Closed Bug 519962 Opened 11 years ago Closed 9 years ago

startup crash [@ nsPrefBranch::Observe(nsISupports*, char const*, unsigned short const*)] or [@ _purecall | nsPrefBranch::freeObserverList() | nsPrefBranch::Observe(nsISupports*, char const*, unsigned short const*)] (version3.0only)

Categories

(Core :: Preferences: Backend, defect, critical)

x86
All
defect
Not set
critical

Tracking

()

RESOLVED FIXED
Tracking Status
blocking1.9.2 --- needed
status1.9.2 --- .13-fixed
blocking1.9.1 --- -
status1.9.1 --- .16-fixed

People

(Reporter: wsmwk, Assigned: Bienvenu)

References

()

Details

(Keywords: crash, topcrash+, Whiteboard: [tb30xwanted][tbird topcrash][tb31wanted][gs][fixed by bug 533355])

Crash Data

startup crash [@nsPrefBranch::Observe(nsISupports*, char const*, unsigned short const*)]
#10 crash for 3.0b4, but it's not clear whether it's just a few people or not.
not new. goes back at least to 3.0b1. but the number of crashes significantly increased with 3.0b3 and 3.0b4, so this deserves QA going into 3.0 

also exists in firefox, but extremely rare http://crash-stats.mozilla.com/query/query?product=Firefox&version=ALL%3AALL&date=&range_value=4&range_unit=weeks&query_search=signature&query_type=exact&query=nsPrefBranch%3A%3AObserve%28nsISupports*%2C+char+const*%2C+unsigned+short+const*%29&do_query=1

bp-03259a37-08f4-4b93-be04-968262090926
I was trying to go back to an earlier version because the latest security update will not let me send mail - I can receive it. Also I did not like the way the folders were changed around between the main account, the local account and what you call the smart account. I currently have no folders in Thunderbird and cannot find a way to retreive my information.
Frame	Module	Signature [Expand]	Source
0		@0x6769726f	
1	thunderbird.exe	nsPrefBranch::Observe	modules/libpref/src/nsPrefBranch.cpp:722
2	xpcom_core.dll	nsObserverList::NotifyObservers	xpcom/ds/nsObserverList.cpp:128
3	xpcom_core.dll	nsObserverService::NotifyObservers	xpcom/ds/nsObserverService.cpp:181
4	xpcom_core.dll	NS_ShutdownXPCOM_P	xpcom/build/nsXPComInit.cpp:764
5	thunderbird.exe	ScopedXPCOMStartup::~ScopedXPCOMStartup	toolkit/xre/nsAppRunner.cpp:956
6	thunderbird.exe	XRE_main	toolkit/xre/nsAppRunner.cpp:3115
7	thunderbird.exe	NS_internal_main	mail/app/nsMailApp.cpp:103
this is now #3 crash for 3.0 with 2.9% of crashes, though likely many of these are duplicates via a few users.
and socorro has morphed it to _purecall | nsPrefBranch::freeObserverList() | nsPrefBranch::Observe(nsISupports*, char const*, unsigned short const*)

bp-2184e456-d855-44c2-969b-2ed1c2091125
0	ntdll.dll	KiFastSystemCallRet	
1	ntdll.dll	ZwWaitForSingleObject	
2	kernel32.dll	WaitForSingleObjectEx	
3	kernel32.dll	WaitForSingleObject	
4	thunderbird.exe	google_breakpad::ExceptionHandler::WriteMinidumpOnHandlerThread	toolkit/crashreporter/google-breakpad/src/client/windows/handler/exception_handler.cc:562
5	thunderbird.exe	google_breakpad::ExceptionHandler::HandlePureVirtualCall	toolkit/crashreporter/google-breakpad/src/client/windows/handler/exception_handler.cc:506
6	mozcrt19.dll	_purecall	objdir-tb/mozilla/memory/jemalloc/src/purevirt.c:47
7	thunderbird.exe	nsPrefBranch::freeObserverList	modules/libpref/src/nsPrefBranch.cpp:781
8	thunderbird.exe	nsPrefBranch::Observe	modules/libpref/src/nsPrefBranch.cpp:722
9	xpcom_core.dll	nsObserverList::NotifyObservers	xpcom/ds/nsObserverList.cpp:128
10	xpcom_core.dll	nsObserverService::NotifyObservers	xpcom/ds/nsObserverService.cpp:181
11	xpcom_core.dll	mozilla::ShutdownXPCOM	xpcom/build/nsXPComInit.cpp:776
12	thunderbird.exe	ScopedXPCOMStartup::~ScopedXPCOMStartup	toolkit/xre/nsAppRunner.cpp:956
blocking-thunderbird3.0: --- → ?
Summary: startup crash [@nsPrefBranch::Observe(nsISupports*, char const*, unsigned short const*)] → startup crash [@ nsPrefBranch::Observe(nsISupports*, char const*, unsigned short const*)] or [@ _purecall | nsPrefBranch::freeObserverList() | nsPrefBranch::Observe(nsISupports*, char const*, unsigned short const*)]
Keywords: topcrash
bp-88e68da6-5619-4f42-ba55-4cb332091119 has a comment of "importing eudora 7 mailboxes and crashed when Avast showed a message of infection in a nstemp file"
Component: General → XPCOM
Product: Thunderbird → Core
QA Contact: general → xpcom
Chris, is there any magic analysis you can apply here?

#1 crash for thunderbird 3.0. 
An amazing number of comments are not in english, more than the normal ratio IMO
Whiteboard: [tb30xwants]
Depends on: 533355
Blocks: 535029
Serge, how will bug 533355 help?  Note, if it does, it will be hard to tell if 533355 helps unless there is a test...because crashes almost don't exist in nightlies

given the ratio of international crash comments, and the upsurge, any potential suspects that came in a big way in b3/b4 ... like tabs, gloda, smartfolders?
(In reply to comment #5)

> Serge, how will bug 533355 help?  Note, if it does, it will be hard to tell if

No idea: I just guessed it could be related and would be worth monitoring.

> 533355 helps unless there is a test...because crashes almost don't exist in
> nightlies

Sure, a test there/here would be helpful.
I would assume that if (top)crash disappears then that would be a good hint...

***

The reports seem to be on TB 3.0x (or FF 3.5.x) only.
Is this bug actually Trunk or m-1.9.1 only?
(In reply to comment #6)
> (In reply to comment #5)
> 
> > Serge, how will bug 533355 help?  Note, if it does, it will be hard to tell if
> 
> No idea: I just guessed it could be related and would be worth monitoring.
> 
> > 533355 helps unless there is a test...because crashes almost don't exist in
> > nightlies
> 
> Sure, a test there/here would be helpful.
> I would assume that if (top)crash disappears then that would be a good hint...

once released publicly in a point release, yes. 

 
> The reports seem to be on TB 3.0x (or FF 3.5.x) only.
> Is this bug actually Trunk or m-1.9.1 only?

for whatever reason, there are no reports of TB trunk crashes, nor in TB nightlies. In the absence of information to the contrary however, I'm inclined to assume this is due to conditions in the field, and not due to differences in code.
Sorry for the spam, I need to temporarily move this back to a TB product to clear a requested flag.
Component: XPCOM → Build Config
Product: Core → MailNews Core
blocking-thunderbird3.0: ? → ---
Component: Build Config → XPCOM
Product: MailNews Core → Core
Component: XPCOM → Preferences: Backend
QA Contact: xpcom → preferences-backend
what's needed to move this bug forward?  
a testcase, a dev, ...?
blocking1.9.1: --- → ?
more detail for drivers ...

because it is primarily a startup crash, it is a blocker for affected users to be able to even use version 3.  it is therefore probably a must fix before we can turn on auto-update from v2 to v3.

#2 crash for thunderbird 3.0.0 accounting for 5% of crashes.
On the support forums you ask for detail on how to reproduce to be posted here. I'm not sure I can help much but I am willing to send more info about my machine and software if you can guide me about what would be useful.

Steps to reproduce:
1. Click on look for updates from within Thunderbird 2 client
2. Click to download Thunderbird 3
3. Disk image opens automatically, follow instructions to drag TB icon over app folder icon (50+ MB files copied)
4. Click on application icon in dock - immediate crash
5. Click on application icon in apps folder - immediate crash
6. Download from Mozilla website British English version (repeats 1 - 5)
7. Download from Mozilla website US English version (repeats 1 - 5)
8. Drag application to trash and repeat 7.
PS one of my many crash reports would be:
bp-554f7b3b-5b4f-4752-bf71-200db2100107
Dolske/Dietrich/Gavin: can you guys figure out who's got some time to look at this? It blocks the TB3 major update, though not a Firefox release.
blocking1.9.1: ? → -
This problem began with an update to 3.5.7. Since then I cannot open Firefox. It instantly brings up the Mozilla Crash Reporter. I tried downloading twice with the same result.
#1 crash for Thunderbird 3.0.0 adding numbers for two other signatures:
* nsPrefBranch::freeObserverList() | nsPrefBranch::Observe(nsISupports*, char const*, unsigned short const*)  -- Bug 535029 Mac only
* @0x0 | nsPrefBranch::Observe(nsISupports*, char const*, unsigned short const*)


John, Bill, thanks for comments.  Bill, you were doing an update, and not "installing" add-ons?  Do you have any dictionary installed? (bill's crash is bp-1cedebf6-3e99-4843-a6dc-47c202100107)

devs...
* note: extension manager hasn't completed startup, all these crashes have "WARNING: Json file missing Add-ons" - ref bug 527012 comment 5.
* note comments 6 and 7 about no crashes for Trunk or 1.9.2 ... but perhaps not unexpected if this is specific to upgrades?
* getsatisfaction reports http://gsfn.us/t/nr1o
* the predominate crash sigs are:  
 - FF nsPrefBranch::freeObserverList() | nsPrefBranch::Observe(nsISupports*, char const*, unsigned short const*)
 - TB _purecall | nsPrefBranch::freeObserverList() | nsPrefBranch::Observe(nsISupports*, char const*, unsigned short const*)


p.s. does anyone "own" or regularly check the preferences component bugs?


Mac + Win XP _Thunderbird_ (very small number of crashes)
@0x0 | nsPrefBranch::Observe(nsISupports*, char const*, unsigned short const*)
bp-b5216fa8-996c-4b50-ad38-bdf2e2100107
1  	thunderbird-bin  	nsPrefBranch::Observe  	 modules/libpref/src/nsPrefBranch.cpp:722
2 	libxpcom_core.dylib 	nsObserverList::NotifyObservers 	xpcom/ds/nsObserverList.cpp:128
3 	libxpcom_core.dylib 	nsObserverService::NotifyObservers 	xpcom/ds/nsObserverService.cpp:181
Keywords: topcrashtopcrash+
Whiteboard: [tb30xwants] → [tb30xwants][#1 thunderbird crasher]
Wayne, correct. This was an update install prompted by a "New Version" notice when I opened my older version of FF. I downloaded the update, dragged it to "Applications" and tried to open. After a couple of attempts I downloaded the update a second time, trashed the previous update and dragged the new one to "Applications" with the same result. I've never had a dictionary installed in FF.

I have looked for a Mozilla or Firefox Preferences file thinking I could trash it and see if that helped but no luck finding one.

(In reply to comment #15)
> #1 crash for Thunderbird 3.0.0 adding numbers for two other signatures:
> * nsPrefBranch::freeObserverList() | nsPrefBranch::Observe(nsISupports*, char
> const*, unsigned short const*)  -- Bug 535029 Mac only
> * @0x0 | nsPrefBranch::Observe(nsISupports*, char const*, unsigned short
> const*)
> 
> 
> John, Bill, thanks for comments.  Bill, you were doing an update, and not
> "installing" add-ons?  Do you have any dictionary installed? (bill's crash is
> bp-1cedebf6-3e99-4843-a6dc-47c202100107)
> 
> devs...
> * note: extension manager hasn't completed startup, all these crashes have
> "WARNING: Json file missing Add-ons" - ref bug 527012 comment 5.
> * note comments 6 and 7 about no crashes for Trunk or 1.9.2 ... but perhaps not
> unexpected if this is specific to upgrades?
> * getsatisfaction reports http://gsfn.us/t/nr1o
> * the predominate crash sigs are:  
>  - FF nsPrefBranch::freeObserverList() | nsPrefBranch::Observe(nsISupports*,
> char const*, unsigned short const*)
>  - TB _purecall | nsPrefBranch::freeObserverList() |
> nsPrefBranch::Observe(nsISupports*, char const*, unsigned short const*)
> 
> 
> p.s. does anyone "own" or regularly check the preferences component bugs?
> 
> 
> Mac + Win XP _Thunderbird_ (very small number of crashes)
> @0x0 | nsPrefBranch::Observe(nsISupports*, char const*, unsigned short const*)
> bp-b5216fa8-996c-4b50-ad38-bdf2e2100107
> 1      thunderbird-bin      nsPrefBranch::Observe      
> modules/libpref/src/nsPrefBranch.cpp:722
> 2     libxpcom_core.dylib     nsObserverList::NotifyObservers    
> xpcom/ds/nsObserverList.cpp:128
> 3     libxpcom_core.dylib     nsObserverService::NotifyObservers    
> xpcom/ds/nsObserverService.cpp:181
Bill, from your steps it appears that you are downloading the dmg and installing it manually?
Robert,
Yes, per the instructions. Is there another option?
The builtin app update mechanism.
Well, I just downloaded the 3.6 beta, dragged it to Applications and it seems to work fine, so my problem is solved!
Thanks Bill. It appears this bug doesn't have anything to do with application update though it might have to do with running a newer version.
Dolske/Dietrich/Gavin - are these stacks of any use in resolving this, or do we need steps?


nsPrefBranch::Observe(nsISupports*, char const*, unsigned short const*) is still strong as #8 crash [1].  most common comment is crash after doing system restore:
https://crash-stats.mozilla.com/report/index/696f4a5b-8d74-424a-b9d6-87ff92100206
https://crash-stats.mozilla.com/report/index/06ee4635-7d7e-4609-884e-6b0062100211

Correlations

Below are tables of correlation data generated by dbaron's scripts. If this crash signature is a top crasher, then we should be able to load it from the origional text files. Only the first 2MB of each Platform's output is loaded.

    *
      Modules Next

      18% (13/74) vs.   2% (114/5385) IMJP12.IME
      18% (13/74) vs.   2% (114/5385) IMJP12K.DLL
      82% (61/74) vs.  74% (3994/5385) msctf.dll
      18% (13/74) vs.   9% (506/5385) msvcp80.dll
      8% (6/74) vs.   0% (6/5385) DeskHook.dll
      8% (6/74) vs.   0% (8/5385) CnsMin.dll
      9% (7/74) vs.   2% (96/5385) IMJP9.IME
      9% (7/74) vs.   2% (96/5385) IMJP9K.DLL
      42% (31/74) vs.  34% (1845/5385) powrprof.dll
      42% (31/74) vs.  34% (1846/5385) nsi.dll
      42% (31/74) vs.  34% (1846/5385) propsys.dll
      8% (6/74) vs.   1% (43/5385) jwdsrch.dll
      5% (4/74) vs.   0% (5/5385) prio.dll

[1] https://crash-stats.mozilla.com/report/list?version=Thunderbird%3A3.0.1&build_id=&query_search=signature&query_type=exact&query=&date=&range_value=2&range_unit=weeks&process_type=all&plugin_field=&plugin_query_type=&plugin_query=&do_query=&signature=nsPrefBranch%3A%3AObserve%28nsISupports*%2C%20char%20const*%2C%20unsigned%20short%20const*%29&missing_sig=&page=1


one stack, all ms-windows, is gone as of 3.0.1  _purecall | nsPrefBranch::freeObserverList() | nsPrefBranch::Observe(nsISupports*, char const*, unsigned short const*)
it now exists only in 3.0.0 per https://crash-stats.mozilla.com/report/list?product=Thunderbird&build_id=&query_search=signature&query_type=exact&query=_purecall%20|%20nsPrefBranch%3A%3AfreeObserverList%28%29%20|%20nsPrefBranch%3A%3AObserve%28nsISupports*%2C%20char%20const*%2C%20unsigned%20short%20const*%29&date=&range_value=4&range_unit=weeks&process_type=all&plugin_field=&plugin_query_type=&plugin_query=&do_query=1&signature=_purecall%20|%20nsPrefBranch%3A%3AfreeObserverList%28%29%20|%20nsPrefBranch%3A%3AObserve%28nsISupports*%2C%20char%20const*%2C%20unsigned%20short%20const*%29&missing_sig=&page=1

unsure if it morphed or was fixed by one of these 3.0.1-fixed (if any) https://bugzilla.mozilla.org/buglist.cgi?status_whiteboard_type=notregexp;query_format=advanced;field0-0-0=cf_status_thunderbird30;status_whiteboard=fixed%20RC.*%20build;type0-0-0=equals;value0-0-0=.1-fixed
and it looks like it's crashing too early to list extensions
One set of steps to reproduce: run reftests a whole bunch.

http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1269230585.1269231030.29887.gz
WINNT 5.2 mozilla-central opt test reftest on 2010/03/21 21:03:05
s: win32-slave16
At this stage we probably want to care more about getting this into 1.9.2 than 1.9.1. And, depending now how well the crashes of this bug match to philor's comment 24, we may still need someone to reproduce.  do the failed reftests mean bug 533355 doesn't/won't help?
I suspect that generally this crash forces people to leave thunderbird, or stick with version 2.  Some firefox and thunderbird crashes state they can't get it started.
#15 rank for thunderbird 3.0.4. Examples in TB 3.1, so recheck post release.

bp-80e4bf57-cb18-4841-b83a-d10662100529  3.1
Thunderbird Build ID	20100526103039
Branch	1.9.2
OS	Windows NT
OS Version	5.1.2600 Szervizcsomag 3

Last firefox trunk crash seen is build 20100407114617. Almost non-existent for FF 3.6
Whiteboard: [tb30xwants][#1 thunderbird crasher] → [tb30xwants][#15 thunderbird crasher]
#2 crash for version 3.1

many of these crashes happen shortly after installs - unclear if it's because the user is seeking relief from a crash in a prior version, or if it's a one-off crash related to just the new version, like an extension conflict - although the later shouldn't happen going from 3.0.x to 3.0.5 - examples:
* bp-8e90211a-4bd6-46c0-9b59-2c9292100623 3.0.5
* bp-b53cbde0-60f0-4b99-a7b1-e24092100624 3.0.5 Mac
* bp-1fe3349b-0561-4c14-96db-b52a52100622 3.0.2 weird, why install 3.0.2?
* bp-c67627b0-9d77-469a-83d5-61bc22100622 3.0.3 perhaps this is the same person?)

bp-d58b6f69-3355-4e7b-aae6-6d3c92100627 3.1
bp-bc6612f4-0f72-4ec1-86a9-dedf62100626 3.1 (anju)
Whiteboard: [tb30xwants][#15 thunderbird crasher] → [tb30xwants][#15 thunderbird crasher][tb31wants]
I had this bug and could not install Thunderbird 3.0 on my office MacBook Pro. It installed fine on my home iMac. I just tried deleting all files with 'Thunderbird' in the name, restarting my MacBook and doing a fresh installation of 3.1. It fails with the same symptoms.

e.g. crash report bp-2e41fabc-2b78-40df-95d8-270822100702
raising visbility with "blocking1.9.2?"

I'm back to emphasize this is needed for Thunderbird because it is #9 crash for v3.1. Comments are roughly equal mix of :
* can't start thunderbird at all
* can't send mail

Not affecting 90% of our users, but still, this sucks pretty bad for those that encounter it.

bp-acaaa6b5-5976-4035-b020-551c22100628 I upgraded and haven't been able to get my email since!
bp-175190c2-cb31-4968-9ea3-9141d2100704 Newly downloaded program won't launch.
bp-1917148e-f5c7-4da2-9808-19c832100704 cannogt send
bp-0f437737-832b-438e-a752-3b3f52100628 can't send (meteagle)

90% of firefox crashes are Mac, but rare - about 3/week - like
bp-7a020304-300b-43a9-aee1-73da42100626
0		@0x4d656e69	
1	XUL	nsPrefBranch::Observe
2	XUL	nsObserverList::NotifyObservers
3	XUL	nsObserverService::NotifyObservers
4	XUL	mozilla::ShutdownXPCOM	xpcom/build/nsXPComInit.cpp:851 

and bp-0188fcef-6138-41a4-9713-2b76f2100621
blocking1.9.2: --- → ?
Whiteboard: [tb30xwants][#15 thunderbird crasher][tb31wants] → [tb30xwants][#9 thunderbird crasher][tb31wants]
This really looks like more of an application-level mis-use, we can't really "block" Gecko updates on it without more to go on. Marking it blocking unfortunately doesn't magically get things fixed.
Assignee: nobody → bienvenu
blocking1.9.2: ? → needed
Dan, the patch in bug 533355 is alleged to fix this, I believe, and has already landed on trunk.
(In reply to comment #31)
> Dan, the patch in bug 533355 is alleged to fix this, I believe, and has already
> landed on trunk.

hopefully :)  thanks.  But does the fact that FF trunk crashes still happen go against this hope?  
example: bp-70f817ac-d999-4d79-b757-55ea02100509 build 20100315090913
query: https://crash-stats.mozilla.com/report/list?product=Firefox&branch=1.9.2&branch=1.9.3&build_id=&query_search=signature&query_type=exact&query=nsPrefBranch%3A%3AObserve%28nsISupports*%2C%20char%20const*%2C%20unsigned%20short%20const*%29&date=05%2F16%2F2010%2012%3A02%3A45&range_value=4&range_unit=weeks&hang_type=any&process_type=any&plugin_field=&plugin_query_type=&plugin_query=&do_query=1&signature=nsPrefBranch%3A%3AObserve%28nsISupports*%2C%20char%20const*%2C%20unsigned%20short%20const*%29&missing_sig=&page=1

The Thunderbird crash reporters with whom I've exchanged mail are a mixed bag of issues, but at least 5 clear clusters of reports have emerged:
* Mac crash reporters who resolved their issue by simply reinstalling
* some claim to have been (master?) password related
* can send but not receive mail (receive crashes)
* doesn't crash in safe mode
* no relief found - forced back to Thunderbird v2
reporter of bp-c308a835-4217-40cd-90ba-7be0b2100625 (thomas) writes 

crash [continuously] right after updating thunderbird from V3.0.x to V3.0.4 as administrator. Vista reported DEP-Error. started Thunderbird once in safe mode (as administrator). Now all is o.k. with the new version, also in my user account.
~25 crashes contacted. ~7 responses. 2 people went to Windows Live Mail
--
bp-bc6612f4-0f72-4ec1-86a9-dedf62100626 (CoolAvenues) Crash when upgraded to 3.1. degraded to 2.x branch, it worked.
bp-acaaa6b5-5976-4035-b020-551c22100628 (tom) also - 3.1 won't work at all.
--
bp-02779e9d-6fd7-432b-bca8-ae1982100715  (wendy) Something in the "inbox" that didn't show up on the screen that was huge that was preventing incoming mail. We deleted everything in that folder and now it works fine.
Whiteboard: [tb30xwants][#9 thunderbird crasher][tb31wants] → [tb30xwants][#9 thunderbird crasher][tb31wants][gs]
I just removed the crashing TB 3.1 from my OSX PowerBook as per instructions at http://www.mozillamessaging.com/en-US/thunderbird/3.0.1/releasenotes/

I dowloaded and installed 3.1.1 from the website and all seems to be well. TB now starts normally and I was able to set up my account which appears to be working.

The only thing of note to do with this machine that I can think of (TB is working fine on my home machine) is a recent Disk Permissions Repair operation carried out 5 days ago.
(In reply to comment #35)
> I just removed the crashing TB 3.1 from my OSX PowerBook as per instructions at
> http://www.mozillamessaging.com/en-US/thunderbird/3.0.1/releasenotes/

John, what section and item in release notes are you referring to?


> I dowloaded and installed 3.1.1 from the website and all seems to be well. TB
> now starts normally and I was able to set up my account which appears to be
> working.
> 
> The only thing of note to do with this machine that I can think of (TB is
> working fine on my home machine) is a recent Disk Permissions Repair operation
> carried out 5 days ago.

Do you recall what Disk Repair reported as being wrong?
It was the section on Uninstalling Thunderbird 3. I removed the Thunderbird folder at ~/Library/Thunderbird (Mac OSX instructions) in addition to moving the Application to Trash. This was OK for me because this is was a new install of Thunderbird.

I can't recall all of the permissions errors reported (there were a lot). I can recall that most related to Airport wireless, which I can't believe is connected to this issue.
new signature for this in version 3.1.4 - #10 crash - morkStore::IsOpenMdbObject(nsIMdbEnv*, unsigned char*). perhaps one of the signatures in the bug summary has morphed.

fix to bug 533355 is still needed on 1.9.2 branch
bienvenu, we still need to move this forward, correct? ...
(In reply to comment #31)
> Dan, the patch in bug 533355 is alleged to fix this, I believe, and has already
> landed on trunk.


mork follow up to comment 38

morkCursor::IsOpenMdbObject(nsIMdbEnv*, unsigned char*) is #7 crash for v3.1.4
morkStore::IsOpenMdbObject(nsIMdbEnv*, unsigned char*) is #2 crash for v3.1.3(now defunct)
neither crash sig appears in each other's version.

> new signature for this in version 3.1.4 - #10 crash -
> morkStore::IsOpenMdbObject(nsIMdbEnv*, unsigned char*). perhaps one of the
> signatures in the bug summary has morphed.

bp-9a140b47-70c6-475b-aa65-220bf2101009 v3.1.4
0	thunderbird.exe	morkCursor::IsOpenMdbObject	db/mork/src/morkCursor.cpp:163
1	thunderbird.exe	nsPrefBranch::freeObserverList	modules/libpref/src/nsPrefBranch.cpp:781
2	thunderbird.exe	nsPrefBranch::Observe	modules/libpref/src/nsPrefBranch.cpp:722
3	xpcom_core.dll	nsObserverList::NotifyObservers	xpcom/ds/nsObserverList.cpp:130
4	xpcom_core.dll	nsObserverService::NotifyObservers	xpcom/ds/nsObserverService.cpp:182
5	xpcom_core.dll	mozilla::ShutdownXPCOM	xpcom/build/nsXPComInit.cpp:851
6	thunderbird.exe	ScopedXPCOMStartup::~ScopedXPCOMStartup	toolkit/xre/nsAppRunner.cpp:1052
7	thunderbird.exe	XRE_main	toolkit/xre/nsAppRunner.cpp:3274
8	thunderbird.exe	NS_internal_main	mail/app/nsMailApp.cpp:101
9	thunderbird.exe	wmain	toolkit/xre/nsWindowsWMain.cpp:120
10	thunderbird.exe	__tmainCRTStartup	objdir-tb/mozilla/memory/jemalloc/crtsrc/crtexe.c:591
11	kernel32.dll	BaseProcessStart
Whiteboard: [tb30xwants][#9 thunderbird crasher][tb31wants][gs] → [tb30xwants][tbird topcrash][tb31wants][gs]
Mork isn't a pref observer, so that part of the stack seems a bit bogus.
(In reply to comment #39)
> morkCursor::IsOpenMdbObject(nsIMdbEnv*, unsigned char*) is #7 crash for v3.1.4
> morkStore::IsOpenMdbObject(nsIMdbEnv*, unsigned char*) is #2 crash for
> v3.1.3(now defunct)
> neither crash sig appears in each other's version.

morkCursor::IsOpenMdbObject is now totally gone from v3.1.5 crashes.
And (no shock) nsPrefBranch::Observe(.... has returned as a topcrash, top 10.

I examined the last 8 months of both of these signatures, and there are zero occurrences on trunk. (again, no shock)
Blocks: 597951
Fixed in Thunderbird 3.1.8/3.1.9 - no crashes

yay neil and serge and all who helped!
Status: NEW → RESOLVED
Closed: 9 years ago
Resolution: --- → FIXED
Summary: startup crash [@ nsPrefBranch::Observe(nsISupports*, char const*, unsigned short const*)] or [@ _purecall | nsPrefBranch::freeObserverList() | nsPrefBranch::Observe(nsISupports*, char const*, unsigned short const*)] → startup crash [@ nsPrefBranch::Observe(nsISupports*, char const*, unsigned short const*)] or [@ _purecall | nsPrefBranch::freeObserverList() | nsPrefBranch::Observe(nsISupports*, char const*, unsigned short const*)] (version3.0only)
Whiteboard: [tb30xwants][tbird topcrash][tb31wants][gs] → [tb30xwanted][tbird topcrash][tb31wanted][gs][fixed by bug 533355]
Duplicate of this bug: 535029
Crash Signature: [@ nsPrefBranch::Observe(nsISupports*, char const*, unsigned short const*)] [@ _purecall | nsPrefBranch::freeObserverList() | nsPrefBranch::Observe(nsISupports*, char const*, unsigned short const*)]
Issue is resolved - clearing old keywords - qa-wanted clean-up
Crash Signature: [@ nsPrefBranch::Observe(nsISupports*, char const*, unsigned short const*)] [@ _purecall | nsPrefBranch::freeObserverList() | nsPrefBranch::Observe(nsISupports*, char const*, unsigned short const*)] → [@ nsPrefBranch::Observe(nsISupports*, char const*, unsigned short const*)] [@ _purecall | nsPrefBranch::freeObserverList() | nsPrefBranch::Observe(nsISupports*, char const*, unsigned short const*)]
Keywords: qawanted
No longer blocks: 535029
You need to log in before you can comment on or make changes to this bug.