[Asterisk-Dev] moh issues
rizzo at icir.org
Tue Jun 14 00:37:26 MST 2005
followup to myself -- reading from the net, it seems that there are
issues with the (userland) threading library on FreeBSD 4.x,
which might explain this particular problem i am seeing.
I will move the discussion to the *-bsd list...
On Sat, Jun 11, 2005 at 05:33:39AM -0700, Luigi Rizzo wrote:
> i tried to enable music-on-hold (res_musiconhold.c) on my system
> (asterisk-current with FreeBSD 4.11 and the modified chan_oss.c
> that is on mantis) but i am having problems which i am not clear how
> to debug.
> I suspect something in the thread scheduling, but it might well
> be some kind of deadlock caused by the activation of moh.
> the bottom line is that i'd like to know whether i am trying to
> use something that is known broken, or whether moh does actually work,
> in which case i'd like to know on what platform.
> --- Detailed problem description: ---
> Symptoms are that when the moh thread starts (because it founds a
> valid directory in moh.conf), after a few iterations it remains
> blocked on the usleep() near line 488 in monmp3thread(), despite
> the value passed is a reasonably small one (< 100ms).
> As far as i can tell there are no moh clients.
> Then sometimes if i type a few commands on the CLI the thread,
> or there are some network events, the usleep() unblocks and
> monmp3thread does one more iteration, but basically most
> threads seem to be blocked, CPU usage is close to 0,
> mpg123 is blocked on a select()
> The audio device (which is unrelated, as mpg123 should talk
> to stdout not to the audio device, but it is managed by
> a separate thread) does not respond anymore.
> SIP transactions also seem to progress randomly.
> If i try to stop asterisk ("stop now" from the cli) behaviour
> varies - sometimes it terminates reasonably well, most of the
> times enters an infinite loop which a bit of debugging shows to
> be in monmp3thread(), and trying to break out of it with control-C
> causes eventually a few free()s of invalid pointers and a core dump.
> Any suggestions on what to look for in order to debug this problem ?
> Asterisk-Dev mailing list
> Asterisk-Dev at lists.digium.com
> To UNSUBSCRIBE or update options visit:
More information about the asterisk-dev