| Summary: | [abrt] evolution-data-server-3.0.1-1.fc15: Process /usr/libexec/e-calendar-factory was killed by signal 11 (SIGSEGV) | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Product: | [Fedora] Fedora | Reporter: | Peter Robinson <pbrobinson> | ||||||||
| Component: | evolution-data-server | Assignee: | Matthew Barnes <mbarnes> | ||||||||
| Status: | CLOSED CURRENTRELEASE | QA Contact: | Fedora Extras Quality Assurance <extras-qa> | ||||||||
| Severity: | unspecified | Docs Contact: | |||||||||
| Priority: | unspecified | ||||||||||
| Version: | 15 | CC: | ark, christoph.maser, jlockman3, mbarnes, mcrha, mitko.69 | ||||||||
| Target Milestone: | --- | ||||||||||
| Target Release: | --- | ||||||||||
| Hardware: | x86_64 | ||||||||||
| OS: | Unspecified | ||||||||||
| Whiteboard: | abrt_hash:c8c5ef1e3589d74018612194eba3c34edf52fbcb | ||||||||||
| Fixed In Version: | evolution-mapi-3.2.3 | Doc Type: | Bug Fix | ||||||||
| Doc Text: | Story Points: | --- | |||||||||
| Clone Of: | Environment: | ||||||||||
| Last Closed: | 2012-02-14 08:36:31 UTC | Type: | --- | ||||||||
| Regression: | --- | Mount Type: | --- | ||||||||
| Documentation: | --- | CRM: | |||||||||
| Verified Versions: | Category: | --- | |||||||||
| oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |||||||||
| Cloudforms Team: | --- | Target Upstream Version: | |||||||||
| Attachments: |
|
||||||||||
|
Description
Peter Robinson
2011-05-13 08:59:58 UTC
Created attachment 498726 [details]
File: maps
Created attachment 498727 [details]
File: backtrace
Package: evolution-data-server-3.0.1-1.fc15 Architecture: x86_64 OS Release: Fedora release 15 (Lovelock) Comment ----- creating cal appointement using mapi to an exchange server *** Bug 704469 has been marked as a duplicate of this bug. *** *** Bug 704237 has been marked as a duplicate of this bug. *** BTW I don't think this should make any difference but its running over a forwarded X session (over ssh). Aha, I see, there failed recurrence blob parsing. Could you try with this test build [1], please? It uses an upstream patch done by Seanius, and I'm wondering whether it'll fixes your issue too or not. If not, then we'll try to investigate more. Thanks in advance. (In case it'll work then it'll be rather funny, because the patch was ready before 3.0.0 release, I only didn't want to change the code just before the release. I didn't think it's safe doing so.) [1] http://koji.fedoraproject.org/koji/taskinfo?taskID=3069427 It in fact makes it worse as it pretty much instantly crashes evolution (I'm glad you didn't make the change :) The abrt processing fingers a number of bugs so here's the output: Logging into Bugzilla at https://bugzilla.redhat.com Checking for duplicates Bug is already reported: 699766 Bug 699766 is a duplicate, using parent bug 656221 Bug 656221 is a duplicate, using parent bug 649400 Logging out Status: CLOSED UPSTREAM https://bugzilla.redhat.com/show_bug.cgi?id=649400 Ultimately it ends up back at https://bugzilla.gnome.org/show_bug.cgi?id=330728 Actually just an update to that. It crashed with that last comment. I rolled back the evolution-mapi version and it still crashed. So I connected to the VM console and ran evolution from gnome in the VM. It no longer crashed. Upgraded to the new version of mapi and tried again. No crash but I still get the error in the screenshot in bug 704237 with a red error box in evo with this same e-calendar-factory crash. (In reply to comment #8) > It in fact makes it worse as it pretty much instantly crashes evolution (I'm > glad you didn't make the change :) > > Ultimately it ends up back at https://bugzilla.gnome.org/show_bug.cgi?id=330728 It seems this bug happens when I start up in the calendar component. I start with "evolution --component mail" to resolve it and its fine. Not sure if there's a little used codepath there that still has the above bug on startup (yes, separate issues I know). (In reply to comment #8) > Logging into Bugzilla at https://bugzilla.redhat.com > Checking for duplicates > Bug is already reported: 699766 > Bug 699766 is a duplicate, using parent bug 656221 > Bug 656221 is a duplicate, using parent bug 649400 I see, it is not caused by the evo-mapi update, it's something different. (In reply to comment #9) > ...but I still get the error in the screenshot in bug 704237 with a red error > box in evo with this same e-calendar-factory crash. OK, so the modified code also suffers of this issue. I think it's because your event was modified in a way which evo-mapi reads incorrectly, and gets out-of-memory-bounds because of that. Could you run e-calendar-factory from a console like this: $ MAPI_DEBUG=10 /usr/libexec/e-calendar-factory &>log.txt keep it for later use and attach here only last 2KB of that file, please? I would like to keep it for later if more output will be needed. The log should give us raw data, one of them recurrence blob which it was trying to parse. With that I should be able to reproduce the issue and possibly fix it. (The other option is to give you a debugging package, with a patch for exact debug prints in the area I'm focusing to, instead of this raw chatty log).
> $ MAPI_DEBUG=10 /usr/libexec/e-calendar-factory &>log.txt
> keep it for later use and attach here only last 2KB of that file, please? I
Sure. Does it contain any personal data?
(In reply to comment #12) > > $ MAPI_DEBUG=10 /usr/libexec/e-calendar-factory &>log.txt > > keep it for later use and attach here only last 2KB of that file, please? I > > Sure. Does it contain any personal data? It does at the beginning, it's prints all raw communication between client and server. Bottom 2KB usually doesn't show much personal data, it's mostly about components and values got from the server. I got this from e-calendar-factory: "(e-calendar-factory:6077): libecal-CRITICAL **: e_cal_component_set_summary: assertion `summary->value != NULL' failed Bye." Version 3.0.2-1 Package: evolution-data-server-3.0.2-1.fc15 Architecture: x86_64 OS Release: Fedora release 15 (Lovelock) Comment ----- opened evolution Package: evolution-data-server-3.0.2-1.fc15 Architecture: x86_64 OS Release: Fedora release 15 (Lovelock) Comment ----- Unsure on what triggerred it I'm sorry it took so long, this got quite low in my todo list, but as the new bits for rrule binary blob will reach its production time soon, then it's quite bad we do not have this issue sorted out yet. I built a test package for evolution-mapi in Fedora 15 [1] which prints binary data on e-calendar-factory console, which are being parsed by the function. I would like to ask you for a help, to install this package and give me those debug prints. The best if you are able to reproduce the issue, but if not, then moving away local calendar cache from ~/.cache/evolution/calendar/mapi* will result into a new refetch of all calendar items, thus it may step over that special one+ too. You can get the log by steps like these: a) close evolution b) make sure e-calendar-factory is not running c) run e-calendar-factory from a console like this: $ /usr/libexec/e-calendar-factory &>log.txt d) after few seconds (when the factory is up and running) run also evolution from another console and go to Calendar view, and let it fetch items or reproduce the issue. The log.txt file may be larger by amount of blobs for recurring events. The log may not contain much private information - if I understood it properly, then mostly changed Subject texts for particular events, if any. The used patch, apart of debugging info, contains also new bits for rrule binary parsing, those, which made it crash to Peter. [1] http://koji.fedoraproject.org/koji/taskinfo?taskID=3252011 Created attachment 516901 [details]
Log from debug-build
here is a log as requested. hth, thanks for the update. have a nice weekend.
Thanks, it contains data I was looking for. One thing I'm not sure about, did the factory crash for you like for Peter, or it survived with the new recurrence blob parsing code? I'm asking, because I would expect seeing some error in the log too, a note about factory crash. The thing is, if it doesn't crash, then the new parsing works for you correctly. Package: evolution-data-server-3.0.3-1.fc15 Architecture: x86_64 OS Release: Fedora release 15 (Lovelock) Comment ----- Sudden crash. No established root cause Package: evolution-data-server-3.0.2-1.fc15 Architecture: x86_64 OS Release: Fedora release 15 (Lovelock) Comment ----- Just crashed I'm wondering, were anybody able to reproduce this under Fedora 16, aka evolution-mapi-3.2.x, please? I see all duplicates from Fedora 15, and even that only with 3.0.1, thus I tend to believe that this was fixed in 3.2.0+, but I would rather have a confirmation from you. I didn't see any crashes like this with 3.2.3-1.fc16 I've not seen this particular one on F-16 (that I can remember) so I think it can be closed. Thanks for the update. Feel free to reopen, in case you'll have an event which can reproduce this. There is a possibility to clear your local cache, under ~/.cache/evolution/calendar/mapi..../, while e-calendar-factory is not running, and let evolution reload all the events from the server. This way it may step on the offending one too, in case it wasn't deleted from the server yet. I'm not asking you, as this is an expensive operation (slow with server download). |