Apache OpenOffice (AOO) Bugzilla – Issue 24978
1.1.1a crashes when saving this file after modification
Last modified: 2013-08-07 14:43:45 UTC
Hi, 1.1.1a as well as Czech builds from fix2 tree crash when you open this file, do a simple modification and save it back to Word format. Originally reported by Martin Kamin on Czech users mailing list, confirmed by me and also by CPH on IRC.
Created attachment 12827 [details] The victim document
Users on Czech mailing list confirmed that 1.1.0 final on Win XP and also on GNU/Linux is OK. 1.1.1a on Win XP crashes too, so this is: - reproducible on both Windows and GNU/Linux - it is regression from 1.1.0 One user also tried to save it with modification to sxw and reopen and save as DOC. It crashed too. Hope this helps.
reassigned to mru @mru: I tried this on Solaris and it happens there, too... An Error Report has been sent...
MRU->CMC: This should be fixed for OO 1.1.2 I think.
Backtrace: Thread 1 (process 22895 thread 0x60b): #0 0x07d9e8e4 in PlcDrawObj::WritePlc(SwWW8Writer&) const () #1 0x07db2974 in SwWW8Writer::WriteFkpPlcUsw() () #2 0x07db2ba4 in SwWW8Writer::StoreDoc1() () #3 0x07db370c in SwWW8Writer::StoreDoc() () #4 0x07db3ee4 in SwWW8Writer::WriteStorage() () #5 0x07d8b38c in StgWriter::Write(SwPaM&, SvStorage&, String const*) () #6 0x07cec454 in SwWriter::Write(WriterRef&, String const*) () #7 0x07e452f0 in SwDocShell::ConvertTo(SfxMedium&) () #8 0x05055da4 in SfxObjectShell::SaveTo_Impl(SfxMedium&, SfxItemSet const*, unsigned char) () #9 0x05059da4 in SfxObjectShell::DoSave_Impl(SfxItemSet const*) () #10 0x0505a148 in SfxObjectShell::Save_Impl(SfxItemSet const*) () #11 0x050671e8 in SfxObjectShell::ExecFile_Impl(SfxRequest&) () #12 0x050e0074 in SfxDispatcher::Call_Impl(SfxShell&, SfxSlot const&, SfxRequest&, unsigned char) () #13 0x050e26b8 in SfxDispatcher::PostMsgHandler(SfxRequest*) () #14 0x050fe2f4 in SfxHintPoster::Event(SfxHint*) () #15 0x050fe260 in SfxHintPoster::LinkStubDoEvent_Impl(void*, void*) () #16 0x005e3e44 in ImplHandleUserEvent(ImplSVEvent*) () #17 0x005e4608 in ImplWindowFrameProc(void*, SalFrame*, unsigned short, void const*) () #18 0x0062fc84 in SalFrameData::HandleClientMessage(XClientMessageEvent*) () #19 0x006306b8 in SalFrameData::Dispatch(_XEvent*) () #20 0x006539b0 in SalDisplay::Dispatch(_XEvent*) () #21 0x00653654 in SalDisplay::Yield(unsigned char) () #22 0x0064fa58 in DisplayYield(int, SalDisplay*) () #23 0x0064e9c4 in SalXLib::Yield(unsigned char) () #24 0x004dc984 in Application::Yield() () #25 0x004dc86c in Application::Execute() () #26 0x00008a50 in desktop::Desktop::Main() () #27 0x004e0a2c in SVMain() () #28 0x0064d764 in main () #29 0x0000232c in _start (argc=1, argv=0x0, envp=0x6c4000) at /SourceCache/Csu/Csu-46/crt.c: 267 #30 0x000021a0 in start () (gdb)
cmc->mru: No crash now with changes for kildarefilterteam22 for 1.1.2
Checked fix in CWS kildarefilterteam22.
Verified.
What is the exact change that make it work? I'd like to implement it to Czech 1.1.1 so we do not have this regression.
Not sure :-), one of the previous changes made along the cws_srx645_kildarefilterteam22 tag fixed it.
The change that fixed this particular issue is http://tmp.janik.cz/OpenOffice.org/24978/24978.diff In case we decide to put this into 1.1.1 as well, there is also complete diff of sw project between kildarefilterteam22 and its anchor tag.
set target to 1.1.1
Reopened.
MRU->MMAHER: this also has to be ntroduced into OO 1.1.1. Thanks!
I ran into this exact same stack dump on OOo1.1.1rc with a different document, http://www.ieee.org/organizations/pubs/transactions/TRANS-JOUR.DOC, while playing with issue 13857. Raising priority to P2 since it happens in not just an isolated Czech case, but also in an English document from the IEEE representative of IEEE journal articles.
Created attachment 13586 [details] Extracted patch
Dan, thanks for confirmation. I have just tested two OOo versions with your document (simple one letter change and save back to DOC): 1.1.1RC English -> crash 1.1.1RC Czech with 24978.diff applied. OK, no crash! Is it possible for you to rebuild OOo with this patch applied?
Sadly, as my RSI prevents me from typing much, unless it's Real Easy to set up, I can't do it. If you can provide a shell script to run that will download the source and patch and build everything unattended, I can handle it, though. BTW see http://kegel.com/crosstool for an existence proof for such a script for a similarly devilish large project :-)
Grumble. I might have been confused; it's possible my crash actually matches issue 25635, judging by the stack traceback.
Dan, yes :-)
Fixed in ooo111regression4
close issue.