1371057 - Intermittent test_nuke_sandbox_event_listeners.js | application crashed [@ mozalloc_abort(char const * const)]

Status: RESOLVED INCOMPLETE

(Core :: XPCOM, defect)

Description

[Treeherder Bug Filer]

Filed by: wkocher [at] mozilla.com

https://treeherder.mozilla.org/logviewer.html#?job_id=105202579&repo=mozilla-inbound

https://archive.mozilla.org/pub/firefox/tinderbox-builds/mozilla-inbound-win64-debug/1496835210/mozilla-inbound_win8_64-debug_test-xpcshell-bm110-tests1-windows-build3738.txt.gz

Comment 1

[Andrew McCreight [:mccr8]]

Hitting an assert in nsEventQueue::~nsEventQueue().

Comment 2

[Nathan Froyd [:froydnj]]

(In reply to Andrew McCreight [:mccr8] from comment #1)
> Hitting an assert in nsEventQueue::~nsEventQueue().

We're, uh, deleting the main thread there, I guess?  That's quite peculiar!

Comment 3

[Bill McCloskey [inactive unless it's an emergency] (:billm)]

(In reply to Nathan Froyd [:froydnj] from comment #2)
> (In reply to Andrew McCreight [:mccr8] from comment #1)
> > Hitting an assert in nsEventQueue::~nsEventQueue().
> 
> We're, uh, deleting the main thread there, I guess?  That's quite peculiar!

Is it? We've already shut down the thread manager at this point. I don't know of anything else that would be intentionally holding on to the main thread.

What is strange is that we pump the main thread right before we shut it down in the thread manager. So it's not clear how new events would show up after that. Maybe the pumping is failing to do its job, or maybe someone is somehow sneaking an event in? We could add some assertions to check for those things.