User-Agent: Build Identifier: Open Office quits with the error: Main memory shortage. Please quit other applications or close some windows before continuing. [OK] An unrecoverable error has occurred. All modified files have been saved and can probably be recovered at next program restart.[OK] x 2 Reproducible: Always Steps to Reproduce: 1. Open xls spreadsheet using ooffice sheet.xls 2. Close it using File> Close 3. Open it again using File> 1 (recent items list) 4. Error Top shows: [snip] Mem: 515792k total, 513700k used, 2092k free, 64056k buffers Swap: 1048312k total, 344k used, 1047968k free, 227452k cached [snip] 19728 user 24 0 156m 67m 100m R 40.8 13.4 0:52.91 soffice.bin Stack attached.
Created attachment 105827 [details] stack
Reported upstream: http://www.openoffice.org/issues/show_bug.cgi?id=36280
Created attachment 105895 [details] *bugbuddy* backtrace Seems like oo.org relies on bug-buddy being installed. Installing it, gives the following error on crash: Application Error/usr/share/bug-buddy/gdb-cmd:2: Error in sourced command file: Previous frame inner to this frame (corrupt stack?) *bugbuddy* backtrace attached
OO.org cannot replicate this crash on 1.1.3 Either this is a RH build bug, or a 1.1.2 bug.
Doesn't crash on RH9 either. Installing the official build of OOo.org (latest -v1.1.3) on FC3t3 doesn't help. Instead of a crash, I get a hang. A ps ax shows: 14385 pts/1 Sl+ 0:05 /home/box/OpenOffice.org1.1.3/program/soffice.bin simple.xls 14403 pts/1 S+ 0:00 /home/box/OpenOffice.org1.1.3/program/getstyle-gnome --vcl-system-settings-window 50331649 # strace -p 14385 Process 14385 attached - interrupt to quit select(7, [6], NULL, NULL, NULL # strace -p 14403 Process 14403 attached - interrupt to quit ioctl(3, FIONREAD, [0]) = 0 poll(
Update: OOo 1.1.3 now works. 1.1.2 is the culprit.
Created attachment 106008 [details] patch to fix caolanm->dcbw: the source .xls does not have a SummaryInformation Stream, sfx2 goes looking for it, but the test to see if its not there fails and so fx2 thinks it has one and allocs mem based on junk contents. Its pure luck that it crashes the second time depending on the random contents of memory and not the first, and probably pure luck that it's not blowing up for everyone else. The attached patch fixes.
Excellent! :)
Caolan, this fix doesn't appear to be upstream. Can you shepherd that through? It should be able to go into the "obvious fix" CWS I would think. http://framework.openoffice.org/source/browse/framework/sfx2/source/doc/docinf.cxx?rev=1.28&content-type=text/vnd.viewcvs-markup
Hmm, even if 1.1.3 works, it probably covers up the issue rather than fixing it at the source.
OOo upstream is http://qa.openoffice.org/issues/show_bug.cgi?id=36280
should be in 1.1.2-11
*** Bug 135517 has been marked as a duplicate of this bug. ***
working in 1.1.2-11