Re: mISDN_send: error -12 strikes again

Short update: I've written two python scripts that access asterisk via
the management interface, two instances of the first script try to
generate new calls, the second script kills them -- randomly via both
sides of the call. I have a hfc-multi card crossover connected (two NT,
two TE ports).

With this setup I can reproduce the deadlock in minutes not days.

Looks like several threads are waiting for our chan_lock, but some
others wait for a lock in __ao2_lock in main/astobj2.c. 

Unfortunately I've not yet turned on lock debugging in asterisk (and
this experiment is with a newer not-yet-instrumented version of
asterisk). So I currently can't find out which thread is the first to
obtain one of these locks.

I'll keep you updated :-)

Ralf

-- 
Ralf Schlatterbeck             email: [email protected]