Don't think I enjoy talking to myself, now.  Have you ever
spent too long writing a clarification, that turns out to
be not quite correct?  Well, I do it all the time.  I said:

> fully awake, but I can see the need for clarification in my
> USB description, to wit:

> > directly to the OHCI hub.  With NetBSD, everything connected
> > directly to the internal OHCI hub is seen (except for a USB2
> > hub, which manages to be used successfully as USB1 with
> > FreeBSD), but nothing connected to the external USB1 hub is
> > recognized.

Ignore what I wrote.  The problem appears to be a race condition
that my particular setup reveals, where I have my root filesystem
on what should appear as sd0, and the kernel gets to the point of
boot-on-foo, root-on-bar, swap-on-baz before the USB devices that
appear as sd* get recognized.

After successfully getting the FreeBSD-based firewire code into
a kernel and booting from it, I see no problems at all.  Here my
sd0 root disk is found early enough that all the devices attached
to the USB ports get probed, just as they should be.

Perhaps that was an issue with the original firewire code too in
my case.  Or not.

So the only real issue is the external hub overcurrent problem,
though I would also like to test OHCI/umass data integrity as I
see problems there with my FreeBSD, but I'll have to figure out
a way to get my boot disk usable on USB.


Is there something, perhaps an `#if 1' in the source somewhere,
that I can tweak, that will cause the probing and attaching of
USB (and firewire) devices to be completed before the kernel
goes ahead with the root-on-whatever step?  Or am I out of luck
with my idea to use a USB drive as my root disk?

Kernel source should be up-to-date as of yesterday-ish.


thanks
barry bouwsma