Compare commits

..

88 Commits

Author SHA1 Message Date
Fernando Sahmkow
f9b940a442 Reaper: Set minimum cleaning limit on OGL. 2021-06-22 22:07:17 +02:00
Fernando Sahmkow
569a1962c0 Reaper: Guarantee correct deletion. 2021-06-20 19:11:41 +02:00
Fernando Sahmkow
865dd615ca Reaper: Upgrade label from unsafe to experimental as no regressions are known now. 2021-06-20 12:35:19 +02:00
Fernando Sahmkow
719a6dd5a1 Reaper: Correct size calculation on Vulkan. 2021-06-17 08:48:41 +02:00
Fernando Sahmkow
ca6f47c686 Reaper: Change memory restrictions on TC depending on host memory on VK. 2021-06-17 00:29:48 +02:00
Fernando Sahmkow
0dd98842bf Reaper: Address Feedback. 2021-06-16 21:35:03 +02:00
Fernando Sahmkow
954ad2a61e Reaper: Setup settings and final tuning. 2021-06-16 21:35:03 +02:00
Fernando Sahmkow
d8ad6aa187 Reaper: Tune it up to be an smart GC. 2021-06-16 21:35:02 +02:00
ReinUsesLisp
a11bc4a382 Initial Reaper Setup
WIP
2021-06-16 21:35:02 +02:00
ReinUsesLisp
5b1efe522e vulkan_memory_allocator: Release allocations with no commits 2021-06-16 21:35:01 +02:00
bunnei
973bf306ed Merge pull request #6464 from ameerj/disable-astc
textures: Add a toggle for GPU Accelerated ASTC decoder
2021-06-16 11:29:10 -07:00
Morph
92942fe01b Merge pull request #6460 from Morph1984/fs-access-log-fix
fsp_srv: Fix filesystem access logging
2021-06-16 14:03:01 -04:00
Morph
74790e4f33 common: fs: file: Remove redundant call to WriteStringToFile
The Append open mode will create a new file if said file does not exist at a given path, making this call redundant.
2021-06-16 00:06:02 -04:00
Morph
0f48292de1 fsp_srv: Fix filesystem access logging
This introduces a new setting Enable FS Access Log which saves the filesystem access log to sdmc:/FsAccessLog.txt

If this setting is not enabled, this will indicate to FS to not call OutputAccessLogToSdCard.

Fixes softlocks during loading in Xenoblade Chronicles 2 when certain DLC is enabled.
2021-06-16 00:06:02 -04:00
bunnei
78651b5476 Merge pull request #6462 from Morph1984/proper-flush
common: fs: file: Flush the file to the disk when Flush() is called
2021-06-15 19:03:19 -07:00
ameerj
5fc8393125 astc_decoder: Fix LDR CEM1 endpoint calculation
Per the spec, L1 is clamped to the value 0xff if it is greater than 0xff. An oversight caused us to take the maximum of L1 and 0xff, rather than the minimum.

Huge thanks to wwylele for finding this.

Co-Authored-By: Weiyi Wang <wwylele@gmail.com>
2021-06-15 20:19:01 -04:00
ameerj
f9bfeaa2bc yuzu_cmd/config: Add Accelerate ASTC and missing NVDEC emulation settings 2021-06-15 20:19:00 -04:00
ameerj
b2955479e5 configure_graphics: Add Accelerate ASTC decoding setting 2021-06-15 20:19:00 -04:00
ameerj
c4ff7ecf51 textures: Reintroduce CPU ASTC decoder
Users may want to fall back to the CPU ASTC texture decoder due to hangs
and crashes that may be caused by keeping the GPU under compute heavy
loads for extended periods of time. This is especially the case in games
such as Astral Chain which make extensive use of ASTC textures.
2021-06-15 20:19:00 -04:00
bunnei
f3caf53648 Merge pull request #6470 from ameerj/lm-silence
lm: Demote LM guest logs to LOG_DEBUG
2021-06-14 20:40:25 -07:00
ameerj
422a15ee75 lm: Demote guest logs to LOG_DEBUG
Guest logs are not very useful, as they are intended for use by the game developers during development. As such, they provide little meaning to be logged by yuzu and tend to overwhelm the log output at times.
2021-06-14 22:23:27 -04:00
bunnei
c8f86edee6 Merge pull request #6456 from Morph1984/very-important-changes
configure_cpu_debug: Clarify settings behavior
2021-06-14 15:05:44 -07:00
Fernando Sahmkow
8d4dfc98ec Merge pull request #6448 from Morph1984/recursive-dir-iterator
common: fs: Use the normal directory iterator in *Recursively functions
2021-06-14 15:47:04 +02:00
Mai M
de12dbb01a Merge pull request #6463 from Morph1984/restructure-logging
common: logging: Restructure logging backend
2021-06-13 14:29:39 -04:00
Morph
a4454329c1 general: Remove extraneous includes 2021-06-13 11:32:43 -04:00
Morph
391e823c79 common: logging: Restructure backend code 2021-06-13 11:05:58 -04:00
Morph
8150c65c07 common: logging: backend: Wrap IOFile in a unique_ptr
Allows us to forward declare Common::FS::IOFile.
2021-06-13 11:05:58 -04:00
Morph
a98b6c8f07 common: fs: file: Flush the file to the disk when Flush() is called
std::fflush does not guarantee that file buffers are flushed to the disk.

Use _commit on Windows and fsync on all other OSes to ensure that the file is flushed to the disk.
2021-06-13 07:47:57 -04:00
Morph
56afd4ab4b Merge pull request #6452 from german77/sixaxis_firmware_stub
hid: Stub IsFirmwareUpdateAvailableForSixAxisSensor
2021-06-13 05:28:32 -04:00
Morph
c11b4c45e1 configure_cpu_debug: Clarify settings behavior
This makes it clear that the disabled settings only take effect when CPU Accuracy is set to Debug Mode.
2021-06-13 00:27:33 -04:00
Morph
c978f3144c common: fs: Use the normal directory iterator in *Recursively functions
MSVC's implementation of recursive_directory_iterator throws an exception on an error despite a std::error_code being passed into its constructor. This is most likely a bug in MSVC's implementation since directory_iterator does not throw an exception on an error.

We can replace the usage of recursive_directory_iterator for now until MSVC fixes their implementation of it.
2021-06-12 01:39:07 -04:00
bunnei
8b5655a98e Merge pull request #6453 from lat9nq/libusb-fix-msvc
externals: Don't set FOUND or VERSION on LIBUSB
2021-06-11 16:29:34 -07:00
lat9nq
2817ef1a53 externals: Don't set FOUND or VERSION on LIBUSB
Fixes an issue where libusb.h wouldn't be found when building yuzu on
MSVC.

This only affects the "traditional" CMake pathway for linking libusb to
yuzu AKA MSVC. For autotools we still want to set these variables before
configuring SDL.
2021-06-11 16:57:04 -04:00
bunnei
58180f9fa8 Merge pull request #6451 from Morph1984/check-disk-space-dump
yuzu: main: Ensure enough space is available for RomFS dumping
2021-06-11 13:23:23 -07:00
german77
827483409b hid: Stub IsFirmwareUpdateAvailableForSixAxisSensor 2021-06-11 14:44:46 -05:00
Mai M
9951322e5a Merge pull request #6422 from FernandoS27/i-am-the-senate
Implement/Port Fastmem from Citra to Yuzu
2021-06-11 14:26:54 -04:00
Morph
fa2aac1bf5 yuzu: main: Ensure enough space is available for RomFS dumping
This warns the user if there isn't enough free space to dump the entire RomFS to disk. It requires at least the size of the extracted RomFS + 1 GiB as a buffer of free space.
2021-06-11 14:04:11 -04:00
bunnei
0c0c1a039e Merge pull request #6443 from Morph1984/k-light-condition-variable
kernel: KLightConditionVariable: Update implementation to 12.x
2021-06-11 11:03:55 -07:00
Markus Wick
7f85abb281 common/host_memory: Implement a fallback if fastmem fails.
This falls back to the old approach of using a virtual buffer.

Windows is untested, but this build should fix support for Windows < 10 v1803. However without fastmem support at all.
2021-06-11 17:27:17 +02:00
ReinUsesLisp
f332d4a9b5 common/host_shader: Load Windows 10 functions dynamically
Workaround old headers and libraries shipped on MinGW.
2021-06-11 17:27:17 +02:00
Fernando Sahmkow
588ab44470 GPUTHread: Remove async reads from Normal Accuracy. 2021-06-11 17:27:17 +02:00
ReinUsesLisp
7b0d8bd1fb rasterizer: Update pages in batches 2021-06-11 17:27:17 +02:00
ReinUsesLisp
ee67460ff0 host_memory: Support staged VirtualProtect calls 2021-06-11 17:27:17 +02:00
FernandoS27
5ba28325b2 General: Add settings for fastmem and disabling adress space check. 2021-06-11 17:27:17 +02:00
Markus Wick
c4609c92ee common/host_memory: Optimize for huge tables.
In theory, if we have 2 MB continously mapped, this should save one layer of TLB.
Let's make it at least more likely by aligning the memory.
2021-06-11 17:27:06 +02:00
Markus Wick
621f3f5f47 core: Make use of fastmem 2021-06-11 17:27:06 +02:00
ReinUsesLisp
740edacc8d tests: Add tests for host memory 2021-06-11 17:27:06 +02:00
Markus Wick
5105318bbc common/host_memory: Add Linux implementation 2021-06-11 17:27:06 +02:00
ReinUsesLisp
a7837a3791 common/host_memory: Add interface and Windows implementation 2021-06-11 17:27:06 +02:00
Morph
fbb170857f Merge pull request #6450 from lat9nq/update-sdl
externals: Update SDL to 2f248a2a
2021-06-11 06:33:50 -04:00
lat9nq
f738c6b231 externals: Update SDL to 2f248a2a 2021-06-11 04:40:16 -04:00
bunnei
c1b8e59ea0 Merge pull request #6407 from lat9nq/fix-libusb-2
cmake: Use autotools for libusb linking generally on GNU, and cleanup
2021-06-10 23:35:30 -07:00
bunnei
46ec0ee55b Merge pull request #6445 from degasus/fix_ubsn
Fix GCC undefined behavior sanitizer.
2021-06-10 22:17:33 -07:00
Morph
ebd38d66db kernel: Unconditionally set thread state when appropriate 2021-06-11 00:58:04 -04:00
Morph
aa79ca7a7a kernel: KLightConditionVariable: Update implementation to 12.x
Updates the implementation of KLightConditionVariable to FW 12.x
2021-06-11 00:58:04 -04:00
bunnei
4547b2735a Merge pull request #6444 from bunnei/fix-sm-sessions
hle: service: sm: Remove redundant session reservation, etc.
2021-06-10 12:17:13 -07:00
Markus Wick
6755025310 Fix GCC undefined behavior sanitizer.
* Wrong alignment in u64 LOG_DEBUG -> memcpy.
* Huge shift exponent in stride calculation for linear buffer, unused result -> skipped.
* Large shift in buffer cache if word = 0, skip checking for set bits.

Non of those were critical, so this should not change any behavior.
At least with the assumption, that the last one used masking behavior, which always yield continuous_bits = 0.
2021-06-10 21:07:27 +02:00
bunnei
781c85b951 hle: service: sm: Remove redundant session reservation, etc.
- We were double-reserving, causing us to run out of sessions in Pokemon Sword & Shield.
2021-06-10 11:34:41 -07:00
bunnei
fa8a0065ca hle: service: Increase arbitrary max sessions limit.
- Pokemon Sword/Shield are still hitting this for some reason, causing an svcBreak.
2021-06-10 00:08:09 -07:00
bunnei
74f0087bfa Merge pull request #6441 from bunnei/fix-session-handler
hle: kernel: KServerSession: Fix client disconnected.
2021-06-09 22:53:25 -07:00
bunnei
b259e95c09 hle: kernel: KClientPort: Add an assert for session count.
- Prevents us from over decrementing num_sessions.
2021-06-09 22:36:42 -07:00
bunnei
ec5674a6ad hle: service: sm: Fix GetService setup of session & port. 2021-06-09 22:29:18 -07:00
bunnei
2aa6a8d889 hle: service: Use correct size for ServerSessionCountMax. 2021-06-09 22:04:36 -07:00
bunnei
b2971b48ed hle: kernel: KServerSession: Fix client disconnected.
- Prevents a cloned session's handler from being overwritten by another disconnected session.
- Fixes session handler nullptr asserts with Pokemon Sword & Shield.
2021-06-09 21:37:11 -07:00
Ameer J
86d832ab9a Merge pull request #6439 from lat9nq/ci-no-7z
ci: common: Remove 7z packaging
2021-06-09 19:47:08 -04:00
Mai M
61c7a81ec8 Merge pull request #6440 from bunnei/cancel-synch
kernel: svc: Add missing error check to CancelSynchronization.
2021-06-09 19:08:36 -04:00
lat9nq
fbad68de0f ci: windows: Compress using xz
Use XZ instead of gzip for packing. Should save about 10 MB.
2021-06-09 18:54:23 -04:00
bunnei
c63ea608aa kernel: svc: Add missing error check to CancelSynchronization.
- Avoids a potential crash if the handle is invalid, and also makes this code accurate to real kernel behavior.
2021-06-09 15:24:46 -07:00
lat9nq
6eeb532c96 ci: common: Remove 7z packaging
Removes the 7z from being package during CI, as only .tar.xz preserves
information needed on Linux, and otherwise is just extremely redundant
to package in addition to the .tar.xz.  This affects Linux releases and
PR-verify artifacts only. MSVC releases do not use this script to my
knowledge.
2021-06-09 17:16:29 -04:00
Mai M
5857067a18 Merge pull request #6436 from liushuyu/master
src/common/CMakeLists.txt: fix variable escaping
2021-06-09 15:38:56 -04:00
bunnei
2d32fc2318 hle: service: Increase arbitrary max sessions limit.
- Pokemon Sword/Shield are still hitting this for some reason, causing an svcBreak.
2021-06-09 11:59:34 -07:00
bunnei
75a4ac12c6 Merge pull request #6413 from Kewlan/limitable_input_dialog_limit
limitable_input_dialog: Implement character limiter
2021-06-09 11:55:36 -07:00
liushuyu
eb9deffab6 src/common/CMakeLists.txt: fix variable escaping 2021-06-09 02:20:55 -06:00
Morph
15483c07c6 Merge pull request #6435 from lioncash/nodisc2
common/fs/path_util: Remove [[nodiscard]] from function with void return
2021-06-09 02:44:41 -04:00
bunnei
f9c3e2e872 Merge pull request #6434 from lioncash/tcontext
configure_ui: Add translation context for file-scope strings
2021-06-08 19:36:44 -07:00
bunnei
3c621d37f0 Merge pull request #6428 from bunnei/service-thread-crash-fix
hle: kernel: Remove service thread manager and use weak_ptr.
2021-06-08 16:43:55 -07:00
Lioncash
dd8577e91d common/fs/path_util: Remove [[nodiscard]] from function with void return
We can't make use of the return value here, since we don't a return
value to work with.
2021-06-08 19:36:09 -04:00
Lioncash
b3eb08254b configure_ui: Add translation context for file-scope strings
Allows for these strings to show up in the translation files.
2021-06-08 19:33:23 -04:00
Mai M
f09c9b5fcc Merge pull request #6426 from lat9nq/context-menu-start
yuzu qt: Start games from context menu
2021-06-08 17:09:25 -04:00
bunnei
3b5673daca Merge pull request #6412 from clementgallet/yuzu-cmd-window-gl
yuzu-cmd: Fix OpenGL rendering
2021-06-07 21:12:17 -07:00
lat9nq
5ac018d1df yuzu qt: Start games from context menu
This connects the BootGame function to the context menu. In addition,
there is an option to boot without using the custom configuration.
2021-06-07 20:27:51 -04:00
Kewlan
058196a089 limitable_input_dialog: Implement character limiter
When using GetText() you can now choose what set of characters the user can't enter.
2021-06-06 09:07:55 +02:00
Clément Gallet
2e1c58b905 Avoid -Wshadow warning
Co-authored-by: Mai M. <mathew1800@gmail.com>
2021-06-05 18:43:10 +02:00
Clément Gallet
9ff8504452 yuzu-cmd: Fix OpenGL rendering 2021-06-04 11:39:04 +02:00
lat9nq
7395cd3124 externals: libusb: Call program names not full paths 2021-06-03 04:53:01 -04:00
lat9nq
890acfa2c0 externals: libusb: Link libusb statically on Linux
Turns out that this is possible. Also addresses my own review comment.
2021-06-03 04:38:29 -04:00
lat9nq
ddc47e6df8 cmake: General improvements to libusb linking
Delegates libusb external communication to externals/CMakeLists.txt
Ensures an interface library `usb` for every pathway
input_common just links to the `usb` library now
externals/libusb/CMakeLists.txt sets variables to override SDL2's libusb
finding
Other minor cleanup
2021-06-03 03:49:35 -04:00
lat9nq
55dd027115 cmake: Use autotools to build libusb generally for GNU
Building libusb was also broken on GCC (and maybe Clang) on our
CMakeLists after upgrading to 1.0.24, but it was not being checked
because our 18.04 container had libusb installed on it.
This builds on the MinGW work from earlier and extends it to the rest of
the GNU toolchains. In addition we make use of pkg-config when present
to find libusb. pkg-config is preferrable because we can specify a
minimum required version.
2021-06-03 02:49:53 -04:00
104 changed files with 3806 additions and 636 deletions

View File

@@ -9,11 +9,5 @@ cp "${REV_NAME}-source.tar.xz" "$DIR_NAME"
tar $COMPRESSION_FLAGS "$ARCHIVE_NAME" "$DIR_NAME"
mv "$DIR_NAME" $RELEASE_NAME
mv "${REV_NAME}-source.tar.xz" $RELEASE_NAME
7z a "$REV_NAME.7z" $RELEASE_NAME
# move the compiled archive into the artifacts directory to be uploaded by travis releases
mv "$ARCHIVE_NAME" "${ARTIFACTS_DIR}/"
mv "$REV_NAME.7z" "${ARTIFACTS_DIR}/"

View File

@@ -3,8 +3,8 @@
. .ci/scripts/common/pre-upload.sh
REV_NAME="yuzu-windows-mingw-${GITDATE}-${GITREV}"
ARCHIVE_NAME="${REV_NAME}.tar.gz"
COMPRESSION_FLAGS="-czvf"
ARCHIVE_NAME="${REV_NAME}.tar.xz"
COMPRESSION_FLAGS="-cJvf"
if [ "${RELEASE_NAME}" = "mainline" ]; then
DIR_NAME="${REV_NAME}"

View File

@@ -23,6 +23,8 @@ option(ENABLE_WEB_SERVICE "Enable web services (telemetry, etc.)" ON)
option(YUZU_USE_BUNDLED_BOOST "Download bundled Boost" OFF)
option(YUZU_USE_BUNDLED_LIBUSB "Compile bundled libusb" OFF)
CMAKE_DEPENDENT_OPTION(YUZU_USE_BUNDLED_FFMPEG "Download/Build bundled FFmpeg" ON "WIN32" OFF)
option(YUZU_USE_QT_WEB_ENGINE "Use QtWebEngine for web applet implementation" OFF)
@@ -420,14 +422,22 @@ elseif (TARGET Boost::boost)
endif()
# Ensure libusb is properly configured (based on dolphin libusb include)
if(NOT APPLE)
if(NOT APPLE AND NOT YUZU_USE_BUNDLED_LIBUSB)
include(FindPkgConfig)
find_package(LibUSB)
endif()
if (NOT LIBUSB_FOUND)
add_subdirectory(externals/libusb)
set(LIBUSB_INCLUDE_DIR "${CMAKE_CURRENT_SOURCE_DIR}/externals/libusb/libusb/libusb")
set(LIBUSB_LIBRARIES usb)
if (PKG_CONFIG_FOUND)
pkg_check_modules(LIBUSB QUIET libusb-1.0>=1.0.24)
else()
find_package(LibUSB)
endif()
if (LIBUSB_FOUND)
add_library(usb INTERFACE)
target_include_directories(usb INTERFACE "${LIBUSB_INCLUDE_DIRS}")
target_link_libraries(usb INTERFACE "${LIBUSB_LIBRARIES}")
else()
message(WARNING "libusb not found, falling back to externals")
set(YUZU_USE_BUNDLED_LIBUSB ON)
endif()
endif()
# List of all FFmpeg components required

View File

@@ -45,6 +45,11 @@ target_include_directories(microprofile INTERFACE ./microprofile)
add_library(unicorn-headers INTERFACE)
target_include_directories(unicorn-headers INTERFACE ./unicorn/include)
# libusb
if (NOT LIBUSB_FOUND OR YUZU_USE_BUNDLED_LIBUSB)
add_subdirectory(libusb)
endif()
# SDL2
if (NOT SDL2_FOUND AND ENABLE_SDL2)
if (NOT WIN32)

2
externals/SDL vendored

View File

@@ -1,10 +1,13 @@
if (MINGW)
# The MinGW toolchain for some reason doesn't work with this CMakeLists file after updating to
# 1.0.24, so we do it the old-fashioned way for now. We may want to move native Linux toolchains
# to here, too (TODO lat9nq?).
if (MINGW OR (${CMAKE_SYSTEM_NAME} MATCHES "Linux"))
set(LIBUSB_FOUND ON CACHE BOOL "libusb is present" FORCE)
set(LIBUSB_VERSION "1.0.24" CACHE STRING "libusb version string" FORCE)
# GNU toolchains for some reason doesn't work with the later half of this CMakeLists after
# updating to 1.0.24, so we do it the old-fashioned way for now.
set(LIBUSB_PREFIX "${CMAKE_CURRENT_BINARY_DIR}/libusb")
set(LIBUSB_SRC_DIR "${CMAKE_CURRENT_SOURCE_DIR}/libusb")
# Workarounds for MSYS/MinGW
if (MSYS)
# CMake on Windows passes `C:/`, but we need `/C/` or `/c/` to use `configure`
@@ -19,36 +22,42 @@ if (MINGW)
set(LIBUSB_CONFIGURE "${LIBUSB_SRC_DIR}/configure")
set(LIBUSB_MAKEFILE "${LIBUSB_PREFIX}/Makefile")
set(LIBUSB_LIBRARY "${LIBUSB_PREFIX}/libusb/.libs/libusb-1.0.dll.a")
set(LIBUSB_SHARED_LIBRARY "${LIBUSB_PREFIX}/libusb/.libs/libusb-1.0.dll")
set(LIBUSB_SHARED_LIBRARY_DEST "${CMAKE_BINARY_DIR}/bin/libusb-1.0.dll")
# Causes "externals/libusb/libusb/libusb/os/windows_winusb.c:1427:2: error: conversion to non-scalar type requested", so cannot statically link it for now.
# set(LIBUSB_CFLAGS "-DGUID_DEVINTERFACE_USB_DEVICE=\\(GUID\\){0xA5DCBF10,0x6530,0x11D2,{0x90,0x1F,0x00,0xC0,0x4F,0xB9,0x51,0xED}}")
if (MINGW)
set(LIBUSB_LIBRARIES "${LIBUSB_PREFIX}/libusb/.libs/libusb-1.0.dll.a" CACHE PATH "libusb library path" FORCE)
set(LIBUSB_SHARED_LIBRARY "${LIBUSB_PREFIX}/libusb/.libs/libusb-1.0.dll")
set(LIBUSB_SHARED_LIBRARY_DEST "${CMAKE_BINARY_DIR}/bin/libusb-1.0.dll")
set(LIBUSB_CONFIGURE_ARGS --host=x86_64-w64-mingw32 --build=x86_64-windows)
else()
set(LIBUSB_LIBRARIES "${LIBUSB_PREFIX}/libusb/.libs/libusb-1.0.a" CACHE PATH "libusb library path" FORCE)
endif()
set(LIBUSB_INCLUDE_DIRS "${LIBUSB_SRC_DIR}/libusb" CACHE PATH "libusb headers path" FORCE)
# MINGW: causes "externals/libusb/libusb/libusb/os/windows_winusb.c:1427:2: error: conversion to non-scalar type requested", so cannot statically link it for now.
if (NOT MINGW)
set(LIBUSB_CFLAGS "-DGUID_DEVINTERFACE_USB_DEVICE=\\(GUID\\){0xA5DCBF10,0x6530,0x11D2,{0x90,0x1F,0x00,0xC0,0x4F,0xB9,0x51,0xED}}")
endif()
make_directory("${LIBUSB_PREFIX}")
add_custom_command(
OUTPUT
"${LIBUSB_LIBRARY}"
"${LIBUSB_LIBRARIES}"
COMMAND
make
WORKING_DIRECTORY
"${LIBUSB_PREFIX}"
)
# We may use this path for other GNU toolchains, so put all of the MinGW-specific stuff here
if (MINGW)
set(LIBUSB_CONFIGURE_ARGS --host=x86_64-w64-mingw32 --build=x86_64-windows)
endif()
add_custom_command(
OUTPUT
"${LIBUSB_MAKEFILE}"
COMMAND
# /bin/env
# CFLAGS="${LIBUSB_CFLAGS}"
/bin/sh "${LIBUSB_CONFIGURE}"
env
CFLAGS="${LIBUSB_CFLAGS}"
sh "${LIBUSB_CONFIGURE}"
${LIBUSB_CONFIGURE_ARGS}
--srcdir="${LIBUSB_SRC_DIR}"
WORKING_DIRECTORY
@@ -59,7 +68,7 @@ if (MINGW)
OUTPUT
"${LIBUSB_CONFIGURE}"
COMMAND
/bin/sh "${LIBUSB_SRC_DIR}/bootstrap.sh"
sh "${LIBUSB_SRC_DIR}/bootstrap.sh"
WORKING_DIRECTORY
"${LIBUSB_SRC_DIR}"
)
@@ -68,19 +77,30 @@ if (MINGW)
OUTPUT
"${LIBUSB_SHARED_LIBRARY_DEST}"
COMMAND
/bin/cp "${LIBUSB_SHARED_LIBRARY}" "${LIBUSB_SHARED_LIBRARY_DEST}"
cp "${LIBUSB_SHARED_LIBRARY}" "${LIBUSB_SHARED_LIBRARY_DEST}"
)
add_custom_target(usb-bootstrap ALL DEPENDS "${LIBUSB_CONFIGURE}")
add_custom_target(usb-configure ALL DEPENDS "${LIBUSB_MAKEFILE}" usb-bootstrap)
add_custom_target(usb-build ALL DEPENDS "${LIBUSB_LIBRARY}" usb-configure)
add_custom_target(usb-bootstrap DEPENDS "${LIBUSB_CONFIGURE}")
add_custom_target(usb-configure DEPENDS "${LIBUSB_MAKEFILE}" usb-bootstrap)
add_custom_target(usb-build ALL DEPENDS "${LIBUSB_LIBRARIES}" usb-configure)
# Workaround since static linking didn't work out -- We need to copy the DLL to the bin directory
add_custom_target(usb-copy ALL DEPENDS "${LIBUSB_SHARED_LIBRARY_DEST}" usb-build)
# Make `usb` alias to LIBUSB_LIBRARY
add_library(usb INTERFACE)
target_link_libraries(usb INTERFACE "${LIBUSB_LIBRARY}")
else() # MINGW
add_dependencies(usb usb-copy)
target_link_libraries(usb INTERFACE "${LIBUSB_LIBRARIES}")
target_include_directories(usb INTERFACE "${LIBUSB_INCLUDE_DIRS}")
if (${CMAKE_SYSTEM_NAME} MATCHES "Linux")
Include(FindPkgConfig)
pkg_check_modules(LIBUDEV REQUIRED libudev)
if (LIBUDEV_FOUND)
target_include_directories(usb INTERFACE "${LIBUDEV_INCLUDE_DIRS}")
target_link_libraries(usb INTERFACE "${LIBUDEV_STATIC_LIBRARIES}")
endif()
endif()
else() # MINGW OR (${CMAKE_SYSTEM_NAME} MATCHES "Linux")
# Ensure libusb compiles with UTF-8 encoding on MSVC
if(MSVC)
add_compile_options(/utf-8)
@@ -236,4 +256,4 @@ else() # MINGW
configure_file(config.h.in config.h)
endif() # MINGW
endif() # MINGW OR (${CMAKE_SYSTEM_NAME} MATCHES "Linux")

View File

@@ -21,14 +21,14 @@ find_package(Git QUIET)
add_custom_command(OUTPUT scm_rev.cpp
COMMAND ${CMAKE_COMMAND}
-DSRC_DIR="${CMAKE_SOURCE_DIR}"
-DBUILD_REPOSITORY="${BUILD_REPOSITORY}"
-DTITLE_BAR_FORMAT_IDLE="${TITLE_BAR_FORMAT_IDLE}"
-DTITLE_BAR_FORMAT_RUNNING="${TITLE_BAR_FORMAT_RUNNING}"
-DBUILD_TAG="${BUILD_TAG}"
-DBUILD_ID="${DISPLAY_VERSION}"
-DGIT_EXECUTABLE="${GIT_EXECUTABLE}"
-P "${CMAKE_SOURCE_DIR}/CMakeModules/GenerateSCMRev.cmake"
-DSRC_DIR=${CMAKE_SOURCE_DIR}
-DBUILD_REPOSITORY=${BUILD_REPOSITORY}
-DTITLE_BAR_FORMAT_IDLE=${TITLE_BAR_FORMAT_IDLE}
-DTITLE_BAR_FORMAT_RUNNING=${TITLE_BAR_FORMAT_RUNNING}
-DBUILD_TAG=${BUILD_TAG}
-DBUILD_ID=${DISPLAY_VERSION}
-DGIT_EXECUTABLE=${GIT_EXECUTABLE}
-P ${CMAKE_SOURCE_DIR}/CMakeModules/GenerateSCMRev.cmake
DEPENDS
# WARNING! It was too much work to try and make a common location for this list,
# so if you need to change it, please update CMakeModules/GenerateSCMRev.cmake as well
@@ -92,6 +92,7 @@ add_custom_command(OUTPUT scm_rev.cpp
"${CMAKE_CURRENT_SOURCE_DIR}/scm_rev.h"
# technically we should regenerate if the git version changed, but its not worth the effort imo
"${CMAKE_SOURCE_DIR}/CMakeModules/GenerateSCMRev.cmake"
VERBATIM
)
add_library(common STATIC
@@ -130,6 +131,8 @@ add_library(common STATIC
hash.h
hex_util.cpp
hex_util.h
host_memory.cpp
host_memory.h
intrusive_red_black_tree.h
logging/backend.cpp
logging/backend.h
@@ -138,6 +141,7 @@ add_library(common STATIC
logging/log.h
logging/text_formatter.cpp
logging/text_formatter.h
logging/types.h
lz4_compression.cpp
lz4_compression.h
math_util.h

View File

@@ -24,6 +24,7 @@ enum : u64 {
Size_128_MB = 128ULL * Size_1_MB,
Size_448_MB = 448ULL * Size_1_MB,
Size_507_MB = 507ULL * Size_1_MB,
Size_512_MB = 512ULL * Size_1_MB,
Size_562_MB = 562ULL * Size_1_MB,
Size_1554_MB = 1554ULL * Size_1_MB,
Size_2048_MB = 2048ULL * Size_1_MB,

View File

@@ -183,10 +183,6 @@ size_t WriteStringToFile(const std::filesystem::path& path, FileType type,
size_t AppendStringToFile(const std::filesystem::path& path, FileType type,
std::string_view string) {
if (!Exists(path)) {
return WriteStringToFile(path, type, string);
}
if (!IsFile(path)) {
return 0;
}
@@ -309,7 +305,11 @@ bool IOFile::Flush() const {
errno = 0;
const auto flush_result = std::fflush(file) == 0;
#ifdef _WIN32
const auto flush_result = std::fflush(file) == 0 && _commit(fileno(file)) == 0;
#else
const auto flush_result = std::fflush(file) == 0 && fsync(fileno(file)) == 0;
#endif
if (!flush_result) {
const auto ec = std::error_code{errno, std::generic_category()};

View File

@@ -71,7 +71,7 @@ template <typename Path>
/**
* Writes a string to a file at path and returns the number of characters successfully written.
* If an file already exists at path, its contents will be erased.
* If a file already exists at path, its contents will be erased.
* If the filesystem object at path is not a file, this function returns 0.
*
* @param path Filesystem path
@@ -95,7 +95,6 @@ template <typename Path>
/**
* Appends a string to a file at path and returns the number of characters successfully written.
* If a file does not exist at path, WriteStringToFile is called instead.
* If the filesystem object at path is not a file, this function returns 0.
*
* @param path Filesystem path

View File

@@ -321,7 +321,8 @@ bool RemoveDirContentsRecursively(const fs::path& path) {
std::error_code ec;
for (const auto& entry : fs::recursive_directory_iterator(path, ec)) {
// TODO (Morph): Replace this with recursive_directory_iterator once it's fixed in MSVC.
for (const auto& entry : fs::directory_iterator(path, ec)) {
if (ec) {
LOG_ERROR(Common_Filesystem,
"Failed to completely enumerate the directory at path={}, ec_message={}",
@@ -337,6 +338,12 @@ bool RemoveDirContentsRecursively(const fs::path& path) {
PathToUTF8String(entry.path()), ec.message());
break;
}
// TODO (Morph): Remove this when MSVC fixes recursive_directory_iterator.
// recursive_directory_iterator throws an exception despite passing in a std::error_code.
if (entry.status().type() == fs::file_type::directory) {
return RemoveDirContentsRecursively(entry.path());
}
}
if (ec) {
@@ -475,7 +482,8 @@ void IterateDirEntriesRecursively(const std::filesystem::path& path,
std::error_code ec;
for (const auto& entry : fs::recursive_directory_iterator(path, ec)) {
// TODO (Morph): Replace this with recursive_directory_iterator once it's fixed in MSVC.
for (const auto& entry : fs::directory_iterator(path, ec)) {
if (ec) {
break;
}
@@ -495,6 +503,12 @@ void IterateDirEntriesRecursively(const std::filesystem::path& path,
break;
}
}
// TODO (Morph): Remove this when MSVC fixes recursive_directory_iterator.
// recursive_directory_iterator throws an exception despite passing in a std::error_code.
if (entry.status().type() == fs::file_type::directory) {
IterateDirEntriesRecursively(entry.path(), callback, filter);
}
}
if (callback_error || ec) {

View File

@@ -209,7 +209,7 @@ void SetYuzuPath(YuzuPath yuzu_path, const std::filesystem::path& new_path);
#ifdef _WIN32
template <typename Path>
[[nodiscard]] void SetYuzuPath(YuzuPath yuzu_path, const Path& new_path) {
void SetYuzuPath(YuzuPath yuzu_path, const Path& new_path) {
if constexpr (IsChar<typename Path::value_type>) {
SetYuzuPath(yuzu_path, ToU8String(new_path));
} else {

538
src/common/host_memory.cpp Normal file
View File

@@ -0,0 +1,538 @@
#ifdef _WIN32
#include <iterator>
#include <unordered_map>
#include <boost/icl/separate_interval_set.hpp>
#include <windows.h>
#include "common/dynamic_library.h"
#elif defined(__linux__) // ^^^ Windows ^^^ vvv Linux vvv
#ifndef _GNU_SOURCE
#define _GNU_SOURCE
#endif
#include <fcntl.h>
#include <sys/mman.h>
#include <unistd.h>
#endif // ^^^ Linux ^^^
#include <mutex>
#include "common/alignment.h"
#include "common/assert.h"
#include "common/host_memory.h"
#include "common/logging/log.h"
#include "common/scope_exit.h"
namespace Common {
constexpr size_t PageAlignment = 0x1000;
constexpr size_t HugePageSize = 0x200000;
#ifdef _WIN32
// Manually imported for MinGW compatibility
#ifndef MEM_RESERVE_PLACEHOLDER
#define MEM_RESERVE_PLACEHOLDER 0x0004000
#endif
#ifndef MEM_REPLACE_PLACEHOLDER
#define MEM_REPLACE_PLACEHOLDER 0x00004000
#endif
#ifndef MEM_COALESCE_PLACEHOLDERS
#define MEM_COALESCE_PLACEHOLDERS 0x00000001
#endif
#ifndef MEM_PRESERVE_PLACEHOLDER
#define MEM_PRESERVE_PLACEHOLDER 0x00000002
#endif
using PFN_CreateFileMapping2 = _Ret_maybenull_ HANDLE(WINAPI*)(
_In_ HANDLE File, _In_opt_ SECURITY_ATTRIBUTES* SecurityAttributes, _In_ ULONG DesiredAccess,
_In_ ULONG PageProtection, _In_ ULONG AllocationAttributes, _In_ ULONG64 MaximumSize,
_In_opt_ PCWSTR Name,
_Inout_updates_opt_(ParameterCount) MEM_EXTENDED_PARAMETER* ExtendedParameters,
_In_ ULONG ParameterCount);
using PFN_VirtualAlloc2 = _Ret_maybenull_ PVOID(WINAPI*)(
_In_opt_ HANDLE Process, _In_opt_ PVOID BaseAddress, _In_ SIZE_T Size,
_In_ ULONG AllocationType, _In_ ULONG PageProtection,
_Inout_updates_opt_(ParameterCount) MEM_EXTENDED_PARAMETER* ExtendedParameters,
_In_ ULONG ParameterCount);
using PFN_MapViewOfFile3 = _Ret_maybenull_ PVOID(WINAPI*)(
_In_ HANDLE FileMapping, _In_opt_ HANDLE Process, _In_opt_ PVOID BaseAddress,
_In_ ULONG64 Offset, _In_ SIZE_T ViewSize, _In_ ULONG AllocationType, _In_ ULONG PageProtection,
_Inout_updates_opt_(ParameterCount) MEM_EXTENDED_PARAMETER* ExtendedParameters,
_In_ ULONG ParameterCount);
using PFN_UnmapViewOfFile2 = BOOL(WINAPI*)(_In_ HANDLE Process, _In_ PVOID BaseAddress,
_In_ ULONG UnmapFlags);
template <typename T>
static void GetFuncAddress(Common::DynamicLibrary& dll, const char* name, T& pfn) {
if (!dll.GetSymbol(name, &pfn)) {
LOG_CRITICAL(HW_Memory, "Failed to load {}", name);
throw std::bad_alloc{};
}
}
class HostMemory::Impl {
public:
explicit Impl(size_t backing_size_, size_t virtual_size_)
: backing_size{backing_size_}, virtual_size{virtual_size_}, process{GetCurrentProcess()},
kernelbase_dll("Kernelbase") {
if (!kernelbase_dll.IsOpen()) {
LOG_CRITICAL(HW_Memory, "Failed to load Kernelbase.dll");
throw std::bad_alloc{};
}
GetFuncAddress(kernelbase_dll, "CreateFileMapping2", pfn_CreateFileMapping2);
GetFuncAddress(kernelbase_dll, "VirtualAlloc2", pfn_VirtualAlloc2);
GetFuncAddress(kernelbase_dll, "MapViewOfFile3", pfn_MapViewOfFile3);
GetFuncAddress(kernelbase_dll, "UnmapViewOfFile2", pfn_UnmapViewOfFile2);
// Allocate backing file map
backing_handle =
pfn_CreateFileMapping2(INVALID_HANDLE_VALUE, nullptr, FILE_MAP_WRITE | FILE_MAP_READ,
PAGE_READWRITE, SEC_COMMIT, backing_size, nullptr, nullptr, 0);
if (!backing_handle) {
LOG_CRITICAL(HW_Memory, "Failed to allocate {} MiB of backing memory",
backing_size >> 20);
throw std::bad_alloc{};
}
// Allocate a virtual memory for the backing file map as placeholder
backing_base = static_cast<u8*>(pfn_VirtualAlloc2(process, nullptr, backing_size,
MEM_RESERVE | MEM_RESERVE_PLACEHOLDER,
PAGE_NOACCESS, nullptr, 0));
if (!backing_base) {
Release();
LOG_CRITICAL(HW_Memory, "Failed to reserve {} MiB of virtual memory",
backing_size >> 20);
throw std::bad_alloc{};
}
// Map backing placeholder
void* const ret = pfn_MapViewOfFile3(backing_handle, process, backing_base, 0, backing_size,
MEM_REPLACE_PLACEHOLDER, PAGE_READWRITE, nullptr, 0);
if (ret != backing_base) {
Release();
LOG_CRITICAL(HW_Memory, "Failed to map {} MiB of virtual memory", backing_size >> 20);
throw std::bad_alloc{};
}
// Allocate virtual address placeholder
virtual_base = static_cast<u8*>(pfn_VirtualAlloc2(process, nullptr, virtual_size,
MEM_RESERVE | MEM_RESERVE_PLACEHOLDER,
PAGE_NOACCESS, nullptr, 0));
if (!virtual_base) {
Release();
LOG_CRITICAL(HW_Memory, "Failed to reserve {} GiB of virtual memory",
virtual_size >> 30);
throw std::bad_alloc{};
}
}
~Impl() {
Release();
}
void Map(size_t virtual_offset, size_t host_offset, size_t length) {
std::unique_lock lock{placeholder_mutex};
if (!IsNiechePlaceholder(virtual_offset, length)) {
Split(virtual_offset, length);
}
ASSERT(placeholders.find({virtual_offset, virtual_offset + length}) == placeholders.end());
TrackPlaceholder(virtual_offset, host_offset, length);
MapView(virtual_offset, host_offset, length);
}
void Unmap(size_t virtual_offset, size_t length) {
std::lock_guard lock{placeholder_mutex};
// Unmap until there are no more placeholders
while (UnmapOnePlaceholder(virtual_offset, length)) {
}
}
void Protect(size_t virtual_offset, size_t length, bool read, bool write) {
DWORD new_flags{};
if (read && write) {
new_flags = PAGE_READWRITE;
} else if (read && !write) {
new_flags = PAGE_READONLY;
} else if (!read && !write) {
new_flags = PAGE_NOACCESS;
} else {
UNIMPLEMENTED_MSG("Protection flag combination read={} write={}", read, write);
}
const size_t virtual_end = virtual_offset + length;
std::lock_guard lock{placeholder_mutex};
auto [it, end] = placeholders.equal_range({virtual_offset, virtual_end});
while (it != end) {
const size_t offset = std::max(it->lower(), virtual_offset);
const size_t protect_length = std::min(it->upper(), virtual_end) - offset;
DWORD old_flags{};
if (!VirtualProtect(virtual_base + offset, protect_length, new_flags, &old_flags)) {
LOG_CRITICAL(HW_Memory, "Failed to change virtual memory protect rules");
}
++it;
}
}
const size_t backing_size; ///< Size of the backing memory in bytes
const size_t virtual_size; ///< Size of the virtual address placeholder in bytes
u8* backing_base{};
u8* virtual_base{};
private:
/// Release all resources in the object
void Release() {
if (!placeholders.empty()) {
for (const auto& placeholder : placeholders) {
if (!pfn_UnmapViewOfFile2(process, virtual_base + placeholder.lower(),
MEM_PRESERVE_PLACEHOLDER)) {
LOG_CRITICAL(HW_Memory, "Failed to unmap virtual memory placeholder");
}
}
Coalesce(0, virtual_size);
}
if (virtual_base) {
if (!VirtualFree(virtual_base, 0, MEM_RELEASE)) {
LOG_CRITICAL(HW_Memory, "Failed to free virtual memory");
}
}
if (backing_base) {
if (!pfn_UnmapViewOfFile2(process, backing_base, MEM_PRESERVE_PLACEHOLDER)) {
LOG_CRITICAL(HW_Memory, "Failed to unmap backing memory placeholder");
}
if (!VirtualFreeEx(process, backing_base, 0, MEM_RELEASE)) {
LOG_CRITICAL(HW_Memory, "Failed to free backing memory");
}
}
if (!CloseHandle(backing_handle)) {
LOG_CRITICAL(HW_Memory, "Failed to free backing memory file handle");
}
}
/// Unmap one placeholder in the given range (partial unmaps are supported)
/// Return true when there are no more placeholders to unmap
bool UnmapOnePlaceholder(size_t virtual_offset, size_t length) {
const auto it = placeholders.find({virtual_offset, virtual_offset + length});
const auto begin = placeholders.begin();
const auto end = placeholders.end();
if (it == end) {
return false;
}
const size_t placeholder_begin = it->lower();
const size_t placeholder_end = it->upper();
const size_t unmap_begin = std::max(virtual_offset, placeholder_begin);
const size_t unmap_end = std::min(virtual_offset + length, placeholder_end);
ASSERT(unmap_begin >= placeholder_begin && unmap_begin < placeholder_end);
ASSERT(unmap_end <= placeholder_end && unmap_end > placeholder_begin);
const auto host_pointer_it = placeholder_host_pointers.find(placeholder_begin);
ASSERT(host_pointer_it != placeholder_host_pointers.end());
const size_t host_offset = host_pointer_it->second;
const bool split_left = unmap_begin > placeholder_begin;
const bool split_right = unmap_end < placeholder_end;
if (!pfn_UnmapViewOfFile2(process, virtual_base + placeholder_begin,
MEM_PRESERVE_PLACEHOLDER)) {
LOG_CRITICAL(HW_Memory, "Failed to unmap placeholder");
}
// If we have to remap memory regions due to partial unmaps, we are in a data race as
// Windows doesn't support remapping memory without unmapping first. Avoid adding any extra
// logic within the panic region described below.
// Panic region, we are in a data race right now
if (split_left || split_right) {
Split(unmap_begin, unmap_end - unmap_begin);
}
if (split_left) {
MapView(placeholder_begin, host_offset, unmap_begin - placeholder_begin);
}
if (split_right) {
MapView(unmap_end, host_offset + unmap_end - placeholder_begin,
placeholder_end - unmap_end);
}
// End panic region
size_t coalesce_begin = unmap_begin;
if (!split_left) {
// Try to coalesce pages to the left
coalesce_begin = it == begin ? 0 : std::prev(it)->upper();
if (coalesce_begin != placeholder_begin) {
Coalesce(coalesce_begin, unmap_end - coalesce_begin);
}
}
if (!split_right) {
// Try to coalesce pages to the right
const auto next = std::next(it);
const size_t next_begin = next == end ? virtual_size : next->lower();
if (placeholder_end != next_begin) {
// We can coalesce to the right
Coalesce(coalesce_begin, next_begin - coalesce_begin);
}
}
// Remove and reinsert placeholder trackers
UntrackPlaceholder(it);
if (split_left) {
TrackPlaceholder(placeholder_begin, host_offset, unmap_begin - placeholder_begin);
}
if (split_right) {
TrackPlaceholder(unmap_end, host_offset + unmap_end - placeholder_begin,
placeholder_end - unmap_end);
}
return true;
}
void MapView(size_t virtual_offset, size_t host_offset, size_t length) {
if (!pfn_MapViewOfFile3(backing_handle, process, virtual_base + virtual_offset, host_offset,
length, MEM_REPLACE_PLACEHOLDER, PAGE_READWRITE, nullptr, 0)) {
LOG_CRITICAL(HW_Memory, "Failed to map placeholder");
}
}
void Split(size_t virtual_offset, size_t length) {
if (!VirtualFreeEx(process, reinterpret_cast<LPVOID>(virtual_base + virtual_offset), length,
MEM_RELEASE | MEM_PRESERVE_PLACEHOLDER)) {
LOG_CRITICAL(HW_Memory, "Failed to split placeholder");
}
}
void Coalesce(size_t virtual_offset, size_t length) {
if (!VirtualFreeEx(process, reinterpret_cast<LPVOID>(virtual_base + virtual_offset), length,
MEM_RELEASE | MEM_COALESCE_PLACEHOLDERS)) {
LOG_CRITICAL(HW_Memory, "Failed to coalesce placeholders");
}
}
void TrackPlaceholder(size_t virtual_offset, size_t host_offset, size_t length) {
placeholders.insert({virtual_offset, virtual_offset + length});
placeholder_host_pointers.emplace(virtual_offset, host_offset);
}
void UntrackPlaceholder(boost::icl::separate_interval_set<size_t>::iterator it) {
placeholders.erase(it);
placeholder_host_pointers.erase(it->lower());
}
/// Return true when a given memory region is a "nieche" and the placeholders don't have to be
/// splitted.
bool IsNiechePlaceholder(size_t virtual_offset, size_t length) const {
const auto it = placeholders.upper_bound({virtual_offset, virtual_offset + length});
if (it != placeholders.end() && it->lower() == virtual_offset + length) {
const bool is_root = it == placeholders.begin() && virtual_offset == 0;
return is_root || std::prev(it)->upper() == virtual_offset;
}
return false;
}
HANDLE process{}; ///< Current process handle
HANDLE backing_handle{}; ///< File based backing memory
DynamicLibrary kernelbase_dll;
PFN_CreateFileMapping2 pfn_CreateFileMapping2{};
PFN_VirtualAlloc2 pfn_VirtualAlloc2{};
PFN_MapViewOfFile3 pfn_MapViewOfFile3{};
PFN_UnmapViewOfFile2 pfn_UnmapViewOfFile2{};
std::mutex placeholder_mutex; ///< Mutex for placeholders
boost::icl::separate_interval_set<size_t> placeholders; ///< Mapped placeholders
std::unordered_map<size_t, size_t> placeholder_host_pointers; ///< Placeholder backing offset
};
#elif defined(__linux__) // ^^^ Windows ^^^ vvv Linux vvv
class HostMemory::Impl {
public:
explicit Impl(size_t backing_size_, size_t virtual_size_)
: backing_size{backing_size_}, virtual_size{virtual_size_} {
bool good = false;
SCOPE_EXIT({
if (!good) {
Release();
}
});
// Backing memory initialization
fd = memfd_create("HostMemory", 0);
if (fd == -1) {
LOG_CRITICAL(HW_Memory, "memfd_create failed: {}", strerror(errno));
throw std::bad_alloc{};
}
// Defined to extend the file with zeros
int ret = ftruncate(fd, backing_size);
if (ret != 0) {
LOG_CRITICAL(HW_Memory, "ftruncate failed with {}, are you out-of-memory?",
strerror(errno));
throw std::bad_alloc{};
}
backing_base = static_cast<u8*>(
mmap(nullptr, backing_size, PROT_READ | PROT_WRITE, MAP_SHARED, fd, 0));
if (backing_base == MAP_FAILED) {
LOG_CRITICAL(HW_Memory, "mmap failed: {}", strerror(errno));
throw std::bad_alloc{};
}
// Virtual memory initialization
virtual_base = static_cast<u8*>(
mmap(nullptr, virtual_size, PROT_NONE, MAP_PRIVATE | MAP_ANONYMOUS, -1, 0));
if (virtual_base == MAP_FAILED) {
LOG_CRITICAL(HW_Memory, "mmap failed: {}", strerror(errno));
throw std::bad_alloc{};
}
good = true;
}
~Impl() {
Release();
}
void Map(size_t virtual_offset, size_t host_offset, size_t length) {
void* ret = mmap(virtual_base + virtual_offset, length, PROT_READ | PROT_WRITE,
MAP_SHARED | MAP_FIXED, fd, host_offset);
ASSERT_MSG(ret != MAP_FAILED, "mmap failed: {}", strerror(errno));
}
void Unmap(size_t virtual_offset, size_t length) {
// The method name is wrong. We're still talking about the virtual range.
// We don't want to unmap, we want to reserve this memory.
void* ret = mmap(virtual_base + virtual_offset, length, PROT_NONE,
MAP_PRIVATE | MAP_ANONYMOUS | MAP_FIXED, -1, 0);
ASSERT_MSG(ret != MAP_FAILED, "mmap failed: {}", strerror(errno));
}
void Protect(size_t virtual_offset, size_t length, bool read, bool write) {
int flags = 0;
if (read) {
flags |= PROT_READ;
}
if (write) {
flags |= PROT_WRITE;
}
int ret = mprotect(virtual_base + virtual_offset, length, flags);
ASSERT_MSG(ret == 0, "mprotect failed: {}", strerror(errno));
}
const size_t backing_size; ///< Size of the backing memory in bytes
const size_t virtual_size; ///< Size of the virtual address placeholder in bytes
u8* backing_base{reinterpret_cast<u8*>(MAP_FAILED)};
u8* virtual_base{reinterpret_cast<u8*>(MAP_FAILED)};
private:
/// Release all resources in the object
void Release() {
if (virtual_base != MAP_FAILED) {
int ret = munmap(virtual_base, virtual_size);
ASSERT_MSG(ret == 0, "munmap failed: {}", strerror(errno));
}
if (backing_base != MAP_FAILED) {
int ret = munmap(backing_base, backing_size);
ASSERT_MSG(ret == 0, "munmap failed: {}", strerror(errno));
}
if (fd != -1) {
int ret = close(fd);
ASSERT_MSG(ret == 0, "close failed: {}", strerror(errno));
}
}
int fd{-1}; // memfd file descriptor, -1 is the error value of memfd_create
};
#else // ^^^ Linux ^^^ vvv Generic vvv
class HostMemory::Impl {
public:
explicit Impl(size_t /*backing_size */, size_t /* virtual_size */) {
// This is just a place holder.
// Please implement fastmem in a propper way on your platform.
throw std::bad_alloc{};
}
void Map(size_t virtual_offset, size_t host_offset, size_t length) {}
void Unmap(size_t virtual_offset, size_t length) {}
void Protect(size_t virtual_offset, size_t length, bool read, bool write) {}
u8* backing_base{nullptr};
u8* virtual_base{nullptr};
};
#endif // ^^^ Generic ^^^
HostMemory::HostMemory(size_t backing_size_, size_t virtual_size_)
: backing_size(backing_size_), virtual_size(virtual_size_) {
try {
// Try to allocate a fastmem arena.
// The implementation will fail with std::bad_alloc on errors.
impl = std::make_unique<HostMemory::Impl>(AlignUp(backing_size, PageAlignment),
AlignUp(virtual_size, PageAlignment) +
3 * HugePageSize);
backing_base = impl->backing_base;
virtual_base = impl->virtual_base;
if (virtual_base) {
virtual_base += 2 * HugePageSize - 1;
virtual_base -= reinterpret_cast<size_t>(virtual_base) & (HugePageSize - 1);
virtual_base_offset = virtual_base - impl->virtual_base;
}
} catch (const std::bad_alloc&) {
LOG_CRITICAL(HW_Memory,
"Fastmem unavailable, falling back to VirtualBuffer for memory allocation");
fallback_buffer = std::make_unique<Common::VirtualBuffer<u8>>(backing_size);
backing_base = fallback_buffer->data();
virtual_base = nullptr;
}
}
HostMemory::~HostMemory() = default;
HostMemory::HostMemory(HostMemory&&) noexcept = default;
HostMemory& HostMemory::operator=(HostMemory&&) noexcept = default;
void HostMemory::Map(size_t virtual_offset, size_t host_offset, size_t length) {
ASSERT(virtual_offset % PageAlignment == 0);
ASSERT(host_offset % PageAlignment == 0);
ASSERT(length % PageAlignment == 0);
ASSERT(virtual_offset + length <= virtual_size);
ASSERT(host_offset + length <= backing_size);
if (length == 0 || !virtual_base || !impl) {
return;
}
impl->Map(virtual_offset + virtual_base_offset, host_offset, length);
}
void HostMemory::Unmap(size_t virtual_offset, size_t length) {
ASSERT(virtual_offset % PageAlignment == 0);
ASSERT(length % PageAlignment == 0);
ASSERT(virtual_offset + length <= virtual_size);
if (length == 0 || !virtual_base || !impl) {
return;
}
impl->Unmap(virtual_offset + virtual_base_offset, length);
}
void HostMemory::Protect(size_t virtual_offset, size_t length, bool read, bool write) {
ASSERT(virtual_offset % PageAlignment == 0);
ASSERT(length % PageAlignment == 0);
ASSERT(virtual_offset + length <= virtual_size);
if (length == 0 || !virtual_base || !impl) {
return;
}
impl->Protect(virtual_offset + virtual_base_offset, length, read, write);
}
} // namespace Common

70
src/common/host_memory.h Normal file
View File

@@ -0,0 +1,70 @@
// Copyright 2019 yuzu Emulator Project
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
#pragma once
#include <memory>
#include "common/common_types.h"
#include "common/virtual_buffer.h"
namespace Common {
/**
* A low level linear memory buffer, which supports multiple mappings
* Its purpose is to rebuild a given sparse memory layout, including mirrors.
*/
class HostMemory {
public:
explicit HostMemory(size_t backing_size_, size_t virtual_size_);
~HostMemory();
/**
* Copy constructors. They shall return a copy of the buffer without the mappings.
* TODO: Implement them with COW if needed.
*/
HostMemory(const HostMemory& other) = delete;
HostMemory& operator=(const HostMemory& other) = delete;
/**
* Move constructors. They will move the buffer and the mappings to the new object.
*/
HostMemory(HostMemory&& other) noexcept;
HostMemory& operator=(HostMemory&& other) noexcept;
void Map(size_t virtual_offset, size_t host_offset, size_t length);
void Unmap(size_t virtual_offset, size_t length);
void Protect(size_t virtual_offset, size_t length, bool read, bool write);
[[nodiscard]] u8* BackingBasePointer() noexcept {
return backing_base;
}
[[nodiscard]] const u8* BackingBasePointer() const noexcept {
return backing_base;
}
[[nodiscard]] u8* VirtualBasePointer() noexcept {
return virtual_base;
}
[[nodiscard]] const u8* VirtualBasePointer() const noexcept {
return virtual_base;
}
private:
size_t backing_size{};
size_t virtual_size{};
// Low level handler for the platform dependent memory routines
class Impl;
std::unique_ptr<Impl> impl;
u8* backing_base{};
u8* virtual_base{};
size_t virtual_base_offset{};
// Fallback if fastmem is not supported on this platform
std::unique_ptr<Common::VirtualBuffer<u8>> fallback_buffer;
};
} // namespace Common

View File

@@ -17,6 +17,7 @@
#endif
#include "common/assert.h"
#include "common/fs/file.h"
#include "common/fs/fs.h"
#include "common/logging/backend.h"
#include "common/logging/log.h"
@@ -140,10 +141,14 @@ private:
std::chrono::steady_clock::time_point time_origin{std::chrono::steady_clock::now()};
};
ConsoleBackend::~ConsoleBackend() = default;
void ConsoleBackend::Write(const Entry& entry) {
PrintMessage(entry);
}
ColorConsoleBackend::~ColorConsoleBackend() = default;
void ColorConsoleBackend::Write(const Entry& entry) {
PrintColoredMessage(entry);
}
@@ -157,16 +162,19 @@ FileBackend::FileBackend(const std::filesystem::path& filename) {
void(FS::RemoveFile(old_filename));
void(FS::RenameFile(filename, old_filename));
file = FS::IOFile(filename, FS::FileAccessMode::Write, FS::FileType::TextFile);
file =
std::make_unique<FS::IOFile>(filename, FS::FileAccessMode::Write, FS::FileType::TextFile);
}
FileBackend::~FileBackend() = default;
void FileBackend::Write(const Entry& entry) {
// prevent logs from going over the maximum size (in case its spamming and the user doesn't
// know)
constexpr std::size_t MAX_BYTES_WRITTEN = 100 * 1024 * 1024;
constexpr std::size_t MAX_BYTES_WRITTEN_EXTENDED = 1024 * 1024 * 1024;
if (!file.IsOpen()) {
if (!file->IsOpen()) {
return;
}
@@ -176,147 +184,20 @@ void FileBackend::Write(const Entry& entry) {
return;
}
bytes_written += file.WriteString(FormatLogMessage(entry).append(1, '\n'));
bytes_written += file->WriteString(FormatLogMessage(entry).append(1, '\n'));
if (entry.log_level >= Level::Error) {
void(file.Flush());
void(file->Flush());
}
}
DebuggerBackend::~DebuggerBackend() = default;
void DebuggerBackend::Write(const Entry& entry) {
#ifdef _WIN32
::OutputDebugStringW(UTF8ToUTF16W(FormatLogMessage(entry).append(1, '\n')).c_str());
#endif
}
/// Macro listing all log classes. Code should define CLS and SUB as desired before invoking this.
#define ALL_LOG_CLASSES() \
CLS(Log) \
CLS(Common) \
SUB(Common, Filesystem) \
SUB(Common, Memory) \
CLS(Core) \
SUB(Core, ARM) \
SUB(Core, Timing) \
CLS(Config) \
CLS(Debug) \
SUB(Debug, Emulated) \
SUB(Debug, GPU) \
SUB(Debug, Breakpoint) \
SUB(Debug, GDBStub) \
CLS(Kernel) \
SUB(Kernel, SVC) \
CLS(Service) \
SUB(Service, ACC) \
SUB(Service, Audio) \
SUB(Service, AM) \
SUB(Service, AOC) \
SUB(Service, APM) \
SUB(Service, ARP) \
SUB(Service, BCAT) \
SUB(Service, BPC) \
SUB(Service, BGTC) \
SUB(Service, BTDRV) \
SUB(Service, BTM) \
SUB(Service, Capture) \
SUB(Service, ERPT) \
SUB(Service, ETicket) \
SUB(Service, EUPLD) \
SUB(Service, Fatal) \
SUB(Service, FGM) \
SUB(Service, Friend) \
SUB(Service, FS) \
SUB(Service, GRC) \
SUB(Service, HID) \
SUB(Service, IRS) \
SUB(Service, LBL) \
SUB(Service, LDN) \
SUB(Service, LDR) \
SUB(Service, LM) \
SUB(Service, Migration) \
SUB(Service, Mii) \
SUB(Service, MM) \
SUB(Service, NCM) \
SUB(Service, NFC) \
SUB(Service, NFP) \
SUB(Service, NIFM) \
SUB(Service, NIM) \
SUB(Service, NPNS) \
SUB(Service, NS) \
SUB(Service, NVDRV) \
SUB(Service, OLSC) \
SUB(Service, PCIE) \
SUB(Service, PCTL) \
SUB(Service, PCV) \
SUB(Service, PM) \
SUB(Service, PREPO) \
SUB(Service, PSC) \
SUB(Service, PSM) \
SUB(Service, SET) \
SUB(Service, SM) \
SUB(Service, SPL) \
SUB(Service, SSL) \
SUB(Service, TCAP) \
SUB(Service, Time) \
SUB(Service, USB) \
SUB(Service, VI) \
SUB(Service, WLAN) \
CLS(HW) \
SUB(HW, Memory) \
SUB(HW, LCD) \
SUB(HW, GPU) \
SUB(HW, AES) \
CLS(IPC) \
CLS(Frontend) \
CLS(Render) \
SUB(Render, Software) \
SUB(Render, OpenGL) \
SUB(Render, Vulkan) \
CLS(Audio) \
SUB(Audio, DSP) \
SUB(Audio, Sink) \
CLS(Input) \
CLS(Network) \
CLS(Loader) \
CLS(CheatEngine) \
CLS(Crypto) \
CLS(WebService)
// GetClassName is a macro defined by Windows.h, grrr...
const char* GetLogClassName(Class log_class) {
switch (log_class) {
#define CLS(x) \
case Class::x: \
return #x;
#define SUB(x, y) \
case Class::x##_##y: \
return #x "." #y;
ALL_LOG_CLASSES()
#undef CLS
#undef SUB
case Class::Count:
break;
}
return "Invalid";
}
const char* GetLevelName(Level log_level) {
#define LVL(x) \
case Level::x: \
return #x
switch (log_level) {
LVL(Trace);
LVL(Debug);
LVL(Info);
LVL(Warning);
LVL(Error);
LVL(Critical);
case Level::Count:
break;
}
#undef LVL
return "Invalid";
}
void SetGlobalFilter(const Filter& filter) {
Impl::Instance().SetGlobalFilter(filter);
}

View File

@@ -1,36 +1,24 @@
// Copyright 2014 Citra Emulator Project
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
#pragma once
#include <chrono>
#include <filesystem>
#include <memory>
#include <string>
#include <string_view>
#include "common/fs/file.h"
#include "common/logging/filter.h"
#include "common/logging/log.h"
namespace Common::FS {
class IOFile;
}
namespace Common::Log {
class Filter;
/**
* A log entry. Log entries are store in a structured format to permit more varied output
* formatting on different frontends, as well as facilitating filtering and aggregation.
*/
struct Entry {
std::chrono::microseconds timestamp;
Class log_class{};
Level log_level{};
const char* filename = nullptr;
unsigned int line_num = 0;
std::string function;
std::string message;
bool final_entry = false;
};
/**
* Interface for logging backends. As loggers can be created and removed at runtime, this can be
* used by a frontend for adding a custom logging backend as needed
@@ -38,6 +26,7 @@ struct Entry {
class Backend {
public:
virtual ~Backend() = default;
virtual void SetFilter(const Filter& new_filter) {
filter = new_filter;
}
@@ -53,6 +42,8 @@ private:
*/
class ConsoleBackend : public Backend {
public:
~ConsoleBackend() override;
static const char* Name() {
return "console";
}
@@ -67,6 +58,8 @@ public:
*/
class ColorConsoleBackend : public Backend {
public:
~ColorConsoleBackend() override;
static const char* Name() {
return "color_console";
}
@@ -83,6 +76,7 @@ public:
class FileBackend : public Backend {
public:
explicit FileBackend(const std::filesystem::path& filename);
~FileBackend() override;
static const char* Name() {
return "file";
@@ -95,7 +89,7 @@ public:
void Write(const Entry& entry) override;
private:
FS::IOFile file;
std::unique_ptr<FS::IOFile> file;
std::size_t bytes_written = 0;
};
@@ -104,6 +98,8 @@ private:
*/
class DebuggerBackend : public Backend {
public:
~DebuggerBackend() override;
static const char* Name() {
return "debugger";
}
@@ -119,17 +115,6 @@ void RemoveBackend(std::string_view backend_name);
Backend* GetBackend(std::string_view backend_name);
/**
* Returns the name of the passed log class as a C-string. Subclasses are separated by periods
* instead of underscores as in the enumeration.
*/
const char* GetLogClassName(Class log_class);
/**
* Returns the name of the passed log level as a C-string.
*/
const char* GetLevelName(Level log_level);
/**
* The global filter will prevent any messages from even being processed if they are filtered. Each
* backend can have a filter, but if the level is lower than the global filter, the backend will

View File

@@ -3,7 +3,6 @@
// Refer to the license.txt file included.
#include <algorithm>
#include "common/logging/backend.h"
#include "common/logging/filter.h"
#include "common/string_util.h"
@@ -22,7 +21,7 @@ Level GetLevelByName(const It begin, const It end) {
template <typename It>
Class GetClassByName(const It begin, const It end) {
for (ClassType i = 0; i < static_cast<ClassType>(Class::Count); ++i) {
for (u8 i = 0; i < static_cast<u8>(Class::Count); ++i) {
const char* level_name = GetLogClassName(static_cast<Class>(i));
if (Common::ComparePartialString(begin, end, level_name)) {
return static_cast<Class>(i);
@@ -62,6 +61,135 @@ bool ParseFilterRule(Filter& instance, Iterator begin, Iterator end) {
}
} // Anonymous namespace
/// Macro listing all log classes. Code should define CLS and SUB as desired before invoking this.
#define ALL_LOG_CLASSES() \
CLS(Log) \
CLS(Common) \
SUB(Common, Filesystem) \
SUB(Common, Memory) \
CLS(Core) \
SUB(Core, ARM) \
SUB(Core, Timing) \
CLS(Config) \
CLS(Debug) \
SUB(Debug, Emulated) \
SUB(Debug, GPU) \
SUB(Debug, Breakpoint) \
SUB(Debug, GDBStub) \
CLS(Kernel) \
SUB(Kernel, SVC) \
CLS(Service) \
SUB(Service, ACC) \
SUB(Service, Audio) \
SUB(Service, AM) \
SUB(Service, AOC) \
SUB(Service, APM) \
SUB(Service, ARP) \
SUB(Service, BCAT) \
SUB(Service, BPC) \
SUB(Service, BGTC) \
SUB(Service, BTDRV) \
SUB(Service, BTM) \
SUB(Service, Capture) \
SUB(Service, ERPT) \
SUB(Service, ETicket) \
SUB(Service, EUPLD) \
SUB(Service, Fatal) \
SUB(Service, FGM) \
SUB(Service, Friend) \
SUB(Service, FS) \
SUB(Service, GRC) \
SUB(Service, HID) \
SUB(Service, IRS) \
SUB(Service, LBL) \
SUB(Service, LDN) \
SUB(Service, LDR) \
SUB(Service, LM) \
SUB(Service, Migration) \
SUB(Service, Mii) \
SUB(Service, MM) \
SUB(Service, NCM) \
SUB(Service, NFC) \
SUB(Service, NFP) \
SUB(Service, NIFM) \
SUB(Service, NIM) \
SUB(Service, NPNS) \
SUB(Service, NS) \
SUB(Service, NVDRV) \
SUB(Service, OLSC) \
SUB(Service, PCIE) \
SUB(Service, PCTL) \
SUB(Service, PCV) \
SUB(Service, PM) \
SUB(Service, PREPO) \
SUB(Service, PSC) \
SUB(Service, PSM) \
SUB(Service, SET) \
SUB(Service, SM) \
SUB(Service, SPL) \
SUB(Service, SSL) \
SUB(Service, TCAP) \
SUB(Service, Time) \
SUB(Service, USB) \
SUB(Service, VI) \
SUB(Service, WLAN) \
CLS(HW) \
SUB(HW, Memory) \
SUB(HW, LCD) \
SUB(HW, GPU) \
SUB(HW, AES) \
CLS(IPC) \
CLS(Frontend) \
CLS(Render) \
SUB(Render, Software) \
SUB(Render, OpenGL) \
SUB(Render, Vulkan) \
CLS(Audio) \
SUB(Audio, DSP) \
SUB(Audio, Sink) \
CLS(Input) \
CLS(Network) \
CLS(Loader) \
CLS(CheatEngine) \
CLS(Crypto) \
CLS(WebService)
// GetClassName is a macro defined by Windows.h, grrr...
const char* GetLogClassName(Class log_class) {
switch (log_class) {
#define CLS(x) \
case Class::x: \
return #x;
#define SUB(x, y) \
case Class::x##_##y: \
return #x "." #y;
ALL_LOG_CLASSES()
#undef CLS
#undef SUB
case Class::Count:
break;
}
return "Invalid";
}
const char* GetLevelName(Level log_level) {
#define LVL(x) \
case Level::x: \
return #x
switch (log_level) {
LVL(Trace);
LVL(Debug);
LVL(Info);
LVL(Warning);
LVL(Error);
LVL(Critical);
case Level::Count:
break;
}
#undef LVL
return "Invalid";
}
Filter::Filter(Level default_level) {
ResetAll(default_level);
}

View File

@@ -5,12 +5,24 @@
#pragma once
#include <array>
#include <chrono>
#include <cstddef>
#include <string_view>
#include "common/logging/log.h"
namespace Common::Log {
/**
* Returns the name of the passed log class as a C-string. Subclasses are separated by periods
* instead of underscores as in the enumeration.
*/
const char* GetLogClassName(Class log_class);
/**
* Returns the name of the passed log level as a C-string.
*/
const char* GetLevelName(Level log_level);
/**
* Implements a log message filter which allows different log classes to have different minimum
* severity levels. The filter can be changed at runtime and can be parsed from a string to allow

View File

@@ -5,7 +5,7 @@
#pragma once
#include <fmt/format.h>
#include "common/common_types.h"
#include "common/logging/types.h"
namespace Common::Log {
@@ -18,124 +18,6 @@ constexpr const char* TrimSourcePath(std::string_view source) {
return source.data() + idx;
}
/// Specifies the severity or level of detail of the log message.
enum class Level : u8 {
Trace, ///< Extremely detailed and repetitive debugging information that is likely to
///< pollute logs.
Debug, ///< Less detailed debugging information.
Info, ///< Status information from important points during execution.
Warning, ///< Minor or potential problems found during execution of a task.
Error, ///< Major problems found during execution of a task that prevent it from being
///< completed.
Critical, ///< Major problems during execution that threaten the stability of the entire
///< application.
Count ///< Total number of logging levels
};
typedef u8 ClassType;
/**
* Specifies the sub-system that generated the log message.
*
* @note If you add a new entry here, also add a corresponding one to `ALL_LOG_CLASSES` in
* backend.cpp.
*/
enum class Class : ClassType {
Log, ///< Messages about the log system itself
Common, ///< Library routines
Common_Filesystem, ///< Filesystem interface library
Common_Memory, ///< Memory mapping and management functions
Core, ///< LLE emulation core
Core_ARM, ///< ARM CPU core
Core_Timing, ///< CoreTiming functions
Config, ///< Emulator configuration (including commandline)
Debug, ///< Debugging tools
Debug_Emulated, ///< Debug messages from the emulated programs
Debug_GPU, ///< GPU debugging tools
Debug_Breakpoint, ///< Logging breakpoints and watchpoints
Debug_GDBStub, ///< GDB Stub
Kernel, ///< The HLE implementation of the CTR kernel
Kernel_SVC, ///< Kernel system calls
Service, ///< HLE implementation of system services. Each major service
///< should have its own subclass.
Service_ACC, ///< The ACC (Accounts) service
Service_AM, ///< The AM (Applet manager) service
Service_AOC, ///< The AOC (AddOn Content) service
Service_APM, ///< The APM (Performance) service
Service_ARP, ///< The ARP service
Service_Audio, ///< The Audio (Audio control) service
Service_BCAT, ///< The BCAT service
Service_BGTC, ///< The BGTC (Background Task Controller) service
Service_BPC, ///< The BPC service
Service_BTDRV, ///< The Bluetooth driver service
Service_BTM, ///< The BTM service
Service_Capture, ///< The capture service
Service_ERPT, ///< The error reporting service
Service_ETicket, ///< The ETicket service
Service_EUPLD, ///< The error upload service
Service_Fatal, ///< The Fatal service
Service_FGM, ///< The FGM service
Service_Friend, ///< The friend service
Service_FS, ///< The FS (Filesystem) service
Service_GRC, ///< The game recording service
Service_HID, ///< The HID (Human interface device) service
Service_IRS, ///< The IRS service
Service_LBL, ///< The LBL (LCD backlight) service
Service_LDN, ///< The LDN (Local domain network) service
Service_LDR, ///< The loader service
Service_LM, ///< The LM (Logger) service
Service_Migration, ///< The migration service
Service_Mii, ///< The Mii service
Service_MM, ///< The MM (Multimedia) service
Service_NCM, ///< The NCM service
Service_NFC, ///< The NFC (Near-field communication) service
Service_NFP, ///< The NFP service
Service_NIFM, ///< The NIFM (Network interface) service
Service_NIM, ///< The NIM service
Service_NPNS, ///< The NPNS service
Service_NS, ///< The NS services
Service_NVDRV, ///< The NVDRV (Nvidia driver) service
Service_OLSC, ///< The OLSC service
Service_PCIE, ///< The PCIe service
Service_PCTL, ///< The PCTL (Parental control) service
Service_PCV, ///< The PCV service
Service_PM, ///< The PM service
Service_PREPO, ///< The PREPO (Play report) service
Service_PSC, ///< The PSC service
Service_PSM, ///< The PSM service
Service_SET, ///< The SET (Settings) service
Service_SM, ///< The SM (Service manager) service
Service_SPL, ///< The SPL service
Service_SSL, ///< The SSL service
Service_TCAP, ///< The TCAP service.
Service_Time, ///< The time service
Service_USB, ///< The USB (Universal Serial Bus) service
Service_VI, ///< The VI (Video interface) service
Service_WLAN, ///< The WLAN (Wireless local area network) service
HW, ///< Low-level hardware emulation
HW_Memory, ///< Memory-map and address translation
HW_LCD, ///< LCD register emulation
HW_GPU, ///< GPU control emulation
HW_AES, ///< AES engine emulation
IPC, ///< IPC interface
Frontend, ///< Emulator UI
Render, ///< Emulator video output and hardware acceleration
Render_Software, ///< Software renderer backend
Render_OpenGL, ///< OpenGL backend
Render_Vulkan, ///< Vulkan backend
Audio, ///< Audio emulation
Audio_DSP, ///< The HLE implementation of the DSP
Audio_Sink, ///< Emulator audio output backend
Loader, ///< ROM loader
CheatEngine, ///< Memory manipulation and engine VM functions
Crypto, ///< Cryptographic engine/functions
Input, ///< Input emulation
Network, ///< Network emulation
WebService, ///< Interface to yuzu Web Services
Count ///< Total number of logging classes
};
/// Logs a message to the global logger, using fmt
void FmtLogMessageImpl(Class log_class, Level log_level, const char* filename,
unsigned int line_num, const char* function, const char* format,

View File

@@ -11,7 +11,7 @@
#include "common/assert.h"
#include "common/common_funcs.h"
#include "common/logging/backend.h"
#include "common/logging/filter.h"
#include "common/logging/log.h"
#include "common/logging/text_formatter.h"
#include "common/string_util.h"

142
src/common/logging/types.h Normal file
View File

@@ -0,0 +1,142 @@
// Copyright 2021 yuzu Emulator Project
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
#include <chrono>
#include "common/common_types.h"
namespace Common::Log {
/// Specifies the severity or level of detail of the log message.
enum class Level : u8 {
Trace, ///< Extremely detailed and repetitive debugging information that is likely to
///< pollute logs.
Debug, ///< Less detailed debugging information.
Info, ///< Status information from important points during execution.
Warning, ///< Minor or potential problems found during execution of a task.
Error, ///< Major problems found during execution of a task that prevent it from being
///< completed.
Critical, ///< Major problems during execution that threaten the stability of the entire
///< application.
Count ///< Total number of logging levels
};
/**
* Specifies the sub-system that generated the log message.
*
* @note If you add a new entry here, also add a corresponding one to `ALL_LOG_CLASSES` in
* filter.cpp.
*/
enum class Class : u8 {
Log, ///< Messages about the log system itself
Common, ///< Library routines
Common_Filesystem, ///< Filesystem interface library
Common_Memory, ///< Memory mapping and management functions
Core, ///< LLE emulation core
Core_ARM, ///< ARM CPU core
Core_Timing, ///< CoreTiming functions
Config, ///< Emulator configuration (including commandline)
Debug, ///< Debugging tools
Debug_Emulated, ///< Debug messages from the emulated programs
Debug_GPU, ///< GPU debugging tools
Debug_Breakpoint, ///< Logging breakpoints and watchpoints
Debug_GDBStub, ///< GDB Stub
Kernel, ///< The HLE implementation of the CTR kernel
Kernel_SVC, ///< Kernel system calls
Service, ///< HLE implementation of system services. Each major service
///< should have its own subclass.
Service_ACC, ///< The ACC (Accounts) service
Service_AM, ///< The AM (Applet manager) service
Service_AOC, ///< The AOC (AddOn Content) service
Service_APM, ///< The APM (Performance) service
Service_ARP, ///< The ARP service
Service_Audio, ///< The Audio (Audio control) service
Service_BCAT, ///< The BCAT service
Service_BGTC, ///< The BGTC (Background Task Controller) service
Service_BPC, ///< The BPC service
Service_BTDRV, ///< The Bluetooth driver service
Service_BTM, ///< The BTM service
Service_Capture, ///< The capture service
Service_ERPT, ///< The error reporting service
Service_ETicket, ///< The ETicket service
Service_EUPLD, ///< The error upload service
Service_Fatal, ///< The Fatal service
Service_FGM, ///< The FGM service
Service_Friend, ///< The friend service
Service_FS, ///< The FS (Filesystem) service
Service_GRC, ///< The game recording service
Service_HID, ///< The HID (Human interface device) service
Service_IRS, ///< The IRS service
Service_LBL, ///< The LBL (LCD backlight) service
Service_LDN, ///< The LDN (Local domain network) service
Service_LDR, ///< The loader service
Service_LM, ///< The LM (Logger) service
Service_Migration, ///< The migration service
Service_Mii, ///< The Mii service
Service_MM, ///< The MM (Multimedia) service
Service_NCM, ///< The NCM service
Service_NFC, ///< The NFC (Near-field communication) service
Service_NFP, ///< The NFP service
Service_NIFM, ///< The NIFM (Network interface) service
Service_NIM, ///< The NIM service
Service_NPNS, ///< The NPNS service
Service_NS, ///< The NS services
Service_NVDRV, ///< The NVDRV (Nvidia driver) service
Service_OLSC, ///< The OLSC service
Service_PCIE, ///< The PCIe service
Service_PCTL, ///< The PCTL (Parental control) service
Service_PCV, ///< The PCV service
Service_PM, ///< The PM service
Service_PREPO, ///< The PREPO (Play report) service
Service_PSC, ///< The PSC service
Service_PSM, ///< The PSM service
Service_SET, ///< The SET (Settings) service
Service_SM, ///< The SM (Service manager) service
Service_SPL, ///< The SPL service
Service_SSL, ///< The SSL service
Service_TCAP, ///< The TCAP service.
Service_Time, ///< The time service
Service_USB, ///< The USB (Universal Serial Bus) service
Service_VI, ///< The VI (Video interface) service
Service_WLAN, ///< The WLAN (Wireless local area network) service
HW, ///< Low-level hardware emulation
HW_Memory, ///< Memory-map and address translation
HW_LCD, ///< LCD register emulation
HW_GPU, ///< GPU control emulation
HW_AES, ///< AES engine emulation
IPC, ///< IPC interface
Frontend, ///< Emulator UI
Render, ///< Emulator video output and hardware acceleration
Render_Software, ///< Software renderer backend
Render_OpenGL, ///< OpenGL backend
Render_Vulkan, ///< Vulkan backend
Audio, ///< Audio emulation
Audio_DSP, ///< The HLE implementation of the DSP
Audio_Sink, ///< Emulator audio output backend
Loader, ///< ROM loader
CheatEngine, ///< Memory manipulation and engine VM functions
Crypto, ///< Cryptographic engine/functions
Input, ///< Input emulation
Network, ///< Network emulation
WebService, ///< Interface to yuzu Web Services
Count ///< Total number of logging classes
};
/**
* A log entry. Log entries are store in a structured format to permit more varied output
* formatting on different frontends, as well as facilitating filtering and aggregation.
*/
struct Entry {
std::chrono::microseconds timestamp;
Class log_class{};
Level log_level{};
const char* filename = nullptr;
unsigned int line_num = 0;
std::string function;
std::string message;
bool final_entry = false;
};
} // namespace Common::Log

View File

@@ -111,6 +111,8 @@ struct PageTable {
VirtualBuffer<u64> backing_addr;
size_t current_address_space_width_in_bits;
u8* fastmem_arena;
};
} // namespace Common

View File

@@ -55,9 +55,11 @@ void LogSettings() {
log_setting("Renderer_UseAsynchronousGpuEmulation",
values.use_asynchronous_gpu_emulation.GetValue());
log_setting("Renderer_UseNvdecEmulation", values.use_nvdec_emulation.GetValue());
log_setting("Renderer_AccelerateASTC", values.accelerate_astc.GetValue());
log_setting("Renderer_UseVsync", values.use_vsync.GetValue());
log_setting("Renderer_UseAssemblyShaders", values.use_assembly_shaders.GetValue());
log_setting("Renderer_UseAsynchronousShaders", values.use_asynchronous_shaders.GetValue());
log_setting("Renderer_UseGarbageCollection", values.use_caches_gc.GetValue());
log_setting("Renderer_AnisotropicFilteringLevel", values.max_anisotropy.GetValue());
log_setting("Audio_OutputEngine", values.sink_id);
log_setting("Audio_EnableAudioStretching", values.enable_audio_stretching.GetValue());
@@ -90,6 +92,13 @@ bool IsGPULevelHigh() {
values.gpu_accuracy.GetValue() == GPUAccuracy::High;
}
bool IsFastmemEnabled() {
if (values.cpu_accuracy.GetValue() == CPUAccuracy::DebugMode) {
return values.cpuopt_fastmem;
}
return true;
}
float Volume() {
if (values.audio_muted) {
return 0.0f;
@@ -115,6 +124,7 @@ void RestoreGlobalState(bool is_powered_on) {
values.cpuopt_unsafe_unfuse_fma.SetGlobal(true);
values.cpuopt_unsafe_reduce_fp_error.SetGlobal(true);
values.cpuopt_unsafe_inaccurate_nan.SetGlobal(true);
values.cpuopt_unsafe_fastmem_check.SetGlobal(true);
// Renderer
values.renderer_backend.SetGlobal(true);
@@ -127,10 +137,12 @@ void RestoreGlobalState(bool is_powered_on) {
values.gpu_accuracy.SetGlobal(true);
values.use_asynchronous_gpu_emulation.SetGlobal(true);
values.use_nvdec_emulation.SetGlobal(true);
values.accelerate_astc.SetGlobal(true);
values.use_vsync.SetGlobal(true);
values.use_assembly_shaders.SetGlobal(true);
values.use_asynchronous_shaders.SetGlobal(true);
values.use_fast_gpu_time.SetGlobal(true);
values.use_caches_gc.SetGlobal(true);
values.bg_red.SetGlobal(true);
values.bg_green.SetGlobal(true);
values.bg_blue.SetGlobal(true);

View File

@@ -125,10 +125,12 @@ struct Values {
bool cpuopt_const_prop;
bool cpuopt_misc_ir;
bool cpuopt_reduce_misalign_checks;
bool cpuopt_fastmem;
Setting<bool> cpuopt_unsafe_unfuse_fma;
Setting<bool> cpuopt_unsafe_reduce_fp_error;
Setting<bool> cpuopt_unsafe_inaccurate_nan;
Setting<bool> cpuopt_unsafe_fastmem_check;
// Renderer
Setting<RendererBackend> renderer_backend;
@@ -145,10 +147,12 @@ struct Values {
Setting<GPUAccuracy> gpu_accuracy;
Setting<bool> use_asynchronous_gpu_emulation;
Setting<bool> use_nvdec_emulation;
Setting<bool> accelerate_astc;
Setting<bool> use_vsync;
Setting<bool> use_assembly_shaders;
Setting<bool> use_asynchronous_shaders;
Setting<bool> use_fast_gpu_time;
Setting<bool> use_caches_gc;
Setting<float> bg_red;
Setting<float> bg_green;
@@ -216,6 +220,7 @@ struct Values {
std::string program_args;
bool dump_exefs;
bool dump_nso;
bool enable_fs_access_log;
bool reporting_services;
bool quest_flag;
bool disable_macro_jit;
@@ -249,6 +254,8 @@ void SetConfiguringGlobal(bool is_global);
bool IsGPULevelExtreme();
bool IsGPULevelHigh();
bool IsFastmemEnabled();
float Volume();
std::string GetTimeZoneString();

View File

@@ -128,6 +128,7 @@ std::shared_ptr<Dynarmic::A32::Jit> ARM_Dynarmic_32::MakeJit(Common::PageTable*
if (page_table) {
config.page_table = reinterpret_cast<std::array<std::uint8_t*, NUM_PAGE_TABLE_ENTRIES>*>(
page_table->pointers.data());
config.fastmem_pointer = page_table->fastmem_arena;
}
config.absolute_offset_page_table = true;
config.page_table_pointer_mask_bits = Common::PageTable::ATTRIBUTE_BITS;
@@ -143,7 +144,7 @@ std::shared_ptr<Dynarmic::A32::Jit> ARM_Dynarmic_32::MakeJit(Common::PageTable*
// Code cache size
config.code_cache_size = 512 * 1024 * 1024;
config.far_code_offset = 256 * 1024 * 1024;
config.far_code_offset = 400 * 1024 * 1024;
// Safe optimizations
if (Settings::values.cpu_accuracy.GetValue() == Settings::CPUAccuracy::DebugMode) {
@@ -171,6 +172,9 @@ std::shared_ptr<Dynarmic::A32::Jit> ARM_Dynarmic_32::MakeJit(Common::PageTable*
if (!Settings::values.cpuopt_reduce_misalign_checks) {
config.only_detect_misalignment_via_page_table_on_page_boundary = false;
}
if (!Settings::values.cpuopt_fastmem) {
config.fastmem_pointer = nullptr;
}
}
// Unsafe optimizations

View File

@@ -160,6 +160,10 @@ std::shared_ptr<Dynarmic::A64::Jit> ARM_Dynarmic_64::MakeJit(Common::PageTable*
config.absolute_offset_page_table = true;
config.detect_misaligned_access_via_page_table = 16 | 32 | 64 | 128;
config.only_detect_misalignment_via_page_table_on_page_boundary = true;
config.fastmem_pointer = page_table->fastmem_arena;
config.fastmem_address_space_bits = address_space_bits;
config.silently_mirror_fastmem = false;
}
// Multi-process state
@@ -181,7 +185,7 @@ std::shared_ptr<Dynarmic::A64::Jit> ARM_Dynarmic_64::MakeJit(Common::PageTable*
// Code cache size
config.code_cache_size = 512 * 1024 * 1024;
config.far_code_offset = 256 * 1024 * 1024;
config.far_code_offset = 400 * 1024 * 1024;
// Safe optimizations
if (Settings::values.cpu_accuracy.GetValue() == Settings::CPUAccuracy::DebugMode) {
@@ -209,6 +213,9 @@ std::shared_ptr<Dynarmic::A64::Jit> ARM_Dynarmic_64::MakeJit(Common::PageTable*
if (!Settings::values.cpuopt_reduce_misalign_checks) {
config.only_detect_misalignment_via_page_table_on_page_boundary = false;
}
if (!Settings::values.cpuopt_fastmem) {
config.fastmem_pointer = nullptr;
}
}
// Unsafe optimizations
@@ -223,6 +230,9 @@ std::shared_ptr<Dynarmic::A64::Jit> ARM_Dynarmic_64::MakeJit(Common::PageTable*
if (Settings::values.cpuopt_unsafe_inaccurate_nan.GetValue()) {
config.optimizations |= Dynarmic::OptimizationFlag::Unsafe_InaccurateNaN;
}
if (Settings::values.cpuopt_unsafe_fastmem_check.GetValue()) {
config.fastmem_address_space_bits = 64;
}
}
return std::make_shared<Dynarmic::A64::Jit>(config);

View File

@@ -6,7 +6,7 @@
namespace Core {
DeviceMemory::DeviceMemory() : buffer{DramMemoryMap::Size} {}
DeviceMemory::DeviceMemory() : buffer{DramMemoryMap::Size, 1ULL << 39} {}
DeviceMemory::~DeviceMemory() = default;
} // namespace Core

View File

@@ -5,7 +5,7 @@
#pragma once
#include "common/common_types.h"
#include "common/virtual_buffer.h"
#include "common/host_memory.h"
namespace Core {
@@ -21,27 +21,30 @@ enum : u64 {
};
}; // namespace DramMemoryMap
class DeviceMemory : NonCopyable {
class DeviceMemory {
public:
explicit DeviceMemory();
~DeviceMemory();
DeviceMemory& operator=(const DeviceMemory&) = delete;
DeviceMemory(const DeviceMemory&) = delete;
template <typename T>
PAddr GetPhysicalAddr(const T* ptr) const {
return (reinterpret_cast<uintptr_t>(ptr) - reinterpret_cast<uintptr_t>(buffer.data())) +
return (reinterpret_cast<uintptr_t>(ptr) -
reinterpret_cast<uintptr_t>(buffer.BackingBasePointer())) +
DramMemoryMap::Base;
}
u8* GetPointer(PAddr addr) {
return buffer.data() + (addr - DramMemoryMap::Base);
return buffer.BackingBasePointer() + (addr - DramMemoryMap::Base);
}
const u8* GetPointer(PAddr addr) const {
return buffer.data() + (addr - DramMemoryMap::Base);
return buffer.BackingBasePointer() + (addr - DramMemoryMap::Base);
}
private:
Common::VirtualBuffer<u8> buffer;
Common::HostMemory buffer;
};
} // namespace Core

View File

@@ -150,7 +150,9 @@ void ProgramMetadata::Print() const {
LOG_DEBUG(Service_FS, " > Is Retail: {}", acid_header.is_retail ? "YES" : "NO");
LOG_DEBUG(Service_FS, "Title ID Min: 0x{:016X}", acid_header.title_id_min);
LOG_DEBUG(Service_FS, "Title ID Max: 0x{:016X}", acid_header.title_id_max);
LOG_DEBUG(Service_FS, "Filesystem Access: 0x{:016X}\n", acid_file_access.permissions);
u64_le permissions_l; // local copy to fix alignment error
std::memcpy(&permissions_l, &acid_file_access.permissions, sizeof(permissions_l));
LOG_DEBUG(Service_FS, "Filesystem Access: 0x{:016X}\n", permissions_l);
// Begin ACI0 printing (actual perms, unsigned)
LOG_DEBUG(Service_FS, "Magic: {:.4}", aci_header.magic.data());

View File

@@ -6,7 +6,6 @@
#include <numeric>
#include <string>
#include "common/fs/path_util.h"
#include "common/logging/backend.h"
#include "core/file_sys/mode.h"
#include "core/file_sys/vfs.h"

View File

@@ -14,7 +14,6 @@
#endif
#include "common/fs/path_util.h"
#include "common/logging/backend.h"
#include "core/file_sys/vfs.h"
#include "core/file_sys/vfs_libzip.h"
#include "core/file_sys/vfs_vector.h"

View File

@@ -57,11 +57,11 @@ bool SessionRequestManager::HasSessionRequestHandler(const HLERequestContext& co
}
void SessionRequestHandler::ClientConnected(KServerSession* session) {
session->SetSessionHandler(shared_from_this());
session->ClientConnected(shared_from_this());
}
void SessionRequestHandler::ClientDisconnected(KServerSession* session) {
session->SetSessionHandler(nullptr);
session->ClientDisconnected();
}
HLERequestContext::HLERequestContext(KernelCore& kernel_, Core::Memory::Memory& memory_,

View File

@@ -28,6 +28,9 @@ void KClientPort::Initialize(KPort* parent_port_, s32 max_sessions_, std::string
void KClientPort::OnSessionFinalized() {
KScopedSchedulerLock sl{kernel};
// This might happen if a session was improperly used with this port.
ASSERT_MSG(num_sessions > 0, "num_sessions is invalid");
const auto prev = num_sessions--;
if (prev == max_sessions) {
this->NotifyAvailable();
@@ -66,7 +69,7 @@ ResultCode KClientPort::CreateSession(KClientSession** out,
// Update the session counts.
{
// Atomically increment the number of sessions.
s32 new_sessions;
s32 new_sessions{};
{
const auto max = max_sessions;
auto cur_sessions = num_sessions.load(std::memory_order_acquire);

View File

@@ -18,41 +18,58 @@ class KernelCore;
class KLightConditionVariable {
public:
explicit KLightConditionVariable(KernelCore& kernel_)
: thread_queue(kernel_), kernel(kernel_) {}
explicit KLightConditionVariable(KernelCore& kernel_) : kernel{kernel_} {}
void Wait(KLightLock* lock, s64 timeout = -1) {
WaitImpl(lock, timeout);
lock->Lock();
void Wait(KLightLock* lock, s64 timeout = -1, bool allow_terminating_thread = true) {
WaitImpl(lock, timeout, allow_terminating_thread);
}
void Broadcast() {
KScopedSchedulerLock lk{kernel};
while (thread_queue.WakeupFrontThread() != nullptr) {
// We want to signal all threads, and so should continue waking up until there's nothing
// to wake.
// Signal all threads.
for (auto& thread : wait_list) {
thread.SetState(ThreadState::Runnable);
}
}
private:
void WaitImpl(KLightLock* lock, s64 timeout) {
void WaitImpl(KLightLock* lock, s64 timeout, bool allow_terminating_thread) {
KThread* owner = GetCurrentThreadPointer(kernel);
// Sleep the thread.
{
KScopedSchedulerLockAndSleep lk(kernel, owner, timeout);
lock->Unlock();
KScopedSchedulerLockAndSleep lk{kernel, owner, timeout};
if (!thread_queue.SleepThread(owner)) {
if (!allow_terminating_thread && owner->IsTerminationRequested()) {
lk.CancelSleep();
return;
}
lock->Unlock();
// Set the thread as waiting.
GetCurrentThread(kernel).SetState(ThreadState::Waiting);
// Add the thread to the queue.
wait_list.push_back(GetCurrentThread(kernel));
}
// Remove the thread from the wait list.
{
KScopedSchedulerLock sl{kernel};
wait_list.erase(wait_list.iterator_to(GetCurrentThread(kernel)));
}
// Cancel the task that the sleep setup.
kernel.TimeManager().UnscheduleTimeEvent(owner);
// Re-acquire the lock.
lock->Lock();
}
KThreadQueue thread_queue;
KernelCore& kernel;
KThread::WaiterList wait_list{};
};
} // namespace Kernel

View File

@@ -59,11 +59,7 @@ void KLightLock::LockSlowPath(uintptr_t _owner, uintptr_t _cur_thread) {
owner_thread->AddWaiter(cur_thread);
// Set thread states.
if (cur_thread->GetState() == ThreadState::Runnable) {
cur_thread->SetState(ThreadState::Waiting);
} else {
KScheduler::SetSchedulerUpdateNeeded(kernel);
}
cur_thread->SetState(ThreadState::Waiting);
if (owner_thread->IsSuspended()) {
owner_thread->ContinueIfHasKernelWaiters();
@@ -73,10 +69,9 @@ void KLightLock::LockSlowPath(uintptr_t _owner, uintptr_t _cur_thread) {
// We're no longer waiting on the lock owner.
{
KScopedSchedulerLock sl{kernel};
KThread* owner_thread = cur_thread->GetLockOwner();
if (owner_thread) {
if (KThread* owner_thread = cur_thread->GetLockOwner(); owner_thread != nullptr) {
owner_thread->RemoveWaiter(cur_thread);
KScheduler::SetSchedulerUpdateNeeded(kernel);
}
}
}
@@ -95,17 +90,13 @@ void KLightLock::UnlockSlowPath(uintptr_t _cur_thread) {
// Pass the lock to the next owner.
uintptr_t next_tag = 0;
if (next_owner) {
if (next_owner != nullptr) {
next_tag = reinterpret_cast<uintptr_t>(next_owner);
if (num_waiters > 1) {
next_tag |= 0x1;
}
if (next_owner->GetState() == ThreadState::Waiting) {
next_owner->SetState(ThreadState::Runnable);
} else {
KScheduler::SetSchedulerUpdateNeeded(kernel);
}
next_owner->SetState(ThreadState::Runnable);
if (next_owner->IsSuspended()) {
next_owner->ContinueIfHasKernelWaiters();

View File

@@ -201,17 +201,15 @@ bool KProcess::ReleaseUserException(KThread* thread) {
// Remove waiter thread.
s32 num_waiters{};
KThread* next = thread->RemoveWaiterByKey(
std::addressof(num_waiters),
reinterpret_cast<uintptr_t>(std::addressof(exception_thread)));
if (next != nullptr) {
if (next->GetState() == ThreadState::Waiting) {
next->SetState(ThreadState::Runnable);
} else {
KScheduler::SetSchedulerUpdateNeeded(kernel);
}
if (KThread* next = thread->RemoveWaiterByKey(
std::addressof(num_waiters),
reinterpret_cast<uintptr_t>(std::addressof(exception_thread)));
next != nullptr) {
next->SetState(ThreadState::Runnable);
}
KScheduler::SetSchedulerUpdateNeeded(kernel);
return true;
} else {
return false;

View File

@@ -117,7 +117,7 @@ bool KResourceLimit::Reserve(LimitableResource which, s64 value, s64 timeout) {
if (current_hints[index] + value <= limit_values[index] &&
(timeout < 0 || core_timing->GetGlobalTimeNs().count() < timeout)) {
waiter_count++;
cond_var.Wait(&lock, timeout);
cond_var.Wait(&lock, timeout, false);
waiter_count--;
} else {
break;

View File

@@ -62,15 +62,14 @@ public:
void OnClientClosed();
/**
* Sets the HLE handler for the session. This handler will be called to service IPC requests
* instead of the regular IPC machinery. (The regular IPC machinery is currently not
* implemented.)
*/
void SetSessionHandler(SessionRequestHandlerPtr handler) {
void ClientConnected(SessionRequestHandlerPtr handler) {
manager->SetSessionHandler(std::move(handler));
}
void ClientDisconnected() {
manager = nullptr;
}
/**
* Handle a sync request from the emulated application.
*

View File

@@ -449,8 +449,8 @@ static ResultCode CancelSynchronization(Core::System& system, Handle handle) {
// Get the thread from its handle.
KScopedAutoObject thread =
system.Kernel().CurrentProcess()->GetHandleTable().GetObject<KThread>(
static_cast<Handle>(handle));
system.Kernel().CurrentProcess()->GetHandleTable().GetObject<KThread>(handle);
R_UNLESS(thread.IsNotNull(), ResultInvalidHandle);
// Cancel the thread's wait.
thread->WaitCancel();

View File

@@ -19,7 +19,6 @@
#include "common/fs/fs.h"
#include "common/fs/path_util.h"
#include "common/hex_util.h"
#include "common/logging/backend.h"
#include "common/logging/log.h"
#include "common/settings.h"
#include "core/core.h"

View File

@@ -13,6 +13,7 @@
#include "common/common_types.h"
#include "common/hex_util.h"
#include "common/logging/log.h"
#include "common/settings.h"
#include "common/string_util.h"
#include "core/core.h"
#include "core/file_sys/directory.h"
@@ -785,6 +786,10 @@ FSP_SRV::FSP_SRV(Core::System& system_)
};
// clang-format on
RegisterHandlers(functions);
if (Settings::values.enable_fs_access_log) {
access_log_mode = AccessLogMode::SdCard;
}
}
FSP_SRV::~FSP_SRV() = default;
@@ -1041,9 +1046,9 @@ void FSP_SRV::DisableAutoSaveDataCreation(Kernel::HLERequestContext& ctx) {
void FSP_SRV::SetGlobalAccessLogMode(Kernel::HLERequestContext& ctx) {
IPC::RequestParser rp{ctx};
log_mode = rp.PopEnum<LogMode>();
access_log_mode = rp.PopEnum<AccessLogMode>();
LOG_DEBUG(Service_FS, "called, log_mode={:08X}", log_mode);
LOG_DEBUG(Service_FS, "called, access_log_mode={}", access_log_mode);
IPC::ResponseBuilder rb{ctx, 2};
rb.Push(ResultSuccess);
@@ -1054,7 +1059,7 @@ void FSP_SRV::GetGlobalAccessLogMode(Kernel::HLERequestContext& ctx) {
IPC::ResponseBuilder rb{ctx, 3};
rb.Push(ResultSuccess);
rb.PushEnum(log_mode);
rb.PushEnum(access_log_mode);
}
void FSP_SRV::OutputAccessLogToSdCard(Kernel::HLERequestContext& ctx) {
@@ -1062,9 +1067,9 @@ void FSP_SRV::OutputAccessLogToSdCard(Kernel::HLERequestContext& ctx) {
auto log = Common::StringFromFixedZeroTerminatedBuffer(
reinterpret_cast<const char*>(raw.data()), raw.size());
LOG_DEBUG(Service_FS, "called, log='{}'", log);
LOG_DEBUG(Service_FS, "called");
reporter.SaveFilesystemAccessReport(log_mode, std::move(log));
reporter.SaveFSAccessLog(log);
IPC::ResponseBuilder rb{ctx, 2};
rb.Push(ResultSuccess);

View File

@@ -24,11 +24,10 @@ enum class AccessLogVersion : u32 {
Latest = V7_0_0,
};
enum class LogMode : u32 {
Off,
enum class AccessLogMode : u32 {
None,
Log,
RedirectToSdCard,
LogToSdCard = Log | RedirectToSdCard,
SdCard,
};
class FSP_SRV final : public ServiceFramework<FSP_SRV> {
@@ -59,13 +58,12 @@ private:
FileSystemController& fsc;
const FileSys::ContentProvider& content_provider;
const Core::Reporter& reporter;
FileSys::VirtualFile romfs;
u64 current_process_id = 0;
u32 access_log_program_index = 0;
LogMode log_mode = LogMode::LogToSdCard;
const Core::Reporter& reporter;
AccessLogMode access_log_mode = AccessLogMode::None;
};
} // namespace Service::FileSystem

View File

@@ -236,7 +236,7 @@ Hid::Hid(Core::System& system_) : ServiceFramework{system_, "hid"} {
{80, &Hid::GetGyroscopeZeroDriftMode, "GetGyroscopeZeroDriftMode"},
{81, &Hid::ResetGyroscopeZeroDriftMode, "ResetGyroscopeZeroDriftMode"},
{82, &Hid::IsSixAxisSensorAtRest, "IsSixAxisSensorAtRest"},
{83, nullptr, "IsFirmwareUpdateAvailableForSixAxisSensor"},
{83, &Hid::IsFirmwareUpdateAvailableForSixAxisSensor, "IsFirmwareUpdateAvailableForSixAxisSensor"},
{91, &Hid::ActivateGesture, "ActivateGesture"},
{100, &Hid::SetSupportedNpadStyleSet, "SetSupportedNpadStyleSet"},
{101, &Hid::GetSupportedNpadStyleSet, "GetSupportedNpadStyleSet"},
@@ -710,6 +710,27 @@ void Hid::IsSixAxisSensorAtRest(Kernel::HLERequestContext& ctx) {
.IsSixAxisSensorAtRest());
}
void Hid::IsFirmwareUpdateAvailableForSixAxisSensor(Kernel::HLERequestContext& ctx) {
IPC::RequestParser rp{ctx};
struct Parameters {
Controller_NPad::DeviceHandle sixaxis_handle;
INSERT_PADDING_WORDS_NOINIT(1);
u64 applet_resource_user_id;
};
const auto parameters{rp.PopRaw<Parameters>()};
LOG_WARNING(
Service_HID,
"(STUBBED) called, npad_type={}, npad_id={}, device_index={}, applet_resource_user_id={}",
parameters.sixaxis_handle.npad_type, parameters.sixaxis_handle.npad_id,
parameters.sixaxis_handle.device_index, parameters.applet_resource_user_id);
IPC::ResponseBuilder rb{ctx, 3};
rb.Push(ResultSuccess);
rb.Push(false);
}
void Hid::ActivateGesture(Kernel::HLERequestContext& ctx) {
IPC::RequestParser rp{ctx};
struct Parameters {

View File

@@ -100,6 +100,7 @@ private:
void GetGyroscopeZeroDriftMode(Kernel::HLERequestContext& ctx);
void ResetGyroscopeZeroDriftMode(Kernel::HLERequestContext& ctx);
void IsSixAxisSensorAtRest(Kernel::HLERequestContext& ctx);
void IsFirmwareUpdateAvailableForSixAxisSensor(Kernel::HLERequestContext& ctx);
void ActivateGesture(Kernel::HLERequestContext& ctx);
void SetSupportedNpadStyleSet(Kernel::HLERequestContext& ctx);
void GetSupportedNpadStyleSet(Kernel::HLERequestContext& ctx);

View File

@@ -51,6 +51,24 @@ struct hash<Service::LM::LogPacketHeaderEntry> {
} // namespace std
namespace Service::LM {
namespace {
std::string_view NameOf(LogSeverity severity) {
switch (severity) {
case LogSeverity::Trace:
return "TRACE";
case LogSeverity::Info:
return "INFO";
case LogSeverity::Warning:
return "WARNING";
case LogSeverity::Error:
return "ERROR";
case LogSeverity::Fatal:
return "FATAL";
default:
return "UNKNOWN";
}
}
} // Anonymous namespace
enum class LogDestination : u32 {
TargetManager = 1 << 0,
@@ -262,33 +280,8 @@ private:
if (text_log) {
output_log += fmt::format("Log Text: {}\n", *text_log);
}
switch (entry.severity) {
case LogSeverity::Trace:
LOG_DEBUG(Service_LM, "LogManager TRACE ({}):\n{}", DestinationToString(destination),
output_log);
break;
case LogSeverity::Info:
LOG_INFO(Service_LM, "LogManager INFO ({}):\n{}", DestinationToString(destination),
output_log);
break;
case LogSeverity::Warning:
LOG_WARNING(Service_LM, "LogManager WARNING ({}):\n{}",
DestinationToString(destination), output_log);
break;
case LogSeverity::Error:
LOG_ERROR(Service_LM, "LogManager ERROR ({}):\n{}", DestinationToString(destination),
output_log);
break;
case LogSeverity::Fatal:
LOG_CRITICAL(Service_LM, "LogManager FATAL ({}):\n{}", DestinationToString(destination),
output_log);
break;
default:
LOG_CRITICAL(Service_LM, "LogManager UNKNOWN ({}):\n{}",
DestinationToString(destination), output_log);
break;
}
LOG_DEBUG(Service_LM, "LogManager {} ({}):\n{}", NameOf(entry.severity),
DestinationToString(destination), output_log);
}
static std::string DestinationToString(LogDestination destination) {

View File

@@ -40,9 +40,11 @@ namespace SM {
class ServiceManager;
}
static const int kMaxPortSize = 8; ///< Maximum size of a port name (8 characters)
/// Arbitrary default number of maximum connections to an HLE service.
static const u32 DefaultMaxSessions = 64;
/// Default number of maximum connections to a server session.
static constexpr u32 ServerSessionCountMax = 0x40;
static_assert(ServerSessionCountMax == 0x40,
"ServerSessionCountMax isn't 0x40 somehow, this assert is a reminder that this will "
"break lots of things");
/**
* This is an non-templated base of ServiceFramework to reduce code bloat and compilation times, it
@@ -178,7 +180,7 @@ protected:
* connected to this service at the same time.
*/
explicit ServiceFramework(Core::System& system_, const char* service_name_,
u32 max_sessions_ = DefaultMaxSessions)
u32 max_sessions_ = ServerSessionCountMax)
: ServiceFrameworkBase(system_, service_name_, max_sessions_, Invoker) {}
/// Registers handlers in the service.

View File

@@ -151,31 +151,23 @@ ResultVal<Kernel::KClientSession*> SM::GetServiceImpl(Kernel::HLERequestContext&
std::string name(PopServiceName(rp));
// Find the named port.
auto result = service_manager.GetServicePort(name);
if (result.Failed()) {
LOG_ERROR(Service_SM, "called service={} -> error 0x{:08X}", name, result.Code().raw);
return result.Code();
auto port_result = service_manager.GetServicePort(name);
if (port_result.Failed()) {
LOG_ERROR(Service_SM, "called service={} -> error 0x{:08X}", name, port_result.Code().raw);
return port_result.Code();
}
auto* port = result.Unwrap();
// Reserve a new session from the process resource limit.
Kernel::KScopedResourceReservation session_reservation(
kernel.CurrentProcess()->GetResourceLimit(), Kernel::LimitableResource::Sessions);
R_UNLESS(session_reservation.Succeeded(), Kernel::ResultLimitReached);
auto& port = port_result.Unwrap()->GetClientPort();
// Create a new session.
auto* session = Kernel::KSession::Create(kernel);
session->Initialize(&port->GetClientPort(), std::move(name));
// Commit the session reservation.
session_reservation.Commit();
// Enqueue the session with the named port.
port->EnqueueSession(&session->GetServerSession());
Kernel::KClientSession* session{};
if (const auto result = port.CreateSession(std::addressof(session)); result.IsError()) {
LOG_ERROR(Service_SM, "called service={} -> error 0x{:08X}", name, result.raw);
return result;
}
LOG_DEBUG(Service_SM, "called service={} -> session={}", name, session->GetId());
return MakeResult(&session->GetClientSession());
return MakeResult(session);
}
void SM::RegisterService(Kernel::HLERequestContext& ctx) {

View File

@@ -12,6 +12,7 @@
#include "common/common_types.h"
#include "common/logging/log.h"
#include "common/page_table.h"
#include "common/settings.h"
#include "common/swap.h"
#include "core/arm/arm_interface.h"
#include "core/core.h"
@@ -32,6 +33,7 @@ struct Memory::Impl {
void SetCurrentPageTable(Kernel::KProcess& process, u32 core_id) {
current_page_table = &process.PageTable().PageTableImpl();
current_page_table->fastmem_arena = system.DeviceMemory().buffer.VirtualBasePointer();
const std::size_t address_space_width = process.PageTable().GetAddressSpaceWidth();
@@ -41,13 +43,23 @@ struct Memory::Impl {
void MapMemoryRegion(Common::PageTable& page_table, VAddr base, u64 size, PAddr target) {
ASSERT_MSG((size & PAGE_MASK) == 0, "non-page aligned size: {:016X}", size);
ASSERT_MSG((base & PAGE_MASK) == 0, "non-page aligned base: {:016X}", base);
ASSERT_MSG(target >= DramMemoryMap::Base && target < DramMemoryMap::End,
"Out of bounds target: {:016X}", target);
MapPages(page_table, base / PAGE_SIZE, size / PAGE_SIZE, target, Common::PageType::Memory);
if (Settings::IsFastmemEnabled()) {
system.DeviceMemory().buffer.Map(base, target - DramMemoryMap::Base, size);
}
}
void UnmapRegion(Common::PageTable& page_table, VAddr base, u64 size) {
ASSERT_MSG((size & PAGE_MASK) == 0, "non-page aligned size: {:016X}", size);
ASSERT_MSG((base & PAGE_MASK) == 0, "non-page aligned base: {:016X}", base);
MapPages(page_table, base / PAGE_SIZE, size / PAGE_SIZE, 0, Common::PageType::Unmapped);
if (Settings::IsFastmemEnabled()) {
system.DeviceMemory().buffer.Unmap(base, size);
}
}
bool IsValidVirtualAddress(const Kernel::KProcess& process, const VAddr vaddr) const {
@@ -466,6 +478,12 @@ struct Memory::Impl {
if (vaddr == 0) {
return;
}
if (Settings::IsFastmemEnabled()) {
const bool is_read_enable = Settings::IsGPULevelHigh() || !cached;
system.DeviceMemory().buffer.Protect(vaddr, size, is_read_enable, !cached);
}
// Iterate over a contiguous CPU address space, which corresponds to the specified GPU
// address space, marking the region as un/cached. The region is marked un/cached at a
// granularity of CPU pages, hence why we iterate on a CPU page basis (note: GPU page size

View File

@@ -195,7 +195,9 @@ json GetHLERequestContextData(Kernel::HLERequestContext& ctx, Core::Memory::Memo
namespace Core {
Reporter::Reporter(System& system_) : system(system_) {}
Reporter::Reporter(System& system_) : system(system_) {
ClearFSAccessLog();
}
Reporter::~Reporter() = default;
@@ -362,22 +364,12 @@ 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;
void Reporter::SaveFSAccessLog(std::string_view log_message) const {
const auto access_log_path =
Common::FS::GetYuzuPath(Common::FS::YuzuPath::SDMCDir) / "FsAccessLog.txt";
const auto timestamp = GetTimestamp();
const auto title_id = system.CurrentProcess()->GetTitleID();
json out;
out["yuzu_version"] = GetYuzuVersionData();
out["report_common"] = GetReportCommonData(title_id, ResultSuccess, 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(Common::FS::AppendStringToFile(access_log_path, Common::FS::FileType::TextFile,
log_message));
}
void Reporter::SaveUserReport() const {
@@ -392,6 +384,18 @@ void Reporter::SaveUserReport() const {
GetPath("user_report", title_id, timestamp));
}
void Reporter::ClearFSAccessLog() const {
const auto access_log_path =
Common::FS::GetYuzuPath(Common::FS::YuzuPath::SDMCDir) / "FsAccessLog.txt";
Common::FS::IOFile access_log_file{access_log_path, Common::FS::FileAccessMode::Write,
Common::FS::FileType::TextFile};
if (!access_log_file.IsOpen()) {
LOG_ERROR(Common_Filesystem, "Failed to clear the filesystem access log.");
}
}
bool Reporter::IsReportingEnabled() const {
return Settings::values.reporting_services;
}

View File

@@ -16,10 +16,6 @@ namespace Kernel {
class HLERequestContext;
} // namespace Kernel
namespace Service::FileSystem {
enum class LogMode : u32;
}
namespace Service::LM {
struct LogMessage;
} // namespace Service::LM
@@ -69,14 +65,15 @@ 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 SaveFSAccessLog(std::string_view log_message) const;
// Can be used anywhere to generate a backtrace and general info report at any point during
// execution. Not intended to be used for anything other than debugging or testing.
void SaveUserReport() const;
private:
void ClearFSAccessLog() const;
bool IsReportingEnabled() const;
System& system;

View File

@@ -230,6 +230,7 @@ void TelemetrySession::AddInitialInfo(Loader::AppLoader& app_loader,
Settings::values.use_asynchronous_gpu_emulation.GetValue());
AddField(field_type, "Renderer_UseNvdecEmulation",
Settings::values.use_nvdec_emulation.GetValue());
AddField(field_type, "Renderer_AccelerateASTC", Settings::values.accelerate_astc.GetValue());
AddField(field_type, "Renderer_UseVsync", Settings::values.use_vsync.GetValue());
AddField(field_type, "Renderer_UseAssemblyShaders",
Settings::values.use_assembly_shaders.GetValue());

View File

@@ -71,8 +71,7 @@ if (ENABLE_SDL2)
target_compile_definitions(input_common PRIVATE HAVE_SDL2)
endif()
target_include_directories(input_common SYSTEM PRIVATE ${LIBUSB_INCLUDE_DIR})
target_link_libraries(input_common PRIVATE ${LIBUSB_LIBRARIES})
target_link_libraries(input_common PRIVATE usb)
create_target_directory_groups(input_common)
target_link_libraries(input_common PUBLIC core PRIVATE common Boost::boost)

View File

@@ -2,6 +2,7 @@ add_executable(tests
common/bit_field.cpp
common/cityhash.cpp
common/fibers.cpp
common/host_memory.cpp
common/param_package.cpp
common/ring_buffer.cpp
core/core_timing.cpp

View File

@@ -0,0 +1,183 @@
// Copyright 2021 yuzu emulator team
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
#include <catch2/catch.hpp>
#include "common/host_memory.h"
using Common::HostMemory;
static constexpr size_t VIRTUAL_SIZE = 1ULL << 39;
static constexpr size_t BACKING_SIZE = 4ULL * 1024 * 1024 * 1024;
TEST_CASE("HostMemory: Initialize and deinitialize", "[common]") {
{ HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE); }
{ HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE); }
}
TEST_CASE("HostMemory: Simple map", "[common]") {
HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
mem.Map(0x5000, 0x8000, 0x1000);
volatile u8* const data = mem.VirtualBasePointer() + 0x5000;
data[0] = 50;
REQUIRE(data[0] == 50);
}
TEST_CASE("HostMemory: Simple mirror map", "[common]") {
HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
mem.Map(0x5000, 0x3000, 0x2000);
mem.Map(0x8000, 0x4000, 0x1000);
volatile u8* const mirror_a = mem.VirtualBasePointer() + 0x5000;
volatile u8* const mirror_b = mem.VirtualBasePointer() + 0x8000;
mirror_b[0] = 76;
REQUIRE(mirror_a[0x1000] == 76);
}
TEST_CASE("HostMemory: Simple unmap", "[common]") {
HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
mem.Map(0x5000, 0x3000, 0x2000);
volatile u8* const data = mem.VirtualBasePointer() + 0x5000;
data[75] = 50;
REQUIRE(data[75] == 50);
mem.Unmap(0x5000, 0x2000);
}
TEST_CASE("HostMemory: Simple unmap and remap", "[common]") {
HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
mem.Map(0x5000, 0x3000, 0x2000);
volatile u8* const data = mem.VirtualBasePointer() + 0x5000;
data[0] = 50;
REQUIRE(data[0] == 50);
mem.Unmap(0x5000, 0x2000);
mem.Map(0x5000, 0x3000, 0x2000);
REQUIRE(data[0] == 50);
mem.Map(0x7000, 0x2000, 0x5000);
REQUIRE(data[0x3000] == 50);
}
TEST_CASE("HostMemory: Nieche allocation", "[common]") {
HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
mem.Map(0x0000, 0, 0x20000);
mem.Unmap(0x0000, 0x4000);
mem.Map(0x1000, 0, 0x2000);
mem.Map(0x3000, 0, 0x1000);
mem.Map(0, 0, 0x1000);
}
TEST_CASE("HostMemory: Full unmap", "[common]") {
HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
mem.Map(0x8000, 0, 0x4000);
mem.Unmap(0x8000, 0x4000);
mem.Map(0x6000, 0, 0x16000);
}
TEST_CASE("HostMemory: Right out of bounds unmap", "[common]") {
HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
mem.Map(0x0000, 0, 0x4000);
mem.Unmap(0x2000, 0x4000);
mem.Map(0x2000, 0x80000, 0x4000);
}
TEST_CASE("HostMemory: Left out of bounds unmap", "[common]") {
HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
mem.Map(0x8000, 0, 0x4000);
mem.Unmap(0x6000, 0x4000);
mem.Map(0x8000, 0, 0x2000);
}
TEST_CASE("HostMemory: Multiple placeholder unmap", "[common]") {
HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
mem.Map(0x0000, 0, 0x4000);
mem.Map(0x4000, 0, 0x1b000);
mem.Unmap(0x3000, 0x1c000);
mem.Map(0x3000, 0, 0x20000);
}
TEST_CASE("HostMemory: Unmap between placeholders", "[common]") {
HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
mem.Map(0x0000, 0, 0x4000);
mem.Map(0x4000, 0, 0x4000);
mem.Unmap(0x2000, 0x4000);
mem.Map(0x2000, 0, 0x4000);
}
TEST_CASE("HostMemory: Unmap to origin", "[common]") {
HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
mem.Map(0x4000, 0, 0x4000);
mem.Map(0x8000, 0, 0x4000);
mem.Unmap(0x4000, 0x4000);
mem.Map(0, 0, 0x4000);
mem.Map(0x4000, 0, 0x4000);
}
TEST_CASE("HostMemory: Unmap to right", "[common]") {
HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
mem.Map(0x4000, 0, 0x4000);
mem.Map(0x8000, 0, 0x4000);
mem.Unmap(0x8000, 0x4000);
mem.Map(0x8000, 0, 0x4000);
}
TEST_CASE("HostMemory: Partial right unmap check bindings", "[common]") {
HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
mem.Map(0x4000, 0x10000, 0x4000);
volatile u8* const ptr = mem.VirtualBasePointer() + 0x4000;
ptr[0x1000] = 17;
mem.Unmap(0x6000, 0x2000);
REQUIRE(ptr[0x1000] == 17);
}
TEST_CASE("HostMemory: Partial left unmap check bindings", "[common]") {
HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
mem.Map(0x4000, 0x10000, 0x4000);
volatile u8* const ptr = mem.VirtualBasePointer() + 0x4000;
ptr[0x3000] = 19;
ptr[0x3fff] = 12;
mem.Unmap(0x4000, 0x2000);
REQUIRE(ptr[0x3000] == 19);
REQUIRE(ptr[0x3fff] == 12);
}
TEST_CASE("HostMemory: Partial middle unmap check bindings", "[common]") {
HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
mem.Map(0x4000, 0x10000, 0x4000);
volatile u8* const ptr = mem.VirtualBasePointer() + 0x4000;
ptr[0x0000] = 19;
ptr[0x3fff] = 12;
mem.Unmap(0x1000, 0x2000);
REQUIRE(ptr[0x0000] == 19);
REQUIRE(ptr[0x3fff] == 12);
}
TEST_CASE("HostMemory: Partial sparse middle unmap and check bindings", "[common]") {
HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
mem.Map(0x4000, 0x10000, 0x2000);
mem.Map(0x6000, 0x20000, 0x2000);
volatile u8* const ptr = mem.VirtualBasePointer() + 0x4000;
ptr[0x0000] = 19;
ptr[0x3fff] = 12;
mem.Unmap(0x5000, 0x2000);
REQUIRE(ptr[0x0000] == 19);
REQUIRE(ptr[0x3fff] == 12);
}

View File

@@ -237,6 +237,7 @@ add_library(video_core STATIC
texture_cache/util.cpp
texture_cache/util.h
textures/astc.h
textures/astc.cpp
textures/decoders.cpp
textures/decoders.h
textures/texture.cpp

View File

@@ -256,6 +256,16 @@ public:
stream_score += score;
}
/// Sets the new frame tick
void SetFrameTick(u64 new_frame_tick) noexcept {
frame_tick = new_frame_tick;
}
/// Returns the new frame tick
[[nodiscard]] u64 FrameTick() const noexcept {
return frame_tick;
}
/// Returns the likeliness of this being a stream buffer
[[nodiscard]] int StreamScore() const noexcept {
return stream_score;
@@ -476,6 +486,9 @@ private:
current_size = 0;
on_going = false;
}
if (empty_bits == PAGES_PER_WORD) {
break;
}
page += empty_bits;
const int continuous_bits = std::countr_one(word >> page);
@@ -583,6 +596,7 @@ private:
RasterizerInterface* rasterizer = nullptr;
VAddr cpu_addr = 0;
Words words;
u64 frame_tick = 0;
BufferFlagBits flags{};
int stream_score = 0;
};

View File

@@ -16,6 +16,7 @@
#include <boost/container/small_vector.hpp>
#include "common/common_sizes.h"
#include "common/common_types.h"
#include "common/div_ceil.h"
#include "common/microprofile.h"
@@ -65,6 +66,9 @@ class BufferCache {
static constexpr BufferId NULL_BUFFER_ID{0};
static constexpr u64 EXPECTED_MEMORY = Common::Size_512_MB;
static constexpr u64 CRITICAL_MEMORY = Common::Size_1_GB;
using Maxwell = Tegra::Engines::Maxwell3D::Regs;
using Runtime = typename P::Runtime;
@@ -102,6 +106,8 @@ public:
void TickFrame();
void RunGarbageCollector();
void WriteMemory(VAddr cpu_addr, u64 size);
void CachedWriteMemory(VAddr cpu_addr, u64 size);
@@ -243,6 +249,8 @@ private:
template <bool insert>
void ChangeRegister(BufferId buffer_id);
void TouchBuffer(Buffer& buffer) const noexcept;
bool SynchronizeBuffer(Buffer& buffer, VAddr cpu_addr, u32 size);
bool SynchronizeBufferImpl(Buffer& buffer, VAddr cpu_addr, u32 size);
@@ -255,6 +263,10 @@ private:
void MappedUploadMemory(Buffer& buffer, u64 total_size_bytes, std::span<BufferCopy> copies);
void DownloadBufferMemory(Buffer& buffer_id);
void DownloadBufferMemory(Buffer& buffer_id, VAddr cpu_addr, u64 size);
void DeleteBuffer(BufferId buffer_id);
void ReplaceBufferDownloads(BufferId old_buffer_id, BufferId new_buffer_id);
@@ -319,6 +331,10 @@ private:
size_t immediate_buffer_capacity = 0;
std::unique_ptr<u8[]> immediate_buffer_alloc;
typename SlotVector<Buffer>::Iterator deletion_iterator;
u64 frame_tick = 0;
u64 total_used_memory = 0;
std::array<BufferId, ((1ULL << 39) >> PAGE_BITS)> page_table;
};
@@ -332,6 +348,28 @@ BufferCache<P>::BufferCache(VideoCore::RasterizerInterface& rasterizer_,
gpu_memory{gpu_memory_}, cpu_memory{cpu_memory_}, runtime{runtime_} {
// Ensure the first slot is used for the null buffer
void(slot_buffers.insert(runtime, NullBufferParams{}));
deletion_iterator = slot_buffers.end();
}
template <class P>
void BufferCache<P>::RunGarbageCollector() {
const bool aggressive_gc = total_used_memory >= CRITICAL_MEMORY;
const u64 ticks_to_destroy = aggressive_gc ? 60 : 120;
int num_iterations = aggressive_gc ? 64 : 32;
for (; num_iterations > 0; --num_iterations) {
if (deletion_iterator == slot_buffers.end()) {
deletion_iterator = slot_buffers.begin();
}
++deletion_iterator;
if (deletion_iterator == slot_buffers.end()) {
break;
}
const auto [buffer_id, buffer] = *deletion_iterator;
if (buffer->FrameTick() + ticks_to_destroy < frame_tick) {
DownloadBufferMemory(*buffer);
DeleteBuffer(buffer_id);
}
}
}
template <class P>
@@ -349,6 +387,10 @@ void BufferCache<P>::TickFrame() {
const bool skip_preferred = hits * 256 < shots * 251;
uniform_buffer_skip_cache_size = skip_preferred ? DEFAULT_SKIP_CACHE_SIZE : 0;
if (Settings::values.use_caches_gc.GetValue() && total_used_memory >= EXPECTED_MEMORY) {
RunGarbageCollector();
}
++frame_tick;
delayed_destruction_ring.Tick();
}
@@ -371,50 +413,8 @@ void BufferCache<P>::CachedWriteMemory(VAddr cpu_addr, u64 size) {
template <class P>
void BufferCache<P>::DownloadMemory(VAddr cpu_addr, u64 size) {
ForEachBufferInRange(cpu_addr, size, [&](BufferId, Buffer& buffer) {
boost::container::small_vector<BufferCopy, 1> copies;
u64 total_size_bytes = 0;
u64 largest_copy = 0;
buffer.ForEachDownloadRange(cpu_addr, size, [&](u64 range_offset, u64 range_size) {
copies.push_back(BufferCopy{
.src_offset = range_offset,
.dst_offset = total_size_bytes,
.size = range_size,
});
total_size_bytes += range_size;
largest_copy = std::max(largest_copy, range_size);
});
if (total_size_bytes == 0) {
return;
}
MICROPROFILE_SCOPE(GPU_DownloadMemory);
if constexpr (USE_MEMORY_MAPS) {
auto download_staging = runtime.DownloadStagingBuffer(total_size_bytes);
const u8* const mapped_memory = download_staging.mapped_span.data();
const std::span<BufferCopy> copies_span(copies.data(), copies.data() + copies.size());
for (BufferCopy& copy : copies) {
// Modify copies to have the staging offset in mind
copy.dst_offset += download_staging.offset;
}
runtime.CopyBuffer(download_staging.buffer, buffer, copies_span);
runtime.Finish();
for (const BufferCopy& copy : copies) {
const VAddr copy_cpu_addr = buffer.CpuAddr() + copy.src_offset;
// Undo the modified offset
const u64 dst_offset = copy.dst_offset - download_staging.offset;
const u8* copy_mapped_memory = mapped_memory + dst_offset;
cpu_memory.WriteBlockUnsafe(copy_cpu_addr, copy_mapped_memory, copy.size);
}
} else {
const std::span<u8> immediate_buffer = ImmediateBuffer(largest_copy);
for (const BufferCopy& copy : copies) {
buffer.ImmediateDownload(copy.src_offset, immediate_buffer.subspan(0, copy.size));
const VAddr copy_cpu_addr = buffer.CpuAddr() + copy.src_offset;
cpu_memory.WriteBlockUnsafe(copy_cpu_addr, immediate_buffer.data(), copy.size);
}
}
});
ForEachBufferInRange(cpu_addr, size,
[&](BufferId, Buffer& buffer) { DownloadBufferMemory(buffer); });
}
template <class P>
@@ -640,6 +640,7 @@ bool BufferCache<P>::IsRegionGpuModified(VAddr addr, size_t size) {
template <class P>
void BufferCache<P>::BindHostIndexBuffer() {
Buffer& buffer = slot_buffers[index_buffer.buffer_id];
TouchBuffer(buffer);
const u32 offset = buffer.Offset(index_buffer.cpu_addr);
const u32 size = index_buffer.size;
SynchronizeBuffer(buffer, index_buffer.cpu_addr, size);
@@ -658,6 +659,7 @@ void BufferCache<P>::BindHostVertexBuffers() {
for (u32 index = 0; index < NUM_VERTEX_BUFFERS; ++index) {
const Binding& binding = vertex_buffers[index];
Buffer& buffer = slot_buffers[binding.buffer_id];
TouchBuffer(buffer);
SynchronizeBuffer(buffer, binding.cpu_addr, binding.size);
if (!flags[Dirty::VertexBuffer0 + index]) {
continue;
@@ -693,6 +695,7 @@ void BufferCache<P>::BindHostGraphicsUniformBuffer(size_t stage, u32 index, u32
const VAddr cpu_addr = binding.cpu_addr;
const u32 size = binding.size;
Buffer& buffer = slot_buffers[binding.buffer_id];
TouchBuffer(buffer);
const bool use_fast_buffer = binding.buffer_id != NULL_BUFFER_ID &&
size <= uniform_buffer_skip_cache_size &&
!buffer.IsRegionGpuModified(cpu_addr, size);
@@ -744,6 +747,7 @@ void BufferCache<P>::BindHostGraphicsStorageBuffers(size_t stage) {
ForEachEnabledBit(enabled_storage_buffers[stage], [&](u32 index) {
const Binding& binding = storage_buffers[stage][index];
Buffer& buffer = slot_buffers[binding.buffer_id];
TouchBuffer(buffer);
const u32 size = binding.size;
SynchronizeBuffer(buffer, binding.cpu_addr, size);
@@ -766,6 +770,7 @@ void BufferCache<P>::BindHostTransformFeedbackBuffers() {
for (u32 index = 0; index < NUM_TRANSFORM_FEEDBACK_BUFFERS; ++index) {
const Binding& binding = transform_feedback_buffers[index];
Buffer& buffer = slot_buffers[binding.buffer_id];
TouchBuffer(buffer);
const u32 size = binding.size;
SynchronizeBuffer(buffer, binding.cpu_addr, size);
@@ -784,6 +789,7 @@ void BufferCache<P>::BindHostComputeUniformBuffers() {
ForEachEnabledBit(enabled_compute_uniform_buffers, [&](u32 index) {
const Binding& binding = compute_uniform_buffers[index];
Buffer& buffer = slot_buffers[binding.buffer_id];
TouchBuffer(buffer);
const u32 size = binding.size;
SynchronizeBuffer(buffer, binding.cpu_addr, size);
@@ -803,6 +809,7 @@ void BufferCache<P>::BindHostComputeStorageBuffers() {
ForEachEnabledBit(enabled_compute_storage_buffers, [&](u32 index) {
const Binding& binding = compute_storage_buffers[index];
Buffer& buffer = slot_buffers[binding.buffer_id];
TouchBuffer(buffer);
const u32 size = binding.size;
SynchronizeBuffer(buffer, binding.cpu_addr, size);
@@ -1101,6 +1108,7 @@ BufferId BufferCache<P>::CreateBuffer(VAddr cpu_addr, u32 wanted_size) {
const OverlapResult overlap = ResolveOverlaps(cpu_addr, wanted_size);
const u32 size = static_cast<u32>(overlap.end - overlap.begin);
const BufferId new_buffer_id = slot_buffers.insert(runtime, rasterizer, overlap.begin, size);
TouchBuffer(slot_buffers[new_buffer_id]);
for (const BufferId overlap_id : overlap.ids) {
JoinOverlap(new_buffer_id, overlap_id, !overlap.has_stream_leap);
}
@@ -1122,8 +1130,14 @@ template <class P>
template <bool insert>
void BufferCache<P>::ChangeRegister(BufferId buffer_id) {
const Buffer& buffer = slot_buffers[buffer_id];
const auto size = buffer.SizeBytes();
if (insert) {
total_used_memory += Common::AlignUp(size, 1024);
} else {
total_used_memory -= Common::AlignUp(size, 1024);
}
const VAddr cpu_addr_begin = buffer.CpuAddr();
const VAddr cpu_addr_end = cpu_addr_begin + buffer.SizeBytes();
const VAddr cpu_addr_end = cpu_addr_begin + size;
const u64 page_begin = cpu_addr_begin / PAGE_SIZE;
const u64 page_end = Common::DivCeil(cpu_addr_end, PAGE_SIZE);
for (u64 page = page_begin; page != page_end; ++page) {
@@ -1135,6 +1149,11 @@ void BufferCache<P>::ChangeRegister(BufferId buffer_id) {
}
}
template <class P>
void BufferCache<P>::TouchBuffer(Buffer& buffer) const noexcept {
buffer.SetFrameTick(frame_tick);
}
template <class P>
bool BufferCache<P>::SynchronizeBuffer(Buffer& buffer, VAddr cpu_addr, u32 size) {
if (buffer.CpuAddr() == 0) {
@@ -1211,6 +1230,57 @@ void BufferCache<P>::MappedUploadMemory(Buffer& buffer, u64 total_size_bytes,
runtime.CopyBuffer(buffer, upload_staging.buffer, copies);
}
template <class P>
void BufferCache<P>::DownloadBufferMemory(Buffer& buffer) {
DownloadBufferMemory(buffer, buffer.CpuAddr(), buffer.SizeBytes());
}
template <class P>
void BufferCache<P>::DownloadBufferMemory(Buffer& buffer, VAddr cpu_addr, u64 size) {
boost::container::small_vector<BufferCopy, 1> copies;
u64 total_size_bytes = 0;
u64 largest_copy = 0;
buffer.ForEachDownloadRange(cpu_addr, size, [&](u64 range_offset, u64 range_size) {
copies.push_back(BufferCopy{
.src_offset = range_offset,
.dst_offset = total_size_bytes,
.size = range_size,
});
total_size_bytes += range_size;
largest_copy = std::max(largest_copy, range_size);
});
if (total_size_bytes == 0) {
return;
}
MICROPROFILE_SCOPE(GPU_DownloadMemory);
if constexpr (USE_MEMORY_MAPS) {
auto download_staging = runtime.DownloadStagingBuffer(total_size_bytes);
const u8* const mapped_memory = download_staging.mapped_span.data();
const std::span<BufferCopy> copies_span(copies.data(), copies.data() + copies.size());
for (BufferCopy& copy : copies) {
// Modify copies to have the staging offset in mind
copy.dst_offset += download_staging.offset;
}
runtime.CopyBuffer(download_staging.buffer, buffer, copies_span);
runtime.Finish();
for (const BufferCopy& copy : copies) {
const VAddr copy_cpu_addr = buffer.CpuAddr() + copy.src_offset;
// Undo the modified offset
const u64 dst_offset = copy.dst_offset - download_staging.offset;
const u8* copy_mapped_memory = mapped_memory + dst_offset;
cpu_memory.WriteBlockUnsafe(copy_cpu_addr, copy_mapped_memory, copy.size);
}
} else {
const std::span<u8> immediate_buffer = ImmediateBuffer(largest_copy);
for (const BufferCopy& copy : copies) {
buffer.ImmediateDownload(copy.src_offset, immediate_buffer.subspan(0, copy.size));
const VAddr copy_cpu_addr = buffer.CpuAddr() + copy.src_offset;
cpu_memory.WriteBlockUnsafe(copy_cpu_addr, immediate_buffer.data(), copy.size);
}
}
}
template <class P>
void BufferCache<P>::DeleteBuffer(BufferId buffer_id) {
const auto scalar_replace = [buffer_id](Binding& binding) {
@@ -1236,6 +1306,7 @@ void BufferCache<P>::DeleteBuffer(BufferId buffer_id) {
Unregister(buffer_id);
delayed_destruction_ring.Push(std::move(slot_buffers[buffer_id]));
slot_buffers.erase(buffer_id);
NotifyBufferDeletion();
}

View File

@@ -99,25 +99,13 @@ void ThreadManager::FlushRegion(VAddr addr, u64 size) {
PushCommand(FlushRegionCommand(addr, size));
return;
}
// Asynchronous GPU mode
switch (Settings::values.gpu_accuracy.GetValue()) {
case Settings::GPUAccuracy::Normal:
PushCommand(FlushRegionCommand(addr, size));
break;
case Settings::GPUAccuracy::High:
// TODO(bunnei): Is this right? Preserving existing behavior for now
break;
case Settings::GPUAccuracy::Extreme: {
auto& gpu = system.GPU();
u64 fence = gpu.RequestFlush(addr, size);
PushCommand(GPUTickCommand(), true);
ASSERT(fence <= gpu.CurrentFlushRequestFence());
break;
}
default:
UNIMPLEMENTED_MSG("Unsupported gpu_accuracy {}", Settings::values.gpu_accuracy.GetValue());
if (!Settings::IsGPULevelExtreme()) {
return;
}
auto& gpu = system.GPU();
u64 fence = gpu.RequestFlush(addr, size);
PushCommand(GPUTickCommand(), true);
ASSERT(fence <= gpu.CurrentFlushRequestFence());
}
void ThreadManager::InvalidateRegion(VAddr addr, u64 size) {

View File

@@ -763,7 +763,7 @@ void ComputeEndpoints(out uvec4 ep1, out uvec4 ep2, uint color_endpoint_mode) {
case 1: {
READ_UINT_VALUES(2)
uint L0 = (v[0] >> 2) | (v[1] & 0xC0);
uint L1 = max(L0 + (v[1] & 0x3F), 0xFFU);
uint L1 = min(L0 + (v[1] & 0x3F), 0xFFU);
ep1 = uvec4(0xFF, L0, L0, L0);
ep2 = uvec4(0xFF, L1, L1, L1);
break;

View File

@@ -2,6 +2,8 @@
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
#include <atomic>
#include "common/assert.h"
#include "common/common_types.h"
#include "common/div_ceil.h"
@@ -10,35 +12,59 @@
namespace VideoCore {
RasterizerAccelerated::RasterizerAccelerated(Core::Memory::Memory& cpu_memory_)
: cpu_memory{cpu_memory_} {}
using namespace Core::Memory;
RasterizerAccelerated::RasterizerAccelerated(Memory& cpu_memory_) : cpu_memory{cpu_memory_} {}
RasterizerAccelerated::~RasterizerAccelerated() = default;
void RasterizerAccelerated::UpdatePagesCachedCount(VAddr addr, u64 size, int delta) {
const auto page_end = Common::DivCeil(addr + size, Core::Memory::PAGE_SIZE);
for (auto page = addr >> Core::Memory::PAGE_BITS; page != page_end; ++page) {
auto& count = cached_pages.at(page >> 2).Count(page);
u64 uncache_begin = 0;
u64 cache_begin = 0;
u64 uncache_bytes = 0;
u64 cache_bytes = 0;
std::atomic_thread_fence(std::memory_order_acquire);
const u64 page_end = Common::DivCeil(addr + size, PAGE_SIZE);
for (u64 page = addr >> PAGE_BITS; page != page_end; ++page) {
std::atomic_uint16_t& count = cached_pages.at(page >> 2).Count(page);
if (delta > 0) {
ASSERT_MSG(count < UINT16_MAX, "Count may overflow!");
ASSERT_MSG(count.load(std::memory_order::relaxed) < UINT16_MAX, "Count may overflow!");
} else if (delta < 0) {
ASSERT_MSG(count > 0, "Count may underflow!");
ASSERT_MSG(count.load(std::memory_order::relaxed) > 0, "Count may underflow!");
} else {
ASSERT_MSG(true, "Delta must be non-zero!");
ASSERT_MSG(false, "Delta must be non-zero!");
}
// Adds or subtracts 1, as count is a unsigned 8-bit value
count += static_cast<u16>(delta);
count.fetch_add(static_cast<u16>(delta), std::memory_order_release);
// Assume delta is either -1 or 1
if (count == 0) {
cpu_memory.RasterizerMarkRegionCached(page << Core::Memory::PAGE_BITS,
Core::Memory::PAGE_SIZE, false);
} else if (count == 1 && delta > 0) {
cpu_memory.RasterizerMarkRegionCached(page << Core::Memory::PAGE_BITS,
Core::Memory::PAGE_SIZE, true);
if (count.load(std::memory_order::relaxed) == 0) {
if (uncache_bytes == 0) {
uncache_begin = page;
}
uncache_bytes += PAGE_SIZE;
} else if (uncache_bytes > 0) {
cpu_memory.RasterizerMarkRegionCached(uncache_begin << PAGE_BITS, uncache_bytes, false);
uncache_bytes = 0;
}
if (count.load(std::memory_order::relaxed) == 1 && delta > 0) {
if (cache_bytes == 0) {
cache_begin = page;
}
cache_bytes += PAGE_SIZE;
} else if (cache_bytes > 0) {
cpu_memory.RasterizerMarkRegionCached(cache_begin << PAGE_BITS, cache_bytes, true);
cache_bytes = 0;
}
}
if (uncache_bytes > 0) {
cpu_memory.RasterizerMarkRegionCached(uncache_begin << PAGE_BITS, uncache_bytes, false);
}
if (cache_bytes > 0) {
cpu_memory.RasterizerMarkRegionCached(cache_begin << PAGE_BITS, cache_bytes, true);
}
}

View File

@@ -9,6 +9,8 @@
#include <glad/glad.h>
#include "common/settings.h"
#include "video_core/renderer_opengl/gl_device.h"
#include "video_core/renderer_opengl/gl_shader_manager.h"
#include "video_core/renderer_opengl/gl_state_tracker.h"
@@ -307,7 +309,9 @@ void ApplySwizzle(GLuint handle, PixelFormat format, std::array<SwizzleSource, 4
[[nodiscard]] bool CanBeAccelerated(const TextureCacheRuntime& runtime,
const VideoCommon::ImageInfo& info) {
return !runtime.HasNativeASTC() && IsPixelFormatASTC(info.format);
if (IsPixelFormatASTC(info.format)) {
return !runtime.HasNativeASTC() && Settings::values.accelerate_astc.GetValue();
}
// Disable other accelerated uploads for now as they don't implement swizzled uploads
return false;
switch (info.type) {
@@ -733,6 +737,8 @@ Image::Image(TextureCacheRuntime& runtime, const VideoCommon::ImageInfo& info_,
}
}
Image::~Image() = default;
void Image::UploadMemory(const ImageBufferMap& map,
std::span<const VideoCommon::BufferImageCopy> copies) {
glBindBuffer(GL_PIXEL_UNPACK_BUFFER, map.buffer);

View File

@@ -143,6 +143,14 @@ public:
explicit Image(TextureCacheRuntime&, const VideoCommon::ImageInfo& info, GPUVAddr gpu_addr,
VAddr cpu_addr);
~Image();
Image(const Image&) = delete;
Image& operator=(const Image&) = delete;
Image(Image&&) = default;
Image& operator=(Image&&) = default;
void UploadMemory(const ImageBufferMap& map,
std::span<const VideoCommon::BufferImageCopy> copies);
@@ -235,6 +243,7 @@ struct TextureCacheParams {
static constexpr bool ENABLE_VALIDATION = true;
static constexpr bool FRAMEBUFFER_BLITS = true;
static constexpr bool HAS_EMULATED_COPIES = true;
static constexpr bool HAS_DEVICE_MEMORY_INFO = false;
using Runtime = OpenGL::TextureCacheRuntime;
using Image = OpenGL::Image;

View File

@@ -8,6 +8,7 @@
#include <vector>
#include "common/bit_cast.h"
#include "common/settings.h"
#include "video_core/engines/fermi_2d.h"
#include "video_core/renderer_vulkan/blit_image.h"
@@ -817,6 +818,10 @@ void TextureCacheRuntime::CopyImage(Image& dst, Image& src,
});
}
u64 TextureCacheRuntime::GetDeviceLocalMemory() const {
return device.GetDeviceLocalMemory();
}
Image::Image(TextureCacheRuntime& runtime, const ImageInfo& info_, GPUVAddr gpu_addr_,
VAddr cpu_addr_)
: VideoCommon::ImageBase(info_, gpu_addr_, cpu_addr_), scheduler{&runtime.scheduler},
@@ -828,7 +833,11 @@ Image::Image(TextureCacheRuntime& runtime, const ImageInfo& info_, GPUVAddr gpu_
commit = runtime.memory_allocator.Commit(buffer, MemoryUsage::DeviceLocal);
}
if (IsPixelFormatASTC(info.format) && !runtime.device.IsOptimalAstcSupported()) {
flags |= VideoCommon::ImageFlagBits::AcceleratedUpload;
if (Settings::values.accelerate_astc.GetValue()) {
flags |= VideoCommon::ImageFlagBits::AcceleratedUpload;
} else {
flags |= VideoCommon::ImageFlagBits::Converted;
}
}
if (runtime.device.HasDebuggingToolAttached()) {
if (image) {
@@ -871,6 +880,8 @@ Image::Image(TextureCacheRuntime& runtime, const ImageInfo& info_, GPUVAddr gpu_
}
}
Image::~Image() = default;
void Image::UploadMemory(const StagingBufferRef& map, std::span<const BufferImageCopy> copies) {
// TODO: Move this to another API
scheduler->RequestOutsideRenderPassOperationContext();

View File

@@ -97,6 +97,8 @@ struct TextureCacheRuntime {
// All known Vulkan drivers can natively handle BGR textures
return true;
}
u64 GetDeviceLocalMemory() const;
};
class Image : public VideoCommon::ImageBase {
@@ -104,6 +106,14 @@ public:
explicit Image(TextureCacheRuntime&, const VideoCommon::ImageInfo& info, GPUVAddr gpu_addr,
VAddr cpu_addr);
~Image();
Image(const Image&) = delete;
Image& operator=(const Image&) = delete;
Image(Image&&) = default;
Image& operator=(Image&&) = default;
void UploadMemory(const StagingBufferRef& map,
std::span<const VideoCommon::BufferImageCopy> copies);
@@ -257,6 +267,7 @@ struct TextureCacheParams {
static constexpr bool ENABLE_VALIDATION = true;
static constexpr bool FRAMEBUFFER_BLITS = false;
static constexpr bool HAS_EMULATED_COPIES = false;
static constexpr bool HAS_DEVICE_MEMORY_INFO = true;
using Runtime = Vulkan::TextureCacheRuntime;
using Image = Vulkan::Image;

View File

@@ -283,4 +283,11 @@ std::pair<u32, u32> GetASTCBlockSize(PixelFormat format) {
return {DefaultBlockWidth(format), DefaultBlockHeight(format)};
}
u64 EstimatedDecompressedSize(u64 base_size, PixelFormat format) {
constexpr u64 RGBA8_PIXEL_SIZE = 4;
const u64 base_block_size = static_cast<u64>(DefaultBlockWidth(format)) *
static_cast<u64>(DefaultBlockHeight(format)) * RGBA8_PIXEL_SIZE;
return (base_size * base_block_size) / BytesPerBlock(format);
}
} // namespace VideoCore::Surface

View File

@@ -462,4 +462,6 @@ bool IsPixelFormatSRGB(PixelFormat format);
std::pair<u32, u32> GetASTCBlockSize(PixelFormat format);
u64 EstimatedDecompressedSize(u64 base_size, PixelFormat format);
} // namespace VideoCore::Surface

View File

@@ -113,6 +113,43 @@ void ImageBase::InsertView(const ImageViewInfo& view_info, ImageViewId image_vie
image_view_ids.push_back(image_view_id);
}
bool ImageBase::IsSafeDownload() const noexcept {
// Skip images that were not modified from the GPU
if (False(flags & ImageFlagBits::GpuModified)) {
return false;
}
// Skip images that .are. modified from the CPU
// We don't want to write sensitive data from the guest
if (True(flags & ImageFlagBits::CpuModified)) {
return false;
}
if (info.num_samples > 1) {
LOG_WARNING(HW_GPU, "MSAA image downloads are not implemented");
return false;
}
return true;
}
void ImageBase::CheckBadOverlapState() {
if (False(flags & ImageFlagBits::BadOverlap)) {
return;
}
if (!overlapping_images.empty()) {
return;
}
flags &= ~ImageFlagBits::BadOverlap;
}
void ImageBase::CheckAliasState() {
if (False(flags & ImageFlagBits::Alias)) {
return;
}
if (!aliased_images.empty()) {
return;
}
flags &= ~ImageFlagBits::Alias;
}
void AddImageAlias(ImageBase& lhs, ImageBase& rhs, ImageId lhs_id, ImageId rhs_id) {
static constexpr auto OPTIONS = RelaxedOptions::Size | RelaxedOptions::Format;
ASSERT(lhs.info.type == rhs.info.type);

View File

@@ -25,6 +25,12 @@ enum class ImageFlagBits : u32 {
Strong = 1 << 5, ///< Exists in the image table, the dimensions are can be trusted
Registered = 1 << 6, ///< True when the image is registered
Picked = 1 << 7, ///< Temporary flag to mark the image as picked
// Garbage Collection Flags
BadOverlap = 1 << 8, ///< This image overlaps other but doesn't fit, has higher
///< garbage collection priority
Alias = 1 << 9, ///< This image has aliases and has priority on garbage
///< collection
};
DECLARE_ENUM_FLAG_OPERATORS(ImageFlagBits)
@@ -44,11 +50,16 @@ struct ImageBase {
void InsertView(const ImageViewInfo& view_info, ImageViewId image_view_id);
[[nodiscard]] bool IsSafeDownload() const noexcept;
[[nodiscard]] bool Overlaps(VAddr overlap_cpu_addr, size_t overlap_size) const noexcept {
const VAddr overlap_end = overlap_cpu_addr + overlap_size;
return cpu_addr < overlap_end && overlap_cpu_addr < cpu_addr_end;
}
void CheckBadOverlapState();
void CheckAliasState();
ImageInfo info;
u32 guest_size_bytes = 0;
@@ -72,6 +83,7 @@ struct ImageBase {
std::vector<SubresourceBase> slice_subresources;
std::vector<AliasedImage> aliased_images;
std::vector<ImageId> overlapping_images;
};
struct ImageAllocBase {

View File

@@ -5,6 +5,7 @@
#pragma once
#include <array>
#include <bit>
#include <concepts>
#include <numeric>
#include <type_traits>
@@ -32,6 +33,60 @@ template <class T>
requires std::is_nothrow_move_assignable_v<T>&&
std::is_nothrow_move_constructible_v<T> class SlotVector {
public:
class Iterator {
friend SlotVector<T>;
public:
constexpr Iterator() = default;
Iterator& operator++() noexcept {
const u64* const bitset = slot_vector->stored_bitset.data();
const u32 size = static_cast<u32>(slot_vector->stored_bitset.size()) * 64;
if (id.index < size) {
do {
++id.index;
} while (id.index < size && !IsValid(bitset));
if (id.index == size) {
id.index = SlotId::INVALID_INDEX;
}
}
return *this;
}
Iterator operator++(int) noexcept {
const Iterator copy{*this};
++*this;
return copy;
}
bool operator==(const Iterator& other) const noexcept {
return id.index == other.id.index;
}
bool operator!=(const Iterator& other) const noexcept {
return id.index != other.id.index;
}
std::pair<SlotId, T*> operator*() const noexcept {
return {id, std::addressof((*slot_vector)[id])};
}
T* operator->() const noexcept {
return std::addressof((*slot_vector)[id]);
}
private:
Iterator(SlotVector<T>* slot_vector_, SlotId id_) noexcept
: slot_vector{slot_vector_}, id{id_} {}
bool IsValid(const u64* bitset) const noexcept {
return ((bitset[id.index / 64] >> (id.index % 64)) & 1) != 0;
}
SlotVector<T>* slot_vector;
SlotId id;
};
~SlotVector() noexcept {
size_t index = 0;
for (u64 bits : stored_bitset) {
@@ -70,6 +125,20 @@ public:
ResetStorageBit(id.index);
}
[[nodiscard]] Iterator begin() noexcept {
const auto it = std::ranges::find_if(stored_bitset, [](u64 value) { return value != 0; });
if (it == stored_bitset.end()) {
return end();
}
const u32 word_index = static_cast<u32>(std::distance(it, stored_bitset.begin()));
const SlotId first_id{word_index * 64 + static_cast<u32>(std::countr_zero(*it))};
return Iterator(this, first_id);
}
[[nodiscard]] Iterator end() noexcept {
return Iterator(this, SlotId{SlotId::INVALID_INDEX});
}
private:
struct NonTrivialDummy {
NonTrivialDummy() noexcept {}
@@ -140,7 +209,6 @@ private:
Entry* values = nullptr;
size_t values_capacity = 0;
size_t values_size = 0;
std::vector<u64> stored_bitset;
std::vector<u32> free_list;

View File

@@ -20,8 +20,10 @@
#include "common/alignment.h"
#include "common/common_funcs.h"
#include "common/common_sizes.h"
#include "common/common_types.h"
#include "common/logging/log.h"
#include "common/settings.h"
#include "video_core/compatible_formats.h"
#include "video_core/delayed_destruction_ring.h"
#include "video_core/dirty_flags.h"
@@ -69,12 +71,17 @@ class TextureCache {
static constexpr bool FRAMEBUFFER_BLITS = P::FRAMEBUFFER_BLITS;
/// True when some copies have to be emulated
static constexpr bool HAS_EMULATED_COPIES = P::HAS_EMULATED_COPIES;
/// True when the API can provide info about the memory of the device.
static constexpr bool HAS_DEVICE_MEMORY_INFO = P::HAS_DEVICE_MEMORY_INFO;
/// Image view ID for null descriptors
static constexpr ImageViewId NULL_IMAGE_VIEW_ID{0};
/// Sampler ID for bugged sampler ids
static constexpr SamplerId NULL_SAMPLER_ID{0};
static constexpr u64 DEFAULT_EXPECTED_MEMORY = Common::Size_1_GB;
static constexpr u64 DEFAULT_CRITICAL_MEMORY = Common::Size_2_GB;
using Runtime = typename P::Runtime;
using Image = typename P::Image;
using ImageAlloc = typename P::ImageAlloc;
@@ -103,6 +110,9 @@ public:
/// Notify the cache that a new frame has been queued
void TickFrame();
/// Runs the Garbage Collector.
void RunGarbageCollector();
/// Return a constant reference to the given image view id
[[nodiscard]] const ImageView& GetImageView(ImageViewId id) const noexcept;
@@ -333,6 +343,10 @@ private:
std::unordered_map<u64, std::vector<ImageId>, IdentityHash<u64>> page_table;
bool has_deleted_images = false;
u64 total_used_memory = 0;
u64 minimum_memory;
u64 expected_memory;
u64 critical_memory;
SlotVector<Image> slot_images;
SlotVector<ImageView> slot_image_views;
@@ -353,6 +367,7 @@ private:
u64 modification_tick = 0;
u64 frame_tick = 0;
typename SlotVector<Image>::Iterator deletion_iterator;
};
template <class P>
@@ -373,11 +388,94 @@ TextureCache<P>::TextureCache(Runtime& runtime_, VideoCore::RasterizerInterface&
// This way the null resource becomes a compile time constant
void(slot_image_views.insert(runtime, NullImageParams{}));
void(slot_samplers.insert(runtime, sampler_descriptor));
deletion_iterator = slot_images.begin();
if constexpr (HAS_DEVICE_MEMORY_INFO) {
const auto device_memory = runtime.GetDeviceLocalMemory();
const u64 possible_expected_memory = (device_memory * 3) / 10;
const u64 possible_critical_memory = (device_memory * 6) / 10;
expected_memory = std::max(possible_expected_memory, DEFAULT_EXPECTED_MEMORY);
critical_memory = std::max(possible_critical_memory, DEFAULT_CRITICAL_MEMORY);
minimum_memory = 0;
} else {
// on OGL we can be more conservatives as the driver takes care.
expected_memory = DEFAULT_EXPECTED_MEMORY + Common::Size_512_MB;
critical_memory = DEFAULT_CRITICAL_MEMORY + Common::Size_1_GB;
minimum_memory = expected_memory;
}
}
template <class P>
void TextureCache<P>::RunGarbageCollector() {
const bool high_priority_mode = total_used_memory >= expected_memory;
const bool aggressive_mode = total_used_memory >= critical_memory;
const u64 ticks_to_destroy = high_priority_mode ? 60 : 100;
int num_iterations = aggressive_mode ? 256 : (high_priority_mode ? 128 : 64);
for (; num_iterations > 0; --num_iterations) {
if (deletion_iterator == slot_images.end()) {
deletion_iterator = slot_images.begin();
if (deletion_iterator == slot_images.end()) {
break;
}
}
auto [image_id, image_tmp] = *deletion_iterator;
Image* image = image_tmp; // fix clang error.
const bool is_alias = True(image->flags & ImageFlagBits::Alias);
const bool is_bad_overlap = True(image->flags & ImageFlagBits::BadOverlap);
const bool must_download = image->IsSafeDownload();
bool should_care = is_bad_overlap || is_alias || (high_priority_mode && !must_download);
const u64 ticks_needed =
is_bad_overlap
? ticks_to_destroy >> 4
: ((should_care && aggressive_mode) ? ticks_to_destroy >> 1 : ticks_to_destroy);
should_care |= aggressive_mode;
if (should_care && image->frame_tick + ticks_needed < frame_tick) {
if (is_bad_overlap) {
const bool overlap_check = std::ranges::all_of(
image->overlapping_images, [&, image](const ImageId& overlap_id) {
auto& overlap = slot_images[overlap_id];
return overlap.frame_tick >= image->frame_tick;
});
if (!overlap_check) {
++deletion_iterator;
continue;
}
}
if (!is_bad_overlap && must_download) {
const bool alias_check = std::ranges::none_of(
image->aliased_images, [&, image](const AliasedImage& alias) {
auto& alias_image = slot_images[alias.id];
return (alias_image.frame_tick < image->frame_tick) ||
(alias_image.modification_tick < image->modification_tick);
});
if (alias_check) {
auto map = runtime.DownloadStagingBuffer(image->unswizzled_size_bytes);
const auto copies = FullDownloadCopies(image->info);
image->DownloadMemory(map, copies);
runtime.Finish();
SwizzleImage(gpu_memory, image->gpu_addr, image->info, copies, map.mapped_span);
}
}
if (True(image->flags & ImageFlagBits::Tracked)) {
UntrackImage(*image);
}
UnregisterImage(image_id);
DeleteImage(image_id);
if (is_bad_overlap) {
++num_iterations;
}
}
++deletion_iterator;
}
}
template <class P>
void TextureCache<P>::TickFrame() {
// Tick sentenced resources in this order to ensure they are destroyed in the right order
if (Settings::values.use_caches_gc.GetValue() && total_used_memory > minimum_memory) {
RunGarbageCollector();
}
sentenced_images.Tick();
sentenced_framebuffers.Tick();
sentenced_image_view.Tick();
@@ -568,17 +666,7 @@ template <class P>
void TextureCache<P>::DownloadMemory(VAddr cpu_addr, size_t size) {
std::vector<ImageId> images;
ForEachImageInRegion(cpu_addr, size, [this, &images](ImageId image_id, ImageBase& image) {
// Skip images that were not modified from the GPU
if (False(image.flags & ImageFlagBits::GpuModified)) {
return;
}
// Skip images that .are. modified from the CPU
// We don't want to write sensitive data from the guest
if (True(image.flags & ImageFlagBits::CpuModified)) {
return;
}
if (image.info.num_samples > 1) {
LOG_WARNING(HW_GPU, "MSAA image downloads are not implemented");
if (!image.IsSafeDownload()) {
return;
}
image.flags &= ~ImageFlagBits::GpuModified;
@@ -967,6 +1055,7 @@ ImageId TextureCache<P>::JoinImages(const ImageInfo& info, GPUVAddr gpu_addr, VA
std::vector<ImageId> overlap_ids;
std::vector<ImageId> left_aliased_ids;
std::vector<ImageId> right_aliased_ids;
std::vector<ImageId> bad_overlap_ids;
ForEachImageInRegion(cpu_addr, size_bytes, [&](ImageId overlap_id, ImageBase& overlap) {
if (info.type != overlap.info.type) {
return;
@@ -992,9 +1081,14 @@ ImageId TextureCache<P>::JoinImages(const ImageInfo& info, GPUVAddr gpu_addr, VA
const ImageBase new_image_base(new_info, gpu_addr, cpu_addr);
if (IsSubresource(new_info, overlap, gpu_addr, options, broken_views, native_bgr)) {
left_aliased_ids.push_back(overlap_id);
overlap.flags |= ImageFlagBits::Alias;
} else if (IsSubresource(overlap.info, new_image_base, overlap.gpu_addr, options,
broken_views, native_bgr)) {
right_aliased_ids.push_back(overlap_id);
overlap.flags |= ImageFlagBits::Alias;
} else {
bad_overlap_ids.push_back(overlap_id);
overlap.flags |= ImageFlagBits::BadOverlap;
}
});
const ImageId new_image_id = slot_images.insert(runtime, new_info, gpu_addr, cpu_addr);
@@ -1022,10 +1116,18 @@ ImageId TextureCache<P>::JoinImages(const ImageInfo& info, GPUVAddr gpu_addr, VA
for (const ImageId aliased_id : right_aliased_ids) {
ImageBase& aliased = slot_images[aliased_id];
AddImageAlias(new_image_base, aliased, new_image_id, aliased_id);
new_image.flags |= ImageFlagBits::Alias;
}
for (const ImageId aliased_id : left_aliased_ids) {
ImageBase& aliased = slot_images[aliased_id];
AddImageAlias(aliased, new_image_base, aliased_id, new_image_id);
new_image.flags |= ImageFlagBits::Alias;
}
for (const ImageId aliased_id : bad_overlap_ids) {
ImageBase& aliased = slot_images[aliased_id];
aliased.overlapping_images.push_back(new_image_id);
new_image.overlapping_images.push_back(aliased_id);
new_image.flags |= ImageFlagBits::BadOverlap;
}
RegisterImage(new_image_id);
return new_image_id;
@@ -1195,6 +1297,13 @@ void TextureCache<P>::RegisterImage(ImageId image_id) {
image.flags |= ImageFlagBits::Registered;
ForEachPage(image.cpu_addr, image.guest_size_bytes,
[this, image_id](u64 page) { page_table[page].push_back(image_id); });
u64 tentative_size = std::max(image.guest_size_bytes, image.unswizzled_size_bytes);
if ((IsPixelFormatASTC(image.info.format) &&
True(image.flags & ImageFlagBits::AcceleratedUpload)) ||
True(image.flags & ImageFlagBits::Converted)) {
tentative_size = EstimatedDecompressedSize(tentative_size, image.info.format);
}
total_used_memory += Common::AlignUp(tentative_size, 1024);
}
template <class P>
@@ -1203,6 +1312,14 @@ void TextureCache<P>::UnregisterImage(ImageId image_id) {
ASSERT_MSG(True(image.flags & ImageFlagBits::Registered),
"Trying to unregister an already registered image");
image.flags &= ~ImageFlagBits::Registered;
image.flags &= ~ImageFlagBits::BadOverlap;
u64 tentative_size = std::max(image.guest_size_bytes, image.unswizzled_size_bytes);
if ((IsPixelFormatASTC(image.info.format) &&
True(image.flags & ImageFlagBits::AcceleratedUpload)) ||
True(image.flags & ImageFlagBits::Converted)) {
tentative_size = EstimatedDecompressedSize(tentative_size, image.info.format);
}
total_used_memory -= Common::AlignUp(tentative_size, 1024);
ForEachPage(image.cpu_addr, image.guest_size_bytes, [this, image_id](u64 page) {
const auto page_it = page_table.find(page);
if (page_it == page_table.end()) {
@@ -1276,9 +1393,19 @@ void TextureCache<P>::DeleteImage(ImageId image_id) {
std::erase_if(other_image.aliased_images, [image_id](const AliasedImage& other_alias) {
return other_alias.id == image_id;
});
other_image.CheckAliasState();
ASSERT_MSG(num_removed_aliases == 1, "Invalid number of removed aliases: {}",
num_removed_aliases);
}
for (const ImageId overlap_id : image.overlapping_images) {
ImageBase& other_image = slot_images[overlap_id];
[[maybe_unused]] const size_t num_removed_overlaps = std::erase_if(
other_image.overlapping_images,
[image_id](const ImageId other_overlap_id) { return other_overlap_id == image_id; });
other_image.CheckBadOverlapState();
ASSERT_MSG(num_removed_overlaps == 1, "Invalid number of removed overlapps: {}",
num_removed_overlaps);
}
for (const ImageViewId image_view_id : image_view_ids) {
sentenced_image_view.Push(std::move(slot_image_views[image_view_id]));
slot_image_views.erase(image_view_id);

View File

@@ -47,6 +47,7 @@
#include "video_core/texture_cache/formatter.h"
#include "video_core/texture_cache/samples_helper.h"
#include "video_core/texture_cache/util.h"
#include "video_core/textures/astc.h"
#include "video_core/textures/decoders.h"
namespace VideoCommon {
@@ -580,6 +581,8 @@ void SwizzleBlockLinearImage(Tegra::MemoryManager& gpu_memory, GPUVAddr gpu_addr
for (s32 layer = 0; layer < info.resources.layers; ++layer) {
const std::span<const u8> src = input.subspan(host_offset);
gpu_memory.ReadBlockUnsafe(gpu_addr + guest_offset, dst.data(), dst.size_bytes());
SwizzleTexture(dst, src, bytes_per_block, num_tiles.width, num_tiles.height,
num_tiles.depth, block.height, block.depth);
@@ -647,6 +650,9 @@ u32 CalculateLayerSize(const ImageInfo& info) noexcept {
}
LevelArray CalculateMipLevelOffsets(const ImageInfo& info) noexcept {
if (info.type == ImageType::Linear) {
return {};
}
ASSERT(info.resources.levels <= static_cast<s32>(MAX_MIP_LEVELS));
const LevelInfo level_info = MakeLevelInfo(info);
LevelArray offsets{};
@@ -881,8 +887,16 @@ void ConvertImage(std::span<const u8> input, const ImageInfo& info, std::span<u8
ASSERT(copy.image_extent == mip_size);
ASSERT(copy.buffer_row_length == Common::AlignUp(mip_size.width, tile_size.width));
ASSERT(copy.buffer_image_height == Common::AlignUp(mip_size.height, tile_size.height));
DecompressBC4(input.subspan(copy.buffer_offset), copy.image_extent,
output.subspan(output_offset));
if (IsPixelFormatASTC(info.format)) {
ASSERT(copy.image_extent.depth == 1);
Tegra::Texture::ASTC::Decompress(input.subspan(copy.buffer_offset),
copy.image_extent.width, copy.image_extent.height,
copy.image_subresource.num_layers, tile_size.width,
tile_size.height, output.subspan(output_offset));
} else {
DecompressBC4(input.subspan(copy.buffer_offset), copy.image_extent,
output.subspan(output_offset));
}
copy.buffer_offset = output_offset;
copy.buffer_row_length = mip_size.width;
copy.buffer_image_height = mip_size.height;

File diff suppressed because it is too large Load Diff

View File

@@ -129,4 +129,7 @@ struct AstcBufferData {
decltype(REPLICATE_BYTE_TO_16_TABLE) replicate_byte_to_16 = REPLICATE_BYTE_TO_16_TABLE;
} constexpr ASTC_BUFFER_DATA;
void Decompress(std::span<const uint8_t> data, uint32_t width, uint32_t height, uint32_t depth,
uint32_t block_width, uint32_t block_height, std::span<uint8_t> output);
} // namespace Tegra::Texture::ASTC

View File

@@ -408,6 +408,7 @@ Device::Device(VkInstance instance_, vk::PhysicalDevice physical_, VkSurfaceKHR
}
logical = vk::Device::Create(physical, queue_cis, extensions, first_next, dld);
CollectPhysicalMemoryInfo();
CollectTelemetryParameters();
CollectToolingInfo();
@@ -818,6 +819,17 @@ void Device::CollectTelemetryParameters() {
}
}
void Device::CollectPhysicalMemoryInfo() {
const auto mem_properties = physical.GetMemoryProperties();
const std::size_t num_properties = mem_properties.memoryHeapCount;
device_access_memory = 0;
for (std::size_t element = 0; element < num_properties; element++) {
if ((mem_properties.memoryHeaps[element].flags & VK_MEMORY_HEAP_DEVICE_LOCAL_BIT) != 0) {
device_access_memory += mem_properties.memoryHeaps[element].size;
}
}
}
void Device::CollectToolingInfo() {
if (!ext_tooling_info) {
return;

View File

@@ -225,6 +225,10 @@ public:
return use_asynchronous_shaders;
}
u64 GetDeviceLocalMemory() const {
return device_access_memory;
}
private:
/// Checks if the physical device is suitable.
void CheckSuitability(bool requires_swapchain) const;
@@ -244,6 +248,9 @@ private:
/// Collects information about attached tools.
void CollectToolingInfo();
/// Collects information about the device's local memory.
void CollectPhysicalMemoryInfo();
/// Returns a list of queue initialization descriptors.
std::vector<VkDeviceQueueCreateInfo> GetDeviceQueueCreateInfos() const;
@@ -302,6 +309,8 @@ private:
/// Nsight Aftermath GPU crash tracker
std::unique_ptr<NsightAftermathTracker> nsight_aftermath_tracker;
u64 device_access_memory;
};
} // namespace Vulkan

View File

@@ -69,10 +69,10 @@ constexpr VkExportMemoryAllocateInfo EXPORT_ALLOCATE_INFO{
class MemoryAllocation {
public:
explicit MemoryAllocation(vk::DeviceMemory memory_, VkMemoryPropertyFlags properties,
u64 allocation_size_, u32 type)
: memory{std::move(memory_)}, allocation_size{allocation_size_}, property_flags{properties},
shifted_memory_type{1U << type} {}
explicit MemoryAllocation(MemoryAllocator* const allocator_, vk::DeviceMemory memory_,
VkMemoryPropertyFlags properties, u64 allocation_size_, u32 type)
: allocator{allocator_}, memory{std::move(memory_)}, allocation_size{allocation_size_},
property_flags{properties}, shifted_memory_type{1U << type} {}
#if defined(_WIN32) || defined(__unix__)
~MemoryAllocation() {
@@ -106,6 +106,10 @@ public:
const auto it = std::ranges::find(commits, begin, &Range::begin);
ASSERT_MSG(it != commits.end(), "Invalid commit");
commits.erase(it);
if (commits.empty()) {
// Do not call any code involving 'this' after this call, the object will be destroyed
allocator->ReleaseMemory(this);
}
}
[[nodiscard]] std::span<u8> Map() {
@@ -171,6 +175,7 @@ private:
return candidate;
}
MemoryAllocator* const allocator; ///< Parent memory allocation.
const vk::DeviceMemory memory; ///< Vulkan memory allocation handler.
const u64 allocation_size; ///< Size of this allocation.
const VkMemoryPropertyFlags property_flags; ///< Vulkan memory property flags.
@@ -275,10 +280,17 @@ bool MemoryAllocator::TryAllocMemory(VkMemoryPropertyFlags flags, u32 type_mask,
return false;
}
}
allocations.push_back(std::make_unique<MemoryAllocation>(std::move(memory), flags, size, type));
allocations.push_back(
std::make_unique<MemoryAllocation>(this, std::move(memory), flags, size, type));
return true;
}
void MemoryAllocator::ReleaseMemory(MemoryAllocation* alloc) {
const auto it = std::ranges::find(allocations, alloc, &std::unique_ptr<MemoryAllocation>::get);
ASSERT(it != allocations.end());
allocations.erase(it);
}
std::optional<MemoryCommit> MemoryAllocator::TryCommit(const VkMemoryRequirements& requirements,
VkMemoryPropertyFlags flags) {
for (auto& allocation : allocations) {

View File

@@ -69,6 +69,8 @@ private:
/// Memory allocator container.
/// Allocates and releases memory allocations on demand.
class MemoryAllocator {
friend MemoryAllocation;
public:
/**
* Construct memory allocator
@@ -104,6 +106,9 @@ private:
/// Tries to allocate a chunk of memory.
bool TryAllocMemory(VkMemoryPropertyFlags flags, u32 type_mask, u64 size);
/// Releases a chunk of memory.
void ReleaseMemory(MemoryAllocation* alloc);
/// Tries to allocate a memory commit.
std::optional<MemoryCommit> TryCommit(const VkMemoryRequirements& requirements,
VkMemoryPropertyFlags flags);

View File

@@ -647,6 +647,8 @@ void Config::ReadDebuggingValues() {
ReadSetting(QStringLiteral("program_args"), QString{}).toString().toStdString();
Settings::values.dump_exefs = ReadSetting(QStringLiteral("dump_exefs"), false).toBool();
Settings::values.dump_nso = ReadSetting(QStringLiteral("dump_nso"), false).toBool();
Settings::values.enable_fs_access_log =
ReadSetting(QStringLiteral("enable_fs_access_log"), false).toBool();
Settings::values.reporting_services =
ReadSetting(QStringLiteral("reporting_services"), false).toBool();
Settings::values.quest_flag = ReadSetting(QStringLiteral("quest_flag"), false).toBool();
@@ -756,6 +758,8 @@ void Config::ReadCpuValues() {
QStringLiteral("cpuopt_unsafe_reduce_fp_error"), true);
ReadSettingGlobal(Settings::values.cpuopt_unsafe_inaccurate_nan,
QStringLiteral("cpuopt_unsafe_inaccurate_nan"), true);
ReadSettingGlobal(Settings::values.cpuopt_unsafe_fastmem_check,
QStringLiteral("cpuopt_unsafe_fastmem_check"), true);
if (global) {
Settings::values.cpuopt_page_tables =
@@ -774,6 +778,8 @@ void Config::ReadCpuValues() {
ReadSetting(QStringLiteral("cpuopt_misc_ir"), true).toBool();
Settings::values.cpuopt_reduce_misalign_checks =
ReadSetting(QStringLiteral("cpuopt_reduce_misalign_checks"), true).toBool();
Settings::values.cpuopt_fastmem =
ReadSetting(QStringLiteral("cpuopt_fastmem"), true).toBool();
}
qt_config->endGroup();
@@ -803,6 +809,7 @@ void Config::ReadRendererValues() {
QStringLiteral("use_asynchronous_gpu_emulation"), true);
ReadSettingGlobal(Settings::values.use_nvdec_emulation, QStringLiteral("use_nvdec_emulation"),
true);
ReadSettingGlobal(Settings::values.accelerate_astc, QStringLiteral("accelerate_astc"), true);
ReadSettingGlobal(Settings::values.use_vsync, QStringLiteral("use_vsync"), true);
ReadSettingGlobal(Settings::values.use_assembly_shaders, QStringLiteral("use_assembly_shaders"),
false);
@@ -810,6 +817,7 @@ void Config::ReadRendererValues() {
QStringLiteral("use_asynchronous_shaders"), false);
ReadSettingGlobal(Settings::values.use_fast_gpu_time, QStringLiteral("use_fast_gpu_time"),
true);
ReadSettingGlobal(Settings::values.use_caches_gc, QStringLiteral("use_caches_gc"), false);
ReadSettingGlobal(Settings::values.bg_red, QStringLiteral("bg_red"), 0.0);
ReadSettingGlobal(Settings::values.bg_green, QStringLiteral("bg_green"), 0.0);
ReadSettingGlobal(Settings::values.bg_blue, QStringLiteral("bg_blue"), 0.0);
@@ -1254,6 +1262,8 @@ void Config::SaveDebuggingValues() {
QString::fromStdString(Settings::values.program_args), QString{});
WriteSetting(QStringLiteral("dump_exefs"), Settings::values.dump_exefs, false);
WriteSetting(QStringLiteral("dump_nso"), Settings::values.dump_nso, false);
WriteSetting(QStringLiteral("enable_fs_access_log"), Settings::values.enable_fs_access_log,
false);
WriteSetting(QStringLiteral("quest_flag"), Settings::values.quest_flag, false);
WriteSetting(QStringLiteral("use_debug_asserts"), Settings::values.use_debug_asserts, false);
WriteSetting(QStringLiteral("disable_macro_jit"), Settings::values.disable_macro_jit, false);
@@ -1332,6 +1342,8 @@ void Config::SaveCpuValues() {
Settings::values.cpuopt_unsafe_reduce_fp_error, true);
WriteSettingGlobal(QStringLiteral("cpuopt_unsafe_inaccurate_nan"),
Settings::values.cpuopt_unsafe_inaccurate_nan, true);
WriteSettingGlobal(QStringLiteral("cpuopt_unsafe_fastmem_check"),
Settings::values.cpuopt_unsafe_fastmem_check, true);
if (global) {
WriteSetting(QStringLiteral("cpuopt_page_tables"), Settings::values.cpuopt_page_tables,
@@ -1348,6 +1360,7 @@ void Config::SaveCpuValues() {
WriteSetting(QStringLiteral("cpuopt_misc_ir"), Settings::values.cpuopt_misc_ir, true);
WriteSetting(QStringLiteral("cpuopt_reduce_misalign_checks"),
Settings::values.cpuopt_reduce_misalign_checks, true);
WriteSetting(QStringLiteral("cpuopt_fastmem"), Settings::values.cpuopt_fastmem, true);
}
qt_config->endGroup();
@@ -1381,6 +1394,7 @@ void Config::SaveRendererValues() {
Settings::values.use_asynchronous_gpu_emulation, true);
WriteSettingGlobal(QStringLiteral("use_nvdec_emulation"), Settings::values.use_nvdec_emulation,
true);
WriteSettingGlobal(QStringLiteral("accelerate_astc"), Settings::values.accelerate_astc, true);
WriteSettingGlobal(QStringLiteral("use_vsync"), Settings::values.use_vsync, true);
WriteSettingGlobal(QStringLiteral("use_assembly_shaders"),
Settings::values.use_assembly_shaders, false);
@@ -1388,6 +1402,7 @@ void Config::SaveRendererValues() {
Settings::values.use_asynchronous_shaders, false);
WriteSettingGlobal(QStringLiteral("use_fast_gpu_time"), Settings::values.use_fast_gpu_time,
true);
WriteSettingGlobal(QStringLiteral("use_caches_gc"), Settings::values.use_caches_gc, false);
// Cast to double because Qt's written float values are not human-readable
WriteSettingGlobal(QStringLiteral("bg_red"), Settings::values.bg_red, 0.0);
WriteSettingGlobal(QStringLiteral("bg_green"), Settings::values.bg_green, 0.0);

View File

@@ -35,12 +35,15 @@ void ConfigureCpu::SetConfiguration() {
ui->cpuopt_unsafe_unfuse_fma->setEnabled(runtime_lock);
ui->cpuopt_unsafe_reduce_fp_error->setEnabled(runtime_lock);
ui->cpuopt_unsafe_inaccurate_nan->setEnabled(runtime_lock);
ui->cpuopt_unsafe_fastmem_check->setEnabled(runtime_lock);
ui->cpuopt_unsafe_unfuse_fma->setChecked(Settings::values.cpuopt_unsafe_unfuse_fma.GetValue());
ui->cpuopt_unsafe_reduce_fp_error->setChecked(
Settings::values.cpuopt_unsafe_reduce_fp_error.GetValue());
ui->cpuopt_unsafe_inaccurate_nan->setChecked(
Settings::values.cpuopt_unsafe_inaccurate_nan.GetValue());
ui->cpuopt_unsafe_fastmem_check->setChecked(
Settings::values.cpuopt_unsafe_fastmem_check.GetValue());
if (Settings::IsConfiguringGlobal()) {
ui->accuracy->setCurrentIndex(static_cast<int>(Settings::values.cpu_accuracy.GetValue()));
@@ -84,6 +87,9 @@ void ConfigureCpu::ApplyConfiguration() {
ConfigurationShared::ApplyPerGameSetting(&Settings::values.cpuopt_unsafe_inaccurate_nan,
ui->cpuopt_unsafe_inaccurate_nan,
cpuopt_unsafe_inaccurate_nan);
ConfigurationShared::ApplyPerGameSetting(&Settings::values.cpuopt_unsafe_fastmem_check,
ui->cpuopt_unsafe_fastmem_check,
cpuopt_unsafe_fastmem_check);
if (Settings::IsConfiguringGlobal()) {
// Guard if during game and set to game-specific value
@@ -134,4 +140,7 @@ void ConfigureCpu::SetupPerGameUI() {
ConfigurationShared::SetColoredTristate(ui->cpuopt_unsafe_inaccurate_nan,
Settings::values.cpuopt_unsafe_inaccurate_nan,
cpuopt_unsafe_inaccurate_nan);
ConfigurationShared::SetColoredTristate(ui->cpuopt_unsafe_fastmem_check,
Settings::values.cpuopt_unsafe_fastmem_check,
cpuopt_unsafe_fastmem_check);
}

View File

@@ -41,4 +41,5 @@ private:
ConfigurationShared::CheckState cpuopt_unsafe_unfuse_fma;
ConfigurationShared::CheckState cpuopt_unsafe_reduce_fp_error;
ConfigurationShared::CheckState cpuopt_unsafe_inaccurate_nan;
ConfigurationShared::CheckState cpuopt_unsafe_fastmem_check;
};

View File

@@ -123,6 +123,18 @@
</property>
</widget>
</item>
<item>
<widget class="QCheckBox" name="cpuopt_unsafe_fastmem_check">
<property name="toolTip">
<string>
&lt;div&gt;This option improves speed by eliminating a safety check before every memory read/write in guest. Disabling it may allow a game to read/write the emulator's memory.&lt;/div&gt;
</string>
</property>
<property name="text">
<string>Disable address space checks</string>
</property>
</widget>
</item>
</layout>
</widget>
</item>

View File

@@ -39,6 +39,8 @@ void ConfigureCpuDebug::SetConfiguration() {
ui->cpuopt_misc_ir->setChecked(Settings::values.cpuopt_misc_ir);
ui->cpuopt_reduce_misalign_checks->setEnabled(runtime_lock);
ui->cpuopt_reduce_misalign_checks->setChecked(Settings::values.cpuopt_reduce_misalign_checks);
ui->cpuopt_fastmem->setEnabled(runtime_lock);
ui->cpuopt_fastmem->setChecked(Settings::values.cpuopt_fastmem);
}
void ConfigureCpuDebug::ApplyConfiguration() {
@@ -50,6 +52,7 @@ void ConfigureCpuDebug::ApplyConfiguration() {
Settings::values.cpuopt_const_prop = ui->cpuopt_const_prop->isChecked();
Settings::values.cpuopt_misc_ir = ui->cpuopt_misc_ir->isChecked();
Settings::values.cpuopt_reduce_misalign_checks = ui->cpuopt_reduce_misalign_checks->isChecked();
Settings::values.cpuopt_fastmem = ui->cpuopt_fastmem->isChecked();
}
void ConfigureCpuDebug::changeEvent(QEvent* event) {

View File

@@ -34,7 +34,7 @@
&lt;br&gt;
If you're not sure what these do, keep all of these enabled.
&lt;br&gt;
These settings only take effect when CPU Accuracy is "Debug Mode".
These settings, when disabled, only take effect when CPU Accuracy is "Debug Mode".
&lt;/div&gt;
</string>
</property>
@@ -139,6 +139,20 @@
</property>
</widget>
</item>
<item>
<widget class="QCheckBox" name="cpuopt_fastmem">
<property name="text">
<string>Enable Host MMU Emulation</string>
</property>
<property name="toolTip">
<string>
&lt;div style="white-space: nowrap"&gt;This optimization speeds up memory accesses by the guest program.&lt;/div&gt;
&lt;div style="white-space: nowrap"&gt;Enabling it causes guest memory reads/writes to be done directly into memory and make use of Host's MMU.&lt;/div&gt;
&lt;div style="white-space: nowrap"&gt;Disabling this forces all memory accesses to use Software MMU Emulation.&lt;/div&gt;
</string>
</property>
</widget>
</item>
</layout>
</widget>
</item>

View File

@@ -28,17 +28,21 @@ ConfigureDebug::ConfigureDebug(QWidget* parent) : QWidget(parent), ui(new Ui::Co
ConfigureDebug::~ConfigureDebug() = default;
void ConfigureDebug::SetConfiguration() {
ui->toggle_console->setEnabled(!Core::System::GetInstance().IsPoweredOn());
const bool runtime_lock = !Core::System::GetInstance().IsPoweredOn();
ui->toggle_console->setEnabled(runtime_lock);
ui->toggle_console->setChecked(UISettings::values.show_console);
ui->log_filter_edit->setText(QString::fromStdString(Settings::values.log_filter));
ui->homebrew_args_edit->setText(QString::fromStdString(Settings::values.program_args));
ui->fs_access_log->setEnabled(runtime_lock);
ui->fs_access_log->setChecked(Settings::values.enable_fs_access_log);
ui->reporting_services->setChecked(Settings::values.reporting_services);
ui->quest_flag->setChecked(Settings::values.quest_flag);
ui->use_debug_asserts->setChecked(Settings::values.use_debug_asserts);
ui->use_auto_stub->setChecked(Settings::values.use_auto_stub);
ui->enable_graphics_debugging->setEnabled(!Core::System::GetInstance().IsPoweredOn());
ui->enable_graphics_debugging->setEnabled(runtime_lock);
ui->enable_graphics_debugging->setChecked(Settings::values.renderer_debug);
ui->disable_macro_jit->setEnabled(!Core::System::GetInstance().IsPoweredOn());
ui->disable_macro_jit->setEnabled(runtime_lock);
ui->disable_macro_jit->setChecked(Settings::values.disable_macro_jit);
ui->extended_logging->setChecked(Settings::values.extended_logging);
}
@@ -47,6 +51,7 @@ void ConfigureDebug::ApplyConfiguration() {
UISettings::values.show_console = ui->toggle_console->isChecked();
Settings::values.log_filter = ui->log_filter_edit->text().toStdString();
Settings::values.program_args = ui->homebrew_args_edit->text().toStdString();
Settings::values.enable_fs_access_log = ui->fs_access_log->isChecked();
Settings::values.reporting_services = ui->reporting_services->isChecked();
Settings::values.quest_flag = ui->quest_flag->isChecked();
Settings::values.use_debug_asserts = ui->use_debug_asserts->isChecked();

View File

@@ -144,9 +144,16 @@
<item>
<widget class="QGroupBox" name="groupBox_5">
<property name="title">
<string>Dump</string>
<string>Debugging</string>
</property>
<layout class="QVBoxLayout" name="verticalLayout_7">
<item>
<widget class="QCheckBox" name="fs_access_log">
<property name="text">
<string>Enable FS Access Log</string>
</property>
</widget>
</item>
<item>
<widget class="QCheckBox" name="reporting_services">
<property name="text">

View File

@@ -70,10 +70,12 @@ void ConfigureGraphics::SetConfiguration() {
ui->use_asynchronous_gpu_emulation->setEnabled(runtime_lock);
ui->use_disk_shader_cache->setEnabled(runtime_lock);
ui->use_nvdec_emulation->setEnabled(runtime_lock);
ui->accelerate_astc->setEnabled(runtime_lock);
ui->use_disk_shader_cache->setChecked(Settings::values.use_disk_shader_cache.GetValue());
ui->use_asynchronous_gpu_emulation->setChecked(
Settings::values.use_asynchronous_gpu_emulation.GetValue());
ui->use_nvdec_emulation->setChecked(Settings::values.use_nvdec_emulation.GetValue());
ui->accelerate_astc->setChecked(Settings::values.accelerate_astc.GetValue());
if (Settings::IsConfiguringGlobal()) {
ui->api->setCurrentIndex(static_cast<int>(Settings::values.renderer_backend.GetValue()));
@@ -118,6 +120,8 @@ void ConfigureGraphics::ApplyConfiguration() {
use_asynchronous_gpu_emulation);
ConfigurationShared::ApplyPerGameSetting(&Settings::values.use_nvdec_emulation,
ui->use_nvdec_emulation, use_nvdec_emulation);
ConfigurationShared::ApplyPerGameSetting(&Settings::values.accelerate_astc, ui->accelerate_astc,
accelerate_astc);
if (Settings::IsConfiguringGlobal()) {
// Guard if during game and set to game-specific value
@@ -254,6 +258,7 @@ void ConfigureGraphics::SetupPerGameUI() {
ui->use_asynchronous_gpu_emulation->setEnabled(
Settings::values.use_asynchronous_gpu_emulation.UsingGlobal());
ui->use_nvdec_emulation->setEnabled(Settings::values.use_nvdec_emulation.UsingGlobal());
ui->accelerate_astc->setEnabled(Settings::values.accelerate_astc.UsingGlobal());
ui->use_disk_shader_cache->setEnabled(Settings::values.use_disk_shader_cache.UsingGlobal());
ui->bg_button->setEnabled(Settings::values.bg_red.UsingGlobal());
@@ -269,6 +274,8 @@ void ConfigureGraphics::SetupPerGameUI() {
ui->use_disk_shader_cache, Settings::values.use_disk_shader_cache, use_disk_shader_cache);
ConfigurationShared::SetColoredTristate(
ui->use_nvdec_emulation, Settings::values.use_nvdec_emulation, use_nvdec_emulation);
ConfigurationShared::SetColoredTristate(ui->accelerate_astc, Settings::values.accelerate_astc,
accelerate_astc);
ConfigurationShared::SetColoredTristate(ui->use_asynchronous_gpu_emulation,
Settings::values.use_asynchronous_gpu_emulation,
use_asynchronous_gpu_emulation);

View File

@@ -47,6 +47,7 @@ private:
QColor bg_color;
ConfigurationShared::CheckState use_nvdec_emulation;
ConfigurationShared::CheckState accelerate_astc;
ConfigurationShared::CheckState use_disk_shader_cache;
ConfigurationShared::CheckState use_asynchronous_gpu_emulation;

View File

@@ -104,6 +104,13 @@
</property>
</widget>
</item>
<item>
<widget class="QCheckBox" name="accelerate_astc">
<property name="text">
<string>Accelerate ASTC texture decoding</string>
</property>
</widget>
</item>
<item>
<widget class="QWidget" name="fullscreen_mode_layout" native="true">
<layout class="QHBoxLayout" name="horizontalLayout_1">

View File

@@ -30,6 +30,7 @@ void ConfigureGraphicsAdvanced::SetConfiguration() {
ui->use_vsync->setChecked(Settings::values.use_vsync.GetValue());
ui->use_assembly_shaders->setChecked(Settings::values.use_assembly_shaders.GetValue());
ui->use_asynchronous_shaders->setChecked(Settings::values.use_asynchronous_shaders.GetValue());
ui->use_caches_gc->setChecked(Settings::values.use_caches_gc.GetValue());
ui->use_fast_gpu_time->setChecked(Settings::values.use_fast_gpu_time.GetValue());
if (Settings::IsConfiguringGlobal()) {
@@ -62,6 +63,8 @@ void ConfigureGraphicsAdvanced::ApplyConfiguration() {
ConfigurationShared::ApplyPerGameSetting(&Settings::values.use_asynchronous_shaders,
ui->use_asynchronous_shaders,
use_asynchronous_shaders);
ConfigurationShared::ApplyPerGameSetting(&Settings::values.use_caches_gc, ui->use_caches_gc,
use_caches_gc);
ConfigurationShared::ApplyPerGameSetting(&Settings::values.use_fast_gpu_time,
ui->use_fast_gpu_time, use_fast_gpu_time);
@@ -101,6 +104,7 @@ void ConfigureGraphicsAdvanced::SetupPerGameUI() {
ui->use_asynchronous_shaders->setEnabled(
Settings::values.use_asynchronous_shaders.UsingGlobal());
ui->use_fast_gpu_time->setEnabled(Settings::values.use_fast_gpu_time.UsingGlobal());
ui->use_caches_gc->setEnabled(Settings::values.use_caches_gc.UsingGlobal());
ui->anisotropic_filtering_combobox->setEnabled(
Settings::values.max_anisotropy.UsingGlobal());
@@ -115,6 +119,8 @@ void ConfigureGraphicsAdvanced::SetupPerGameUI() {
use_asynchronous_shaders);
ConfigurationShared::SetColoredTristate(ui->use_fast_gpu_time,
Settings::values.use_fast_gpu_time, use_fast_gpu_time);
ConfigurationShared::SetColoredTristate(ui->use_caches_gc, Settings::values.use_caches_gc,
use_caches_gc);
ConfigurationShared::SetColoredComboBox(
ui->gpu_accuracy, ui->label_gpu_accuracy,
static_cast<int>(Settings::values.gpu_accuracy.GetValue(true)));

View File

@@ -38,4 +38,5 @@ private:
ConfigurationShared::CheckState use_assembly_shaders;
ConfigurationShared::CheckState use_asynchronous_shaders;
ConfigurationShared::CheckState use_fast_gpu_time;
ConfigurationShared::CheckState use_caches_gc;
};

View File

@@ -103,6 +103,16 @@
</property>
</widget>
</item>
<item>
<widget class="QCheckBox" name="use_caches_gc">
<property name="toolTip">
<string>Enables garbage collection for the GPU caches, this will try to keep VRAM within 3-4 GB by flushing the least used textures/buffers. May cause issues in a few games.</string>
</property>
<property name="text">
<string>Enable GPU cache garbage collection (experimental)</string>
</property>
</widget>
</item>
<item>
<widget class="QWidget" name="af_layout" native="true">
<layout class="QHBoxLayout" name="horizontalLayout_1">

View File

@@ -1395,7 +1395,8 @@ void ConfigureInputPlayer::keyPressEvent(QKeyEvent* event) {
void ConfigureInputPlayer::CreateProfile() {
const auto profile_name =
LimitableInputDialog::GetText(this, tr("New Profile"), tr("Enter a profile name:"), 1, 20);
LimitableInputDialog::GetText(this, tr("New Profile"), tr("Enter a profile name:"), 1, 20,
LimitableInputDialog::InputLimiter::Filesystem);
if (profile_name.isEmpty()) {
return;

View File

@@ -17,17 +17,30 @@
namespace {
constexpr std::array default_icon_sizes{
std::make_pair(0, QT_TR_NOOP("None")),
std::make_pair(32, QT_TR_NOOP("Small (32x32)")),
std::make_pair(64, QT_TR_NOOP("Standard (64x64)")),
std::make_pair(128, QT_TR_NOOP("Large (128x128)")),
std::make_pair(256, QT_TR_NOOP("Full Size (256x256)")),
std::make_pair(0, QT_TRANSLATE_NOOP("ConfigureUI", "None")),
std::make_pair(32, QT_TRANSLATE_NOOP("ConfigureUI", "Small (32x32)")),
std::make_pair(64, QT_TRANSLATE_NOOP("ConfigureUI", "Standard (64x64)")),
std::make_pair(128, QT_TRANSLATE_NOOP("ConfigureUI", "Large (128x128)")),
std::make_pair(256, QT_TRANSLATE_NOOP("ConfigureUI", "Full Size (256x256)")),
};
// clang-format off
constexpr std::array row_text_names{
QT_TR_NOOP("Filename"), QT_TR_NOOP("Filetype"), QT_TR_NOOP("Title ID"),
QT_TR_NOOP("Title Name"), QT_TR_NOOP("None"),
QT_TRANSLATE_NOOP("ConfigureUI", "Filename"),
QT_TRANSLATE_NOOP("ConfigureUI", "Filetype"),
QT_TRANSLATE_NOOP("ConfigureUI", "Title ID"),
QT_TRANSLATE_NOOP("ConfigureUI", "Title Name"),
QT_TRANSLATE_NOOP("ConfigureUI", "None"),
};
// clang-format on
QString GetTranslatedIconSize(size_t index) {
return QCoreApplication::translate("ConfigureUI", default_icon_sizes[index].second);
}
QString GetTranslatedRowTextName(size_t index) {
return QCoreApplication::translate("ConfigureUI", row_text_names[index]);
}
} // Anonymous namespace
ConfigureUi::ConfigureUi(QWidget* parent) : QWidget(parent), ui(new Ui::ConfigureUi) {
@@ -121,11 +134,11 @@ void ConfigureUi::RetranslateUI() {
ui->retranslateUi(this);
for (int i = 0; i < ui->icon_size_combobox->count(); i++) {
ui->icon_size_combobox->setItemText(i, tr(default_icon_sizes[i].second));
ui->icon_size_combobox->setItemText(i, GetTranslatedIconSize(static_cast<size_t>(i)));
}
for (int i = 0; i < ui->row_1_text_combobox->count(); i++) {
const QString name = tr(row_text_names[i]);
const QString name = GetTranslatedRowTextName(static_cast<size_t>(i));
ui->row_1_text_combobox->setItemText(i, name);
ui->row_2_text_combobox->setItemText(i, name);
@@ -152,8 +165,9 @@ void ConfigureUi::InitializeLanguageComboBox() {
}
void ConfigureUi::InitializeIconSizeComboBox() {
for (const auto& size : default_icon_sizes) {
ui->icon_size_combobox->addItem(QString::fromUtf8(size.second), size.first);
for (size_t i = 0; i < default_icon_sizes.size(); i++) {
const auto size = default_icon_sizes[i].first;
ui->icon_size_combobox->addItem(GetTranslatedIconSize(i), size);
}
}
@@ -170,7 +184,7 @@ void ConfigureUi::UpdateFirstRowComboBox(bool init) {
ui->row_1_text_combobox->clear();
for (std::size_t i = 0; i < row_text_names.size(); i++) {
const QString row_text_name = QString::fromUtf8(row_text_names[i]);
const QString row_text_name = GetTranslatedRowTextName(i);
ui->row_1_text_combobox->addItem(row_text_name, QVariant::fromValue(i));
}
@@ -189,7 +203,7 @@ void ConfigureUi::UpdateSecondRowComboBox(bool init) {
ui->row_2_text_combobox->clear();
for (std::size_t i = 0; i < row_text_names.size(); ++i) {
const QString row_text_name = QString::fromUtf8(row_text_names[i]);
const QString row_text_name = GetTranslatedRowTextName(i);
ui->row_2_text_combobox->addItem(row_text_name, QVariant::fromValue(i));
}

View File

@@ -505,6 +505,10 @@ void GameList::PopupContextMenu(const QPoint& menu_location) {
void GameList::AddGamePopup(QMenu& context_menu, u64 program_id, const std::string& path) {
QAction* favorite = context_menu.addAction(tr("Favorite"));
context_menu.addSeparator();
QAction* start_game = context_menu.addAction(tr("Start Game"));
QAction* start_game_global =
context_menu.addAction(tr("Start Game without Custom Configuration"));
context_menu.addSeparator();
QAction* open_save_location = context_menu.addAction(tr("Open Save Data Location"));
QAction* open_mod_location = context_menu.addAction(tr("Open Mod Data Location"));
QAction* open_transferable_shader_cache =
@@ -540,6 +544,12 @@ void GameList::AddGamePopup(QMenu& context_menu, u64 program_id, const std::stri
connect(open_save_location, &QAction::triggered, [this, program_id, path]() {
emit OpenFolderRequested(program_id, GameListOpenTarget::SaveData, path);
});
connect(start_game, &QAction::triggered, [this, path]() {
emit BootGame(QString::fromStdString(path), 0, StartGameType::Normal);
});
connect(start_game_global, &QAction::triggered, [this, path]() {
emit BootGame(QString::fromStdString(path), 0, StartGameType::Global);
});
connect(open_mod_location, &QAction::triggered, [this, program_id, path]() {
emit OpenFolderRequested(program_id, GameListOpenTarget::ModData, path);
});

View File

@@ -28,6 +28,7 @@ class GameListWorker;
class GameListSearchField;
class GameListDir;
class GMainWindow;
enum class StartGameType;
namespace FileSys {
class ManualContentProvider;
@@ -82,6 +83,7 @@ public:
static const QStringList supported_file_extensions;
signals:
void BootGame(const QString& game_path, std::size_t program_index, StartGameType type);
void GameChosen(const QString& game_path);
void ShouldCancelWorker();
void OpenFolderRequested(u64 program_id, GameListOpenTarget target,

View File

@@ -1094,6 +1094,7 @@ void GMainWindow::OnAppFocusStateChanged(Qt::ApplicationState state) {
}
void GMainWindow::ConnectWidgetEvents() {
connect(game_list, &GameList::BootGame, this, &GMainWindow::BootGame);
connect(game_list, &GameList::GameChosen, this, &GMainWindow::OnGameListLoadFile);
connect(game_list, &GameList::OpenDirectory, this, &GMainWindow::OnGameListOpenDirectory);
connect(game_list, &GameList::OpenFolderRequested, this, &GMainWindow::OnGameListOpenFolder);
@@ -1320,7 +1321,7 @@ void GMainWindow::SelectAndSetCurrentUser() {
Settings::values.current_user = dialog.GetIndex();
}
void GMainWindow::BootGame(const QString& filename, std::size_t program_index) {
void GMainWindow::BootGame(const QString& filename, std::size_t program_index, StartGameType type) {
LOG_INFO(Frontend, "yuzu starting...");
StoreRecentFile(filename); // Put the filename on top of the list
@@ -1332,7 +1333,8 @@ void GMainWindow::BootGame(const QString& filename, std::size_t program_index) {
const auto v_file = Core::GetGameFileFromPath(vfs, filename.toUtf8().constData());
const auto loader = Loader::GetLoader(system, v_file, program_index);
if (!(loader == nullptr || loader->ReadProgramId(title_id) != Loader::ResultStatus::Success)) {
if (loader != nullptr && loader->ReadProgramId(title_id) == Loader::ResultStatus::Success &&
type == StartGameType::Normal) {
// Load per game settings
const auto file_path = std::filesystem::path{filename.toStdU16String()};
const auto config_file_name = title_id == 0
@@ -1944,6 +1946,18 @@ void GMainWindow::OnGameListDumpRomFS(u64 program_id, const std::string& game_pa
const auto full = res == selections.constFirst();
const auto entry_size = CalculateRomFSEntrySize(extracted, full);
// The minimum required space is the size of the extracted RomFS + 1 GiB
const auto minimum_free_space = extracted->GetSize() + 0x40000000;
if (full && Common::FS::GetFreeSpaceSize(path) < minimum_free_space) {
QMessageBox::warning(this, tr("RomFS Extraction Failed!"),
tr("There is not enough free space at %1 to extract the RomFS. Please "
"free up space or select a different dump directory at "
"Emulation > Configure > System > Filesystem > Dump Root")
.arg(QString::fromStdString(path)));
return;
}
QProgressDialog progress(tr("Extracting RomFS..."), tr("Cancel"), 0,
static_cast<s32>(entry_size), this);
progress.setWindowModality(Qt::WindowModal);

View File

@@ -39,6 +39,11 @@ class GameListPlaceholder;
class QtSoftwareKeyboardDialog;
enum class StartGameType {
Normal, // Can use custom configuration
Global, // Only uses global configuration
};
namespace Core::Frontend {
struct ControllerParameters;
struct InlineAppearParameters;
@@ -181,7 +186,8 @@ private:
void AllowOSSleep();
bool LoadROM(const QString& filename, std::size_t program_index);
void BootGame(const QString& filename, std::size_t program_index = 0);
void BootGame(const QString& filename, std::size_t program_index = 0,
StartGameType with_config = StartGameType::Normal);
void ShutdownGame();
void ShowTelemetryCallout();

View File

@@ -21,11 +21,13 @@ void LimitableInputDialog::CreateUI() {
text_label = new QLabel(this);
text_entry = new QLineEdit(this);
text_label_invalid = new QLabel(this);
buttons = new QDialogButtonBox(QDialogButtonBox::Ok | QDialogButtonBox::Cancel, this);
auto* const layout = new QVBoxLayout;
layout->addWidget(text_label);
layout->addWidget(text_entry);
layout->addWidget(text_label_invalid);
layout->addWidget(buttons);
setLayout(layout);
@@ -37,18 +39,36 @@ void LimitableInputDialog::ConnectEvents() {
}
QString LimitableInputDialog::GetText(QWidget* parent, const QString& title, const QString& text,
int min_character_limit, int max_character_limit) {
int min_character_limit, int max_character_limit,
InputLimiter limit_type) {
Q_ASSERT(min_character_limit <= max_character_limit);
LimitableInputDialog dialog{parent};
dialog.setWindowTitle(title);
dialog.text_label->setText(text);
dialog.text_entry->setMaxLength(max_character_limit);
dialog.text_label_invalid->show();
switch (limit_type) {
case InputLimiter::Filesystem:
dialog.invalid_characters = QStringLiteral("<>:;\"/\\|,.!?*");
break;
default:
dialog.invalid_characters.clear();
dialog.text_label_invalid->hide();
break;
}
dialog.text_label_invalid->setText(
tr("The text can't contain any of the following characters:\n%1")
.arg(dialog.invalid_characters));
auto* const ok_button = dialog.buttons->button(QDialogButtonBox::Ok);
ok_button->setEnabled(false);
connect(dialog.text_entry, &QLineEdit::textEdited, [&](const QString& new_text) {
ok_button->setEnabled(new_text.length() >= min_character_limit);
connect(dialog.text_entry, &QLineEdit::textEdited, [&] {
if (!dialog.invalid_characters.isEmpty()) {
dialog.RemoveInvalidCharacters();
}
ok_button->setEnabled(dialog.text_entry->text().length() >= min_character_limit);
});
if (dialog.exec() != QDialog::Accepted) {
@@ -57,3 +77,15 @@ QString LimitableInputDialog::GetText(QWidget* parent, const QString& title, con
return dialog.text_entry->text();
}
void LimitableInputDialog::RemoveInvalidCharacters() {
auto cpos = text_entry->cursorPosition();
for (int i = 0; i < text_entry->text().length(); i++) {
if (invalid_characters.contains(text_entry->text().at(i))) {
text_entry->setText(text_entry->text().remove(i, 1));
i--;
cpos--;
}
}
text_entry->setCursorPosition(cpos);
}

View File

@@ -18,14 +18,24 @@ public:
explicit LimitableInputDialog(QWidget* parent = nullptr);
~LimitableInputDialog() override;
enum class InputLimiter {
None,
Filesystem,
};
static QString GetText(QWidget* parent, const QString& title, const QString& text,
int min_character_limit, int max_character_limit);
int min_character_limit, int max_character_limit,
InputLimiter limit_type = InputLimiter::None);
private:
void CreateUI();
void ConnectEvents();
void RemoveInvalidCharacters();
QString invalid_characters;
QLabel* text_label;
QLineEdit* text_entry;
QLabel* text_label_invalid;
QDialogButtonBox* buttons;
};

Some files were not shown because too many files have changed in this diff Show More