pidgin crash on FreeBSD

[GoogleCodeExporter]

What steps will reproduce the problem?
1. install Skype (non-DBus version) and Pidgin
2. run skype and pidgin -d
3. enable Skype account

What is the expected output? What do you see instead?

Pidgin should show the Skype contacts that are online. Instead, Pidgin quits.

What version of the plugin are you using? On what operating system?

rev604 from SVN, FreeBSD 8.2-PRERELEASE amd64, Skype 2.0.0.72.

Please provide any additional information below.

The debug log ends with:

(14:54:32) prefs: /purple/savedstatus/default changed, scheduling save.
(14:54:32) account: Connecting to account (myskypename).
(14:54:32) connection: Connecting. gc = 0x80c65d880
(14:54:32) skype_x11: Set the XErrorHandler
(14:54:32) skype_x11: Opened display
(14:54:32) skype_x11: Created X11 messaging window
(14:54:32) skype_x11: Created skype Atom
(14:54:32) skype_x11: Skype instance found
(14:54:32) skype_x11: Charging lasers...
(14:54:32) GThread: file gthread-posix.c: line 254 
(g_cond_timed_wait_posix_impl): error 'Operation not permitted' during 
'pthread_cond_timedwait'

Original issue reported on code.google.com by matthias...@gmx.de on 19 Dec 2010 at 1:57

[GoogleCodeExporter]

"non-dbus version" means: create an account for the non-dbus "Skype" account, 
not "Skype (DBus)".

Original comment by matthias...@gmx.de on 19 Dec 2010 at 2:00

[GoogleCodeExporter]

FreeBSD port preview at http://mandree.home.pages.de/freebsd/

Original comment by matthias...@gmx.de on 19 Dec 2010 at 2:11

[GoogleCodeExporter]

Hmm.  Not really sure, sorry.  Does FreeBSD have pthreads?

Original comment by eionrobb on 19 Dec 2010 at 7:44

[GoogleCodeExporter]

Yes it does. The relevant manual page is online at 
<http://www.freebsd.org/cgi/man.cgi?query=pthread&apropos=0&sektion=0&manpath=Fr
eeBSD+8.1-RELEASE&format=html>

Original comment by matthias...@gmx.de on 19 Dec 2010 at 10:43

[GoogleCodeExporter]

also see Issue 104, which may be related: 
<http://code.google.com/p/skype4pidgin/issues/detail?id=104>

Original comment by matthias...@gmx.de on 19 Dec 2010 at 10:44

[GoogleCodeExporter]

And it doesn't die if you use the DBus version?
A workaround might be to change skype_messaging.c:220 from
#ifdef _WIN32
to
#if _WIN32 || __FreeBSD__

Original comment by eionrobb on 19 Dec 2010 at 11:10

[GoogleCodeExporter]

Oh, wait no.  That's some win32 specific code in there.  Lemme make up a new 
skype_messaging for you to try.

Original comment by eionrobb on 19 Dec 2010 at 11:11

[GoogleCodeExporter]

Try with this

Original comment by eionrobb on 19 Dec 2010 at 11:14

Attachments:

• skype_messaging.c

[GoogleCodeExporter]

Issue 104 has been merged into this issue.

Original comment by eionrobb on 19 Dec 2010 at 11:15

[GoogleCodeExporter]

The Linux-Skype that we use on FreeBSD doesn't have dbus support, it complains 
about unrecognized options '--enable-dbus' and '--use-session-dbus', so I'm not 
even building the dbus version.

The problem persists with the new skype_messaging.c, however.  The relevant 
pthread_cond_timedwait() manual page documents for EPERM: The specified mutex 
was not locked by the calling thread.

Original comment by matthias...@gmx.de on 19 Dec 2010 at 11:59

[GoogleCodeExporter]

I'm assuming you modified the attached skype_messaging.c to actually compile? :)
Can you get a backtrace of the assert, just so I can see which line it's 
complaining about?

Original comment by eionrobb on 20 Dec 2010 at 12:12

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

That previous comment was unintelligible. I have extended the #ifdef _WIN32 
around the timeout definition to comprise __FreeBSD__, and at least it no 
longer crashes now.

However it seems that my Skype has issues on the particular computer I'm 
currently testing on, so I can't say if it works, or just not crashes.

Original comment by matthias...@gmx.de on 20 Dec 2010 at 1:18

[GoogleCodeExporter]

So - I've sorted the Skype issues out, FTR: snd_hda needs to be loaded, or 
possibly some other sound driver.  With that loaded, with your fix, and with my 
fix as described in Comment 13, it just works.

Find attached the patch I'm using against rev 604. I think I'll release the 
port as soon as the re-rolle r604 tarballs have propagated across the FTP 
mirrors.

Original comment by matthias...@gmx.de on 20 Dec 2010 at 1:43

Attachments:

• patch-skype_messaging.c

[GoogleCodeExporter]

This issue was closed by revision r605.

Original comment by eionrobb on 20 Dec 2010 at 2:49

• Changed state: Fixed

[GoogleCodeExporter]

Thanks for your help testing :)

Original comment by eionrobb on 20 Dec 2010 at 2:49