Closed Bug 141560 Opened 22 years ago Closed 22 years ago

Trunk crash [@ nsLocalFile::Append][@ nsPromiseFlatString::nsPromiseFlatString]

Categories

(Core :: Networking, defect, P1)

Product:
Core
Component:
Networking
Platform:
x86
Windows NT
Type:
defect

Tracking

()

Status:
VERIFIED INVALID
Milestone:
mozilla1.0

People

(Reporter: jay, Assigned: darin.moz)

Details

(Keywords: crash, qawanted, topcrash)

Crash Data

This crash first showed up on the MozillaTrunk on 4/27 builds.  At first I
thought it might be related to bug 137886, but Seth advised me to log a new bug
and assign it to Darin.  Here is the latest from Talkback:

Rank    StackSignature    Count  

4   nsLocalFile::Append   57 

 
 	Source File : nsLocalFileCommon.cpp line : 271
 
====================================================================================================
     Count   Offset    Real Signature
[ 22   nsLocalFile::Append 4e3724cc - nsLocalFile::Append ]
[ 3   nsLocalFile::Append d3d190bd - nsLocalFile::Append ]
[ 3   nsLocalFile::Append 7414ae98 - nsLocalFile::Append ]
[ 3   nsLocalFile::Append 3b1cd475 - nsLocalFile::Append ]
[ 3   nsLocalFile::Append 2b4da16a - nsLocalFile::Append ]
[ 2   nsLocalFile::Append 8c889f4f - nsLocalFile::Append ]
 
     Crash date range: 2002-04-27 to 2002-04-30
     Min/Max Seconds since last crash: 14 - 59828
     Min/Max Runtime: 91 - 88853
     Keyword List : mail(6),  
     Count   Platform List 
     21   Windows NT 5.0 build 2195
     6   Windows 98 4.10 build 67766446
     5   Windows 98 4.10 build 67766222
     3   Windows 98 4.90 build 73010104
     1   Windows NT 4.0 build 1381
 
     Count   Build Id List 
     13   2002042908
     12   2002042708
     9   2002042808
     2   2002042803
 
     No of Unique Users        15
 
 Stack trace(Frame) 

	 nsLocalFile::Append
[nsLocalFileCommon.cpp  line 271] 
	 CSpellChecker::CSpellChecker
[spellchk.cpp  line 369] 
	 nsSpellCheckGlue::ReplaceAllOccurrences
[nsSpellCheckGlue.cpp  line 1137] 
	 nsSpellCheckGlue::NextMisspelledWord
[nsSpellCheckGlue.cpp  line 272] 
	 nsQueryElementAt::nsQueryElementAt  
	 nsComponentManager::CreateInstance
[nsComponentManagerObsolete.cpp  line 115] 
	 nsEditorShell::InitSpellChecker
[nsEditorShell.cpp  line 3843] 
	 XPTC_InvokeByIndex
[xptcinvoke.cpp  line 106] 
	 XPCWrappedNative::CallMethod
[xpcwrappednative.cpp  line 1995] 
	 XPC_WN_CallMethod
[xpcwrappednativejsops.cpp  line 1267] 
	 js_Invoke
[jsinterp.c  line 790] 
	 js_Interpret
[jsinterp.c  line 2744] 
	 js_Invoke
[jsinterp.c  line 806] 
	 js_InternalInvoke
[jsinterp.c  line 881] 
	 JS_CallFunctionValue
[jsapi.c  line 3414] 
	 nsJSContext::CallEventHandler
[nsJSEnvironment.cpp  line 1019] 
	 nsJSEventListener::HandleEvent
[nsJSEventListener.cpp  line 184] 
	 nsEventListenerManager::HandleEventSubType
[nsEventListenerManager.cpp  line 1220] 
	 nsEventListenerManager::HandleEvent
[nsEventListenerManager.cpp  line 1895] 
	 GlobalWindowImpl::HandleDOMEvent
[nsGlobalWindow.cpp  line 755] 
	 DocumentViewerImpl::LoadComplete
[nsDocumentViewer.cpp  line 1538] 
	 nsDocShell::EndPageLoad
[nsDocShell.cpp  line 3958] 
	 nsWebShell::EndPageLoad
[nsWebShell.cpp  line 731] 
	 nsDocShell::OnStateChange
[nsDocShell.cpp  line 3875] 
	 nsDocLoaderImpl::FireOnStateChange
[nsDocLoader.cpp  line 1105] 
	 nsDocLoaderImpl::doStopDocumentLoad
[nsDocLoader.cpp  line 744] 
	 nsDocLoaderImpl::DocLoaderIsEmpty
[nsDocLoader.cpp  line 642] 
	 nsDocLoaderImpl::OnStopRequest
[nsDocLoader.cpp  line 573] 
	 nsLoadGroup::RemoveRequest
[nsLoadGroup.cpp  line 531] 
	 nsJARChannel::OnStopRequest
[nsJARChannel.cpp  line 612] 
	 nsOnStopRequestEvent::HandleEvent
[nsRequestObserverProxy.cpp  line 213] 
	 PL_HandleEvent
[plevent.c  line 597] 
	 PL_ProcessPendingEvents
[plevent.c  line 530] 
	 _md_EventReceiverProc
[plevent.c  line 1078]  
 
     (5798359)	Comments: I can't post to Newsgroups anymore. I have been using Mozilla
since .91 and I think whoever is doing builds now and the RC1 release has caused
stability problems in the product. Too bad  as .99 was pretty close to a product
but I am losing confidence in
     (5798359)	Comments:  Mozilla. I hate Microsoft products. Come on you guys don't
release a piece of crap!
     (5798167)	Comments: The Spell Checker seems to be crashing Mozilla in the 2002042908
nightly build.
     (5797529)	Comments: Just clicked on the spell check button to run spell check on an
e-mail.
     (5781190)	URL: mozilla.org
     (5781190)	Comments: trying to run .xpi spellchecker - crashes mozilla with the
following details:MOZILLA caused an invalid page fault inmodule XPCOM.DLL at
015f:60e4d4cf.Registers:EAX=73726570 CS=015f EIP=60e4d4cf
EFLGS=00010202EBX=60e7a0b3 SS=0167 ESP=0064c6d8
     (5781190)	Comments:  EBP=0064c76cECX=10006254 DS=0167 ESI=10006254 FS=1d27EDX=3c8ed489
ES=0167 EDI=00000000 GS=0000Bytes at CS:EIP:ff 50 14 85 c0 74 5e 83 3d fc 14 ea
60 00 75 55 Stack dump:80000000 10006264 00000000 60e7a0bf 00000000 60e476d4
60e7a0b3 80000000
     (5781190)	Comments:  00000000 60e7dad8 10006264 00000005 00000001 00775e10 60e7a25f
00000000 
     (5781097)	URL: mozilla.org
     (5781097)	Comments: trying to run .xpi spellchecker after re-installing.  had worked
previously - then crashed mozilla every time.  reinstalled from mozilla.org 
still crashes
     (5753440)	Comments: Launching Spellcheck in blank composer window
     (5751286)	Comments: Trying to activate the spellchecker from spellcheck.mozdev.org
     (5722419)	Comments: send email; then it crashes. (occurred also in last night's build)
     (5713849)	Comments: Done composing email; Pressed "Send" button
     (5704684)	Comments: "Send" mail can repeat always...
     (5704353)	Comments: clicked on "send" to send a mail... arrr!!!!
     (5698989)	Comments: try to send email
     (5681592)	Comments: I ran spellcheck.
     (5681504)	Comments: I'm not sure what I did.
 
====================================================================================================
     Count   Offset    Real Signature
[ 8   nsLocalFile::Append 76b7e511 - nsLocalFile::Append ]
[ 4   nsLocalFile::Append da3cd0bb - nsLocalFile::Append ]
[ 2   nsLocalFile::Append a44e7cd7 - nsLocalFile::Append ]
[ 2   nsLocalFile::Append 5cd24d00 - nsLocalFile::Append ]
 
     Crash date range: 2002-04-27 to 2002-04-30
     Min/Max Seconds since last crash: 27 - 3517
     Min/Max Runtime: 141 - 86037
     Keyword List :  
     Count   Platform List 
     8   Windows NT 5.1 build 2600
     4   Windows NT 5.0 build 2195
     2   Windows 98 4.10 build 67766446
     2   Windows 98 4.10 build 67766222
 
     Count   Build Id List 
     9   2002042708
     4   2002042808
     3   2002042908
 
     No of Unique Users         9
 
 Stack trace(Frame) 

	 nsLocalFile::Append
[nsLocalFileCommon.cpp  line 271] 
	 CSpellChecker::~CSpellChecker
[spellchk.cpp  line 381] 
	 nsSpellCheckGlue::MapLanguageToCharset
[nsSpellCheckGlue.cpp  line 1204] 
	 nsSpellCheckGlue::NextMisspelledWord
[nsSpellCheckGlue.cpp  line 299] 
	 nsSpellCheckGlue::nsSpellCheckGlue
[nsSpellCheckGlue.cpp  line 138] 
	 nsComponentManager::CreateInstance
[nsComponentManagerObsolete.cpp  line 115] 
	 nsEditorShell::InitSpellChecker
[nsEditorShell.cpp  line 3843] 
	 XPTC_InvokeByIndex
[xptcinvoke.cpp  line 106] 
	 XPCWrappedNative::CallMethod
[xpcwrappednative.cpp  line 1995] 
	 XPC_WN_CallMethod
[xpcwrappednativejsops.cpp  line 1267] 
	 js_Invoke
[jsinterp.c  line 790] 
	 js_Interpret
[jsinterp.c  line 2744] 
	 js_Invoke
[jsinterp.c  line 806] 
	 js_InternalInvoke
[jsinterp.c  line 881] 
	 JS_CallFunctionValue
[jsapi.c  line 3414] 
	 nsJSContext::CallEventHandler
[nsJSEnvironment.cpp  line 1019] 
	 nsJSEventListener::HandleEvent
[nsJSEventListener.cpp  line 184] 
	 nsEventListenerManager::HandleEventSubType
[nsEventListenerManager.cpp  line 1220] 
	 nsEventListenerManager::HandleEvent
[nsEventListenerManager.cpp  line 1895] 
	 GlobalWindowImpl::HandleDOMEvent
[nsGlobalWindow.cpp  line 755] 
	 DocumentViewerImpl::LoadComplete
[nsDocumentViewer.cpp  line 1538] 
	 nsDocShell::EndPageLoad
[nsDocShell.cpp  line 3958] 
	 nsWebShell::EndPageLoad
[nsWebShell.cpp  line 731] 
	 nsDocShell::OnStateChange
[nsDocShell.cpp  line 3875] 
	 nsDocLoaderImpl::FireOnStateChange
[nsDocLoader.cpp  line 1105] 
	 nsDocLoaderImpl::doStopDocumentLoad
[nsDocLoader.cpp  line 744] 
	 nsDocLoaderImpl::DocLoaderIsEmpty
[nsDocLoader.cpp  line 642] 
	 nsDocLoaderImpl::OnStopRequest
[nsDocLoader.cpp  line 573] 
	 nsLoadGroup::RemoveRequest
[nsLoadGroup.cpp  line 531] 
	 nsJARChannel::OnStopRequest
[nsJARChannel.cpp  line 612] 
	 nsOnStopRequestEvent::HandleEvent
[nsRequestObserverProxy.cpp  line 213] 
	 PL_HandleEvent
[plevent.c  line 597] 
	 PL_ProcessPendingEvents
[plevent.c  line 530] 
	 _md_EventReceiverProc
[plevent.c  line 1078] 
	 USER32.dll + 0x3c076 (0x77d7c076)  
	 USER32.dll + 0x3c076 (0x77d7c076)  
	 USER32.dll + 0x3c076 (0x77d7c076)  
	 USER32.dll + 0x3c076 (0x77d7c076)  
	 _except_handler3()  
	 kernel32.dll + 0x3bb86 (0x77e9bb86)   
 
     (5769839)	Comments: Try to send an e-mail and mozilla crashes.
     (5761867)	Comments: Spellchecker caused crash
     (5715635)	Comments: Trying to use spell-checker with night build of April 28; versions
of spell checker  that failed: "windows xpi for post 0.99 nightlies"  "another
windows xpi for post 0.99 nightlies".
     (5711017)	Comments: Trying out Spellcheck
     (5679622)	Comments: pressed send button in email
     (5679595)	Comments: Pressed 'send' button in messenger.
assigning to darin.
Assignee: new-network-bugs → darin
Adding crash keywords and qawanted to see if we can get this reproduced.  Please
change the component if needed (I wasn't sure which one this crash was in).
Keywords: crash, qawanted, topcrash
Status: NEW → ASSIGNED
Priority: -- → P1
Target Milestone: --- → mozilla1.0
the spellchecker is part of the commercial tree, and this stack traces are all
bogus.

are people using the spellchecker from an older commercial build with the latest
mozilla trunk?  is that what's going on here?
marking INVALID since i'm sure that this bug is explained by comment #3
Status: ASSIGNED → RESOLVED
Closed: 22 years ago
Resolution: --- → INVALID
It looks like most of the users are probably getting the spellchecker .xpi from
 http://spellchecker.mozdev.org/ and trying to run it with nightly builds.  If
we can't do anything about this, perhaps the folks at mozdev.org should be told
about this problem...because regardless of where they get the .xpi, they're
crashing with OUR product...which is still bad (unless we tell the user it's a
NO NO to do what they're doing).
asa: this is the nsIFile related spellchecker crash i mentioned...
jpatel: this is the commercial spellchecker, not the one from mozdev.org!!

see:
http://lxr.mcom.com/commercial/source/modules/spellchk/glue/nsSpellCheckGlue.cpp
Ok, my assumption was wrong.  I based it on this incident:

(5751286)
Comments: Trying to activate the spellchecker from spellcheck.mozdev.org

and assumed others were using the same xpi, but we know at least one person got
the xpi from mozdev.org...

Regardless, marking this verified invalid per Darin's comments.  If we ever
decide to do something about this crash, please reopen.
Status: RESOLVED → VERIFIED
E-mail from Tom to me:

> there appears to be some controversy in Bug 141560 over where the spellchecker 
> hase come from. (See comment #3 in particular) I see in the Talkback database 
> that you crashed twice with the same signature in that bug (I attached some info 
> from Talkback at the bottom to refresh you memory.)
> 
> Any chance you can reproduce it or add any relevant info into the bug? 

I think I can answer your questions. I upgrade to the latest nightly almost
every day (on Win2k). I also use David Einstein's Spellchecker XPI
(http://spellchecker.mozdev.org/) which was formed out of bug 56301.

Up until 2002042608 on Win2k, the "works with 20020326 on my machine"
Spellchecker XPI worked fine. Then, starting with 20020427xx, I noticed that
invoking Spellchecker would crash Mozilla
(http://mozdev.org/bugs/show_bug.cgi?id=1184). In that bug, David mentions that
"Mozilla have rearranged nsIFile for their edification and amusement. 
Theoretically the 1.0 freeze should have stopped this."

So, it was apparently a Spellchecker-problem, related to changes in the Mozilla
code. Pleasantly, though, David just released a new version ("works with
20020501 on my machine") of the Spellchecker XPI
(http://spellchecker.mozdev.org/installation.html) and I'm now happily using
Spellchecker with 2002050208 (on Win2k).
Thanks Alex, that explains a lot...and double verifies this invalid.
It doesn't answer darin's question why commercial build stack trace(spell
checker part) code in the mozilla nightly builds.  Are we shipping Spell Checker
xpi with mozilla nightly builds or it was an accident ?

namachi: The nightly builds don't have Spellchecker XPI in them. So, I have to
reinstall the XPI every time I upgrade builds (which is every day). To be clear,
it's these nightlies that I'm talking about:
http://ftp.mozilla.org/pub/mozilla/nightly/latest/

My guess is that lots of people happen to be doing the same thing that I'm doing
(that is, installing David's Spellchecker XPI into the nightlies).
If the library names for the commercial spellchecker and the mozdev spellchecker
are the same, would talkback show bogus symbols from the commercial one for a
crash that occurs in the mozdev one?  The stacks here look somewhat bogus.

There are similar crashes in talkback in the constructor of nsPromiseFlatString.
Summary: Trunk crash [@ nsLocalFile::Append] → Trunk crash [@ nsLocalFile::Append][@ nsPromiseFlatString::nsPromiseFlatString]
Talkback is showing bogus stack information in this case. I want re-create the
same problem in the Mozilla 1.0 Branch and compare the results.

  This problem may have caused because of gmake builds in the Trunk and nmake
builds in the Branch. gmake builds don't have path information which might have
caused the corruption also. 

  Alex: Can you re-produce the crash with Mozila 1.0 Branch and submit the
crash(using the old spellcheck.xpi).

Thanks,
namachi: Sure, I can do that. (though, just for the record, it's not rocket
science ;). Anyone could go about it just as well as I could.)

Anyhow, I'll generate a crash and post the Talkback ID(s) here.
namachi: Using w32spellchecker20020326.xpi and the following Mozilla build, I
couldn't get a crash:
Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.0rc2) Gecko/20020510

Is there another build you'd like me to try?
Alex: Thanks a lot for all your help. If we the same type of crash in  the
future we go ahead and open a new bug. For now, the change in the api is the
issue and that has been fixed now by re-creating the xpi.
Crash Signature: [@ nsLocalFile::Append] [@ nsPromiseFlatString::nsPromiseFlatString]
You need to log in before you can comment on or make changes to this bug.