1412973 - Crash in <name omitted> | nsIFrame::GetOffsetToCrossDoc

Status: RESOLVED WORKSFORME

(Firefox for Android Graveyard :: General, defect)

Description

[Paul [pwd]]

This bug was filed from the Socorro interface and is 
report bp-8e74eea6-87a0-4db4-8d60-7f8410171030.
=============================================================

Comment 1

[Alex Thompson]

Attachment: logcat.txt with crashes at 12:44, 12:35, 12:35, 12:34 and more

I'm seeing this bug after updating to the latest nightly build (20171030103740).

I had enabled stylo on android, but after disabling it was still present. 

I've attached a logcat output from my phone (Pixel, 2016, Oreo, October security update) in case it is useful.

Comment 2

[Marcia Knous [:marcia]]

I actually see a large spike in crashes in the nightly build: http://bit.ly/2yh1tMd. Over 118 crashes so far. 

Possible regression range based on Build ID: https://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=66f9b72b87297adf712b14be58df13c2333bb3a9&tochange=d3910b7628b8066d3f30d58b17b5824b05768854

I see Bug 1411802 landed in this timeframe. Adding ni on :m_kato in case he has some ideas.

Comment 3

[Makoto Kato [:m_kato]]

(In reply to Alex Thompson from comment #1)

> I had enabled stylo on android, but after disabling it was still present. 

If stylo is enabled, this depends on bug 1411802.  But even if disabled, it seems to be occur. Right?

Another big change of this range is that our toolchan is changed from gcc to clang (bug 1163171).

Comment 4

[Makoto Kato [:m_kato]]

Alex's crash is bug 1411267 from adb logcat log.

Comment 5

[Alex Thompson]

Makoto, are you sure they are all from bug 1411267? I think I've been getting crashes from both recently. They are identified differently in crash-stats. For example, crash id: 632b03be-a865-4ba0-8e36-4af6b0171031 (12:35 on 31/10) has the signature  (<name omitted> | nsIFrame::GetOffsetToCrossDoc). Sorry, I should have been clear about two possible crash sources in my original comment.(In reply to Makoto Kato [:m_kato] from comment #3)

> (In reply to Alex Thompson from comment #1)
> 
> > I had enabled stylo on android, but after disabling it was still present. 
> 
> If stylo is enabled, this depends on bug 1411802.  But even if disabled, it
> seems to be occur. Right?
> 
> Another big change of this range is that our toolchan is changed from gcc to
> clang (bug 1163171).

Yes, I have still seen many instances of this crash after setting servo back to disabled (the default)

Comment 6

[Alex Thompson]

Following up to add, I haven't been able to reproduce this bug after updating to build id: 20171031234943. I was only seeing this with build id: 20171030103740. (Still getting plenty of bug 1411267, but that's a different problem).

Comment 7

[Makoto Kato [:m_kato]]

(In reply to Alex Thompson from comment #5)
> Makoto, are you sure they are all from bug 1411267?

I only found the following crash log.

10-25 12:02:28.430  8672  8682 F libc    : Fatal signal 6 (SIGABRT), code -6 in tid 8682 (HeapTaskDaemon)
10-25 12:02:28.531  8718  8718 F DEBUG   : *** *** *** *** *** *** *** *** *** *** *** *** *** *** *** **
*
10-25 12:02:28.531  8718  8718 F DEBUG   : Build fingerprint: 'google/sailfish/sailfish:8.0.0/OPR1.170623
.027/4292972:user/release-keys'
10-25 12:02:28.531  8718  8718 F DEBUG   : Revision: '0'
10-25 12:02:28.531  8718  8718 F DEBUG   : ABI: 'arm'
10-25 12:02:28.531  8718  8718 F DEBUG   : pid: 8672, tid: 8682, name: HeapTaskDaemon  >>> org.mozilla.fe
nnec_aurora <<<
...

Since crash report hooks other crashes, there is no data. Example, the following is crash.

10-30 15:48:08.869   939  2349 I WindowManager: WIN DEATH: Window{10cc355 u0 org.mozilla.fennec_aurora/or
g.mozilla.gecko.BrowserApp}

Since crash report gets signal, there is no output for SIGSEGV.

> I think I've been
> getting crashes from both recently. They are identified differently in
> crash-stats. For example, crash id: 632b03be-a865-4ba0-8e36-4af6b0171031
> (12:35 on 31/10) has the signature  (<name omitted> |
> nsIFrame::GetOffsetToCrossDoc). Sorry, I should have been clear about two
> possible crash sources in my original comment.

Since telemetry doesn't have servo prefs, I need more information.  But most crash seems on twitter.


(In reply to Alex Thompson from comment #6)
> Following up to add, I haven't been able to reproduce this bug after
> updating to build id: 20171031234943. I was only seeing this with build id:
> 20171030103740. (Still getting plenty of bug 1411267, but that's a different
> problem).

This signature is 20171030103740 only.

Comment 8

[Makoto Kato [:m_kato]]

no crash after 10/31.