From owner-freebsd-scsi  Fri Aug 22 08:30:55 1997
Return-Path: <owner-freebsd-scsi>
Received: (from root@localhost)
          by hub.freebsd.org (8.8.5/8.8.5) id IAA14120
          for freebsd-scsi-outgoing; Fri, 22 Aug 1997 08:30:55 -0700 (PDT)
Received: from godzilla.zeta.org.au (godzilla.zeta.org.au [203.2.228.19])
          by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id IAA14114
          for <freebsd-scsi@FreeBSD.ORG>; Fri, 22 Aug 1997 08:30:50 -0700 (PDT)
Received: (from bde@localhost) by godzilla.zeta.org.au (8.8.5/8.6.9) id BAA30318; Sat, 23 Aug 1997 01:28:25 +1000
Date: Sat, 23 Aug 1997 01:28:25 +1000
From: Bruce Evans <bde@zeta.org.au>
Message-Id: <199708221528.BAA30318@godzilla.zeta.org.au>
To: freebsd-scsi@FreeBSD.ORG, j@uriah.heep.sax.de
Subject: Re: newfs on Fujitsu R640 (2k sector media)
Cc: andrew@zeta.org.au
Sender: owner-freebsd-scsi@FreeBSD.ORG
X-Loop: FreeBSD.org
Precedence: bulk

>> Won't this be foiled when newfs attempts to write the last sector, to
>> verify that it can, and uses a 512-byte write?

It should write in multiples of the fs block size or maybe the fragment
size.  If not, (ab)using the buffered device might work.

>Newfs doesn't write to a last sector to verify something.  That's not
>a PC BIOS...  Newfs just creates the required filesystem structures
>only.

Nope, UTSL, newfs/mkfs.c has full support for PC BIOSes ;-):
	/*
	 * Validate the given file system size.
	 * Verify that its last block can actually be accessed.
	 */
	if (fssize <= 0)
		printf("preposterous size %d\n", fssize), exit(13);
	wtfs(fssize - (realsectorsize / DEV_BSIZE), realsectorsize,
		 (char *)&sblock);

Bruce