636674 - Fennec startup crash [@ nsTextEditRules::nsTextEditRules ]

Status: RESOLVED INVALID

(Core :: DOM: Editor, defect)

Description

[Scoobidiver (away)]

It is a new crash signature that first appeared in Fennec 4.0b6pre/20110222.
It is currently #1 top crasher in Fennec 4.0b6pre.

Signature	nsTextEditRules::nsTextEditRules
UUID	c04d9892-498c-4319-863d-ff37d2110224
Time 	2011-02-24 11:37:10.52316
Uptime	1
Last Crash	33 seconds before submission
Install Age	105 seconds since version was first installed.
Product	Fennec
Version	4.0b6pre
Build ID	20110224042404
Branch	2.0
OS	Linux
OS Version	0.0.0 Linux 2.6.32.9-00000-10.8.2-dirty #16 SMP PREEMPT Mon Jan 3 15:24:07 EST 2011 armv7l
CPU	arm
Crash Reason	SIGBUS
Crash Address	0x0
App Notes 	malata UPC300-2.2
Flextronics/harmony/harmony/:2.2/FRF91/hudson-20110118-193312-TnT_SVN_3588:user/test-keys
Processor Notes 	WARNING: Json file missing Add-ons

Frame 	Module 	Signature [Expand] 	Source
0 	libxul.so 	nsTextEditRules::nsTextEditRules 	editor/libeditor/text/nsTextEditRules.cpp:109
1 	libxul.so 	nsEditor::EndPlaceHolderTransaction 	nsCOMPtr.h:644
2 		@0x1b 	
3 	libxul.so 	nsHtml5AttributeName::releaseStatics 	parser/html/nsHtml5AttributeName.cpp:2557
4 		@0x50890cc7 	
5 	libxul.so 	nsTextEditRules::DidDeleteSelection 	editor/libeditor/text/nsTextEditRules.cpp:949
6 		@0x52f4ffff 	
7 	libxul.so 	nsHtml5AttributeName::initializeStatics 	parser/html/nsHtml5AttributeName.cpp:1082
8 		@0x537345ef 	
9 	libxul.so 	mozilla::_ipdltest::PTestDataStructures::Msg_Test7::~Msg_Test7 	PTestDataStructures.h:3922
10 		@0xfffffffe 	
11 	libxul.so 	nsHtml5AttributeName::initializeStatics 	parser/html/nsHtml5AttributeName.cpp:149
12 		@0x0 	
13 	libxul.so 	nsHtml5AttributeName::initializeStatics 	parser/html/nsHtml5AttributeName.cpp:1139
14 		@0x0 	
15 	libxul.so 	nsHtml5AttributeName::initializeStatics 	parser/html/nsHtml5AttributeName.cpp:149
16 		@0x52f562ff 	
17 	libxul.so 	ComputeIsJITBroken 	basic_ios.h:49
18 		@0xffff0005 	
19 	libc.so 	libc.so@0x104ff 	
20 	libxul.so 	mozilla::plugins::PBrowserStreamChild::Read 	PBrowserStreamChild.cpp:539
21 	libxul.so 	nsPluginHost::WritePluginInfo 	modules/plugin/base/src/nsPluginHost.cpp:2791

The regression range is:
http://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=338a48d9f603&tochange=88204c53761b
The likely culprit is bug 635636.

More reports at:
https://crash-stats.mozilla.com/report/list?range_value=4&range_unit=weeks&signature=nsTextEditRules%3A%3AnsTextEditRules

Comment 1

[(no longer active)]

The stack here doesn't make any sense.  Does anybody know what's up with the stacks?

Comment 2

[(no longer active)]

And FWIW, I can't see how bug 635636 could be relevant here.  The code that it touches should not even run at Fennec startup...

Comment 3

[Scoobidiver (away)]

> The stack here doesn't make any sense.  Does anybody know what's up with the
> stacks?
See bug 634221.

> The code that it touches should not even run at Fennec startup...
May the uptime in crash reports is wrong and it is not a startup crash.

Comment 4

[(no longer active)]

(In reply to comment #3)
> > The code that it touches should not even run at Fennec startup...
> May the uptime in crash reports is wrong and it is not a startup crash.

That looks plausible.  What is the unit of time there?  I see that all of the values for these two bugs are either 0, 1 and 2...

Comment 5

[Scoobidiver (away)]

> What is the unit of time there?
second.

Comment 6

[Josh Matthews [:jdm]]

The uptime is correct, the stack is not.  This is one of many bugs I'm expecting to close as INVALID because of bug 637243.

Comment 7

[Scoobidiver (away)]

> This is one of many bugs I'm expecting to close as INVALID because of bug
> 637243.
This is not because stack traces are buggy that the crash signature is invalid.

Comment 8

[Josh Matthews [:jdm]]

From my point of view, any current crash stacks we have with no STR are completely invalid and not worth spending any time on, since they defy analysis and can't be reproduced to obtain better stacks.  Do you disagree?

Comment 9

[Scoobidiver (away)]

> Do you disagree?
There are different ways to analyze a crash: stack traces and STR are one of them but not the only ones. For a regression, you can look for likely bugs in the regression range.

For this bug, there have been no crashes from 4.0b6pre/20110225, so it can be marked as "workforme".

Comment 10

[(no longer active)]

See bug 634221 comment 4.