Summary: | Openoffice-bin segmentation fault on KDE-3.5.2 compiled with gcc-4.1.0 | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | Krzysztof Raczkowski <raku> |
Component: | [OLD] KDE | Assignee: | Gentoo KDE team <kde> |
Status: | RESOLVED INVALID | ||
Severity: | major | CC: | gentoo3, kiciorek, patrizio.bassi |
Priority: | High | ||
Version: | 2006.0 | ||
Hardware: | x86 | ||
OS: | Linux | ||
Whiteboard: | |||
Package list: | Runtime testing required: | --- |
Description
Krzysztof Raczkowski
2006-04-06 11:34:17 UTC
I can also add that it is possible to run openoffice-bin under KDE like this: kicior ~ $ ssh -X localhost Password: Last login: Wed Apr 5 01:14:21 2006 from localhost kicior ~ $ oowriter2 Here you can read also logs from gdb and strace: 1. http://kicior.umcs.lublin.pl/~leszeks/oogdb.log 2. http://kicior.umcs.lublin.pl/~leszeks/oostrace.log i have the same problem. confirmed. i just was wondering why it worked some days ago, and now no more. my backtrace: different point and function, same lib no debugging symbols found) [Thread debugging using libthread_db enabled] [New Thread -1234855744 (LWP 11283)] Program received signal SIGSEGV, Segmentation fault. [Switching to Thread -1234855744 (LWP 11283)] 0xb57996ba in QMapPrivate<QPopupMenu const*, (anonymous namespace)::ShadowElements>::copy () from /usr/kde/3.5/lib/libkdefx.so.4 (gdb) Workaround (use Gnome instead of KDE decorations): Start OOo with # OOO_FORCE_DESKTOP=gnome ooffice2 here it seems starting, but it loops on "document restore" or similar procedure (In reply to comment #4) > Workaround (use Gnome instead of KDE decorations): > > Start OOo with > > # OOO_FORCE_DESKTOP=gnome ooffice2 > So does this mean the bug is in kdelibs? I ask because I've had enormous problems with KDE since moving to gcc-4.1.x / gcc-2.4* and now I'm also getting this same problem. I couldn't even get the KDE packages to run without changing -Os to -O2. i use -O3 and it's the only kde problem i have actually. I overcame this by removing -ffast-math and -mfpmath=sse from my C(XX)FLAGS, and rebuilding Qt and kdelibs (and the rest of KDE for good measure). From the above, looks like perhaps -ffast-math was the bogey. qt should filter that, so it's useless. i'm now rebuilding my kdelibs, however i have really no troubles with my kde system, so i'm hopeless. i'll report today when the compilation ends. you're right, recompiling kdelibs without --fast-math fixed this issue. i suggest closing this bug and marking as invalid so. Usage of -ffast-math globally is unsafe and not supported. Please don't file bugs unless you can reproduce a problem on a system without it. |