Clemens: Can you explain briefly how the snd-usb-audio driver decides how many URBs to submit and how many isochronous packets to use in each URB? I found the logic in endpoint.c a little confusing. Evidently once the total number of packetss and the number of URBs are known, the packets get divided up as evenly as possible among the URBs. But I don't understand how the totals are determined. In practice, it seems that the pipeline depth is often lower than I would expect. For example, in one test a 48 KHz 16-bit mono full-speed USB audio output device (maxpacket size = 96, interval = 1 frame) ended up getting only two URBs and three packets total. Wouldn't it make more sense to allow for interrupt latencies longer than 1 ms, and perhaps also to try to minimize the rate at which IRQs occur? Two URBs with five packets each seems reasonable. Thanks, 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
