[ltp] cloning 14G drives

[Rob Mayoff]

| I have one spanner left in the works and that is gt 8GB hdd support in
| native mode (works under VMWare and reports hda CHS as 1826/16/63 during
| boot).

That is the correct geometry for the 14G drive.
| 
| This has just become a show stopper for the next stage of what I wanted to
| do. My plan was to use VMWare Linux (yes I have that as well) to build a
| Solaris 7 x86 system to a raw partition that I have reserved (/dev/hda3 to
| be exact) so that I could use it under NT, Linux and native. Problem is
| VMWare cannot see the partitions because the kernel cannot see them so I can
| only build to a virtual disk, that would preclude a native boot of course.

Why can't the kernel see the partitions?  What are the contents of these
files:

	/proc/partitions
	/proc/ide/ide0/hda/capacity
	/proc/ide/ide0/hda/geometry

| 1. Rob this fix you posted to /usr/src/linux/drivers/block/ide-disk.c, is
| this for install time only, or will it get gt 8GB hdd support working with
| the appropriate lilo parameters if I rebuild the kernel? (wow only been
| using Linux for less than two weeks and already talking about rebuilding the
| kernel).

The patch I posted will make the kernel correctly detect the disk size.
You won't need any lilo parameters.

The thing is, without that patch I could not get the system to correctly
see the disk size, even with lilo parameters.  So basically there's no
way that I could find to use the upper 6G of the disk when I was
installing Redhat 6.1.  You have to leave it untouched (can't even have
any partitions there).  Then, after you've installed, you can patch the
kernel, and then add partitions.

The reason the patch is useless for install time is that you would have
to create your own boot image containing the patched kernel, and that's
beyond the ability of most Linux users.

| I am not sure I fully understand what you mean when you wrote: "I know this
| won't help if you're trying to install on a disk that already has partitions
| above 8G.  I had to edit my partition table by hand to work around that
| problem."

When I talked about editing my partition table by hand: I did that
because I already had partitions above 8G, so the Redhat installer
wouldn't work. I had already blown away my old root partition, so I
could not boot my old setup to run fdisk. I was quite screwed. To get
myself out of the jam, I had to copy the first sector of my disk (where
the beginning of the partition table is stored) onto a floppy, edit it
on another Linux box with a hex editor to empty out the partition table,
and then copy the sector back onto my hard drive. Then I was able to
create a new partition table. I was very careful in setting up my new
partitions, so after the install was finished and I patched my kernel, I
was able to use fdisk to recreate my old above-8G partitions and all the
data in them was still intact.

This is not a process I recommend for a self-described Linux newbie.

| 2. Would I be better off just upgrading to kernel 2.2.13 (I have downloaded
| it already and could have a go). Also downloaded XFree86 3.3.5. But if I am
| doing this would it not better to just upgrade to SuSE 6.3 which includes
| kernel 2.2.13 and XFree86 3.3.5 plus more recent PCMCIA and presumably
| StarOffice 5.1 etc.?

The stock 2.2.13 kernel does not correctly detect the 14G disk geometry.
(I know because I installed it this morning.) It is possible that SuSE
comes with a patched kernel that correctly detects the 14G drive. I know
Mandrake does, because I used to run Mandrake 6.0.

----- The Linux ThinkPad mailing list -----
The linux-thinkpad mailing list home page is at:
http://www.bm-soft.com/~bm/tp_mailing.html