1897636 - Crash in [@ IPCError-browser | GPUProcessKill]

Status: RESOLVED FIXED

(Core :: Graphics, defect)

Description

[Gabriele Svelto [:gsvelto]]

Crash report: https://crash-stats.mozilla.org/report/index/ef27fe0f-5027-4d1e-bf5f-09a620240519

Reason: EXCEPTION_BREAKPOINT

Top 10 frames:

0  ntdll.dll  NtWaitForSingleObject
1  KERNELBASE.dll  WaitForSingleObjectEx
2  xul.dll  mozilla::mscom::EnsureMTA::SyncDispatchToPersistentThread(nsIRunnable*)  ipc/mscom/EnsureMTA.cpp:172
3  xul.dll  mozilla::mscom::EnsureMTA::SyncDispatch(nsCOMPtr<nsIRunnable>&&, mozilla::msc...  ipc/mscom/EnsureMTA.cpp:185
3  xul.dll  mozilla::mscom::EnsureMTA::EnsureMTA(mozilla::WMFDecoderModule::Init::<lambda...  ipc/mscom/EnsureMTA.h:73
3  xul.dll  mozilla::WMFDecoderModule::Init(mozilla::WMFDecoderModule::Config)  dom/media/platforms/wmf/WMFDecoderModule.cpp:132
4  xul.dll  mozilla::PDMInitializer::InitGpuPDMs()  dom/media/platforms/PDMFactory.cpp:90
4  xul.dll  mozilla::PDMInitializer::InitPDMs()  dom/media/platforms/PDMFactory.cpp:199
5  xul.dll  mozilla::PDMFactory::EnsureInit::<lambda_0>::operator()() const  dom/media/platforms/PDMFactory.cpp:316
5  xul.dll  mozilla::detail::RunnableFunction<`lambda at /builds/worker/checkouts/gecko/d...  xpcom/threads/nsThreadUtils.h:548

I'm not sure what's going on, but it appears to be a new signature, so maybe a regression?

Comment 1

[BugBot [:suhaib / :marco/ :calixte]]

The bug is linked to a topcrash signature, which matches the following criterion:

• Top 10 desktop browser crashes on nightly (startup)

For more information, please visit BugBot documentation.

Comment 2

[Andrew McCreight [:mccr8]]

About half of these have WMFDecoderModule stuff in the stack like in comment 0, which I think would make this Audio/Video: Playback. The other half (like this one bp-cfc521dd-26df-4c6e-b626-48e3c0240520) have only generic windows event loop stuff in the stack so I'm not sure if that's the same thing or not.

Comment 3

[Sotaro Ikeda [:sotaro]]

:alwu, do you have any idea about the crash?

Comment 4

[Alastor Wu [:alwu]]

The earliest crash signature appeared on May 15th on Fx128, and we started adding GPUProcessKill in bug 1880503 around that time. That can explain why this signature didn't show up before, but it didn't explain why the crash would happen. This crash seems related with the IPC timeout, and PDMFactory performs initialization on the main thread, which could take too much time and cause the time out.

In bug 1885671, we already move the decoder module's initialization off the main thread, and we could do the same thing here as well.

Comment 5

[Alastor Wu [:alwu]]

Attachment: Bug 1897636 - allow the initialization for PDMs on any thread.

Since bug 1885671, we no longer need to run the initialization for
decoder module on the main thread, so we should allow running
InitPDMs() on non-main threads as well.

Comment 6

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/6aa6da68d232
allow the initialization for PDMs on any thread. r=media-playback-reviewers,padenot

Comment 7

[Sandor Molnar[:smolnar]]

https://hg.mozilla.org/mozilla-central/rev/6aa6da68d232

Comment 8

[Atila Butkovits]

Backed out for causing Xpcshell crashes at docshell/test/unit/test_privacy_transition.js (it started failing on central: https://treeherder.mozilla.org/jobs?repo=mozilla-central&selectedTaskRun=YELPcUdIRZWHfytwW8HKrA.0&resultStatus=testfailed%2Cbusted%2Cexception%2Cretry%2Cusercancel&searchStr=windows%2C11%2Cx86%2C22h2%2Cwebrender%2Cdebug%2Cxpcshell%2Ctests%2Ctest-windows11-32-2009-qr%2Fdebug-xpcshell%2Cx3&tochange=71204917f64a971fbd6cf3b0b35b3193a5446f0e&fromchange=71204917f64a971fbd6cf3b0b35b3193a5446f0e )

Backout link: https://hg.mozilla.org/integration/autoland/rev/8b60060d7d6c8fe6ddeb300b39f9d57bf3835f45

Push with failures: https://treeherder.mozilla.org/jobs?repo=autoland&resultStatus=testfailed%2Cbusted%2Cexception%2Cretry%2Cusercancel&searchStr=windows%2C10%2Cx64%2C2009%2Cwebrender%2Cdebug%2Cxpcshell%2Ctests%2Ctest-windows10-64-2009-qr%2Fdebug-xpcshell%2Cx1&revision=6aa6da68d2321c79fa7832f103f00638114d9c12

Failure log: https://treeherder.mozilla.org/logviewer?job_id=459493211&repo=autoland&lineNumber=1656

Comment 9

[Norisz Fay [:noriszfay]]

Backout merged to central: https://hg.mozilla.org/mozilla-central/rev/d7f4e0eadb02

Comment 10

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/182c1293f1ae
allow the initialization for PDMs on any thread. r=media-playback-reviewers,padenot

Comment 11

[Serban Stanca [:SerbanS]]

https://hg.mozilla.org/mozilla-central/rev/182c1293f1ae

Comment 12

[Alastor Wu [:alwu]]

None of crashes happened after landing my patch [1] is related with media stack, so we can file another bug to keep tracking other crashes sharing the same signature.

[1]
https://bit.ly/3X0lbGg
https://bit.ly/4dYijjj
https://bit.ly/3X433LW