>openfirmware is supposed to be able to read off a "raw partition", and
>(I think) you're supposed to be able to run "installboot" on your
>drive in some way such that booting from openfirmware will work. I
>haven't gotten it to work, but I was having other problems anyway.

	Unfortunately I think I might have put NetBSD on too high of a
partition number.  I seem to recall hearing something about OFW not
being able to boot past partition 8 or something like that.  My NetBSD
partition is on 11, and I just get a bad partition error from an OFW
routine if I try to boot hd:11,0 or anything similar.  Already did the
installboot trick, I assume I did it right 'cuz it didn't complain..  =)

>If I boot off IDE, pciide fails, analagously to how SCSI fails if I
>boot off that, or ethernet fails if I don't boot off it (or use the
>patch Tsubai sent).

	My problem seems slightly different.  It seems if I load
ofwboot.elf or the kernel off the HD (by loading ofwboot.elf from my MacOS
partition, or loading the kernel directly from the MacOS partition, or
doing "boot enet:,ofwboot.elf hd:11"  The kernel loads, but then fails
with the pciide0:0:0: lost interrupt.  If I boot the kernel directly over
ethernet or "boot enet:,ofwboot.elf enet" then I can boot kernels with either 
bm0 or wd0a as the boot device with no trouble at all.  However, if before
I do the boot command I do a "dir hd:5,\", then I cannot boot with wd0a as
the root even if I load ofwboot.elf and the kernel over ethernet.  Similarly
if I boot with bm0 as the root device I cannot read wd0's disklabel, I just
get the lost interrupt messages.  So it seems from what I see that NetBSD
can't properly use pciide if OFW has touched it first?  I checked the
.properties on the 0646's OFW entry, and nothing changes from before the
dir to after it, so no help there..  =)  FWIW using stock /sys tree from
the 5/9 tarball, no patches applied.

>I have the same keyboard, but a different mouse, and my mouse has
>never worked when plugged into the keyboard. (My system is 300MHz, so
>maybe this is all just race conditions like someone said.)

	Mine is 350MHz FWIW.  The mouse definitely properly detected attached
to the keyboard with earlier kernels, never actually tested it in any
programs though, so I can't say if it actually worked.

>> However, if I roll the wheel, the X server immediately dies.
>
>See PR 7476

	Thanks.  Will have to grab the xsrc and try the patch out.



                                                          Will Ferry

-----------------
woferry@iname.com