EHCI USB on _top_ of UHCI?

[Bernd Walter]

On Tue, Jun 17, 2003 at 10:46:53PM -0400, David Gilbert wrote:
> I'm running 5.1-RELEASE and have USB2.0 ports in my Dell D800 laptop.
> I also have a USB2.0 drive encloseure with a 120G drive in it... so
> I'm very motivated to get this to work :).
> 
> To start, I posted briefly about this before... and some people
> suggested that this may have a bad interaction with ACPI.  I have
> ruled that out ... disabling ACPI or fixing ACPI (by obtaining fixed
> ACPI code) doesn't change the EHCI behaviour.
> 
> So ... some datapoints.  USB 1.1 (UHCI) works.  If EHCI isn't in the
> kernel all manner of USB devices work.  I have tested the drive, an
> external DVD enclosure and a mouse.  This is good.
> 
> However, if EHCI is in the kernel, attaching the drive gives the
> following messages:
> 
> Jun 13 22:15:46 canoe kernel: usb3: unrecoverable error, controller halted
> Jun 13 22:15:46 canoe kernel: usb3: blocking intrs 0x10
> Jun 13 22:16:05 canoe kernel: uhub3: device problem, disabling port 3
> Jun 13 22:16:48 canoe kernel: usb3: port reset timeout
> Jun 13 22:16:48 canoe kernel: uhub3: port 1 reset failed

Please compile USB_DEBUG into your kernel and do a
sysctl hw.usb.ehci.debug=2
before connecting a device.

> ... and then usb doesn't work thereafter.  Is this something that's

Yes - The ehci controller do only high speed and are sharing the
physical ports with one or more USB 1.x companion controllers, which
are doing full and low speed support.
New devices are check for high speed first and if that controller
stops...

> fixable?  I'm willing to do some work on this, but I'm a little at a
> loss on where to start.  I intend to track -CURRENT with this laptop,
> but my attempt to move from 5.1-RELEASE to -CURRENT was stymied by
> source that wouldn't compile.

Hard to help without errors here...

-- 
B.Walter                   BWCT                http://www.bwct.de
ticso at bwct.de                                  info at bwct.de