Description of problem: soffice crashed when openning a .docx file, dnesg shows: [ 1380.745868] soffice.bin[2553]: segfault at 10 ip 00007f372688a50d sp 00007fff53f15b90 error 4 in libwriterfilterlo.so[7f37266fa000+38a000] Version-Release number of selected component (if applicable): libreoffice-core-3.6.6.2-5.fc18.x86_64 How reproducible: Always. Steps to Reproduce: 1. Go to this web page: http://msdn.microsoft.com/en-US/library/windows/hardware/jj159305 2. Click download for the document, then open it with libreoffice. 3. Actual results: libreoffice doesn't show up the document, it crashed. Expected results: Additional info:
windows-8-brightness-control-integrated-displays.docx right ? doesn't crash for me here though
nothing reported under valgrind either. Is that the right document ? and does it still crash for you ?
(In reply to comment #1) > windows-8-brightness-control-integrated-displays.docx right ? Right. > doesn't crash for me here though That's weird... it still crashes for me, anything I can do to collect some info? Thanks.
This is my installation of libreoffice $ rpm -qa |grep libreoffice libreoffice-impress-3.6.6.2-5.fc18.x86_64 libreoffice-langpack-zh-Hans-3.6.6.2-5.fc18.x86_64 libreoffice-opensymbol-fonts-3.6.6.2-5.fc18.noarch libreoffice-calc-3.6.6.2-5.fc18.x86_64 libreoffice-graphicfilter-3.6.6.2-5.fc18.x86_64 libreoffice-core-3.6.6.2-5.fc18.x86_64 libreoffice-ure-3.6.6.2-5.fc18.x86_64 libreoffice-draw-3.6.6.2-5.fc18.x86_64 libreoffice-pdfimport-3.6.6.2-5.fc18.x86_64 libreoffice-presenter-screen-3.6.6.2-5.fc18.x86_64 libreoffice-writer-3.6.6.2-5.fc18.x86_64 $
you could try... debuginfo-install libreoffice and then gdb /usr/lib64/libreoffice/soffice.bin (gdb) run /path/to/that.docx and get it to crash then... (gdb) bt and paste the backtrace in here. abrt is supposed to do this automatically but appears to be timing out or breaking in some fashion
Created attachment 743442 [details] backtrace generated by abrt I didn't have abrt installed, but hearing what you have said, I installed it. And now when libreoffice crashed, abrt pops out and downloaded debuginfo and generated backtrace. I copied the backtrace to the attached file.
I just tried to use abrt to file the bug, since it provides more information. But in the end, it was closed as a duplicate: --- Running report_Bugzilla --- Logging into Bugzilla at https://bugzilla.redhat.com Checking for duplicates Bug 914917 is CLOSED as DUPLICATE, but it has no DUP_ID (exited with 1) --- Running report_EmergencyAnalysis --- Compressing data Sending /tmp/abrt-upload-2013-05-04-14:13:07-3074.tar.gz to https://retrace.fedoraproject.org/faf/dumpdirs/new/ Successfully sent /tmp/abrt-upload-2013-05-04-14:13:07-3074.tar.gz to https://retrace.fedoraproject.org/faf/dumpdirs/new/ I hope you can find the uploaded abrt data. I have two fedora 18 that have the same problem, the libreoffice shipped in rhel6 does not have this problem. Thanks.
install libreoffice-math and the crash will disappear
*** This bug has been marked as a duplicate of bug 862467 ***
Thanks, the crash disappeared. Should the libreoffice-math be treated as a dependency of libreoffice-writer by yum?
Like I said, the libreoffice shipped in rhel6 does not have this problem, that is libreoffice 3.5.4, and I do not have to install -math there. Installing -math indeed make the crash disappear, but this doesn't feel like the problem is root caused. Is there a plan or is it possible for this problem to be actually fixed? Thanks.
(In reply to comment #10) > Thanks, the crash disappeared. > > Should the libreoffice-math be treated as a dependency of libreoffice-writer > by yum? No, it should not.
(In reply to comment #11) > Like I said, the libreoffice shipped in rhel6 does not have this problem, > that is libreoffice 3.5.4, and I do not have to install -math there. You mean 3.4.5 . That is old. It probably does not contain the math import code yet. > > Installing -math indeed make the crash disappear, but this doesn't feel like > the problem is root caused. Is there a plan or is it possible for this > problem to be actually fixed? Thanks. It was fixed months ago, just nobody proposed the fix to 3.6. It will be in 3.6.7.
(In reply to comment #13) > It was fixed months ago, just nobody proposed the fix to 3.6. It will be in > 3.6.7. Glad to hear this, thanks a lot!