Commit graph

6639 commits

Author SHA1 Message Date
Alexandre Bouvier ed3f9bab11 vma: enable options everywhere 2023-07-31 13:01:21 +02:00
Moonlacer 4007142b91
Revert "Blacklist EDS3 blending from new AMD drivers" 2023-07-26 15:02:48 -05:00
liamwhite d3da1e6517
Merge pull request #10990 from comex/ubsan
Fixes and workarounds to make UBSan happier on macOS
2023-07-26 10:33:28 -04:00
liamwhite 0ea138fb5b
Merge pull request #11098 from GPUCode/texel-buffers
buffer_cache: Increase number of texture buffers
2023-07-22 11:17:27 -04:00
lat9nq 4ecedc4991 vk_buffer_cache: Format 2023-07-18 19:56:20 -04:00
lat9nq 71b3b2a2f0 general: Silence -Wshadow{,-uncaptured-local} warnings
These occur in the latest commits in LLVM Clang.
2023-07-18 19:31:35 -04:00
GPUCode 4347dd26c6 buffer_cache: Increase number of texture buffers 2023-07-15 23:09:58 +03:00
comex d7c532d889 Fixes and workarounds to make UBSan happier on macOS
There are still some other issues not addressed here, but it's a start.

Workarounds for false-positive reports:

- `RasterizerAccelerated`: Put a gigantic array behind a `unique_ptr`,
  because UBSan has a [hardcoded limit](https://stackoverflow.com/questions/64531383/c-runtime-error-using-fsanitize-undefined-object-has-a-possibly-invalid-vp)
  of how big it thinks objects can be, specifically when dealing with
  offset-to-top values used with multiple inheritance.  Hopefully this
  doesn't have a performance impact.

- `QueryCacheBase::QueryCacheBase`: Avoid an operation that UBSan thinks
  is UB even though it at least arguably isn't.  See the link in the
  comment for more information.

Fixes for correct reports:

- `PageTable`, `Memory`: Use `uintptr_t` values instead of pointers to
  avoid UB from pointer overflow (when pointer arithmetic wraps around
  the address space).

- `KScheduler::Reload`: `thread->GetOwnerProcess()` can be `nullptr`;
  avoid calling methods on it in this case.  (The existing code returns
  a garbage reference to a field, which is then passed into
  `LoadWatchpointArray`, and apparently it's never used, so it's
  harmless in practice but still triggers UBSan.)

- `KAutoObject::Close`: This function calls `this->Destroy()`, which
  overwrites the beginning of the object with junk (specifically a free
  list pointer).  Then it calls `this->UnregisterWithKernel()`.  UBSan
  complains about a type mismatch because the vtable has been
  overwritten, and I believe this is indeed UB.  `UnregisterWithKernel`
  also loads `m_kernel` from the 'freed' object, which seems to be
  technically safe (the overwriting doesn't extend as far as that
  field), but seems dubious.  Switch to a `static` method and load
  `m_kernel` in advance.
2023-07-15 12:00:28 -07:00
Alexandre Bouvier c3050c1b48 cmake: allow using system VMA library 2023-07-12 04:51:45 +02:00
bunnei ce7c418e0c
Merge pull request #10996 from Kelebek1/readblock_optimisation
Use spans over guest memory where possible instead of copying data
2023-07-10 18:54:19 -07:00
liamwhite d8eb37fbec
Merge pull request #10994 from liamwhite/ue4-preferred
vulkan_common: use device local preferred for image memory
2023-07-05 09:23:56 -04:00
liamwhite f71140fbd9
Merge pull request #11012 from gidoly/metroid-fix
Fix regression by unreal engine fix pr #11009
2023-07-05 09:23:34 -04:00
bunnei 1462db4694 video_core: vulkan_device: Disable timeline semaphore on Turnip, fix qcom version check. 2023-07-03 19:25:06 -07:00
bunnei 44af2e32a4
Merge pull request #10964 from bunnei/gpu-remove-qcom-check
video_core: vulkan_device: Fix S8Gen2 dynamic state checks.
2023-07-03 16:59:29 -07:00
bunnei cef7aaa8ec video_core: vulkan_device: Change to driver version check. 2023-07-03 14:25:06 -07:00
gidoly 408a9cd50d
oops re open 2023-07-03 20:25:23 +09:00
Kelebek1 6f7cb69c94 Use spans over guest memory where possible instead of copying data. 2023-07-02 23:09:48 +01:00
liamwhite 5e3695ecaa
Merge pull request #10479 from GPUCode/format-list
Add support for VK_KHR_image_format_list
2023-07-02 17:38:21 -04:00
liamwhite eaa62aee98
Merge pull request #10942 from FernandoS27/android-is-a-pain-in-the-a--
Memory Tracking: Add mechanism to register small writes when gpu page is contested by GPU
2023-07-02 11:29:01 -04:00
Liam ad1946b893 vulkan_common: use device local preferred for image memory 2023-07-01 23:44:57 -04:00
Liam 34c448bad4 Revert "texture_cache: Fix incorrect logic for AccelerateDMA"
This reverts commit 1fc47361a1.
2023-07-01 23:37:50 -04:00
liamwhite 146769f44e
Merge pull request #10984 from comex/cob
Minor cleanup in BufferCacheRuntime::ReserveNullBuffer
2023-07-01 22:38:33 -04:00
liamwhite ae7e9b5469
Merge pull request #10974 from Steveice10/macos_vk
vulkan: Improvements to macOS surface creation
2023-07-01 22:38:26 -04:00
liamwhite 971b89b979
Merge pull request #10970 from Morph1984/thing
general: Misc changes that did not deserve their own PRs
2023-07-01 22:38:18 -04:00
comex 1e3b2328a6 Minor cleanup in BufferCacheRuntime::ReserveNullBuffer
As far as I can tell, there is no reason to OR this bit in separately.
2023-07-01 12:00:25 -07:00
GPUCode 272916eeaf renderer_vulkan: Fix some missing view formats
* Many times the format itself wouldn't have been added to the list causing device losses for nvidia GPUs

* Also account for ASTC acceleration storage views
2023-07-01 16:03:35 +03:00
GPUCode 95cefaf993 renderer_vulkan: Add support for VK_KHR_image_format_list 2023-07-01 16:03:29 +03:00
Steveice10 e146a00345 vulkan: Use newer VK_EXT_metal_surface to create surface for MoltenVK. 2023-06-30 23:46:03 -07:00
Morph 5a09fa5012 maxwell_dma: Specify dst_operand.pitch instead of a temp var 2023-06-30 21:49:59 -04:00
Morph 310b6cf4af general: Use ScratchBuffer where possible 2023-06-30 21:49:59 -04:00
Fernando S a10bdaf934
Merge pull request #10953 from FernandoS27/oh-oopsies-yfc
Texture cache: Fix YFC regression due to code testing
2023-06-30 20:25:09 +02:00
Fernando S 9252ad4e10
Merge pull request #10956 from FernandoS27/pikmin-another-game-ill-hate
AccelerateDMA: Don't accelerate 3D texture DMA operations
2023-06-30 09:37:07 +02:00
bunnei ddcd89afd4 video_core: vulkan_device: Scope S8Gen2 checks to just Qualcomm. 2023-06-29 18:41:38 -07:00
bunnei dfa040502a video_core: vulkan_device: Fix S8Gen2 dynamic state checks. 2023-06-29 17:37:42 -07:00
Fernando Sahmkow 596a6132b9 AccelerateDMA: Don't accelerate 3D texture DMA operations 2023-06-29 17:23:29 +02:00
Fernando Sahmkow b62121fd60 Texture cache: Fix YFC regression due to code testing 2023-06-29 11:58:45 +02:00
Matías Locatti ed93cbd462
Blacklist EDS3 blending from new AMD drivers 2023-06-28 20:10:27 -03:00
Fernando Sahmkow da440da9f5 Memory Tracking: Optimize tracking to only use atomic writes when contested with the host GPU 2023-06-28 21:32:45 +02:00
Fernando Sahmkow 47d0d292d5 MemoryTracking: Initial setup of atomic writes. 2023-06-28 19:34:21 +02:00
GPUCode ddcc958336 renderer_vulkan: Prevent crashes when blitting depth stencil 2023-06-27 18:00:09 -07:00
GPUCode eac46ad7ce video_core: Add BCn decoding support 2023-06-27 18:00:09 -07:00
GPUCode b8c96cee5f renderer_vulkan: Add more feature checking 2023-06-27 18:00:09 -07:00
GPUCode 220a42896d renderer_vulkan: Don't assume debug tool with debug renderer
* Causes crashes because mali drivers don't support debug utils
2023-06-27 18:00:09 -07:00
GPUCode 1522b95658 renderer_vulkan: Bump minimum SPIRV version
* 1.3 is guaranteed on all 1.1 drivers
2023-06-27 18:00:09 -07:00
GPUCode c339af37a7 renderer_vulkan: Respect viewport limit 2023-06-27 18:00:09 -07:00
GPUCode a9b44d37e1 renderer_vulkan: Don't add transform feedback flag if unsupported 2023-06-27 18:00:09 -07:00
GPUCode 72e7f5b4dd renderer_vulkan: Add suport for debug report callback 2023-06-27 18:00:09 -07:00
liamwhite 2b3bfafb9e
Merge pull request #10867 from Kelebek1/dma_safe
Use safe reads in DMA engine
2023-06-27 11:21:47 -04:00
liamwhite c6959449d1
Merge pull request #10473 from GPUCode/vma
Use vulkan memory allocator
2023-06-27 11:21:36 -04:00
GPUCode b6c6dcc576 externals: Use cmake subdirectory 2023-06-26 18:59:24 +03:00