1954234 - Crash in [@ mozilla::webgpu::WebGPUParent::MapCallback]

Status: NEW

(Core :: Graphics: WebGPU, defect)

Description

[Andrew McCreight [:mccr8]]

Crash report: https://crash-stats.mozilla.org/report/index/1dd47a53-eebc-4739-a13d-1f6b00250314

MOZ_CRASH Reason:

MOZ_RELEASE_ASSERT(!error.GetError())

Top 10 frames:

0  XUL  MOZ_CrashSequence(void*, long)  mfbt/Assertions.h:272
0  XUL  mozilla::webgpu::WebGPUParent::MapCallback(unsigned char*, mozilla::webgpu::f...  dom/webgpu/ipc/WebGPUParent.cpp:662
1  XUL  <alloc::boxed::Box<F, A> as core::ops::function::FnOnce<Args>>::call_once  /builds/worker/fetches/rustc/lib/rustlib/src/rust/library/alloc/src/boxed.rs:1993
1  XUL  wgpu_core::device::UserClosures::fire  third_party/rust/wgpu-core/src/device/mod.rs:179
2  XUL  wgpu_core::device::global::<impl wgpu_core::global::Global>::poll_all_devices  third_party/rust/wgpu-core/src/device/global.rs:2020
3  XUL  wgpu_server_poll_all_devices  gfx/wgpu_bindings/src/server.rs:180
4  XUL  DispatchToMethod<mozilla::webgpu::WebGPUParent, void (mozilla::webgpu::WebGPU...  ipc/chromium/src/base/tuple.h:381
4  XUL  base::BaseTimer<mozilla::webgpu::WebGPUParent, true>::TimerTask::Run()  ipc/chromium/src/base/timer.h:157
5  XUL  mozilla::DelayedRunnable::Notify(nsITimer*)  xpcom/threads/DelayedRunnable.cpp:92
6  XUL  nsTimerImpl::Fire(int)::$_1::operator()(nsCOMPtr<nsITimerCallback> const&) const  xpcom/threads/nsTimerImpl.cpp:677

I only see two of these crashes, but there's a release assert so maybe there's something useful here.

Comment 1

[Erich Gubler [:ErichDonGubler] (he/him)]

Aggregating on the URLs in the crash reports, these all seem to be based on localhost servers for WebGPU CTS and another unidentified application. I'm not concerned about production yet, but this is definitely a weird case that I can see getting prioritized as we get closer to launch.

Comment 2

[Mayank Bansal]

Bug 1972521 has a testcase that crashed for me with the same signature.

Comment 3

[Erich Gubler [:ErichDonGubler] (he/him)]

Linking to bug 1973591, which has some crash instances in the MapCallback (but crashing for a different reason).