Evolution crashes on startup

[Valentijn Sessink]

Evolution crashes on startup in certain situations.

Will research the issue further, but here's the first results.

Backtrace (and if I can do anything to help, please say so, I have some past
experience with debuggers but that's looooong ago and have no knowledge of gdb
whatsoever, just acting like a dumb monkey here)

(evolution:11056): camel-CRITICAL **: camel_data_wrapper_set_mime_type_field: as
sertion `mime_type != NULL' failed
(no debugging symbols found)
(no debugging symbols found)
(no debugging symbols found)
(no debugging symbols found)
(no debugging symbols found)
[New Thread -1288701008 (LWP 11073)]

Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread -1261560912 (LWP 11070)]
0xb6ccb104 in g_str_hash () from /usr/lib/libglib-2.0.so.0
(gdb) thread apply all bt

Thread 9 (Thread -1288701008 (LWP 11073)):
#0 0xffffe410 in __kernel_vsyscall ()
#1 0xb7acc549 in poll () from /lib/tls/i686/cmov/libc.so.6
#2 0xb6cb58de in g_main_loop_get_context () from /usr/lib/libglib-2.0.so.0
#3 0xb6cb4f57 in g_main_context_dispatch () from /usr/lib/libglib-2.0.so.0
#4 0xb6cb551e in g_main_loop_run () from /usr/lib/libglib-2.0.so.0
#5 0xb73daf28 in link_thread_io_context () from /usr/lib/libORBit-2.so.0
#6 0xb6d0c398 in ?? () from /usr/lib/libglib-2.0.so.0
#7 0xb32ffad8 in ?? ()
#8 0xb6ccd362 in g_static_private_free () from /usr/lib/libglib-2.0.so.0
Previous frame inner to this frame (corrupt stack?)
#0 0xb6ccb104 in g_str_hash () from /usr/lib/libglib-2.0.so.0

http://bugzilla.gnome.org/show_bug.cgi?id=306271: http://bugzilla.gnome.org/show_bug.cgi?id=306271

[Sebastien Bacher (seb128)]

Thanks for your bug. What distribution/version of the evolution package are you
using?
The backtrace is pretty short/useless. Do you get some different result with
"bt" instead "thread apply all bt"?

What do you do exactly to do the crash? Does it happen at random time or while
doing a specific action?

[Valentijn Sessink (valentijn)]

(In reply to comment #1)
> Thanks for your bug. What distribution/version of the evolution package are you
> using?

Sorry for the confusion, here's more information:
Using Hoary, with the following software versions:
Evolution 2.2.1.1-0ubuntu4
libglib2.0-0 version 2.6.3-1

> The backtrace is pretty short/useless. Do you get some different result with
> "bt" instead "thread apply all bt"?

I'll try to find out.

> What do you do exactly to do the crash? Does it happen at random time or while
> doing a specific action?

It seems to happen with specific mail and/or specific mail boxes. I'm not sure
and I've asked the user to try to help me with this. I hope to get a test case
for it.

[Valentijn Sessink (valentijn)]

Created an attachment (id=2589)
this mail will crash evolution

The following e-mail will take down Evolution, when it's opened. After that,
you cannot restart Evolution, as it will immediately go down again upon
encountering this e-mail.

[Sebastien Bacher (seb128)]

Thanks for the mail example. Somebody has opened a bug about this upstream:
http://bugzilla.gnome.org/show_bug.cgi?id=306271

[Valentijn Sessink (valentijn)]

(In reply to comment #4)
> Thanks for the mail example. Somebody has opened a bug about this upstream:
> http://bugzilla.gnome.org/show_bug.cgi?id=306271

Yep, that was us; we confused bugzilla.gnome.org and bugzilla.ubuntu.com here,
so the crashing attachment is in the wrong bugzilla now; we will set pointers to
the crashing attachment in a moment.

[Wouter Hanegraaff (wouter-openoffice)]

This bug was closed upstream since it is supposed to be a dbus bug. I opened a
dbus bug at https://bugs.freedesktop.org/show_bug.cgi?id=3525
.

[Tristan Wibberley (tristan-wibberley)]

The dbus bug appears to have been closed as not theirs and is supposedly being tackled in evolution upstream. There was no link to any gnome bug provided, and the one listed in this bug shows status NOTGNOME. Which I think is wrong. Evolution shouldn't just terminate because dbus is crap - it should do something sensible like give an error message and allow the user to go to the calendars and stuff.

[Sebastien Bacher (seb128)]

Do you still have that issue? What version of Ubuntu do you use now?

[Albert Cardona (cardona)]

I can confirm this disaster. I've apt-get ubuntu-desktop on a kubuntu powerpc dapper installation. After using evolution for 2 days, now it crashes at startup and there's no obvious thing I can do to stop it.

After attempting to start evolution about 10 times, then no other app launches. All crash! Which of the numerous logs is supposed to give some info on what is going on?

[Albert Cardona (cardona)]

Here is the output of running evolution with:
$ DBUS_VERBOSE=1 evolution &> evo.output

CalDAV Eplugin starting up ...

(evolution-2.6:12729): evolution-mail-WARNING **: ignored this junk plugin: not enabled or we have already loaded one

(evolution-2.6:12729): e-utils-WARNING **: Plugin 'Spamassassin junk plugin' failed to load hook 'org.gnome.evolution.mail.junk:1.0'

(evolution-2.6:12729): camel-WARNING **: camel_exception_get_id called with NULL parameter.

(evolution-2.6:12729): camel-WARNING **: camel_exception_get_id called with NULL parameter.

It crashes and shows the message "the app has crashed, do you want to restart, abort, notify developers" or something like that.

[Mike (bild85)]

here's my output of $ DBUS_VERBOSE=1 evolution &> evo.output:

$ cat evo.output
CalDAV Eplugin starting up ...
error : unterminated entity reference Trojans
error : unterminated entity reference Trojans

** (evolution-2.6:12473): CRITICAL **: e2k_global_catalog_lookup: assertion `E2K_IS_GLOBAL_CATALOG (gc)' failed
$

[Sebastien Bacher (seb128)]

That bug seems to list different issues now. The "e2k_global_catalog_lookup: assertion `E2K_IS_GLOBAL_CATALOG (gc)' failed" error looks like http://bugzilla.gnome.org/show_bug.cgi?id=333163 upstream that should be fixed with dapper

Is somebody still having the original issue with dapper?

[Valentijn Sessink (valentijn)]

Yay. It's fixed. I'll leave the rant about the 14 months fix to a *crashing* mail program for another episode, but suffice to say that this one seems fixed in Dapper.

In short: AFAIK it's fixed, you may close the bug.

The long version is that only the crashing part is fixed, but the decoding of the test case (see above) does not work yet: it shows up silly now.

For a test:
sudo apt-get install uudeview
uudeview testcase.rfc822
(see testcase above). You will find a multipart MIME base64 encoded message with four sub-attachments inside, all of them being called "forward", with new base64 attachments inside.

Importing the message in Evolution (file-import) will show a single file called "forward" and a large text file that is not decoded. However, this behaviour is sort-of consistent with Mozilla Thunderbird, so although it is a non-expected behaviour, I will not file another bug report, but feel free to do so (or rename and re-priorize this one). (And if you do so, please note that the testcase and "openwave webengine" (which produced the e-mail) is probably at fault firstly).

So finally, long story short: this bug can be closed - although the test case could be the basis for another bug (which I won't file).

[Sebastien Bacher (seb128)]

Marking the bug as fixed. Sorry for the delay to get it fixed but we get lot of bugs for the number of people and that's not possible to fix everything downstream. Upstream gets lot of issues too which explain that it can takes time.

Feel free to open an another bug about with your incorrect decoding example (on bugzilla.gnome is probably better if you have an account there)