970366 - crash in NS_ABORT_OOM(unsigned int) | mozilla::MediaQueue<mozilla::AudioData>::Push(mozilla::AudioData*)

Status: RESOLVED DUPLICATE of bug 969826

(Core :: Audio/Video, defect)

Description

[Tracy Walker [:tracy]]

This bug was filed from the Socorro interface and is 
report bp-f985f695-f4d4-427b-a1a9-7b7422140209.
=============================================================

New oom crash on Nightly that began with builds on 2014020703 with a high OOM Allocation Size of 33554432 precisely in every crash.

The Nightly change set from 20140206030201 to 20140207030201: http://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=1e9f169c9715&tochange=dedf12c4e805. At a glance, I suspect bug 960873.

Comment 1

[Tracy Walker [:tracy]]

Reported signature is #2 topcrash on Nightly at 5%.  

At #5 is what I believe to be related crash; https://crash-stats.mozilla.com/report/list?product=Firefox&range_value=7&range_unit=days&date=2014-02-10&signature=NS_ABORT_OOM%28unsigned+__int64%29+|+mozilla%3A%3AAudioCompactor%3A%3APush%3Cmozilla%3A%3ADirectShowCopy%3E%28__int64%2C+__int64%2C+int%2C+unsigned+int%2C+unsigned+int%2C+mozilla%3A%3ADirectShowCopy%29&version=Firefox%3A30.0a1#tab-sigsummary

It is AMD64 only with a more random, but high, OOM allocation.
It has the same regression range and suspect bug here. I can file a separate bug if you feel they would have to be fixed independently.

Comment 2

[Tracy Walker [:tracy]]

Ack, #1 Nightly topcrasher at 20% is also related.  https://crash-stats.mozilla.com/report/list?product=Firefox&range_value=7&range_unit=days&date=2014-02-10&signature=mozalloc_abort%28char+const*+const%29+|+mozalloc_handle_oom%28unsigned+int%29+|+moz_xmalloc+|+mozilla%3A%3AAudioCompactor%3A%3APush%3Cmozilla%3A%3ADirectShowCopy%3E%28__int64%2C+__int64%2C+int%2C+unsigned+int%2C+unsigned+int%2C+mozilla%3A%3ADirectShowCopy%29&version=Firefox%3A30.0a1#tab-table

There is a fourth signature at #13, that also related and spiked in the same regression range as the others: https://crash-stats.mozilla.com/report/list?product=Firefox&range_value=7&range_unit=days&date=2014-02-10&signature=mozalloc_abort%28char+const*+const%29+|+mozalloc_handle_oom%28unsigned+__int64%29+|+errno&version=Firefox%3A30.0a1#tab-reports

Again, let me know if you think each signature should be treated independently.

Comment 3

[Benjamin Smedberg]

Then you should probably cc the author/reviewer of that bug ;-)

https://crash-stats.mozilla.com/search/?signature=%3DNS_ABORT_OOM%28unsigned+int%29+|+mozilla%3A%3AMediaQueue%3Cmozilla%3A%3AAudioData%3E%3A%3APush%28mozilla%3A%3AAudioData*%29&_columns=date&_columns=signature&_columns=product&_columns=version&_columns=build_id&_columns=platform&_columns=email&_columns=url&_columns=user_comments&_columns=oom_allocation_size

Comment 4

[Ben Kelly [:bkelly, not reviewing]]

Is this a dupe of bug 969826?  That was a known crasher over the weekend that should have been addressed in yesterday's nightly.

Comment 5

[Tracy Walker [:tracy]]

(In reply to Ben Kelly [:bkelly] from comment #4)
> Is this a dupe of bug 969826?  That was a known crasher over the weekend
> that should have been addressed in yesterday's nightly.

Yes, this is a dupe of 969826. The fix there has stopped all four of the signatures I reported here from happening on builds from 20140210 and 20140211.

I'll add all four signatures to the patched bug.

Thanks Ben!