Maurizio Marini ha scritto: > On Wed July 26 2006 11:12, Massimo Nuvoli wrote: >> This patch corrects a lot of possibile "segmentation fault" in the code. >> >> Now my system passes the 2000 call test happy :-) > the patch for choppy sound is in? > http://www.trixbox.org/modules/newbb/viewtopic.php?viewmode=thread&topic_id=1367&forum=7&post_id=9419#9419 >
I think no :-)
I found the "choppy" sound is coming on the isdn channle when the
system is busy to do something else than voip, example: if there is a
console in "debug" mode with a lot of lines coming out.
Now i am tryng to avoid every "segmentation fault" that is very
dangerous.... i found a lot of points where the chan_visdn or the q931
code is not buggy but unsafe, if the channel disappear after a mutex
or after a system call the system, clear, go in a dead state!
This is on the "user space" part of visdn, i dont think to put my
hands on the "driver space" as is a lot more difficult.
My bigger problem, now, is the driver (i found this) in NT mode that
is unable to recover from some strange condition and goes completly out:
-- visdn1.3: MPH-ACTIVATE-INDICATION
-- set timeout = 3003
-- visdn1.3: MPH-ACTIVATE-INDICATION
-- set timeout = 2987
-- visdn1.3: MPH-ACTIVATE-INDICATION
-- set timeout = 2973
-- visdn1.3: MPH-ACTIVATE-INDICATION
-- set timeout = 2959
-- visdn1.3: MPH-ACTIVATE-INDICATION
-- set timeout = 2944
Using the tester i see that the line come up and after a little time
the tester say "alignment error" and the line go down, in a infinite
while.
The only way to recover from this is to unload the driver, load the
driver and not everytime the channel is recovered.
I found it's very easy to put a channel in this "state", so this is a
big deal!
signature.asc
Description: OpenPGP digital signature
_______________________________________________ Visdn-hackers mailing list [email protected] https://mailman.uli.it/mailman/listinfo/visdn-hackers
