Hi folks,
has anyone here running sofia-sip with ortp (in scheduled mode!) in one
application ?!?!
i found a posting from Abhishek Tiwari on the list - but the questions
is Abhishek, do you use ortp in the scheduled mode ?
in linphone ortp is not used in the scheduled mode ... afaik from the
sources ...
in the "scheduled" mode, ortp spawns a seperate thread where the
jitterbuffer and sending/receiving of rtp packets takes place.
now, when using sofia-sip in UNthreaded mode (su_root_threading(0)) i
get the following behavior:
under some circumstances the voice starts clicking - when measuring with
an oscilloscope it looks like there is some data of the packet missing,
however in wireshark the packet is correct - so we expect a timing
issue.
in my application there is a seperate pthread which does the rtp
communication - which is spawned from the sip callback.
so all in all i have the follwing setup:
- main process running su_root_run ()
- seperate thread running the ortp scheduler
- main process calls sip-callback (when call is ready)
- sip callback spawn ortp communication thread
some time the voice starts clicking, some time not - it's not really
reproduceable ... im sorry.
when i am using the sofia-sip application stand alone - just for the sdp
negotiation and start the rtp communication thread in a seperate process
(via cmdline app) then the voice quality is ok - reproduceable ;)
i am observing exectly the same behaviour when using ccRTP (which also
uses threads) when its built into the sofia-sip application. which is
basically the same setup as described above.
so i guess i have some issues with pthreads ... but what, how and
when ... ?!?!
another point i found out about sofia-sip is the following:
- i registered a su_wait for a socket where lcdprocs LCDd is connected
- when receiving data from the socket the callback gets invoked - thats
ok.
- what i observed is that when processing large amount of data (takes
some time and LCDd is sending more data while the callback is
running) ...
-> the callback gets invoked while it is still running ... ?!?!
- this looks to me like there is a thread invoked in su's callback
mechanism - even if su_root_threading is turend off ?!?!
- otherwise it should be impossible that the callback function
get started a second time while it's still running ...
however, we have implemented some workaround to live with this,
but maybe this has something in common with the rtp-thread
behaviour ... ?!?!
any hints and points to where i can start would be very very
appreciated !!!
thanks for all,
mexx.
-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
Sofia-sip-devel mailing list
Sofia-sip-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/sofia-sip-devel
