[rescue] rescuing an ailing U10...

[der Mouse]

>> Platforms with more sophisticated firmware actually do need a
>> prescribed on-disk layout, as you mentioned.  A survey of where that
>> sort of thing breaks down by platform would be interesting.
> Well don't quote me on this (i.e. I could well be wrong) but I think
> Sun's firmware does a very similar thing to PC firmware ... just
> loads the first few sectors off a disk.

Perhaps some versions.  But the ones I'm familiar with - Sun-3, and
SPARC up to the ultrasparc - don't do that.  (I think the early
ultrasparcs - the U1 and U2 - are similar, but I'm far less familiar
with them and thus far less sure of it for them.)

They do load the first sector, true, but they treat it as a disklabel
rather than just executing it.  They then load the beginning of the
boot partition (usually partition #0 in the label); *that* is what gets
executed.

This first-stage bootloader can do anything that fits into (IIRC) 7=KB.
Normally it then loads a second-stage bootloader which has far fewer
size restrictions (because it's stored in the filesystem, canonically
/boot), and *that* is what loads the kernel.  How the first-stage
bootloader finds the second-stage bootloader varies; some walk the
filesystem to find /boot, others load a fixed set of blocks, set at
bootblock install time, with those block numbers normally set to be the
ones holding /boot's contents.

/~\ The ASCII				der Mouse
\ / Ribbon Campaign
 X  Against HTML	       mouse at rodents.montreal.qc.ca
/ \ Email!	     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B