FYI: Multiple OSes, partition tables, lossage

Benjamin Scott bscott at ntisys.com
Sat Jan 1 15:18:01 EST 2005


  I've been working on getting a couple of legacy OSes working in harmony
with Linux on my main home PC.  I've also been using Fedora Core 3, which
uses the 2.6 kernel.  This has lead to my discovery that the whole partition
table thing has gotten even worse.  Maybe everybody else knows this and I
just didn't get the memo, but if not, the following may prove useful to
others.  Share and enjoy!

  Definition:

  A "legacy OS", in this discussion, is a brain-damaged OS that does not
provide any clue as to why it does things they way it does them, nor does it
provide a way for you to fix it.  Legacy OSes include the latest stuff from
a certain large software company currently headquartered in the North-West
USA.

  Background:

  Certain legacy OSes seem to think that hard disks should have a
multi-dimensional geometry, rather then a simple linear geometry.  This is
called the "CHS geometry" (cylinder, head, sector).  Modern OSes like Linux
thankfully do not need this garbage, but some of us, myself included, are
using legacy OSes, too.  Since any hard disk made within the past 15 or so
years doesn't use CHS, the BIOS invents one to make legacy OSes happy.

  If one only used a single OS, then this did not matter.  When you combine
a modern OS like Linux (which does not use CHS) and a legacy OS which uses
CHS, however, you have to make sure everybody agrees as to what the invented
geometry is.  In this past, the Linux kernel provided some mechanisms for 
this.  They were not perfect, but they worked most of the time.

  New trouble:

  With Linux 2.6, certain powers decided that legacy OSes were Somebody
Else's Problem, and the CHS support in the kernel was changed and reduced.  
This has led to major lossage when dealing with legacy OSes.  Programs (like
parted) which previously depended on the kernel to provide good answers got
hosed.  This could easily render a legacy OS unbootable.

  Fixes include paying attention to the order of operations, using certain
partitioning tools for certain tasks, and/or fixing up partition table
geometry entries manually.

  It is apparent that not everyone agrees on the issues at hand.

  Things are likely to continue to change and evolve.

  References:

A thread on bug-parted about the impact of the kernel changes
http://lists.gnu.org/archive/html/bug-parted/2004-12/msg00022.html

A thread on LKML about changes in kernel support for disk geometry
http://tinyurl.com/6elmz

A message that explains how to prevent and/or fix geometry corruption
http://www.redhat.com/archives/fedora-devel-list/2004-May/msg00908.html

Large Disk HOWTO
http://www.win.tue.nl/~aeb/linux/Large-Disk.html

-- 
Ben Scott <bscott at ntisys.com>
| The opinions expressed in this message are those of the author and do  |
| not represent the views or policy of any other person or organization. |
| All information is provided without warranty of any kind.              |




More information about the gnhlug-discuss mailing list