From owner-freebsd-current  Sat Apr 29 15:46:17 1995
Return-Path: current-owner
Received: (from majordom@localhost)
          by freefall.cdrom.com (8.6.10/8.6.6) id PAA21971
          for current-outgoing; Sat, 29 Apr 1995 15:46:17 -0700
Received: from pelican.com (pelican.com [134.24.4.62])
          by freefall.cdrom.com (8.6.10/8.6.6) with SMTP id PAA21959
          for <current@freebsd.org>; Sat, 29 Apr 1995 15:46:11 -0700
Received: by pelican.com (Smail3.1.28.1 #5)
	id m0s5LH1-000K0jC; Sat, 29 Apr 95 15:46 WET DST
Message-Id: <m0s5LH1-000K0jC@pelican.com>
From: pete@pelican.com (Pete Carah)
Subject: fdisk+disklabel, etc
To: current@FreeBSD.org
Date: Sat, 29 Apr 1995 15:46:02 -0700 (PDT)
X-Mailer: ELM [version 2.4 PL23]
Content-Type: text
Content-Length: 1655      
Sender: current-owner@FreeBSD.org
Precedence: bulk

I'm having a bear of a time getting a disk labelled - whatever I do,
disklabel won't label the disk.  This is a quantum 540 scsi with
1057758 sectors.  This doesn't factor into anything which will fit
into dos (or bios)'s idea of cyl-hd-record.  

To use these scsi drives, there has to be a way of getting fdisk and
disklabel to buy the drive size without paying any attention *at all*
to the slice table's cyl-hd-rec data.  I saw some description fly
by but it won't work for this drive no matter what I try (even go down
to the next "cyl" value using xxx-64-32 or xxx-32-32 or xxx-16-63)
I don't want to lose a meg at the end if possible, and it doesn't work
even with that loss.

We need a decent way to 
1. Get the slice manager to buy ONLY beginning/size and totally ignore
   the bios data.

2. Fake out the size so it can be bigger than cyl*hd*sec as long as the
   real drive size isn't exceeded.

The latter is supposed to work if #cyl is 1023 but this drive doesn't
conveniently partition so that works.

I also have to allocate a 1-cyl dos partition for all time to get the
slice table accepted at boot time, no matter what - if I delete it and
write the bsd slice as the whole disk then things fail when I try to
disklabel.  (actually, now, the fail when I try to disklabel anyhow,
independent of the fbsd size.  This is with disklabel aimed at:
sd1
/dev/rsd1
/dev/rsd1c
/dev/rsd1s4
/dev/rsd1s4c.)

Mostly it complains that the a partition extends past the end of unit,
even though the beg is 0 and size matches that in the slice table.
The last complained about no such file.

We need a GOOD writeup on this; I know it can be done.

-- Pete