From owner-freebsd-hackers  Fri May 11 15:13: 2 2001
Delivered-To: freebsd-hackers@freebsd.org
Received: from bubbette.cs.rice.edu (bubbette.cs.rice.edu [128.42.3.109])
	by hub.freebsd.org (Postfix) with ESMTP id 0417537B423
	for <freebsd-hackers@FreeBSD.org>; Fri, 11 May 2001 15:13:00 -0700 (PDT)
	(envelope-from ssiyer@bubbette.cs.rice.edu)
Received: (from ssiyer@localhost)
	by bubbette.cs.rice.edu (8.11.3/8.9.3) id f4BMAXM04978
	for freebsd-hackers@FreeBSD.org; Fri, 11 May 2001 17:10:33 -0500 (CDT)
	(envelope-from ssiyer)
Date: Fri, 11 May 2001 17:10:33 -0500
From: Sitaram Iyer <ssiyer@cs.rice.edu>
To: freebsd-hackers@FreeBSD.org
Subject: possible bug in/around dscheck()?
Message-ID: <20010511171033.A4960@cs.rice.edu>
Reply-To: Sitaram Iyer <ssiyer@cs.rice.edu>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
X-Mailer: Mutt 1.0.1i
X-url: http://www.cs.rice.edu/~ssiyer/
Sender: owner-freebsd-hackers@FreeBSD.ORG
Precedence: bulk
X-Loop: FreeBSD.ORG

Hi,

I was mirroring some disks (happily ignoring badblocks), using
  rsh source_machine dd if=/dev/ad0 bs=1MB | dd of=/dev/ad0 bs=1MB
with FreeBSD-4.3-release running both machines.

This sometimes (disastrously) aborts when a partition boundary is crossed.
Sure enough, dscheck() invoked from diskstrategy() complains about b_bcount
not on a sector boundary (happens in one of three locations within dscheck).

I think this condition should not be checked when dd'ing entire disks.
Incidentally, I think the 2.2.6-release kernel does the right thing.

On a related note, is there any reason why dd should abort after 8GB?
I'm not sure this always happens, but I've definitely noticed it once.

PS - please Cc any replies to me. Thanks.
Sitaram

To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-hackers" in the body of the message