Closed Bug 1251895 Opened 4 years ago Closed 4 years ago

Frequent crashes [@ PL_HashTableRawLookup][@ GetSerialNumber] when getting stacks for leaks in automation

Categories

(Core :: XPCOM, defect)

defect
Not set

Tracking

()

RESOLVED FIXED
mozilla47
Tracking Status
firefox47 --- fixed

People

(Reporter: RyanVM, Assigned: froydnj)

Details

(Keywords: crash)

Attachments

(1 file)

I've been trying to get stacks for the various intermittent leaks we currently have and am seeing a pretty frequent crash that appears to be in the stack getting code itself.

https://treeherder.mozilla.org/logviewer.html#?job_id=17321578&repo=try
https://treeherder.mozilla.org/logviewer.html#?job_id=17321964&repo=try
Do you have a set of tests/chunks that you frequently run to get this?

Are you trying to log specific object types or object IDs when you do this?  We shouldn't even be touching GetSerialNumber if we're not logging specific things...
Flags: needinfo?(ryanvm)
Ah, trying to log particular objects on all of those.  I think I understand what's going on here; I'll try to put together a patch on Monday.
Assignee: nobody → nfroyd
I think this solves Ryan's crashes (try run in progress): GetSerialNumber is
accessing global state, sometimes accessed under a lock (for logging
AddRef/Release) and sometimes not (for logging COMPtr AddRef/Release).  We
should be accessing it under the lock all the time...
Attachment #8724884 - Flags: review?(continuation)
Comment on attachment 8724884 [details] [diff] [review]
don't race on nsTraceRefcnt's object serial number tables

Review of attachment 8724884 [details] [diff] [review]:
-----------------------------------------------------------------

Yikes!
Attachment #8724884 - Flags: review?(continuation) → review+
https://hg.mozilla.org/mozilla-central/rev/b0d4a61fc0f0
Status: NEW → RESOLVED
Closed: 4 years ago
Resolution: --- → FIXED
Target Milestone: --- → mozilla47
You need to log in before you can comment on or make changes to this bug.