On Tuesday, August 07, 2012 11:41:28 PM Michael Schwendt wrote: > On Tue, 7 Aug 2012 22:29:18 +0200, Vladimir Nadvornik wrote: > > Dne úterý 07 Srpen 2012 17:16:59 Michael Schwendt napsal(a): > > > On Tue, 07 Aug 2012 17:05:58 +0200, Vladimir Nadvornik wrote: > > > > > removing unknown sidecar > > > > > /home/misc/tmp/stresstest/000454.JPG: �Jd8 fd magick > > > > > mismatch at filedata.c:629 > > > > > ** > > > > > ERROR:filedata.c:448:file_data_ref_debug: assertion failed: > > > > > (fd->magick == 0x12345678) Aborted > > > > > > > > Hmm, this could be corrupted anywhere. > > > > > > My brute-force change from last year is this: > > > > > > > > > http://pkgs.fedoraproject.org/cgit/geeqie.git/plain/geeqie-1.0-vflis > > > t-refr esh.patch > > > > IMHO the only possibility how this could fix it is that this is a bug in > > gtk. On the other hand it would explain that it works for me - > > Better not say "it works for me". You can't be sure unless you've got some > stresstest script already. With just a simple manual test like copying a > few thousand images to the working dir, it may not be easy to reproduce. > It's not a crash that's easy to reproduce, but something racy or a > corner-case, and some users have been hit by it (also in the bug tracker). > And vflist_setup_iter_recursive is recursive, so making it a bit more > simple was an obvious attempt. > > > I probably have a different gtk version. > > $ rpm -q gtk2 glib2 > gtk2-2.24.11-1.fc17.x86_64 > glib2-2.32.4-1.fc17.x86_64 > > Geeqie 1.0 has crashed several distribution releases before already. Not > for me. Until I've adjusted my usage patterns after reading how bug > reporters use it. > > Oh, I might give any recent openSUSE release a try, too, if time permits. :) > > Eventually, I'll need to find out how Geeqie 1.1 compares with 1.0 patched, > so applying individual patches one by one may be necessary. It might be > a combination of more than one patches. Watch this spec %changelog entry: > > * Sun Oct 16 2011 Michael Schwendt - 1.0-13 > - Cherry-pick a few commits (from Vladimir Nadvornik, Klaus Ethgen > and Vladislav Naumov). With the modified filelist_sort_compare_filedata > method, Geeqie passes another stress test I've created in order > to track down rare file_data_unref crashes. > > * Tue Aug 09 2011 Michael Schwendt - 1.0-11 > - Patch another place where not to exif_free_fd NULL ptr (#728802). > > * Fri Apr 15 2011 Michael Schwendt - 1.0-10 > - Let's see how we do with a simpler vflist_setup_iter_recursive(). > > So, according to my changelog, the modified filelist_sort_compare_filedata > method has added further stability, too. > However, if 1.1 crashes again, either something's missing, or it's some > new bug, ... and I hope we'll figure it out. I'll send updates to this > topic, hopefully prior to this weekend.
OK, I have finally reproduced it. So far I can tell that it crashes also with view_file_icon so vflist_setup_iter_recursive does not seem to be the cause. I will continue debugging in the evening. Vladimir ------------------------------------------------------------------------------ Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ _______________________________________________ Geeqie-devel mailing list Geeqie-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/geeqie-devel