Yes, something seem broken in the reset on VIA, but I'm not sure
what.  It's kinda hard to debug it without access to a machine and
a USB protocol analyzer.

    -- Lennart

Gary Duzan wrote:

>    Unfortunately, it seems that the uhci driver is somewhat broken
> for VIA chipset implementations. I just upgraded my old FIC SD11
> setup to a new DFI AK74 (KX133-based) one and see the same problems
> on both when trying to use a USB Zip drive. In this case, attaching
> the drive fails with a number of "reset timeout" messages. If the
> drive is connected at boot time, the boot never finishes. The last
> comment I remember on the subject (from Lennart, I believe) was
> that the uhci reset code was broken and needed to be rearchitected.
>    I'll attach some USB/UHCI/UMASS debugging output that I collected
> from the SD11 a couple weeks ago in case anyone cares to explore
> the problem further. This was on a then-current 1.5_ALPHA2.
>
>                                         Gary D. Duzan
>
> #text/plain [usb-debug.txt] usb-debug.txt
>
> In Message <200010190358.e9J3wJV00298@js1.jsnet> ,
>    Joseph Sarkes <joe@js1.jsnet> wrote:
>
> =>I have a couple usb thingies I'm playing with on my
> =>asus A7V athlon motherboard with built in usb ports.
> =>
> =>basically my problem is that when I plug in a logitech
> =>remote keyboard/mouse pair of usb plugs and then plug
> =>in my usr 56K voice fax modem usb connector, the usb
> =>probing stuff goes into a loop and finally panics the
> =>machine. In addition, when I power cycle the  modem,
> =>the whole usb chain of hubs seems to go through a whole
> =>shutdown/reset instead of just the one affected port.
> =>
> =>Do I have some misunderstanding of what I should be seeing?
> =>Is something still in progress on usb support?
> =>Will batman get out of this predicament? Tune in again
> =>folks, same bat time, same bat channel.
> =>
> =>BTW, my reason for playing with usb on the modem is that
> =>my moving hard drives around didn't actually fix the silo
> =>overflow problem I'm having with the serial port/modem
> =>combo. I figured that usb would potentially bypass the whole
> =>thing, but...
> =>
> =>another question related to this is how does one make the
> =>usb keyboard/mouse stuff take over for the ps/2 style keyboard
> =>mouse under netbsd? The bios sees the usb keyboard if the
> =>ps/2 one is unplugged, but then the machine won't work for
> =>console logins until a ps/2 keyboard is plugged in after the
> =>boot process is finished. I did manage to wsconscfg -k the
> =>second keyboard into operation afterwards.
> =>
> =>--
> =>Joseph Sarkes         jsarkes@tiac.net