Date: Wed, 26 Nov 1997 04:45:52 -0800 (PST) From: dan@math.berkeley.edu (Dan Strick) To: freebsd-hardware@FreeBSD.ORG Cc: dan@math.berkeley.edu, kent@tfd.com Subject: Re: large IDE disks Message-ID: <199711261245.EAA15871@math.berkeley.edu>
next in thread | raw e-mail | index | archive | help
> > I'm putting a 2.2 GB IDE disk in my laptop. The problem is the
> > bad sector information. With the straightforward approach, you
> > can only store 126 bad sectors. A lot more than that were found.
>
> A friend (Terry Carroll, Munich (still not on the net)
> told me he used fdisk to carve his disc into 4 slices first,
> then he ran bad144 * 4.
> Ive no idea if this is valid or not,
> no doubt other wiser people will comment :-)
My attitude toward IDE drive maintenance may be seriously outdated,
but here it is anyway: ("no doubt other wiser people will comment")
IDE disk systems simulate an ancient IBM PC/AT ST506 disk controller
with a few semi-standard extensions required to support modern disk
drives. The logic that emulates the old controller is almost entirely
embedded in the IDE disk drives. The behavior of an IDE drive when
emulating a special controller function (such as formatting) is less
well standardized. Since modern drives and modern PCs have modern
performance characteristics, a literal interpretation of old drive
formatting commands may be inappropriate.
Vendors of IDE drives used to provide special drive maintenance
programs that invoked proprietary disk drive features. If you
attempted to do a low level reformatting of your drive using the
standard DOS utilities, you might well end up with an inappropriately
formatted disk (e.g. with sector interleaving). Reformatting an IDE
disk was a bad idea unless you knew that the reformatting program
was appropriate for that specific IDE disk drive.
Vendors no longer provide special drive maintenance programs for
IDE drives, perhaps because the DOS marketplace never really
understood what it called "low level" formatting. I think they
expect the end user to leave the "low level" formatting alone.
If bad sectors develop, then the end user reruns the DOS FORMAT
program or some other utility to remove the bad sectors from the
active file system. If the "low level" format becomes really
degraded, then the end user thinks the drive is busted and buys
a new drive.
What has this got to do with "bad144"? BAD144 is an ancient DEC
standard for bad sector forwarding. It is totally inappropriate
for modern drives which can have thousands of bad spots that are
made invisible to unsophisticated disk driver programs by other
more elegant means. I don't see anything on the bad144 man page
that suggests it has been kept technologically up-to-date.
I doubt that you are really expected to use it.
Modern IDE drives may be smart enough to ignore the details
of a low level drive command and do the "right thing" instead
(including surface analysis) when you tickle them with an
obsolete format command. Perhaps the thing to do is use the
PC BIOS disk formatting feature and test the drive carefully
afterwards (perhaps with the bad144 program). If the drive
still seems to have bad spots or has developed antisocial
behavior (like bad performance), buy a new one. Better yet,
buy a SCSI card.
Dan Strick
dan@math.berkeley.edu
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199711261245.EAA15871>