GNOME Bugzilla – Bug 330157
Reproducible Crash on closing Evo
Last modified: 2013-09-10 13:57:29 UTC
On closing Evolution, it reproducibly always crashes for me. Stacktrace generated by bug-buddy. Started Evolution, wait a few seconds till automatic check for new messages finished for my 3 IMAP accounts, closed Evo. Backtrace was generated from '/opt/gnome-2.12/bin/evolution-2.4' Using host libthread_db library "/lib/tls/libthread_db.so.1". [Thread debugging using libthread_db enabled] [New Thread -1229457728 (LWP 10086)] [New Thread -1303577680 (LWP 10099)] [Thread debugging using libthread_db enabled] [New Thread -1229457728 (LWP 10086)] [New Thread -1303577680 (LWP 10099)] [New Thread -1295184976 (LWP 10098)] [New Thread -1278215248 (LWP 10094)] [New Thread -1268470864 (LWP 10093)] [New Thread -1260078160 (LWP 10091)] [New Thread -1251292240 (LWP 10090)] [New Thread -1242899536 (LWP 10089)] [New Thread -1234506832 (LWP 10088)] [Thread debugging using libthread_db enabled] [New Thread -1229457728 (LWP 10086)] [New Thread -1303577680 (LWP 10099)] 0xffffe410 in ?? ()
+ Trace 65952
*** Bug 323138 has been marked as a duplicate of this bug. ***
This bug obsoletes my previously filed bug. Decided to create a fresh stacktrace to enable searching for the stacktrace at all.
Hmm, looks like a duplicate of bug 314912. Same stacktrace down to frame #25, e_shell_quit (). *** This bug has been marked as a duplicate of 314912 ***
No, this is *not* fixed. The fix 314912 and bug 320772 is suppsed to be in gnome-2-12 branch since 2005-11-07 (see bug 320772 comment 1). Evolution 2.4.2.1 was released on 2005-12-07, 2.4.2 on 2005-11-29. Either the patch did not go in into gnome-2-12 branch, or this is a different issue. REOPENing. Sorry for the noise. Cc'ing Vivek, since he seems to have worked on the other bug.
Actually, removing Vivek from the Cc list again... *sigh*
I tried reproducing this with the current build from HEAD (2.5.9) - and it seems to work fine. Not sure which patch actually fixed this - Hmm.... (me starts to check view cvs). This is going to take sometime. Have you seen this happen on the head build guenther??
Another stacktrace generated by bug-buddy, same procedure as in the original report. This time with debugging symbols. Backtrace was generated from '/opt/gnome-2.12/bin/evolution-2.4' Using host libthread_db library "/lib/tls/libthread_db.so.1". [Thread debugging using libthread_db enabled] [New Thread -1229457728 (LWP 30762)] [New Thread -1293976656 (LWP 30775)] [Thread debugging using libthread_db enabled] [New Thread -1229457728 (LWP 30762)] [New Thread -1293976656 (LWP 30775)] [Thread debugging using libthread_db enabled] [New Thread -1229457728 (LWP 30762)] [New Thread -1293976656 (LWP 30775)] [New Thread -1285399632 (LWP 30774)] [New Thread -1277006928 (LWP 30770)] [New Thread -1268470864 (LWP 30769)] [New Thread -1260078160 (LWP 30767)] [New Thread -1251292240 (LWP 30766)] [New Thread -1242899536 (LWP 30765)] [New Thread -1234506832 (LWP 30764)] 0xffffe410 in ?? ()
+ Trace 66085
Partha: I didn't really use HEAD builds yet. Anyway, there is another "crashes on close" bug out there I commented on, which might be related. See bug 272556 comment 29: > IIRC Evo did not crash for me on closing, after setting up a fresh one. At > some point later, it started to, again. I believe I have went though this > more than once already... This might be the same reason. I don't know. Fact is, I previously noticed a reprodicuble crash on exit, which was no issue at some point later. Most likely when starting fresh. Not sure. But it always started to crash later on... Thus I am not surprised that you can not reproduce this... *sigh* Need to do some more testing. Feel free to poke me on IRC. Meanwhile, have fun with the fresh stacktrace. :)
Another trace, same steps as previously. Evolution and e-d-s built with no custom CFLAGS for optimization this time, debugging symbols enabled. As vanilla as can be. Backtrace was generated from '/opt/gnome-2.12/bin/evolution-2.4' Using host libthread_db library "/lib/tls/libthread_db.so.1". [Thread debugging using libthread_db enabled] [New Thread -1229859136 (LWP 22829)] [New Thread -1287816272 (LWP 22894)] [Thread debugging using libthread_db enabled] [New Thread -1229859136 (LWP 22829)] [New Thread -1287816272 (LWP 22894)] [Thread debugging using libthread_db enabled] [New Thread -1229859136 (LWP 22829)] [New Thread -1287816272 (LWP 22894)] [New Thread -1279423568 (LWP 22893)] [New Thread -1270887504 (LWP 22839)] [New Thread -1262490704 (LWP 22837)] [New Thread -1254098000 (LWP 22836)] [New Thread -1243522128 (LWP 22835)] [New Thread -1235129424 (LWP 22834)] 0xffffe410 in ?? ()
+ Trace 66116
its a crash for sure - but am not able to reproduce this at all. Sigh!
See comment 8. I already told you, dude... What else do you need beside a plain vanilla stacktrace with debugging symbols? btw, the same with 2.5.91.
Additional note: Evo crashes only, if the Mailer was used, i.e. I opened the Mailer component. (All IMAP accounts are configured to periodically and automatically check for new mail.) If I start Evo, only use the Address Book, and close Evo after that, I do not get this crash. Still, it crashes always on close, when I switched to the Mailer component.
guenther, any news on this? still reproducible? guess so, code has not changed in the last six months... partha:
+ Trace 68897
retargetting.
Andre, didn't have a look at fresh stacktraces in a while, but the issue did not change at all. Perfectly reproducible crash. Please see all the details I mentioned above. The stacktrace is identical to others (which are closed as fixed), but this crash is *NOT* fixed.
*** Bug 346569 has been marked as a duplicate of this bug. ***
See duplicate bug 346569 -- *still* an issue with 2.6.x.
Sorry for tge dupped bug 346569. If you need any other trace or something to check this, please ask. I will be glad to help.
Set target milestone to 2.8
https://launchpad.net/distros/ubuntu/+source/evolution/+bug/57975 has a similar backtrace.
*** Bug 354791 has been marked as a duplicate of this bug. ***
*** Bug 356156 has been marked as a duplicate of this bug. ***
varadhan: stacktrace with uptodate line numbers available at bug 356156
*** Bug 356484 has been marked as a duplicate of this bug. ***
The Ubuntu bug submitter stated it's fixed for him with evolution 2.8.0
*** Bug 349521 has been marked as a duplicate of this bug. ***
up to date stacktrace with symbols available at bug 357064
*** Bug 357064 has been marked as a duplicate of this bug. ***
*** Bug 358330 has been marked as a duplicate of this bug. ***
bug 360099 provides an up-to-date stacktrace with line numbers. this is getting urgent as we get more and more duplicates.
*** Bug 360099 has been marked as a duplicate of this bug. ***
*** Bug 360252 has been marked as a duplicate of this bug. ***
*** Bug 360352 has been marked as a duplicate of this bug. ***
bug 360352 has a stack trace with line numbers that might be newer...
*** Bug 362597 has been marked as a duplicate of this bug. ***
related to bug 316071
*** Bug 363681 has been marked as a duplicate of this bug. ***
*** Bug 367641 has been marked as a duplicate of this bug. ***
https://launchpad.net/distros/ubuntu/+source/evolution/+bug/57975/comments/16 has another backtrace with evo symbols.
Created attachment 75922 [details] [review] Proposed patch
I have tried this patch and it works perfectly and Im not able to see the crash at all. I have reproduced this in the following scenario continuously. If your search folder structure looks like below ---------------- > FOOACCOUNT - INBOX - FOOBOX >Search Folders - Foo - Baa - Foo123 - Baa123 - Unmatched ----------------------- 1. Create Foo123 in a way that the sources for this vfolder are INBOX and Foo (search Folder). 2. Have the focus in Baa and close you can see this crash continuously. If you have the focus in Foo123 or Foo, this crash wont occur. The reason for the crash is that when a VFolder is a source/sub to a VFolder, the refcounts arenot handled right. When the source/sub folders of a VFOlder are unreffed, I decided to unref the source/sub only if it is not a vfolder. There could be other ways to fix this bug also, leaving this as a work around. I would let the maintainer, review and decide on it. Thx.
The patch targets the problem. But i'm not sure if just this would end up fixing the problem. The issue is that there are unrefs on an object which has its ref count set to 0. I'm not able to trace down the exact path. Fixed a few unref issues but still no where close to fixing the complete issue. But yes, the patch from srini surely is a hack.
*** Bug 369807 has been marked as a duplicate of this bug. ***
Partha, it fixes the bug. It doesnt have any side effect, not even a leak. I'll commit the patch for now as it is annoying to have a crash on shutdown 100% all the times.
Committed to HEAD and Stable. We can look at a better fix as we go on. Thanks.
*** Bug 371605 has been marked as a duplicate of this bug. ***
*** Bug 374568 has been marked as a duplicate of this bug. ***
*** Bug 272556 has been marked as a duplicate of this bug. ***
Running 2.9.3 in Ubuntu Feisty, I no longer see this issue. Thanks!
*** Bug 383749 has been marked as a duplicate of this bug. ***
*** Bug 380117 has been marked as a duplicate of this bug. ***
*** Bug 387222 has been marked as a duplicate of this bug. ***
*** Bug 387902 has been marked as a duplicate of this bug. ***
*** Bug 394626 has been marked as a duplicate of this bug. ***
*** Bug 395557 has been marked as a duplicate of this bug. ***
*** Bug 395492 has been marked as a duplicate of this bug. ***
*** Bug 399159 has been marked as a duplicate of this bug. ***
*** Bug 399766 has been marked as a duplicate of this bug. ***
*** Bug 403184 has been marked as a duplicate of this bug. ***
*** Bug 403630 has been marked as a duplicate of this bug. ***
*** Bug 407040 has been marked as a duplicate of this bug. ***
*** Bug 406088 has been marked as a duplicate of this bug. ***
*** Bug 417890 has been marked as a duplicate of this bug. ***
*** Bug 425483 has been marked as a duplicate of this bug. ***
*** Bug 423791 has been marked as a duplicate of this bug. ***
*** Bug 429669 has been marked as a duplicate of this bug. ***
*** Bug 444516 has been marked as a duplicate of this bug. ***
*** Bug 451396 has been marked as a duplicate of this bug. ***