[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [cobalt-users] Raq4 - Kernel 2.4 Upgrade Woes [UPDATE]



Josh Trutwin wrote:

I installed Lilo into the MBR so I could boot up the Debian kernel
easilly, the instructions said that Cobalt ignores the MBR so this would
only serve as a convenience for booting while the disk was in the PC.  I
guess I could try another install and not add Lilo, just use boot
floppies, do you think that would make a difference?


It shouldn't.  The ROM doesn't have any use for lilo or the MBR,
and should ignore both.  It loads the in-rom kernel to memory,
jumps to it and uses its IDE and filesystem drivers (ext2 only
for 2.3.40) to mount the root device and read an ELF kernel
image from /boot/vmlinux.gz.

Then the kernel passes control back to the ROM and it will take
care of uncompressing the new kernel and jumping to it.
This is commonly referred to as a "two kernel monty"  :)

Its got a number of advantages and disadvantages to the normal
PC-style of booting, but the main reason for Cobalt was that
lilo and such use legacy BIOS calls to do their dirty work...

Anyway, I've just been catching up on this thread and I noticed
that over the weekend you said you tried again and used
"set_boot_dev hda1" and "set_params root=/dev/hda1", but the
serial console log from the first message shows that hdc1 is
the installed drive..  Was this a typo?

-duncan