On Sun, 26 Apr 2009 salvatorelionetti@xxxxxxxx wrote: > Alan, after your instruction i've included 'usbmon' in my kernel. > Surfing in menuconfig, i've been tempted to include also: > - Root Hub TT > - Improved TT scheduling > > Situation changed after this recompilation: > - Error -28 disappear !!! > - Playback work like previous build, > (i've no applied quantitative markers different from my hears) > - Record go on but sound seems really bad. > > Going deeply on such 'badness', i've seen that > - Right channel is always ok (in quality is similar to playback) > - Left one has spikes error over samples. In detail: > o say 90% of urbs have one single error (one 16 bit sample is wrong) > o say 5% of urbs have 2 error > o say 5% of urbs have 0 error > > Spike error is 10:100 times greater (few time less) so resulting data > is seriousely damaged. Strangly error follow sign of data. > > I attach you requested data. Does that data represent one single recording? It looks like it might. There are gaps in the log -- places where no URB was submitted for several milliseconds at a time. For example, look in the region around timestamp 2636828034. All three URBs complete and they aren't resubmitted until about 5 ms later. Whatever driver you're using, it doesn't have enough URBs in its ring buffer. Instead of using 3 URBs it should use more like 20 or 30. Even better would be to continue using 3 URBs, but make each URB contain 10 packets of data instead of just one. > Can you suggest me how to proceed to found the cause of this problem? > Is possible for example to save binary data about transfer (not only > arrive time, size, direction), eventually separated from HUB -> > DRIVER HCD and DRIVER HCD -> MY DRIVER? Have you even heard about > such 'spike error' behavior? It is possible to get the binary data, but not with the command-line version of usbmon. You have to use the usbmon program; see http://people.redhat.com/zaitcev/linux/usbmon-5.4.tar.gz However until the problem of the gaps is fixed, there's no point looking at the binary data. Alan Stern -- To unsubscribe from this list: send the line "unsubscribe linux-usb" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
