Johnny Billquist wrote:
Anders Magnusson wrote:
Johnny Billquist wrote:
Brad Parker wrote:
I seem to have my foot stuck in a gopher hole. I'm confused about a
few things.
- the hp.c driver in boot/ ignores the partition argument (making it
impossible
to boot a miniroot I think). fortunately the other drivers in boot/
(ra & mfm)
do honor the partition arg.
- if one does get a miniroot kernel to boot, the partition info is not
communicated to the new kernel. so it thinks you booted from
partition 0
('a') and uses that as root. rather confusing.
am I correct about that?
4.3bsd sends the boot partition in r10. It seems like using one of the
"rpb" registers (like r4) might be a better plan. Is there any reason
not to send the boot partition info in r4? (i.e. rpb_bootr4)
other platforms (1-2 anyway) seem to use MAKEBOOTDEV()
Not sure how much we care, but VMB have reserved meanings for R0-R5,
which we would break by this idea.
It shouldn't matter here, this code is executed long after VMB is left
:-)
Hey. If I need to set R4 to a specific value to make VMB happy, I can't
really set it to something else to make boot happy, can I?
Eh, of course :-/
Hm, I don't think Ultrix can give partition number, can it? If it can, we
should use the same way, otherwise it would be backward-compatible to
encode it in the upper word of R3 maybe? Or use R4, the comment in your
boot file says that it is only used if R5 bit 3 is set.