1283069 - Crash in _preextrapolate_helper

Status: RESOLVED INCOMPLETE

(Core :: WebRTC: Audio/Video, defect, P1)

Description

[tony.lepmets]

This bug was filed from the Socorro interface and is 
report bp-fa016cb2-8558-4765-8ce4-397602160629.
=============================================================

Crash occurs when ending a WebRTC call, eg Chrome and Firefox are in a video call, when Chrome ends the call on their side (by closing the tab for example) then shortly after Firefox crashes. This doesn't happen 100% of the time but when establishing and ending the call several times in a row then the crash can occur mid call as well.

This crash does not happen with 48 and did not happen before 47.

Comment 1

[Byron Campen [:bwc]]

I see lots of webrtc.org stuff going on in other threads, but no code of our own (there are 7 of these crashes in the last week on crash-stats, all on OS X, https://crash-stats.mozilla.com/signature/?signature=_preextrapolate_helper&_columns=date&_columns=product&_columns=version&_columns=build_id&_columns=platform&_columns=reason&_columns=address&page=1#reports)

Comment 2

[Nils Ohlmeier [:drno]]

And they all seem to have happened on Mac OSX 10.10.5 14F27, which is the latest Yosemite update from Apple from August 2015.
Do you have any way to confirm if the crash also happens on El Capitan (10.11)?

Comment 3

[indrek]

I tried to reproduce it:
1) locally (peer connection) - everything worked
2) using google stun server - everything worked
3) using custom turn server (coturn) - browser crashed
4) using custom turn server (reTurnServer) - browser crashed
Will try to create a minimal page to replicate it.

Comment 4

[tony.lepmets]

Also crashes on El Capitan: https://crash-stats.mozilla.com/report/index/16b37840-95f1-4eeb-b94f-606b52160701

Comment 5

[Maire Reavy [:mreavy]]

Tony/indrek - can you re-verify it doesn't happen after 47?  Specifically 48 and Nightly 50.

Comment 6

[Nils Ohlmeier [:drno]]

(In reply to Byron Campen [:bwc] from comment #1)
> I see lots of webrtc.org stuff going on in other threads, but no code of our
> own (there are 7 of these crashes in the last week on crash-stats, all on OS
> X,
> https://crash-stats.mozilla.com/signature/
> ?signature=_preextrapolate_helper&_columns=date&_columns=product&_columns=ver
> sion&_columns=build_id&_columns=platform&_columns=reason&_columns=address&pag
> e=1#reports)

What happened here? Does Socorro suffer from amnesia or Alzheimer?

I'm 100% sure that when Byron and me commented on this the above search showed something like 10 crashes. And now Socorro claims that their is no a single dupe reports of this? It does not even find the two reports linked here in the bug report directly as the same... ?!?

Comment 7

[Nils Ohlmeier [:drno]]

Turns out that crash-stats has an implicit 7 days filter build in, and apparently this crash did not happen within the last 7 days any more.
Here is the same query with an extra date filter so we can still see the old crashes: https://crash-stats.mozilla.com/signature/?date=%3E2016-06-20&signature=_preextrapolate_helper&_columns=date&_columns=product&_columns=version&_columns=build_id&_columns=platform&_columns=reason&_columns=address&_sort=-date&page=1

Comment 8

[indrek]

Crash does not happen in FF48.

Comment 9

[Randell Jesup [:jesup] (needinfo me)]

No reports from FF48 or later.  Likely fixed.