This is strange--Using 2.6.11.2, I have build into the kernel the
appropriate USB drivers for my card and enabled USB Human Interface
Device support. Hotplug support is also enabled.

When I boot my machine, my keyboard is not recognized. But if I unplug
it and plug it in again, it is recognized, and works just fine from
then on.

My current solution is to use a USB-to-PS2 adapter for my keyboard, but
I'd prefer just to plug it into the USB port.

Is there a way to explicitly tell linux to scan the USB bus for
devices?
How?

Is there an appropriate place in the init scripts to put such a
command?
Where?

Motherboard is a supermicor P6-DGU (dual PIII from about 1998). It has
built-in USB-1.1, but that's turned off in BIOS. Instead, I'm using a
PCI usb 2.0 card from Wal-Mart.

Thanks for your kind support!

kernel update should help a lot.

Hmm...I've tried as late as 2.6.15.2 with no help; seems to behave just
the same w.r.t. keyboard.

Though k3b does send the SCSI controller out into the weeds with
2.6.15.2.
Unfortunately, I haven't had time to try a bunch of kernel versions &
see where it broke; I've got a workaround for the kb, and there's
nothing in 2.6.15 (16?) that I think I really need.

Just curious more than anything if anybody had seen similar behaviour
with the keyboard.

I rsynced slackware at work today, so this weekend while I'm screwing
around I may see what usb commands I can put in rc.local or whatever to
rescan the bus.

The keyboard is kind of a PITA to investigate because there's a reboot
each time I try something (gotta have something to type on!), and my
old machine takes a while just to get through BIOS.

thrugoodmarshall@hotmail.com wrote:
> When I boot my machine, my keyboard is not recognized. But if I unplug
> it and plug it in again, it is recognized, and works just fine from
> then on.

I had a similar problem recently. s/keyboard/mouse/ I believe I forgot
to compile in /proc/usb or some similar USB filesystem. Thus the boot
scripts couldn't scan for devices in the standard places. Hotplug
doesn't work on devices that were already plugged in when the machine is
started.

Later,
Daniel

On Fri, 17 Feb 2006 12:41:06 -0800, thrugoodmarshall@hotmail.com wrote:

> Hmm...I've tried as late as 2.6.15.2 with no help; seems to behave just
> the same w.r.t. keyboard.
> Though k3b does send the SCSI controller out into the weeds with
> 2.6.15.2.

cant help you there.
but it might be helpful to upgrade scsi controller firmware, and the
burners firmware (to see if theres any difference before reporting a bug)

> Unfortunately, I haven't had time to try a bunch of kernel versions &
> see where it broke; I've got a workaround for the kb, and there's
> nothing in 2.6.15 (16?) that I think I really need.
> Just curious more than anything if anybody had seen similar behaviour
> with the keyboard.

hmm, check hotplug blacklist.

> I rsynced slackware at work today, so this weekend while I'm screwing
> around I may see what usb commands I can put in rc.local or whatever to
> rescan the bus.
> The keyboard is kind of a PITA to investigate because there's a reboot
> each time I try something (gotta have something to type on!), and my
> old machine takes a while just to get through BIOS.

well, one more thing... bios update. and reconfiguring it again. watch for
options like usb keyboard and stuff.

Hmmmm...well, the mobo is pretty old (supermicro P6dgu, (c) 1997 or
so), so maybe there's firmware available.

Thing is, I'm using the USB controller in a PCI board for everything
USB (the board only had 1.1).

So, I'm guessing that the keyboard/mouse scan takes place in the kernel
BEFORE the USB controller is up & running.

Now, I'm thinking I'll have to either re-enable the USB1.1 stuff on the
mobo or build the USB HID stuff as modules & modprobe them after
everything else is up.

I'll follow up later if any of this works.

thrugoodmarshall@hotmail.com wrote:
> Hmmmm...well, the mobo is pretty old (supermicro P6dgu, (c) 1997 or
> so), so maybe there's firmware available.
>
> Thing is, I'm using the USB controller in a PCI board for everything
> USB (the board only had 1.1).
>
> So, I'm guessing that the keyboard/mouse scan takes place in the kernel
> BEFORE the USB controller is up & running.

Keyboard? Maybe. Mouse? No; it just needs to be ready by the time X
starts.

> Now, I'm thinking I'll have to either re-enable the USB1.1 stuff on the
> mobo or ...

This would be a good idea. HID devices are low bandwidth; save the
high-speed slots for something which can use them.

Daniel