> Hello,
>
> >>> Once I compiled that into the kernel, the keyboard works. Apparently
> >>> the kernel does not pick up the keyboard until it is completely
> >>> loaded
> >>> and finished autoconf...?
> >> That sounds familiar - I'm pretty sure someone else(?) mentioned a
> >> problem like this a while ago. As far as I know the kernel doesn't
> >> need
> >> any userspace help to set up USB keyboards ( it would be pretty silly
> >> if it would... ) so I'd consider this a bug and therefore a candidate
> >> for send-pr. I guess there's already a PR about this but better have a
> >> look.
> >
> > kern/28582
> >
> > I was told to man usb and use a flag in the kernel config file. When I
> > brought up that this is not an ideal solution for machines with no
> > legacy
> > ports like my current i386 machine, I never received a reply. It's
> > possible that this is the same issue.
>
> Interesting - for some reason I assumed this was macppc-specific.
> Probably because I used a Sun Type 6 USB keyboard with NetBSD/i386 for
> ages, but only with 1.6.x. On the other hand - I never encountered the
> root fs prompt.
> Hmm, IPL_USB is the same as IPL_BIO, the PS/2 keyboard is on IPL_TTY
> which is considerably higher.


Lennart sent me this WRT the PR a while back:

"From the usb(4) man page:

The flags argument to the usb device affects the order in which the
device detection happens during cold boot.  Normally, only the USB host
controller and the usb device are detected during the autoconfiguration
when the machine is booted.  The rest of the devices are detected once
the system becomes functional and the kernel thread for the usb device is
started.  Sometimes it is desirable to have a device detected early in
the boot process, e.g., the console keyboard.  To achieve this use a
flags value of 1."