Issue 1857 - Writer crashed during autosave
Summary: Writer crashed during autosave
Status: CLOSED FIXED
Alias: None
Product: Writer
Classification: Application
Component: code (show other issues)
Version: 638
Hardware: PC Windows 98
: P3 Trivial (vote)
Target Milestone: ---
Assignee: Mathias_Bauer
QA Contact: issues@sw
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2001-10-10 13:10 UTC by simonbr
Modified: 2003-09-08 16:56 UTC (History)
1 user (show)

See Also:
Issue Type: DEFECT
Latest Confirmation in: ---
Developer Difficulty: ---


Attachments

Note You need to log in before you can comment on or make changes to this issue.
Description simonbr 2001-10-10 13:10:02 UTC
I was editing a drawing in Writer; the autosave prompt popped up just when I 
typed ENTER, immediately dismissing it; about halfway through the autosave the 
program crashed.

SOFFICE caused an invalid page fault in
module SFX638MI.DLL at 015f:1d5f3dc7.
Registers:
EAX=00000000 CS=015f EIP=1d5f3dc7 EFLGS=00010246
EBX=1e550114 SS=0167 ESP=0086ec90 EBP=0086ecfc
ECX=00000000 DS=0167 ESI=024d8e5c FS=352f
EDX=0249a92c ES=0167 EDI=00714008 GS=0000
Bytes at CS:EIP:
8b 10 8b 4d 9c ff 52 14 89 45 0c 8b 45 9c 8b 48 
Stack dump:
00714008 00714008 00000000 00990e30 004e4198 004e4198 1c6202fd 00714008 
00714008 1d50e937 1e351646 006ff6a4 1e35168a 1e351610 1e42648d 1e351610
Comment 1 stefan.baltzer 2001-10-29 12:06:35 UTC
1.   Does this ALWAYS happen when autosave starts to work?
1.a. (If not): What does it take to make this occur (start from an empty document up to the type 
of doc where it happened) 
2.   Is it Text documents only? (Try a spreadsheet with the same path/name/ same objects inside)
3.   Was that file saved before?
4.a. (If so): Where was the file stored (complete path)?
4.b. (If not): Where is your  "default saving path" (complete path)?
5.   Did you enable autosave "with prompt" or without?

Please comment. Thank you.
Comment 2 simonbr 2001-10-30 08:54:49 UTC
> 1.   Does this ALWAYS happen when autosave starts to work?
No, it was just this once.

> 1.a. (If not): What does it take to make this occur (start from an 
> empty document up to the type of doc where it happened)
It was a small document (<1 page) in Writer. I was using the drawing 
tools. It might have to do with the fact that the autosave prompt 
received OK (I was just pressing the ENTER key) almost immediately it 
came up. There is nothing more specific I could put my finger on.

> 2.   Is it Text documents only? (Try a spreadsheet with the same 
> path/name/ same objects inside)
I doubt if I can even reproduce the error with a text document.

> 3.   Was that file saved before?
I'm not sure.

> 4.b. (If not): Where is your  "default saving path" (complete path)?
I think it is "C:\My Documents". I am not sure if it was at the time 
this error occurred.

> 5.   Did you enable autosave "with prompt" or without?
With prompt.

I don't have the impression this error is likely to occur, so it may 
be quite difficult to reproduce. This is the reason I included the 
stack trace in the hope you could use it to find out what happened.

I later switched autosaving off because OOo is supposed to have an 
autorecovery mechanism analogous to that in MS Office, which is what 
I actually want. I haven't been able to figure out what additional 
value "autosaving" could actually have?

Best regards, Simon.
Comment 3 stefan.baltzer 2001-10-30 16:09:52 UTC
Mathias, If you can't figure out how to reproduce this, I'm afraid we have to leave it as not 
reproducible and hope that things get better along with the improvement of stability in general.
Reassigned.
Comment 4 Mathias_Bauer 2001-11-09 10:32:44 UTC
Sorry, but the stacktrace doesn't give a valid information to me 
Comment 5 Martin Hollmichel 2001-11-09 10:59:13 UTC
mh->mba: I think, that this could be SfxShell::GetSlotState(...).
Comment 6 Mathias_Bauer 2001-11-09 11:25:46 UTC
Sorry, the last comment was incomplete because I accidently hit the
return key.
What I wanted to say: the stacktrace doesn't give a valid information
to me, because I examined the function where the crash occured in
close detail, and I didn't find any chance for a crash.
We made some changes there that made the code a little bit safer, and
this happened after the 638 build. There was one code line that could
end in a NULL pointer access under some rare conditions. May be you
had  one?! Anyway, this NULL pointer access will not happen anymore.
I couldn't reproduce this bug in any build (638, 641), so I can only
assume that my code change hopefully fixed the bug.

Comment 7 Mathias_Bauer 2002-09-11 08:44:20 UTC
Seems to be fixed, got no response from submitter.
Comment 8 simonbr 2002-09-11 16:47:26 UTC
As the error is probably hard to reproduce, it is difficult to 
determine if if it has actually been fixed. But I fully agree with 
closing the issue. Thanks!