On Tue, Feb 20, 2018 at 03:05:33PM +0100, Martin Pieuchot wrote: > Diff below implements the USBPcap interface for dumping USB isochronous > frames. It can be very useful to analyze what the stack is doing. Note > that tcpdump(8)'s snaplen default is too small to capture such frames. I > used the following to analyze uaudio(4) traffic on ehci(4): > > # tcpdump -s 3303 -w usnd.pcap -i usb0 > > The first capture I did was with AUDIODEVICE=snd/0 then I did a second > one with AUDIODEVICE=rsnd/0. Like Stefan I hear some crackling noise > in the first case. > > What I could see after opening the pcap(3) files in wireshark confirmed > my theory. When using the raw audio device delay between frames is less > than 8us. However in the general case it jumps to 35us. > > It is too early to say what needs to be fixed, but at least I'd like to > get the USBPcap part in, ok?
this is very cool! the diff looks correct, ok ratchov.
