Closed Bug 1227718 Opened 9 years ago Closed 8 years ago

Intermittent Main app process exited normally | application crashed [@ DebugBreak + 0x2] after Assertion failure: lock->owner == 0, at prulock.c:194

Categories

(Core :: Graphics: Layers, defect)

defect
Not set
normal

Tracking

()

RESOLVED FIXED
mozilla48
Tracking Status
e10s + ---
firefox47 --- affected
firefox48 --- fixed

People

(Reporter: KWierso, Assigned: bas.schouten)

References

(Blocks 1 open bug)

Details

(Keywords: intermittent-failure, Whiteboard: [e10s-orangeblockers])

This is a lock inside AsyncTransactionTrackersHolder, that we're deleting while it's still held by someone.  The process uptime in one log I pulled was only 6 seconds, so we might be still starting up.
Component: NSPR → Graphics: Layers
Product: NSPR → Core
Version: other → unspecified
Sotaro, can you reproduce this locally?
Flags: needinfo?(milan) → needinfo?(sotaro.ikeda.g)
It seems difficult to reproduce it locally. AsyncTransactionTrackersHolder::Finalize() expects that Layers are already shut down. The problem seems to suggest that layers are not shutdown yet.

One simple way to address seems that deferring AsyncTransactionTrackersHolder::Finalize() call more.
I am going to remove the mutex in Bug 1252835.
See Also: → 1252835
See Also: → 1234997
See Also: → 1237257
I wonder if this is a similar underlying problem to the various e10s work_queue_.empty() bugs we've got as well.
See Also: → 1252283
Assignee: wtc → sotaro.ikeda.g
IPC::Message leaks appear to routinely ride along with these as well.
See Also: → 1252349
See Also: → 1242221
Whiteboard: [e10s-orangeblockers]
Bug 1254829 might be related.
Flags: needinfo?(sotaro.ikeda.g)
See Also: → 1254829
See Also: 1252283
See Also: 1254829
No hits since bug 1262898 landed. Optimistically declaring victory.
Assignee: sotaro.ikeda.g → bas
Status: NEW → RESOLVED
Closed: 8 years ago
Depends on: 1262898
Resolution: --- → FIXED
See Also: 1252835, 1242221
Target Milestone: --- → mozilla48
See Also: 1234997
See Also: 1237257
You need to log in before you can comment on or make changes to this bug.