Am Mittwoch 25 November 2009 23:38:25 schrieb Alan Stern: > On Wed, 25 Nov 2009, Stephan Diestelhorst wrote: > You should add a dump_stack() line to handshake_on_error_set_halt(), > just after the ehci_err() call. Then we'd be able to see the failure > pathway and determine whether it really was caused by the delay being > too short. Will do that now and report back. > > It might well be the case that the delay is too small. I will change > > to a higher value (this and the device ID should eventually be > > converted into symbolic constants). > > Otherwise a stress test is the only option. I have bumped the delay up to 1000 us, and could not trigger the issue, even with doing modprobe uvcvideo; rmmod uvcvideo in a loop for several hundred interations, with varying delays between the insert and removal. However, I am not sure, whether this is actually stress testing anything. > > Does anybody have a clue how to > > properly determine the neccessary delay, either with a stress test or > > by someone (Intel) telling us the hardware delay? > > There hasn't been any feedback from Intel. If you can find the right > person to ask, go ahead. Nope. Cheers, Stephan -- 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
