619601 - Crash [@ _moz_cairo_surface_create_similar ]

Status: RESOLVED WONTFIX

(Core :: Graphics, defect, P3)

Description

[Alice0775 White]

Build Identifier:
Mozilla/5.0 (Windows NT 6.1; WOW64; rv:2.0b9pre) Gecko/20101215 Firefox/4.0b9pre ID:20101215030322

Browser crashes randomly with crash reports after loading URL, and also crashes during restore session.
Maybe this crash is due to add-ons.


Reproducible: Always

Steps to Reproduce:
1. Start Minefield with new profile
2. Open any page in tabs more than 15.
3. Open URL ( http://www.youtube.com/watch?v=BH0BNbwqNK8&webm=1&html5=1 )
4. Restart browser with session restore

Actual Results:
 Browser crashes randomly in step 3 or step 4.

Expected Results:
 No crashes.

Comment 1

[Alice0775 White]

Crash report
bp-63ed7053-d94f-4ed1-8936-bae1a2101215

Comment 2

[Alice0775 White]

Regression window:
Works:
http://hg.mozilla.org/mozilla-central/rev/1a149c7a4fa2
Mozilla/5.0 (Windows NT 6.1; WOW64; rv:2.0b8pre) Gecko/20101209 Firefox/4.0b8pre ID:20101209191612
Fails:
http://hg.mozilla.org/mozilla-central/rev/8be85c96f560
Mozilla/5.0 (Windows NT 6.1; WOW64; rv:2.0b8pre) Gecko/20101209 Firefox/4.0b8pre ID:20101209203804
Pushlog:
http://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=1a149c7a4fa2&tochange=8be85c96f560

Comment 3

[Arie Paap [:wildmyron]]

dup of bug 619044, but might be better to dupe that one forward

Comment 5

[Timothy Nikkel (:tnikkel)]

Attachment: patch

I guess we have to restore the null check for now, and try to reproduce and investigate this crash.

Comment 6

[Timothy Nikkel (:tnikkel)]

The null check patch landed
http://hg.mozilla.org/mozilla-central/rev/a5413c3c1013
but I'm keeping this open to investigate the root cause.

Comment 7

[(no longer active)]

This seems to have caused 3.33% Ts regression on OSX 10.5 and 1.6% Ts regression on XP.

Is this expected?

Comment 8

[Timothy Nikkel (:tnikkel)]

That's probably impossible. More likely is that the places merge on revision before is the cause. I've seen the automated mails blame one or more changesets after the one that caused the improvement/regression many times before.

Comment 9

[Joe Drew (not getting mail)]

Timothy, do you have any intention of investigating this? If not, we could probably find someone to, though it doesn't seem like a high priority :/

Comment 10

[Timothy Nikkel (:tnikkel)]

I tried reproducing but failed to do so, and then that was it.

Comment 11

[u279076]

I still see reports of this in the wild but at extremely low volume. As an example: bp-a5d70c36-f73c-41c3-b5d6-79b412160427 (Firefox 45.0.2 on Windows XP).

Comment 12

[BugBot [:suhaib / :marco/ :calixte]]

Closing because no crashes reported for 12 weeks.