STUBBED) called mode={} config={}", static_cast<u32>(mode),
- config);
+ const auto mode = rp.PopEnum<PerformanceMode>();
+ const auto config = rp.PopEnum<PerformanceConfiguration>();
+ LOG_DEBUG(Service_APM, "called mode={} config={}", static_cast<u32>(mode),
+ static_cast<u32>(config));
+
+ controller.SetPerformanceConfiguration(mode, config);
IPC::ResponseBuilder rb{ctx, 2};
rb.Push(RESULT_SUCCESS);
@@ -50,20 +39,23 @@ private:
void GetPerformanceConfiguration(Kernel::HLERequestContext& ctx) {
IPC::RequestParser rp{ctx};
- auto mode = static_cast<PerformanceMode>(rp.Pop<u32>());
- LOG_WARNING(Service_APM, "(STUBBED) called mode={}", static_cast<u32>(mode));
+ const auto mode = rp.PopEnum<PerformanceMode>();
+ LOG_DEBUG(Service_APM, "called mode={}", static_cast<u32>(mode));
IPC::ResponseBuilder rb{ctx, 3};
rb.Push(RESULT_SUCCESS);
- rb.Push<u32>(static_cast<u32>(PerformanceConfiguration::Config1));
+ rb.PushEnum(controller.GetCurrentPerformanceConfiguration(mode));
}
+
+ Controller& controller;
};
-APM::APM(std::shared_ptr<Module> apm, const char* name)
- : ServiceFramework(name), apm(std::move(apm)) {
+APM::APM(std::shared_ptr<Module> apm, Controller& controller, const char* name)
+ : ServiceFramework(name), apm(std::move(apm)), controller(controller) {
static const FunctionInfo functions[] = {
{0, &APM::OpenSession, "OpenSession"},
- {1, nullptr, "GetPerformanceMode"},
+ {1, &APM::GetPerformanceMode, "GetPerformanceMode"},
+ {6, nullptr, "IsCpuOverclockEnabled"},
};
RegisterHandlers(functions);
}
@@ -75,10 +67,17 @@ void APM::OpenSession(Kernel::HLERequestContext& ctx) {
IPC::ResponseBuilder rb{ctx, 2, 0, 1};
rb.Push(RESULT_SUCCESS);
- rb.PushIpcInterface<ISession>();
+ rb.PushIpcInterface<ISession>(controller);
+}
+
+void APM::GetPerformanceMode(Kernel::HLERequestContext& ctx) {
+ LOG_DEBUG(Service_APM, "called");
+
+ IPC::ResponseBuilder rb{ctx, 2};
+ rb.PushEnum(controller.GetCurrentPerformanceMode());
}
-APM_Sys::APM_Sys() : ServiceFramework{"apm:sys"} {
+APM_Sys::APM_Sys(Controller& controller) : ServiceFramework{"apm:sys"}, controller(controller) {
// clang-format off
static const FunctionInfo functions[] = {
{0, nullptr, "RequestPerformanceMode"},
@@ -87,8 +86,8 @@ APM_Sys::APM_Sys() : ServiceFramework{"apm:sys"} {
{3, nullptr, "GetLastThrottlingState"},
{4, nullptr, "ClearLastThrottlingState"},
{5, nullptr, "LoadAndApplySettings"},
- {6, nullptr, "SetCpuBoostMode"},
- {7, nullptr, "GetCurrentPerformanceConfiguration"},
+ {6, &APM_Sys::SetCpuBoostMode, "SetCpuBoostMode"},
+ {7, &APM_Sys::GetCurrentPerformanceConfiguration, "GetCurrentPerformanceConfiguration"},
};
// clang-format on
@@ -102,7 +101,28 @@ void APM_Sys::GetPerformanceEvent(Kernel::HLERequestContext& ctx) {
IPC::ResponseBuilder rb{ctx, 2, 0, 1};
rb.Push(RESULT_SUCCESS);
- rb.PushIpcInterface<ISession>();
+ rb.PushIpcInterface<ISession>(controller);
+}
+
+void APM_Sys::SetCpuBoostMode(Kernel::HLERequestContext& ctx) {
+ IPC::RequestParser rp{ctx};
+ const auto mode = rp.PopEnum<CpuBoostMode>();
+
+ LOG_DEBUG(Service_APM, "called, mode={:08X}", static_cast<u32>(mode));
+
+ controller.SetFromCpuBoostMode(mode);
+
+ IPC::ResponseBuilder rb{ctx, 2};
+ rb.Push(RESULT_SUCCESS);
+}
+
+void APM_Sys::GetCurrentPerformanceConfiguration(Kernel::HLERequestContext& ctx) {
+ LOG_DEBUG(Service_APM, "called");
+
+ IPC::ResponseBuilder rb{ctx, 3};
+ rb.Push(RESULT_SUCCESS);
+ rb.PushEnum(
+ controller.GetCurrentPerformanceConfiguration(controller.GetCurrentPerformanceMode()));
}
} // namespace Service::APM
diff --git a/src/core/hle/service/apm/interface.h b/src/core/hle/service/apm/interface.h
index 773541aa4..de1b89437 100644
--- a/src/core/hle/service/apm/interface.h
+++ b/src/core/hle/service/apm/interface.h
@@ -8,24 +8,34 @@
namespace Service::APM {
+class Controller;
+class Module;
+
class APM final : public ServiceFramework<APM> {
public:
- explicit APM(std::shared_ptr<Module> apm, const char* name);
+ explicit APM(std::shared_ptr<Module> apm, Controller& controller, const char* name);
~APM() override;
private:
void OpenSession(Kernel::HLERequestContext& ctx);
+ void GetPerformanceMode(Kernel::HLERequestContext& ctx);
std::shared_ptr<Module> apm;
+ Controller& controller;
};
class APM_Sys final : public ServiceFramework<APM_Sys> {
public:
- explicit APM_Sys();
+ explicit APM_Sys(Controller& controller);
~APM_Sys() override;
+ void SetCpuBoostMode(Kernel::HLERequestContext& ctx);
+
private:
void GetPerformanceEvent(Kernel::HLERequestContext& ctx);
+ void GetCurrentPerformanceConfiguration(Kernel::HLERequestContext& ctx);
+
+ Controller& controller;
};
} // namespace Service::APM
diff --git a/src/core/hle/service/audio/audren_u.cpp b/src/core/hle/service/audio/audren_u.cpp
index 3711e1ea1..679299f68 100644
--- a/src/core/hle/service/audio/audren_u.cpp
+++ b/src/core/hle/service/audio/audren_u.cpp
@@ -25,7 +25,8 @@ namespace Service::Audio {
class IAudioRenderer final : public ServiceFramework<IAudioRenderer> {
public:
- explicit IAudioRenderer(AudioCore::AudioRendererParameter audren_params)
+ explicit IAudioRenderer(AudioCore::AudioRendererParameter audren_params,
+ const std::size_t instance_number)
: ServiceFramework("IAudioRenderer") {
// clang-format off
static const FunctionInfo functions[] = {
@@ -48,8 +49,8 @@ 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);
+ renderer = std::make_unique<AudioCore::AudioRenderer>(
+ system.CoreTiming(), audren_params, system_event.writable, instance_number);
}
private:
@@ -607,7 +608,7 @@ void AudRenU::OpenAudioRendererImpl(Kernel::HLERequestContext& ctx) {
IPC::ResponseBuilder rb{ctx, 2, 0, 1};
rb.Push(RESULT_SUCCESS);
- rb.PushIpcInterface<IAudioRenderer>(params);
+ rb.PushIpcInterface<IAudioRenderer>(params, audren_instance_count++);
}
bool AudRenU::IsFeatureSupported(AudioFeatures feature, u32_le revision) const {
diff --git a/src/core/hle/service/audio/audren_u.h b/src/core/hle/service/audio/audren_u.h
index 1d3c8df61..49f2733cf 100644
--- a/src/core/hle/service/audio/audren_u.h
+++ b/src/core/hle/service/audio/audren_u.h
@@ -33,6 +33,7 @@ private:
};
bool IsFeatureSupported(AudioFeatures feature, u32_le revision) const;
+ std::size_t audren_instance_count = 0;
};
} // namespace Service::Audio
diff --git a/src/core/hle/service/filesystem/filesystem.cpp b/src/core/hle/service/filesystem/filesystem.cpp
index 1ebfeb4bf..8ce110dd1 100644
--- a/src/core/hle/service/filesystem/filesystem.cpp
+++ b/src/core/hle/service/filesystem/filesystem.cpp
@@ -472,12 +472,12 @@ void CreateFactories(FileSys::VfsFilesystem& vfs, bool overwrite) {
}
}
-void InstallInterfaces(SM::ServiceManager& service_manager, FileSys::VfsFilesystem& vfs) {
+void InstallInterfaces(Core::System& system) {
romfs_factory = nullptr;
- CreateFactories(vfs, false);
- std::make_shared<FSP_LDR>()->InstallAsService(service_manager);
- std::make_shared<FSP_PR>()->InstallAsService(service_manager);
- std::make_shared<FSP_SRV>()->InstallAsService(service_manager);
+ CreateFactories(*system.GetFilesystem(), false);
+ std::make_shared<FSP_LDR>()->InstallAsService(system.ServiceManager());
+ std::make_shared<FSP_PR>()->InstallAsService(system.ServiceManager());
+ std::make_shared<FSP_SRV>(system.GetReporter())->InstallAsService(system.ServiceManager());
}
} // namespace Service::FileSystem
diff --git a/src/core/hle/service/filesystem/filesystem.h b/src/core/hle/service/filesystem/filesystem.h
index 6481f237c..3849dd89e 100644
--- a/src/core/hle/service/filesystem/filesystem.h
+++ b/src/core/hle/service/filesystem/filesystem.h
@@ -65,7 +65,7 @@ FileSys::VirtualDir GetModificationDumpRoot(u64 title_id);
// above is called.
void CreateFactories(FileSys::VfsFilesystem& vfs, bool overwrite = true);
-void InstallInterfaces(SM::ServiceManager& service_manager, FileSys::VfsFilesystem& vfs);
+void InstallInterfaces(Core::System& system);
// A class that wraps a VfsDirectory with methods that return ResultVal and ResultCode instead of
// pointers and booleans. This makes using a VfsDirectory with switch services much easier and
diff --git a/src/core/hle/service/filesystem/fsp_srv.cpp b/src/core/hle/service/filesystem/fsp_srv.cpp
index e7df8fd98..d3cd46a9b 100644
--- a/src/core/hle/service/filesystem/fsp_srv.cpp
+++ b/src/core/hle/service/filesystem/fsp_srv.cpp
@@ -26,6 +26,7 @@
#include "core/hle/kernel/process.h"
#include "core/hle/service/filesystem/filesystem.h"
#include "core/hle/service/filesystem/fsp_srv.h"
+#include "core/reporter.h"
namespace Service::FileSystem {
@@ -613,7 +614,7 @@ private:
u64 next_entry_index = 0;
};
-FSP_SRV::FSP_SRV() : ServiceFramework("fsp-srv") {
+FSP_SRV::FSP_SRV(const Core::Reporter& reporter) : ServiceFramework("fsp-srv"), reporter(reporter) {
// clang-format off
static const FunctionInfo functions[] = {
{0, nullptr, "OpenFileSystem"},
@@ -710,14 +711,14 @@ FSP_SRV::FSP_SRV() : ServiceFramework("fsp-srv") {
{1001, nullptr, "SetSaveDataSize"},
{1002, nullptr, "SetSaveDataRootPath"},
{1003, nullptr, "DisableAutoSaveDataCreation"},
- {1004, nullptr, "SetGlobalAccessLogMode"},
+ {1004, &FSP_SRV::SetGlobalAccessLogMode, "SetGlobalAccessLogMode"},
{1005, &FSP_SRV::GetGlobalAccessLogMode, "GetGlobalAccessLogMode"},
- {1006, nullptr, "OutputAccessLogToSdCard"},
+ {1006, &FSP_SRV::OutputAccessLogToSdCard, "OutputAccessLogToSdCard"},
{1007, nullptr, "RegisterUpdatePartition"},
{1008, nullptr, "OpenRegisteredUpdatePartition"},
{1009, nullptr, "GetAndClearMemoryReportInfo"},
{1010, nullptr, "SetDataStorageRedirectTarget"},
- {1011, nullptr, "OutputAccessLogToSdCard2"},
+ {1011, &FSP_SRV::GetAccessLogVersionInfo, "GetAccessLogVersionInfo"},
{1100, nullptr, "OverrideSaveDataTransferTokenSignVerificationKey"},
{1110, nullptr, "CorruptSaveDataFileSystemBySaveDataSpaceId2"},
{1200, nullptr, "OpenMultiCommitManager"},
@@ -814,21 +815,22 @@ void FSP_SRV::OpenSaveDataInfoReaderBySaveDataSpaceId(Kernel::HLERequestContext&
rb.PushIpcInterface<ISaveDataInfoReader>(std::make_shared<ISaveDataInfoReader>(space));
}
-void FSP_SRV::GetGlobalAccessLogMode(Kernel::HLERequestContext& ctx) {
- LOG_WARNING(Service_FS, "(STUBBED) called");
+void FSP_SRV::SetGlobalAccessLogMode(Kernel::HLERequestContext& ctx) {
+ IPC::RequestParser rp{ctx};
+ log_mode = rp.PopEnum<LogMode>();
- enum class LogMode : u32 {
- Off,
- Log,
- RedirectToSdCard,
- LogToSdCard = Log | RedirectToSdCard,
- };
+ LOG_DEBUG(Service_FS, "called, log_mode={:08X}", static_cast<u32>(log_mode));
+
+ IPC::ResponseBuilder rb{ctx, 2};
+ rb.Push(RESULT_SUCCESS);
+}
+
+void FSP_SRV::GetGlobalAccessLogMode(Kernel::HLERequestContext& ctx) {
+ LOG_DEBUG(Service_FS, "called");
- // Given we always want to receive logging information,
- // we always specify logging as enabled.
IPC::ResponseBuilder rb{ctx, 3};
rb.Push(RESULT_SUCCESS);
- rb.PushEnum(LogMode::Log);
+ rb.PushEnum(log_mode);
}
void FSP_SRV::OpenDataStorageByCurrentProcess(Kernel::HLERequestContext& ctx) {
@@ -902,4 +904,26 @@ void FSP_SRV::OpenPatchDataStorageByCurrentProcess(Kernel::HLERequestContext& ct
rb.Push(FileSys::ERROR_ENTITY_NOT_FOUND);
}
+void FSP_SRV::OutputAccessLogToSdCard(Kernel::HLERequestContext& ctx) {
+ const auto raw = ctx.ReadBuffer();
+ auto log = Common::StringFromFixedZeroTerminatedBuffer(
+ reinterpret_cast<const char*>(raw.data()), raw.size());
+
+ LOG_DEBUG(Service_FS, "called, log='{}'", log);
+
+ reporter.SaveFilesystemAccessReport(log_mode, std::move(log));
+
+ IPC::ResponseBuilder rb{ctx, 2};
+ rb.Push(RESULT_SUCCESS);
+}
+
+void FSP_SRV::GetAccessLogVersionInfo(Kernel::HLERequestContext& ctx) {
+ LOG_DEBUG(Service_FS, "called");
+
+ IPC::ResponseBuilder rb{ctx, 4};
+ rb.Push(RESULT_SUCCESS);
+ rb.PushEnum(AccessLogVersion::Latest);
+ rb.Push(access_log_program_index);
+}
+
} // namespace Service::FileSystem
diff --git a/src/core/hle/service/filesystem/fsp_srv.h b/src/core/hle/service/filesystem/fsp_srv.h
index d7572ba7a..b5486a193 100644
--- a/src/core/hle/service/filesystem/fsp_srv.h
+++ b/src/core/hle/service/filesystem/fsp_srv.h
@@ -7,15 +7,32 @@
#include <memory>
#include "core/hle/service/service.h"
+namespace Core {
+class Reporter;
+}
+
namespace FileSys {
class FileSystemBackend;
}
namespace Service::FileSystem {
+enum class AccessLogVersion : u32 {
+ V7_0_0 = 2,
+
+ Latest = V7_0_0,
+};
+
+enum class LogMode : u32 {
+ Off,
+ Log,
+ RedirectToSdCard,
+ LogToSdCard = Log | RedirectToSdCard,
+};
+
class FSP_SRV final : public ServiceFramework<FSP_SRV> {
public:
- explicit FSP_SRV();
+ explicit FSP_SRV(const Core::Reporter& reporter);
~FSP_SRV() override;
private:
@@ -26,13 +43,20 @@ private:
void OpenSaveDataFileSystem(Kernel::HLERequestContext& ctx);
void OpenReadOnlySaveDataFileSystem(Kernel::HLERequestContext& ctx);
void OpenSaveDataInfoReaderBySaveDataSpaceId(Kernel::HLERequestContext& ctx);
+ void SetGlobalAccessLogMode(Kernel::HLERequestContext& ctx);
void GetGlobalAccessLogMode(Kernel::HLERequestContext& ctx);
void OpenDataStorageByCurrentProcess(Kernel::HLERequestContext& ctx);
void OpenDataStorageByDataId(Kernel::HLERequestContext& ctx);
void OpenPatchDataStorageByCurrentProcess(Kernel::HLERequestContext& ctx);
+ void OutputAccessLogToSdCard(Kernel::HLERequestContext& ctx);
+ void GetAccessLogVersionInfo(Kernel::HLERequestContext& ctx);
FileSys::VirtualFile romfs;
u64 current_process_id = 0;
+ u32 access_log_program_index = 0;
+ LogMode log_mode = LogMode::LogToSdCard;
+
+ const Core::Reporter& reporter;
};
} // namespace Service::FileSystem
diff --git a/src/core/hle/service/friend/friend.cpp b/src/core/hle/service/friend/friend.cpp
index dec541f2e..d1ec12ef9 100644
--- a/src/core/hle/service/friend/friend.cpp
+++ b/src/core/hle/service/friend/friend.cpp
@@ -22,7 +22,7 @@ public:
{0, nullptr, "GetCompletionEvent"},
{1, nullptr, "Cancel"},
{10100, nullptr, "GetFriendListIds"},
- {10101, nullptr, "GetFriendList"},
+ {10101, &IFriendService::GetFriendList, "GetFriendList"},
{10102, nullptr, "UpdateFriendInfo"},
{10110, nullptr, "GetFriendProfileImage"},
{10200, nullptr, "SendFriendRequestForApplication"},
@@ -99,6 +99,23 @@ 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");
@@ -112,6 +129,22 @@ 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> {
diff --git a/src/core/hle/service/hid/controllers/npad.cpp b/src/core/hle/service/hid/controllers/npad.cpp
index fdd6d79a2..1e81f776f 100644
--- a/src/core/hle/service/hid/controllers/npad.cpp
+++ b/src/core/hle/service/hid/controllers/npad.cpp
@@ -548,6 +548,37 @@ void Controller_NPad::DisconnectNPad(u32 npad_id) {
connected_controllers[NPadIdToIndex(npad_id)].is_connected = false;
}
+void Controller_NPad::StartLRAssignmentMode() {
+ // Nothing internally is used for lr assignment mode. Since we have the ability to set the
+ // controller types from boot, it doesn't really matter about showing a selection screen
+ is_in_lr_assignment_mode = true;
+}
+
+void Controller_NPad::StopLRAssignmentMode() {
+ is_in_lr_assignment_mode = false;
+}
+
+bool Controller_NPad::SwapNpadAssignment(u32 npad_id_1, u32 npad_id_2) {
+ if (npad_id_1 == NPAD_HANDHELD || npad_id_2 == NPAD_HANDHELD || npad_id_1 == NPAD_UNKNOWN ||
+ npad_id_2 == NPAD_UNKNOWN) {
+ return true;
+ }
+ const auto npad_index_1 = NPadIdToIndex(npad_id_1);
+ const auto npad_index_2 = NPadIdToIndex(npad_id_2);
+
+ if (!IsControllerSupported(connected_controllers[npad_index_1].type) ||
+ !IsControllerSupported(connected_controllers[npad_index_2].type)) {
+ return false;
+ }
+
+ std::swap(connected_controllers[npad_index_1].type, connected_controllers[npad_index_2].type);
+
+ InitNewlyAddedControler(npad_index_1);
+ InitNewlyAddedControler(npad_index_2);
+
+ return true;
+}
+
bool Controller_NPad::IsControllerSupported(NPadControllerType controller) {
if (controller == NPadControllerType::Handheld) {
// Handheld is not even a supported type, lets stop here
diff --git a/src/core/hle/service/hid/controllers/npad.h b/src/core/hle/service/hid/controllers/npad.h
index 4ff50b3cd..4b6c1083f 100644
--- a/src/core/hle/service/hid/controllers/npad.h
+++ b/src/core/hle/service/hid/controllers/npad.h
@@ -124,6 +124,10 @@ public:
void ConnectAllDisconnectedControllers();
void ClearAllControllers();
+ void StartLRAssignmentMode();
+ void StopLRAssignmentMode();
+ bool SwapNpadAssignment(u32 npad_id_1, u32 npad_id_2);
+
// Logical OR for all buttons presses on all controllers
// Specifically for cheat engine and other features.
u32 GetAndResetPressState();
@@ -321,5 +325,6 @@ private:
void RequestPadStateUpdate(u32 npad_id);
std::array<ControllerPad, 10> npad_pad_states{};
bool IsControllerSupported(NPadControllerType controller);
+ bool is_in_lr_assignment_mode{false};
};
} // namespace Service::HID
diff --git a/src/core/hle/service/hid/errors.h b/src/core/hle/service/hid/errors.h
new file mode 100644
index 000000000..3583642e7
--- /dev/null
+++ b/src/core/hle/service/hid/errors.h
@@ -0,0 +1,13 @@
+// Copyright 2019 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "core/hle/result.h"
+
+namespace Service::HID {
+
+constexpr ResultCode ERR_NPAD_NOT_CONNECTED{ErrorModule::HID, 710};
+
+} // namespace Service::HID
diff --git a/src/core/hle/service/hid/hid.cpp b/src/core/hle/service/hid/hid.cpp
index a4ad95d96..0bd24b8eb 100644
--- a/src/core/hle/service/hid/hid.cpp
+++ b/src/core/hle/service/hid/hid.cpp
@@ -16,6 +16,7 @@
#include "core/hle/kernel/readable_event.h"
#include "core/hle/kernel/shared_memory.h"
#include "core/hle/kernel/writable_event.h"
+#include "core/hle/service/hid/errors.h"
#include "core/hle/service/hid/hid.h"
#include "core/hle/service/hid/irs.h"
#include "core/hle/service/hid/xcd.h"
@@ -202,11 +203,11 @@ Hid::Hid() : ServiceFramework("hid") {
{123, nullptr, "SetNpadJoyAssignmentModeSingleByDefault"},
{124, &Hid::SetNpadJoyAssignmentModeDual, "SetNpadJoyAssignmentModeDual"},
{125, &Hid::MergeSingleJoyAsDualJoy, "MergeSingleJoyAsDualJoy"},
- {126, nullptr, "StartLrAssignmentMode"},
- {127, nullptr, "StopLrAssignmentMode"},
+ {126, &Hid::StartLrAssignmentMode, "StartLrAssignmentMode"},
+ {127, &Hid::StopLrAssignmentMode, "StopLrAssignmentMode"},
{128, &Hid::SetNpadHandheldActivationMode, "SetNpadHandheldActivationMode"},
{129, nullptr, "GetNpadHandheldActivationMode"},
- {130, nullptr, "SwapNpadAssignment"},
+ {130, &Hid::SwapNpadAssignment, "SwapNpadAssignment"},
{131, nullptr, "IsUnintendedHomeButtonInputProtectionEnabled"},
{132, nullptr, "EnableUnintendedHomeButtonInputProtection"},
{133, nullptr, "SetNpadJoyAssignmentModeSingleWithDestination"},
@@ -733,6 +734,49 @@ void Hid::SetPalmaBoostMode(Kernel::HLERequestContext& ctx) {
rb.Push(RESULT_SUCCESS);
}
+void Hid::StartLrAssignmentMode(Kernel::HLERequestContext& ctx) {
+ IPC::RequestParser rp{ctx};
+ const auto applet_resource_user_id{rp.Pop<u64>()};
+
+ LOG_DEBUG(Service_HID, "called, applet_resource_user_id={}", applet_resource_user_id);
+ auto& controller = applet_resource->GetController<Controller_NPad>(HidController::NPad);
+ controller.StartLRAssignmentMode();
+
+ IPC::ResponseBuilder rb{ctx, 2};
+ rb.Push(RESULT_SUCCESS);
+}
+
+void Hid::StopLrAssignmentMode(Kernel::HLERequestContext& ctx) {
+ IPC::RequestParser rp{ctx};
+ const auto applet_resource_user_id{rp.Pop<u64>()};
+
+ LOG_DEBUG(Service_HID, "called, applet_resource_user_id={}", applet_resource_user_id);
+ auto& controller = applet_resource->GetController<Controller_NPad>(HidController::NPad);
+ controller.StopLRAssignmentMode();
+
+ IPC::ResponseBuilder rb{ctx, 2};
+ rb.Push(RESULT_SUCCESS);
+}
+
+void Hid::SwapNpadAssignment(Kernel::HLERequestContext& ctx) {
+ IPC::RequestParser rp{ctx};
+ const auto npad_1{rp.Pop<u32>()};
+ const auto npad_2{rp.Pop<u32>()};
+ const auto applet_resource_user_id{rp.Pop<u64>()};
+
+ LOG_DEBUG(Service_HID, "called, applet_resource_user_id={}, npad_1={}, npad_2={}",
+ applet_resource_user_id, npad_1, npad_2);
+
+ auto& controller = applet_resource->GetController<Controller_NPad>(HidController::NPad);
+ IPC::ResponseBuilder rb{ctx, 2};
+ if (controller.SwapNpadAssignment(npad_1, npad_2)) {
+ rb.Push(RESULT_SUCCESS);
+ } else {
+ LOG_ERROR(Service_HID, "Npads are not connected!");
+ rb.Push(ERR_NPAD_NOT_CONNECTED);
+ }
+}
+
class HidDbg final : public ServiceFramework<HidDbg> {
public:
explicit HidDbg() : ServiceFramework{"hid:dbg"} {
diff --git a/src/core/hle/service/hid/hid.h b/src/core/hle/service/hid/hid.h
index d3660cad2..28260ef1b 100644
--- a/src/core/hle/service/hid/hid.h
+++ b/src/core/hle/service/hid/hid.h
@@ -119,6 +119,9 @@ private:
void StopSixAxisSensor(Kernel::HLERequestContext& ctx);
void SetIsPalmaAllConnectable(Kernel::HLERequestContext& ctx);
void SetPalmaBoostMode(Kernel::HLERequestContext& ctx);
+ void StartLrAssignmentMode(Kernel::HLERequestContext& ctx);
+ void StopLrAssignmentMode(Kernel::HLERequestContext& ctx);
+ void SwapNpadAssignment(Kernel::HLERequestContext& ctx);
std::shared_ptr<IAppletResource> applet_resource;
};
diff --git a/src/core/hle/service/ldr/ldr.cpp b/src/core/hle/service/ldr/ldr.cpp
index b839303ac..8ddad8682 100644
--- a/src/core/hle/service/ldr/ldr.cpp
+++ b/src/core/hle/service/ldr/ldr.cpp
@@ -345,14 +345,16 @@ public:
vm_manager
.MirrorMemory(*map_address, nro_address, nro_size, Kernel::MemoryState::ModuleCode)
.IsSuccess());
- ASSERT(vm_manager.UnmapRange(nro_address, nro_size).IsSuccess());
+ ASSERT(vm_manager.ReprotectRange(nro_address, nro_size, Kernel::VMAPermission::None)
+ .IsSuccess());
if (bss_size > 0) {
ASSERT(vm_manager
.MirrorMemory(*map_address + nro_size, bss_address, bss_size,
Kernel::MemoryState::ModuleCode)
.IsSuccess());
- ASSERT(vm_manager.UnmapRange(bss_address, bss_size).IsSuccess());
+ ASSERT(vm_manager.ReprotectRange(bss_address, bss_size, Kernel::VMAPermission::None)
+ .IsSuccess());
}
vm_manager.ReprotectRange(*map_address, header.text_size,
@@ -364,7 +366,8 @@ public:
Core::System::GetInstance().InvalidateCpuInstructionCaches();
- nro.insert_or_assign(*map_address, NROInfo{hash, nro_size + bss_size});
+ nro.insert_or_assign(*map_address,
+ NROInfo{hash, nro_address, nro_size, bss_address, bss_size});
IPC::ResponseBuilder rb{ctx, 4};
rb.Push(RESULT_SUCCESS);
@@ -409,9 +412,23 @@ public:
}
auto& vm_manager = Core::CurrentProcess()->VMManager();
- const auto& nro_size = iter->second.size;
+ const auto& nro_info = iter->second;
- ASSERT(vm_manager.UnmapRange(nro_address, nro_size).IsSuccess());
+ // 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());
+ }
Core::System::GetInstance().InvalidateCpuInstructionCaches();
@@ -473,7 +490,10 @@ private:
struct NROInfo {
SHA256Hash hash;
- u64 size;
+ VAddr nro_address;
+ u64 nro_size;
+ VAddr bss_address;
+ u64 bss_size;
};
bool initialized = false;
diff --git a/src/core/hle/service/mii/mii.cpp b/src/core/hle/service/mii/mii.cpp
index ce84e25ed..0b3923ad9 100644
--- a/src/core/hle/service/mii/mii.cpp
+++ b/src/core/hle/service/mii/mii.cpp
@@ -48,7 +48,7 @@ public:
{19, nullptr, "Export"},
{20, nullptr, "IsBrokenDatabaseWithClearFlag"},
{21, &IDatabaseService::GetIndex, "GetIndex"},
- {22, nullptr, "SetInterfaceVersion"},
+ {22, &IDatabaseService::SetInterfaceVersion, "SetInterfaceVersion"},
{23, nullptr, "Convert"},
};
// clang-format on
@@ -350,8 +350,22 @@ 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{};
};
diff --git a/src/core/hle/service/pm/pm.cpp b/src/core/hle/service/pm/pm.cpp
index ebcc41a43..fe6b5f798 100644
--- a/src/core/hle/service/pm/pm.cpp
+++ b/src/core/hle/service/pm/pm.cpp
@@ -3,11 +3,44 @@
// 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"} {
@@ -41,14 +74,15 @@ private:
class DebugMonitor final : public ServiceFramework<DebugMonitor> {
public:
- explicit DebugMonitor() : ServiceFramework{"pm:dmnt"} {
+ explicit DebugMonitor(const Kernel::KernelCore& kernel)
+ : ServiceFramework{"pm:dmnt"}, kernel(kernel) {
// clang-format off
static const FunctionInfo functions[] = {
{0, nullptr, "GetDebugProcesses"},
{1, nullptr, "StartDebugProcess"},
- {2, nullptr, "GetTitlePid"},
+ {2, &DebugMonitor::GetTitlePid, "GetTitlePid"},
{3, nullptr, "EnableDebugForTitleId"},
- {4, nullptr, "GetApplicationPid"},
+ {4, &DebugMonitor::GetApplicationPid, "GetApplicationPid"},
{5, nullptr, "EnableDebugForApplication"},
{6, nullptr, "DisableDebug"},
};
@@ -56,21 +90,77 @@ 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() : ServiceFramework{"pm:info"} {
+ explicit Info(const std::vector<Kernel::SharedPtr<Kernel::Process>>& process_list)
+ : ServiceFramework{"pm:info"}, process_list(process_list) {
static const FunctionInfo functions[] = {
- {0, nullptr, "GetTitleId"},
+ {0, &Info::GetTitleId, "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() : ServiceFramework{"pm:shell"} {
+ explicit Shell(const Kernel::KernelCore& kernel)
+ : ServiceFramework{"pm:shell"}, kernel(kernel) {
// clang-format off
static const FunctionInfo functions[] = {
{0, nullptr, "LaunchProcess"},
@@ -79,21 +169,31 @@ public:
{3, nullptr, "GetProcessEventWaiter"},
{4, nullptr, "GetProcessEventType"},
{5, nullptr, "NotifyBootFinished"},
- {6, nullptr, "GetApplicationPid"},
+ {6, &Shell::GetApplicationPid, "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(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);
+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());
}
} // namespace Service::PM
diff --git a/src/core/hle/service/pm/pm.h b/src/core/hle/service/pm/pm.h
index cc8d3f215..852e7050c 100644
--- a/src/core/hle/service/pm/pm.h
+++ b/src/core/hle/service/pm/pm.h
@@ -4,8 +4,8 @@
#pragma once
-namespace Service::SM {
-class ServiceManager;
+namespace Core {
+class System;
}
namespace Service::PM {
@@ -16,6 +16,6 @@ enum class SystemBootMode {
};
/// Registers all PM services with the specified service manager.
-void InstallInterfaces(SM::ServiceManager& service_manager);
+void InstallInterfaces(Core::System& system);
} // namespace Service::PM
diff --git a/src/core/hle/service/service.cpp b/src/core/hle/service/service.cpp
index 5fc7d3cab..7eefd733f 100644
--- a/src/core/hle/service/service.cpp
+++ b/src/core/hle/service/service.cpp
@@ -195,8 +195,7 @@ ResultCode ServiceFrameworkBase::HandleSyncRequest(Kernel::HLERequestContext& co
// Module interface
/// Initialize ServiceManager
-void Init(std::shared_ptr<SM::ServiceManager>& sm, Core::System& system,
- FileSys::VfsFilesystem& vfs) {
+void Init(std::shared_ptr<SM::ServiceManager>& sm, Core::System& system) {
// NVFlinger needs to be accessed by several services like Vi and AppletOE so we instantiate it
// here and pass it into the respective InstallInterfaces functions.
auto nv_flinger = std::make_shared<NVFlinger::NVFlinger>(system.CoreTiming());
@@ -206,7 +205,7 @@ void Init(std::shared_ptr<SM::ServiceManager>& sm, Core::System& system,
Account::InstallInterfaces(system);
AM::InstallInterfaces(*sm, nv_flinger, system);
AOC::InstallInterfaces(*sm);
- APM::InstallInterfaces(*sm);
+ APM::InstallInterfaces(system);
Audio::InstallInterfaces(*sm);
BCAT::InstallInterfaces(*sm);
BPC::InstallInterfaces(*sm);
@@ -218,7 +217,7 @@ void Init(std::shared_ptr<SM::ServiceManager>& sm, Core::System& system,
EUPLD::InstallInterfaces(*sm);
Fatal::InstallInterfaces(*sm);
FGM::InstallInterfaces(*sm);
- FileSystem::InstallInterfaces(*sm, vfs);
+ FileSystem::InstallInterfaces(system);
Friend::InstallInterfaces(*sm);
Glue::InstallInterfaces(system);
GRC::InstallInterfaces(*sm);
@@ -242,7 +241,7 @@ void Init(std::shared_ptr<SM::ServiceManager>& sm, Core::System& system,
PCTL::InstallInterfaces(*sm);
PCV::InstallInterfaces(*sm);
PlayReport::InstallInterfaces(*sm);
- PM::InstallInterfaces(*sm);
+ PM::InstallInterfaces(system);
PSC::InstallInterfaces(*sm);
PSM::InstallInterfaces(*sm);
Set::InstallInterfaces(*sm);
diff --git a/src/core/hle/service/service.h b/src/core/hle/service/service.h
index abbfe5524..c6c4bdae5 100644
--- a/src/core/hle/service/service.h
+++ b/src/core/hle/service/service.h
@@ -182,8 +182,7 @@ private:
};
/// Initialize ServiceManager
-void Init(std::shared_ptr<SM::ServiceManager>& sm, Core::System& system,
- FileSys::VfsFilesystem& vfs);
+void Init(std::shared_ptr<SM::ServiceManager>& sm, Core::System& system);
/// Shutdown ServiceManager
void Shutdown();
diff --git a/src/core/reporter.cpp b/src/core/reporter.cpp
index 6ea26fda7..5d4c3e6ea 100644
--- a/src/core/reporter.cpp
+++ b/src/core/reporter.cpp
@@ -350,6 +350,24 @@ void Reporter::SaveErrorReport(u64 title_id, ResultCode result,
SaveToFile(std::move(out), GetPath("error_report", title_id, timestamp));
}
+void Reporter::SaveFilesystemAccessReport(Service::FileSystem::LogMode log_mode,
+ std::string log_message) const {
+ if (!IsReportingEnabled())
+ return;
+
+ const auto timestamp = GetTimestamp();
+ const auto title_id = system.CurrentProcess()->GetTitleID();
+ json out;
+
+ out["yuzu_version"] = GetYuzuVersionData();
+ out["report_common"] = GetReportCommonData(title_id, RESULT_SUCCESS, timestamp);
+
+ out["log_mode"] = fmt::format("{:08X}", static_cast<u32>(log_mode));
+ out["log_message"] = std::move(log_message);
+
+ SaveToFile(std::move(out), GetPath("filesystem_access_report", title_id, timestamp));
+}
+
void Reporter::SaveUserReport() const {
if (!IsReportingEnabled()) {
return;
diff --git a/src/core/reporter.h b/src/core/reporter.h
index 4266ca550..44256de50 100644
--- a/src/core/reporter.h
+++ b/src/core/reporter.h
@@ -16,6 +16,10 @@ namespace Kernel {
class HLERequestContext;
} // namespace Kernel
+namespace Service::FileSystem {
+enum class LogMode : u32;
+}
+
namespace Core {
class System;
@@ -49,6 +53,9 @@ public:
std::optional<std::string> custom_text_main = {},
std::optional<std::string> custom_text_detail = {}) const;
+ void SaveFilesystemAccessReport(Service::FileSystem::LogMode log_mode,
+ std::string log_message) const;
+
void SaveUserReport() const;
private:
diff --git a/src/core/settings.cpp b/src/core/settings.cpp
index 63aa59690..0dd1632ac 100644
--- a/src/core/settings.cpp
+++ b/src/core/settings.cpp
@@ -85,7 +85,6 @@ 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);
diff --git a/src/core/settings.h b/src/core/settings.h
index acf18d653..6638ce8f9 100644
--- a/src/core/settings.h
+++ b/src/core/settings.h
@@ -378,7 +378,6 @@ struct Values {
std::atomic_bool is_device_reload_pending{true};
// Core
- bool cpu_jit_enabled;
bool use_multi_core;
// Data Storage
diff --git a/src/core/telemetry_session.cpp b/src/core/telemetry_session.cpp
index 98f49042a..793d102d3 100644
--- a/src/core/telemetry_session.cpp
+++ b/src/core/telemetry_session.cpp
@@ -168,7 +168,6 @@ 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",
diff --git a/src/video_core/CMakeLists.txt b/src/video_core/CMakeLists.txt
index 6839abe71..7c18c27b3 100644
--- a/src/video_core/CMakeLists.txt
+++ b/src/video_core/CMakeLists.txt
@@ -1,4 +1,5 @@
add_library(video_core STATIC
+ buffer_cache.h
dma_pusher.cpp
dma_pusher.h
debug_utils/debug_utils.cpp
@@ -43,8 +44,6 @@ 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
@@ -103,6 +102,8 @@ 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
diff --git a/src/video_core/buffer_cache.h b/src/video_core/buffer_cache.h
new file mode 100644
index 000000000..6f868b8b4
--- /dev/null
+++ b/src/video_core/buffer_cache.h
@@ -0,0 +1,299 @@
+// 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
diff --git a/src/video_core/engines/maxwell_3d.h b/src/video_core/engines/maxwell_3d.h
index 13e314944..8d15c8a48 100644
--- a/src/video_core/engines/maxwell_3d.h
+++ b/src/video_core/engines/maxwell_3d.h
@@ -67,6 +67,7 @@ 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,
diff --git a/src/video_core/engines/shader_bytecode.h b/src/video_core/engines/shader_bytecode.h
index b92921d0f..79d469b88 100644
--- a/src/video_core/engines/shader_bytecode.h
+++ b/src/video_core/engines/shader_bytecode.h
@@ -1278,6 +1278,7 @@ 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.
@@ -1368,6 +1369,20 @@ 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
} out;
@@ -1464,6 +1479,7 @@ public:
BFE_IMM,
BFI_IMM_R,
BRA,
+ BRX,
PBK,
LD_A,
LD_L,
@@ -1738,6 +1754,7 @@ 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"),
@@ -1760,7 +1777,7 @@ 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("1101101---------", Id::TLDS, Type::Texture, "TLDS"),
+ INST("1101-01---------", 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"),
diff --git a/src/video_core/gpu.cpp b/src/video_core/gpu.cpp
index 52706505b..1b4975498 100644
--- a/src/video_core/gpu.cpp
+++ b/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>(rasterizer);
+ memory_manager = std::make_unique<Tegra::MemoryManager>(system, 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);
diff --git a/src/video_core/macro_interpreter.cpp b/src/video_core/macro_interpreter.cpp
index c766ed692..9f59a2dc1 100644
--- a/src/video_core/macro_interpreter.cpp
+++ b/src/video_core/macro_interpreter.cpp
@@ -4,14 +4,18 @@
#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);
diff --git a/src/video_core/memory_manager.cpp b/src/video_core/memory_manager.cpp
index 322453116..bffae940c 100644
--- a/src/video_core/memory_manager.cpp
+++ b/src/video_core/memory_manager.cpp
@@ -5,13 +5,17 @@
#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(VideoCore::RasterizerInterface& rasterizer) : rasterizer{rasterizer} {
+MemoryManager::MemoryManager(Core::System& system, VideoCore::RasterizerInterface& rasterizer)
+ : rasterizer{rasterizer}, system{system} {
std::fill(page_table.pointers.begin(), page_table.pointers.end(), nullptr);
std::fill(page_table.attributes.begin(), page_table.attributes.end(),
Common::PageType::Unmapped);
@@ -49,6 +53,11 @@ 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;
}
@@ -59,7 +68,11 @@ 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;
}
@@ -68,9 +81,16 @@ 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;
}
diff --git a/src/video_core/memory_manager.h b/src/video_core/memory_manager.h
index 43a84bd52..aea010087 100644
--- a/src/video_core/memory_manager.h
+++ b/src/video_core/memory_manager.h
@@ -14,6 +14,10 @@ namespace VideoCore {
class RasterizerInterface;
}
+namespace Core {
+class System;
+}
+
namespace Tegra {
/**
@@ -47,7 +51,7 @@ struct VirtualMemoryArea {
class MemoryManager final {
public:
- explicit MemoryManager(VideoCore::RasterizerInterface& rasterizer);
+ explicit MemoryManager(Core::System& system, VideoCore::RasterizerInterface& rasterizer);
~MemoryManager();
GPUVAddr AllocateSpace(u64 size, u64 align);
@@ -173,6 +177,8 @@ private:
Common::PageTable page_table{page_bits};
VMAMap vma_map;
VideoCore::RasterizerInterface& rasterizer;
+
+ Core::System& system;
};
} // namespace Tegra
diff --git a/src/video_core/rasterizer_interface.h b/src/video_core/rasterizer_interface.h
index 5ee4f8e8e..2b7367568 100644
--- a/src/video_core/rasterizer_interface.h
+++ b/src/video_core/rasterizer_interface.h
@@ -47,6 +47,9 @@ 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,
diff --git a/src/video_core/renderer_opengl/gl_buffer_cache.cpp b/src/video_core/renderer_opengl/gl_buffer_cache.cpp
index 2b9bd142e..2a9b523f5 100644
--- a/src/video_core/renderer_opengl/gl_buffer_cache.cpp
+++ b/src/video_core/renderer_opengl/gl_buffer_cache.cpp
@@ -2,103 +2,57 @@
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
-#include <cstring>
#include <memory>
-#include "common/alignment.h"
-#include "core/core.h"
-#include "video_core/memory_manager.h"
+#include <glad/glad.h>
+
+#include "common/assert.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 {
-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(RasterizerOpenGL& rasterizer, std::size_t size)
- : RasterizerCache{rasterizer}, stream_buffer(size, true) {}
-
-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);
- }
- }
+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)} {}
- AlignBuffer(alignment);
- const GLintptr uploaded_offset = buffer_offset;
+OGLBufferCache::~OGLBufferCache() = default;
- 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;
+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::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::ToHandle(const OGLBuffer& buffer) {
+ return &buffer.handle;
}
-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;
+const GLuint* OGLBufferCache::GetEmptyBuffer(std::size_t) {
+ static const GLuint null_buffer = 0;
+ return &null_buffer;
}
-void OGLBufferCache::Unmap() {
- stream_buffer.Unmap(buffer_offset - buffer_offset_base);
+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);
}
-GLuint OGLBufferCache::GetHandle() const {
- return stream_buffer.GetHandle();
+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);
}
-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;
+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));
}
} // namespace OpenGL
diff --git a/src/video_core/renderer_opengl/gl_buffer_cache.h b/src/video_core/renderer_opengl/gl_buffer_cache.h
index f2347581b..8c8ac4038 100644
--- a/src/video_core/renderer_opengl/gl_buffer_cache.h
+++ b/src/video_core/renderer_opengl/gl_buffer_cache.h
@@ -4,80 +4,44 @@
#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 CachedBufferEntry final : public RasterizerCacheObject {
-public:
- explicit CachedBufferEntry(VAddr cpu_addr, std::size_t size, GLintptr offset,
- std::size_t alignment, u8* host_ptr);
-
- 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>> {
+class OGLBufferCache final : public VideoCommon::BufferCache<OGLBuffer, GLuint, OGLStreamBuffer> {
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);
+ explicit OGLBufferCache(RasterizerOpenGL& rasterizer, Core::System& system,
+ std::size_t stream_size);
+ ~OGLBufferCache();
- /// 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;
+ const GLuint* GetEmptyBuffer(std::size_t) override;
protected:
- void AlignBuffer(std::size_t alignment);
+ OGLBuffer CreateBuffer(std::size_t size) override;
+
+ 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;
- u8* buffer_ptr = nullptr;
- GLintptr buffer_offset = 0;
- GLintptr buffer_offset_base = 0;
+ void CopyBufferData(const OGLBuffer& src, const OGLBuffer& dst, std::size_t src_offset,
+ std::size_t dst_offset, std::size_t size) override;
};
} // namespace OpenGL
diff --git a/src/video_core/renderer_opengl/gl_device.cpp b/src/video_core/renderer_opengl/gl_device.cpp
index 4b1f22f7a..85424a4c9 100644
--- a/src/video_core/renderer_opengl/gl_device.cpp
+++ b/src/video_core/renderer_opengl/gl_device.cpp
@@ -24,6 +24,7 @@ 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;
diff --git a/src/video_core/renderer_opengl/gl_device.h b/src/video_core/renderer_opengl/gl_device.h
index 7a3a0675b..dc883722d 100644
--- a/src/video_core/renderer_opengl/gl_device.h
+++ b/src/video_core/renderer_opengl/gl_device.h
@@ -18,6 +18,10 @@ public:
return uniform_buffer_alignment;
}
+ std::size_t GetShaderStorageBufferAlignment() const {
+ return shader_storage_alignment;
+ }
+
u32 GetMaxVertexAttributes() const {
return max_vertex_attributes;
}
@@ -43,6 +47,7 @@ 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{};
diff --git a/src/video_core/renderer_opengl/gl_global_cache.cpp b/src/video_core/renderer_opengl/gl_global_cache.cpp
deleted file mode 100644
index d5e385151..000000000
--- a/src/video_core/renderer_opengl/gl_global_cache.cpp
+++ /dev/null
@@ -1,102 +0,0 @@
-// 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
diff --git a/src/video_core/renderer_opengl/gl_global_cache.h b/src/video_core/renderer_opengl/gl_global_cache.h
deleted file mode 100644
index 2d467a240..000000000
--- a/src/video_core/renderer_opengl/gl_global_cache.h
+++ /dev/null
@@ -1,82 +0,0 @@
-// 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
diff --git a/src/video_core/renderer_opengl/gl_rasterizer.cpp b/src/video_core/renderer_opengl/gl_rasterizer.cpp
index f45a3c5ef..0bb5c068c 100644
--- a/src/video_core/renderer_opengl/gl_rasterizer.cpp
+++ b/src/video_core/renderer_opengl/gl_rasterizer.cpp
@@ -20,6 +20,7 @@
#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"
@@ -80,11 +81,25 @@ 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},
- global_cache{*this}, system{system}, screen_info{info},
- buffer_cache(*this, STREAM_BUFFER_SIZE) {
+ system{system}, screen_info{info}, buffer_cache{*this, system, STREAM_BUFFER_SIZE} {
OpenGLState::ApplyDefaultState();
shader_program_manager = std::make_unique<GLShader::ProgramManager>();
@@ -129,8 +144,6 @@ 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
@@ -197,11 +210,11 @@ void RasterizerOpenGL::SetupVertexBuffer(GLuint vao) {
ASSERT(end > start);
const u64 size = end - start + 1;
- const GLintptr vertex_buffer_offset = buffer_cache.UploadMemory(start, size);
+ const auto [vertex_buffer, vertex_buffer_offset] = buffer_cache.UploadMemory(start, size);
// Bind the vertex array to the buffer at the current offset.
- glVertexArrayVertexBuffer(vao, index, buffer_cache.GetHandle(), vertex_buffer_offset,
- vertex_array.stride);
+ vertex_array_pushbuffer.SetVertexBuffer(index, vertex_buffer, vertex_buffer_offset,
+ vertex_array.stride);
if (regs.instanced_arrays.IsInstancingEnabled(index) && vertex_array.divisor != 0) {
// Enable vertex buffer instancing with the specified divisor.
@@ -215,7 +228,19 @@ void RasterizerOpenGL::SetupVertexBuffer(GLuint vao) {
gpu.dirty_flags.vertex_array.reset();
}
-DrawParameters RasterizerOpenGL::SetupDraw() {
+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) {
const auto& gpu = system.GPU().Maxwell3D();
const auto& regs = gpu.regs;
const bool is_indexed = accelerate_draw == AccelDraw::Indexed;
@@ -227,11 +252,9 @@ DrawParameters RasterizerOpenGL::SetupDraw() {
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 =
- buffer_cache.UploadMemory(regs.index_array.IndexStart(), CalculateIndexBufferSize());
+ params.index_buffer_offset = index_buffer_offset;
params.base_vertex = static_cast<GLint>(regs.vb_element_base);
} else {
params.count = regs.vertex_buffer.count;
@@ -247,10 +270,6 @@ 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)};
@@ -271,12 +290,11 @@ void RasterizerOpenGL::SetupShaders(GLenum primitive_mode) {
GLShader::MaxwellUniformData ubo{};
ubo.SetFromRegs(gpu, stage);
- const GLintptr offset =
+ const auto [buffer, offset] =
buffer_cache.UploadHostMemory(&ubo, sizeof(ubo), device.GetUniformBufferAlignment());
// Bind the emulation info buffer
- bind_ubo_pushbuffer.Push(buffer_cache.GetHandle(), offset,
- static_cast<GLsizeiptr>(sizeof(ubo)));
+ bind_ubo_pushbuffer.Push(buffer, offset, static_cast<GLsizeiptr>(sizeof(ubo)));
Shader shader{shader_cache.GetStageProgram(program)};
@@ -321,9 +339,6 @@ 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;
@@ -634,26 +649,46 @@ void RasterizerOpenGL::DrawArrays() {
Maxwell::MaxShaderStage;
// Add space for at least 18 constant buffers
- buffer_size +=
- Maxwell::MaxConstBuffers * (MaxConstbufferSize + device.GetUniformBufferAlignment());
+ buffer_size += Maxwell::MaxConstBuffers *
+ (Maxwell::MaxConstBufferSize + device.GetUniformBufferAlignment());
- 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 the vertex array.
+ buffer_cache.Map(buffer_size);
+ // 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();
- DrawParameters params = SetupDraw();
+ // 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.
texture_cache.GuardSamplers(true);
SetupShaders(params.primitive_mode);
texture_cache.GuardSamplers(false);
ConfigureFramebuffers(state);
- buffer_cache.Unmap();
+ // 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();
+ }
shader_program_manager->ApplyTo(state);
state.Apply();
@@ -675,7 +710,7 @@ void RasterizerOpenGL::FlushRegion(CacheAddr addr, u64 size) {
return;
}
texture_cache.FlushRegion(addr, size);
- global_cache.FlushRegion(addr, size);
+ buffer_cache.FlushRegion(addr, size);
}
void RasterizerOpenGL::InvalidateRegion(CacheAddr addr, u64 size) {
@@ -685,7 +720,6 @@ 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);
}
@@ -696,6 +730,10 @@ 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) {
@@ -739,11 +777,9 @@ 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 (u32 bindpoint = 0; bindpoint < entries.size(); ++bindpoint) {
- const auto& entry = entries[bindpoint];
+ for (const auto& entry : shader->GetShaderEntries().const_buffers) {
SetupConstBuffer(shader_stage.const_buffers[entry.GetIndex()], entry);
}
}
@@ -752,46 +788,34 @@ 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(0, 0, 0);
+ bind_ubo_pushbuffer.Push(buffer_cache.GetEmptyBuffer(sizeof(float)), 0, sizeof(float));
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.
- size = Common::AlignUp(size, sizeof(GLvec4));
- ASSERT_MSG(size <= MaxConstbufferSize, "Constant buffer is too big");
+ const std::size_t size = Common::AlignUp(GetConstBufferSize(buffer, entry), sizeof(GLvec4));
- 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);
+ const auto alignment = device.GetUniformBufferAlignment();
+ const auto [cbuf, offset] = buffer_cache.UploadMemory(buffer.address, size, alignment);
+ bind_ubo_pushbuffer.Push(cbuf, offset, size);
}
void RasterizerOpenGL::SetupGlobalRegions(Tegra::Engines::Maxwell3D::Regs::ShaderStage stage,
const Shader& shader) {
- 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()));
+ 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));
}
}
diff --git a/src/video_core/renderer_opengl/gl_rasterizer.h b/src/video_core/renderer_opengl/gl_rasterizer.h
index bf67e3a70..40b571d58 100644
--- a/src/video_core/renderer_opengl/gl_rasterizer.h
+++ b/src/video_core/renderer_opengl/gl_rasterizer.h
@@ -24,7 +24,6 @@
#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"
@@ -63,6 +62,7 @@ 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;
@@ -73,11 +73,6 @@ 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{};
@@ -98,17 +93,19 @@ private:
/**
* Configures the color and depth framebuffer states.
- * @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 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 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 use_color_fb = true, bool using_depth_fb = true,
+ OpenGLState& current_state, bool using_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.
@@ -189,7 +186,6 @@ private:
TextureCacheOpenGL texture_cache;
ShaderCacheOpenGL shader_cache;
- GlobalRegionCacheOpenGL global_cache;
SamplerCacheOpenGL sampler_cache;
FramebufferCacheOpenGL framebuffer_cache;
@@ -208,6 +204,7 @@ 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};
@@ -220,7 +217,9 @@ private:
void SetupVertexBuffer(GLuint vao);
- DrawParameters SetupDraw();
+ GLintptr SetupIndexBuffer();
+
+ DrawParameters SetupDraw(GLintptr index_buffer_offset);
void SetupShaders(GLenum primitive_mode);
diff --git a/src/video_core/renderer_opengl/gl_sampler_cache.h b/src/video_core/renderer_opengl/gl_sampler_cache.h
index defbc2d81..34ee37f00 100644
--- a/src/video_core/renderer_opengl/gl_sampler_cache.h
+++ b/src/video_core/renderer_opengl/gl_sampler_cache.h
@@ -17,9 +17,9 @@ public:
~SamplerCacheOpenGL();
protected:
- OGLSampler CreateSampler(const Tegra::Texture::TSCEntry& tsc) const;
+ OGLSampler CreateSampler(const Tegra::Texture::TSCEntry& tsc) const override;
- GLuint ToSamplerType(const OGLSampler& sampler) const;
+ GLuint ToSamplerType(const OGLSampler& sampler) const override;
};
} // namespace OpenGL
diff --git a/src/video_core/renderer_opengl/gl_shader_cache.cpp b/src/video_core/renderer_opengl/gl_shader_cache.cpp
index 7fea92713..32dd9eae7 100644
--- a/src/video_core/renderer_opengl/gl_shader_cache.cpp
+++ b/src/video_core/renderer_opengl/gl_shader_cache.cpp
@@ -129,9 +129,11 @@ 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) {
- u64 unique_identifier =
- Common::CityHash64(reinterpret_cast<const char*>(code.data()), CalculateProgramSize(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);
if (program_type != Maxwell::ShaderProgram::VertexA) {
return unique_identifier;
}
@@ -140,8 +142,11 @@ u64 GetUniqueIdentifier(Maxwell::ShaderProgram program_type, const ProgramCode&
std::size_t seed = 0;
boost::hash_combine(seed, unique_identifier);
- const u64 identifier_b = Common::CityHash64(reinterpret_cast<const char*>(code_b.data()),
- CalculateProgramSize(code_b));
+ if (size_b == 0) {
+ size_b = CalculateProgramSize(code_b);
+ }
+ const u64 identifier_b =
+ Common::CityHash64(reinterpret_cast<const char*>(code_b.data()), size_b);
boost::hash_combine(seed, identifier_b);
return static_cast<u64>(seed);
}
@@ -150,14 +155,17 @@ 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.unique_identifier = GetUniqueIdentifier(
+ program_type, program_code, program_code_b, setup.program.size_a, setup.program.size_b);
switch (program_type) {
case Maxwell::ShaderProgram::VertexA:
diff --git a/src/video_core/renderer_opengl/gl_shader_decompiler.cpp b/src/video_core/renderer_opengl/gl_shader_decompiler.cpp
index 3408e6c41..119073776 100644
--- a/src/video_core/renderer_opengl/gl_shader_decompiler.cpp
+++ b/src/video_core/renderer_opengl/gl_shader_decompiler.cpp
@@ -47,7 +47,7 @@ using TextureArgument = std::pair<Type, Node>;
using TextureIR = std::variant<TextureAoffi, TextureArgument>;
constexpr u32 MAX_CONSTBUFFER_ELEMENTS =
- static_cast<u32>(RasterizerOpenGL::MaxConstbufferSize) / (4 * sizeof(float));
+ static_cast<u32>(Maxwell::MaxConstBufferSize) / (4 * sizeof(float));
class ShaderWriter {
public:
@@ -192,10 +192,12 @@ 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.
- 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));
+ 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));
+ }
}
code.AddLine("while (true) {{");
@@ -1602,6 +1604,14 @@ 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]);
@@ -1836,6 +1846,7 @@ private:
&GLSLDecompiler::ImageStore,
&GLSLDecompiler::Branch,
+ &GLSLDecompiler::BranchIndirect,
&GLSLDecompiler::PushFlowStack,
&GLSLDecompiler::PopFlowStack,
&GLSLDecompiler::Exit,
diff --git a/src/video_core/renderer_opengl/gl_shader_gen.cpp b/src/video_core/renderer_opengl/gl_shader_gen.cpp
index 9148629ec..f9ee8429e 100644
--- a/src/video_core/renderer_opengl/gl_shader_gen.cpp
+++ b/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);
+ const ShaderIR program_ir(setup.program.code, PROGRAM_OFFSET, setup.program.size_a);
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);
+ const ShaderIR program_ir_b(setup.program.code_b, PROGRAM_OFFSET, setup.program.size_b);
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);
+ const ShaderIR program_ir(setup.program.code, PROGRAM_OFFSET, setup.program.size_a);
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);
+ const ShaderIR program_ir(setup.program.code, PROGRAM_OFFSET, setup.program.size_a);
ProgramResult program =
Decompile(device, program_ir, Maxwell3D::Regs::ShaderStage::Fragment, "fragment");
diff --git a/src/video_core/renderer_opengl/gl_shader_gen.h b/src/video_core/renderer_opengl/gl_shader_gen.h
index 0536c8a03..7cbc590f8 100644
--- a/src/video_core/renderer_opengl/gl_shader_gen.h
+++ b/src/video_core/renderer_opengl/gl_shader_gen.h
@@ -27,6 +27,8 @@ 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
diff --git a/src/video_core/renderer_opengl/gl_state.cpp b/src/video_core/renderer_opengl/gl_state.cpp
index d86e137ac..0eae98afe 100644
--- a/src/video_core/renderer_opengl/gl_state.cpp
+++ b/src/video_core/renderer_opengl/gl_state.cpp
@@ -6,8 +6,11 @@
#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;
@@ -524,6 +527,7 @@ void OpenGLState::ApplySamplers() const {
}
void OpenGLState::Apply() const {
+ MICROPROFILE_SCOPE(OpenGL_State);
ApplyFramebufferState();
ApplyVertexArrayState();
ApplyShaderProgram();
diff --git a/src/video_core/renderer_opengl/gl_texture_cache.cpp b/src/video_core/renderer_opengl/gl_texture_cache.cpp
index 08ae1a429..b1f6bc7c2 100644
--- a/src/video_core/renderer_opengl/gl_texture_cache.cpp
+++ b/src/video_core/renderer_opengl/gl_texture_cache.cpp
@@ -31,6 +31,8 @@ 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 {
@@ -535,6 +537,7 @@ 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);
diff --git a/src/video_core/renderer_opengl/renderer_opengl.cpp b/src/video_core/renderer_opengl/renderer_opengl.cpp
index b142521ec..9ecdddb0d 100644
--- a/src/video_core/renderer_opengl/renderer_opengl.cpp
+++ b/src/video_core/renderer_opengl/renderer_opengl.cpp
@@ -101,7 +101,6 @@ 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) {
@@ -130,6 +129,8 @@ void RendererOpenGL::SwapBuffers(
DrawScreen(render_window.GetFramebufferLayout());
+ rasterizer->TickFrame();
+
render_window.SwapBuffers();
}
@@ -262,7 +263,6 @@ void RendererOpenGL::CreateRasterizer() {
if (rasterizer) {
return;
}
- // Initialize sRGB Usage
OpenGLState::ClearsRGBUsed();
rasterizer = std::make_unique<RasterizerOpenGL>(system, emu_window, screen_info);
}
diff --git a/src/video_core/renderer_opengl/utils.cpp b/src/video_core/renderer_opengl/utils.cpp
index 68c36988d..c504a2c1a 100644
--- a/src/video_core/renderer_opengl/utils.cpp
+++ b/src/video_core/renderer_opengl/utils.cpp
@@ -13,29 +13,67 @@
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_;
- buffers.clear();
+ buffer_pointers.clear();
offsets.clear();
sizes.clear();
}
-void BindBuffersRangePushBuffer::Push(GLuint buffer, GLintptr offset, GLsizeiptr size) {
- buffers.push_back(buffer);
+void BindBuffersRangePushBuffer::Push(const GLuint* buffer, GLintptr offset, GLsizeiptr size) {
+ buffer_pointers.push_back(buffer);
offsets.push_back(offset);
sizes.push_back(size);
}
-void BindBuffersRangePushBuffer::Bind() const {
- const std::size_t count{buffers.size()};
+void BindBuffersRangePushBuffer::Bind() {
+ // Ensure sizes are valid.
+ const std::size_t count{buffer_pointers.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());
}
diff --git a/src/video_core/renderer_opengl/utils.h b/src/video_core/renderer_opengl/utils.h
index 4a752f3b4..6c2b45546 100644
--- a/src/video_core/renderer_opengl/utils.h
+++ b/src/video_core/renderer_opengl/utils.h
@@ -11,20 +11,49 @@
namespace OpenGL {
-class BindBuffersRangePushBuffer {
+class VertexArrayPushBuffer final {
public:
- BindBuffersRangePushBuffer(GLenum target);
+ 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();
void Setup(GLuint first_);
- void Push(GLuint buffer, GLintptr offset, GLsizeiptr size);
+ void Push(const GLuint* buffer, GLintptr offset, GLsizeiptr size);
- void Bind() const;
+ void Bind();
private:
- GLenum target;
- GLuint first;
+ GLenum target{};
+ GLuint first{};
+ std::vector<const GLuint*> buffer_pointers;
+
std::vector<GLuint> buffers;
std::vector<GLintptr> offsets;
std::vector<GLsizeiptr> sizes;
diff --git a/src/video_core/renderer_vulkan/vk_buffer_cache.cpp b/src/video_core/renderer_vulkan/vk_buffer_cache.cpp
index 02a9f5ecb..d2e9f4031 100644
--- a/src/video_core/renderer_vulkan/vk_buffer_cache.cpp
+++ b/src/video_core/renderer_vulkan/vk_buffer_cache.cpp
@@ -109,8 +109,8 @@ void VKBufferCache::Reserve(std::size_t max_size) {
}
}
-VKExecutionContext VKBufferCache::Send(VKExecutionContext exctx) {
- return stream_buffer->Send(exctx, buffer_offset - buffer_offset_base);
+void VKBufferCache::Send() {
+ stream_buffer->Send(buffer_offset - buffer_offset_base);
}
void VKBufferCache::AlignBuffer(std::size_t alignment) {
diff --git a/src/video_core/renderer_vulkan/vk_buffer_cache.h b/src/video_core/renderer_vulkan/vk_buffer_cache.h
index 3edf460df..49f13bcdc 100644
--- a/src/video_core/renderer_vulkan/vk_buffer_cache.h
+++ b/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.
- [[nodiscard]] VKExecutionContext Send(VKExecutionContext exctx);
+ void Send();
/// Returns the buffer cache handle.
vk::Buffer GetBuffer() const {
diff --git a/src/video_core/renderer_vulkan/vk_sampler_cache.h b/src/video_core/renderer_vulkan/vk_sampler_cache.h
index 771b05c73..1f73b716b 100644
--- a/src/video_core/renderer_vulkan/vk_sampler_cache.h
+++ b/src/video_core/renderer_vulkan/vk_sampler_cache.h
@@ -4,9 +4,6 @@
#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"
@@ -21,9 +18,9 @@ public:
~VKSamplerCache();
protected:
- UniqueSampler CreateSampler(const Tegra::Texture::TSCEntry& tsc) const;
+ UniqueSampler CreateSampler(const Tegra::Texture::TSCEntry& tsc) const override;
- vk::Sampler ToSamplerType(const UniqueSampler& sampler) const;
+ vk::Sampler ToSamplerType(const UniqueSampler& sampler) const override;
private:
const VKDevice& device;
diff --git a/src/video_core/renderer_vulkan/vk_scheduler.cpp b/src/video_core/renderer_vulkan/vk_scheduler.cpp
index f1fea1871..0f8116458 100644
--- a/src/video_core/renderer_vulkan/vk_scheduler.cpp
+++ b/src/video_core/renderer_vulkan/vk_scheduler.cpp
@@ -19,23 +19,19 @@ VKScheduler::VKScheduler(const VKDevice& device, VKResourceManager& resource_man
VKScheduler::~VKScheduler() = default;
-VKExecutionContext VKScheduler::GetExecutionContext() const {
- return VKExecutionContext(current_fence, current_cmdbuf);
-}
-
-VKExecutionContext VKScheduler::Flush(vk::Semaphore semaphore) {
+void VKScheduler::Flush(bool release_fence, vk::Semaphore semaphore) {
SubmitExecution(semaphore);
- current_fence->Release();
+ if (release_fence)
+ current_fence->Release();
AllocateNewContext();
- return GetExecutionContext();
}
-VKExecutionContext VKScheduler::Finish(vk::Semaphore semaphore) {
+void VKScheduler::Finish(bool release_fence, vk::Semaphore semaphore) {
SubmitExecution(semaphore);
current_fence->Wait();
- current_fence->Release();
+ if (release_fence)
+ current_fence->Release();
AllocateNewContext();
- return GetExecutionContext();
}
void VKScheduler::SubmitExecution(vk::Semaphore semaphore) {
diff --git a/src/video_core/renderer_vulkan/vk_scheduler.h b/src/video_core/renderer_vulkan/vk_scheduler.h
index cfaf5376f..0e5b49c7f 100644
--- a/src/video_core/renderer_vulkan/vk_scheduler.h
+++ b/src/video_core/renderer_vulkan/vk_scheduler.h
@@ -10,10 +10,43 @@
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 {
@@ -21,16 +54,21 @@ public:
explicit VKScheduler(const VKDevice& device, VKResourceManager& resource_manager);
~VKScheduler();
- /// Gets the current execution context.
- [[nodiscard]] VKExecutionContext GetExecutionContext() const;
+ /// Gets a reference to the current fence.
+ VKFenceView GetFence() const {
+ return current_fence;
+ }
+
+ /// 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. It invalidates
- /// the current execution context and returns a new one.
- VKExecutionContext Finish(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);
private:
void SubmitExecution(vk::Semaphore semaphore);
@@ -44,26 +82,4 @@ 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
diff --git a/src/video_core/renderer_vulkan/vk_shader_decompiler.cpp b/src/video_core/renderer_vulkan/vk_shader_decompiler.cpp
index 6071c6d99..9b2d8e987 100644
--- a/src/video_core/renderer_vulkan/vk_shader_decompiler.cpp
+++ b/src/video_core/renderer_vulkan/vk_shader_decompiler.cpp
@@ -947,6 +947,14 @@ 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);
@@ -1332,6 +1340,7 @@ private:
&SPIRVDecompiler::ImageStore,
&SPIRVDecompiler::Branch,
+ &SPIRVDecompiler::BranchIndirect,
&SPIRVDecompiler::PushFlowStack,
&SPIRVDecompiler::PopFlowStack,
&SPIRVDecompiler::Exit,
diff --git a/src/video_core/renderer_vulkan/vk_stream_buffer.cpp b/src/video_core/renderer_vulkan/vk_stream_buffer.cpp
index 58ffa42f2..62f1427f5 100644
--- a/src/video_core/renderer_vulkan/vk_stream_buffer.cpp
+++ b/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()};
}
-VKExecutionContext VKStreamBuffer::Send(VKExecutionContext exctx, u64 size) {
+void VKStreamBuffer::Send(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.
- exctx = scheduler.Flush();
+ scheduler.Flush();
std::for_each(watches.begin(), watches.begin() + *invalidation_mark,
[&](auto& resource) { resource->Wait(); });
invalidation_mark = std::nullopt;
@@ -62,11 +62,9 @@ VKExecutionContext VKStreamBuffer::Send(VKExecutionContext exctx, u64 size) {
ReserveWatches(WATCHES_RESERVE_CHUNK);
}
// Add a watch for this allocation.
- watches[used_watches++]->Watch(exctx.GetFence());
+ watches[used_watches++]->Watch(scheduler.GetFence());
offset += size;
-
- return exctx;
}
void VKStreamBuffer::CreateBuffers(VKMemoryManager& memory_manager, vk::BufferUsageFlags usage) {
diff --git a/src/video_core/renderer_vulkan/vk_stream_buffer.h b/src/video_core/renderer_vulkan/vk_stream_buffer.h
index 69d036ccd..842e54162 100644
--- a/src/video_core/renderer_vulkan/vk_stream_buffer.h
+++ b/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.
- [[nodiscard]] VKExecutionContext Send(VKExecutionContext exctx, u64 size);
+ void Send(u64 size);
vk::Buffer GetBuffer() const {
return *buffer;
diff --git a/src/video_core/shader/control_flow.cpp b/src/video_core/shader/control_flow.cpp
new file mode 100644
index 000000000..fdcc970ff
--- /dev/null
+++ b/src/video_core/shader/control_flow.cpp
@@ -0,0 +1,476 @@
+// 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
diff --git a/src/video_core/shader/control_flow.h b/src/video_core/shader/control_flow.h
new file mode 100644
index 000000000..5e8ea3271
--- /dev/null
+++ b/src/video_core/shader/control_flow.h
@@ -0,0 +1,63 @@
+// Copyright 2019 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <cstring>
+#include <list>
+#include <optional>
+#include <unordered_set>
+
+#include "video_core/engines/shader_bytecode.h"
+#include "video_core/shader/shader_ir.h"
+
+namespace VideoCommon::Shader {
+
+using Tegra::Shader::ConditionCode;
+using Tegra::Shader::Pred;
+
+constexpr s32 exit_branch = -1;
+
+struct Condition {
+ Pred predicate{Pred::UnusedIndex};
+ ConditionCode cc{ConditionCode::T};
+
+ bool IsUnconditional() const {
+ return predicate == Pred::UnusedIndex && cc == ConditionCode::T;
+ }
+ bool operator==(const Condition& other) const {
+ return std::tie(predicate, cc) == std::tie(other.predicate, other.cc);
+ }
+};
+
+struct ShaderBlock {
+ u32 start{};
+ u32 end{};
+ bool ignore_branch{};
+ struct Branch {
+ Condition cond{};
+ bool kills{};
+ s32 address{};
+ bool operator==(const Branch& b) const {
+ return std::tie(cond, kills, address) == std::tie(b.cond, b.kills, b.address);
+ }
+ } branch{};
+ bool operator==(const ShaderBlock& sb) const {
+ return std::tie(start, end, ignore_branch, branch) ==
+ std::tie(sb.start, sb.end, sb.ignore_branch, sb.branch);
+ }
+};
+
+struct ShaderCharacteristics {
+ std::list<ShaderBlock> blocks{};
+ bool decompilable{};
+ u32 start{};
+ u32 end{};
+ std::unordered_set<u32> labels{};
+};
+
+std::optional<ShaderCharacteristics> ScanFlow(const ProgramCode& program_code, u32 program_size,
+ u32 start_address);
+
+} // namespace VideoCommon::Shader
diff --git a/src/video_core/shader/decode.cpp b/src/video_core/shader/decode.cpp
index 2c9ff28f2..29c8895c5 100644
--- a/src/video_core/shader/decode.cpp
+++ b/src/video_core/shader/decode.cpp
@@ -11,6 +11,7 @@
#include "common/common_types.h"
#include "video_core/engines/shader_bytecode.h"
#include "video_core/engines/shader_header.h"
+#include "video_core/shader/control_flow.h"
#include "video_core/shader/node_helper.h"
#include "video_core/shader/shader_ir.h"
@@ -21,20 +22,6 @@ using Tegra::Shader::OpCode;
namespace {
-/// Merges exit method of two parallel branches.
-constexpr ExitMethod ParallelExit(ExitMethod a, ExitMethod b) {
- if (a == ExitMethod::Undetermined) {
- return b;
- }
- if (b == ExitMethod::Undetermined) {
- return a;
- }
- if (a == b) {
- return a;
- }
- return ExitMethod::Conditional;
-}
-
/**
* Returns whether the instruction at the specified offset is a 'sched' instruction.
* Sched instructions always appear before a sequence of 3 instructions.
@@ -51,85 +38,104 @@ constexpr bool IsSchedInstruction(u32 offset, u32 main_offset) {
void ShaderIR::Decode() {
std::memcpy(&header, program_code.data(), sizeof(Tegra::Shader::Header));
- std::set<u32> labels;
- const ExitMethod exit_method = Scan(main_offset, MAX_PROGRAM_LENGTH, labels);
- if (exit_method != ExitMethod::AlwaysEnd) {
- UNREACHABLE_MSG("Program does not always end");
- }
-
- if (labels.empty()) {
- basic_blocks.insert({main_offset, DecodeRange(main_offset, MAX_PROGRAM_LENGTH)});
+ disable_flow_stack = false;
+ const auto info = ScanFlow(program_code, program_size, main_offset);
+ if (info) {
+ const auto& shader_info = *info;
+ coverage_begin = shader_info.start;
+ coverage_end = shader_info.end;
+ if (shader_info.decompilable) {
+ disable_flow_stack = true;
+ const auto insert_block = ([this](NodeBlock& nodes, u32 label) {
+ if (label == exit_branch) {
+ return;
+ }
+ basic_blocks.insert({label, nodes});
+ });
+ const auto& blocks = shader_info.blocks;
+ NodeBlock current_block;
+ u32 current_label = exit_branch;
+ for (auto& block : blocks) {
+ if (shader_info.labels.count(block.start) != 0) {
+ insert_block(current_block, current_label);
+ current_block.clear();
+ current_label = block.start;
+ }
+ if (!block.ignore_branch) {
+ DecodeRangeInner(current_block, block.start, block.end);
+ InsertControlFlow(current_block, block);
+ } else {
+ DecodeRangeInner(current_block, block.start, block.end + 1);
+ }
+ }
+ insert_block(current_block, current_label);
+ return;
+ }
+ LOG_WARNING(HW_GPU, "Flow Stack Removing Failed! Falling back to old method");
+ // we can't decompile it, fallback to standard method
+ for (const auto& block : shader_info.blocks) {
+ basic_blocks.insert({block.start, DecodeRange(block.start, block.end + 1)});
+ }
return;
}
+ LOG_WARNING(HW_GPU, "Flow Analysis Failed! Falling back to brute force compiling");
+
+ // Now we need to deal with an undecompilable shader. We need to brute force
+ // a shader that captures every position.
+ coverage_begin = main_offset;
+ const u32 shader_end = static_cast<u32>(program_size / sizeof(u64));
+ coverage_end = shader_end;
+ for (u32 label = main_offset; label < shader_end; label++) {
+ basic_blocks.insert({label, DecodeRange(label, label + 1)});
+ }
+}
- labels.insert(main_offset);
-
- for (const u32 label : labels) {
- const auto next_it = labels.lower_bound(label + 1);
- const u32 next_label = next_it == labels.end() ? MAX_PROGRAM_LENGTH : *next_it;
+NodeBlock ShaderIR::DecodeRange(u32 begin, u32 end) {
+ NodeBlock basic_block;
+ DecodeRangeInner(basic_block, begin, end);
+ return basic_block;
+}
- basic_blocks.insert({label, DecodeRange(label, next_label)});
+void ShaderIR::DecodeRangeInner(NodeBlock& bb, u32 begin, u32 end) {
+ for (u32 pc = begin; pc < (begin > end ? MAX_PROGRAM_LENGTH : end);) {
+ pc = DecodeInstr(bb, pc);
}
}
-ExitMethod ShaderIR::Scan(u32 begin, u32 end, std::set<u32>& labels) {
- const auto [iter, inserted] =
- exit_method_map.emplace(std::make_pair(begin, end), ExitMethod::Undetermined);
- ExitMethod& exit_method = iter->second;
- if (!inserted)
- return exit_method;
-
- for (u32 offset = begin; offset != end && offset != MAX_PROGRAM_LENGTH; ++offset) {
- coverage_begin = std::min(coverage_begin, offset);
- coverage_end = std::max(coverage_end, offset + 1);
-
- const Instruction instr = {program_code[offset]};
- const auto opcode = OpCode::Decode(instr);
- if (!opcode)
- continue;
- switch (opcode->get().GetId()) {
- case OpCode::Id::EXIT: {
- // The EXIT instruction can be predicated, which means that the shader can conditionally
- // end on this instruction. We have to consider the case where the condition is not met
- // and check the exit method of that other basic block.
- using Tegra::Shader::Pred;
- if (instr.pred.pred_index == static_cast<u64>(Pred::UnusedIndex)) {
- return exit_method = ExitMethod::AlwaysEnd;
- } else {
- const ExitMethod not_met = Scan(offset + 1, end, labels);
- return exit_method = ParallelExit(ExitMethod::AlwaysEnd, not_met);
- }
+void ShaderIR::InsertControlFlow(NodeBlock& bb, const ShaderBlock& block) {
+ const auto apply_conditions = ([&](const Condition& cond, Node n) -> Node {
+ Node result = n;
+ if (cond.cc != ConditionCode::T) {
+ result = Conditional(GetConditionCode(cond.cc), {result});
}
- case OpCode::Id::BRA: {
- const u32 target = offset + instr.bra.GetBranchTarget();
- labels.insert(target);
- const ExitMethod no_jmp = Scan(offset + 1, end, labels);
- const ExitMethod jmp = Scan(target, end, labels);
- return exit_method = ParallelExit(no_jmp, jmp);
- }
- case OpCode::Id::SSY:
- case OpCode::Id::PBK: {
- // The SSY and PBK use a similar encoding as the BRA instruction.
- UNIMPLEMENTED_IF_MSG(instr.bra.constant_buffer != 0,
- "Constant buffer branching is not supported");
- const u32 target = offset + instr.bra.GetBranchTarget();
- labels.insert(target);
- // Continue scanning for an exit method.
- break;
+ if (cond.predicate != Pred::UnusedIndex) {
+ u32 pred = static_cast<u32>(cond.predicate);
+ const bool is_neg = pred > 7;
+ if (is_neg) {
+ pred -= 8;
+ }
+ result = Conditional(GetPredicate(pred, is_neg), {result});
}
- default:
- break;
+ return result;
+ });
+ if (block.branch.address < 0) {
+ if (block.branch.kills) {
+ Node n = Operation(OperationCode::Discard);
+ n = apply_conditions(block.branch.cond, n);
+ bb.push_back(n);
+ global_code.push_back(n);
+ return;
}
+ Node n = Operation(OperationCode::Exit);
+ n = apply_conditions(block.branch.cond, n);
+ bb.push_back(n);
+ global_code.push_back(n);
+ return;
}
- return exit_method = ExitMethod::AlwaysReturn;
-}
-
-NodeBlock ShaderIR::DecodeRange(u32 begin, u32 end) {
- NodeBlock basic_block;
- for (u32 pc = begin; pc < (begin > end ? MAX_PROGRAM_LENGTH : end);) {
- pc = DecodeInstr(basic_block, pc);
- }
- return basic_block;
+ Node n = Operation(OperationCode::Branch, Immediate(block.branch.address));
+ n = apply_conditions(block.branch.cond, n);
+ bb.push_back(n);
+ global_code.push_back(n);
}
u32 ShaderIR::DecodeInstr(NodeBlock& bb, u32 pc) {
@@ -140,15 +146,18 @@ u32 ShaderIR::DecodeInstr(NodeBlock& bb, u32 pc) {
const Instruction instr = {program_code[pc]};
const auto opcode = OpCode::Decode(instr);
+ const u32 nv_address = ConvertAddressToNvidiaSpace(pc);
// Decoding failure
if (!opcode) {
UNIMPLEMENTED_MSG("Unhandled instruction: {0:x}", instr.value);
+ bb.push_back(Comment(fmt::format("{:05x} Unimplemented Shader instruction (0x{:016x})",
+ nv_address, instr.value)));
return pc + 1;
}
- bb.push_back(
- Comment(fmt::format("{}: {} (0x{:016x})", pc, opcode->get().GetName(), instr.value)));
+ bb.push_back(Comment(
+ fmt::format("{:05x} {} (0x{:016x})", nv_address, opcode->get().GetName(), instr.value)));
using Tegra::Shader::Pred;
UNIMPLEMENTED_IF_MSG(instr.pred.full_pred == Pred::NeverExecute,
diff --git a/src/video_core/shader/decode/other.cpp b/src/video_core/shader/decode/other.cpp
index d46a8ab82..42e3de02f 100644
--- a/src/video_core/shader/decode/other.cpp
+++ b/src/video_core/shader/decode/other.cpp
@@ -91,11 +91,46 @@ u32 ShaderIR::DecodeOther(NodeBlock& bb, u32 pc) {
break;
}
case OpCode::Id::BRA: {
- UNIMPLEMENTED_IF_MSG(instr.bra.constant_buffer != 0,
- "BRA with constant buffers are not implemented");
+ Node branch;
+ if (instr.bra.constant_buffer == 0) {
+ const u32 target = pc + instr.bra.GetBranchTarget();
+ branch = Operation(OperationCode::Branch, Immediate(target));
+ } else {
+ const u32 target = pc + 1;
+ const Node op_a = GetConstBuffer(instr.cbuf36.index, instr.cbuf36.GetOffset());
+ const Node convert = SignedOperation(OperationCode::IArithmeticShiftRight, true,
+ PRECISE, op_a, Immediate(3));
+ const Node operand =
+ Operation(OperationCode::IAdd, PRECISE, convert, Immediate(target));
+ branch = Operation(OperationCode::BranchIndirect, convert);
+ }
- const u32 target = pc + instr.bra.GetBranchTarget();
- const Node branch = Operation(OperationCode::Branch, Immediate(target));
+ const Tegra::Shader::ConditionCode cc = instr.flow_condition_code;
+ if (cc != Tegra::Shader::ConditionCode::T) {
+ bb.push_back(Conditional(GetConditionCode(cc), {branch}));
+ } else {
+ bb.push_back(branch);
+ }
+ break;
+ }
+ case OpCode::Id::BRX: {
+ Node operand;
+ if (instr.brx.constant_buffer != 0) {
+ const s32 target = pc + 1;
+ const Node index = GetRegister(instr.gpr8);
+ const Node op_a =
+ GetConstBufferIndirect(instr.cbuf36.index, instr.cbuf36.GetOffset() + 0, index);
+ const Node convert = SignedOperation(OperationCode::IArithmeticShiftRight, true,
+ PRECISE, op_a, Immediate(3));
+ operand = Operation(OperationCode::IAdd, PRECISE, convert, Immediate(target));
+ } else {
+ const s32 target = pc + instr.brx.GetBranchExtend();
+ const Node op_a = GetRegister(instr.gpr8);
+ const Node convert = SignedOperation(OperationCode::IArithmeticShiftRight, true,
+ PRECISE, op_a, Immediate(3));
+ operand = Operation(OperationCode::IAdd, PRECISE, convert, Immediate(target));
+ }
+ const Node branch = Operation(OperationCode::BranchIndirect, operand);
const Tegra::Shader::ConditionCode cc = instr.flow_condition_code;
if (cc != Tegra::Shader::ConditionCode::T) {
@@ -109,6 +144,10 @@ u32 ShaderIR::DecodeOther(NodeBlock& bb, u32 pc) {
UNIMPLEMENTED_IF_MSG(instr.bra.constant_buffer != 0,
"Constant buffer flow is not supported");
+ if (disable_flow_stack) {
+ break;
+ }
+
// The SSY opcode tells the GPU where to re-converge divergent execution paths with SYNC.
const u32 target = pc + instr.bra.GetBranchTarget();
bb.push_back(
@@ -119,6 +158,10 @@ u32 ShaderIR::DecodeOther(NodeBlock& bb, u32 pc) {
UNIMPLEMENTED_IF_MSG(instr.bra.constant_buffer != 0,
"Constant buffer PBK is not supported");
+ if (disable_flow_stack) {
+ break;
+ }
+
// PBK pushes to a stack the address where BRK will jump to.
const u32 target = pc + instr.bra.GetBranchTarget();
bb.push_back(
@@ -130,6 +173,10 @@ u32 ShaderIR::DecodeOther(NodeBlock& bb, u32 pc) {
UNIMPLEMENTED_IF_MSG(cc != Tegra::Shader::ConditionCode::T, "SYNC condition code used: {}",
static_cast<u32>(cc));
+ if (disable_flow_stack) {
+ break;
+ }
+
// The SYNC opcode jumps to the address previously set by the SSY opcode
bb.push_back(Operation(OperationCode::PopFlowStack, MetaStackClass::Ssy));
break;
@@ -138,6 +185,9 @@ u32 ShaderIR::DecodeOther(NodeBlock& bb, u32 pc) {
const Tegra::Shader::ConditionCode cc = instr.flow_condition_code;
UNIMPLEMENTED_IF_MSG(cc != Tegra::Shader::ConditionCode::T, "BRK condition code used: {}",
static_cast<u32>(cc));
+ if (disable_flow_stack) {
+ break;
+ }
// The BRK opcode jumps to the address previously set by the PBK opcode
bb.push_back(Operation(OperationCode::PopFlowStack, MetaStackClass::Pbk));
diff --git a/src/video_core/shader/decode/texture.cpp b/src/video_core/shader/decode/texture.cpp
index cb480be9b..323be3f14 100644
--- a/src/video_core/shader/decode/texture.cpp
+++ b/src/video_core/shader/decode/texture.cpp
@@ -269,7 +269,13 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) {
LOG_WARNING(HW_GPU, "TLDS.NODEP implementation is incomplete");
}
- WriteTexsInstructionFloat(bb, instr, GetTldsCode(instr, texture_type, is_array));
+ const Node4 components = GetTldsCode(instr, texture_type, is_array);
+
+ if (instr.tlds.fp32_flag) {
+ WriteTexsInstructionFloat(bb, instr, components);
+ } else {
+ WriteTexsInstructionHalfFloat(bb, instr, components);
+ }
break;
}
default:
diff --git a/src/video_core/shader/node.h b/src/video_core/shader/node.h
index 0ac83fcf0..7427ed896 100644
--- a/src/video_core/shader/node.h
+++ b/src/video_core/shader/node.h
@@ -148,11 +148,12 @@ enum class OperationCode {
ImageStore, /// (MetaImage, float[N] coords) -> void
- Branch, /// (uint branch_target) -> void
- PushFlowStack, /// (uint branch_target) -> void
- PopFlowStack, /// () -> void
- Exit, /// () -> void
- Discard, /// () -> void
+ Branch, /// (uint branch_target) -> void
+ BranchIndirect, /// (uint branch_target) -> void
+ PushFlowStack, /// (uint branch_target) -> void
+ PopFlowStack, /// () -> void
+ Exit, /// () -> void
+ Discard, /// () -> void
EmitVertex, /// () -> void
EndPrimitive, /// () -> void
diff --git a/src/video_core/shader/shader_ir.cpp b/src/video_core/shader/shader_ir.cpp
index fd80dd116..caa409788 100644
--- a/src/video_core/shader/shader_ir.cpp
+++ b/src/video_core/shader/shader_ir.cpp
@@ -22,8 +22,8 @@ using Tegra::Shader::PredCondition;
using Tegra::Shader::PredOperation;
using Tegra::Shader::Register;
-ShaderIR::ShaderIR(const ProgramCode& program_code, u32 main_offset)
- : program_code{program_code}, main_offset{main_offset} {
+ShaderIR::ShaderIR(const ProgramCode& program_code, u32 main_offset, const std::size_t size)
+ : program_code{program_code}, main_offset{main_offset}, program_size{size} {
Decode();
}
diff --git a/src/video_core/shader/shader_ir.h b/src/video_core/shader/shader_ir.h
index a656761ea..03c888def 100644
--- a/src/video_core/shader/shader_ir.h
+++ b/src/video_core/shader/shader_ir.h
@@ -22,18 +22,12 @@
namespace VideoCommon::Shader {
+struct ShaderBlock;
+
using ProgramCode = std::vector<u64>;
constexpr u32 MAX_PROGRAM_LENGTH = 0x1000;
-/// Describes the behaviour of code path of a given entry point and a return point.
-enum class ExitMethod {
- Undetermined, ///< Internal value. Only occur when analyzing JMP loop.
- AlwaysReturn, ///< All code paths reach the return point.
- Conditional, ///< Code path reaches the return point or an END instruction conditionally.
- AlwaysEnd, ///< All code paths reach a END instruction.
-};
-
class ConstBuffer {
public:
explicit ConstBuffer(u32 max_offset, bool is_indirect)
@@ -73,7 +67,7 @@ struct GlobalMemoryUsage {
class ShaderIR final {
public:
- explicit ShaderIR(const ProgramCode& program_code, u32 main_offset);
+ explicit ShaderIR(const ProgramCode& program_code, u32 main_offset, std::size_t size);
~ShaderIR();
const std::map<u32, NodeBlock>& GetBasicBlocks() const {
@@ -141,12 +135,20 @@ public:
return header;
}
+ bool IsFlowStackDisabled() const {
+ return disable_flow_stack;
+ }
+
+ u32 ConvertAddressToNvidiaSpace(const u32 address) const {
+ return (address - main_offset) * sizeof(Tegra::Shader::Instruction);
+ }
+
private:
void Decode();
- ExitMethod Scan(u32 begin, u32 end, std::set<u32>& labels);
-
NodeBlock DecodeRange(u32 begin, u32 end);
+ void DecodeRangeInner(NodeBlock& bb, u32 begin, u32 end);
+ void InsertControlFlow(NodeBlock& bb, const ShaderBlock& block);
/**
* Decodes a single instruction from Tegra to IR.
@@ -338,10 +340,11 @@ private:
const ProgramCode& program_code;
const u32 main_offset;
+ const std::size_t program_size;
+ bool disable_flow_stack{};
u32 coverage_begin{};
u32 coverage_end{};
- std::map<std::pair<u32, u32>, ExitMethod> exit_method_map;
std::map<u32, NodeBlock> basic_blocks;
NodeBlock global_code;
diff --git a/src/video_core/texture_cache/surface_base.cpp b/src/video_core/texture_cache/surface_base.cpp
index 7a0fdb19b..6af9044ca 100644
--- a/src/video_core/texture_cache/surface_base.cpp
+++ b/src/video_core/texture_cache/surface_base.cpp
@@ -75,9 +75,12 @@ MatchStructureResult SurfaceBaseImpl::MatchesStructure(const SurfaceParams& rhs)
// Linear Surface check
if (!params.is_tiled) {
- if (std::tie(params.width, params.height, params.pitch) ==
- std::tie(rhs.width, rhs.height, rhs.pitch)) {
- return MatchStructureResult::FullMatch;
+ if (std::tie(params.height, params.pitch) == std::tie(rhs.height, rhs.pitch)) {
+ if (params.width == rhs.width) {
+ return MatchStructureResult::FullMatch;
+ } else {
+ return MatchStructureResult::SemiMatch;
+ }
}
return MatchStructureResult::None;
}
diff --git a/src/video_core/texture_cache/surface_base.h b/src/video_core/texture_cache/surface_base.h
index 8ba386a8a..bcce8d863 100644
--- a/src/video_core/texture_cache/surface_base.h
+++ b/src/video_core/texture_cache/surface_base.h
@@ -200,8 +200,9 @@ public:
modification_tick = tick;
}
- void MarkAsRenderTarget(const bool is_target) {
+ void MarkAsRenderTarget(const bool is_target, const u32 index) {
this->is_target = is_target;
+ this->index = index;
}
void MarkAsPicked(const bool is_picked) {
@@ -221,6 +222,10 @@ public:
return is_target;
}
+ u32 GetRenderTarget() const {
+ return index;
+ }
+
bool IsRegistered() const {
return is_registered;
}
@@ -307,10 +312,13 @@ private:
return view;
}
+ static constexpr u32 NO_RT = 0xFFFFFFFF;
+
bool is_modified{};
bool is_target{};
bool is_registered{};
bool is_picked{};
+ u32 index{NO_RT};
u64 modification_tick{};
};
diff --git a/src/video_core/texture_cache/surface_params.cpp b/src/video_core/texture_cache/surface_params.cpp
index 9c56e2b4f..fd5472451 100644
--- a/src/video_core/texture_cache/surface_params.cpp
+++ b/src/video_core/texture_cache/surface_params.cpp
@@ -290,12 +290,19 @@ std::size_t SurfaceParams::GetLayerSize(bool as_host_size, bool uncompressed) co
std::size_t SurfaceParams::GetInnerMipmapMemorySize(u32 level, bool as_host_size,
bool uncompressed) const {
- const bool tiled{as_host_size ? false : is_tiled};
const u32 width{GetMipmapSize(uncompressed, GetMipWidth(level), GetDefaultBlockWidth())};
const u32 height{GetMipmapSize(uncompressed, GetMipHeight(level), GetDefaultBlockHeight())};
const u32 depth{is_layered ? 1U : GetMipDepth(level)};
- return Tegra::Texture::CalculateSize(tiled, GetBytesPerPixel(), width, height, depth,
- GetMipBlockHeight(level), GetMipBlockDepth(level));
+ if (is_tiled) {
+ return Tegra::Texture::CalculateSize(!as_host_size, GetBytesPerPixel(), width, height,
+ depth, GetMipBlockHeight(level),
+ GetMipBlockDepth(level));
+ } else if (as_host_size || IsBuffer()) {
+ return GetBytesPerPixel() * width * height * depth;
+ } else {
+ // Linear Texture Case
+ return pitch * height * depth;
+ }
}
bool SurfaceParams::operator==(const SurfaceParams& rhs) const {
diff --git a/src/video_core/texture_cache/texture_cache.h b/src/video_core/texture_cache/texture_cache.h
index c9e72531a..7f9623c62 100644
--- a/src/video_core/texture_cache/texture_cache.h
+++ b/src/video_core/texture_cache/texture_cache.h
@@ -133,11 +133,11 @@ public:
regs.zeta.memory_layout.block_depth, regs.zeta.memory_layout.type)};
auto surface_view = GetSurface(gpu_addr, depth_params, preserve_contents, true);
if (depth_buffer.target)
- depth_buffer.target->MarkAsRenderTarget(false);
+ depth_buffer.target->MarkAsRenderTarget(false, NO_RT);
depth_buffer.target = surface_view.first;
depth_buffer.view = surface_view.second;
if (depth_buffer.target)
- depth_buffer.target->MarkAsRenderTarget(true);
+ depth_buffer.target->MarkAsRenderTarget(true, DEPTH_RT);
return surface_view.second;
}
@@ -167,11 +167,11 @@ public:
auto surface_view = GetSurface(gpu_addr, SurfaceParams::CreateForFramebuffer(system, index),
preserve_contents, true);
if (render_targets[index].target)
- render_targets[index].target->MarkAsRenderTarget(false);
+ render_targets[index].target->MarkAsRenderTarget(false, NO_RT);
render_targets[index].target = surface_view.first;
render_targets[index].view = surface_view.second;
if (render_targets[index].target)
- render_targets[index].target->MarkAsRenderTarget(true);
+ render_targets[index].target->MarkAsRenderTarget(true, static_cast<u32>(index));
return surface_view.second;
}
@@ -191,7 +191,7 @@ public:
if (depth_buffer.target == nullptr) {
return;
}
- depth_buffer.target->MarkAsRenderTarget(false);
+ depth_buffer.target->MarkAsRenderTarget(false, NO_RT);
depth_buffer.target = nullptr;
depth_buffer.view = nullptr;
}
@@ -200,7 +200,7 @@ public:
if (render_targets[index].target == nullptr) {
return;
}
- render_targets[index].target->MarkAsRenderTarget(false);
+ render_targets[index].target->MarkAsRenderTarget(false, NO_RT);
render_targets[index].target = nullptr;
render_targets[index].view = nullptr;
}
@@ -270,6 +270,16 @@ protected:
// and reading it from a sepparate buffer.
virtual void BufferCopy(TSurface& src_surface, TSurface& dst_surface) = 0;
+ void ManageRenderTargetUnregister(TSurface& surface) {
+ auto& maxwell3d = system.GPU().Maxwell3D();
+ const u32 index = surface->GetRenderTarget();
+ if (index == DEPTH_RT) {
+ maxwell3d.dirty_flags.zeta_buffer = true;
+ } else {
+ maxwell3d.dirty_flags.color_buffer.set(index, true);
+ }
+ }
+
void Register(TSurface surface) {
const GPUVAddr gpu_addr = surface->GetGpuAddr();
const CacheAddr cache_ptr = ToCacheAddr(system.GPU().MemoryManager().GetPointer(gpu_addr));
@@ -294,6 +304,9 @@ protected:
if (guard_render_targets && surface->IsProtected()) {
return;
}
+ if (!guard_render_targets && surface->IsRenderTarget()) {
+ ManageRenderTargetUnregister(surface);
+ }
const GPUVAddr gpu_addr = surface->GetGpuAddr();
const CacheAddr cache_ptr = surface->GetCacheAddr();
const std::size_t size = surface->GetSizeInBytes();
@@ -649,15 +662,6 @@ private:
}
return {current_surface, *view};
}
- // The next case is unsafe, so if we r in accurate GPU, just skip it
- if (Settings::values.use_accurate_gpu_emulation) {
- return RecycleSurface(overlaps, params, gpu_addr, preserve_contents,
- MatchTopologyResult::FullMatch);
- }
- // This is the case the texture is a part of the parent.
- if (current_surface->MatchesSubTexture(params, gpu_addr)) {
- return RebuildSurface(current_surface, params, is_render);
- }
} else {
// If there are many overlaps, odds are they are subtextures of the candidate
// surface. We try to construct a new surface based on the candidate parameters,
@@ -793,6 +797,9 @@ private:
static constexpr u64 registry_page_size{1 << registry_page_bits};
std::unordered_map<CacheAddr, std::vector<TSurface>> registry;
+ static constexpr u32 DEPTH_RT = 8;
+ static constexpr u32 NO_RT = 0xFFFFFFFF;
+
// The L1 Cache is used for fast texture lookup before checking the overlaps
// This avoids calculating size and other stuffs.
std::unordered_map<CacheAddr, TSurface> l1_cache;
diff --git a/src/yuzu/configuration/config.cpp b/src/yuzu/configuration/config.cpp
index 73978ff5b..b7f3fdf75 100644
--- a/src/yuzu/configuration/config.cpp
+++ b/src/yuzu/configuration/config.cpp
@@ -436,8 +436,6 @@ void Config::ReadControlValues() {
void Config::ReadCoreValues() {
qt_config->beginGroup(QStringLiteral("Core"));
- Settings::values.cpu_jit_enabled =
- ReadSetting(QStringLiteral("cpu_jit_enabled"), true).toBool();
Settings::values.use_multi_core = ReadSetting(QStringLiteral("use_multi_core"), false).toBool();
qt_config->endGroup();
@@ -831,7 +829,6 @@ void Config::SaveControlValues() {
void Config::SaveCoreValues() {
qt_config->beginGroup(QStringLiteral("Core"));
- WriteSetting(QStringLiteral("cpu_jit_enabled"), Settings::values.cpu_jit_enabled, true);
WriteSetting(QStringLiteral("use_multi_core"), Settings::values.use_multi_core, false);
qt_config->endGroup();
diff --git a/src/yuzu_cmd/config.cpp b/src/yuzu_cmd/config.cpp
index 30b22341b..067d58d80 100644
--- a/src/yuzu_cmd/config.cpp
+++ b/src/yuzu_cmd/config.cpp
@@ -340,7 +340,6 @@ void Config::ReadValues() {
}
// Core
- Settings::values.cpu_jit_enabled = sdl2_config->GetBoolean("Core", "cpu_jit_enabled", true);
Settings::values.use_multi_core = sdl2_config->GetBoolean("Core", "use_multi_core", false);
// Renderer
diff --git a/src/yuzu_cmd/default_ini.h b/src/yuzu_cmd/default_ini.h
index 4f1add434..0cfc111a6 100644
--- a/src/yuzu_cmd/default_ini.h
+++ b/src/yuzu_cmd/default_ini.h
@@ -76,10 +76,6 @@ motion_device=
touch_device=
[Core]
-# Whether to use the Just-In-Time (JIT) compiler for CPU emulation
-# 0: Interpreter (slow), 1 (default): JIT (fast)
-cpu_jit_enabled =
-
# Whether to use multi-core for CPU emulation
# 0 (default): Disabled, 1: Enabled
use_multi_core=
diff --git a/src/yuzu_tester/config.cpp b/src/yuzu_tester/config.cpp
index b96b7d279..9a11dc6c3 100644
--- a/src/yuzu_tester/config.cpp
+++ b/src/yuzu_tester/config.cpp
@@ -114,7 +114,6 @@ void Config::ReadValues() {
}
// Core
- Settings::values.cpu_jit_enabled = sdl2_config->GetBoolean("Core", "cpu_jit_enabled", true);
Settings::values.use_multi_core = sdl2_config->GetBoolean("Core", "use_multi_core", false);
// Renderer
diff --git a/src/yuzu_tester/default_ini.h b/src/yuzu_tester/default_ini.h
index 0f880d8c7..9a3e86d68 100644
--- a/src/yuzu_tester/default_ini.h
+++ b/src/yuzu_tester/default_ini.h
@@ -8,10 +8,6 @@ namespace DefaultINI {
const char* sdl2_config_file = R"(
[Core]
-# Whether to use the Just-In-Time (JIT) compiler for CPU emulation
-# 0: Interpreter (slow), 1 (default): JIT (fast)
-cpu_jit_enabled =
-
# Whether to use multi-core for CPU emulation
# 0 (default): Disabled, 1: Enabled
use_multi_core=