Looping forever while loop causes Dawn to Hang

[Max-Sepp]

Looping forever while loop causes Dawn to Hang

When running on a Windows 11 laptop with an AMD gpu and a Windows laptop with an Intel gpu fatal errors are produced when running crash.html.
The fatal errors are not produced when run on a phone with a Qualcomm gpu or an ARM gpu

The fatal errors are:

Device lost: ID3D12Device::GetDeviceRemovedReason failed with DXGI_ERROR_DEVICE_HUNG (0x887A0006)
 - While handling unexpected error type Internal when allowed errors are (Validation|DeviceLost).
    at CheckHRESULTImpl (..\..\third_party\dawn\src\dawn\native\d3d\D3DError.cpp:119)

Backend messages:
 * Device removed reason: DXGI_ERROR_DEVICE_HUNG (0x887A0006)

and

Failed to execute 'mapAsync' on 'GPUBuffer': [Device] is lost.

Whilst running Chrome freezes and then resumes.

After running the shader attempting to run any other shaders in a different tab results in: No available adapters

When running shaders consecutively one after each other in the same page the following is given:

"Failed to request device: Failed to execute 'requestDevice' on 'GPUAdapter': D3D12 create command queue failed with DXGI_ERROR_DEVICE_REMOVED (0x887A0005)
    at CheckHRESULTImpl (..\\..\\third_party\\dawn\\src\\dawn\\native\\d3d\\D3DError.cpp:117)

The code believed to cause the error is:

var x = 1f;
while (x == 1) {
    if (dot4U8Packed(u32(ub.values[7].value), 438797897u) < 1147979299u)
    {
        continue;
    }
    x = x - 1;
}
return vec4(1, x, x, 1);

The above loop loops forever.

Intel Laptop System Specification

139.0.7258.128 (Official Build) (64-bit) (cohort: 139.0.7258.154 Rollout)

Vendor: intel

Architecture: gen-121p

about gpu report: intel-about-gpu-2025-08-27T15-18-42-581Z.txt

AMD Laptop System Specification

139.0.7258.155 (Official Build) (64-bit) (cohort: Stable Installs & Version Pins)

Vendor: amd

Architecture: rdna-2

about gpu report: amd-about-gpu-2025-08-27T15-34-09-079Z.txt

[afd]

This behaviour is likely acceptable, as I don't think WGSL shaders are allowed to have non-terminating loops. I would expect this to be an undefined behaviour or something similar. Can you check that the WGSL and WebGPU specifications say about nonterminating shaders?

[Max-Sepp]

From the WGSL specification:

A dynamic error occurs if the loop would execute an unbounded number of iterations. This may result in early termination of the loop, other non-local effects, or even device loss.
So looks like acceptable behaviour for this tab.

I was wondering if the fact that it then prevents shaders from running after it executes was a problem. This problem is fixed if chrome is restarted.

[Max-Sepp]

I just ran a test and this can effect shaders that are currently running e.g. running crash.html will causes a shader running in a different window to also crash with the same error

[Max-Sepp]

Video of crash:

Hang_device.mov