xsane busted with usb2

[Hans Petter Selasky]

On Tuesday 06 January 2009, M. Warner Losh wrote:
> In message: <200901061721.08498.hselasky at c2i.net>
>
>             Hans Petter Selasky <hselasky at c2i.net> writes:
> : On Tuesday 06 January 2009, M. Warner Losh wrote:
> : > In message: <200901061653.11745.hselasky at c2i.net>
> : >
> : >             Hans Petter Selasky <hselasky at c2i.net> writes:
> : > : On Tuesday 06 January 2009, M. Warner Losh wrote:
> :
> : I agree that we should put something somewhere about the most common
> : problems switching to USB2 and its solutions. Maybe in UPDATING?
>
> I think we need to have a pointer from UPDATING to a wiki page that
> has these entries on it.  Do you have an account on wiki.freebsd.org?

Yes, I do.

>
> : > : > I'm still finding odd things
> : > : > that don't work with usb2, and so far my fallback has been to just
> : > : > use the old stack for those things...
> : > :
> : > : Can you explain a little bit more what you mean. Are these things
> : > : Host Controller drivers, middle ware or USB device drivers?
> : >
> : > Devices are what are failing.
> : >
> : > : Are you sure that you have tried every possible command in usbconfig
> : > : and your device does still not work?
> : >
> : > I haven't.  But usually it is things like burning a DVD from a usb
> : > disk doesn't work to a USB DVD player (using storage_ata).  Burning it
> : > from a firewire disk does seem to work.  There's no I/O errors when
> : > this happens.  The uscanner thing that maybe this will fix.  I've had
> : > trouble unloading usb2 modules since sometimes they hang.  It can be
> : > quite time consuming to file bug reports on all these things, since I
> : > don't want to file one that wastes your time for being too vague or
> : > unreproducible.  I've only recently started using usb2 heavily for
> : > day-to-day tasks rather than just-test-it-out tasks I'd been doing
> : > before.
> :
> : Hi,
> :
> : If the USB device does not respond, its firmware might have crashed!
> : Simply put: The new USB stack is in some cases several times faster than
> : the old one. Some USB devices simply die because the firmware on the USB
> : device is badly designed.
>
> Well, the usb device doesn't crash, usually.  The problem smells more
> like issues in the usb stack (eg, heavy load only on that device
> works, but heavy loading with multiple devices doesn't with usb2,
> similar workloads work with usb1), but I've not done enough of a root
> cause analysis to be able to say for sure.  This is the part where
> I've not had the time to do proper testing of the exact workload to
> give you good feedback.  I haven't even eliminated the possibility
> that this might be a hardware problem with, say, the disk in the
> enclosure...
>
> One thing I'm seeing at the moment is that with your suggested quirk
> changes is an inability to stream on the burning of the DVD....
>
> : a) Send the device to me for debugging (I cannot promise that you get it
> : back).
> : b) Sit down an adjust the timing of the control transfers until you hit
> : the needle.
>
> These alternatives are going to be not very viable...
>

--HPS