Closed Bug 671916 Opened 8 years ago Closed 8 years ago

crash @ mozalloc_abort | NS_DebugBreak_P | X11Error during windowless Flash Player painting

Categories

(Core :: Plug-ins, defect, critical)

All
Linux
defect
Not set
critical

Tracking

()

RESOLVED FIXED
mozilla8

People

(Reporter: Logan, Assigned: BenWa)

References

(Depends on 1 open bug)

Details

(Keywords: crash, regression)

Crash Data

Attachments

(1 file)

This bug was filed from the Socorro interface and is 
report bp-3ac6f8ab-dd9e-44f7-90d7-7a2382110715 .
============================================================= 
I've now received this crash twice (see also https://crash-stats.mozilla.com/report/index/bp-4b1d89fa-103e-4c75-bff4-201f42110715).  The Flash plugin is now intermittently crashing on the latest trunk updates under Ubuntu 11.04.
I've noticed this a lot, as well. It seems to crash for no obvious reason, even when flash isn't running. Started on July 13, maybe a regression? I'll look into a range.
Hardware: x86 → All
https://crash-stats.mozilla.com/report/list?signature=mozalloc_abort%20%7C%20NS_DebugBreak_P%20%7C%20X11Error

Frame 	Module 	Signature [Expand] 	Source
0 	libmozalloc.so 	mozalloc_abort 	memory/mozalloc/mozalloc_abort.cpp:66
1 	libxul.so 	NS_DebugBreak_P 	xpcom/base/nsDebugImpl.cpp:388
2 	libxul.so 	X11Error 	toolkit/xre/nsX11ErrorHandler.cpp:199
3 	libX11.so.6.3.0 	libX11.so.6.3.0@0x3b298 	
4 	libX11.so.6.3.0 	libX11.so.6.3.0@0x4192e 	
5 	libX11.so.6.3.0 	libX11.so.6.3.0@0x41fb5 	
6 	libX11.so.6.3.0 	libX11.so.6.3.0@0x35846 	
7 	libxul.so 	mozilla::plugins::PluginInstanceChild::ShowPluginFrame 	dom/plugins/ipc/PluginInstanceChild.cpp:3174
8 	libxul.so 	mozilla::plugins::PluginInstanceChild::InvalidateRectDelayed 	dom/plugins/ipc/PluginInstanceChild.cpp:3271
9 	libxul.so 	RunnableMethod<mozilla::plugins::PluginInstanceChild, void , Tuple0>::Run 	ipc/chromium/src/base/tuple.h:383
10 	libxul.so 	MessageLoop::DeferOrRunPendingTask 	ipc/chromium/src/base/message_loop.cc:342
11 	libxul.so 	MessageLoop::DoWork 	ipc/chromium/src/base/message_loop.cc:450
12 	libxul.so 	base::MessagePumpForUI::RunWithDispatcher 	ipc/chromium/src/base/message_pump_glib.cc:199
13 	libxul.so 	base::MessagePumpForUI::Run 	ipc/chromium/src/base/message_pump_glib.h:59
14 	libxul.so 	MessageLoop::Run 	ipc/chromium/src/base/message_loop.cc:218
15 	libxul.so 	XRE_InitChildProcess 	toolkit/xre/nsEmbedFunctions.cpp:510
16 	plugin-container 	main 	ipc/app/MozillaRuntimeMain.cpp:81
17 	libc-2.11.1.so 	libc-2.11.1.so@0x16bd5 	
18 	plugin-container 	plugin-container@0xe40 	
19 	plugin-container 	plugin-container@0xeff 	
20 	ld-2.11.1.so 	ld-2.11.1.so@0xe02f 	
21 	ld-2.11.1.so 	ld-2.11.1.so@0x1c8f7
This seems to happen with a whole range of Linux (kernel) versions, so it's also not a single person or distro, must be more generic.
Also, it's 8.0a1-trunk-only and first started happening on 2011-07-13 versions. Who is a Linux dev guy we could get into the loop here? Any Linux-specific checkins on the 12th or perhaps 13th that could have triggered that?
Summary: crash mozalloc_abort → crash @ mozalloc_abort | NS_DebugBreak_P | X11Error
Possible pushlog regression range: http://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=8753de11b181&tochange=931f06b80727.

There was a merge from mozilla inbound that day.
Karl, bsmedberg, I wasn't sure who could help with this bug but dbaron suggested I cc you both. This crash started showing up on 7/13. Any ideas?
(In reply to comment #4)
> Possible pushlog regression range:
> http://hg.mozilla.org/mozilla-central/
> pushloghtml?fromchange=8753de11b181&tochange=931f06b80727.

Marcia, it could also be in the next "slot" of nightly, as it's possible that those very few reports from 13th build IDs were on "hourly" builds. A looking at http://hg.mozilla.org/mozilla-central/pushloghtml?startdate=2011-07-12&enddate=2011-07-14 I see some very X11/GTK-specific changesets at the top of the list.

Ginn, could one of your changes be related?
It's a day when beta version of flash 11 has been released. :-)
http://labs.adobe.com/downloads/flashplayer11.html
(In reply to comment #7)
> It's a day when beta version of flash 11 has been released. :-)
> http://labs.adobe.com/downloads/flashplayer11.html

That's an interesting data point, but unfortunately doesn't explain why this happens on 8.0a1 only and not on any other versions.
Product: Firefox → Core
QA Contact: general → general
Bug 644707 is the reason why we see this as a SIGSEGV only on 8.
Component: General → Plug-ins
Depends on: 639725
QA Contact: general → plugins
Summary: crash @ mozalloc_abort | NS_DebugBreak_P | X11Error → crash @ mozalloc_abort | NS_DebugBreak_P | X11Error during windowless Flash Player painting
We aren't getting any data about Flash versions for this one so we can't correlate it specifically to Flash 11. It's the top plugin crash on the trunk right now.
I'm still getting this crash with Flash 10.3.181.34, so it's definitely not Flash Player 11-specific (if at all).
Are there any console messages to indicate the X Error?
Do you have steps to reproduce.(In reply to comment #10)

> We aren't getting any data about Flash versions

FWIW we can map the debug identifier for libflashplayer.so to versions.
When I try to track this down using Linux I end up in a Mac OS X bug. Anyway, this is what I've found using FlashPlayer 11.0.1.60:

bp-128a6a3b-242c-42d6-8842-c99842110809
[@ mozalloc_abort | NS_DebugBreak_P | X11Error ] 

Last good nightly: 2011-07-13
First bad nightly: 2011-07-14

Pushlog:
http://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=931f06b80727&tochange=34b0b3bc6984

The first bad revision is:
changeset:   72713:5cea57e451dd
user:        Benoit Girard <b56girard@gmail.com>
date:        Tue Jul 12 10:31:18 2011 -0400
summary:     Bug 663259 - Enable Mac Async plugin by default. r=cjones,mattwoodrow

http://hg.mozilla.org/mozilla-central/rev/5cea57e451dd

-------

STR:

1. Visit http://chessinkorea.com/xe/3959 with a new, clean profile
2. Shake the right scroll bar fast for 20 seconds

Actual Results:
FlashPlayer crash

Expected Results:
No FlashPlayer crash
Thanks very much, Thomas!
Blocks: 663259
Thanks for finding the regression range! The problem is here:
http://hg.mozilla.org/mozilla-central/rev/5cea57e451dd#l6.228

I will post a patch first thing in the morning and will spin off a try build.
Attached patch patch v1Splinter Review
See: http://hg.mozilla.org/mozilla-central/rev/5cea57e451dd#l6.228
This line was not meant to be changed for non osx platforms.
Attachment #552071 - Flags: review?(jones.chris.g)
Attachment #552071 - Flags: review?(jones.chris.g) → review+
Can someone confirm that the problem is fixed before we land this?

Try builds:
https://ftp.mozilla.org/pub/mozilla.org/firefox/try-builds/b56girard@gmail.com-92f1de674411/
I've tested https://ftp.mozilla.org/pub/mozilla.org/firefox/try-builds/b56girard@gmail.com-92f1de674411/try-linux64/firefox-8.0a1.en-US.linux-x86_64.tar.bz2 together with the STR from comment 13 and have NOT been able to reproduce the bug. Thus, it looks fixed to me.
Merged:
http://hg.mozilla.org/mozilla-central/rev/0e60623e3607
Assignee: nobody → bgirard
Status: NEW → RESOLVED
Closed: 8 years ago
Flags: in-testsuite?
Resolution: --- → FIXED
Whiteboard: [inbound]
Target Milestone: --- → mozilla8
So technically we still see this crash in recent builds but the volume is way down. I don't see a reason to track this.

+ 9.0a1 - 7 crashes in the past 2 weeks.
+ 8.0a2 - 3 crashes in the past 2 weeks.
(In reply to Sheila Mooney from comment #22)
> So technically we still see this crash in recent builds but the volume is
> way down. I don't see a reason to track this.

Yes, the fix landed for 8, so we should be OK. The remaining ones are probably not this regression at least (or it's old builds).
You need to log in before you can comment on or make changes to this bug.