On 2001-04-21, Benjamin Black wrote:
> hi everyone.
>
> a few days ago, my partition tables were fine. proof of fact: i
> installed the new kernel-image-2.4.3-686 that was uploaded a few days
> ago into my boot sector, and lilo did not complain; i am now running
> that image.
Hi. I am seeing the same behavior on two unstable branch systems that I
upgraded to kernel-image-2.4.3-k6/7. They were both running 2.4.3 kernels
that I had compiled with no problem.
> today, i was trying to set up my cd writer, so i added
> append=3D"hdd=3Dide-scsi" to lilo.conf, but now when i run lilo, i got
> this=
> :
>
> Warning: current root is not on the first disk
> Device 0x0340: Invalid partition table, 2nd entry
> 3D address: 1/0/128 (129024)
> Linear address: 1/0/2040 (2056320)
I get the same error. "lilo -P ignore" seems to work around it, in that
it installs the boot sectors and everything works. If you run lilo under
an earlier kernel (ie. not a debian 2.4.3 kernel) then lilo will stop
complaining.
However, "lilo -P fix" seems to screw things up quite badly. I tried that
on one system, and now if I go back to an earlier kernel (the
2.2.18pre21 from potato, which is one of the few I have boot disks
for) and try to run lilo, I get the same error. Also, I cannot currently
get it to boot from that lilo boot block.
> so i ran fdisk on my two drives, and this is what i see:
>
> Disk /dev/hda: 16 heads, 63 sectors, 19590 cylinders
>
> Units =3D cylinders of 1008 * 512 bytes
I suspect that if you boot another kernel, your disk geometry will look
quite different. When I boot with the 2.2.18pre21 kernel I get a much
different disk translation, and the partition sizes all work out okay. The
cylinder units were bigger, too, so there were less cylinders.
Also, in response to another poster's suggestion: if you resize the
partitions using gpart, then you ever have to go back to an older kernel,
won't you will have the same problem?
Brad
