GNOME Bugzilla – Bug 274240
LDAP address autocomplete crash at Evolution-DataServer-Addressbook-stubs.c:46
Last modified: 2008-06-18 22:17:54 UTC
Distribution: Fedora Core release 3 (Heidelberg) Package: Evolution Priority: Normal Version: GNOME2.8.0 unspecified Gnome-Distributor: Red Hat, Inc Synopsis: Crash on drag & drop of email Bugzilla-Product: Evolution Bugzilla-Component: Mailer Bugzilla-Version: unspecified BugBuddy-GnomeVersion: 2.0 (2.8.0) Description: Description of the crash: Selected a couple of messages and dragged to a different folder. Both folders are on a Cyrus IMAP server. Evo crashed. Steps to reproduce the crash: 1. Select one or more messages 2. Drag to a different folder 3. Expected Results: Messages moved How often does this happen? First time. Additional Information: KDE 3.4, Rh FC3 + updates. Debugging Information: Backtrace was generated from '/opt/evolution/libexec/evolution-data-server-1.2' Using host libthread_db library "/lib/tls/libthread_db.so.1". [Thread debugging using libthread_db enabled] [New Thread -1208113472 (LWP 21988)] [New Thread -1220916304 (LWP 22001)] [Thread debugging using libthread_db enabled] [New Thread -1208113472 (LWP 21988)] [New Thread -1220916304 (LWP 22001)] [Thread debugging using libthread_db enabled] [New Thread -1208113472 (LWP 21988)] [New Thread -1220916304 (LWP 22001)] [New Thread -1210426448 (LWP 21989)] 0x00af67a2 in _dl_sysinfo_int80 () from /lib/ld-linux.so.2
+ Trace 57526
Thread 3 (Thread -1210426448 (LWP 21989))
Setting qa contact to the default for this product. This bug either had no qa contact or an invalid one.
looks like a crash in ldap lookups
*** Bug 304354 has been marked as a duplicate of this bug. ***
*** Bug 306662 has been marked as a duplicate of this bug. ***
Here's another example backtrace, again from an FC3 machine attempting to auto-complete the "To" field of a forwarded message from an MS Exchange 5.5 LDAP server. Distribution: Fedora Core release 3 (Heidelberg) Package: Evolution-Data-Server Severity: normal Version: GNOME2.8.0 unspecified Gnome-Distributor: Red Hat, Inc Synopsis: EDS crash Bugzilla-Product: Evolution-Data-Server Bugzilla-Component: Contacts Bugzilla-Version: unspecified BugBuddy-GnomeVersion: 2.0 (2.8.0) Description: Debugging Information: Backtrace was generated from '/usr/libexec/evolution-data-server-1.0' Using host libthread_db library "/lib/tls/libthread_db.so.1". [Thread debugging using libthread_db enabled] [New Thread -1208146240 (LWP 20333)] [New Thread -1210434640 (LWP 20334)] [Thread debugging using libthread_db enabled] [New Thread -1208146240 (LWP 20333)] [New Thread -1210434640 (LWP 20334)] [Thread debugging using libthread_db enabled] [New Thread -1208146240 (LWP 20333)] [New Thread -1210434640 (LWP 20334)] 0x004877a2 in _dl_sysinfo_int80 () from /lib/ld-linux.so.2
+ Trace 61097
Thread 2 (Thread -1210434640 (LWP 20334))
reassigning old bugs to get rid of triage@ximian.com as assignee. do you still see this in current versions?
>> do you still see this in current versions? Yes, frequently. Currently on Evo 2.4.1, e-d-s 1.4.1.1. Installed on two machines (home and office), both on FC4. Both crash e-d-s several times a day, always during an LDAP lookup (address autocompletion). However most lookups do not crash; it appears to be completely random (e.g. it doesn't depend on the address being looked up).
okay, updating version number. thanks, patrick. :-)
I should qualify my earlier statement: I haven't seen any crash of 2.4.x specifically while using drag-and-drop. They've all been during message composition. This may or may not be the same bug as I originally reported, but is almost certainly related to LDAP (I haven't so far collected any core dumps of e-d-s to confirm).
Adjusting the summary as per patrick's last comment.
*** Bug 300157 has been marked as a duplicate of this bug. ***
Isn't it a bug in e-d-s and not in evo? I'm wondering why the bugzilla product is set to Evolution.
jan, you're right indeed.
gar....sorry for the noise
Created attachment 75930 [details] Backtrace from e-d-s 1.10 This bug seems to still be present.
Bug 354442 may be a dulicate. It has a commited patch attached.
The backtrace I posted here today is more likely the same bug as Bug 354442 (or a related one) . It did not involve any moving of messages. Entering an address in the To: or Cc: field is enough to trigger the bug, and the behaviour is exactly the same as I reported in that bug. I only read the title of this bug when I posted the backtrace - and did not find #354442 at the time (even if it was me who submitted that one in the first place). This is with Evolution-data-server 1.10, so the patch/fix from that bug does either not fix the problem, or another problem with the same symptoms has surfaced.
Created attachment 77814 [details] New backtrace The following fedora-packages: evolution-2.9.2-3.fc7 evolution-debuginfo-2.9.2-3.fc7 evolution-data-server-1.9.2-3.fc7 evolution-data-server-debuginfo-1.9.2-3.fc7
Ola, can you please install the glib and libbonobo packages as well?
Created attachment 77904 [details] backtrace with more debuginfo glib-1.2.10-23.fc6 glib-debuginfo-1.2.10-23.fc6 glib2-2.12.4-2.fc7 glib2-debuginfo-2.12.4-2.fc7 evolution-2.9.2-3.fc7 evolution-debuginfo-2.9.2-3.fc7 evolution-data-server-1.9.2-3.fc7 evolution-data-server-debuginfo-1.9.2-3.fc7 libbonobo-2.16.0-1.fc6 libbonobo-debuginfo-2.16.0-1.fc6 All packages from Fedora Development (RawHide)
uhm sorry ... can you install the libORBit package as well, please?
Closing this bug report as no further information has been provided. Please feel free to reopen this bug if you can provide the information asked for. Thanks!
It doesn't crash for me with evo 2.9.92 and e-d-s 1.9.92, although it doesn't work completely well. But that's different story and I'll file it as a different bug when I gather enough problem information.
Reopening, here is a stacktrace with evo 2.18.0 / e-d-s 1.10.0. It isn't fully complete (some debug packages were missing) but I get this crash from time to time, so next time, it will be more complete :) Distribution: Mandriva Linux release 2007.1 (Official) for i586 Gnome Release: 2.18.0 2007-03-15 (Mandriva) BugBuddy Version: 2.18.0 System: Linux 2.6.17-13mdv #1 SMP Fri Mar 23 19:03:31 UTC 2007 i686 X Vendor: The X.Org Foundation X Vendor Release: 70200000 Selinux: No Accessibility: Disabled GTK+ Theme: Ia Ora Blue Icon Theme: gnome Memory status: size: 71016448 vsize: 71016448 resident: 4079616 share: 2674688 rss: 4079616 rss_rlim: 4294967295 CPU usage: start_time: 1176717847 rtime: 66 utime: 34 stime: 32 cutime:0 cstime: 0 timeout: 0 it_real_value: 0 frequency: 100 Backtrace was generated from '/usr/lib/evolution-data-server-1.10' Using host libthread_db library "/lib/i686/libthread_db.so.1". [Thread debugging using libthread_db enabled] [New Thread -1222060352 (LWP 4372)] [New Thread -1233577056 (LWP 12093)] [New Thread -1241969760 (LWP 4397)] [New Thread -1224688736 (LWP 4373)] 0xbfffe410 in __kernel_vsyscall ()
+ Trace 130451
Thread 2 (Thread -1233577056 (LWP 12093))
Better stacktrace (evo 2.18.0 / e-d-s 1.10.0) : Distribution: Mandriva Linux release 2007.1 (Official) for i586 Gnome Release: 2.18.0 2007-03-15 (Mandriva) BugBuddy Version: 2.18.0 System: Linux 2.6.17-13mdv #1 SMP Fri Mar 23 19:03:31 UTC 2007 i686 X Vendor: The X.Org Foundation X Vendor Release: 70200000 Selinux: No Accessibility: Disabled GTK+ Theme: Ia Ora Blue Icon Theme: gnome Memory status: size: 482623488 vsize: 482623488 resident: 273104896 share: 14135296 rss: 273104896 rss_rlim: 4294967295 CPU usage: start_time: 1178800505 rtime: 1369079 utime: 758136 stime: 610943 cutime:136 cstime: 386 timeout: 0 it_real_value: 0 frequency: 100 Backtrace was generated from '/usr/bin/evolution' Using host libthread_db library "/lib/i686/libthread_db.so.1". [Thread debugging using libthread_db enabled] [New Thread -1237145888 (LWP 14937)] [New Thread -1476396128 (LWP 14962)] [New Thread -1333789792 (LWP 14960)] [New Thread -1270420576 (LWP 14959)] [New Thread -1308755040 (LWP 14953)] [New Thread -1310721120 (LWP 14952)] [New Thread -1300362336 (LWP 14949)] [New Thread -1280312416 (LWP 14947)] [New Thread -1261991008 (LWP 14943)] [New Thread -1253598304 (LWP 14942)] 0xbfffe410 in __kernel_vsyscall ()
+ Trace 134084
Thread 1 (Thread -1237145888 (LWP 14937))
(In reply to comment #21) > Closing this bug report as no further information has been provided. Please > feel free to reopen this bug if you can provide the information asked for. > Thanks! > Please, don't. We're currently deploying Evolution in a Linux-based distribution in more than 4.5K users which mostly use autocompletion, for which we provide an OpenLDAP server, and we're currently facing a strong opposition since Evo tends to hang while autocompleting or when sending a message with lots of recipients or when closing the compose message window, when it has LDAP autocompletion activated. I can't provide a backtrace ATM, since I don't have access to every machine of every user, but I'm eager to try any patch in a wide basis and to provide any other information you might need to solve this problem. (BTW, we're using evo 2.8, from Debian Etch) Thanks for your time!
José, this bug is not closed - see the status.
There are atleast 3 different crashers reported in this bug.
I dont think the first (trace) crasher is valid any more. It was part of the numerous bugs that got fixed with Devashish's ldap rewrite around 2.8/2.10 time. The second crasher is already fixed in head. The poll_ldap and ldap_op_finished seems to be unique crashers now. Let me see it. I have seen this ldap_op_finished crash on GAL as well.
Similar bug from GAL http://bugzilla.gnome.org/show_bug.cgi?id=332979
Fixed ldap_op_finished bug in head. (Similar fix has been pushed to GAL also bug #332979) Sending addressbook/ChangeLog Sending addressbook/backends/ldap/e-book-backend-ldap.c Transmitting file data .. Committed revision 7964. Im not sure of the poll_ldap crash, that could have been a side effect of this. Even then, I will try and reproduce more to get it off the code. Ill keep this bug open to see if we still get poll_ldap crashes after this. This fix should be part of GNOME 2.19.91 release. (Evolution 2.11.91)
I'm closing this bug. Please reopen if this occurs again.
This bug is definately not fixed for me... I'm running evolution 2.12.2, data-server 1.12.3, on gentoo It's crashing all the time while autocompleting against my ldap addressbook. It's easily reproducible: * Create a new mail * Type a few letters into the To: field, perhaps hit delete, to force a few ldap queries * Evolution freezes If I can help in any way by posting stack-traces/debug info etc..., I would be pleased to, just send me brief instructions as to what you need.
Runger: Sure. Please attach to gdb when it hangs both for Evolution and EDS. Only then it would be easier to see if it still happens. Possible that it is a different hang. Since this is a crash.
Hi! I haven't forgotten about this - the constant evo hangs are really driving me crazy... it just seems a little more complicated than I originally thought, and I am no longer sure how to best post the bug-reports. Looks like I have a least 2 seperate problems: 1. The hang on autocomplete as described above. As you write, it is probably actually a different bug, since it is definately a hang - evo freezes up, rather than a crash. 2. A problem where the message preview pane freezes. This happens with certain messages. The rest of evo keeps working fine, but the message preview pane is stuck on that one message, with "Formatting message..." displayed int he status bar. I've got a trace for problem #2. Should I open 2 new bugs for these items, or should we handle them here, what is preferable? Thanks, Richard
runger, I feel that they are same. Addresslookups hang for you. Check if Mail Preferences->Html Images-> Loading Images == Load images if messages in contacts. If so it will do ldap lookup. Change it to never load or always load. You may not face it. But get the gdb traces of eds and evolution for debugging.
Hi! Yes, the "Load Images" setting was set to "Contacts only", so that could definately be the same bug then. Since changing the setting to "Always load Images" I have not had any more preview pane freeze-ups. Below you will find a backtrace for evolution and eds after one of the autocompletion freezes. Does it contain anything useful? Thanks, Richard ----------------------------------------------------------------------------- Evolution Trace: (gdb) thread apply all ba
+ Trace 192579
Ok, here is another instance of the 'hang'. Not sure about the EDS trace on this one Are these traces useful? Thanks, Richard Evolution: (gdb) thread apply all ba
+ Trace 192787
And another instance of the evolution auto-complete hang: Evo: (gdb) thread apply all ba
+ Trace 194029
Thread 1 (Thread -1247131872 (LWP 29926))
This bug is really driving me crazy - any news on this issue? Thanks, regards from Vienna, Richard
You need to have debuginfo pkgs of Evolution, EDS, Exchange, Gtk+, Glib, Orbit, Bonobo. Only then we have the real traces.
Here is another, more complete backtrace... Any other debug symbols I should generate? libmono? Evolution: (gdb) thread apply all ba
+ Trace 195193
Thanks! Regards from Vienna, Richard Unger
If you have enabled/installed mono plugins disable it. It sort of corrupts your nice traces and I can see a bit of effect here. Just try to get another round of traces with that disabled, so that it is much clear and useful. Which crashed here Evo/EDS ?
Ok, here is another trace of the hang, this time with mono disabled. Looks like some kind of deadlock... Thanks, Regards from Vienna, Richard Evolution: (gdb) thread apply all ba
+ Trace 196613
Thread 1 (Thread -1244997888 (LWP 19374))
0 0xb7f9c410 in __kernel_vsyscall ()
+ Trace 196917
Runger, sorry to bother you. Can you try 2.22.1.1 or 2.23.1 ? In that case, It will be great. We have added few ldap locks in 2.23.x and fix for some composer hangs post 2.22.
Hi! Thanks, this bug does indeed seem to have disappeared for me completely in 2.22.2. Regards, Richard