[Freeswitch-dev] probs with RTP timestamps (again?)

Sluschny, Thomas Thomas.Sluschny at SIEMENS.COM
Wed Jan 23 08:14:29 EST 2008


hi,
 
i just tested a new version from 21.01. (FreeSWITCH Version 1.0.pre3
(7309M)).
 
If i call a phone from FreeSwitch i hear in phone robot like voice,
direction phone -> FreeSwitch is ok (i use FS on WinXP).
 
A look to wireshark shows that FreeSwitch send all 60ms 3 packets with
the same timestamp,
in versions before timestamp was always incremented by 160 to be
synchron to RTP data bytes.
 
I remembered the same error was already there:
http://jira.freeswitch.org/browse/MODENDP-40
and than fixed.
Is it now back? or another bug?
 
In this version i got also some CRITs, may be this has the same cause:
2008-01-23 11:47:50 [CRIT] switch_time.c:387 softtimer_runtime() Forward
Clock Skew Detected!
2008-01-23 11:48:03 [CRIT] switch_time.c:387 softtimer_runtime() Forward
Clock Skew Detected!
2008-01-23 11:48:54 [CRIT] switch_time.c:387 softtimer_runtime() Forward
Clock Skew Detected!
2008-01-23 11:50:56 [CRIT] switch_time.c:375 softtimer_runtime() Reverse
Clock Skew Detected!
2008-01-23 11:55:36 [CRIT] switch_time.c:387 softtimer_runtime() Forward
Clock Skew Detected!
 
so i went little bit deeper and come to Sleep() method from windows.
this sleep is very inaccurate,
it counts in 15ms steps, so a sleep(1) is the same as sleep(15).
 
my idea was to implement a better timer with WaitableTimer and Wait For
SingleObject.
now i have a timer with 1ms resolution, but the problem remains :(
 
regards,
Thomas
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.freeswitch.org/pipermail/freeswitch-dev/attachments/20080123/02549f9a/attachment.html 


More information about the Freeswitch-dev mailing list