fix(naga): don't crash on zero value of dynamically-sized array

[mooori]

Connections

Fixes #8442

Description

Usage of dynamically sized arrays is restricted to a few special cases. let a = array<f32>() is trying to construct the ZeroValue of a dynamically sized array, which is not allowed. With this PR, ZeroValue construction of dynamically sized arrays gets validated out.

Compiling the wgsl from #8442 doesn't crash naga anymore but instead throws an error:

$ RUST_BACKTRACE=1 cargo run -p naga-cli -- canary.wgsl canary.hlsl

[ERROR naga::valid::zero_value] Constructing zero value of runtime-sized array
error: Entry point main at Compute is invalid
  ┌─ canary.wgsl:3:13
  │
3 │     let a = array<f32>();
  │             ^^^^^^^^^^^ naga::ir::Expression [0]
  │
  = Expression [0] is invalid
  = ZeroValue construction of runtime-sized array is not allowed

Testing

Adds test zero_value_dyn_array_error.

Squash or Rebase?

Squash

Checklist

• Run cargo fmt.
• Run taplo format.
• Run cargo clippy --tests. If applicable, add:
  • --target wasm32-unknown-unknown
• Run cargo xtask test to run tests.
• If this contains user-facing changes, add a CHANGELOG.md entry.

[andyleiserson]

There is a relevant CTS test webgpu:shader,validation,expression,call,builtin,value_constructor:array_zero_value:case="invalid_rta", but it isn't failing, because it expects any validation error, and is getting one related to constructibility. (I'm looking at fixing this in the CTS.)

There are also some related CTS tests that expose other broken cases (atomics, override-sized arrays). Also see #4720.

Submitting as comments right now, but this is all nits, there's nothing here that is essential to fix before merging this.

[andyleiserson]

There is a relevant CTS test webgpu:shader,validation,expression,call,builtin,value_constructor:array_zero_value:case="invalid_rta", but it isn't failing, because it expects any validation error, and is getting one related to constructibility. (I'm looking at fixing this in the CTS.)

gpuweb/cts#4550 has my proposed change to the CTS.

[andyleiserson]

Thanks for the contribution!