Re: fdisk, lilo problems
> Interesting. What vintage of disk is this? The new IDE disks report
> back a heads/sectors/cylinders number that is entirely synthetic as
> they map more sectors on the outer tracks than the inner ones and they
> transparently and autonomously relocate bad blocks, and the geometry is
> deeply buried after that. It sounds like your disk is an older one,
> and your BIOS isn't helping the situation.
The disk is a Western Digital "Caviar 2450". It has the following
info on a label:
Part No 99-004137-000
CCC A1 14 MAR 94
So it looks to be just under two years since manufacture. I had to
move my office Linux system into the lab, and I'm trying to set up the
machine I inherited to replace it.
> I believe the new fdisk only cares that partitions begin on a track
> boundary for DOS compatibility, but I haven't experimented much.
Me neither. I was pretty deep inside LILO sources a year or so ago
tracking down a problem, but don't remember much. I wasn't looking
hard at this part of it anyhow.
>From the disparities between some calculations being done with 16
heads and other calculations being done with 32 heads, it looks like
there might be some question of "What's a block?". I still haven't
doped out where all the calculated numbers come from, but it looks
like the calculations might be calculating two 32-head blocks per
one 16-head block and running into trouble with a partition boundary
at the end of an odd-numbered 32-head block where there's no corresponding
16-head block boundary.