Am Montag, 31. Oktober 2005 17:10 schrieb Daniele Orlandi: > On Monday 31 October 2005 16:44, Florian Heese wrote: > > 1) After module loading I get one time for each card : > > "hfc-pci: pci-0000:00:0e.0: sync lost, pci performance too low!" > > This is something you shouldn't worry about. My experience is that AMD's > PowerNOW! and probably other clock throttling technologies make the PCI bus > unavailable to busmastering devices for a short time. > > HFC-S PCI A chips are unable to access the bus during that time and they > complain about that, however, it's often short enough that you lose just > one octect, which is quite unnoticeable in voice payload and results in a > corrupted frame in packeted data. > > Try to disable clock throttling on the box and see if the problem arises > again. > > > 2) call from 1111 to 2222: - realtime audio in both directions, works > > fine > > > > 3) call from 2222 to 1111: - realtime audio from 2222 to 1111 > > - delayed audio from 1111 to 2222 > > > > 4) If I pickup "ISDN-Phone I" you can hear the last frames from the last > > call, befor the dialtone is audible. > > Uhm... this is sort of known bug. There probably is a race condition which > leads to unemptied fifo in some direction. The delay should be 2-seconds > long, isn't it? > yes about 2 seconds
> I'm working on the voice layer right to make it more robust and a fix for > that will go in for sure. perfect... > > Bye, best regards F.Heese _______________________________________________ Visdn-hackers mailing list [email protected] https://mailman.uli.it/mailman/listinfo/visdn-hackers
