Compare commits

...

185 Commits

Author SHA1 Message Date
Liam dea5de91d2 buffer_cache: reset cached write bits after flushing invalidations 2022-03-23 20:59:01 -04:00
Ameer J fb4d80b169 Merge pull request #8073 from Shoegzer/compile_fixes
Add include to fix compiling
2022-03-23 16:31:02 -04:00
Shoegzer a8896e940a Add include to fix compiling 2022-03-23 16:18:48 -04:00
bunnei 749efbb84b Merge pull request #8031 from Morph1984/cleanup-mii-please
applets: Cleanup MiiEdit applet implementation
2022-03-22 21:56:42 -07:00
bunnei f8773e4ba0 Merge pull request #8070 from yuzu-emu/revert-8016-kill-mem-use
Revert "dynarmic: Reduce size of code caches"
2022-03-22 18:34:19 -07:00
bunnei e439c31c9e Revert "dynarmic: Reduce size of code caches" 2022-03-22 18:32:54 -07:00
Morph 6e232b9efe applets/mii: Remove unused include 2022-03-21 23:57:31 -04:00
Morph bcca849d0c applets/mii: Remove frontend parameters
These are unused for now as we do not support a frontend implementation.
2022-03-21 23:57:31 -04:00
Morph c50401903f applets/mii: Cleanup MiiEdit applet implementation
This also enables proper support for MiiEdit applets which are used in games with firmware versions prior to 10.2.0 by handling the 2 different versions of applet inputs and outputs.
2022-03-21 23:57:31 -04:00
Morph 1d5f6b173d applets/mii: Cleanup MiiEdit applet types 2022-03-21 23:57:31 -04:00
Morph 7c44a4ede3 applets/mii: Move MiiEdit applet types into its own file 2022-03-21 23:57:31 -04:00
Morph aa44ef7b64 service: Move mii enums and structs into its own file
Moves these into types.h, since other files also make use of these types.
2022-03-21 23:57:31 -04:00
Morph e05e6853fa applets: Rename Mii to MiiEdit 2022-03-21 23:57:31 -04:00
Narr the Reg f83cd2e8b9 Merge pull request #8067 from ameerj/qt-browser-include
qt_web_browser: Add missing includes
2022-03-21 21:33:09 -06:00
ameerj b3cfccdb80 qt_web_browser: Add missing includes 2022-03-21 22:42:14 -04:00
Ameer J 75046a3351 Merge pull request #8038 from liamwhite/exit-register-detection
shader_recompiler/EXIT: increment output register on failed enable test
2022-03-21 21:24:07 -04:00
bunnei 82ac66f8a4 Merge pull request #8048 from ameerj/include-purge
general: Reduce unused includes across the project
2022-03-21 18:03:54 -07:00
bunnei ff2e891022 Merge pull request #7812 from FernandoS27/made-straight-from-the-nut
BufferCache: Find direction of the stream buffer increase.
2022-03-20 15:23:53 -07:00
Fernando Sahmkow 3b0d233cbd BufferCache: Find direction of the stream buffer increase. 2022-03-20 21:37:23 +01:00
Mai M 628534a9ac Merge pull request #8054 from merryhime/dynarmic
dynarmic: Accelerate SHA256 and implement for A32 frontend
2022-03-20 10:37:38 -04:00
Merry d1c0cdf4f2 dynarmic: Accelerate SHA256 and implement for A32 frontend
* Implements hardware acceleration for SHA256 instructions.
 * Adds SHA256 instructions introduced in ARMv8 to A32 frontend.
 * Implements polyfill for processors that do not support hardware
   accelerated SHA instructions.
2022-03-20 14:06:17 +00:00
Fernando S 7c05c987a3 Merge pull request #8036 from ameerj/starbit-nv
vk_texture_cache: Do not reinterpret DepthStencil source images
2022-03-20 13:35:59 +01:00
bunnei 150f6db4d1 Merge pull request #7840 from lioncash/bitor
texture_cache: Amend unintended bitwise OR in SynchronizeAliases
2022-03-20 02:19:30 -07:00
ameerj 866b7c20a8 general: Fix clang/gcc build errors 2022-03-20 02:25:09 -04:00
ameerj a367db44cf yuzu_cmd: Reduce unused includes 2022-03-20 02:25:09 -04:00
ameerj 936829e873 yuzu: Reduce unused includes 2022-03-20 02:25:09 -04:00
ameerj 9b505758dc web_service: Reduce unused includes 2022-03-20 02:25:09 -04:00
ameerj 967ed01fcf input_common: Reduce unused includes 2022-03-20 02:25:09 -04:00
ameerj 574a2c4b77 shader_recompiler: Reduce unused includes 2022-03-20 02:25:08 -04:00
bunnei 5960d54722 Merge pull request #8040 from Morph1984/handle-table
KHandleTable: Optimize table entry layout
2022-03-19 23:17:37 -07:00
bunnei 474318ee37 Merge pull request #8047 from ameerj/msvc-test-disable
.ci/build-msvc: Disable YUZU_TESTS cmake variable
2022-03-19 16:28:17 -07:00
ameerj 923decae5a common: Reduce unused includes 2022-03-19 15:01:31 -04:00
ameerj 1bc7d61b57 video_core: Reduce unused includes 2022-03-19 15:01:31 -04:00
bunnei 17ebe211ec Merge pull request #8025 from lat9nq/cmd-specify-config
yuzu_cmd: Allow user to specify config file location
2022-03-19 01:33:50 -07:00
ameerj c85a3e5a28 build-msvc: Disable tests 2022-03-19 02:35:09 -04:00
ameerj 8a8ea65fae common: Reduce unused includes 2022-03-19 02:23:33 -04:00
ameerj ade596121b core: Reduce unused includes 2022-03-19 02:23:32 -04:00
bunnei 8c8b5359f2 Merge pull request #8028 from v1993/patch-9
bsd: Allow inexact match for address length in AcceptImpl
2022-03-18 18:06:13 -07:00
Liam 536d7ed7b1 Address review comments 2022-03-18 15:55:46 -04:00
Liam d400b618a7 shader_recompiler/EXIT: skip render targets with no outputs 2022-03-18 09:26:25 -04:00
Morph fe1182e916 Merge pull request #8039 from ameerj/core-include
general: Reduce core.h includes
2022-03-18 02:45:30 -04:00
ameerj d618bba8a6 general: Reduce core.h includes 2022-03-18 02:13:02 -04:00
Morph 8b7d571b66 KHandleTable: Optimize table entry layout
Since the handle type is not being used, we can reduce the amount of space each entry takes up by 4 bytes.
2022-03-18 00:28:25 -04:00
Liam 6fa17f3372 shader_recompiler/EXIT: increment output register on failed enable test 2022-03-17 22:09:31 -04:00
ameerj 4d840aa903 vk_texture_cache: Do not reinterpret DepthStencil source images
Fixes star pointer interactions in Super Mario Galaxy on some drivers, notably Nvidia.

Co-Authored-By: Fernando S. <1731197+fernandos27@users.noreply.github.com>
2022-03-17 20:55:05 -04:00
Fernando S cb86e7941b Merge pull request #8024 from liamwhite/const-indexing
Add shader support for const buffer indirect addressing
2022-03-18 00:36:31 +01:00
Fernando S a616f49864 Merge pull request #8030 from liamwhite/s8d24-conversion
Vulkan: convert S8D24 <-> ABGR8
2022-03-18 00:36:06 +01:00
Liam 3009d0bd7d Address review comments 2022-03-17 14:48:18 -04:00
Liam e228a40db8 shader_recompiler: Use functions for indirect const buffer accesses 2022-03-17 13:30:21 -04:00
Liam 3ac522ba41 Address review comments 2022-03-17 09:30:41 -04:00
bunnei f55af65e82 Merge pull request #7964 from german77/miiii
applet: mii: Simple implementation of mii applet
2022-03-16 21:37:53 -07:00
Liam 6407f16d81 Address review comments 2022-03-16 18:00:42 -04:00
Liam 1415542f73 shader_recompiler: Implement LDC.IS address mode 2022-03-16 11:05:04 -04:00
Fernando S 2db5076ec9 Merge pull request #8013 from bunnei/kernel-slab-rework-v2
Kernel Memory Updates (Part 6): Use guest memory for slab heaps & update TLS.
2022-03-16 12:15:33 +01:00
Fernando S c3c351e2c2 Merge pull request #8023 from ameerj/kirby-pop-in
maxwell_3d: Implement a safer CB data upload
2022-03-16 12:14:08 +01:00
bunnei 613558867c Merge pull request #8026 from lat9nq/ext-mem-ini
default_ini: List use_extended_memory_layout in default config file
2022-03-15 18:12:10 -07:00
Liam bcc2d7e69b Vulkan: convert S8D24 <-> ABGR8 2022-03-15 20:05:21 -04:00
Valeri 9e633999d6 bsd: Allow inexact match for address length in AcceptImpl
Minecraft passes in zero for length, but this should account for all possible cases
2022-03-15 14:06:34 +03:00
lat9nq 24d51e1c92 yuzu_cmd: Allow user to specify config file location
Adds an option `-c` or `--config` with one required argument that allows
the user to specify to where the config file is located. Useful for
scripts that run specific games with different preferences for settings.
2022-03-15 03:48:40 -04:00
lat9nq cb32d9aff8 default_ini: List use_extended_memory_layout in default config file 2022-03-15 03:13:55 -04:00
bunnei 59d2a38daa Merge pull request #8006 from BytesGalore/fix_cmake_missing_qt5_dbus
build(cmake): fix missing Qt5::DBus link target for bundled linux package
2022-03-14 18:56:39 -07:00
bunnei e95bb782f0 core: hle: kernel: init_slab_setup: Move CalculateSlabHeapGapSize to global namespace. 2022-03-14 18:14:54 -07:00
bunnei 5f3e77d93e core: hle: kernel: Allocate dummy threads on host thread storage.
- Fixes a crash where on subsequent boots, long-lived host threads would have their dummy threads freed.
2022-03-14 18:14:54 -07:00
bunnei 82a2463062 core: hle: kernel: Downgrade dangling objects warning to debug.
- It is not impossible to leak kernel objects, so this is not really any issue anymore (albeit, still interesting).
2022-03-14 18:14:54 -07:00
bunnei f7d1929816 core: hle: kernel: Make object list container global and ensure it is reset on each emulation session. 2022-03-14 18:14:54 -07:00
bunnei 51589c5e21 core: hle: kernel: Remove server session tracking.
- These are now allocated/managed by emulated memory, so we do not need to track and free them on shutdown.
2022-03-14 18:14:54 -07:00
bunnei 0defac2f2a core: hle: kernel: k_process: Remove handle table finalize, reset page table. 2022-03-14 18:14:54 -07:00
bunnei 813b2ef253 core: hle: kernel: k_process: Implement thread local storage accurately. 2022-03-14 18:14:54 -07:00
bunnei 3210bc2767 core: hle: kernel: k_page_table: Add implementations of MapPages, UnmapPages, and FindFreeArea for TLS. 2022-03-14 18:14:54 -07:00
bunnei 15d9b0418f core: hle: kernel: k_slab_heap: Refresh to use guest allocations. 2022-03-14 18:14:54 -07:00
bunnei a25cd4bb4b core: hle: kernel: Update init_slab_heap, use device memory, and add KThreadLocalPage and KPageBuffer.
- Refreshes our slab initialization code to latest known behavior.
- Moves all guest kernel slabs into emulated device memory.
- Adds KThreadLocalPage and KPageBuffer, which we will use for accurate TLS management.
2022-03-14 18:14:54 -07:00
bunnei 91819726b1 core: hle: kernel: k_page_buffer: Add KThreadLocalPage primitive. 2022-03-14 18:14:53 -07:00
bunnei 08434842b3 core: hle: kernel: k_page_buffer: Add KPageBuffer primitive. 2022-03-14 18:14:53 -07:00
bunnei 4a28d8cebb core: hle: kernel: k_thread: Ensure host Fiber is freed. 2022-03-14 18:14:53 -07:00
bunnei ed67e1dd10 core: hle: kernel: k_server_session: Ensure SessionRequestManager is freed. 2022-03-14 18:14:53 -07:00
bunnei bfc4823e36 core: hle: service: kernel_helpers: Use system resource limit. 2022-03-14 18:14:53 -07:00
bunnei 8873c0c3db core: hle: service: sm: Fix KPort reference count. 2022-03-14 18:14:53 -07:00
bunnei 25c0acc388 core: hle: kernel: k_thread: Update to reflect tree changes. 2022-03-14 18:14:53 -07:00
bunnei 07c9d9bdbd core: hle: kernel: Use weak_ptr where possible for SessionRequestHandler and SessionRequestManager. 2022-03-14 18:14:53 -07:00
bunnei ce33503adf core: hle: kernel: k_memory_layout: Update kernel slab memory sizes. 2022-03-14 18:14:53 -07:00
bunnei 0f0e1c25bc core: hle: kernel: svc_types: Add ThreadLocalRegionSize. 2022-03-14 18:14:53 -07:00
bunnei 944d9186ca core: hle: kernel: k_condition_variable: Update to reflect tree changes. 2022-03-14 18:14:53 -07:00
bunnei 158c5845ab core: hle: kernel: k_address_arbiter: Update to reflect tree changes. 2022-03-14 18:14:53 -07:00
bunnei 0fdf1d2a60 common: tree: Various updates. 2022-03-14 18:14:53 -07:00
bunnei 69c2faeb6a common: intrusive_red_black_tree: Various updates. 2022-03-14 18:14:53 -07:00
Liam 52895fab67 shader: add support for const buffer indirect addressing 2022-03-14 19:43:32 -04:00
ameerj 5119a57614 maxwell_3d: Implement a safer CB data upload
This makes constant buffer uploads safer and more accurate by updating the GPU memory as soon as the CB Data method is invoked. The previous implementation was deferring the updates until a different maxwell 3d method was detected, then writing all CB data at once.
2022-03-14 19:18:36 -04:00
Fernando S cd07a43724 Merge pull request #8008 from ameerj/rescale-offsets-array
rescaling_pass: Fix rescaling Color2DArray ImageFetch offsets
2022-03-15 00:08:22 +01:00
Fernando S f9e1f559b1 Merge pull request #8000 from liamwhite/hagi
Initial support for Wii Hagi emulator
2022-03-15 00:08:05 +01:00
bunnei cc285b9924 Merge pull request #8015 from FernandoS27/fix-global-mem
Shader decompiler: Fix storage tracking in deko3d.
2022-03-14 16:03:23 -07:00
byte[] be0e6a2bb4 Maxwell3D: Link to override constant definition in nouveau 2022-03-14 11:06:25 -04:00
Fernando S 0331b8d799 Merge pull request #8016 from merryhime/kill-mem-use
dynarmic: Reduce size of code caches
2022-03-14 16:04:46 +01:00
byte[] 364c67e49b Maxwell3D: restore original topology when topology overrides are disabled 2022-03-14 11:00:08 -04:00
Liam 37aa472269 Maxwell3D: Use override constants from nouveau
This fixes some incorrect rendering in Sunshine
2022-03-14 10:11:58 -04:00
Merry 220674d0d6 dynarmic: Reduce size of code caches 2022-03-13 22:17:14 +00:00
Fernando Sahmkow 185fc03c3c Shader decompiler: do constant propgation before texture pass. 2022-03-13 21:49:40 +01:00
Fernando Sahmkow ec9f0f064e Shader decompiler: Fix storage tracking in deko3d. 2022-03-13 17:41:16 +01:00
bunnei 8decc8d1a5 Merge pull request #8007 from ameerj/vs-2022-errors
emit_spirv, vk_compute_pass: Resolve VS2022 compiler errors
2022-03-13 03:43:06 -07:00
merry 1f6bbb6257 Merge pull request #8009 from ameerj/dynarmic-exclusives-config
config: Write dynarmic exclusive memory configs
2022-03-13 07:42:38 +00:00
ameerj 6b164a80a1 config: Write dynarmic exclusive memory configs
Ensures the configs are written and saved between boots
2022-03-12 03:42:50 -05:00
ameerj f87f8d4610 rescaling_pass: Fix rescaling Color2DArray ImageFetch offsets
ImageFetch offsets for 2D array coordinates have a different composite size than the coordinates. The rescaling pass was not taking this into account.

Fixes broken shaders when scaling is enabled in Astral Chain, and likely other titles.
2022-03-12 03:31:56 -05:00
ameerj e8c50e709e emit_spirv, vk_compute_pass: Resolve VS2022 compiler errors 2022-03-12 02:54:33 -05:00
BytesGalore 948f6e1112 build(cmake): fix missing Qt5::DBus link target for bundled linux package 2022-03-12 08:40:33 +01:00
bunnei 27cc7b6a73 Merge pull request #7997 from Wunkolo/cpu_detect_more
cpu_detect: Add additional x86 flags and telemetry
2022-03-11 17:26:41 -08:00
Liam 56c646d82c Maxwell3D: Restrict topology override effect to after the register is set 2022-03-11 19:42:12 -05:00
bunnei 5c74dd6462 Merge pull request #8003 from yuzu-emu/revert-7982-fix_cmake_missing_qt5_dbus
Revert "build(cmake): fix missing Qt5::DBus target on linux"
2022-03-11 15:22:30 -08:00
bunnei 15fdc2cd09 Revert "build(cmake): fix missing Qt5::DBus target on linux" 2022-03-11 15:22:24 -08:00
Wunkolo d248c1203e cpu_detect: Add additional x86 flags and telemetry
Adds detection of additional CPU flags to cpu_detect and additions to telemetry output.

This is not exhaustive but guided by features that [dynarmic utilizes](https://github.com/merryhime/dynarmic/blob/bcfe377aaa5138af740e90af5be7a7dff7b62a52/src/dynarmic/backend/x64/host_feature.h#L12-L33) as well as features that are currently utilized but not reported to telemetry(invariant_tsc). This is intended to guide future optimizations.

AVX512 in particular is broken up into its individual subsets and some other processor features such as [sha](https://en.wikipedia.org/wiki/Intel_SHA_extensions) and [gfni](https://en.wikipedia.org/wiki/AVX-512#GFNI) are added to have some forward-facing data-points.

What used to be a single `CPU_Extension_x64_AVX512` telemetry field
is also broken up into individual `CPU_Extension_x64_AVX512{F,VL,CD,...}` fields.
2022-03-11 10:27:00 -08:00
Wunkolo 29a7a61806 common/telemetry: Update AddField name type to string_view
Non-owning `string_view` is flexable and
avoids some of the many redundant copies made over `std::string`
2022-03-11 10:26:59 -08:00
Liam 70e632f153 Maxwell3D: mark index buffers as dirty after updating counts 2022-03-11 08:51:22 -05:00
bunnei 8180b262fc Merge pull request #7982 from BytesGalore/fix_cmake_missing_qt5_dbus
build(cmake): fix missing Qt5::DBus target on linux
2022-03-10 23:12:33 -08:00
Liam 82c3042c0f TextureCacheRuntime: allow converting D24S8 to ABGR8
I can't see how this would be useful, but Galaxy uses it.
2022-03-10 20:25:34 -05:00
Liam f1521183f8 Maxwell3D: read small-index draw and primitive topology override registers
This allows Galaxy and Sunshine to render for the first time.
2022-03-10 19:21:04 -05:00
Mai M e200161982 Merge pull request #7999 from merryhime/fix-7992
backend: Ensure backend_thread is destructed before message_queue
2022-03-10 08:07:41 -05:00
Merry 22f50c6bc1 backend: Ensure backend_thread is destructed before message_queue
Ensures that stop_token signals that stop has been requested before destruction of conditional_variable
2022-03-10 10:49:15 +00:00
Morph 52f8f00434 Merge pull request #7998 from Wunkolo/cpuid_array
cpu_detect: Revert `__cpuid{ex}` array-type argument
2022-03-10 00:09:36 -05:00
Wunkolo d9b1199ffb cpu_detect: Revert __cpuid{ex} array-type argument
Restores compatibility with MSVC's `__cpuid` intrinsic.
2022-03-09 19:50:01 -08:00
bunnei 9a97ef4647 Merge pull request #7936 from Wunkolo/cpu_detect
cpu_detect: Refactor detection of processor features
2022-03-09 15:34:42 -08:00
Wunkolo 873a9fa7e5 cpu_detect: Add missing lzcnt detection 2022-03-09 13:57:47 -08:00
Wunkolo ec5f3351b6 cpu_detect: Refactor cpu/manufacturer identification
Set the zero-enum value to Unknown
Move the Manufacterer enum into the CPUCaps structure namespace
Add "ParseManufacturer" utility-function
Fix cpu/brand string buffer sizes(!)
2022-03-09 13:57:47 -08:00
Wunkolo 86e9e60f07 cpu_detect: Update array-types to span and array
Update some uses of `int` into some more explicitly sized types as well
2022-03-09 13:57:47 -08:00
Wunkolo 3c33ba7f18 cpu_detect: Utilize Bit<N> utility function 2022-03-09 13:57:47 -08:00
Wunkolo d233de8194 cpu_detect: Compact capability fields
As this structure gets more explicit, bools can be bitfields and
small enums can use smaller types for their span of values.
2022-03-09 13:57:47 -08:00
Wunkolo add2cfcb96 bit_util: Add bit utility function
Extracts a singular bit, as a bool, from the specified compile-time index.
2022-03-09 13:57:47 -08:00
bunnei 6f670381cf Merge pull request #7975 from bunnei/ldr-fix
hle: service: ldr: Use deterministic addresses when mapping NROs.
2022-03-08 17:39:03 -08:00
bunnei 853e58e593 hle: service: ldr: Use deterministic addresses when mapping NROs.
- Instead of randomization, choose in-order addresses for where to map NROs into memory.
- This results in predictable behavior when debugging and consistent behavior when reproducing issues.
2022-03-08 17:38:20 -08:00
bunnei f2743b41b0 Merge pull request #7986 from lat9nq/vk-callback
core, video_core: Fix two crashes when failing to create the emulated GPU instance
2022-03-08 12:36:57 -08:00
Fernando S 35309f27ed Merge pull request #7989 from degasus/maxwell_LUT3
shader_recompiler/LOP3: Use brute force python results within switch/case.
2022-03-08 15:40:31 +01:00
Markus Wick c78c8190d5 shader_recompiler/LOP3: Use brute force python results within switch/case.
Thanks to @asLody for optimizing this function. This raised the focus that this function should be optimized more.

The current table assumes that the host GPU is able to invert for free, so only AND,OR,XOR are accumulated in the performance metrik.

Performance results:

Instructions
0: 8
1: 30
2: 114
3: 80
4: 24

Latency
0: 8
1: 30
2: 194
3: 24
2022-03-08 09:44:28 +01:00
bunnei 1f37896066 Merge pull request #7974 from bunnei/improve-code-mem
Kernel Memory Updates (Part 5): Revamp MapCodeMemory and UnmapCodeMemory.
2022-03-07 20:28:39 -08:00
bunnei 749f76e6fe hle: kernel: KPageTable: Improve implementations of MapCodeMemory and UnmapCodeMemory.
- This makes these functions more accurate to the real HOS implementations.
- Fixes memory access issues in Super Smash Bros. Ultimate that occur when un/mapping NROs.
2022-03-07 17:18:20 -08:00
lat9nq b5e60ae1b0 video_core: Cancel Scoped's exit call on GPU failure
When CreateRenderer fails, the GraphicsContext that was std::move'd into
it is destroyed before the Scoped that was created to manage its
currency. In that case, the GraphicsContext::Scoped will still call its
destructor at the ending of the function. And because the context is
destroyed, the Scoped will cause a crash as it attempts to call a
destroyed object's DoneCurrent function.

Since we know when the call would be invalid, call the Scoped's Cancel
method. This prevents it from calling a method on a destroyed object.
2022-03-07 18:21:56 -05:00
lat9nq 1f24a4e520 emu_window: Create a way to Cancel the exit of a Scoped
If a GraphicsContext is destroyed before its Scoped is destroyed, this
causes a crash as the Scoped tries to call a method in the destroyed
context on exit.

Add a way to Cancel the call when we know that calling the
GraphicsContext will not work.
2022-03-07 18:21:56 -05:00
Fernando S 58b52f4884 Merge pull request #7930 from asLody/dma-semaphore
MaxwellDMA: Implement semaphore operations
2022-03-07 21:53:38 +01:00
lat9nq 381f1dd2c9 core: Don't shutdown a null GPU
When CreateGPU fails, yuzu would try and shutdown the GPU instance
regardless of whether any instance was actually created.

Check for nullptr before calling its methods to prevent a crash.
2022-03-07 15:25:20 -05:00
Lody 4498908e72 MaxwellDMA: Implement semaphore operations 2022-03-07 13:46:18 +08:00
Ameer J 370e480c8c gl_graphics_pipeline: Improve shader builder synchronization using fences (#7969)
* gl_graphics_pipeline: Improve shader builder synchronization

Make use of GLsync objects to ensure better synchronization between shader builder threads and the main context

* gl_graphics_pipeline: Make built_fence access threadsafe

* gl_graphics_pipeline: Use GLsync objects only when building in parallel

* gl_graphics_pipeline: Replace GetSync calls with non-blocking waits

The spec states that a ClientWait on a Fence object ensures the changes propagate to the calling context
2022-03-06 16:46:49 +01:00
BytesGalore fc84649aab build(cmake): fix missing Qt5::DBus link target 2022-03-06 12:21:46 +01:00
Fernando S 5192c64991 Merge pull request #7973 from Morph1984/debug-crash
host_memory: Fix fastmem crashes in debug builds
2022-03-06 04:49:27 +01:00
bunnei a31c195749 Merge pull request #7935 from Wunkolo/logging-join-fix
logging: Convert `backend_thread` into an `std::jthread`
2022-03-02 19:09:26 -08:00
bunnei 3ab82e7582 Merge pull request #7956 from bunnei/improve-mem-manager
Kernel Memory Updates (Part 4): Revamp KMemoryManager & other fixes
2022-03-02 17:55:51 -08:00
Morph b33f23cc46 host_memory: Fix fastmem crashes in debug builds
It is possible for virtual_offset to not be 0 when the iterator is at the beginning, and thus, std::prev(it) may be evaluated, leading to a crash in debug mode.

Co-Authored-By: Fernando S. <1731197+FernandoS27@users.noreply.github.com>
2022-03-02 18:36:59 -05:00
Fernando S e06a133717 Merge pull request #7959 from merryhime/cmpxchg
dynarmic: Inline exclusive memory accesses
2022-03-01 22:50:52 +01:00
Mai M 3c47570563 Merge pull request #7967 from zhaobot/tx-update-20220301023432
Update translations (2022-03-01)
2022-03-01 00:50:28 -05:00
The yuzu Community b4c919eba8 Update translations (2022-03-01) 2022-03-01 02:34:54 +00:00
german77 03d671fabc applet: mii: Simple implementation of mii applet 2022-02-28 18:53:41 -06:00
Mai M 96dbb9233a Merge pull request #7963 from ameerj/gl_fence_nit
gl_fence_manager: Minor optimization to signal querying
2022-02-28 19:53:24 -05:00
Mai M fdf81cd112 Merge pull request #7966 from merryhime/cmake-ffmpeg-endif
ffmpeg: Fix mis-maching argument warning for endif
2022-02-28 19:51:10 -05:00
Merry c4df0d8cdc ffmpeg: Fix mis-maching argument warning for endif 2022-02-28 21:08:12 +00:00
merry ec9689f200 dynarmic: Update to latest master 2022-02-28 20:10:13 +00:00
bunnei 14d28a043d hle: kernel: Re-create memory layout at initialization.
- As this can only be derived once.
2022-02-27 18:00:09 -08:00
bunnei 16e5954fcb hle: kernel: Remove unused pool locals. 2022-02-27 18:00:09 -08:00
bunnei f87f076162 hle: kernel: k_memory_manager: Rework for latest kernel behavior.
- Updates the KMemoryManager implementation against latest documentation.
- Reworks KMemoryLayout to be accessed throughout the kernel.
- Fixes an issue with pool sizes being incorrectly reported.
2022-02-27 18:00:09 -08:00
Wunkolo 913c2bd2cb logging: Convert backend_thread into an std::jthread
Was getting an unhandled `invalid_argument` [exception](https://en.cppreference.com/w/cpp/thread/thread/join) during
shutdown on my linux machine. This removes the need for a `StopBackendThread` function entirely since `jthread`
[automatically handles both checking if the thread is joinable and stopping the token before attempting to join](https://en.cppreference.com/w/cpp/thread/jthread/~jthread) in the case that `StartBackendThread` was never called.
2022-02-27 16:23:52 -08:00
ameerj 7f7df43da2 gl_fence_manager: Minor optimization to signal querying
Per the spec, bufSize is the number of integers that will be written, in this case, 1.

Also, the length argument is optional if the information of the number of elements written is not needed.
2022-02-27 17:57:33 -05:00
merry 16784e5bb3 dynarmic: Inline exclusive memory accesses
Inlines implementation of exclusive instructions into JITted code,
improving performance of applications relying heavily on these
instructions.

We also fastmem these instructions for additional speed, with
support for appropriate recompilation on fastmem failure.

An unsafe optimization to disable the intercore global_monitor is also
provided, should one wish to rely solely on cmpxchg semantics for
safety.

See also: merryhime/dynarmic#664
2022-02-27 19:40:05 +00:00
bunnei adbb9c2b00 hle: kernel: k_page_heap: GetPhysicalAddr can be const. 2022-02-27 10:34:02 -08:00
bunnei f7e65eb971 hle: kernel: k_page_heap: Remove superfluous consexpr. 2022-02-27 10:34:02 -08:00
bunnei 06e2b76c75 hle: kernel: k_page_heap: Various updates and improvements.
- KPageHeap tracks physical addresses, not virtual addresses.
- Various updates and improvements to match latest documentation for this type.
2022-02-27 10:34:02 -08:00
bunnei 5d1a81520c hle: kernel: Add initial_process.h header. 2022-02-27 10:34:02 -08:00
bunnei a6496deeed hle: kernel: board: nx: Add k_memory_layout.h header. 2022-02-27 10:34:02 -08:00
bunnei 9b5e7971dc hle: kernel: k_system_control: Add GetRealMemorySize and update GetKernelPhysicalBaseAddress. 2022-02-27 10:34:02 -08:00
bunnei 18e77a54c3 hle: kernel: k_memory_layout: Add GetPhysicalLinearRegion. 2022-02-27 10:34:02 -08:00
bunnei 06a21ac229 hle: kernel: k_memory_region_types: Update for new regions. 2022-02-27 10:34:02 -08:00
bunnei 96d90be59f Merge pull request #7955 from bunnei/update-dynarmic
externals: Dynarmic: Update to latest rev.
2022-02-26 10:52:47 -08:00
bunnei ce46aa4721 externals: Dynarmic: Update to latest rev.
- Fixes inaccurate size reporting in SpaceRemaining, which caused crashes in yuzu with SSBU.
2022-02-26 09:51:41 -08:00
bunnei 20e9501b0d Merge pull request #7932 from bunnei/extended-mem-layout
Add extended memory layout (6GB) support and improve KResourceLimit management
2022-02-26 01:41:08 -08:00
bunnei 56d9052b73 Merge pull request #7953 from ameerj/radv-rdna2-crash
vulkan_device: Blacklist RADV on RDNA2 from VK_EXT_vertex_input_dynamic_state
2022-02-26 01:19:13 -08:00
Mai M 91d802f68d Merge pull request #7948 from Morph1984/11-11-10-float
maxwell_to_(gl/vk): Add 11_11_10 float vertex format
2022-02-26 00:09:44 -05:00
Ameer J 5f8d6bc504 vulkan_device: Blacklist RADV on RDNA2 from VK_EXT_vertex_input_dynamic_state
RDNA2 devices running under the RADV driver were crashing when VK_EXT_vertex_input_dynamic_state was enabled.

Blacklisting these devices until a proper fix is established.
2022-02-25 23:09:03 -05:00
Morph e292b2d991 maxwell_to_(gl/vk): Add 11_11_10 float vertex format
- Used by パワプロクンポケットR
2022-02-25 17:11:17 -05:00
bunnei 6c9b9b72da Merge pull request #7939 from asLody/fb-format-gbra8
vk_blit_screen: Add missing framebuffer format
2022-02-25 00:58:28 -08:00
bunnei f582a4482d Merge pull request #7927 from german77/amiibo
yuzu: Disconnect amiibos on drag and drop
2022-02-24 23:13:32 -08:00
Lody 6978093404 vk_blit_screen: Add missing format bgra8 2022-02-24 19:25:39 +08:00
bunnei 71f62a346d hle: kernel: KSystemControl: Use 6GB memory layout when "use_extended_memory_layout" setting is enabled.
- This uses a larger 6GB DRAM memory layout, which is useful for some mods that require more memory.
2022-02-21 13:07:19 -08:00
bunnei c0e45a3c78 core: device_memory: Use memory size reported by KSystemControl.
- That way, we can consolidate the memory layout to one place.
2022-02-21 13:07:19 -08:00
bunnei 70482e6b26 settings: Add a new "use_extended_memory_layout" setting.
- This will be used to enable emulation of a larger memory arrangement.
2022-02-21 13:07:19 -08:00
bunnei ca5e843bf6 core: hle: kernel: Remove resource limit hack for PhysicalMemory.
- With prior changes, we now report the correct amount of physical memory available to the emulated process.
2022-02-21 12:41:31 -08:00
bunnei a74fddc98f core: hle: kernel: KProcess: Pass in KResourceLimit on process creation.
- This allows us to have a resource limit per process, rather than use the global system resource limit.
2022-02-21 12:41:31 -08:00
bunnei 57ebcbf2c4 core: hle: kernel: KEvent: Pass in owner KProcess on event creation.
- This is necessary to ensure resource limits are freed from the right process.
2022-02-21 12:41:06 -08:00
bunnei c7019db6f4 core: hle: kernel: KResourceLimit: Add a helper function for creating a KResourceLimit for a process. 2022-02-21 12:40:09 -08:00
german77 b504df9d02 yuzu: Remove amiibos on drag and drop 2022-02-20 14:28:21 -06:00
Lioncash e015dc8264 texture_cache: Ensure has_blacklisted is always initialized
Resolves a -Wmaybe_uninitialized warning
2022-02-02 14:37:27 -05:00
Lioncash 7367e55d1d texture_cache: Remove dead code within SynchronizeAliases
Since these were being copied by value, none of the changes applied in
the loop would be reflected.

However, from the looks of it, this would already be applied within
CopyImage() anyways, so this can be removed.
2022-02-02 14:37:22 -05:00
Lioncash 856f576c05 texture_cache: Amend unintended bitwise OR in SynchronizeAliases 2022-02-02 14:20:58 -05:00
389 changed files with 20172 additions and 12624 deletions
+1 -1
View File
@@ -8,7 +8,7 @@ steps:
displayName: 'Install vulkan-sdk'
- script: python -m pip install --upgrade pip conan
displayName: 'Install conan'
- script: refreshenv && mkdir build && cd build && cmake -G "Visual Studio 16 2019" -A x64 -DYUZU_USE_BUNDLED_QT=1 -DYUZU_USE_BUNDLED_SDL2=1 -DYUZU_USE_QT_WEB_ENGINE=ON -DENABLE_COMPATIBILITY_LIST_DOWNLOAD=ON -DYUZU_ENABLE_COMPATIBILITY_REPORTING=${COMPAT} -DUSE_DISCORD_PRESENCE=ON -DENABLE_QT_TRANSLATION=ON -DDISPLAY_VERSION=${{ parameters['version'] }} -DCMAKE_BUILD_TYPE=Release .. && cd ..
- script: refreshenv && mkdir build && cd build && cmake -G "Visual Studio 16 2019" -A x64 -DYUZU_USE_BUNDLED_QT=1 -DYUZU_USE_BUNDLED_SDL2=1 -DYUZU_USE_QT_WEB_ENGINE=ON -DENABLE_COMPATIBILITY_LIST_DOWNLOAD=ON -DYUZU_ENABLE_COMPATIBILITY_REPORTING=${COMPAT} -DYUZU_TESTS=OFF -DUSE_DISCORD_PRESENCE=ON -DENABLE_QT_TRANSLATION=ON -DDISPLAY_VERSION=${{ parameters['version'] }} -DCMAKE_BUILD_TYPE=Release .. && cd ..
displayName: 'Configure CMake'
- task: MSBuild@1
displayName: 'Build'
+5 -1
View File
@@ -363,7 +363,11 @@ if(ENABLE_QT)
set(YUZU_QT_NO_CMAKE_SYSTEM_PATH "NO_CMAKE_SYSTEM_PATH")
endif()
find_package(Qt5 ${QT_VERSION} REQUIRED COMPONENTS Widgets ${QT_PREFIX_HINT} ${YUZU_QT_NO_CMAKE_SYSTEM_PATH})
if ((${CMAKE_SYSTEM_NAME} STREQUAL "Linux") AND YUZU_USE_BUNDLED_QT)
find_package(Qt5 ${QT_VERSION} REQUIRED COMPONENTS Widgets DBus ${QT_PREFIX_HINT} ${YUZU_QT_NO_CMAKE_SYSTEM_PATH})
else()
find_package(Qt5 ${QT_VERSION} REQUIRED COMPONENTS Widgets ${QT_PREFIX_HINT} ${YUZU_QT_NO_CMAKE_SYSTEM_PATH})
endif()
if (YUZU_USE_QT_WEB_ENGINE)
find_package(Qt5 COMPONENTS WebEngineCore WebEngineWidgets)
endif()
+423 -454
View File
File diff suppressed because it is too large Load Diff
+404 -435
View File
File diff suppressed because it is too large Load Diff
+403 -434
View File
File diff suppressed because it is too large Load Diff
+404 -435
View File
File diff suppressed because it is too large Load Diff
+404 -435
View File
File diff suppressed because it is too large Load Diff
+403 -434
View File
File diff suppressed because it is too large Load Diff
+405 -436
View File
File diff suppressed because it is too large Load Diff
+426 -457
View File
File diff suppressed because it is too large Load Diff
+404 -435
View File
File diff suppressed because it is too large Load Diff
+490 -507
View File
File diff suppressed because it is too large Load Diff
+408 -439
View File
File diff suppressed because it is too large Load Diff
+403 -434
View File
File diff suppressed because it is too large Load Diff
+403 -434
View File
File diff suppressed because it is too large Load Diff
+472 -497
View File
File diff suppressed because it is too large Load Diff
+408 -439
View File
File diff suppressed because it is too large Load Diff
+655 -661
View File
File diff suppressed because it is too large Load Diff
+404 -435
View File
File diff suppressed because it is too large Load Diff
+403 -434
View File
File diff suppressed because it is too large Load Diff
+404 -435
View File
File diff suppressed because it is too large Load Diff
+6202
View File
File diff suppressed because it is too large Load Diff
+405 -436
View File
File diff suppressed because it is too large Load Diff
+404 -435
View File
File diff suppressed because it is too large Load Diff
+406 -437
View File
File diff suppressed because it is too large Load Diff
+1 -1
View File
@@ -214,6 +214,6 @@ else(WIN32)
set(FFmpeg_LDFLAGS "${FFmpeg_LDFLAGS}" PARENT_SCOPE)
set(FFmpeg_LIBRARIES "${FFmpeg_LIBRARIES}" PARENT_SCOPE)
set(FFmpeg_INCLUDE_DIR "${FFmpeg_INCLUDE_DIR}" PARENT_SCOPE)
endif(WIN32)
endif()
unset(FFmpeg_COMPONENTS)
+2 -3
View File
@@ -4,13 +4,12 @@
#pragma once
#include <cstring>
#include <memory>
#include "common/common_types.h"
#if _MSC_VER
#include <intrin.h>
#else
#include <cstring>
#endif
namespace Common {
-1
View File
@@ -33,7 +33,6 @@
#include <cstddef>
#include <limits>
#include <type_traits>
#include "common/common_funcs.h"
#include "common/swap.h"
/*
+7
View File
@@ -57,4 +57,11 @@ requires std::is_integral_v<T>
return static_cast<T>(1ULL << ((8U * sizeof(T)) - std::countl_zero(value - 1U)));
}
template <size_t bit_index, typename T>
requires std::is_integral_v<T>
[[nodiscard]] constexpr bool Bit(const T value) {
static_assert(bit_index < BitSize<T>(), "bit_index must be smaller than size of T");
return ((value >> bit_index) & T(1)) == T(1);
}
} // namespace Common
-1
View File
@@ -2,7 +2,6 @@
// Licensed under GPLv2+
// Refer to the license.txt file included.
#include <cstring>
#include <string>
#include <utility>
-1
View File
@@ -4,7 +4,6 @@
#include "common/fs/file.h"
#include "common/fs/fs.h"
#include "common/fs/path_util.h"
#include "common/logging/log.h"
#ifdef _WIN32
-2
View File
@@ -6,10 +6,8 @@
#include <cstdio>
#include <filesystem>
#include <fstream>
#include <span>
#include <type_traits>
#include <vector>
#include "common/concepts.h"
#include "common/fs/fs_types.h"
-1
View File
@@ -7,7 +7,6 @@
#include <functional>
#include "common/common_funcs.h"
#include "common/common_types.h"
namespace Common::FS {
-1
View File
@@ -8,7 +8,6 @@
#include <filesystem>
#include <span>
#include <string>
#include <string_view>
#include "common/common_types.h"
-1
View File
@@ -7,7 +7,6 @@
#include <array>
#include <cstddef>
#include <string>
#include <type_traits>
#include <vector>
#include <fmt/format.h>
#include "common/common_types.h"
+3 -3
View File
@@ -18,6 +18,7 @@
#include <fcntl.h>
#include <sys/mman.h>
#include <unistd.h>
#include "common/scope_exit.h"
#endif // ^^^ Linux ^^^
@@ -27,7 +28,6 @@
#include "common/assert.h"
#include "common/host_memory.h"
#include "common/logging/log.h"
#include "common/scope_exit.h"
namespace Common {
@@ -327,8 +327,8 @@ private:
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 it == placeholders.begin() ? virtual_offset == 0
: std::prev(it)->upper() == virtual_offset;
}
return false;
}
+211 -183
View File
@@ -4,6 +4,7 @@
#pragma once
#include "common/common_funcs.h"
#include "common/parent_of_member.h"
#include "common/tree.h"
@@ -15,32 +16,33 @@ class IntrusiveRedBlackTreeImpl;
}
#pragma pack(push, 4)
struct IntrusiveRedBlackTreeNode {
YUZU_NON_COPYABLE(IntrusiveRedBlackTreeNode);
public:
using EntryType = RBEntry<IntrusiveRedBlackTreeNode>;
constexpr IntrusiveRedBlackTreeNode() = default;
void SetEntry(const EntryType& new_entry) {
entry = new_entry;
}
[[nodiscard]] EntryType& GetEntry() {
return entry;
}
[[nodiscard]] const EntryType& GetEntry() const {
return entry;
}
using RBEntry = freebsd::RBEntry<IntrusiveRedBlackTreeNode>;
private:
EntryType entry{};
RBEntry m_entry;
friend class impl::IntrusiveRedBlackTreeImpl;
public:
explicit IntrusiveRedBlackTreeNode() = default;
template <class, class, class>
friend class IntrusiveRedBlackTree;
[[nodiscard]] constexpr RBEntry& GetRBEntry() {
return m_entry;
}
[[nodiscard]] constexpr const RBEntry& GetRBEntry() const {
return m_entry;
}
constexpr void SetRBEntry(const RBEntry& entry) {
m_entry = entry;
}
};
static_assert(sizeof(IntrusiveRedBlackTreeNode) ==
3 * sizeof(void*) + std::max<size_t>(sizeof(freebsd::RBColor), 4));
#pragma pack(pop)
template <class T, class Traits, class Comparator>
class IntrusiveRedBlackTree;
@@ -48,12 +50,17 @@ class IntrusiveRedBlackTree;
namespace impl {
class IntrusiveRedBlackTreeImpl {
YUZU_NON_COPYABLE(IntrusiveRedBlackTreeImpl);
private:
template <class, class, class>
friend class ::Common::IntrusiveRedBlackTree;
using RootType = RBHead<IntrusiveRedBlackTreeNode>;
RootType root;
private:
using RootType = freebsd::RBHead<IntrusiveRedBlackTreeNode>;
private:
RootType m_root;
public:
template <bool Const>
@@ -81,149 +88,150 @@ public:
IntrusiveRedBlackTreeImpl::reference>;
private:
pointer node;
pointer m_node;
public:
explicit Iterator(pointer n) : node(n) {}
constexpr explicit Iterator(pointer n) : m_node(n) {}
bool operator==(const Iterator& rhs) const {
return this->node == rhs.node;
constexpr bool operator==(const Iterator& rhs) const {
return m_node == rhs.m_node;
}
bool operator!=(const Iterator& rhs) const {
constexpr bool operator!=(const Iterator& rhs) const {
return !(*this == rhs);
}
pointer operator->() const {
return this->node;
constexpr pointer operator->() const {
return m_node;
}
reference operator*() const {
return *this->node;
constexpr reference operator*() const {
return *m_node;
}
Iterator& operator++() {
this->node = GetNext(this->node);
constexpr Iterator& operator++() {
m_node = GetNext(m_node);
return *this;
}
Iterator& operator--() {
this->node = GetPrev(this->node);
constexpr Iterator& operator--() {
m_node = GetPrev(m_node);
return *this;
}
Iterator operator++(int) {
constexpr Iterator operator++(int) {
const Iterator it{*this};
++(*this);
return it;
}
Iterator operator--(int) {
constexpr Iterator operator--(int) {
const Iterator it{*this};
--(*this);
return it;
}
operator Iterator<true>() const {
return Iterator<true>(this->node);
constexpr operator Iterator<true>() const {
return Iterator<true>(m_node);
}
};
private:
// Define accessors using RB_* functions.
bool EmptyImpl() const {
return root.IsEmpty();
constexpr bool EmptyImpl() const {
return m_root.IsEmpty();
}
IntrusiveRedBlackTreeNode* GetMinImpl() const {
return RB_MIN(const_cast<RootType*>(&root));
constexpr IntrusiveRedBlackTreeNode* GetMinImpl() const {
return freebsd::RB_MIN(const_cast<RootType&>(m_root));
}
IntrusiveRedBlackTreeNode* GetMaxImpl() const {
return RB_MAX(const_cast<RootType*>(&root));
constexpr IntrusiveRedBlackTreeNode* GetMaxImpl() const {
return freebsd::RB_MAX(const_cast<RootType&>(m_root));
}
IntrusiveRedBlackTreeNode* RemoveImpl(IntrusiveRedBlackTreeNode* node) {
return RB_REMOVE(&root, node);
constexpr IntrusiveRedBlackTreeNode* RemoveImpl(IntrusiveRedBlackTreeNode* node) {
return freebsd::RB_REMOVE(m_root, node);
}
public:
static IntrusiveRedBlackTreeNode* GetNext(IntrusiveRedBlackTreeNode* node) {
return RB_NEXT(node);
static constexpr IntrusiveRedBlackTreeNode* GetNext(IntrusiveRedBlackTreeNode* node) {
return freebsd::RB_NEXT(node);
}
static IntrusiveRedBlackTreeNode* GetPrev(IntrusiveRedBlackTreeNode* node) {
return RB_PREV(node);
static constexpr IntrusiveRedBlackTreeNode* GetPrev(IntrusiveRedBlackTreeNode* node) {
return freebsd::RB_PREV(node);
}
static const IntrusiveRedBlackTreeNode* GetNext(const IntrusiveRedBlackTreeNode* node) {
static constexpr IntrusiveRedBlackTreeNode const* GetNext(
IntrusiveRedBlackTreeNode const* node) {
return static_cast<const IntrusiveRedBlackTreeNode*>(
GetNext(const_cast<IntrusiveRedBlackTreeNode*>(node)));
}
static const IntrusiveRedBlackTreeNode* GetPrev(const IntrusiveRedBlackTreeNode* node) {
static constexpr IntrusiveRedBlackTreeNode const* GetPrev(
IntrusiveRedBlackTreeNode const* node) {
return static_cast<const IntrusiveRedBlackTreeNode*>(
GetPrev(const_cast<IntrusiveRedBlackTreeNode*>(node)));
}
public:
constexpr IntrusiveRedBlackTreeImpl() {}
constexpr IntrusiveRedBlackTreeImpl() = default;
// Iterator accessors.
iterator begin() {
constexpr iterator begin() {
return iterator(this->GetMinImpl());
}
const_iterator begin() const {
constexpr const_iterator begin() const {
return const_iterator(this->GetMinImpl());
}
iterator end() {
constexpr iterator end() {
return iterator(static_cast<IntrusiveRedBlackTreeNode*>(nullptr));
}
const_iterator end() const {
constexpr const_iterator end() const {
return const_iterator(static_cast<const IntrusiveRedBlackTreeNode*>(nullptr));
}
const_iterator cbegin() const {
constexpr const_iterator cbegin() const {
return this->begin();
}
const_iterator cend() const {
constexpr const_iterator cend() const {
return this->end();
}
iterator iterator_to(reference ref) {
return iterator(&ref);
constexpr iterator iterator_to(reference ref) {
return iterator(std::addressof(ref));
}
const_iterator iterator_to(const_reference ref) const {
return const_iterator(&ref);
constexpr const_iterator iterator_to(const_reference ref) const {
return const_iterator(std::addressof(ref));
}
// Content management.
bool empty() const {
constexpr bool empty() const {
return this->EmptyImpl();
}
reference back() {
constexpr reference back() {
return *this->GetMaxImpl();
}
const_reference back() const {
constexpr const_reference back() const {
return *this->GetMaxImpl();
}
reference front() {
constexpr reference front() {
return *this->GetMinImpl();
}
const_reference front() const {
constexpr const_reference front() const {
return *this->GetMinImpl();
}
iterator erase(iterator it) {
constexpr iterator erase(iterator it) {
auto cur = std::addressof(*it);
auto next = GetNext(cur);
this->RemoveImpl(cur);
@@ -234,16 +242,16 @@ public:
} // namespace impl
template <typename T>
concept HasLightCompareType = requires {
{ std::is_same<typename T::LightCompareType, void>::value } -> std::convertible_to<bool>;
concept HasRedBlackKeyType = requires {
{ std::is_same<typename T::RedBlackKeyType, void>::value } -> std::convertible_to<bool>;
};
namespace impl {
template <typename T, typename Default>
consteval auto* GetLightCompareType() {
if constexpr (HasLightCompareType<T>) {
return static_cast<typename T::LightCompareType*>(nullptr);
consteval auto* GetRedBlackKeyType() {
if constexpr (HasRedBlackKeyType<T>) {
return static_cast<typename T::RedBlackKeyType*>(nullptr);
} else {
return static_cast<Default*>(nullptr);
}
@@ -252,16 +260,17 @@ namespace impl {
} // namespace impl
template <typename T, typename Default>
using LightCompareType = std::remove_pointer_t<decltype(impl::GetLightCompareType<T, Default>())>;
using RedBlackKeyType = std::remove_pointer_t<decltype(impl::GetRedBlackKeyType<T, Default>())>;
template <class T, class Traits, class Comparator>
class IntrusiveRedBlackTree {
YUZU_NON_COPYABLE(IntrusiveRedBlackTree);
public:
using ImplType = impl::IntrusiveRedBlackTreeImpl;
private:
ImplType impl{};
ImplType m_impl;
public:
template <bool Const>
@@ -277,9 +286,9 @@ public:
using iterator = Iterator<false>;
using const_iterator = Iterator<true>;
using light_value_type = LightCompareType<Comparator, value_type>;
using const_light_pointer = const light_value_type*;
using const_light_reference = const light_value_type&;
using key_type = RedBlackKeyType<Comparator, value_type>;
using const_key_pointer = const key_type*;
using const_key_reference = const key_type&;
template <bool Const>
class Iterator {
@@ -298,183 +307,201 @@ public:
IntrusiveRedBlackTree::reference>;
private:
ImplIterator iterator;
ImplIterator m_impl;
private:
explicit Iterator(ImplIterator it) : iterator(it) {}
constexpr explicit Iterator(ImplIterator it) : m_impl(it) {}
explicit Iterator(typename std::conditional<Const, ImplType::const_iterator,
ImplType::iterator>::type::pointer ptr)
: iterator(ptr) {}
constexpr explicit Iterator(typename ImplIterator::pointer p) : m_impl(p) {}
ImplIterator GetImplIterator() const {
return this->iterator;
constexpr ImplIterator GetImplIterator() const {
return m_impl;
}
public:
bool operator==(const Iterator& rhs) const {
return this->iterator == rhs.iterator;
constexpr bool operator==(const Iterator& rhs) const {
return m_impl == rhs.m_impl;
}
bool operator!=(const Iterator& rhs) const {
constexpr bool operator!=(const Iterator& rhs) const {
return !(*this == rhs);
}
pointer operator->() const {
return Traits::GetParent(std::addressof(*this->iterator));
constexpr pointer operator->() const {
return Traits::GetParent(std::addressof(*m_impl));
}
reference operator*() const {
return *Traits::GetParent(std::addressof(*this->iterator));
constexpr reference operator*() const {
return *Traits::GetParent(std::addressof(*m_impl));
}
Iterator& operator++() {
++this->iterator;
constexpr Iterator& operator++() {
++m_impl;
return *this;
}
Iterator& operator--() {
--this->iterator;
constexpr Iterator& operator--() {
--m_impl;
return *this;
}
Iterator operator++(int) {
constexpr Iterator operator++(int) {
const Iterator it{*this};
++this->iterator;
++m_impl;
return it;
}
Iterator operator--(int) {
constexpr Iterator operator--(int) {
const Iterator it{*this};
--this->iterator;
--m_impl;
return it;
}
operator Iterator<true>() const {
return Iterator<true>(this->iterator);
constexpr operator Iterator<true>() const {
return Iterator<true>(m_impl);
}
};
private:
static int CompareImpl(const IntrusiveRedBlackTreeNode* lhs,
const IntrusiveRedBlackTreeNode* rhs) {
static constexpr int CompareImpl(const IntrusiveRedBlackTreeNode* lhs,
const IntrusiveRedBlackTreeNode* rhs) {
return Comparator::Compare(*Traits::GetParent(lhs), *Traits::GetParent(rhs));
}
static int LightCompareImpl(const void* elm, const IntrusiveRedBlackTreeNode* rhs) {
return Comparator::Compare(*static_cast<const_light_pointer>(elm), *Traits::GetParent(rhs));
static constexpr int CompareKeyImpl(const_key_reference key,
const IntrusiveRedBlackTreeNode* rhs) {
return Comparator::Compare(key, *Traits::GetParent(rhs));
}
// Define accessors using RB_* functions.
IntrusiveRedBlackTreeNode* InsertImpl(IntrusiveRedBlackTreeNode* node) {
return RB_INSERT(&impl.root, node, CompareImpl);
constexpr IntrusiveRedBlackTreeNode* InsertImpl(IntrusiveRedBlackTreeNode* node) {
return freebsd::RB_INSERT(m_impl.m_root, node, CompareImpl);
}
IntrusiveRedBlackTreeNode* FindImpl(const IntrusiveRedBlackTreeNode* node) const {
return RB_FIND(const_cast<ImplType::RootType*>(&impl.root),
const_cast<IntrusiveRedBlackTreeNode*>(node), CompareImpl);
constexpr IntrusiveRedBlackTreeNode* FindImpl(IntrusiveRedBlackTreeNode const* node) const {
return freebsd::RB_FIND(const_cast<ImplType::RootType&>(m_impl.m_root),
const_cast<IntrusiveRedBlackTreeNode*>(node), CompareImpl);
}
IntrusiveRedBlackTreeNode* NFindImpl(const IntrusiveRedBlackTreeNode* node) const {
return RB_NFIND(const_cast<ImplType::RootType*>(&impl.root),
const_cast<IntrusiveRedBlackTreeNode*>(node), CompareImpl);
constexpr IntrusiveRedBlackTreeNode* NFindImpl(IntrusiveRedBlackTreeNode const* node) const {
return freebsd::RB_NFIND(const_cast<ImplType::RootType&>(m_impl.m_root),
const_cast<IntrusiveRedBlackTreeNode*>(node), CompareImpl);
}
IntrusiveRedBlackTreeNode* FindLightImpl(const_light_pointer lelm) const {
return RB_FIND_LIGHT(const_cast<ImplType::RootType*>(&impl.root),
static_cast<const void*>(lelm), LightCompareImpl);
constexpr IntrusiveRedBlackTreeNode* FindKeyImpl(const_key_reference key) const {
return freebsd::RB_FIND_KEY(const_cast<ImplType::RootType&>(m_impl.m_root), key,
CompareKeyImpl);
}
IntrusiveRedBlackTreeNode* NFindLightImpl(const_light_pointer lelm) const {
return RB_NFIND_LIGHT(const_cast<ImplType::RootType*>(&impl.root),
static_cast<const void*>(lelm), LightCompareImpl);
constexpr IntrusiveRedBlackTreeNode* NFindKeyImpl(const_key_reference key) const {
return freebsd::RB_NFIND_KEY(const_cast<ImplType::RootType&>(m_impl.m_root), key,
CompareKeyImpl);
}
constexpr IntrusiveRedBlackTreeNode* FindExistingImpl(
IntrusiveRedBlackTreeNode const* node) const {
return freebsd::RB_FIND_EXISTING(const_cast<ImplType::RootType&>(m_impl.m_root),
const_cast<IntrusiveRedBlackTreeNode*>(node), CompareImpl);
}
constexpr IntrusiveRedBlackTreeNode* FindExistingKeyImpl(const_key_reference key) const {
return freebsd::RB_FIND_EXISTING_KEY(const_cast<ImplType::RootType&>(m_impl.m_root), key,
CompareKeyImpl);
}
public:
constexpr IntrusiveRedBlackTree() = default;
// Iterator accessors.
iterator begin() {
return iterator(this->impl.begin());
constexpr iterator begin() {
return iterator(m_impl.begin());
}
const_iterator begin() const {
return const_iterator(this->impl.begin());
constexpr const_iterator begin() const {
return const_iterator(m_impl.begin());
}
iterator end() {
return iterator(this->impl.end());
constexpr iterator end() {
return iterator(m_impl.end());
}
const_iterator end() const {
return const_iterator(this->impl.end());
constexpr const_iterator end() const {
return const_iterator(m_impl.end());
}
const_iterator cbegin() const {
constexpr const_iterator cbegin() const {
return this->begin();
}
const_iterator cend() const {
constexpr const_iterator cend() const {
return this->end();
}
iterator iterator_to(reference ref) {
return iterator(this->impl.iterator_to(*Traits::GetNode(std::addressof(ref))));
constexpr iterator iterator_to(reference ref) {
return iterator(m_impl.iterator_to(*Traits::GetNode(std::addressof(ref))));
}
const_iterator iterator_to(const_reference ref) const {
return const_iterator(this->impl.iterator_to(*Traits::GetNode(std::addressof(ref))));
constexpr const_iterator iterator_to(const_reference ref) const {
return const_iterator(m_impl.iterator_to(*Traits::GetNode(std::addressof(ref))));
}
// Content management.
bool empty() const {
return this->impl.empty();
constexpr bool empty() const {
return m_impl.empty();
}
reference back() {
return *Traits::GetParent(std::addressof(this->impl.back()));
constexpr reference back() {
return *Traits::GetParent(std::addressof(m_impl.back()));
}
const_reference back() const {
return *Traits::GetParent(std::addressof(this->impl.back()));
constexpr const_reference back() const {
return *Traits::GetParent(std::addressof(m_impl.back()));
}
reference front() {
return *Traits::GetParent(std::addressof(this->impl.front()));
constexpr reference front() {
return *Traits::GetParent(std::addressof(m_impl.front()));
}
const_reference front() const {
return *Traits::GetParent(std::addressof(this->impl.front()));
constexpr const_reference front() const {
return *Traits::GetParent(std::addressof(m_impl.front()));
}
iterator erase(iterator it) {
return iterator(this->impl.erase(it.GetImplIterator()));
constexpr iterator erase(iterator it) {
return iterator(m_impl.erase(it.GetImplIterator()));
}
iterator insert(reference ref) {
constexpr iterator insert(reference ref) {
ImplType::pointer node = Traits::GetNode(std::addressof(ref));
this->InsertImpl(node);
return iterator(node);
}
iterator find(const_reference ref) const {
constexpr iterator find(const_reference ref) const {
return iterator(this->FindImpl(Traits::GetNode(std::addressof(ref))));
}
iterator nfind(const_reference ref) const {
constexpr iterator nfind(const_reference ref) const {
return iterator(this->NFindImpl(Traits::GetNode(std::addressof(ref))));
}
iterator find_light(const_light_reference ref) const {
return iterator(this->FindLightImpl(std::addressof(ref)));
constexpr iterator find_key(const_key_reference ref) const {
return iterator(this->FindKeyImpl(ref));
}
iterator nfind_light(const_light_reference ref) const {
return iterator(this->NFindLightImpl(std::addressof(ref)));
constexpr iterator nfind_key(const_key_reference ref) const {
return iterator(this->NFindKeyImpl(ref));
}
constexpr iterator find_existing(const_reference ref) const {
return iterator(this->FindExistingImpl(Traits::GetNode(std::addressof(ref))));
}
constexpr iterator find_existing_key(const_key_reference ref) const {
return iterator(this->FindExistingKeyImpl(ref));
}
};
template <auto T, class Derived = impl::GetParentType<T>>
template <auto T, class Derived = Common::impl::GetParentType<T>>
class IntrusiveRedBlackTreeMemberTraits;
template <class Parent, IntrusiveRedBlackTreeNode Parent::*Member, class Derived>
@@ -498,19 +525,16 @@ private:
return std::addressof(parent->*Member);
}
static constexpr Derived* GetParent(IntrusiveRedBlackTreeNode* node) {
return GetParentPointer<Member, Derived>(node);
static Derived* GetParent(IntrusiveRedBlackTreeNode* node) {
return Common::GetParentPointer<Member, Derived>(node);
}
static constexpr Derived const* GetParent(const IntrusiveRedBlackTreeNode* node) {
return GetParentPointer<Member, Derived>(node);
static Derived const* GetParent(IntrusiveRedBlackTreeNode const* node) {
return Common::GetParentPointer<Member, Derived>(node);
}
private:
static constexpr TypedStorage<Derived> DerivedStorage = {};
};
template <auto T, class Derived = impl::GetParentType<T>>
template <auto T, class Derived = Common::impl::GetParentType<T>>
class IntrusiveRedBlackTreeMemberTraitsDeferredAssert;
template <class Parent, IntrusiveRedBlackTreeNode Parent::*Member, class Derived>
@@ -521,11 +545,6 @@ public:
IntrusiveRedBlackTree<Derived, IntrusiveRedBlackTreeMemberTraitsDeferredAssert, Comparator>;
using TreeTypeImpl = impl::IntrusiveRedBlackTreeImpl;
static constexpr bool IsValid() {
TypedStorage<Derived> DerivedStorage = {};
return GetParent(GetNode(GetPointer(DerivedStorage))) == GetPointer(DerivedStorage);
}
private:
template <class, class, class>
friend class IntrusiveRedBlackTree;
@@ -540,30 +559,36 @@ private:
return std::addressof(parent->*Member);
}
static constexpr Derived* GetParent(IntrusiveRedBlackTreeNode* node) {
return GetParentPointer<Member, Derived>(node);
static Derived* GetParent(IntrusiveRedBlackTreeNode* node) {
return Common::GetParentPointer<Member, Derived>(node);
}
static constexpr Derived const* GetParent(const IntrusiveRedBlackTreeNode* node) {
return GetParentPointer<Member, Derived>(node);
static Derived const* GetParent(IntrusiveRedBlackTreeNode const* node) {
return Common::GetParentPointer<Member, Derived>(node);
}
};
template <class Derived>
class IntrusiveRedBlackTreeBaseNode : public IntrusiveRedBlackTreeNode {
class alignas(void*) IntrusiveRedBlackTreeBaseNode : public IntrusiveRedBlackTreeNode {
public:
using IntrusiveRedBlackTreeNode::IntrusiveRedBlackTreeNode;
constexpr Derived* GetPrev() {
return static_cast<Derived*>(impl::IntrusiveRedBlackTreeImpl::GetPrev(this));
return static_cast<Derived*>(static_cast<IntrusiveRedBlackTreeBaseNode*>(
impl::IntrusiveRedBlackTreeImpl::GetPrev(this)));
}
constexpr const Derived* GetPrev() const {
return static_cast<const Derived*>(impl::IntrusiveRedBlackTreeImpl::GetPrev(this));
return static_cast<const Derived*>(static_cast<const IntrusiveRedBlackTreeBaseNode*>(
impl::IntrusiveRedBlackTreeImpl::GetPrev(this)));
}
constexpr Derived* GetNext() {
return static_cast<Derived*>(impl::IntrusiveRedBlackTreeImpl::GetNext(this));
return static_cast<Derived*>(static_cast<IntrusiveRedBlackTreeBaseNode*>(
impl::IntrusiveRedBlackTreeImpl::GetNext(this)));
}
constexpr const Derived* GetNext() const {
return static_cast<const Derived*>(impl::IntrusiveRedBlackTreeImpl::GetNext(this));
return static_cast<const Derived*>(static_cast<const IntrusiveRedBlackTreeBaseNode*>(
impl::IntrusiveRedBlackTreeImpl::GetNext(this)));
}
};
@@ -581,19 +606,22 @@ private:
friend class impl::IntrusiveRedBlackTreeImpl;
static constexpr IntrusiveRedBlackTreeNode* GetNode(Derived* parent) {
return static_cast<IntrusiveRedBlackTreeNode*>(parent);
return static_cast<IntrusiveRedBlackTreeNode*>(
static_cast<IntrusiveRedBlackTreeBaseNode<Derived>*>(parent));
}
static constexpr IntrusiveRedBlackTreeNode const* GetNode(Derived const* parent) {
return static_cast<const IntrusiveRedBlackTreeNode*>(parent);
return static_cast<const IntrusiveRedBlackTreeNode*>(
static_cast<const IntrusiveRedBlackTreeBaseNode<Derived>*>(parent));
}
static constexpr Derived* GetParent(IntrusiveRedBlackTreeNode* node) {
return static_cast<Derived*>(node);
return static_cast<Derived*>(static_cast<IntrusiveRedBlackTreeBaseNode<Derived>*>(node));
}
static constexpr Derived const* GetParent(const IntrusiveRedBlackTreeNode* node) {
return static_cast<const Derived*>(node);
static constexpr Derived const* GetParent(IntrusiveRedBlackTreeNode const* node) {
return static_cast<const Derived*>(
static_cast<const IntrusiveRedBlackTreeBaseNode<Derived>*>(node));
}
};
+5 -15
View File
@@ -5,10 +5,8 @@
#include <atomic>
#include <chrono>
#include <climits>
#include <exception>
#include <stop_token>
#include <thread>
#include <vector>
#include <fmt/format.h>
@@ -218,19 +216,17 @@ private:
Impl(const std::filesystem::path& file_backend_filename, const Filter& filter_)
: filter{filter_}, file_backend{file_backend_filename} {}
~Impl() {
StopBackendThread();
}
~Impl() = default;
void StartBackendThread() {
backend_thread = std::thread([this] {
backend_thread = std::jthread([this](std::stop_token stop_token) {
Common::SetCurrentThreadName("yuzu:Log");
Entry entry;
const auto write_logs = [this, &entry]() {
ForEachBackend([&entry](Backend& backend) { backend.Write(entry); });
};
while (!stop.stop_requested()) {
entry = message_queue.PopWait(stop.get_token());
while (!stop_token.stop_requested()) {
entry = message_queue.PopWait(stop_token);
if (entry.filename != nullptr) {
write_logs();
}
@@ -244,11 +240,6 @@ private:
});
}
void StopBackendThread() {
stop.request_stop();
backend_thread.join();
}
Entry CreateEntry(Class log_class, Level log_level, const char* filename, unsigned int line_nr,
const char* function, std::string&& message) const {
using std::chrono::duration_cast;
@@ -283,10 +274,9 @@ private:
ColorConsoleBackend color_console_backend{};
FileBackend file_backend;
std::stop_source stop;
std::thread backend_thread;
MPSCQueue<Entry, true> message_queue{};
std::chrono::steady_clock::time_point time_origin{std::chrono::steady_clock::now()};
std::jthread backend_thread;
};
} // namespace
-1
View File
@@ -4,7 +4,6 @@
#pragma once
#include <filesystem>
#include "common/logging/filter.h"
namespace Common::Log {
-1
View File
@@ -7,7 +7,6 @@
#include <array>
#include <chrono>
#include <cstddef>
#include <string_view>
#include "common/logging/log.h"
namespace Common::Log {
-2
View File
@@ -10,12 +10,10 @@
#endif
#include "common/assert.h"
#include "common/common_funcs.h"
#include "common/logging/filter.h"
#include "common/logging/log.h"
#include "common/logging/log_entry.h"
#include "common/logging/text_formatter.h"
#include "common/string_util.h"
namespace Common::Log {
-1
View File
@@ -4,7 +4,6 @@
#pragma once
#include <cstddef>
#include <string>
namespace Common::Log {
+1 -1
View File
@@ -70,4 +70,4 @@ const MemoryInfo& GetMemInfo() {
return mem_info;
}
} // namespace Common
} // namespace Common
-1
View File
@@ -6,7 +6,6 @@
#include <fmt/format.h>
#include "common/fs/file.h"
#include "common/fs/fs.h"
#include "common/fs/path_util.h"
#include "common/nvidia_flags.h"
-1
View File
@@ -5,7 +5,6 @@
#pragma once
#include <atomic>
#include <tuple>
#include "common/common_types.h"
#include "common/virtual_buffer.h"
-1
View File
@@ -7,7 +7,6 @@
#include <type_traits>
#include "common/assert.h"
#include "common/common_types.h"
namespace Common {
namespace detail {
-1
View File
@@ -12,7 +12,6 @@
#include <new>
#include <type_traits>
#include <vector>
#include "common/common_types.h"
namespace Common {
+2
View File
@@ -167,6 +167,7 @@ void RestoreGlobalState(bool is_powered_on) {
// Core
values.use_multi_core.SetGlobal(true);
values.use_extended_memory_layout.SetGlobal(true);
// CPU
values.cpu_accuracy.SetGlobal(true);
@@ -175,6 +176,7 @@ void RestoreGlobalState(bool is_powered_on) {
values.cpuopt_unsafe_ignore_standard_fpcr.SetGlobal(true);
values.cpuopt_unsafe_inaccurate_nan.SetGlobal(true);
values.cpuopt_unsafe_fastmem_check.SetGlobal(true);
values.cpuopt_unsafe_ignore_global_monitor.SetGlobal(true);
// Renderer
values.renderer_backend.SetGlobal(true);
+4
View File
@@ -466,6 +466,7 @@ struct Values {
// Core
Setting<bool> use_multi_core{true, "use_multi_core"};
Setting<bool> use_extended_memory_layout{false, "use_extended_memory_layout"};
// Cpu
RangedSetting<CPUAccuracy> cpu_accuracy{CPUAccuracy::Auto, CPUAccuracy::Auto,
@@ -483,12 +484,15 @@ struct Values {
BasicSetting<bool> cpuopt_misc_ir{true, "cpuopt_misc_ir"};
BasicSetting<bool> cpuopt_reduce_misalign_checks{true, "cpuopt_reduce_misalign_checks"};
BasicSetting<bool> cpuopt_fastmem{true, "cpuopt_fastmem"};
BasicSetting<bool> cpuopt_fastmem_exclusives{true, "cpuopt_fastmem_exclusives"};
BasicSetting<bool> cpuopt_recompile_exclusives{true, "cpuopt_recompile_exclusives"};
Setting<bool> cpuopt_unsafe_unfuse_fma{true, "cpuopt_unsafe_unfuse_fma"};
Setting<bool> cpuopt_unsafe_reduce_fp_error{true, "cpuopt_unsafe_reduce_fp_error"};
Setting<bool> cpuopt_unsafe_ignore_standard_fpcr{true, "cpuopt_unsafe_ignore_standard_fpcr"};
Setting<bool> cpuopt_unsafe_inaccurate_nan{true, "cpuopt_unsafe_inaccurate_nan"};
Setting<bool> cpuopt_unsafe_fastmem_check{true, "cpuopt_unsafe_fastmem_check"};
Setting<bool> cpuopt_unsafe_ignore_global_monitor{true, "cpuopt_unsafe_ignore_global_monitor"};
// Renderer
RangedSetting<RendererBackend> renderer_backend{
-2
View File
@@ -5,11 +5,9 @@
#include <algorithm>
#include <cctype>
#include <codecvt>
#include <cstdlib>
#include <locale>
#include <sstream>
#include "common/logging/log.h"
#include "common/string_util.h"
#ifdef _WIN32
+44 -17
View File
@@ -4,7 +4,6 @@
#include <algorithm>
#include <cstring>
#include "common/assert.h"
#include "common/scm_rev.h"
#include "common/telemetry.h"
@@ -55,22 +54,50 @@ void AppendBuildInfo(FieldCollection& fc) {
void AppendCPUInfo(FieldCollection& fc) {
#ifdef ARCHITECTURE_x86_64
fc.AddField(FieldType::UserSystem, "CPU_Model", Common::GetCPUCaps().cpu_string);
fc.AddField(FieldType::UserSystem, "CPU_BrandString", Common::GetCPUCaps().brand_string);
fc.AddField(FieldType::UserSystem, "CPU_Extension_x64_AES", Common::GetCPUCaps().aes);
fc.AddField(FieldType::UserSystem, "CPU_Extension_x64_AVX", Common::GetCPUCaps().avx);
fc.AddField(FieldType::UserSystem, "CPU_Extension_x64_AVX2", Common::GetCPUCaps().avx2);
fc.AddField(FieldType::UserSystem, "CPU_Extension_x64_AVX512", Common::GetCPUCaps().avx512);
fc.AddField(FieldType::UserSystem, "CPU_Extension_x64_BMI1", Common::GetCPUCaps().bmi1);
fc.AddField(FieldType::UserSystem, "CPU_Extension_x64_BMI2", Common::GetCPUCaps().bmi2);
fc.AddField(FieldType::UserSystem, "CPU_Extension_x64_FMA", Common::GetCPUCaps().fma);
fc.AddField(FieldType::UserSystem, "CPU_Extension_x64_FMA4", Common::GetCPUCaps().fma4);
fc.AddField(FieldType::UserSystem, "CPU_Extension_x64_SSE", Common::GetCPUCaps().sse);
fc.AddField(FieldType::UserSystem, "CPU_Extension_x64_SSE2", Common::GetCPUCaps().sse2);
fc.AddField(FieldType::UserSystem, "CPU_Extension_x64_SSE3", Common::GetCPUCaps().sse3);
fc.AddField(FieldType::UserSystem, "CPU_Extension_x64_SSSE3", Common::GetCPUCaps().ssse3);
fc.AddField(FieldType::UserSystem, "CPU_Extension_x64_SSE41", Common::GetCPUCaps().sse4_1);
fc.AddField(FieldType::UserSystem, "CPU_Extension_x64_SSE42", Common::GetCPUCaps().sse4_2);
const auto& caps = Common::GetCPUCaps();
const auto add_field = [&fc](std::string_view field_name, const auto& field_value) {
fc.AddField(FieldType::UserSystem, field_name, field_value);
};
add_field("CPU_Model", caps.cpu_string);
add_field("CPU_BrandString", caps.brand_string);
add_field("CPU_Extension_x64_SSE", caps.sse);
add_field("CPU_Extension_x64_SSE2", caps.sse2);
add_field("CPU_Extension_x64_SSE3", caps.sse3);
add_field("CPU_Extension_x64_SSSE3", caps.ssse3);
add_field("CPU_Extension_x64_SSE41", caps.sse4_1);
add_field("CPU_Extension_x64_SSE42", caps.sse4_2);
add_field("CPU_Extension_x64_AVX", caps.avx);
add_field("CPU_Extension_x64_AVX_VNNI", caps.avx_vnni);
add_field("CPU_Extension_x64_AVX2", caps.avx2);
// Skylake-X/SP level AVX512, for compatibility with the previous telemetry field
add_field("CPU_Extension_x64_AVX512",
caps.avx512f && caps.avx512cd && caps.avx512vl && caps.avx512dq && caps.avx512bw);
add_field("CPU_Extension_x64_AVX512F", caps.avx512f);
add_field("CPU_Extension_x64_AVX512CD", caps.avx512cd);
add_field("CPU_Extension_x64_AVX512VL", caps.avx512vl);
add_field("CPU_Extension_x64_AVX512DQ", caps.avx512dq);
add_field("CPU_Extension_x64_AVX512BW", caps.avx512bw);
add_field("CPU_Extension_x64_AVX512BITALG", caps.avx512bitalg);
add_field("CPU_Extension_x64_AVX512VBMI", caps.avx512vbmi);
add_field("CPU_Extension_x64_AES", caps.aes);
add_field("CPU_Extension_x64_BMI1", caps.bmi1);
add_field("CPU_Extension_x64_BMI2", caps.bmi2);
add_field("CPU_Extension_x64_F16C", caps.f16c);
add_field("CPU_Extension_x64_FMA", caps.fma);
add_field("CPU_Extension_x64_FMA4", caps.fma4);
add_field("CPU_Extension_x64_GFNI", caps.gfni);
add_field("CPU_Extension_x64_INVARIANT_TSC", caps.invariant_tsc);
add_field("CPU_Extension_x64_LZCNT", caps.lzcnt);
add_field("CPU_Extension_x64_MOVBE", caps.movbe);
add_field("CPU_Extension_x64_PCLMULQDQ", caps.pclmulqdq);
add_field("CPU_Extension_x64_POPCNT", caps.popcnt);
add_field("CPU_Extension_x64_SHA", caps.sha);
#else
fc.AddField(FieldType::UserSystem, "CPU_Model", "Other");
#endif
+3 -3
View File
@@ -55,8 +55,8 @@ class Field : public FieldInterface {
public:
YUZU_NON_COPYABLE(Field);
Field(FieldType type_, std::string name_, T value_)
: name(std::move(name_)), type(type_), value(std::move(value_)) {}
Field(FieldType type_, std::string_view name_, T value_)
: name(name_), type(type_), value(std::move(value_)) {}
~Field() override = default;
@@ -123,7 +123,7 @@ public:
* @param value Value for the field to add.
*/
template <typename T>
void AddField(FieldType type, const char* name, T value) {
void AddField(FieldType type, std::string_view name, T value) {
return AddField(std::make_unique<Field<T>>(type, name, std::move(value)));
}
+444 -387
View File
@@ -43,246 +43,445 @@
* The maximum height of a red-black tree is 2lg (n+1).
*/
#include "common/assert.h"
namespace Common::freebsd {
namespace Common {
template <typename T>
class RBHead {
public:
[[nodiscard]] T* Root() {
return rbh_root;
}
[[nodiscard]] const T* Root() const {
return rbh_root;
}
void SetRoot(T* root) {
rbh_root = root;
}
[[nodiscard]] bool IsEmpty() const {
return Root() == nullptr;
}
private:
T* rbh_root = nullptr;
};
enum class EntryColor {
Black,
Red,
enum class RBColor {
RB_BLACK = 0,
RB_RED = 1,
};
#pragma pack(push, 4)
template <typename T>
class RBEntry {
public:
[[nodiscard]] T* Left() {
return rbe_left;
constexpr RBEntry() = default;
[[nodiscard]] constexpr T* Left() {
return m_rbe_left;
}
[[nodiscard]] constexpr const T* Left() const {
return m_rbe_left;
}
[[nodiscard]] const T* Left() const {
return rbe_left;
constexpr void SetLeft(T* e) {
m_rbe_left = e;
}
void SetLeft(T* left) {
rbe_left = left;
[[nodiscard]] constexpr T* Right() {
return m_rbe_right;
}
[[nodiscard]] constexpr const T* Right() const {
return m_rbe_right;
}
[[nodiscard]] T* Right() {
return rbe_right;
constexpr void SetRight(T* e) {
m_rbe_right = e;
}
[[nodiscard]] const T* Right() const {
return rbe_right;
[[nodiscard]] constexpr T* Parent() {
return m_rbe_parent;
}
[[nodiscard]] constexpr const T* Parent() const {
return m_rbe_parent;
}
void SetRight(T* right) {
rbe_right = right;
constexpr void SetParent(T* e) {
m_rbe_parent = e;
}
[[nodiscard]] T* Parent() {
return rbe_parent;
[[nodiscard]] constexpr bool IsBlack() const {
return m_rbe_color == RBColor::RB_BLACK;
}
[[nodiscard]] constexpr bool IsRed() const {
return m_rbe_color == RBColor::RB_RED;
}
[[nodiscard]] constexpr RBColor Color() const {
return m_rbe_color;
}
[[nodiscard]] const T* Parent() const {
return rbe_parent;
}
void SetParent(T* parent) {
rbe_parent = parent;
}
[[nodiscard]] bool IsBlack() const {
return rbe_color == EntryColor::Black;
}
[[nodiscard]] bool IsRed() const {
return rbe_color == EntryColor::Red;
}
[[nodiscard]] EntryColor Color() const {
return rbe_color;
}
void SetColor(EntryColor color) {
rbe_color = color;
constexpr void SetColor(RBColor c) {
m_rbe_color = c;
}
private:
T* rbe_left = nullptr;
T* rbe_right = nullptr;
T* rbe_parent = nullptr;
EntryColor rbe_color{};
T* m_rbe_left{};
T* m_rbe_right{};
T* m_rbe_parent{};
RBColor m_rbe_color{RBColor::RB_BLACK};
};
#pragma pack(pop)
template <typename T>
struct CheckRBEntry {
static constexpr bool value = false;
};
template <typename T>
struct CheckRBEntry<RBEntry<T>> {
static constexpr bool value = true;
};
template <typename Node>
[[nodiscard]] RBEntry<Node>& RB_ENTRY(Node* node) {
return node->GetEntry();
template <typename T>
concept IsRBEntry = CheckRBEntry<T>::value;
template <typename T>
concept HasRBEntry = requires(T& t, const T& ct) {
{ t.GetRBEntry() } -> std::same_as<RBEntry<T>&>;
{ ct.GetRBEntry() } -> std::same_as<const RBEntry<T>&>;
};
template <typename T>
requires HasRBEntry<T>
class RBHead {
private:
T* m_rbh_root = nullptr;
public:
[[nodiscard]] constexpr T* Root() {
return m_rbh_root;
}
[[nodiscard]] constexpr const T* Root() const {
return m_rbh_root;
}
constexpr void SetRoot(T* root) {
m_rbh_root = root;
}
[[nodiscard]] constexpr bool IsEmpty() const {
return this->Root() == nullptr;
}
};
template <typename T>
requires HasRBEntry<T>
[[nodiscard]] constexpr RBEntry<T>& RB_ENTRY(T* t) {
return t->GetRBEntry();
}
template <typename T>
requires HasRBEntry<T>
[[nodiscard]] constexpr const RBEntry<T>& RB_ENTRY(const T* t) {
return t->GetRBEntry();
}
template <typename Node>
[[nodiscard]] const RBEntry<Node>& RB_ENTRY(const Node* node) {
return node->GetEntry();
template <typename T>
requires HasRBEntry<T>
[[nodiscard]] constexpr T* RB_LEFT(T* t) {
return RB_ENTRY(t).Left();
}
template <typename T>
requires HasRBEntry<T>
[[nodiscard]] constexpr const T* RB_LEFT(const T* t) {
return RB_ENTRY(t).Left();
}
template <typename Node>
[[nodiscard]] Node* RB_PARENT(Node* node) {
return RB_ENTRY(node).Parent();
template <typename T>
requires HasRBEntry<T>
[[nodiscard]] constexpr T* RB_RIGHT(T* t) {
return RB_ENTRY(t).Right();
}
template <typename T>
requires HasRBEntry<T>
[[nodiscard]] constexpr const T* RB_RIGHT(const T* t) {
return RB_ENTRY(t).Right();
}
template <typename Node>
[[nodiscard]] const Node* RB_PARENT(const Node* node) {
return RB_ENTRY(node).Parent();
template <typename T>
requires HasRBEntry<T>
[[nodiscard]] constexpr T* RB_PARENT(T* t) {
return RB_ENTRY(t).Parent();
}
template <typename T>
requires HasRBEntry<T>
[[nodiscard]] constexpr const T* RB_PARENT(const T* t) {
return RB_ENTRY(t).Parent();
}
template <typename Node>
void RB_SET_PARENT(Node* node, Node* parent) {
return RB_ENTRY(node).SetParent(parent);
template <typename T>
requires HasRBEntry<T>
constexpr void RB_SET_LEFT(T* t, T* e) {
RB_ENTRY(t).SetLeft(e);
}
template <typename T>
requires HasRBEntry<T>
constexpr void RB_SET_RIGHT(T* t, T* e) {
RB_ENTRY(t).SetRight(e);
}
template <typename T>
requires HasRBEntry<T>
constexpr void RB_SET_PARENT(T* t, T* e) {
RB_ENTRY(t).SetParent(e);
}
template <typename Node>
[[nodiscard]] Node* RB_LEFT(Node* node) {
return RB_ENTRY(node).Left();
template <typename T>
requires HasRBEntry<T>
[[nodiscard]] constexpr bool RB_IS_BLACK(const T* t) {
return RB_ENTRY(t).IsBlack();
}
template <typename T>
requires HasRBEntry<T>
[[nodiscard]] constexpr bool RB_IS_RED(const T* t) {
return RB_ENTRY(t).IsRed();
}
template <typename Node>
[[nodiscard]] const Node* RB_LEFT(const Node* node) {
return RB_ENTRY(node).Left();
template <typename T>
requires HasRBEntry<T>
[[nodiscard]] constexpr RBColor RB_COLOR(const T* t) {
return RB_ENTRY(t).Color();
}
template <typename Node>
void RB_SET_LEFT(Node* node, Node* left) {
return RB_ENTRY(node).SetLeft(left);
template <typename T>
requires HasRBEntry<T>
constexpr void RB_SET_COLOR(T* t, RBColor c) {
RB_ENTRY(t).SetColor(c);
}
template <typename Node>
[[nodiscard]] Node* RB_RIGHT(Node* node) {
return RB_ENTRY(node).Right();
template <typename T>
requires HasRBEntry<T>
constexpr void RB_SET(T* elm, T* parent) {
auto& rb_entry = RB_ENTRY(elm);
rb_entry.SetParent(parent);
rb_entry.SetLeft(nullptr);
rb_entry.SetRight(nullptr);
rb_entry.SetColor(RBColor::RB_RED);
}
template <typename Node>
[[nodiscard]] const Node* RB_RIGHT(const Node* node) {
return RB_ENTRY(node).Right();
template <typename T>
requires HasRBEntry<T>
constexpr void RB_SET_BLACKRED(T* black, T* red) {
RB_SET_COLOR(black, RBColor::RB_BLACK);
RB_SET_COLOR(red, RBColor::RB_RED);
}
template <typename Node>
void RB_SET_RIGHT(Node* node, Node* right) {
return RB_ENTRY(node).SetRight(right);
}
template <typename Node>
[[nodiscard]] bool RB_IS_BLACK(const Node* node) {
return RB_ENTRY(node).IsBlack();
}
template <typename Node>
[[nodiscard]] bool RB_IS_RED(const Node* node) {
return RB_ENTRY(node).IsRed();
}
template <typename Node>
[[nodiscard]] EntryColor RB_COLOR(const Node* node) {
return RB_ENTRY(node).Color();
}
template <typename Node>
void RB_SET_COLOR(Node* node, EntryColor color) {
return RB_ENTRY(node).SetColor(color);
}
template <typename Node>
void RB_SET(Node* node, Node* parent) {
auto& entry = RB_ENTRY(node);
entry.SetParent(parent);
entry.SetLeft(nullptr);
entry.SetRight(nullptr);
entry.SetColor(EntryColor::Red);
}
template <typename Node>
void RB_SET_BLACKRED(Node* black, Node* red) {
RB_SET_COLOR(black, EntryColor::Black);
RB_SET_COLOR(red, EntryColor::Red);
}
template <typename Node>
void RB_ROTATE_LEFT(RBHead<Node>* head, Node* elm, Node*& tmp) {
template <typename T>
requires HasRBEntry<T>
constexpr void RB_ROTATE_LEFT(RBHead<T>& head, T* elm, T*& tmp) {
tmp = RB_RIGHT(elm);
RB_SET_RIGHT(elm, RB_LEFT(tmp));
if (RB_RIGHT(elm) != nullptr) {
if (RB_SET_RIGHT(elm, RB_LEFT(tmp)); RB_RIGHT(elm) != nullptr) {
RB_SET_PARENT(RB_LEFT(tmp), elm);
}
RB_SET_PARENT(tmp, RB_PARENT(elm));
if (RB_PARENT(tmp) != nullptr) {
if (RB_SET_PARENT(tmp, RB_PARENT(elm)); RB_PARENT(tmp) != nullptr) {
if (elm == RB_LEFT(RB_PARENT(elm))) {
RB_SET_LEFT(RB_PARENT(elm), tmp);
} else {
RB_SET_RIGHT(RB_PARENT(elm), tmp);
}
} else {
head->SetRoot(tmp);
head.SetRoot(tmp);
}
RB_SET_LEFT(tmp, elm);
RB_SET_PARENT(elm, tmp);
}
template <typename Node>
void RB_ROTATE_RIGHT(RBHead<Node>* head, Node* elm, Node*& tmp) {
template <typename T>
requires HasRBEntry<T>
constexpr void RB_ROTATE_RIGHT(RBHead<T>& head, T* elm, T*& tmp) {
tmp = RB_LEFT(elm);
RB_SET_LEFT(elm, RB_RIGHT(tmp));
if (RB_LEFT(elm) != nullptr) {
if (RB_SET_LEFT(elm, RB_RIGHT(tmp)); RB_LEFT(elm) != nullptr) {
RB_SET_PARENT(RB_RIGHT(tmp), elm);
}
RB_SET_PARENT(tmp, RB_PARENT(elm));
if (RB_PARENT(tmp) != nullptr) {
if (RB_SET_PARENT(tmp, RB_PARENT(elm)); RB_PARENT(tmp) != nullptr) {
if (elm == RB_LEFT(RB_PARENT(elm))) {
RB_SET_LEFT(RB_PARENT(elm), tmp);
} else {
RB_SET_RIGHT(RB_PARENT(elm), tmp);
}
} else {
head->SetRoot(tmp);
head.SetRoot(tmp);
}
RB_SET_RIGHT(tmp, elm);
RB_SET_PARENT(elm, tmp);
}
template <typename Node>
void RB_INSERT_COLOR(RBHead<Node>* head, Node* elm) {
Node* parent = nullptr;
Node* tmp = nullptr;
template <typename T>
requires HasRBEntry<T>
constexpr void RB_REMOVE_COLOR(RBHead<T>& head, T* parent, T* elm) {
T* tmp;
while ((elm == nullptr || RB_IS_BLACK(elm)) && elm != head.Root()) {
if (RB_LEFT(parent) == elm) {
tmp = RB_RIGHT(parent);
if (RB_IS_RED(tmp)) {
RB_SET_BLACKRED(tmp, parent);
RB_ROTATE_LEFT(head, parent, tmp);
tmp = RB_RIGHT(parent);
}
if ((RB_LEFT(tmp) == nullptr || RB_IS_BLACK(RB_LEFT(tmp))) &&
(RB_RIGHT(tmp) == nullptr || RB_IS_BLACK(RB_RIGHT(tmp)))) {
RB_SET_COLOR(tmp, RBColor::RB_RED);
elm = parent;
parent = RB_PARENT(elm);
} else {
if (RB_RIGHT(tmp) == nullptr || RB_IS_BLACK(RB_RIGHT(tmp))) {
T* oleft;
if ((oleft = RB_LEFT(tmp)) != nullptr) {
RB_SET_COLOR(oleft, RBColor::RB_BLACK);
}
RB_SET_COLOR(tmp, RBColor::RB_RED);
RB_ROTATE_RIGHT(head, tmp, oleft);
tmp = RB_RIGHT(parent);
}
RB_SET_COLOR(tmp, RB_COLOR(parent));
RB_SET_COLOR(parent, RBColor::RB_BLACK);
if (RB_RIGHT(tmp)) {
RB_SET_COLOR(RB_RIGHT(tmp), RBColor::RB_BLACK);
}
RB_ROTATE_LEFT(head, parent, tmp);
elm = head.Root();
break;
}
} else {
tmp = RB_LEFT(parent);
if (RB_IS_RED(tmp)) {
RB_SET_BLACKRED(tmp, parent);
RB_ROTATE_RIGHT(head, parent, tmp);
tmp = RB_LEFT(parent);
}
if ((RB_LEFT(tmp) == nullptr || RB_IS_BLACK(RB_LEFT(tmp))) &&
(RB_RIGHT(tmp) == nullptr || RB_IS_BLACK(RB_RIGHT(tmp)))) {
RB_SET_COLOR(tmp, RBColor::RB_RED);
elm = parent;
parent = RB_PARENT(elm);
} else {
if (RB_LEFT(tmp) == nullptr || RB_IS_BLACK(RB_LEFT(tmp))) {
T* oright;
if ((oright = RB_RIGHT(tmp)) != nullptr) {
RB_SET_COLOR(oright, RBColor::RB_BLACK);
}
RB_SET_COLOR(tmp, RBColor::RB_RED);
RB_ROTATE_LEFT(head, tmp, oright);
tmp = RB_LEFT(parent);
}
RB_SET_COLOR(tmp, RB_COLOR(parent));
RB_SET_COLOR(parent, RBColor::RB_BLACK);
if (RB_LEFT(tmp)) {
RB_SET_COLOR(RB_LEFT(tmp), RBColor::RB_BLACK);
}
RB_ROTATE_RIGHT(head, parent, tmp);
elm = head.Root();
break;
}
}
}
if (elm) {
RB_SET_COLOR(elm, RBColor::RB_BLACK);
}
}
template <typename T>
requires HasRBEntry<T>
constexpr T* RB_REMOVE(RBHead<T>& head, T* elm) {
T* child = nullptr;
T* parent = nullptr;
T* old = elm;
RBColor color = RBColor::RB_BLACK;
if (RB_LEFT(elm) == nullptr) {
child = RB_RIGHT(elm);
} else if (RB_RIGHT(elm) == nullptr) {
child = RB_LEFT(elm);
} else {
T* left;
elm = RB_RIGHT(elm);
while ((left = RB_LEFT(elm)) != nullptr) {
elm = left;
}
child = RB_RIGHT(elm);
parent = RB_PARENT(elm);
color = RB_COLOR(elm);
if (child) {
RB_SET_PARENT(child, parent);
}
if (parent) {
if (RB_LEFT(parent) == elm) {
RB_SET_LEFT(parent, child);
} else {
RB_SET_RIGHT(parent, child);
}
} else {
head.SetRoot(child);
}
if (RB_PARENT(elm) == old) {
parent = elm;
}
elm->SetRBEntry(old->GetRBEntry());
if (RB_PARENT(old)) {
if (RB_LEFT(RB_PARENT(old)) == old) {
RB_SET_LEFT(RB_PARENT(old), elm);
} else {
RB_SET_RIGHT(RB_PARENT(old), elm);
}
} else {
head.SetRoot(elm);
}
RB_SET_PARENT(RB_LEFT(old), elm);
if (RB_RIGHT(old)) {
RB_SET_PARENT(RB_RIGHT(old), elm);
}
if (parent) {
left = parent;
}
if (color == RBColor::RB_BLACK) {
RB_REMOVE_COLOR(head, parent, child);
}
return old;
}
parent = RB_PARENT(elm);
color = RB_COLOR(elm);
if (child) {
RB_SET_PARENT(child, parent);
}
if (parent) {
if (RB_LEFT(parent) == elm) {
RB_SET_LEFT(parent, child);
} else {
RB_SET_RIGHT(parent, child);
}
} else {
head.SetRoot(child);
}
if (color == RBColor::RB_BLACK) {
RB_REMOVE_COLOR(head, parent, child);
}
return old;
}
template <typename T>
requires HasRBEntry<T>
constexpr void RB_INSERT_COLOR(RBHead<T>& head, T* elm) {
T *parent = nullptr, *tmp = nullptr;
while ((parent = RB_PARENT(elm)) != nullptr && RB_IS_RED(parent)) {
Node* gparent = RB_PARENT(parent);
T* gparent = RB_PARENT(parent);
if (parent == RB_LEFT(gparent)) {
tmp = RB_RIGHT(gparent);
if (tmp && RB_IS_RED(tmp)) {
RB_SET_COLOR(tmp, EntryColor::Black);
RB_SET_COLOR(tmp, RBColor::RB_BLACK);
RB_SET_BLACKRED(parent, gparent);
elm = gparent;
continue;
@@ -300,7 +499,7 @@ void RB_INSERT_COLOR(RBHead<Node>* head, Node* elm) {
} else {
tmp = RB_LEFT(gparent);
if (tmp && RB_IS_RED(tmp)) {
RB_SET_COLOR(tmp, EntryColor::Black);
RB_SET_COLOR(tmp, RBColor::RB_BLACK);
RB_SET_BLACKRED(parent, gparent);
elm = gparent;
continue;
@@ -318,194 +517,14 @@ void RB_INSERT_COLOR(RBHead<Node>* head, Node* elm) {
}
}
RB_SET_COLOR(head->Root(), EntryColor::Black);
RB_SET_COLOR(head.Root(), RBColor::RB_BLACK);
}
template <typename Node>
void RB_REMOVE_COLOR(RBHead<Node>* head, Node* parent, Node* elm) {
Node* tmp;
while ((elm == nullptr || RB_IS_BLACK(elm)) && elm != head->Root() && parent != nullptr) {
if (RB_LEFT(parent) == elm) {
tmp = RB_RIGHT(parent);
if (!tmp) {
ASSERT_MSG(false, "tmp is invalid!");
break;
}
if (RB_IS_RED(tmp)) {
RB_SET_BLACKRED(tmp, parent);
RB_ROTATE_LEFT(head, parent, tmp);
tmp = RB_RIGHT(parent);
}
if ((RB_LEFT(tmp) == nullptr || RB_IS_BLACK(RB_LEFT(tmp))) &&
(RB_RIGHT(tmp) == nullptr || RB_IS_BLACK(RB_RIGHT(tmp)))) {
RB_SET_COLOR(tmp, EntryColor::Red);
elm = parent;
parent = RB_PARENT(elm);
} else {
if (RB_RIGHT(tmp) == nullptr || RB_IS_BLACK(RB_RIGHT(tmp))) {
Node* oleft;
if ((oleft = RB_LEFT(tmp)) != nullptr) {
RB_SET_COLOR(oleft, EntryColor::Black);
}
RB_SET_COLOR(tmp, EntryColor::Red);
RB_ROTATE_RIGHT(head, tmp, oleft);
tmp = RB_RIGHT(parent);
}
RB_SET_COLOR(tmp, RB_COLOR(parent));
RB_SET_COLOR(parent, EntryColor::Black);
if (RB_RIGHT(tmp)) {
RB_SET_COLOR(RB_RIGHT(tmp), EntryColor::Black);
}
RB_ROTATE_LEFT(head, parent, tmp);
elm = head->Root();
break;
}
} else {
tmp = RB_LEFT(parent);
if (RB_IS_RED(tmp)) {
RB_SET_BLACKRED(tmp, parent);
RB_ROTATE_RIGHT(head, parent, tmp);
tmp = RB_LEFT(parent);
}
if (!tmp) {
ASSERT_MSG(false, "tmp is invalid!");
break;
}
if ((RB_LEFT(tmp) == nullptr || RB_IS_BLACK(RB_LEFT(tmp))) &&
(RB_RIGHT(tmp) == nullptr || RB_IS_BLACK(RB_RIGHT(tmp)))) {
RB_SET_COLOR(tmp, EntryColor::Red);
elm = parent;
parent = RB_PARENT(elm);
} else {
if (RB_LEFT(tmp) == nullptr || RB_IS_BLACK(RB_LEFT(tmp))) {
Node* oright;
if ((oright = RB_RIGHT(tmp)) != nullptr) {
RB_SET_COLOR(oright, EntryColor::Black);
}
RB_SET_COLOR(tmp, EntryColor::Red);
RB_ROTATE_LEFT(head, tmp, oright);
tmp = RB_LEFT(parent);
}
RB_SET_COLOR(tmp, RB_COLOR(parent));
RB_SET_COLOR(parent, EntryColor::Black);
if (RB_LEFT(tmp)) {
RB_SET_COLOR(RB_LEFT(tmp), EntryColor::Black);
}
RB_ROTATE_RIGHT(head, parent, tmp);
elm = head->Root();
break;
}
}
}
if (elm) {
RB_SET_COLOR(elm, EntryColor::Black);
}
}
template <typename Node>
Node* RB_REMOVE(RBHead<Node>* head, Node* elm) {
Node* child = nullptr;
Node* parent = nullptr;
Node* old = elm;
EntryColor color{};
const auto finalize = [&] {
if (color == EntryColor::Black) {
RB_REMOVE_COLOR(head, parent, child);
}
return old;
};
if (RB_LEFT(elm) == nullptr) {
child = RB_RIGHT(elm);
} else if (RB_RIGHT(elm) == nullptr) {
child = RB_LEFT(elm);
} else {
Node* left;
elm = RB_RIGHT(elm);
while ((left = RB_LEFT(elm)) != nullptr) {
elm = left;
}
child = RB_RIGHT(elm);
parent = RB_PARENT(elm);
color = RB_COLOR(elm);
if (child) {
RB_SET_PARENT(child, parent);
}
if (parent) {
if (RB_LEFT(parent) == elm) {
RB_SET_LEFT(parent, child);
} else {
RB_SET_RIGHT(parent, child);
}
} else {
head->SetRoot(child);
}
if (RB_PARENT(elm) == old) {
parent = elm;
}
elm->SetEntry(old->GetEntry());
if (RB_PARENT(old)) {
if (RB_LEFT(RB_PARENT(old)) == old) {
RB_SET_LEFT(RB_PARENT(old), elm);
} else {
RB_SET_RIGHT(RB_PARENT(old), elm);
}
} else {
head->SetRoot(elm);
}
RB_SET_PARENT(RB_LEFT(old), elm);
if (RB_RIGHT(old)) {
RB_SET_PARENT(RB_RIGHT(old), elm);
}
if (parent) {
left = parent;
}
return finalize();
}
parent = RB_PARENT(elm);
color = RB_COLOR(elm);
if (child) {
RB_SET_PARENT(child, parent);
}
if (parent) {
if (RB_LEFT(parent) == elm) {
RB_SET_LEFT(parent, child);
} else {
RB_SET_RIGHT(parent, child);
}
} else {
head->SetRoot(child);
}
return finalize();
}
// Inserts a node into the RB tree
template <typename Node, typename CompareFunction>
Node* RB_INSERT(RBHead<Node>* head, Node* elm, CompareFunction cmp) {
Node* parent = nullptr;
Node* tmp = head->Root();
template <typename T, typename Compare>
requires HasRBEntry<T>
constexpr T* RB_INSERT(RBHead<T>& head, T* elm, Compare cmp) {
T* parent = nullptr;
T* tmp = head.Root();
int comp = 0;
while (tmp) {
@@ -529,17 +548,17 @@ Node* RB_INSERT(RBHead<Node>* head, Node* elm, CompareFunction cmp) {
RB_SET_RIGHT(parent, elm);
}
} else {
head->SetRoot(elm);
head.SetRoot(elm);
}
RB_INSERT_COLOR(head, elm);
return nullptr;
}
// Finds the node with the same key as elm
template <typename Node, typename CompareFunction>
Node* RB_FIND(RBHead<Node>* head, Node* elm, CompareFunction cmp) {
Node* tmp = head->Root();
template <typename T, typename Compare>
requires HasRBEntry<T>
constexpr T* RB_FIND(RBHead<T>& head, T* elm, Compare cmp) {
T* tmp = head.Root();
while (tmp) {
const int comp = cmp(elm, tmp);
@@ -555,11 +574,11 @@ Node* RB_FIND(RBHead<Node>* head, Node* elm, CompareFunction cmp) {
return nullptr;
}
// Finds the first node greater than or equal to the search key
template <typename Node, typename CompareFunction>
Node* RB_NFIND(RBHead<Node>* head, Node* elm, CompareFunction cmp) {
Node* tmp = head->Root();
Node* res = nullptr;
template <typename T, typename Compare>
requires HasRBEntry<T>
constexpr T* RB_NFIND(RBHead<T>& head, T* elm, Compare cmp) {
T* tmp = head.Root();
T* res = nullptr;
while (tmp) {
const int comp = cmp(elm, tmp);
@@ -576,13 +595,13 @@ Node* RB_NFIND(RBHead<Node>* head, Node* elm, CompareFunction cmp) {
return res;
}
// Finds the node with the same key as lelm
template <typename Node, typename CompareFunction>
Node* RB_FIND_LIGHT(RBHead<Node>* head, const void* lelm, CompareFunction lcmp) {
Node* tmp = head->Root();
template <typename T, typename U, typename Compare>
requires HasRBEntry<T>
constexpr T* RB_FIND_KEY(RBHead<T>& head, const U& key, Compare cmp) {
T* tmp = head.Root();
while (tmp) {
const int comp = lcmp(lelm, tmp);
const int comp = cmp(key, tmp);
if (comp < 0) {
tmp = RB_LEFT(tmp);
} else if (comp > 0) {
@@ -595,14 +614,14 @@ Node* RB_FIND_LIGHT(RBHead<Node>* head, const void* lelm, CompareFunction lcmp)
return nullptr;
}
// Finds the first node greater than or equal to the search key
template <typename Node, typename CompareFunction>
Node* RB_NFIND_LIGHT(RBHead<Node>* head, const void* lelm, CompareFunction lcmp) {
Node* tmp = head->Root();
Node* res = nullptr;
template <typename T, typename U, typename Compare>
requires HasRBEntry<T>
constexpr T* RB_NFIND_KEY(RBHead<T>& head, const U& key, Compare cmp) {
T* tmp = head.Root();
T* res = nullptr;
while (tmp) {
const int comp = lcmp(lelm, tmp);
const int comp = cmp(key, tmp);
if (comp < 0) {
res = tmp;
tmp = RB_LEFT(tmp);
@@ -616,8 +635,43 @@ Node* RB_NFIND_LIGHT(RBHead<Node>* head, const void* lelm, CompareFunction lcmp)
return res;
}
template <typename Node>
Node* RB_NEXT(Node* elm) {
template <typename T, typename Compare>
requires HasRBEntry<T>
constexpr T* RB_FIND_EXISTING(RBHead<T>& head, T* elm, Compare cmp) {
T* tmp = head.Root();
while (true) {
const int comp = cmp(elm, tmp);
if (comp < 0) {
tmp = RB_LEFT(tmp);
} else if (comp > 0) {
tmp = RB_RIGHT(tmp);
} else {
return tmp;
}
}
}
template <typename T, typename U, typename Compare>
requires HasRBEntry<T>
constexpr T* RB_FIND_EXISTING_KEY(RBHead<T>& head, const U& key, Compare cmp) {
T* tmp = head.Root();
while (true) {
const int comp = cmp(key, tmp);
if (comp < 0) {
tmp = RB_LEFT(tmp);
} else if (comp > 0) {
tmp = RB_RIGHT(tmp);
} else {
return tmp;
}
}
}
template <typename T>
requires HasRBEntry<T>
constexpr T* RB_NEXT(T* elm) {
if (RB_RIGHT(elm)) {
elm = RB_RIGHT(elm);
while (RB_LEFT(elm)) {
@@ -636,8 +690,9 @@ Node* RB_NEXT(Node* elm) {
return elm;
}
template <typename Node>
Node* RB_PREV(Node* elm) {
template <typename T>
requires HasRBEntry<T>
constexpr T* RB_PREV(T* elm) {
if (RB_LEFT(elm)) {
elm = RB_LEFT(elm);
while (RB_RIGHT(elm)) {
@@ -656,30 +711,32 @@ Node* RB_PREV(Node* elm) {
return elm;
}
template <typename Node>
Node* RB_MINMAX(RBHead<Node>* head, bool is_min) {
Node* tmp = head->Root();
Node* parent = nullptr;
template <typename T>
requires HasRBEntry<T>
constexpr T* RB_MIN(RBHead<T>& head) {
T* tmp = head.Root();
T* parent = nullptr;
while (tmp) {
parent = tmp;
if (is_min) {
tmp = RB_LEFT(tmp);
} else {
tmp = RB_RIGHT(tmp);
}
tmp = RB_LEFT(tmp);
}
return parent;
}
template <typename Node>
Node* RB_MIN(RBHead<Node>* head) {
return RB_MINMAX(head, true);
template <typename T>
requires HasRBEntry<T>
constexpr T* RB_MAX(RBHead<T>& head) {
T* tmp = head.Root();
T* parent = nullptr;
while (tmp) {
parent = tmp;
tmp = RB_RIGHT(tmp);
}
return parent;
}
template <typename Node>
Node* RB_MAX(RBHead<Node>* head) {
return RB_MINMAX(head, false);
}
} // namespace Common
} // namespace Common::freebsd
+1 -1
View File
@@ -4,7 +4,6 @@
#pragma once
#include <cstring>
#include <utility>
#ifdef _MSC_VER
@@ -13,6 +12,7 @@
#pragma intrinsic(_umul128)
#pragma intrinsic(_udiv128)
#else
#include <cstring>
#include <x86intrin.h>
#endif
-1
View File
@@ -7,7 +7,6 @@
#include <array>
#include <functional>
#include <string>
#include <string_view>
#include "common/common_types.h"
-1
View File
@@ -4,7 +4,6 @@
#pragma once
#include <type_traits>
#include <utility>
namespace Common {
-2
View File
@@ -2,8 +2,6 @@
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
#include <cstdint>
#include "common/uint128.h"
#include "common/wall_clock.h"
+66 -55
View File
@@ -1,8 +1,11 @@
// Copyright 2013 Dolphin Emulator Project / 2015 Citra Emulator Project
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
// Copyright 2013 Dolphin Emulator Project / 2015 Citra Emulator Project / 2022 Yuzu Emulator
// Project Licensed under GPLv2 or any later version Refer to the license.txt file included.
#include <array>
#include <cstring>
#include <iterator>
#include <string_view>
#include "common/bit_util.h"
#include "common/common_types.h"
#include "common/x64/cpu_detect.h"
@@ -17,7 +20,7 @@
// clang-format on
#endif
static inline void __cpuidex(int info[4], int function_id, int subfunction_id) {
static inline void __cpuidex(int info[4], u32 function_id, u32 subfunction_id) {
#if defined(__DragonFly__) || defined(__FreeBSD__)
// Despite the name, this is just do_cpuid() with ECX as second input.
cpuid_count((u_int)function_id, (u_int)subfunction_id, (u_int*)info);
@@ -30,7 +33,7 @@ static inline void __cpuidex(int info[4], int function_id, int subfunction_id) {
#endif
}
static inline void __cpuid(int info[4], int function_id) {
static inline void __cpuid(int info[4], u32 function_id) {
return __cpuidex(info, function_id, 0);
}
@@ -45,6 +48,17 @@ static inline u64 _xgetbv(u32 index) {
namespace Common {
CPUCaps::Manufacturer CPUCaps::ParseManufacturer(std::string_view brand_string) {
if (brand_string == "GenuineIntel") {
return Manufacturer::Intel;
} else if (brand_string == "AuthenticAMD") {
return Manufacturer::AMD;
} else if (brand_string == "HygonGenuine") {
return Manufacturer::Hygon;
}
return Manufacturer::Unknown;
}
// Detects the various CPU features
static CPUCaps Detect() {
CPUCaps caps = {};
@@ -53,75 +67,74 @@ static CPUCaps Detect() {
// yuzu at all anyway
int cpu_id[4];
memset(caps.brand_string, 0, sizeof(caps.brand_string));
// Detect CPU's CPUID capabilities and grab CPU string
// Detect CPU's CPUID capabilities and grab manufacturer string
__cpuid(cpu_id, 0x00000000);
u32 max_std_fn = cpu_id[0]; // EAX
const u32 max_std_fn = cpu_id[0]; // EAX
std::memcpy(&caps.brand_string[0], &cpu_id[1], sizeof(int));
std::memcpy(&caps.brand_string[4], &cpu_id[3], sizeof(int));
std::memcpy(&caps.brand_string[8], &cpu_id[2], sizeof(int));
if (cpu_id[1] == 0x756e6547 && cpu_id[2] == 0x6c65746e && cpu_id[3] == 0x49656e69)
caps.manufacturer = Manufacturer::Intel;
else if (cpu_id[1] == 0x68747541 && cpu_id[2] == 0x444d4163 && cpu_id[3] == 0x69746e65)
caps.manufacturer = Manufacturer::AMD;
else if (cpu_id[1] == 0x6f677948 && cpu_id[2] == 0x656e6975 && cpu_id[3] == 0x6e65476e)
caps.manufacturer = Manufacturer::Hygon;
else
caps.manufacturer = Manufacturer::Unknown;
std::memset(caps.brand_string, 0, std::size(caps.brand_string));
std::memcpy(&caps.brand_string[0], &cpu_id[1], sizeof(u32));
std::memcpy(&caps.brand_string[4], &cpu_id[3], sizeof(u32));
std::memcpy(&caps.brand_string[8], &cpu_id[2], sizeof(u32));
caps.manufacturer = CPUCaps::ParseManufacturer(caps.brand_string);
// Set reasonable default cpu string even if brand string not available
std::strncpy(caps.cpu_string, caps.brand_string, std::size(caps.brand_string));
__cpuid(cpu_id, 0x80000000);
u32 max_ex_fn = cpu_id[0];
// Set reasonable default brand string even if brand string not available
strcpy(caps.cpu_string, caps.brand_string);
const u32 max_ex_fn = cpu_id[0];
// Detect family and other miscellaneous features
if (max_std_fn >= 1) {
__cpuid(cpu_id, 0x00000001);
if ((cpu_id[3] >> 25) & 1)
caps.sse = true;
if ((cpu_id[3] >> 26) & 1)
caps.sse2 = true;
if ((cpu_id[2]) & 1)
caps.sse3 = true;
if ((cpu_id[2] >> 9) & 1)
caps.ssse3 = true;
if ((cpu_id[2] >> 19) & 1)
caps.sse4_1 = true;
if ((cpu_id[2] >> 20) & 1)
caps.sse4_2 = true;
if ((cpu_id[2] >> 25) & 1)
caps.aes = true;
caps.sse = Common::Bit<25>(cpu_id[3]);
caps.sse2 = Common::Bit<26>(cpu_id[3]);
caps.sse3 = Common::Bit<0>(cpu_id[2]);
caps.pclmulqdq = Common::Bit<1>(cpu_id[2]);
caps.ssse3 = Common::Bit<9>(cpu_id[2]);
caps.sse4_1 = Common::Bit<19>(cpu_id[2]);
caps.sse4_2 = Common::Bit<20>(cpu_id[2]);
caps.movbe = Common::Bit<22>(cpu_id[2]);
caps.popcnt = Common::Bit<23>(cpu_id[2]);
caps.aes = Common::Bit<25>(cpu_id[2]);
caps.f16c = Common::Bit<29>(cpu_id[2]);
// AVX support requires 3 separate checks:
// - Is the AVX bit set in CPUID?
// - Is the XSAVE bit set in CPUID?
// - XGETBV result has the XCR bit set.
if (((cpu_id[2] >> 28) & 1) && ((cpu_id[2] >> 27) & 1)) {
if (Common::Bit<28>(cpu_id[2]) && Common::Bit<27>(cpu_id[2])) {
if ((_xgetbv(_XCR_XFEATURE_ENABLED_MASK) & 0x6) == 0x6) {
caps.avx = true;
if ((cpu_id[2] >> 12) & 1)
if (Common::Bit<12>(cpu_id[2]))
caps.fma = true;
}
}
if (max_std_fn >= 7) {
__cpuidex(cpu_id, 0x00000007, 0x00000000);
// Can't enable AVX2 unless the XSAVE/XGETBV checks above passed
if ((cpu_id[1] >> 5) & 1)
caps.avx2 = caps.avx;
if ((cpu_id[1] >> 3) & 1)
caps.bmi1 = true;
if ((cpu_id[1] >> 8) & 1)
caps.bmi2 = true;
// Checks for AVX512F, AVX512CD, AVX512VL, AVX512DQ, AVX512BW (Intel Skylake-X/SP)
if ((cpu_id[1] >> 16) & 1 && (cpu_id[1] >> 28) & 1 && (cpu_id[1] >> 31) & 1 &&
(cpu_id[1] >> 17) & 1 && (cpu_id[1] >> 30) & 1) {
caps.avx512 = caps.avx2;
// Can't enable AVX{2,512} unless the XSAVE/XGETBV checks above passed
if (caps.avx) {
caps.avx2 = Common::Bit<5>(cpu_id[1]);
caps.avx512f = Common::Bit<16>(cpu_id[1]);
caps.avx512dq = Common::Bit<17>(cpu_id[1]);
caps.avx512cd = Common::Bit<28>(cpu_id[1]);
caps.avx512bw = Common::Bit<30>(cpu_id[1]);
caps.avx512vl = Common::Bit<31>(cpu_id[1]);
caps.avx512vbmi = Common::Bit<1>(cpu_id[2]);
caps.avx512bitalg = Common::Bit<12>(cpu_id[2]);
}
caps.bmi1 = Common::Bit<3>(cpu_id[1]);
caps.bmi2 = Common::Bit<8>(cpu_id[1]);
caps.sha = Common::Bit<29>(cpu_id[1]);
caps.gfni = Common::Bit<8>(cpu_id[2]);
__cpuidex(cpu_id, 0x00000007, 0x00000001);
caps.avx_vnni = caps.avx && Common::Bit<4>(cpu_id[0]);
}
}
@@ -138,15 +151,13 @@ static CPUCaps Detect() {
if (max_ex_fn >= 0x80000001) {
// Check for more features
__cpuid(cpu_id, 0x80000001);
if ((cpu_id[2] >> 16) & 1)
caps.fma4 = true;
caps.lzcnt = Common::Bit<5>(cpu_id[2]);
caps.fma4 = Common::Bit<16>(cpu_id[2]);
}
if (max_ex_fn >= 0x80000007) {
__cpuid(cpu_id, 0x80000007);
if (cpu_id[3] & (1 << 8)) {
caps.invariant_tsc = true;
}
caps.invariant_tsc = Common::Bit<8>(cpu_id[3]);
}
if (max_std_fn >= 0x16) {
+51 -28
View File
@@ -1,42 +1,65 @@
// Copyright 2013 Dolphin Emulator Project / 2015 Citra Emulator Project
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
// Copyright 2013 Dolphin Emulator Project / 2015 Citra Emulator Project / 2022 Yuzu Emulator
// Project Project Licensed under GPLv2 or any later version Refer to the license.txt file included.
#pragma once
namespace Common {
#include <string_view>
#include "common/common_types.h"
enum class Manufacturer : u32 {
Intel = 0,
AMD = 1,
Hygon = 2,
Unknown = 3,
};
namespace Common {
/// x86/x64 CPU capabilities that may be detected by this module
struct CPUCaps {
enum class Manufacturer : u8 {
Unknown = 0,
Intel = 1,
AMD = 2,
Hygon = 3,
};
static Manufacturer ParseManufacturer(std::string_view brand_string);
Manufacturer manufacturer;
char cpu_string[0x21];
char brand_string[0x41];
bool sse;
bool sse2;
bool sse3;
bool ssse3;
bool sse4_1;
bool sse4_2;
bool lzcnt;
bool avx;
bool avx2;
bool avx512;
bool bmi1;
bool bmi2;
bool fma;
bool fma4;
bool aes;
bool invariant_tsc;
char brand_string[13];
char cpu_string[48];
u32 base_frequency;
u32 max_frequency;
u32 bus_frequency;
bool sse : 1;
bool sse2 : 1;
bool sse3 : 1;
bool ssse3 : 1;
bool sse4_1 : 1;
bool sse4_2 : 1;
bool avx : 1;
bool avx_vnni : 1;
bool avx2 : 1;
bool avx512f : 1;
bool avx512dq : 1;
bool avx512cd : 1;
bool avx512bw : 1;
bool avx512vl : 1;
bool avx512vbmi : 1;
bool avx512bitalg : 1;
bool aes : 1;
bool bmi1 : 1;
bool bmi2 : 1;
bool f16c : 1;
bool fma : 1;
bool fma4 : 1;
bool gfni : 1;
bool invariant_tsc : 1;
bool lzcnt : 1;
bool movbe : 1;
bool pclmulqdq : 1;
bool popcnt : 1;
bool sha : 1;
};
/**
-2
View File
@@ -4,8 +4,6 @@
#include <array>
#include <chrono>
#include <limits>
#include <mutex>
#include <thread>
#include "common/atomic_ops.h"
-2
View File
@@ -4,8 +4,6 @@
#pragma once
#include <optional>
#include "common/wall_clock.h"
namespace Common {
+11
View File
@@ -122,6 +122,8 @@ add_library(core STATIC
frontend/applets/error.h
frontend/applets/general_frontend.cpp
frontend/applets/general_frontend.h
frontend/applets/mii_edit.cpp
frontend/applets/mii_edit.h
frontend/applets/profile_select.cpp
frontend/applets/profile_select.h
frontend/applets/software_keyboard.cpp
@@ -152,6 +154,7 @@ add_library(core STATIC
hle/api_version.h
hle/ipc.h
hle/ipc_helpers.h
hle/kernel/board/nintendo/nx/k_memory_layout.h
hle/kernel/board/nintendo/nx/k_system_control.cpp
hle/kernel/board/nintendo/nx/k_system_control.h
hle/kernel/board/nintendo/nx/secure_monitor.h
@@ -164,6 +167,7 @@ add_library(core STATIC
hle/kernel/hle_ipc.h
hle/kernel/init/init_slab_setup.cpp
hle/kernel/init/init_slab_setup.h
hle/kernel/initial_process.h
hle/kernel/k_address_arbiter.cpp
hle/kernel/k_address_arbiter.h
hle/kernel/k_address_space_info.cpp
@@ -205,6 +209,8 @@ add_library(core STATIC
hle/kernel/k_memory_region.h
hle/kernel/k_memory_region_type.h
hle/kernel/k_page_bitmap.h
hle/kernel/k_page_buffer.cpp
hle/kernel/k_page_buffer.h
hle/kernel/k_page_heap.cpp
hle/kernel/k_page_heap.h
hle/kernel/k_page_linked_list.h
@@ -242,6 +248,8 @@ add_library(core STATIC
hle/kernel/k_system_control.h
hle/kernel/k_thread.cpp
hle/kernel/k_thread.h
hle/kernel/k_thread_local_page.cpp
hle/kernel/k_thread_local_page.h
hle/kernel/k_thread_queue.cpp
hle/kernel/k_thread_queue.h
hle/kernel/k_trace.h
@@ -298,6 +306,9 @@ add_library(core STATIC
hle/service/am/applets/applet_error.h
hle/service/am/applets/applet_general_backend.cpp
hle/service/am/applets/applet_general_backend.h
hle/service/am/applets/applet_mii_edit.cpp
hle/service/am/applets/applet_mii_edit.h
hle/service/am/applets/applet_mii_edit_types.h
hle/service/am/applets/applet_profile_select.cpp
hle/service/am/applets/applet_profile_select.h
hle/service/am/applets/applet_software_keyboard.cpp
+12
View File
@@ -137,6 +137,8 @@ std::shared_ptr<Dynarmic::A32::Jit> ARM_Dynarmic_32::MakeJit(Common::PageTable*
config.page_table_pointer_mask_bits = Common::PageTable::ATTRIBUTE_BITS;
config.detect_misaligned_access_via_page_table = 16 | 32 | 64 | 128;
config.only_detect_misalignment_via_page_table_on_page_boundary = true;
config.fastmem_exclusive_access = true;
config.recompile_on_exclusive_fastmem_failure = true;
// Multi-process state
config.processor_id = core_index;
@@ -178,6 +180,12 @@ std::shared_ptr<Dynarmic::A32::Jit> ARM_Dynarmic_32::MakeJit(Common::PageTable*
if (!Settings::values.cpuopt_fastmem) {
config.fastmem_pointer = nullptr;
}
if (!Settings::values.cpuopt_fastmem_exclusives) {
config.fastmem_exclusive_access = false;
}
if (!Settings::values.cpuopt_recompile_exclusives) {
config.recompile_on_exclusive_fastmem_failure = false;
}
}
// Unsafe optimizations
@@ -195,6 +203,9 @@ std::shared_ptr<Dynarmic::A32::Jit> ARM_Dynarmic_32::MakeJit(Common::PageTable*
if (Settings::values.cpuopt_unsafe_inaccurate_nan) {
config.optimizations |= Dynarmic::OptimizationFlag::Unsafe_InaccurateNaN;
}
if (Settings::values.cpuopt_unsafe_ignore_global_monitor) {
config.optimizations |= Dynarmic::OptimizationFlag::Unsafe_IgnoreGlobalMonitor;
}
}
// Curated optimizations
@@ -203,6 +214,7 @@ std::shared_ptr<Dynarmic::A32::Jit> ARM_Dynarmic_32::MakeJit(Common::PageTable*
config.optimizations |= Dynarmic::OptimizationFlag::Unsafe_UnfuseFMA;
config.optimizations |= Dynarmic::OptimizationFlag::Unsafe_IgnoreStandardFPCRValue;
config.optimizations |= Dynarmic::OptimizationFlag::Unsafe_InaccurateNaN;
config.optimizations |= Dynarmic::OptimizationFlag::Unsafe_IgnoreGlobalMonitor;
}
return std::make_unique<Dynarmic::A32::Jit>(config);
+13
View File
@@ -185,6 +185,9 @@ std::shared_ptr<Dynarmic::A64::Jit> ARM_Dynarmic_64::MakeJit(Common::PageTable*
config.fastmem_pointer = page_table->fastmem_arena;
config.fastmem_address_space_bits = address_space_bits;
config.silently_mirror_fastmem = false;
config.fastmem_exclusive_access = true;
config.recompile_on_exclusive_fastmem_failure = true;
}
// Multi-process state
@@ -237,6 +240,12 @@ std::shared_ptr<Dynarmic::A64::Jit> ARM_Dynarmic_64::MakeJit(Common::PageTable*
if (!Settings::values.cpuopt_fastmem) {
config.fastmem_pointer = nullptr;
}
if (!Settings::values.cpuopt_fastmem_exclusives) {
config.fastmem_exclusive_access = false;
}
if (!Settings::values.cpuopt_recompile_exclusives) {
config.recompile_on_exclusive_fastmem_failure = false;
}
}
// Unsafe optimizations
@@ -254,6 +263,9 @@ std::shared_ptr<Dynarmic::A64::Jit> ARM_Dynarmic_64::MakeJit(Common::PageTable*
if (Settings::values.cpuopt_unsafe_fastmem_check) {
config.fastmem_address_space_bits = 64;
}
if (Settings::values.cpuopt_unsafe_ignore_global_monitor) {
config.optimizations |= Dynarmic::OptimizationFlag::Unsafe_IgnoreGlobalMonitor;
}
}
// Curated optimizations
@@ -262,6 +274,7 @@ std::shared_ptr<Dynarmic::A64::Jit> ARM_Dynarmic_64::MakeJit(Common::PageTable*
config.optimizations |= Dynarmic::OptimizationFlag::Unsafe_UnfuseFMA;
config.optimizations |= Dynarmic::OptimizationFlag::Unsafe_InaccurateNaN;
config.fastmem_address_space_bits = 64;
config.optimizations |= Dynarmic::OptimizationFlag::Unsafe_IgnoreGlobalMonitor;
}
return std::make_shared<Dynarmic::A64::Jit>(config);
@@ -37,8 +37,8 @@ u128 DynarmicExclusiveMonitor::ExclusiveRead128(std::size_t core_index, VAddr ad
});
}
void DynarmicExclusiveMonitor::ClearExclusive() {
monitor.Clear();
void DynarmicExclusiveMonitor::ClearExclusive(std::size_t core_index) {
monitor.ClearProcessor(core_index);
}
bool DynarmicExclusiveMonitor::ExclusiveWrite8(std::size_t core_index, VAddr vaddr, u8 value) {
@@ -4,8 +4,6 @@
#pragma once
#include <unordered_map>
#include <dynarmic/interface/exclusive_monitor.h>
#include "common/common_types.h"
@@ -29,7 +27,7 @@ public:
u32 ExclusiveRead32(std::size_t core_index, VAddr addr) override;
u64 ExclusiveRead64(std::size_t core_index, VAddr addr) override;
u128 ExclusiveRead128(std::size_t core_index, VAddr addr) override;
void ClearExclusive() override;
void ClearExclusive(std::size_t core_index) override;
bool ExclusiveWrite8(std::size_t core_index, VAddr vaddr, u8 value) override;
bool ExclusiveWrite16(std::size_t core_index, VAddr vaddr, u16 value) override;
+1 -1
View File
@@ -23,7 +23,7 @@ public:
virtual u32 ExclusiveRead32(std::size_t core_index, VAddr addr) = 0;
virtual u64 ExclusiveRead64(std::size_t core_index, VAddr addr) = 0;
virtual u128 ExclusiveRead128(std::size_t core_index, VAddr addr) = 0;
virtual void ClearExclusive() = 0;
virtual void ClearExclusive(std::size_t core_index) = 0;
virtual bool ExclusiveWrite8(std::size_t core_index, VAddr vaddr, u8 value) = 0;
virtual bool ExclusiveWrite16(std::size_t core_index, VAddr vaddr, u16 value) = 0;
+13 -3
View File
@@ -28,7 +28,9 @@
#include "core/file_sys/vfs_real.h"
#include "core/hardware_interrupt_manager.h"
#include "core/hid/hid_core.h"
#include "core/hle/kernel/k_memory_manager.h"
#include "core/hle/kernel/k_process.h"
#include "core/hle/kernel/k_resource_limit.h"
#include "core/hle/kernel/k_scheduler.h"
#include "core/hle/kernel/kernel.h"
#include "core/hle/kernel/physical_core.h"
@@ -36,7 +38,6 @@
#include "core/hle/service/apm/apm_controller.h"
#include "core/hle/service/filesystem/filesystem.h"
#include "core/hle/service/glue/glue_manager.h"
#include "core/hle/service/hid/hid.h"
#include "core/hle/service/service.h"
#include "core/hle/service/sm/sm.h"
#include "core/hle/service/time/time_manager.h"
@@ -252,9 +253,16 @@ struct System::Impl {
}
telemetry_session->AddInitialInfo(*app_loader, fs_controller, *content_provider);
// Create a resource limit for the process.
const auto physical_memory_size =
kernel.MemoryManager().GetSize(Kernel::KMemoryManager::Pool::Application);
auto* resource_limit = Kernel::CreateResourceLimitForProcess(system, physical_memory_size);
// Create the process.
auto main_process = Kernel::KProcess::Create(system.Kernel());
ASSERT(Kernel::KProcess::Initialize(main_process, system, "main",
Kernel::KProcess::ProcessType::Userland)
Kernel::KProcess::ProcessType::Userland, resource_limit)
.IsSuccess());
const auto [load_result, load_parameters] = app_loader->Load(*main_process, system);
if (load_result != Loader::ResultStatus::Success) {
@@ -317,7 +325,9 @@ struct System::Impl {
is_powered_on = false;
exit_lock = false;
gpu_core->NotifyShutdown();
if (gpu_core != nullptr) {
gpu_core->NotifyShutdown();
}
services.reset();
service_manager.reset();
+4 -1
View File
@@ -3,10 +3,13 @@
// Refer to the license.txt file included.
#include "core/device_memory.h"
#include "hle/kernel/board/nintendo/nx/k_system_control.h"
namespace Core {
DeviceMemory::DeviceMemory() : buffer{DramMemoryMap::Size, 1ULL << 39} {}
DeviceMemory::DeviceMemory()
: buffer{Kernel::Board::Nintendo::Nx::KSystemControl::Init::GetIntendedMemorySize(),
1ULL << 39} {}
DeviceMemory::~DeviceMemory() = default;
} // namespace Core
-4
View File
@@ -12,12 +12,8 @@ namespace Core {
namespace DramMemoryMap {
enum : u64 {
Base = 0x80000000ULL,
Size = 0x100000000ULL,
End = Base + Size,
KernelReserveBase = Base + 0x60000,
SlabHeapBase = KernelReserveBase + 0x85000,
SlapHeapSize = 0xa21000,
SlabHeapEnd = SlabHeapBase + SlapHeapSize,
};
}; // namespace DramMemoryMap
-1
View File
@@ -5,7 +5,6 @@
#pragma once
#include <cstddef>
#include <iterator>
#include "common/common_funcs.h"
#include "common/common_types.h"
+3
View File
@@ -10,7 +10,10 @@
#include "common/hex_util.h"
#include "common/logging/log.h"
#include "common/settings.h"
#ifndef _WIN32
#include "common/string_util.h"
#endif
#include "core/core.h"
#include "core/file_sys/common_funcs.h"
#include "core/file_sys/content_archive.h"
+2
View File
@@ -6,7 +6,9 @@
#include <array>
#include <vector>
#include "common/bit_field.h"
#include "common/common_funcs.h"
#include "common/common_types.h"
#include "common/swap.h"
#include "core/file_sys/vfs_types.h"
+18
View File
@@ -0,0 +1,18 @@
// Copyright 2022 yuzu Emulator Project
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
#include "common/logging/log.h"
#include "core/frontend/applets/mii_edit.h"
namespace Core::Frontend {
MiiEditApplet::~MiiEditApplet() = default;
void DefaultMiiEditApplet::ShowMiiEdit(const std::function<void()>& callback) const {
LOG_WARNING(Service_AM, "(STUBBED) called");
callback();
}
} // namespace Core::Frontend
+23
View File
@@ -0,0 +1,23 @@
// Copyright 2022 yuzu Emulator Project
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
#pragma once
#include <functional>
namespace Core::Frontend {
class MiiEditApplet {
public:
virtual ~MiiEditApplet();
virtual void ShowMiiEdit(const std::function<void()>& callback) const = 0;
};
class DefaultMiiEditApplet final : public MiiEditApplet {
public:
void ShowMiiEdit(const std::function<void()>& callback) const override;
};
} // namespace Core::Frontend
+10 -1
View File
@@ -42,11 +42,20 @@ public:
context.MakeCurrent();
}
~Scoped() {
context.DoneCurrent();
if (active) {
context.DoneCurrent();
}
}
/// In the event that context was destroyed before the Scoped is destroyed, this provides a
/// mechanism to prevent calling a destroyed object's method during the deconstructor
void Cancel() {
active = false;
}
private:
GraphicsContext& context;
bool active{true};
};
/// Calls MakeCurrent on the context and calls DoneCurrent when the scope for the returned value
+1 -1
View File
@@ -385,7 +385,7 @@ public:
T PopRaw();
template <class T>
std::shared_ptr<T> PopIpcInterface() {
std::weak_ptr<T> PopIpcInterface() {
ASSERT(context->Session()->IsDomain());
ASSERT(context->GetDomainMessageHeader().input_object_count > 0);
return context->GetDomainHandler<T>(Pop<u32>() - 1);
@@ -0,0 +1,13 @@
// Copyright 2022 yuzu Emulator Project
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
#pragma once
#include "common/common_types.h"
namespace Kernel {
constexpr inline PAddr MainMemoryAddress = 0x80000000;
} // namespace Kernel
@@ -5,6 +5,7 @@
#include <random>
#include "common/literals.h"
#include "common/settings.h"
#include "core/hle/kernel/board/nintendo/nx/k_system_control.h"
#include "core/hle/kernel/board/nintendo/nx/secure_monitor.h"
@@ -28,33 +29,20 @@ namespace {
using namespace Common::Literals;
u32 GetMemoryModeForInit() {
return 0x01;
}
u32 GetMemorySizeForInit() {
return 0;
return Settings::values.use_extended_memory_layout ? Smc::MemorySize_6GB : Smc::MemorySize_4GB;
}
Smc::MemoryArrangement GetMemoryArrangeForInit() {
switch (GetMemoryModeForInit() & 0x3F) {
case 0x01:
default:
return Smc::MemoryArrangement_4GB;
case 0x02:
return Smc::MemoryArrangement_4GBForAppletDev;
case 0x03:
return Smc::MemoryArrangement_4GBForSystemDev;
case 0x11:
return Smc::MemoryArrangement_6GB;
case 0x12:
return Smc::MemoryArrangement_6GBForAppletDev;
case 0x21:
return Smc::MemoryArrangement_8GB;
}
return Settings::values.use_extended_memory_layout ? Smc::MemoryArrangement_6GB
: Smc::MemoryArrangement_4GB;
}
} // namespace
size_t KSystemControl::Init::GetRealMemorySize() {
return GetIntendedMemorySize();
}
// Initialization.
size_t KSystemControl::Init::GetIntendedMemorySize() {
switch (GetMemorySizeForInit()) {
@@ -69,7 +57,13 @@ size_t KSystemControl::Init::GetIntendedMemorySize() {
}
PAddr KSystemControl::Init::GetKernelPhysicalBaseAddress(u64 base_address) {
return base_address;
const size_t real_dram_size = KSystemControl::Init::GetRealMemorySize();
const size_t intended_dram_size = KSystemControl::Init::GetIntendedMemorySize();
if (intended_dram_size * 2 < real_dram_size) {
return base_address;
} else {
return base_address + ((real_dram_size - intended_dram_size) / 2);
}
}
bool KSystemControl::Init::ShouldIncreaseThreadResourceLimit() {
@@ -13,6 +13,7 @@ public:
class Init {
public:
// Initialization.
static std::size_t GetRealMemorySize();
static std::size_t GetIntendedMemorySize();
static PAddr GetKernelPhysicalBaseAddress(u64 base_address);
static bool ShouldIncreaseThreadResourceLimit();
+1 -5
View File
@@ -17,7 +17,6 @@
#include "core/hle/kernel/k_auto_object.h"
#include "core/hle/kernel/k_handle_table.h"
#include "core/hle/kernel/k_process.h"
#include "core/hle/kernel/k_readable_event.h"
#include "core/hle/kernel/k_server_session.h"
#include "core/hle/kernel/k_thread.h"
#include "core/hle/kernel/kernel.h"
@@ -45,7 +44,7 @@ bool SessionRequestManager::HasSessionRequestHandler(const HLERequestContext& co
LOG_CRITICAL(IPC, "object_id {} is too big!", object_id);
return false;
}
return DomainHandler(object_id - 1) != nullptr;
return DomainHandler(object_id - 1).lock() != nullptr;
} else {
return session_handler != nullptr;
}
@@ -53,9 +52,6 @@ bool SessionRequestManager::HasSessionRequestHandler(const HLERequestContext& co
void SessionRequestHandler::ClientConnected(KServerSession* session) {
session->ClientConnected(shared_from_this());
// Ensure our server session is tracked globally.
kernel.RegisterServerSession(session);
}
void SessionRequestHandler::ClientDisconnected(KServerSession* session) {
+5 -4
View File
@@ -94,6 +94,7 @@ protected:
std::weak_ptr<ServiceThread> service_thread;
};
using SessionRequestHandlerWeakPtr = std::weak_ptr<SessionRequestHandler>;
using SessionRequestHandlerPtr = std::shared_ptr<SessionRequestHandler>;
/**
@@ -139,7 +140,7 @@ public:
}
}
SessionRequestHandlerPtr DomainHandler(std::size_t index) const {
SessionRequestHandlerWeakPtr DomainHandler(std::size_t index) const {
ASSERT_MSG(index < DomainHandlerCount(), "Unexpected handler index {}", index);
return domain_handlers.at(index);
}
@@ -328,10 +329,10 @@ public:
template <typename T>
std::shared_ptr<T> GetDomainHandler(std::size_t index) const {
return std::static_pointer_cast<T>(manager->DomainHandler(index));
return std::static_pointer_cast<T>(manager.lock()->DomainHandler(index).lock());
}
void SetSessionRequestManager(std::shared_ptr<SessionRequestManager> manager_) {
void SetSessionRequestManager(std::weak_ptr<SessionRequestManager> manager_) {
manager = std::move(manager_);
}
@@ -374,7 +375,7 @@ private:
u32 handles_offset{};
u32 domain_offset{};
std::shared_ptr<SessionRequestManager> manager;
std::weak_ptr<SessionRequestManager> manager;
KernelCore& kernel;
Core::Memory::Memory& memory;
+80 -21
View File
@@ -7,19 +7,23 @@
#include "common/common_funcs.h"
#include "common/common_types.h"
#include "core/core.h"
#include "core/device_memory.h"
#include "core/hardware_properties.h"
#include "core/hle/kernel/init/init_slab_setup.h"
#include "core/hle/kernel/k_code_memory.h"
#include "core/hle/kernel/k_event.h"
#include "core/hle/kernel/k_memory_layout.h"
#include "core/hle/kernel/k_memory_manager.h"
#include "core/hle/kernel/k_page_buffer.h"
#include "core/hle/kernel/k_port.h"
#include "core/hle/kernel/k_process.h"
#include "core/hle/kernel/k_resource_limit.h"
#include "core/hle/kernel/k_session.h"
#include "core/hle/kernel/k_shared_memory.h"
#include "core/hle/kernel/k_shared_memory_info.h"
#include "core/hle/kernel/k_system_control.h"
#include "core/hle/kernel/k_thread.h"
#include "core/hle/kernel/k_thread_local_page.h"
#include "core/hle/kernel/k_transfer_memory.h"
namespace Kernel::Init {
@@ -32,9 +36,13 @@ namespace Kernel::Init {
HANDLER(KEvent, (SLAB_COUNT(KEvent)), ##__VA_ARGS__) \
HANDLER(KPort, (SLAB_COUNT(KPort)), ##__VA_ARGS__) \
HANDLER(KSharedMemory, (SLAB_COUNT(KSharedMemory)), ##__VA_ARGS__) \
HANDLER(KSharedMemoryInfo, (SLAB_COUNT(KSharedMemory) * 8), ##__VA_ARGS__) \
HANDLER(KTransferMemory, (SLAB_COUNT(KTransferMemory)), ##__VA_ARGS__) \
HANDLER(KCodeMemory, (SLAB_COUNT(KCodeMemory)), ##__VA_ARGS__) \
HANDLER(KSession, (SLAB_COUNT(KSession)), ##__VA_ARGS__) \
HANDLER(KThreadLocalPage, \
(SLAB_COUNT(KProcess) + (SLAB_COUNT(KProcess) + SLAB_COUNT(KThread)) / 8), \
##__VA_ARGS__) \
HANDLER(KResourceLimit, (SLAB_COUNT(KResourceLimit)), ##__VA_ARGS__)
namespace {
@@ -50,38 +58,46 @@ enum KSlabType : u32 {
// Constexpr counts.
constexpr size_t SlabCountKProcess = 80;
constexpr size_t SlabCountKThread = 800;
constexpr size_t SlabCountKEvent = 700;
constexpr size_t SlabCountKEvent = 900;
constexpr size_t SlabCountKInterruptEvent = 100;
constexpr size_t SlabCountKPort = 256 + 0x20; // Extra 0x20 ports over Nintendo for homebrew.
constexpr size_t SlabCountKPort = 384;
constexpr size_t SlabCountKSharedMemory = 80;
constexpr size_t SlabCountKTransferMemory = 200;
constexpr size_t SlabCountKCodeMemory = 10;
constexpr size_t SlabCountKDeviceAddressSpace = 300;
constexpr size_t SlabCountKSession = 933;
constexpr size_t SlabCountKSession = 1133;
constexpr size_t SlabCountKLightSession = 100;
constexpr size_t SlabCountKObjectName = 7;
constexpr size_t SlabCountKResourceLimit = 5;
constexpr size_t SlabCountKDebug = Core::Hardware::NUM_CPU_CORES;
constexpr size_t SlabCountKAlpha = 1;
constexpr size_t SlabCountKBeta = 6;
constexpr size_t SlabCountKIoPool = 1;
constexpr size_t SlabCountKIoRegion = 6;
constexpr size_t SlabCountExtraKThread = 160;
/// Helper function to translate from the slab virtual address to the reserved location in physical
/// memory.
static PAddr TranslateSlabAddrToPhysical(KMemoryLayout& memory_layout, VAddr slab_addr) {
slab_addr -= memory_layout.GetSlabRegionAddress();
return slab_addr + Core::DramMemoryMap::SlabHeapBase;
}
template <typename T>
VAddr InitializeSlabHeap(Core::System& system, KMemoryLayout& memory_layout, VAddr address,
size_t num_objects) {
// TODO(bunnei): This is just a place holder. We should initialize the appropriate KSlabHeap for
// kernel object type T with the backing kernel memory pointer once we emulate kernel memory.
const size_t size = Common::AlignUp(sizeof(T) * num_objects, alignof(void*));
VAddr start = Common::AlignUp(address, alignof(T));
// This is intentionally empty. Once KSlabHeap is fully implemented, we can replace this with
// the pointer to emulated memory to pass along. Until then, KSlabHeap will just allocate/free
// host memory.
void* backing_kernel_memory{};
// This should use the virtual memory address passed in, but currently, we do not setup the
// kernel virtual memory layout. Instead, we simply map these at a region of physical memory
// that we reserve for the slab heaps.
// TODO(bunnei): Fix this once we support the kernel virtual memory layout.
if (size > 0) {
void* backing_kernel_memory{
system.DeviceMemory().GetPointer(TranslateSlabAddrToPhysical(memory_layout, start))};
const KMemoryRegion* region = memory_layout.FindVirtual(start + size - 1);
ASSERT(region != nullptr);
ASSERT(region->IsDerivedFrom(KMemoryRegionType_KernelSlab));
@@ -91,6 +107,12 @@ VAddr InitializeSlabHeap(Core::System& system, KMemoryLayout& memory_layout, VAd
return start + size;
}
size_t CalculateSlabHeapGapSize() {
constexpr size_t KernelSlabHeapGapSize = 2_MiB - 296_KiB;
static_assert(KernelSlabHeapGapSize <= KernelSlabHeapGapsSizeMax);
return KernelSlabHeapGapSize;
}
} // namespace
KSlabResourceCounts KSlabResourceCounts::CreateDefault() {
@@ -109,8 +131,8 @@ KSlabResourceCounts KSlabResourceCounts::CreateDefault() {
.num_KObjectName = SlabCountKObjectName,
.num_KResourceLimit = SlabCountKResourceLimit,
.num_KDebug = SlabCountKDebug,
.num_KAlpha = SlabCountKAlpha,
.num_KBeta = SlabCountKBeta,
.num_KIoPool = SlabCountKIoPool,
.num_KIoRegion = SlabCountKIoRegion,
};
}
@@ -136,11 +158,34 @@ size_t CalculateTotalSlabHeapSize(const KernelCore& kernel) {
#undef ADD_SLAB_SIZE
// Add the reserved size.
size += KernelSlabHeapGapsSize;
size += CalculateSlabHeapGapSize();
return size;
}
void InitializeKPageBufferSlabHeap(Core::System& system) {
auto& kernel = system.Kernel();
const auto& counts = kernel.SlabResourceCounts();
const size_t num_pages =
counts.num_KProcess + counts.num_KThread + (counts.num_KProcess + counts.num_KThread) / 8;
const size_t slab_size = num_pages * PageSize;
// Reserve memory from the system resource limit.
ASSERT(kernel.GetSystemResourceLimit()->Reserve(LimitableResource::PhysicalMemory, slab_size));
// Allocate memory for the slab.
constexpr auto AllocateOption = KMemoryManager::EncodeOption(
KMemoryManager::Pool::System, KMemoryManager::Direction::FromFront);
const PAddr slab_address =
kernel.MemoryManager().AllocateAndOpenContinuous(num_pages, 1, AllocateOption);
ASSERT(slab_address != 0);
// Initialize the slabheap.
KPageBuffer::InitializeSlabHeap(kernel, system.DeviceMemory().GetPointer(slab_address),
slab_size);
}
void InitializeSlabHeaps(Core::System& system, KMemoryLayout& memory_layout) {
auto& kernel = system.Kernel();
@@ -160,13 +205,13 @@ void InitializeSlabHeaps(Core::System& system, KMemoryLayout& memory_layout) {
}
// Create an array to represent the gaps between the slabs.
const size_t total_gap_size = KernelSlabHeapGapsSize;
const size_t total_gap_size = CalculateSlabHeapGapSize();
std::array<size_t, slab_types.size()> slab_gaps;
for (size_t i = 0; i < slab_gaps.size(); i++) {
for (auto& slab_gap : slab_gaps) {
// Note: This is an off-by-one error from Nintendo's intention, because GenerateRandomRange
// is inclusive. However, Nintendo also has the off-by-one error, and it's "harmless", so we
// will include it ourselves.
slab_gaps[i] = KSystemControl::GenerateRandomRange(0, total_gap_size);
slab_gap = KSystemControl::GenerateRandomRange(0, total_gap_size);
}
// Sort the array, so that we can treat differences between values as offsets to the starts of
@@ -177,13 +222,21 @@ void InitializeSlabHeaps(Core::System& system, KMemoryLayout& memory_layout) {
}
}
for (size_t i = 0; i < slab_types.size(); i++) {
// Track the gaps, so that we can free them to the unused slab tree.
VAddr gap_start = address;
size_t gap_size = 0;
for (size_t i = 0; i < slab_gaps.size(); i++) {
// Add the random gap to the address.
address += (i == 0) ? slab_gaps[0] : slab_gaps[i] - slab_gaps[i - 1];
const auto cur_gap = (i == 0) ? slab_gaps[0] : slab_gaps[i] - slab_gaps[i - 1];
address += cur_gap;
gap_size += cur_gap;
#define INITIALIZE_SLAB_HEAP(NAME, COUNT, ...) \
case KSlabType_##NAME: \
address = InitializeSlabHeap<NAME>(system, memory_layout, address, COUNT); \
if (COUNT > 0) { \
address = InitializeSlabHeap<NAME>(system, memory_layout, address, COUNT); \
} \
break;
// Initialize the slabheap.
@@ -192,7 +245,13 @@ void InitializeSlabHeaps(Core::System& system, KMemoryLayout& memory_layout) {
FOREACH_SLAB_TYPE(INITIALIZE_SLAB_HEAP)
// If we somehow get an invalid type, abort.
default:
UNREACHABLE();
UNREACHABLE_MSG("Unknown slab type: {}", slab_types[i]);
}
// If we've hit the end of a gap, free it.
if (gap_start + gap_size != address) {
gap_start = address;
gap_size = 0;
}
}
}
+3 -2
View File
@@ -32,12 +32,13 @@ struct KSlabResourceCounts {
size_t num_KObjectName;
size_t num_KResourceLimit;
size_t num_KDebug;
size_t num_KAlpha;
size_t num_KBeta;
size_t num_KIoPool;
size_t num_KIoRegion;
};
void InitializeSlabResourceCounts(KernelCore& kernel);
size_t CalculateTotalSlabHeapSize(const KernelCore& kernel);
void InitializeKPageBufferSlabHeap(Core::System& system);
void InitializeSlabHeaps(Core::System& system, KMemoryLayout& memory_layout);
} // namespace Kernel::Init
+23
View File
@@ -0,0 +1,23 @@
// Copyright 2022 yuzu Emulator Project
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
#pragma once
#include "common/common_types.h"
#include "common/literals.h"
#include "core/hle/kernel/board/nintendo/nx/k_memory_layout.h"
#include "core/hle/kernel/board/nintendo/nx/k_system_control.h"
namespace Kernel {
using namespace Common::Literals;
constexpr std::size_t InitialProcessBinarySizeMax = 12_MiB;
static inline PAddr GetInitialProcessBinaryPhysicalAddress() {
return Kernel::Board::Nintendo::Nx::KSystemControl::Init::GetKernelPhysicalBaseAddress(
MainMemoryAddress);
}
} // namespace Kernel
+5 -5
View File
@@ -49,7 +49,7 @@ bool DecrementIfLessThan(Core::System& system, s32* out, VAddr address, s32 valu
}
} else {
// Otherwise, clear our exclusive hold and finish
monitor.ClearExclusive();
monitor.ClearExclusive(current_core);
}
// We're done.
@@ -78,7 +78,7 @@ bool UpdateIfEqual(Core::System& system, s32* out, VAddr address, s32 value, s32
}
} else {
// Otherwise, clear our exclusive hold and finish.
monitor.ClearExclusive();
monitor.ClearExclusive(current_core);
}
// We're done.
@@ -115,7 +115,7 @@ ResultCode KAddressArbiter::Signal(VAddr addr, s32 count) {
{
KScopedSchedulerLock sl(kernel);
auto it = thread_tree.nfind_light({addr, -1});
auto it = thread_tree.nfind_key({addr, -1});
while ((it != thread_tree.end()) && (count <= 0 || num_waiters < count) &&
(it->GetAddressArbiterKey() == addr)) {
// End the thread's wait.
@@ -148,7 +148,7 @@ ResultCode KAddressArbiter::SignalAndIncrementIfEqual(VAddr addr, s32 value, s32
return ResultInvalidState;
}
auto it = thread_tree.nfind_light({addr, -1});
auto it = thread_tree.nfind_key({addr, -1});
while ((it != thread_tree.end()) && (count <= 0 || num_waiters < count) &&
(it->GetAddressArbiterKey() == addr)) {
// End the thread's wait.
@@ -171,7 +171,7 @@ ResultCode KAddressArbiter::SignalAndModifyByWaitingCountIfEqual(VAddr addr, s32
{
[[maybe_unused]] const KScopedSchedulerLock sl(kernel);
auto it = thread_tree.nfind_light({addr, -1});
auto it = thread_tree.nfind_key({addr, -1});
// Determine the updated value.
s32 new_value{};
if (count <= 0) {
-2
View File
@@ -4,8 +4,6 @@
#pragma once
#include <atomic>
#include "common/bit_util.h"
#include "common/common_types.h"
-1
View File
@@ -5,7 +5,6 @@
#include "common/alignment.h"
#include "common/common_types.h"
#include "core/device_memory.h"
#include "core/hle/kernel/k_auto_object.h"
#include "core/hle/kernel/k_code_memory.h"
#include "core/hle/kernel/k_light_lock.h"
#include "core/hle/kernel/k_memory_block.h"
+1 -2
View File
@@ -9,7 +9,6 @@
#include "core/hle/kernel/k_process.h"
#include "core/hle/kernel/k_scheduler.h"
#include "core/hle/kernel/k_scoped_scheduler_lock_and_sleep.h"
#include "core/hle/kernel/k_synchronization_object.h"
#include "core/hle/kernel/k_thread.h"
#include "core/hle/kernel/k_thread_queue.h"
#include "core/hle/kernel/kernel.h"
@@ -244,7 +243,7 @@ void KConditionVariable::Signal(u64 cv_key, s32 count) {
{
KScopedSchedulerLock sl(kernel);
auto it = thread_tree.nfind_light({cv_key, -1});
auto it = thread_tree.nfind_key({cv_key, -1});
while ((it != thread_tree.end()) && (count <= 0 || num_waiters < count) &&
(it->GetConditionVariableKey() == cv_key)) {
KThread* target_thread = std::addressof(*it);
+5 -9
View File
@@ -14,7 +14,7 @@ KEvent::KEvent(KernelCore& kernel_)
KEvent::~KEvent() = default;
void KEvent::Initialize(std::string&& name_) {
void KEvent::Initialize(std::string&& name_, KProcess* owner_) {
// Increment reference count.
// Because reference count is one on creation, this will result
// in a reference count of two. Thus, when both readable and
@@ -30,10 +30,8 @@ void KEvent::Initialize(std::string&& name_) {
writable_event.Initialize(this, name_ + ":Writable");
// Set our owner process.
owner = kernel.CurrentProcess();
if (owner) {
owner->Open();
}
owner = owner_;
owner->Open();
// Mark initialized.
name = std::move(name_);
@@ -47,10 +45,8 @@ void KEvent::Finalize() {
void KEvent::PostDestroy(uintptr_t arg) {
// Release the event count resource the owner process holds.
KProcess* owner = reinterpret_cast<KProcess*>(arg);
if (owner) {
owner->GetResourceLimit()->Release(LimitableResource::Events, 1);
owner->Close();
}
owner->GetResourceLimit()->Release(LimitableResource::Events, 1);
owner->Close();
}
} // namespace Kernel
+1 -1
View File
@@ -22,7 +22,7 @@ public:
explicit KEvent(KernelCore& kernel_);
~KEvent() override;
void Initialize(std::string&& name);
void Initialize(std::string&& name, KProcess* owner_);
void Finalize() override;
+4 -4
View File
@@ -63,7 +63,7 @@ bool KHandleTable::Remove(Handle handle) {
return true;
}
ResultCode KHandleTable::Add(Handle* out_handle, KAutoObject* obj, u16 type) {
ResultCode KHandleTable::Add(Handle* out_handle, KAutoObject* obj) {
KScopedDisableDispatch dd(kernel);
KScopedSpinLock lk(m_lock);
@@ -75,7 +75,7 @@ ResultCode KHandleTable::Add(Handle* out_handle, KAutoObject* obj, u16 type) {
const auto linear_id = this->AllocateLinearId();
const auto index = this->AllocateEntry();
m_entry_infos[index].info = {.linear_id = linear_id, .type = type};
m_entry_infos[index].linear_id = linear_id;
m_objects[index] = obj;
obj->Open();
@@ -116,7 +116,7 @@ void KHandleTable::Unreserve(Handle handle) {
}
}
void KHandleTable::Register(Handle handle, KAutoObject* obj, u16 type) {
void KHandleTable::Register(Handle handle, KAutoObject* obj) {
KScopedDisableDispatch dd(kernel);
KScopedSpinLock lk(m_lock);
@@ -132,7 +132,7 @@ void KHandleTable::Register(Handle handle, KAutoObject* obj, u16 type) {
// Set the entry.
ASSERT(m_objects[index] == nullptr);
m_entry_infos[index].info = {.linear_id = static_cast<u16>(linear_id), .type = type};
m_entry_infos[index].linear_id = static_cast<u16>(linear_id);
m_objects[index] = obj;
obj->Open();
+8 -26
View File
@@ -42,7 +42,7 @@ public:
m_free_head_index = -1;
// Free all entries.
for (s32 i = 0; i < static_cast<s32>(m_table_size); ++i) {
for (s16 i = 0; i < static_cast<s16>(m_table_size); ++i) {
m_objects[i] = nullptr;
m_entry_infos[i].next_free_index = i - 1;
m_free_head_index = i;
@@ -104,17 +104,8 @@ public:
ResultCode Reserve(Handle* out_handle);
void Unreserve(Handle handle);
template <typename T>
ResultCode Add(Handle* out_handle, T* obj) {
static_assert(std::is_base_of_v<KAutoObject, T>);
return this->Add(out_handle, obj, obj->GetTypeObj().GetClassToken());
}
template <typename T>
void Register(Handle handle, T* obj) {
static_assert(std::is_base_of_v<KAutoObject, T>);
return this->Register(handle, obj, obj->GetTypeObj().GetClassToken());
}
ResultCode Add(Handle* out_handle, KAutoObject* obj);
void Register(Handle handle, KAutoObject* obj);
template <typename T>
bool GetMultipleObjects(T** out, const Handle* handles, size_t num_handles) const {
@@ -160,9 +151,6 @@ public:
}
private:
ResultCode Add(Handle* out_handle, KAutoObject* obj, u16 type);
void Register(Handle handle, KAutoObject* obj, u16 type);
s32 AllocateEntry() {
ASSERT(m_count < m_table_size);
@@ -179,7 +167,7 @@ private:
ASSERT(m_count > 0);
m_objects[index] = nullptr;
m_entry_infos[index].next_free_index = m_free_head_index;
m_entry_infos[index].next_free_index = static_cast<s16>(m_free_head_index);
m_free_head_index = index;
@@ -278,19 +266,13 @@ private:
}
union EntryInfo {
struct {
u16 linear_id;
u16 type;
} info;
s32 next_free_index;
u16 linear_id;
s16 next_free_index;
constexpr u16 GetLinearId() const {
return info.linear_id;
return linear_id;
}
constexpr u16 GetType() const {
return info.type;
}
constexpr s32 GetNextFreeIndex() const {
constexpr s16 GetNextFreeIndex() const {
return next_free_index;
}
};
+7 -3
View File
@@ -57,11 +57,11 @@ constexpr std::size_t KernelPageTableHeapSize = GetMaximumOverheadSize(MainMemor
constexpr std::size_t KernelInitialPageHeapSize = 128_KiB;
constexpr std::size_t KernelSlabHeapDataSize = 5_MiB;
constexpr std::size_t KernelSlabHeapGapsSize = 2_MiB - 64_KiB;
constexpr std::size_t KernelSlabHeapSize = KernelSlabHeapDataSize + KernelSlabHeapGapsSize;
constexpr std::size_t KernelSlabHeapGapsSizeMax = 2_MiB - 64_KiB;
constexpr std::size_t KernelSlabHeapSize = KernelSlabHeapDataSize + KernelSlabHeapGapsSizeMax;
// NOTE: This is calculated from KThread slab counts, assuming KThread size <= 0x860.
constexpr std::size_t KernelSlabHeapAdditionalSize = 416_KiB;
constexpr std::size_t KernelSlabHeapAdditionalSize = 0x68000;
constexpr std::size_t KernelResourceSize =
KernelPageTableHeapSize + KernelInitialPageHeapSize + KernelSlabHeapSize;
@@ -173,6 +173,10 @@ public:
return Dereference(FindVirtualLinear(address));
}
const KMemoryRegion& GetPhysicalLinearRegion(PAddr address) const {
return Dereference(FindPhysicalLinear(address));
}
const KMemoryRegion* GetPhysicalKernelTraceBufferRegion() const {
return GetPhysicalMemoryRegionTree().FindFirstDerived(KMemoryRegionType_KernelTraceBuffer);
}
+353 -131
View File
@@ -10,189 +10,411 @@
#include "common/scope_exit.h"
#include "core/core.h"
#include "core/device_memory.h"
#include "core/hle/kernel/initial_process.h"
#include "core/hle/kernel/k_memory_manager.h"
#include "core/hle/kernel/k_page_linked_list.h"
#include "core/hle/kernel/kernel.h"
#include "core/hle/kernel/svc_results.h"
namespace Kernel {
KMemoryManager::KMemoryManager(Core::System& system_) : system{system_} {}
namespace {
std::size_t KMemoryManager::Impl::Initialize(Pool new_pool, u64 start_address, u64 end_address) {
const auto size{end_address - start_address};
// Calculate metadata sizes
const auto ref_count_size{(size / PageSize) * sizeof(u16)};
const auto optimize_map_size{(Common::AlignUp((size / PageSize), 64) / 64) * sizeof(u64)};
const auto manager_size{Common::AlignUp(optimize_map_size + ref_count_size, PageSize)};
const auto page_heap_size{KPageHeap::CalculateManagementOverheadSize(size)};
const auto total_metadata_size{manager_size + page_heap_size};
ASSERT(manager_size <= total_metadata_size);
ASSERT(Common::IsAligned(total_metadata_size, PageSize));
// Setup region
pool = new_pool;
// Initialize the manager's KPageHeap
heap.Initialize(start_address, size, page_heap_size);
// Free the memory to the heap
heap.Free(start_address, size / PageSize);
// Update the heap's used size
heap.UpdateUsedSize();
return total_metadata_size;
constexpr KMemoryManager::Pool GetPoolFromMemoryRegionType(u32 type) {
if ((type | KMemoryRegionType_DramApplicationPool) == type) {
return KMemoryManager::Pool::Application;
} else if ((type | KMemoryRegionType_DramAppletPool) == type) {
return KMemoryManager::Pool::Applet;
} else if ((type | KMemoryRegionType_DramSystemPool) == type) {
return KMemoryManager::Pool::System;
} else if ((type | KMemoryRegionType_DramSystemNonSecurePool) == type) {
return KMemoryManager::Pool::SystemNonSecure;
} else {
UNREACHABLE_MSG("InvalidMemoryRegionType for conversion to Pool");
return {};
}
}
void KMemoryManager::InitializeManager(Pool pool, u64 start_address, u64 end_address) {
ASSERT(pool < Pool::Count);
managers[static_cast<std::size_t>(pool)].Initialize(pool, start_address, end_address);
} // namespace
KMemoryManager::KMemoryManager(Core::System& system_)
: system{system_}, pool_locks{
KLightLock{system_.Kernel()},
KLightLock{system_.Kernel()},
KLightLock{system_.Kernel()},
KLightLock{system_.Kernel()},
} {}
void KMemoryManager::Initialize(VAddr management_region, size_t management_region_size) {
// Clear the management region to zero.
const VAddr management_region_end = management_region + management_region_size;
// Reset our manager count.
num_managers = 0;
// Traverse the virtual memory layout tree, initializing each manager as appropriate.
while (num_managers != MaxManagerCount) {
// Locate the region that should initialize the current manager.
PAddr region_address = 0;
size_t region_size = 0;
Pool region_pool = Pool::Count;
for (const auto& it : system.Kernel().MemoryLayout().GetPhysicalMemoryRegionTree()) {
// We only care about regions that we need to create managers for.
if (!it.IsDerivedFrom(KMemoryRegionType_DramUserPool)) {
continue;
}
// We want to initialize the managers in order.
if (it.GetAttributes() != num_managers) {
continue;
}
const PAddr cur_start = it.GetAddress();
const PAddr cur_end = it.GetEndAddress();
// Validate the region.
ASSERT(cur_end != 0);
ASSERT(cur_start != 0);
ASSERT(it.GetSize() > 0);
// Update the region's extents.
if (region_address == 0) {
region_address = cur_start;
region_size = it.GetSize();
region_pool = GetPoolFromMemoryRegionType(it.GetType());
} else {
ASSERT(cur_start == region_address + region_size);
// Update the size.
region_size = cur_end - region_address;
ASSERT(GetPoolFromMemoryRegionType(it.GetType()) == region_pool);
}
}
// If we didn't find a region, we're done.
if (region_size == 0) {
break;
}
// Initialize a new manager for the region.
Impl* manager = std::addressof(managers[num_managers++]);
ASSERT(num_managers <= managers.size());
const size_t cur_size = manager->Initialize(region_address, region_size, management_region,
management_region_end, region_pool);
management_region += cur_size;
ASSERT(management_region <= management_region_end);
// Insert the manager into the pool list.
const auto region_pool_index = static_cast<u32>(region_pool);
if (pool_managers_tail[region_pool_index] == nullptr) {
pool_managers_head[region_pool_index] = manager;
} else {
pool_managers_tail[region_pool_index]->SetNext(manager);
manager->SetPrev(pool_managers_tail[region_pool_index]);
}
pool_managers_tail[region_pool_index] = manager;
}
// Free each region to its corresponding heap.
size_t reserved_sizes[MaxManagerCount] = {};
const PAddr ini_start = GetInitialProcessBinaryPhysicalAddress();
const PAddr ini_end = ini_start + InitialProcessBinarySizeMax;
const PAddr ini_last = ini_end - 1;
for (const auto& it : system.Kernel().MemoryLayout().GetPhysicalMemoryRegionTree()) {
if (it.IsDerivedFrom(KMemoryRegionType_DramUserPool)) {
// Get the manager for the region.
auto index = it.GetAttributes();
auto& manager = managers[index];
const PAddr cur_start = it.GetAddress();
const PAddr cur_last = it.GetLastAddress();
const PAddr cur_end = it.GetEndAddress();
if (cur_start <= ini_start && ini_last <= cur_last) {
// Free memory before the ini to the heap.
if (cur_start != ini_start) {
manager.Free(cur_start, (ini_start - cur_start) / PageSize);
}
// Open/reserve the ini memory.
manager.OpenFirst(ini_start, InitialProcessBinarySizeMax / PageSize);
reserved_sizes[it.GetAttributes()] += InitialProcessBinarySizeMax;
// Free memory after the ini to the heap.
if (ini_last != cur_last) {
ASSERT(cur_end != 0);
manager.Free(ini_end, cur_end - ini_end);
}
} else {
// Ensure there's no partial overlap with the ini image.
if (cur_start <= ini_last) {
ASSERT(cur_last < ini_start);
} else {
// Otherwise, check the region for general validity.
ASSERT(cur_end != 0);
}
// Free the memory to the heap.
manager.Free(cur_start, it.GetSize() / PageSize);
}
}
}
// Update the used size for all managers.
for (size_t i = 0; i < num_managers; ++i) {
managers[i].SetInitialUsedHeapSize(reserved_sizes[i]);
}
}
VAddr KMemoryManager::AllocateAndOpenContinuous(std::size_t num_pages, std::size_t align_pages,
u32 option) {
// Early return if we're allocating no pages
PAddr KMemoryManager::AllocateAndOpenContinuous(size_t num_pages, size_t align_pages, u32 option) {
// Early return if we're allocating no pages.
if (num_pages == 0) {
return {};
return 0;
}
// Lock the pool that we're allocating from
// Lock the pool that we're allocating from.
const auto [pool, dir] = DecodeOption(option);
const auto pool_index{static_cast<std::size_t>(pool)};
std::lock_guard lock{pool_locks[pool_index]};
KScopedLightLock lk(pool_locks[static_cast<std::size_t>(pool)]);
// Choose a heap based on our page size request
const s32 heap_index{KPageHeap::GetAlignedBlockIndex(num_pages, align_pages)};
// Choose a heap based on our page size request.
const s32 heap_index = KPageHeap::GetAlignedBlockIndex(num_pages, align_pages);
// Loop, trying to iterate from each block
// TODO (bunnei): Support multiple managers
Impl& chosen_manager{managers[pool_index]};
VAddr allocated_block{chosen_manager.AllocateBlock(heap_index, false)};
// If we failed to allocate, quit now
if (!allocated_block) {
return {};
// Loop, trying to iterate from each block.
Impl* chosen_manager = nullptr;
PAddr allocated_block = 0;
for (chosen_manager = this->GetFirstManager(pool, dir); chosen_manager != nullptr;
chosen_manager = this->GetNextManager(chosen_manager, dir)) {
allocated_block = chosen_manager->AllocateBlock(heap_index, true);
if (allocated_block != 0) {
break;
}
}
// If we allocated more than we need, free some
const auto allocated_pages{KPageHeap::GetBlockNumPages(heap_index)};
// If we failed to allocate, quit now.
if (allocated_block == 0) {
return 0;
}
// If we allocated more than we need, free some.
const size_t allocated_pages = KPageHeap::GetBlockNumPages(heap_index);
if (allocated_pages > num_pages) {
chosen_manager.Free(allocated_block + num_pages * PageSize, allocated_pages - num_pages);
chosen_manager->Free(allocated_block + num_pages * PageSize, allocated_pages - num_pages);
}
// Open the first reference to the pages.
chosen_manager->OpenFirst(allocated_block, num_pages);
return allocated_block;
}
ResultCode KMemoryManager::Allocate(KPageLinkedList& page_list, std::size_t num_pages, Pool pool,
Direction dir, u32 heap_fill_value) {
ASSERT(page_list.GetNumPages() == 0);
ResultCode KMemoryManager::AllocatePageGroupImpl(KPageLinkedList* out, size_t num_pages, Pool pool,
Direction dir, bool random) {
// Choose a heap based on our page size request.
const s32 heap_index = KPageHeap::GetBlockIndex(num_pages);
R_UNLESS(0 <= heap_index, ResultOutOfMemory);
// Early return if we're allocating no pages
if (num_pages == 0) {
return ResultSuccess;
}
// Lock the pool that we're allocating from
const auto pool_index{static_cast<std::size_t>(pool)};
std::lock_guard lock{pool_locks[pool_index]};
// Choose a heap based on our page size request
const s32 heap_index{KPageHeap::GetBlockIndex(num_pages)};
if (heap_index < 0) {
return ResultOutOfMemory;
}
// TODO (bunnei): Support multiple managers
Impl& chosen_manager{managers[pool_index]};
// Ensure that we don't leave anything un-freed
auto group_guard = detail::ScopeExit([&] {
for (const auto& it : page_list.Nodes()) {
const auto min_num_pages{std::min<size_t>(
it.GetNumPages(), (chosen_manager.GetEndAddress() - it.GetAddress()) / PageSize)};
chosen_manager.Free(it.GetAddress(), min_num_pages);
// Ensure that we don't leave anything un-freed.
auto group_guard = SCOPE_GUARD({
for (const auto& it : out->Nodes()) {
auto& manager = this->GetManager(system.Kernel().MemoryLayout(), it.GetAddress());
const size_t num_pages_to_free =
std::min(it.GetNumPages(), (manager.GetEndAddress() - it.GetAddress()) / PageSize);
manager.Free(it.GetAddress(), num_pages_to_free);
}
});
// Keep allocating until we've allocated all our pages
for (s32 index{heap_index}; index >= 0 && num_pages > 0; index--) {
const auto pages_per_alloc{KPageHeap::GetBlockNumPages(index)};
while (num_pages >= pages_per_alloc) {
// Allocate a block
VAddr allocated_block{chosen_manager.AllocateBlock(index, false)};
if (!allocated_block) {
break;
}
// Safely add it to our group
{
auto block_guard = detail::ScopeExit(
[&] { chosen_manager.Free(allocated_block, pages_per_alloc); });
if (const ResultCode result{page_list.AddBlock(allocated_block, pages_per_alloc)};
result.IsError()) {
return result;
// Keep allocating until we've allocated all our pages.
for (s32 index = heap_index; index >= 0 && num_pages > 0; index--) {
const size_t pages_per_alloc = KPageHeap::GetBlockNumPages(index);
for (Impl* cur_manager = this->GetFirstManager(pool, dir); cur_manager != nullptr;
cur_manager = this->GetNextManager(cur_manager, dir)) {
while (num_pages >= pages_per_alloc) {
// Allocate a block.
PAddr allocated_block = cur_manager->AllocateBlock(index, random);
if (allocated_block == 0) {
break;
}
block_guard.Cancel();
}
// Safely add it to our group.
{
auto block_guard =
SCOPE_GUARD({ cur_manager->Free(allocated_block, pages_per_alloc); });
R_TRY(out->AddBlock(allocated_block, pages_per_alloc));
block_guard.Cancel();
}
num_pages -= pages_per_alloc;
num_pages -= pages_per_alloc;
}
}
}
// Clear allocated memory.
for (const auto& it : page_list.Nodes()) {
std::memset(system.DeviceMemory().GetPointer(it.GetAddress()), heap_fill_value,
it.GetSize());
}
// Only succeed if we allocated as many pages as we wanted
if (num_pages) {
return ResultOutOfMemory;
}
// Only succeed if we allocated as many pages as we wanted.
R_UNLESS(num_pages == 0, ResultOutOfMemory);
// We succeeded!
group_guard.Cancel();
return ResultSuccess;
}
ResultCode KMemoryManager::Free(KPageLinkedList& page_list, std::size_t num_pages, Pool pool,
Direction dir, u32 heap_fill_value) {
// Early return if we're freeing no pages
if (!num_pages) {
return ResultSuccess;
}
ResultCode KMemoryManager::AllocateAndOpen(KPageLinkedList* out, size_t num_pages, u32 option) {
ASSERT(out != nullptr);
ASSERT(out->GetNumPages() == 0);
// Lock the pool that we're freeing from
const auto pool_index{static_cast<std::size_t>(pool)};
std::lock_guard lock{pool_locks[pool_index]};
// Early return if we're allocating no pages.
R_SUCCEED_IF(num_pages == 0);
// TODO (bunnei): Support multiple managers
Impl& chosen_manager{managers[pool_index]};
// Lock the pool that we're allocating from.
const auto [pool, dir] = DecodeOption(option);
KScopedLightLock lk(pool_locks[static_cast<size_t>(pool)]);
// Free all of the pages
for (const auto& it : page_list.Nodes()) {
const auto min_num_pages{std::min<size_t>(
it.GetNumPages(), (chosen_manager.GetEndAddress() - it.GetAddress()) / PageSize)};
chosen_manager.Free(it.GetAddress(), min_num_pages);
// Allocate the page group.
R_TRY(this->AllocatePageGroupImpl(out, num_pages, pool, dir, false));
// Open the first reference to the pages.
for (const auto& block : out->Nodes()) {
PAddr cur_address = block.GetAddress();
size_t remaining_pages = block.GetNumPages();
while (remaining_pages > 0) {
// Get the manager for the current address.
auto& manager = this->GetManager(system.Kernel().MemoryLayout(), cur_address);
// Process part or all of the block.
const size_t cur_pages =
std::min(remaining_pages, manager.GetPageOffsetToEnd(cur_address));
manager.OpenFirst(cur_address, cur_pages);
// Advance.
cur_address += cur_pages * PageSize;
remaining_pages -= cur_pages;
}
}
return ResultSuccess;
}
std::size_t KMemoryManager::Impl::CalculateManagementOverheadSize(std::size_t region_size) {
const std::size_t ref_count_size = (region_size / PageSize) * sizeof(u16);
const std::size_t optimize_map_size =
ResultCode KMemoryManager::AllocateAndOpenForProcess(KPageLinkedList* out, size_t num_pages,
u32 option, u64 process_id, u8 fill_pattern) {
ASSERT(out != nullptr);
ASSERT(out->GetNumPages() == 0);
// Decode the option.
const auto [pool, dir] = DecodeOption(option);
// Allocate the memory.
{
// Lock the pool that we're allocating from.
KScopedLightLock lk(pool_locks[static_cast<size_t>(pool)]);
// Allocate the page group.
R_TRY(this->AllocatePageGroupImpl(out, num_pages, pool, dir, false));
// Open the first reference to the pages.
for (const auto& block : out->Nodes()) {
PAddr cur_address = block.GetAddress();
size_t remaining_pages = block.GetNumPages();
while (remaining_pages > 0) {
// Get the manager for the current address.
auto& manager = this->GetManager(system.Kernel().MemoryLayout(), cur_address);
// Process part or all of the block.
const size_t cur_pages =
std::min(remaining_pages, manager.GetPageOffsetToEnd(cur_address));
manager.OpenFirst(cur_address, cur_pages);
// Advance.
cur_address += cur_pages * PageSize;
remaining_pages -= cur_pages;
}
}
}
// Set all the allocated memory.
for (const auto& block : out->Nodes()) {
std::memset(system.DeviceMemory().GetPointer(block.GetAddress()), fill_pattern,
block.GetSize());
}
return ResultSuccess;
}
void KMemoryManager::Open(PAddr address, size_t num_pages) {
// Repeatedly open references until we've done so for all pages.
while (num_pages) {
auto& manager = this->GetManager(system.Kernel().MemoryLayout(), address);
const size_t cur_pages = std::min(num_pages, manager.GetPageOffsetToEnd(address));
{
KScopedLightLock lk(pool_locks[static_cast<size_t>(manager.GetPool())]);
manager.Open(address, cur_pages);
}
num_pages -= cur_pages;
address += cur_pages * PageSize;
}
}
void KMemoryManager::Close(PAddr address, size_t num_pages) {
// Repeatedly close references until we've done so for all pages.
while (num_pages) {
auto& manager = this->GetManager(system.Kernel().MemoryLayout(), address);
const size_t cur_pages = std::min(num_pages, manager.GetPageOffsetToEnd(address));
{
KScopedLightLock lk(pool_locks[static_cast<size_t>(manager.GetPool())]);
manager.Close(address, cur_pages);
}
num_pages -= cur_pages;
address += cur_pages * PageSize;
}
}
void KMemoryManager::Close(const KPageLinkedList& pg) {
for (const auto& node : pg.Nodes()) {
Close(node.GetAddress(), node.GetNumPages());
}
}
void KMemoryManager::Open(const KPageLinkedList& pg) {
for (const auto& node : pg.Nodes()) {
Open(node.GetAddress(), node.GetNumPages());
}
}
size_t KMemoryManager::Impl::Initialize(PAddr address, size_t size, VAddr management,
VAddr management_end, Pool p) {
// Calculate management sizes.
const size_t ref_count_size = (size / PageSize) * sizeof(u16);
const size_t optimize_map_size = CalculateOptimizedProcessOverheadSize(size);
const size_t manager_size = Common::AlignUp(optimize_map_size + ref_count_size, PageSize);
const size_t page_heap_size = KPageHeap::CalculateManagementOverheadSize(size);
const size_t total_management_size = manager_size + page_heap_size;
ASSERT(manager_size <= total_management_size);
ASSERT(management + total_management_size <= management_end);
ASSERT(Common::IsAligned(total_management_size, PageSize));
// Setup region.
pool = p;
management_region = management;
page_reference_counts.resize(
Kernel::Board::Nintendo::Nx::KSystemControl::Init::GetIntendedMemorySize() / PageSize);
ASSERT(Common::IsAligned(management_region, PageSize));
// Initialize the manager's KPageHeap.
heap.Initialize(address, size, management + manager_size, page_heap_size);
return total_management_size;
}
size_t KMemoryManager::Impl::CalculateManagementOverheadSize(size_t region_size) {
const size_t ref_count_size = (region_size / PageSize) * sizeof(u16);
const size_t optimize_map_size =
(Common::AlignUp((region_size / PageSize), Common::BitSize<u64>()) /
Common::BitSize<u64>()) *
sizeof(u64);
const std::size_t manager_meta_size =
Common::AlignUp(optimize_map_size + ref_count_size, PageSize);
const std::size_t page_heap_size = KPageHeap::CalculateManagementOverheadSize(region_size);
const size_t manager_meta_size = Common::AlignUp(optimize_map_size + ref_count_size, PageSize);
const size_t page_heap_size = KPageHeap::CalculateManagementOverheadSize(region_size);
return manager_meta_size + page_heap_size;
}
+149 -18
View File
@@ -5,11 +5,12 @@
#pragma once
#include <array>
#include <mutex>
#include <tuple>
#include "common/common_funcs.h"
#include "common/common_types.h"
#include "core/hle/kernel/k_light_lock.h"
#include "core/hle/kernel/k_memory_layout.h"
#include "core/hle/kernel/k_page_heap.h"
#include "core/hle/result.h"
@@ -52,22 +53,33 @@ public:
explicit KMemoryManager(Core::System& system_);
constexpr std::size_t GetSize(Pool pool) const {
return managers[static_cast<std::size_t>(pool)].GetSize();
void Initialize(VAddr management_region, size_t management_region_size);
constexpr size_t GetSize(Pool pool) const {
constexpr Direction GetSizeDirection = Direction::FromFront;
size_t total = 0;
for (auto* manager = this->GetFirstManager(pool, GetSizeDirection); manager != nullptr;
manager = this->GetNextManager(manager, GetSizeDirection)) {
total += manager->GetSize();
}
return total;
}
void InitializeManager(Pool pool, u64 start_address, u64 end_address);
PAddr AllocateAndOpenContinuous(size_t num_pages, size_t align_pages, u32 option);
ResultCode AllocateAndOpen(KPageLinkedList* out, size_t num_pages, u32 option);
ResultCode AllocateAndOpenForProcess(KPageLinkedList* out, size_t num_pages, u32 option,
u64 process_id, u8 fill_pattern);
VAddr AllocateAndOpenContinuous(size_t num_pages, size_t align_pages, u32 option);
ResultCode Allocate(KPageLinkedList& page_list, std::size_t num_pages, Pool pool, Direction dir,
u32 heap_fill_value = 0);
ResultCode Free(KPageLinkedList& page_list, std::size_t num_pages, Pool pool, Direction dir,
u32 heap_fill_value = 0);
static constexpr size_t MaxManagerCount = 10;
static constexpr std::size_t MaxManagerCount = 10;
void Close(PAddr address, size_t num_pages);
void Close(const KPageLinkedList& pg);
void Open(PAddr address, size_t num_pages);
void Open(const KPageLinkedList& pg);
public:
static std::size_t CalculateManagementOverheadSize(std::size_t region_size) {
static size_t CalculateManagementOverheadSize(size_t region_size) {
return Impl::CalculateManagementOverheadSize(region_size);
}
@@ -100,17 +112,26 @@ private:
Impl() = default;
~Impl() = default;
std::size_t Initialize(Pool new_pool, u64 start_address, u64 end_address);
size_t Initialize(PAddr address, size_t size, VAddr management, VAddr management_end,
Pool p);
VAddr AllocateBlock(s32 index, bool random) {
return heap.AllocateBlock(index, random);
}
void Free(VAddr addr, std::size_t num_pages) {
void Free(VAddr addr, size_t num_pages) {
heap.Free(addr, num_pages);
}
constexpr std::size_t GetSize() const {
void SetInitialUsedHeapSize(size_t reserved_size) {
heap.SetInitialUsedSize(reserved_size);
}
constexpr Pool GetPool() const {
return pool;
}
constexpr size_t GetSize() const {
return heap.GetSize();
}
@@ -122,10 +143,88 @@ private:
return heap.GetEndAddress();
}
static std::size_t CalculateManagementOverheadSize(std::size_t region_size);
constexpr size_t GetPageOffset(PAddr address) const {
return heap.GetPageOffset(address);
}
static constexpr std::size_t CalculateOptimizedProcessOverheadSize(
std::size_t region_size) {
constexpr size_t GetPageOffsetToEnd(PAddr address) const {
return heap.GetPageOffsetToEnd(address);
}
constexpr void SetNext(Impl* n) {
next = n;
}
constexpr void SetPrev(Impl* n) {
prev = n;
}
constexpr Impl* GetNext() const {
return next;
}
constexpr Impl* GetPrev() const {
return prev;
}
void OpenFirst(PAddr address, size_t num_pages) {
size_t index = this->GetPageOffset(address);
const size_t end = index + num_pages;
while (index < end) {
const RefCount ref_count = (++page_reference_counts[index]);
ASSERT(ref_count == 1);
index++;
}
}
void Open(PAddr address, size_t num_pages) {
size_t index = this->GetPageOffset(address);
const size_t end = index + num_pages;
while (index < end) {
const RefCount ref_count = (++page_reference_counts[index]);
ASSERT(ref_count > 1);
index++;
}
}
void Close(PAddr address, size_t num_pages) {
size_t index = this->GetPageOffset(address);
const size_t end = index + num_pages;
size_t free_start = 0;
size_t free_count = 0;
while (index < end) {
ASSERT(page_reference_counts[index] > 0);
const RefCount ref_count = (--page_reference_counts[index]);
// Keep track of how many zero refcounts we see in a row, to minimize calls to free.
if (ref_count == 0) {
if (free_count > 0) {
free_count++;
} else {
free_start = index;
free_count = 1;
}
} else {
if (free_count > 0) {
this->Free(heap.GetAddress() + free_start * PageSize, free_count);
free_count = 0;
}
}
index++;
}
if (free_count > 0) {
this->Free(heap.GetAddress() + free_start * PageSize, free_count);
}
}
static size_t CalculateManagementOverheadSize(size_t region_size);
static constexpr size_t CalculateOptimizedProcessOverheadSize(size_t region_size) {
return (Common::AlignUp((region_size / PageSize), Common::BitSize<u64>()) /
Common::BitSize<u64>()) *
sizeof(u64);
@@ -135,13 +234,45 @@ private:
using RefCount = u16;
KPageHeap heap;
std::vector<RefCount> page_reference_counts;
VAddr management_region{};
Pool pool{};
Impl* next{};
Impl* prev{};
};
private:
Impl& GetManager(const KMemoryLayout& memory_layout, PAddr address) {
return managers[memory_layout.GetPhysicalLinearRegion(address).GetAttributes()];
}
const Impl& GetManager(const KMemoryLayout& memory_layout, PAddr address) const {
return managers[memory_layout.GetPhysicalLinearRegion(address).GetAttributes()];
}
constexpr Impl* GetFirstManager(Pool pool, Direction dir) const {
return dir == Direction::FromBack ? pool_managers_tail[static_cast<size_t>(pool)]
: pool_managers_head[static_cast<size_t>(pool)];
}
constexpr Impl* GetNextManager(Impl* cur, Direction dir) const {
if (dir == Direction::FromBack) {
return cur->GetPrev();
} else {
return cur->GetNext();
}
}
ResultCode AllocatePageGroupImpl(KPageLinkedList* out, size_t num_pages, Pool pool,
Direction dir, bool random);
private:
Core::System& system;
std::array<std::mutex, static_cast<std::size_t>(Pool::Count)> pool_locks;
std::array<KLightLock, static_cast<size_t>(Pool::Count)> pool_locks;
std::array<Impl*, MaxManagerCount> pool_managers_head{};
std::array<Impl*, MaxManagerCount> pool_managers_tail{};
std::array<Impl, MaxManagerCount> managers;
size_t num_managers{};
};
} // namespace Kernel
+9 -1
View File
@@ -14,7 +14,8 @@
namespace Kernel {
enum KMemoryRegionType : u32 {
KMemoryRegionAttr_CarveoutProtected = 0x04000000,
KMemoryRegionAttr_CarveoutProtected = 0x02000000,
KMemoryRegionAttr_Uncached = 0x04000000,
KMemoryRegionAttr_DidKernelMap = 0x08000000,
KMemoryRegionAttr_ShouldKernelMap = 0x10000000,
KMemoryRegionAttr_UserReadOnly = 0x20000000,
@@ -239,6 +240,11 @@ static_assert(KMemoryRegionType_VirtualDramHeapBase.GetValue() == 0x1A);
static_assert(KMemoryRegionType_VirtualDramKernelPtHeap.GetValue() == 0x2A);
static_assert(KMemoryRegionType_VirtualDramKernelTraceBuffer.GetValue() == 0x4A);
// UNUSED: .DeriveSparse(2, 2, 0);
constexpr auto KMemoryRegionType_VirtualDramUnknownDebug =
KMemoryRegionType_Dram.DeriveSparse(2, 2, 1);
static_assert(KMemoryRegionType_VirtualDramUnknownDebug.GetValue() == (0x52));
constexpr auto KMemoryRegionType_VirtualDramKernelInitPt =
KMemoryRegionType_VirtualDramHeapBase.Derive(3, 0);
constexpr auto KMemoryRegionType_VirtualDramPoolManagement =
@@ -330,6 +336,8 @@ constexpr KMemoryRegionType GetTypeForVirtualLinearMapping(u32 type_id) {
return KMemoryRegionType_VirtualDramKernelTraceBuffer;
} else if (KMemoryRegionType_DramKernelPtHeap.IsAncestorOf(type_id)) {
return KMemoryRegionType_VirtualDramKernelPtHeap;
} else if ((type_id | KMemoryRegionAttr_ShouldKernelMap) == type_id) {
return KMemoryRegionType_VirtualDramUnknownDebug;
} else {
return KMemoryRegionType_Dram;
}
+19
View File
@@ -0,0 +1,19 @@
// Copyright 2022 yuzu Emulator Project
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
#include "common/alignment.h"
#include "common/assert.h"
#include "core/core.h"
#include "core/device_memory.h"
#include "core/hle/kernel/k_page_buffer.h"
#include "core/hle/kernel/memory_types.h"
namespace Kernel {
KPageBuffer* KPageBuffer::FromPhysicalAddress(Core::System& system, PAddr phys_addr) {
ASSERT(Common::IsAligned(phys_addr, PageSize));
return reinterpret_cast<KPageBuffer*>(system.DeviceMemory().GetPointer(phys_addr));
}
} // namespace Kernel
+28
View File
@@ -0,0 +1,28 @@
// Copyright 2022 yuzu Emulator Project
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
#pragma once
#include <array>
#include "common/common_types.h"
#include "core/hle/kernel/memory_types.h"
#include "core/hle/kernel/slab_helpers.h"
namespace Kernel {
class KPageBuffer final : public KSlabAllocated<KPageBuffer> {
public:
KPageBuffer() = default;
static KPageBuffer* FromPhysicalAddress(Core::System& system, PAddr phys_addr);
private:
[[maybe_unused]] alignas(PageSize) std::array<u8, PageSize> m_buffer{};
};
static_assert(sizeof(KPageBuffer) == PageSize);
static_assert(alignof(KPageBuffer) == PageSize);
} // namespace Kernel

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