Disk partitioning (was: Daemon's Advocate article)

Garance A Drosihn drosih at rpi.edu
Wed Mar 3 14:10:16 PST 2004


At 1:32 PM -0800 3/3/04, Brooks Davis wrote:
On Wed, Mar 03, 2004, Garance A Drosihn wrote:
 > At 12:34 PM -0800 3/3/04, Matthew Dillon wrote:
 > >
 > >    There is actually room for 16, but the dev minor number
 > >    does not have enough bits to specify more then 8.
 >
 > Could we do something slimy, like have another device
 > major-number which means "the second 8-partitions of a disk"?
What's the urgency?  You've got four slices without using
extended one which gives 32 partitions easily.  If you need
more then that on x86,
a) I have already run into situations where I wanted more
   partitions on an i386 disk than FreeBSD/i386 supports.
b) I also run FreeBSD/sparc64, where we don't have the
   slices.  I run out of partitions very rapidly on that.
your time would probalby be better spent figuring out the issues
involved in booting from a GPT partition and cleaning up the code
in sbin/gpt on FreeBSD.
If you mean me, specifically, then my time is better spent on my
own projects.  I know nothing about disk and filesystem-level code,
but I do sometimes have projects where I wish I had more partitions.
(well, actually I do have a snapshot-related idea that I hope to
tackle soon, but let's ignore that for the moment...).
FreeBSD already has problems with too many people tripping over
each other in the low-level code.  It doesn't need me jumping in
and adding to the confusion...  Much better for me to work at the
levels where I have at least some idea of what's going on.
That would give you nearly 16K partitions per disk which ought to
be good enough for almost anyone.  We're going to have to go that
way soon anyway since >2TB disks don't really work with MBR
partitions.
If you meant "your time" as in "whoever has the time to implement
this", then certainly I'm all for this better solution!  :-)  Even
doubling the partitions wouldn't really solve my need for partitions,
but it would help.  Since I don't know much of what happens at the
lower-levels, I really don't know how much work would be needed for
any of these ideas.  I'd be happier with 16 partitions than 8, and
much happier with any option for 64 partitions.  16K partitions
certainly sounds adequate to me!    :-)
--
Garance Alistair Drosehn            =   gad at xxxxxxxxxxxxxxxxxxxx
Senior Systems Programmer           or  gad at xxxxxxxxxxx
Rensselaer Polytechnic Institute    or  drosih at xxxxxxx




More information about the Kernel mailing list