GNOME Bugzilla – Bug 708437
[abrt] Crash on contact removal notification processing
Last modified: 2019-02-25 06:46:28 UTC
Moving this from a downstream bug report: https://bugzilla.redhat.com/show_bug.cgi?id=1010087 Description of problem: Not sure what triggered this, I've been poking at carddav sync. Version-Release number of selected component: evolution-3.9.92-1.fc20 Additional info: reporter: libreport-2.1.7 backtrace_rating: 4 cmdline: evolution crash_function: item_removed executable: /usr/bin/evolution kernel: 3.12.0-0.rc1.git2.2.fc21.x86_64 Core was generated by `evolution'. Program terminated with signal SIGSEGV, Segmentation fault.
+ Trace 232509
Thread 1 (Thread 0x7f98c0937a40 (LWP 13532))
"Thanks for a bug report. Based on the backtrace, a notification about removed contact was received, and UI tried to react on it. I think it'll be harder to reproduce, but I'm moving this upstream anyway. Please see [1] for any further updates. If possible, please CC yourself there, in case upstream developers will have additional questions." That sounds correct: I have a CalDAV server configured in Evo, and I was doing some contact cleanup. This may well have happened when I deleted a contact from the web interface and Evo noticed that change.
I may actually have deleted a user with Evo and it crashed immediately afterwards. I don't 100% recall.
I tried to reproduce the crash, also under valgrind, but no luck. I simulated the "deleted from other server" by opening the second window which had selected the same WebDAV book (thus I saw the chagens in both windows). Valgrind didn't claim anything during this add/remove contact/contact-list tests.
I just ran into this bug twice on F20 with evolution 3.10.4-2.fc20 and eds 3.10.4-3.fc20. In both cases I had a contact edited so it would be a duplicate to another one (same email address or same name). First crash: I renamed a contact (within Evolution) so that two contacts within the same address book had the same first and second name, but with different presentation (one was "A, B", the other one was "B A"). I immediately deleted one of these contacts later. Second crash: I added the email address from contact A to contact B, saved and immediately deleted contact A after that. In both cases I would expect evolution to display the well-known dialog that it found some duplicate but as it looks I was faster with deleting the contact. In both cases my addressbook was consistent after starting evolution again and all changes made before the crash were made, including deleting the contact which led to the crash.
Thanks for the update, Christian. Do you use WebDAV address book too, or a different kind of the book?
No, no WebDAV. Just 2 local address books (afaik this is the default configuration in Evolution): one called "personal", one with addresses automatically collected from mails. Both contain some hundred contacts and are stored in ~/.local/share/evolution/… on a encrypted filesystem on SSD.
Thanks for the update. I probably do something wrong, but I still cannot reproduce this crash. I tried your steps, but it seems the data are more important to trigger the issue (I guess a contact removal also invokes slight sort change, which may move cursor, thus change the view, but that's more a guess than anything else). When I was following your steps, I was asked about similar contact when saving, which reminded me of bug #707121, but it had very different circumstances and results than this bug report.
I cannot reproduce this bug either. It happened twice doing the same thing but I often did this and it didn't reliably crash.
I tend to close this, due to an non-reproducible nature of the issue, and no duplicates since 3.10.x (which can also mean that the users do not delete their contacts). The current stable is 3.16.1, but any reproducible steps for 3.16.x are highly welcome. Opinions?
I haven't seen it lately, but then I haven't deleted any contacts, I don't think.
(In reply to Milan Crha from comment #9) > I tend to close this (In reply to Adam Williamson from comment #10) > I haven't seen it lately Closing this bug report as no further information has been provided. Please feel free to reopen this bug report if you can provide the information that was asked for in a previous comment and if this still happens in a recent supported version. Thanks!