Apache OpenOffice (AOO) Bugzilla – Issue 80952
Crash if searching "Hintergrundbild" in the help
Last modified: 2007-12-05 11:59:30 UTC
I tested OOG680_m2 from Pavel and from Sun with Pavel's langpack and also with the version, which sshould be the Key ID build. When I look for the word "Hintergrundbild" in the help and have no selection set it crashs. If I set one of the two selection it doesn't crash. Some other words also doesn't crash. It seems that it touch only some words. I get error report IDs rwxfbn, ry4fbn, ryyfbn, r8kfbn, r22fbn Mechtilde
If I look for "Hintergrundbilder" (Plural) then it also crash with the first selection "Nur ganz Wörter".
On which operating system does this crash occur? What do you mean by "look for the word"? You are at Help Viewer's Index tab page or at the Find tab page? What do you mean by "set one of the two selection"? Can you please have a look if this occurs on an English version, too?
I use GNU/Linux "Look for a word" similar to find tab. the two selections are 1. Complete words only 2. Find in Headings only I the English Version I can't look for "Hintergrundbild" or "Hintergrundbilder". It seems this is a problem of special words like some similar other Issues. Time later I will look for them
Please have a look. Cannot confirm the crash for an English version, may be a localisation issue
Hello I have testet on Windows XP with m225 only Enlish Version. For Search keyword, I have use "Background" I cannot confirm the error now I will trey with OOG m2 in english @mechtilde: Do you can me take in the CC? Raphael
@ufi wat's your Platform? If you test on Linux then I think it's a problem with the localisation. If you use Windows this bug is maybe only on linux platforms.
Add me on CC
Can verify defect on SO m226 build 9213 for Linux Enter "Hintergrundbild" as Find term in a German version crashes Enter "Watermark" as Find term in an English version does not crash. --> localization issue, possibly in the "Defining Background Colors or Background Graphics" help page
"Verify" was the wrong word. I meant I can "confirm" the crash. Perhaps Peter can help, set him to cc
pb: it is really strange. Perhaps the search engine is the problem. I have no other idea at the moment. So I set Andreas (ab) on cc.
For me it seems similar to the issue 71882 http://qa.openoffice.org/issues/show_bug.cgi?id=71882
Can someone please paste what is in these reports? > I get error report IDs rwxfbn, ry4fbn, ryyfbn, r8kfbn, r22fbn
pb: the stack with OpenOffice OOG680_m2 and german language pack on Windows: ucpchelp1.dll!xmlsearch::qe::QueryHitData::getDocument() Line 166 + 0x16 bytes C++ upchelp1.dll!chelp::ResultSetForQuery::ResultSetForQuery(const com::sun::star::uno::Reference<com::sun::star::lang::XMultiServiceFactory> & xMSF={...}, const com::sun::star::uno::Reference<com::sun::star::ucb::XContentProvider> & xProvider={...}, long nOpenMode=0, const com::sun::star::uno::Sequence<com::sun::star::beans::Property> & seq={...}, const com::sun::star::uno::Sequence<com::sun::star::ucb::NumberedSortingInfo> & seqSort={...}, chelp::URLParameter & aURLParameter={...}, chelp::Databases * pDatabases=0x056362f0) Line 178 + 0x12 bytes C++ ucpchelp1.dll!ResultSetForQueryFactory::createResultSet() Line 349 + 0x56 bytes C++ ucpchelp1.dll!chelp::DynamicResultSet::initStatic() Line 91 + 0x13 bytes C++ ucbhelper4MSC.dll!ucbhelper::ResultSetImplHelper::init(unsigned char bStatic='') Line 333 C++ ucbhelper4MSC.dll!ucbhelper::ResultSetImplHelper::getStaticResultSet() Line 210 C++ sfx680mi.dll!SfxContentHelper::GetResultSet(const String & rURL={...}) Line 553 + 0xa bytes C++ sfx680mi.dll!SearchTabPage_Impl::SearchHdl(PushButton * __formal=0x00000000) Line 1225 + 0xd bytes C++ sfx680mi.dll!SearchTabPage_Impl::LinkStubSearchHdl(void * pThis=0x0568f3e8, void * pCaller=0x00000000) Line 1208 + 0xe bytes C++ tl680mi.dll!Link::Call(void * pCaller=0x00000000) Line 158 + 0xe bytes C++ sfx680mi.dll!SearchBox_Impl::Select() Line 1083 C++ vcl680mi.dll!ComboBox::ImplSelectHdl(void * __formal=0x00000000) Line 502 + 0xd bytes C++ vcl680mi.dll!ComboBox::LinkStubImplSelectHdl(void * pThis=0x0568f5f4, void * pCaller=0x00000000) Line 419 + 0xe bytes C++ tl680mi.dll!Link::Call(void * pCaller=0x00000000) Line 158 + 0xe bytes C++ ... Andreas please take over. Thanks.
Seems to be comparable with #i66664 and probably also related to the help indexer. As long as we don't find a group of words I strongly doubt that this issue (or #i66664) is a release_blocker. As also stated in #i66664 I even doubt that it's P2. The priority guidelines say "P2 marks severe problems which affect a significant number of customers". Of course I don't know how many will search for "Hintergrundbild" but I strongly doubt the number is significant. STARTED anyway
Not a stopper, as discussed in release status meeting. => 2.4
Personally I suspect that this issue, and issue 66664 are all possibly due to 67740. It's definitely fairly futile fixing any of it while the double free exists as it makes OOo fairly random afterwards. Subsequent crashes after the double free may be through no fault of the finally crashing code.
Another "candidate" word for German is Speiche/Speichern (comes from http://www.openoffice.org/ issues/show_bug.cgi?id=77015). Two liner patch http://www.openoffice.org/nonav/issues/showattachment.cgi/38007/xmlhelp.free.patch from http://www.openoffice.org/issues/show_bug.cgi?id=67740 does fix it here. Please reconsider target of this issue.
@ab: please check with QA that patch in a cws. set target 2.3.
On Mac OS X with MallocStackLogging turned on: soffice.bin(13628,0xa000d000) malloc: *** error for object 0x19a439f0: incorrect checksum for freed object - object was probably modified after being freed, break at szone_error to debug when I search for these words.
Creating cws...
Patch applied, seems to work, cws will be Ready for QA tomorrow
FIXED
ab->kla: Please verify
Seen ok in CWS ab41 MultiPatchSets for Linux and Win -> verified
verified in OOG680_m3 -> closed
*** Issue 67740 has been marked as a duplicate of this issue. ***
FWIW, this'll hopefully make a lot of other "weird unexplainable crashes" just go away.
*** Issue 81122 has been marked as a duplicate of this issue. ***