For some reason, the implementation of `Display` for
`ShaderError<WithSpan<ValidationError>>>` is building its own
`Diagnostic` with _only_ the labels populated: no messages or notes.
I don't know why this is, but it seems to have a strict subset of the
information already present in the private implementation of
`WithSpan::diagnostic`.
Fix this by exposing `WithSpan::diagnostic` as `pub(crate)`, and
re-using its output in `impl Display for
ShaderError<WithSpan<ValidationError>>>`.
Improve handling of `Access` expressions whose base is an array or
matrix (not a pointer to such), and whose index is not known at
compile time. SPIR-V does not have instructions that can do this
directly, so spill such values to temporary variables, and perform the
accesses using `OpAccessChain` instructions applied to the
temporaries.
When performing chains of accesses like `a[i].x[j]`, do not reify
intermediate values; generate a single `OpAccessChain` for the entire
thing.
Remove special cases for arrays; the same code now handles arrays and
matrices.
Update validation to permit dynamic indexing of matrices.
For details, see the comments on the new tracking structures in
`naga:🔙:spv::Function`.
Add snapshot test `index-by-value.wgsl`.
Fixes#6358.
Fixes#4337.
Alternative to #6362.
This change uses `include_wgsl!(…)` in usages where an
`include_str!("….wgsl")` was used to construct
a `ShaderModuleDescriptor`'s `source, but the `label` was set to `None`.
This should (1) showcase a nice idiomatic convenience we offer in our
examples better, (2) make code more concise, and (3) get some
automatically generated labels in diagnostics where it seems it won't
hurt.
* add parsing for spirv::Op::AtomicLoad and spirv::Op::AtomicStore
* spv-in parse AtomicExchange and AtomicCompareExchange
* add atomic i decrement
* bookend atomic store statement with emmitter.finish/emitter.start to suppress a double load expression
bookend atomic result expressions with emitter.finish/start to prevent double defs
* add atomic iadd, isub, smin, umin, smax, umax, and, or, xor
* parse atomic flag test and set, parse atomic flag clear
* remove atomic compare exchange work
* changelog
* moved spirv tests into front/spv/mod.rs
* feature gate atomic spv tests because they require wgsl-[in,out]
* BlockContext::get_contained_global_variable returns Result
* Generate spans covering the entire instruction.
Granted, there is pre-existing code in the SPIR-V front end that gets
this wrong, but:
It doesn't make sense to read `self.data_offset`, and then immediately
pass that to `self.span_from_with_op`. The point of that function is
to make the span cover the entire instruction, operands included.
* Move `From` implementation into spv front end
* doc comments, minor cleanups
* remove parsing of OpAtomicFlagClear and OpAtomicFlagTestAndSet
* sync atomic spvasm files
---------
Co-authored-by: Jim Blandy <jimb@red-bean.com>
* Properly handle the case where `Navigator.gpu` is undefined and WebGPU is the only compiled backend.
Previously, `Instance::request_adapter` would invoke a wasm binding with an undefined arg0,
thus crashing the program. Now it will cleanly return `None` instead.
Fixes#6196.
* Fix typo in `Instance::new` doc comment.
* Add note to CHANGELOG.md
* Introduce `DefinedNonNullJsValue` type.
* Assert definedness of self.gpu in surface_get_capabilities.
* Use DefinedNonNullJsValue in signature of get_browser_gpu_property().
* Clarify meaning of gpu field with a comment.
This is just an API change for all the "set_bind_group" calls. Calls
that pass a Some() argument should have unchanged behavior. The None
cases are left as TODOs.
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.
PR #5956 wasn't fully complete and still had some outstanding minor
issues and cleanups to be done, as well as hidden semantic changes.
This addresses a bunch of them:
- Remove unnecessary `Error` mapping to `String` as `windows-rs`'s
`Error` has a more complete `Display` representation by itself.
- Remove `into_result()` as every call could have formatted the
`windows-rs` `Error` in a log call directly.
- Pass `None` instead of a pointer to an empty slice wherever possible
(waiting for https://github.com/microsoft/win32metadata/pull/1971 to
trickle down into `windows-rs`).
- Remove `.clone()` on COM objects (temporarily increasing the refcount)
when it can be avoided by inverting the order of operations on `raw`
variables.
PR #6150 suffered a much larger rebase "hell" than I anticipated. On
my Linux box I made this change, but lost it while force-pushing from
Windows (and created some other compiler errors while at it...).
By disabling all features on `glutin`/`glutin-winit` (the latter only
uses `x11`, and only forwards `wayland` to `glutin`) we may have dropped
a lot of "unused" dependencies for other GL backends, but also made the
crate unable to import X11 (Xlib/Xcb) and Wayland handles into EGL.
Also import the missing `glutin::context::Version` struct again which
was added last-minute to #6150 (to make sure my Intel card on Windows
creates a GLES 3.0+ instead of GLES 2.0 context) while the import was
accidentally squashed into #6152 (not merged yet).
