On Tue, May 31, 2011 at 10:53 PM, DJ Delorie <dj(a)redhat.com> wrote:
> It seems that the smsc95xx module can not be kept responsible for the
> bad performance of your drive.
The opposite seems to be true - activity in the smsc95xx module
*improves* performance. It still smells like a missing interrupt (or
poll) problem, but I still don't know where...
> Other points that might investigation,
> include the USB-controller on your PandaBoard (ehci, ohci, ...) or
[ 1833.619598] usb 1-1.3: new full speed USB device using ehci-omap and address 6
[ 2318.736450] usb 1-1.3: new high speed USB device using ehci-omap and address 7
[ 2695.448394] usb 1-1.3: new full speed USB device using ehci-omap and address 8
ehci-omap is built in to the kernel.
This driver seems to use all the ehci-hcd functions. You can set the
parameters on the kernel-command line like "ehci_hcd.ignore_oc=1" (to
ignore over-currents, which you probably don't have). It's late here,
and I can not concentrate the effects that the options describe in
ehci-hcd.c at the moment. You might find some hints in
linux-2.6/Documentation/usb/ehci.txt as it describes some debugging