Re: initrd and booting SCSI
On Thu, 10 Jun 1999, Gareth wrote:
> G'day Just a quick point me in the right direction question.
>
> I am trying to get my system to boot from the SCSI disk and pass some
> parameters to the kernel (mem=128M is one of them)
> however I cannot get it to boot from the SCSI disk! (it hangs at LI)
I had this problem exactly (but with additional parameters, and AFAIK your
syntax is wrong for mem=, or is not what I found in the doco).
The LILO User's Guide says it's caused by a geometry mismatch. In my case
it was not that, and I fixed it by setting boot=/dev/sda instead of
putting the boot sector on a primary Linux partition. (Someone recently
wondered out loud on this list why anyone would put LILO on the MBR...)
Probably the cause, if not a geometry mismatch, is corruption of the MBR.
You might try lilo -u or lilo -U to uninstall LILO, and start again. Or
restore it from your backup if lilo refuses these requests. Or boot
another operating system -- e.g. use Microsoft's FDISK with the /MBR
option. You should then be able to put LILO on the boot sector of a
primary Linux file system, e.g. boot=/dev/sda1. Or if you are nervous
about a dd onto /dev/sda, just install LILO on the MBR, i.e.
boot=/dev/sda.
My take on the mem= parameter is that I need
append="mem=0x10000000 ..."
for 256 MB of RAM. Maybe decimal megabyte parameters are okay; I'm just
going by what I read.
> I have checked and double checked the LILO stuff and disk partions.
>
> I have been told I shoud use a 2 stage bootup with initrd to load SCSI
> stuff to boot from the SCSI disk. (I am currently using a bootfloppy)
Okay, in full I have
append="mem=0x10000000 reboot=warm aha152x=0x340,10,7,1"
but that is only because I have a second SCSI adapter. I think the first
should be probed. I.e. you should not need a 2 stage bootup with initrd
to load the SCSI driver to boot from the SCSI disk. IMHO.
Reply to: