1549321 - Crash in [@ IPCError-browser | RecvCreateAudioIPCConnection CubebUtils::CreateAudioIPCConnection failed]

Status: RESOLVED FIXED

(Core :: Audio/Video: cubeb, defect, P2)

Description

[Gabriele Svelto [:gsvelto]]

This bug is for crash report bp-865dd2f6-9d4f-455b-9f3e-4ea3b0190505.

Not putting the stack in this one for now because the stacks visible on Socorro are all over the place. I think we're looking at the wrong minidump since there's two in the crash reports. From the signature this is related to this must be related to:

https://searchfox.org/mozilla-central/rev/b2015fdd464f598d645342614593d4ebda922d95/dom/ipc/ContentParent.cpp#4395
https://searchfox.org/mozilla-central/rev/b2015fdd464f598d645342614593d4ebda922d95/dom/media/CubebUtils.cpp#408

I'll try to manually extract a signature later and see what's wrong with Socorro.

Comment 1

[Matthew Gregan [:kinetik]]

I just noticed the crash report is for Windows 10. We shouldn't be in CreateAudioIPCConnection or anything related to AudioIPC as that code is not yet enabled on Windows, so this crash can only occur if the user has manually set the media.cubeb.sandbox pref to true.

Comment 3

[Matthew Gregan [:kinetik]]

I misunderstood the purpose of IPC_FAIL() and how it interacts with async IPC. My expectation was that IPC_FAIL() would cause the promise returned by SendCreateAudioIPCConnection() to be rejected. Per bug 1313200 comment 9 through 11, that's not the case.

Comment 4

[Matthew Gregan [:kinetik]]

Attachment: Bug 1549321 - Pass AudioIPC init failure back to child rather than IPC_FAIL() killing parent. r?achronop

Comment 5

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/ecdebbbb2253
Pass AudioIPC init failure back to child rather than IPC_FAIL() killing parent.  r=achronop

Comment 6

[Andreea Pavel [:apavel]]

https://hg.mozilla.org/mozilla-central/rev/ecdebbbb2253

Comment 7

[Ryan VanderMeulen [:RyanVM]]

Crash volume is low enough that I assume we can just let this ride the trains.