Commit Graph
2543 Commits
Author SHA1 Message Date
teoxoy 6e8d0bd149 add tests for adapter_allowed 2026-06-03 23:36:32 +02:00
teoxoy 939776d5dc check that all limits are at least as high as default limits under
`STRICT_WEBGPU_COMPLIANCE`
2026-06-03 23:36:32 +02:00
teoxoy b61ddc1211 check alignment limits are powers of 2 2026-06-03 23:36:32 +02:00
teoxoy 6aa13649e4 filter out adapters with missing downlevel flags under
`STRICT_WEBGPU_COMPLIANCE`
2026-06-03 23:36:32 +02:00
teoxoy 45f820a4ad disallow LoadOp::DontCare under STRICT_WEBGPU_COMPLIANCE 2026-06-03 23:36:32 +02:00
Teoxoy bf1b57d47f [indirect-draw-validation] make sure we don't encounter u64 offsets in indirect draw validation 2026-06-03 22:43:37 +02:00
Teoxoy eddbd98028 [indirect-draw-validation] use a mask instead of bitshifts 2026-06-03 22:43:37 +02:00
Teoxoy d2a630771f [indirect-draw-validation] rename can_overflow to overflow_permitted 2026-06-03 22:43:37 +02:00
Teoxoy 35ba591a6c [indirect-draw-validation] add test for calculate_src_offsets 2026-06-03 22:43:37 +02:00
Teoxoy 82080096a0 [indirect-draw-validation] fix calculation of out_dynamic_offset
`max_dynamic_offset` should be used to clamp `out_dynamic_offset`
directly; aligning it to `dynamic_offset_stride` clamps
`out_dynamic_offset` to a lower value which then causes `out_offset` to
exceed the binding size.
2026-06-03 22:43:37 +02:00
Teoxoy 3ad17cbed6 [indirect-draw-validation] document output invariants of calculate_src_offsets 2026-06-03 22:43:37 +02:00
Teoxoy 8638b9e461 [indirect-draw-validation] simplify dynamic_offset_stride calculation 2026-06-03 22:43:37 +02:00
Teoxoy 31d22915be [indirect-draw-validation] don't attempt indirect draw validation for mesh shaders since it's not supported 2026-06-03 22:43:37 +02:00
Andy Leiserson 0c78336958 Require full initialization of buffers used in dynamic offset bindings (#9628) 2026-06-03 18:26:31 +00:00
AdrianEddy d73c7560b4 fix(dx12): Handle multi-planar formats in texture copy operations (#9551) 2026-06-03 19:52:20 +02:00
Jim Blandy 6cb7f2170e wgpu_core::track::range::RangedStates: Remove stale TODO. (#9636) 2026-06-03 17:40:58 +00:00
Andy Leiserson 560b763f64 fix(core): Correctly track texture init state after writing one aspect 2026-06-03 19:33:29 +02:00
5unekkuand5unekku ac0f6bd351 wgpu-core: enable is_sorted_by_key debug assert in triage_submissions (#9630)
Co-authored-by: 5unekku <@5unekku>
2026-06-03 17:51:08 +02:00
Luo Zhihao 8b7e9edcdb Mask stencil clear value and reference value with 255 (#9607) 2026-06-02 17:09:58 +00:00
Luo ZhihaoandErich Gubler a7e0298dc5 Validate LoadOp and StoreOp are None for attachments without corresponding depth or stencil aspect (#9567)
Co-authored-by: Erich Gubler <erichdongubler@gmail.com>
2026-05-29 11:28:50 -04:00
Vecvec a1fceed071 Move fence synchronisation into wgpu-hal (#9475) 2026-05-28 14:49:12 -07:00
Andy Leiserson fc1aa76bad fix(core): Don't use {buffer,texture}_init_actions for usage tracking
Fixes #9576
2026-05-25 10:48:20 +02:00
Andy Leiserson d1e15bc3d3 refactor(core): Rename used_*_ranges to *_init_actions 2026-05-25 10:48:20 +02:00
AdrianEddy 6208d80934 Add an explicit initial_state to create_texture_from_hal (#9496)
This is a breaking change for users of the low-level `create_texture_from_hal` API.
2026-05-12 16:59:56 -07:00
Andy Leiserson 6469861478 Small optimization to prepare_surface_texture_for_submit (#9473) 2026-05-08 13:39:47 -07:00
Andy Leiserson 54bf53eb11 fix(core): Check for device loss on all submit paths (#9478) 2026-05-06 19:06:19 +02:00
Jim Blandy 64394cc73a Invoke the mapping callback even if the buffer is invalid. (#9513) 2026-05-06 08:19:12 -07:00
Andy Leiserson e1e4bb1a14 fix(core): Flush pending writes before mapAsync, if needed
Fixes #5173
2026-04-29 10:24:42 -07:00
Andy Leiserson 0aaa7ff4b1 fix(core): Flush pending writes in on_submitted_work_done 2026-04-29 10:24:42 -07:00
Andy Leiserson 8e9f7e572b refactor(core): Make submit_with_pending_writes more general
Extend the command_index lock through submission
2026-04-29 10:24:42 -07:00
Andy Leiserson 78c90239c1 chore(core): Remove an unused error variant 2026-04-29 10:24:42 -07:00
Andy Leiserson 0d32f7e756 fix(core): Order BUFFER_MAP_STATE lock after QUEUE_PENDING_WRITES 2026-04-22 14:36:17 -04:00
Andy Leiserson 90dd7435e9 fix(core): Don't overlap QUEUE_PENDING_WRITES and QUEUE_LIFE_TRACKER locks 2026-04-22 14:36:17 -04:00
Andy Leiserson 458d310916 docs: Update lock graph
Related to #5572
2026-04-22 14:36:17 -04:00
Andy Leiserson 07a7dd4454 refactor(core): Eliminate buffer mapping logic in Global (#9444) 2026-04-21 16:50:55 +02:00
Andy Leiserson 61e76ee746 fix(core): Use more checked arithmetic, part 2b (#9425) 2026-04-16 14:27:01 +02:00
Teodor Tanasoaia cc44c1f616 [d3d12] always expose Features::FLOAT32_BLENDABLE (#9431) 2026-04-15 11:46:36 -04:00
teoxoy c9c7359f5b deduplicate vertex buffer validation in is_ready 2026-04-14 17:20:15 -07:00
teoxoy a1b646243d converge on a single VertexState impl for both render passes and
render bundles

- render passes now defer setting vertex buffers to the start of a draw
  command (same as bundles)
- render bundles now cache vertex/instance limit calculations when
  setting the pipeline or when setting a vertex buffer (same as passes)
2026-04-14 17:20:15 -07:00
teoxoy a9a641c82a remove PipelineState from bundle 2026-04-14 17:20:15 -07:00
teoxoy 4bbed4074c add max_bind_groups_plus_vertex_buffers limit 2026-04-14 17:20:15 -07:00
teoxoy 2060683c29 make vertex buffer slots optional 2026-04-14 17:20:15 -07:00
atlas 20f5c7430e fix(core): use wait_for_idle instead of timeout loop to wait for present in Queue drop 2026-04-14 10:51:03 +02:00
atlas 4c61d61361 fix(core): Keep surface texture alive through present preparation 2026-04-14 10:51:03 +02:00
atlas 12ab856f6c feat(core): Move present from Surface to Queue 2026-04-14 10:51:03 +02:00
atlas 6314157c92 fix(core): Add NothingToPresent error for present/discard without acquired texture 2026-04-14 10:51:03 +02:00
atlas c662f2a595 refactor(core): Factor out submit_pending_writes from Queue::submit 2026-04-14 10:51:03 +02:00
Erich Gubler 10ce22b2ea refactor!: subvert-rename dispatch* to dispatchWorkGroups* (#9362) 2026-04-11 20:58:22 +02:00
teoxoy 8c40370eae [d3d12] lower max_vertex_attributes to 30
See https://github.com/gpuweb/gpuweb/pull/5604#discussion_r2967504832
2026-04-11 00:55:54 -04:00
Teodor Tanasoaia f7b650d55c [wgpu-core] fix potential crash when device is not supplied to validate_render_bundle_encoder_descriptor (#9407) 2026-04-10 19:46:34 +00:00