On 15 Sep 2001, Georges Heinesch wrote:
> Quoting Frederick Bruckman (10-Sep-01 02:53:51):
> >> On Sun, Sep 09, 2001 at 06:21:48PM +0100, Georges Heinesch wrote:
> >> >
> >> > However when booting up from WinNT, PartitionInfo (part of
> >> > Powerquest's PartitionMagic) reported error:
> >> >
> >> > The EndCylinder is past the end of the drive

> > I let PartitionInfo "fix" my CHS, and though now the CHS information
> > makes no sense -- both of the partitions that start after 2G have
> > the exact same CHS -- it doesn't seem to hurt anything on my modern
> > laptop. (I use the NT boot selector on Windows 2000.)

> Sorry Frederick, but I didn't get what you said here. Could you
> provide some more details about what I should try and why I should do
> that. With this kind of manipulations, I usually only have one shot.

At first, I didn't install a boot selector. I intended to change the
active partition via "fdisk -a" on the NetBSD side, and PartitionMagic
on the W2K side. It's a long story -- FIPS refused to partition the
disk, evidently because of the FORMS virus taking up a high numbered
block, I went for the *big* *hammer* (zeroing out the virus with a
disk editor, but without properly restoring the Windows MBR), and it
took me a couple more days with a disk editor to figure out what
happened and fix it. I finally got to the point where the NetBSD boot
blocks would boot NetBSD; after "fdisk -a", W2K booted fine, but then
PartitionMagic said "blah blah blah, do you want to fix it", so I
clicked yes (I'd already backed up the the first 63 sectors with "dd"),
and that worked.

Here's the output of "fdisk" (now):

NetBSD disklabel disk geometry:
cylinders: 16383 heads: 16 sectors/track: 63 (1008 sectors/cylinder)

BIOS disk geometry:
cylinders: 1022 heads: 240 sectors/track: 63 (15120 sectors/cylinder)

Partition table:
0: sysid 12 (Primary DOS with 32 bit FAT - LBA)
    start 63, size 15709617 (7670 MB), flag 0x80
	beg: cylinder    0, head   1, sector  1
	end: cylinder 1023, head 239, sector 63
1: sysid 15 (Ext. partition - LBA)
    start 15709680, size 408240 (199 MB), flag 0x0
	beg: cylinder 1023, head   0, sector  1
	end: cylinder 1023, head 239, sector 63
2: sysid 169 (NetBSD)
    start 16117920, size 7454160 (3639 MB), flag 0x0
	beg: cylinder 1023, head   0, sector  1
	end: cylinder 1023, head 239, sector 63
3: <UNUSED>

Notice that the CHS is simply wrong, especially for partitions 1 & 2.
I can only speculate that pegging the cylinder at 1023 let's you at
least boot the 0 partition on certain particular (very old) BIOS'.
It's obviously not doing anything constructive.

If you already have a boot selector set up (either Window's or
NetBSD's), and you don't enjoy mucking with the partition table with a
disk editor or "dd", I suggest you just leave it alone.

Frederick