The WebGPU spec. `createBindGroup` [states][spec-ref] (emphasis mine):
> Device timeline initialization steps:
>
> …
>
> 2. If any of the following conditions are unsatisfied generate a
> validation error, invalidate _bindGroup_ and return.
>
> …
>
> For each `GPUBindGroupEntry` _bindingDescriptor_ in
> _descriptor_.`entries`:
>
> - …
>
> - If the defined binding member for _layoutBinding_ is:
>
> - …
>
> - `buffer`
>
> - …
>
> - If _layoutBinding_.`buffer`.`type` is
>
> - …
>
> - `"storage"` or `"read-only-storage"`
>
> - …
>
> - effective buffer binding size(_bufferBinding_) is a multiple of 4.
[spec-ref]: https://www.w3.org/TR/webgpu/#dom-gpudevice-createbindgroup
We were not implementing this check of effective buffer binding size.
Check that it's a multiple of 4, including
`webgpu:api,validation,createBindGroup:buffer,effective_buffer_binding_size:*`
that this is now implemented as intended.
Adds a `BindingResource` variant for external textures. In core's
create_bind_group() implementation, allow binding either external
textures or texture views to `BindingType::ExternalTexture` layout
entries.
In either case, provide HAL with a `hal::ExternalTextureBinding`,
consisting of 3 `hal::TextureBinding`s and a `hal::BufferBinding`. In
the texture view case we use the device's default params buffer for
the buffer. When there are fewer than 3 planes we can simply repeat an
existing plane multiple times - the contents of the params buffer will
ensure the shader only accesses the correct number of planes anyway.
Track the view or external texture in `BindGroupStates` to ensure they
remain alive whilst required.
And finally, add the corresponding API to wgpu, with an implementation
for the wgpu-core backend.
Adds a new feature flag, `EXTERNAL_TEXTURE`, indicating device support
for our implementation of WebGPU's `GPUExternalTexture` [1] which will
land in upcoming patches. Conceptually this would make more sense as a
downlevel flag, as it is a core part of the WebGPU spec which we do not
yet support. We do not want, however, to cause applications to reject
adapters because we have not finished implementing this, so for now we
are making it an opt-in feature.
As an initial step towards supporting this feature, this patch adds a
new `BindingType` corresponding to WebGPU's
`GPUExternalTextureBindingLayout` [2]. This binding type dictates that
when creating a bind group the corresponding entry must be either an
external texture or a texture view with certain additional requirements
[3].
As of yet wgpu has no concept of an external texture (that will follow
in later patches) but for now this patch ensures that texture views
corresponding to an external texture binding type are validated
correctly. Note that as the feature flag is not yet supported on any
real backends, bind group layout creation will fail before getting the
chance to attempt to create a bind group. But in the added tests using
the noop backend we can see this validation taking place.
[1] https://www.w3.org/TR/webgpu/#gpuexternaltexture
[1] https://www.w3.org/TR/webgpu/#dictdef-gpuexternaltexturebindinglayout
[2] https://gpuweb.github.io/gpuweb/#bind-group-creation
* Make `wgpu_core::snatch::LockTrace` fully instead of partially conditional.
Now, when `cfg(not(debug_assertions))`, there is no `SNATCH_LOCK_TRACE`
thread local, and `LockTrace` has no data fields.
* Make `std` usage optional for `wgpu-core`.
Adds a `std` feature, enabled by default, to `wgpu-core`. When that
feature is disabled, the following functionality is not available:
* `Send + Sync` for resources.
* `trace` feature.
* `observe_locks` feature.
* Snatch lock recursive locking assertion.
Dependencies on `std` that still exist:
* Locks
* `std::error::Error` (waiting for Rust 1.81)
* `std::os::raw::c_char` for FFI
* `thread_local` and `Backtrace` in `snatch`
* fix more compile errors
* fix err
* fix all compile errors
* fix clippy errors
* fix issues in ray-traced-triangle example
* fix clippy warnings
* fix wasm not compiling
* fix test
* fix clippy (again)
* fix some rustfmt issues
* fix more rustfmt issues
* fix even more rustfmt issues
* fix yet more rustfmt issues
* add #[allow(dead_code)] to some structs to fix clippy on web assembly
* more format fixes
* even more format fixes
* close to final format fixes
* final format fixes (hopefully)
* actual final format fixes (hopefully)
* move ray_cube_cube example and fix bugs
* move other examples
* fix crash when dropping
* fix test
* apply clippy and rustfmt changes
* fix some memory leaks
* fix incompatibility with new wgpu trunk changes
* fix clippy & fmt
* fix clippy errors on github
* update to trunk
* remove a line that printed log "Created blas ..." in device create tlas
* fix merge issues
* fix merge
* fix fmt
* restored example ray scene functionality
* WIP fix rebase
* Fix compile issues
* More fixes
* Fix rebase
* Add wgls backend write_value_type AccelerationStructure
* Add back xtask Cargo.lock
* Revert serde path
* Doc typos
* Add -diff
* Revert more serde
* Revert serde again?
* Fix doc typo
* add test
* clippy & fmt
* fix error "blas does not exist"
* fix various things
* fix unused acceleration structures causing invalid SPIR-V
* fix merge
* fix use after free in a different way (now moved tlas instances to wgpu_core)
* refactor & fmt
* remove accidentally created file
* remove accidentally created file (actually)
* merge & fix
* Fix problems
* fix validation error
* fix validation
* fix merge
* fmt
* fix merge
* use staging buffer trackers in creating the ray-tracing staging buffers (not the tlas_s)
* change one missed staging buffer's tracker
* wip: example
* finished shadow!
* fix merge
* clippy
* remove some uses of blas ids
* fmt
* fix merge
* deduplicate code shared between command_encoder_build_acceleration_structures and command_encoder_build_acceleration_structures_unsafe_tlas
* clippy
* fix merge
* fix warnings
* clippy & fmt
* re-add matching to get_raw_tlas_instance_size and tlas_instance_into_bytes
* fmt
* fix command
* wip: fix merge
* fix merge
* fmt & docs
* fix wasm clippy
* fix issues with tests
* fix issues with tests
* fmt & clippy (wasm)
* wip ray-traced-triangle (non hal)
* unify camera location in ray_shadows
* fix tests
* add readme to ray_traced_triangle
* note that vertex format must be f32x3 (not yet validated)
* changelog
* require VertexFormat::Float32x3
* fix compile error
* when creating and building acceleration structures require Features::RAY_TRACING_ACCELERATION_STRUCTURE is enabled
* remove duplicate workaround
* fix most feedback
* remove all changes from .gitattributes
* fix compile error in test
* more changes along same lines
* format
* apply feedback to more places
* fix `cargo clippy` not just `cargo clippy --all-features`
* cache max instances from tlas creation
* fix clippy
* fix examples
* fix tests
* remove now useless docs
* add more docs
* yet more docs
* still more docs
* Documentation.
* Running out of unique commit messages to say writing docs.
* Add more references to docs.
* Get BindGroup in scope
* Remove unintentional changes
* Remove #[allow(dead_code)]
* Remove old comments
* Add device.poll
* make ray_scene start right way up
* fmt
* move scratch buffer to its own module
* actually remove intentional changes (turns out wgt is inconsistent)
* use better names in ray_scene/shader.wgsl
* fmt
* remove more unintentional changes.
* work on ray_tracing use after free
* spelling
Co-authored-by: Connor Fitzgerald <connorwadefitzgerald@gmail.com>
* grammar
Co-authored-by: Connor Fitzgerald <connorwadefitzgerald@gmail.com>
* move things into struct (should be all vectors with 3 item types or more)
* remove long-standing locks.
* fmt
* move not using index buffer to test
* impl index & index mut for tlas package
* impl index & index mut for tlas package
* fix errors
* fix test errors
* move everything out of ray_tracing.rs
* more index_mut
* fmt
* fix docs links
* move to separate impl block
* fix older request review
* fmt
* remove missed deref
* fix wasm clippy
* change to suggestions
* fmt
* add a note as to why `command_encoder_build_acceleration_structures_unsafe_tlas` and `command_encoder_build_acceleration_structures` are currently separate
* add a todo to reconsider soon `command_encoder_build_acceleration_structures_unsafe_tlas`
* switch to mat4 in ray-tracing examples
* Combine RT Tests Under Single Submodule
* Add More Raytracing Tests and Re-Organize
* Documentation and Formatting Fixes
* Remove Forced Alignment in wgpu-hal
* work on out_of_order_as_build and add out_of_order_as_build_use
* fmt
* update test label naming
* fix one validation error
* fix clippy & fmt
* add acceleration structure creation tests
* make mis-matched index test use the correct function
* move validation into validate_command_buffer
* fix tests
* add blas and tlas to pending writes to remove (now unneeded) `device.poll()`s
* remove commented out code from examples and properly use the constants in ray_shadow
* add examples to examples\README.md
* rename ray tracing to ray queries in readme
* remove unneeded usage from scratch buffer creation
* use snatchables for raw acceleration structures
* format
* add EXPERIMENTAL_ to features
* fix missed comment
---------
Co-authored-by: Daniel Keitel <daniel.mg.keitel@fau.de>
Co-authored-by: JMS55 <47158642+JMS55@users.noreply.github.com>
Co-authored-by: atlas dostal <rodol@rivalrebels.com>
Co-authored-by: vero <email@atlasdostal.com>
Co-authored-by: Connor Fitzgerald <connorwadefitzgerald@gmail.com>
Weaken our dependence on the `once_cell` crate by using functionality
from `std` instead that was upstreamed from `once_cell`.
It's not yet possible to eliminate this dependency entirely, but do what
we can for now.
In `wgpu_hal`:
- Document that `wgpu_hal` guarantees that shaders will not access buffer
contents beyond the bindgroups' bound regions, rounded up to some
adapter-specific alignment. Introduce the term "accessible region" for
the portion of the buffer that shaders can actually get at.
- Document that all bets are off if you disable bounds checks with
`ShaderModuleDescriptor::runtime_checks`.
- Provide this alignment in `wgpu_hal::Alignments`. Update all backends
appropriately.
- In the Vulkan backend, use Naga to inject bounds checks on buffer accesses
unless `robustBufferAccess2` is available; `robustBufferAccess` is not
sufficient. Retrieve `VK_EXT_robustness2`'s properties, as needed to discover
the alignment above.
In `wgpu_core`:
- Use buffer bindings' accessible regions to determine which parts of the buffer
need to be initialized.
In `wgpu_types`:
- Document some of the possible effects of using
`ShaderBoundsChecks::unchecked`.
Fixes#1813.
