Re: [vox-tech] Partition question
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [vox-tech] Partition question
On Thu, 12 Aug 2004, Margo Schulter wrote:
> Currently the system I'm using has a primary MS-DOS 6.22
> partition at hda1 (DOS C:), 256M, and an extended DOS partition
> at hda2 with three logical drives (hda5, hda6, hda7), or DOS
> D:, E:, and F:, respectively 256M, 512M, and 1024M. On hda5 is
> ZipSlack, a fine UMSDOS version of Slackware Linux 10.0. The
> disk is 37G (or 40G decimal), so there's plenty room for a
> native Linux installation of Slackware plus an installation of
> My original idea, given the limit of three regular primary
> partitions plus an extended partition, was to use parted to grow
> the extended partition to fill almost the whole rest of the
> disk, allowing 256M at the end for two primary partitions to be
> used for /boot directories for Slackware and Gentoo, or about
> 128M each. This might make a bit of allowance for kernel bloat
Kernel bloat is one thing, but far more valuable is the ability to have
multiple kernels to boot from in case something is broken.
> In short, however, I learned that in practice, resizing the
> extended partition with parted resulted in a fine lba partition
> which a reboot revealed that DOS wouldn't recognize -- nor the
> three logical drives D:, E:, and F:. While I had made some
> backups with cdrecord, simply running parted again and resizing
> hda2 to its original size nicely restored recognition of those
> Now I'm considering some alternatives, and seeking advice here
> on some of the choices:
> (1) A crude solution: leave the extended partition
> as is and simply use fdisk to add two more
> primary partitions, hda3 and hda4, as swap and
> root for a single Linux distro.
Last time I tried something like this, fdisk was unable to insert
partitions. I would be surprised if this limitation had changed.
> (2) A more refined solution: why not replace MS-DOS
> with a more sophisticated version of DOS such as
> FreeDOS which might not only recognize an extended
> lba partition shared with Linux, but also save
> some DOS disk space by overcoming that cluster
> size disadvantage of FAT-16.
Not having any desire to maintain ties with DOS or DOS compatibility, I
have no idea if such is possible. However, I would expect any such
implementation to run into compatibility problems with DOS... it is,
after all, famous for its limited memory model, so mapping larger and
larger clusters into that memory is going to require some tradeoff in
available ram for programs or use of very large himem blocks.
> (3) Another solution, if practical: use parted to
> resize hda2 while staying within the 8G limit,
> enlarging it by just enough to support two
> Linux swap partitions, one for Slackware and
> one for Gentoo, say 768M each. Then I could add
> two primary partitions taking up the rest of
> the disk, one as a root partition for each
hda1 = C:
hda2 = ? two swap partitions? er, just one swap and any running linux can
hda3 = Slackware root?
hda4 = Gentoo
no slots left for logical partition at this point
I am puzzled, though... this eliminates drives D:, E:, and F:... in what
way is this superior to having a logical partition that DOS cannot access?
hda1 = C:
hda2 = swap
hda3 = logical
hda5 = Slackware root
hda6 = Gentoo root
hda7 = /home or /home/shared
There is even room in here for a partition before the swap partition that
could serve as a drive D: for DOS.
> While the third solution is rather simple -- if it's feasible --
> in comparison to more intricate partitioning schemes for Linux
> distros, it might suffice for a single-user system.
True... though having a common /home or /home/shared might be convenient.
(The former is easy, but sharing configfiles may not work well in some
cases, so keeping files in a common place no matter which distro you use
might work conveniently with a symlink from /home/mschulter/shared to
Personally, I avoid multi-booting operating systems... virtual machines
(e.g. VMWare) or multiple physical machines (with or without KVM switches)
are simply more productive approaches.
Jeff Newmiller The ..... ..... Go Live...
DCN:<firstname.lastname@example.org> Basics: ##.#. ##.#. Live Go...
Live: OO#.. Dead: OO#.. Playing
Research Engineer (Solar/Batteries O.O#. #.O#. with
/Software/Embedded Controllers) .OO#. .OO#. rocks...2k
vox-tech mailing list