[Moved to linux-usb because this is unrelated to libusb] On Mon, 26 Jul 2010, Hans Petter Selasky wrote: > Hi Alan, > > > > For example the use of the alt_next field in the EHCI TD on BULK > > > endpoints, to receive multiple short packets. On some chipsets it works. > > > On others it crashes the hardware. The EHCI spec. is completely silent > > > on the issue. > > I found out that the EHCI chip was using the qtd_next on short packet instead > of alt_next, and then corrupted the total-bytes field: > > http://svn.freebsd.org/viewvc/base?view=revision&revision=197682 > > Search google for the commit ID, and possibly you will find the logs showing > this too. I couldn't find the logs or bug reports. Can you provide any URLs? > > No it isn't. Section 4.10.2 says very explicitly how the alt_next > > field should be used. > > Ok, right, but it doesn't change the fact: Short packets on EHCI TD's are > buggy and I really hope it is better with the XHCI :-) It's odd that I haven't encountered any reports about this problem. Maybe it's because the short packets tend to occur in the last TD of an URB, where we don't care if the queue keeps on running and so the alt_next pointer isn't set. How often do the Intel controllers follow the wrong pointer? 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
