Return to BSD News archive
Path: sserve!newshost.anu.edu.au!munnari.oz.au!news.Hawaii.Edu!ames!agate!spool.mu.edu!howland.reston.ans.net!gatech!prism!gt8134b
From: gt8134b@prism.gatech.EDU (Howlin' Bob)
Newsgroups: comp.os.386bsd.development
Subject: Re: Hard disk geometry translation (was V86 mode ...)
Message-ID: <109172@hydra.gatech.EDU>
Date: 17 Aug 93 20:42:35 GMT
References: <107725@hydra.gatech.EDU> <1993Aug9.224939.19834@fcom.cc.utah.edu> <24cc1hINNo8@kralizec.zeta.org.au> <CBo9C6.9ED@sugar.neosoft.com> <24gt3e$gg7@klaava.Helsinki.FI>
Organization: Georgia Institute of Technology
Lines: 45
In <24gt3e$gg7@klaava.Helsinki.FI> torvalds@klaava.Helsinki.FI (Linus Torvalds) writes:
>In article <CBo9C6.9ED@sugar.neosoft.com> peter@NeoSoft.com (Peter da Silva) writes:
>>In article <24cc1hINNo8@kralizec.zeta.org.au> bde@kralizec.zeta.org.au (Bruce Evans) writes:
>>> Linux just uses the translation given by the BIOS. The only problem
>>> with this is is that the translated geometry may not be appropriate
>>> for file system layout optimizations.
>>
>>Worse, you could lose disk space. For example, the translation given by the
>>Compaq BIOS eats between 30 and 60 MB off the drives I use.
You must use very large drives. Is this loss because the cylinder count
cannot top 1024? In this case, there's no easy way to have a DOS partition
co-exist with your other OS. Linux lets you specify the HD geometry
on the LILO command line, or it can be compiled in. I don't know anything
about the IDE spec, but I suppose you can request geometry information?
If so, then it might be nice to support this in Linux so that very large
drives can be used with no loss of space *if* you don't need to co-exist
with DOS.
> knowing about the real geometry) is mostly a load of bull-sh*t. It
> may have made sense 10-20 years ago, but I doubt the FFS disk
> geometry optimizations are really worth it these days with
> controllers that do sector mapping etc (the BSD 4kB blocks are
Again, I have to wonder about Linux's ext2fs. ext2fs seems fast
to me, but perhaps it is performing some cylinder-group "optimizations"
that actually negatively impact performance?
> no need to worry unnecessarily on a software level. Not to mention
> the fact that it's ugly in the extreme that the filesystem should
> need to know anything about the geometry in the first place.
Really? I'm not so sure. There is no rule set in stone that says that
any layer above the device driver should view all block devices as
linear collections of blocks. Why shouldn't we abstract all block
devices to three dimensional spaces? The linear special case would
still be there, and specific knowledge about a disk's geometry could
be helpful.
--
Robert Sanders
Georgia Institute of Technology, Atlanta Georgia, 30332
uucp: ...!{decvax,hplabs,ncar,purdue,rutgers}!gatech!prism!gt8134b
Internet: gt8134b@prism.gatech.edu