There are several things about the IBM PC architecture that
are essentially broken; we just have to do the best that we can. This
is part of what has made Linux's development interesting to watch.
>> There is an ordered set <1, 2, ...> of controllers seen by the
>> BIOS. When LILO tell the BIOS to read the kernel image, it has to
>> tell the BIOS which controller (host adaptor), which disk (LUN), and
>> which sectors to read.
I may have overstated my case; I haven't actually verified
that lilo really can boot off a second controller. "man lilo.conf"
doesn't say, and I don't have a copy of the lilo sources handy at the
moment. At the minimum, the problem consists of identifying which
controller is first, hence bootable.
>LILO works most of the time because most of users only have IDE only or
>IDE and a single SCSI adapter.
>Even assuming IDE then SCSI order is broken since:
>- Some BIOSes allows to boot from SCSI when IDE + SCSI is present.
>- What about some other subtleties of set-up programs as:
> * Setting devices that boots and those that does'nt.
> * Reverse scanning SCSI devices.
> * Booting from CD/ROM.
>
>I am not a LILO expert, but probably LILO allows to force the BIOS device
>number also. Does it?
Hmmm... not mentioned in the man page. Also, we're straying
from the charter of the kernel newsgroup a bit. It would be
interesting to have a parameter known to lilo (and the kernel?) that
could be used to override the device order if you are planning to,
say, reorder your SCSI drives, including your boot drive.
Craig Milo Rogers
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu