From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; U; Linux x86_64; rv:1.7.3) Gecko/20041020 Firefox/0.10.1 Description of problem: After installing the openoffice.org updates OO will no longer run. At first running ooffice would simply result in a segfault. [hcamp@localhost dhl]$ ooffice Segmentation fault And from the kernel logs: office.bin[9974]: segfault at 0000000000366dc0 rip 00000000004593ad rsp 00000000ffffd2f0 error 4 Then I tried nuking my .openoffice directory and the segfault gained some more information: [hcamp@localhost ~]$ ooffice sh: line 1: 10122 Segmentation fault LD_LIBRARY_PATH=/usr/lib/ooo-1.1:/usr/lib/ooo-1.1/program /usr/lib/ooo-1.1/program/setup -R:/tmp/ooo-wrapper.autoresponse.file.T10120 -v -nogui >~/.openoffice-install-log 2>&1 Installation of OpenOffice.org 1.1.2 failed: 'LD_LIBRARY_PATH=/usr/lib/ooo-1.1:/usr/lib/ooo-1.1/program /usr/lib/ooo-1.1/program/setup -R:/tmp/ooo-wrapper.autoresponse.file.T10120 -v -nogui > ~/.openoffice-install-log 2>&1' Version-Release number of selected component (if applicable): openoffice.org-1.1.2-11.5.fc3 How reproducible: Always Steps to Reproduce: 1. Execute openoffice.org any way you like 2. Watch it go boom. 3. Actual Results: Segmentation fault Expected Results: Memory hungry productivity app appears. Additional info:
->cmc, looks like an rpath/LD_LIBRARY_PATH thing?
caolanm->hcamp: Can you attach the ~/.openoffice-install-log
It's an empty file. What I can tell you, though, is that both soffice.bin and setup.bin segfault right after doing: getrlimit(RLIMIT_STACK, {rlim_cur=-4284481536, rlim_max=0}) = 0 I'm not sure how simply reading such a strange limit could cause a segfault, having not used getrlimit myself before and not knowing how OOo uses the data. It is interesting though that this translates to 0xffffffff00a00000, ie the upper 32 bits all 1s. Then again, OOo is compiled 32 bit, so perhaps the upper 32 bits are only an strace artifact. Anyway, this problem might have more to do with the glibc update that I installed at the same time. glibc-2.3.3-74
I can't think of anything significant which has changed, but OOo is now working.
I'm just checking with a pre-rpath change 1.1.2-10 first, can you do this for me... > export LD_LIBRARY_PATH=/usr/X11R6/lib:/usr/lib/ooo-1.1/program/ > gdb /usr/lib/ooo-1.1/program/soffice.bin (gdb) run I get... #0 0x0070b0ea in _nl_intern_locale_data () from /lib/tls/libc.so.6 #1 0x0070b2e0 in _nl_load_locale () from /lib/tls/libc.so.6 #2 0x0070ae8e in _nl_find_locale () from /lib/tls/libc.so.6 #3 0x0070a6ac in setlocale () from /lib/tls/libc.so.6 #4 0x00c1ef38 in osl_getTextEncodingFromLocale () from /usr/lib/ooo-1.1/program/libsal.so.3 #5 0x00c222a0 in osl_getProcessLocale () from /usr/lib/ooo-1.1/program/libsal.so.3 #6 0x00c1edf8 in osl_getTextEncodingFromLocale () from /usr/lib/ooo-1.1/program/libsal.so.3 #7 0x00c1fcb1 in osl_setThreadKeyData () from /usr/lib/ooo-1.1/program/libsal.so.3 #8 0x00c1f128 in osl_getSystemLocale () from /usr/lib/ooo-1.1/program/libsal.so.3 #9 0x00920c2b in pthread_once () from /lib/tls/libpthread.so.0 #10 0x00c1fcee in osl_getThreadTextEncoding () from /usr/lib/ooo-1.1/program/libsal.so.3 #11 0x00c28b7b in osl_syncFile () from /usr/lib/ooo-1.1/program/libsal.so.3 #12 0x00c1fde6 in osl_loadModule () from /usr/lib/ooo-1.1/program/libsal.so.3 #13 0xf7fc3a7d in vcl::SolarThreadExecutor::execute () from /usr/lib/ooo-1.1/program/libvcl645li.so #14 0xf7fc4812 in CreateSalInstance () from /usr/lib/ooo-1.1/program/libvcl645li.so
It's not crashing anymore, so there is not much to backtrace. If I break out to the gdb prompt after OOo comes up and do a backtrace it's only a couple lines. Program received signal SIGINT, Interrupt. [Switching to Thread 4160596192 (LWP 9354)] 0xffffe405 in ?? () (gdb) bt #0 0xffffe405 in ?? () #1 0xffffd438 in ?? () #2 0x0024b6a4 in poll () from /lib/tls/libc.so.6 Previous frame inner to this frame (corrupt stack?) Also, there don't seem to be any debuginfo packages available for OOo to provide debugging symbols. Do you have some more specific instructions?
So right now in an unaltered terminal > oowriter crashes, while > export LD_LIBRARY_PATH=/usr/X11R6/lib:/usr/lib/ooo-1.1/program/ > oowriter does not crash ?
Neither crash. Openoffice is working perfectly well now. I suppose this is a good thing, but it's quite disconcerting because I have no explanation for it. Could it have been something to do with prelink? Ie. prelink got a chance to run and straightned things out?
I've got the same error, but it has nothing to do with prelink. I've launch prelink, but nothing changes, and OpenOffice is still crashing... By the way, i got this in .openoffice-install-log: (process:3946): GLib-WARNING **: getpwuid_r(): failed due to: Permission denied. I also got a empty file called .openoffice-lock
It seems that the fault is not in openoffice. I've came back to openoffice.org-1.1.2-10, and this one is also crashing, with the same error as above. But if a start it as root, it is working perfectly...
caolanm->Jerome: Is this still broken for you or resolved with the various updates to FC3, e.g. ooo 1.1.3 et al.
Updates fixes the problem. I think you can close this bug.