Bug 212740

Summary: 32bit openoffice crashes saving files on x86_64
Product: [Fedora] Fedora Reporter: Bruno Postle <bruno>
Component: openoffice.orgAssignee: Caolan McNamara <caolanm>
Status: CLOSED CURRENTRELEASE QA Contact:
Severity: medium Docs Contact:
Priority: medium    
Version: 5   
Target Milestone: ---   
Target Release: ---   
Hardware: x86_64   
OS: Linux   
Whiteboard:
Fixed In Version: FC6 Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2006-11-28 11:51:12 UTC Type: ---
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:
Attachments:
Description Flags
strace of crashing openoffice on x86_64 system none

Description Bruno Postle 2006-10-28 16:27:13 UTC
Description of problem: Open Office crashes when trying to save a document. 
Tested with draw and write documents.  Tried switching user accounts and
different file locations.

Version-Release number of selected component (if applicable): 2.0.2-5.19.2

How reproducible: Always

Steps to Reproduce:
1. Create new user account or remove ~/.openoffice.org2.0/
2. Start open office
3. File -> save or save as...
4. enter a filename, click save
  
Actual results:

"Illegal instruction" message and zero byte file created

Comment 1 Caolan McNamara 2006-10-31 10:37:07 UTC
Did you mix any fc-6 rpms with your fc-5 install ? fc-6 binaries are compiled
with a different linker system which could cause such a problem. Otherwise was
there any crash information produced by OOo on exit, i.e. a crash dialog.

Locally here there is no crash, nor has there be similiar sounding crashes, and
Ooo has not been updated for FC-5 in the last month or so

Comment 2 Bruno Postle 2006-10-31 12:52:52 UTC
(In reply to comment #1)
> Did you mix any fc-6 rpms with your fc-5 install ?

Nope, this is a stable system with no apparent other problems.

> Otherwise was there any crash information produced by OOo on exit,
> i.e. a crash dialog.

None, the application vanishes and the "Illegal instruction" message appears on
the console.

This could be something missing in the dependency tree: I had previously removed
all i386 RPMs from the system, openoffice was installed with yum install
'openoffice.org*'.

Comment 3 Caolan McNamara 2006-10-31 13:00:19 UTC
Indeed, i386 OOo on x86_64, perhaps a 

strace /usr/lib/openoffice.org/program/soffice.bin > /tmp/ooorun.log 2>&1 and
attach the log might show us what's missing

Comment 4 Bruno Postle 2006-10-31 21:23:27 UTC
Created attachment 139901 [details]
strace of crashing openoffice on x86_64 system

Comment 5 Bruno Postle 2006-10-31 21:28:17 UTC
(In reply to comment #3)
> 
> strace /usr/lib/openoffice.org/program/soffice.bin > /tmp/ooorun.log 2>&1

I've attached the strace.  The only obvious missing dependency to me was
libGL.so.1, but installing mesa-libGL.i386 doesn't fix anything.

Comment 6 Caolan McNamara 2006-11-01 08:52:16 UTC
do you have a 32bit gnome-vfs2 installed ?

Comment 7 Bruno Postle 2006-11-01 23:10:02 UTC
(In reply to comment #6)
> do you have a 32bit gnome-vfs2 installed ?

Yes, already installed.  I've checked the system for duplicate RPMs, missing
dependencies etc... and it all checks out ok.  If I had a list of the i386 RPMs
installed by default on a x86_64 system I could do a comparison and add the
missing packages one by one.

Comment 8 Caolan McNamara 2006-11-14 08:58:33 UTC
Did you ever figure out what rpm was missing ?

Comment 9 Bruno Postle 2006-11-14 21:00:37 UTC
(In reply to comment #8)
> Did you ever figure out what rpm was missing ?

Nope, after looking at how anaconda does it, I did a `yum groupinstall` for all
the groups that yum claimed were already installed.  This pulled in 105 RPMs,
lots of which looked like likely suspects.

..but I still get the 'Illegal instruction' error, so the problem may be elsewhere.

Comment 10 Caolan McNamara 2006-11-28 11:51:12 UTC
The FC6 OOo is native x86_64 so this class of problem doesn't arise under FC6,
so closed as fixed from that perspective, even though the underlying
dependencies for a 32bit app on 64bit may still exist, but available resources
limit investigation into that.