Skip site navigation (1)Skip section navigation (2)
Date:      Sun, 23 Apr 2006 18:30:49 +0200
From:      Jean-Yves Lefort <jylefort@FreeBSD.org>
To:        "Poul-Henning Kamp" <phk@phk.freebsd.dk>
Cc:        freebsd-geom@FreeBSD.org
Subject:   Re: multiple of sector size I/O restriction
Message-ID:  <20060423183049.0e90ee1e.jylefort@FreeBSD.org>
In-Reply-To: <889.1145808650@critter.freebsd.dk>
References:  <20060423175605.35a26523.jylefort@FreeBSD.org> <889.1145808650@critter.freebsd.dk>

next in thread | previous in thread | raw e-mail | index | archive | help
--Signature=_Sun__23_Apr_2006_18_30_49_+0200_FQ_43J5mh1sDLPJD
Content-Type: text/plain; charset=US-ASCII
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

On Sun, 23 Apr 2006 18:10:50 +0200
"Poul-Henning Kamp" <phk@phk.freebsd.dk> wrote:

> In message <20060423175605.35a26523.jylefort@FreeBSD.org>, Jean-Yves Lefo=
rt wri
> tes:
>=20
> >> >Which causes this:
> >> >
> >> >	# dd if=3D3D/dev/ad1 of=3D3Dtest bs=3D3D512 count=3D3D1
> >> >	1+0 records in
> >> >	1+0 records out
> >> >	512 bytes transferred in 1.889527 secs (271 bytes/sec)
> >> >
> >> >	# dd if=3D3D/dev/ad1 of=3D3Dtest bs=3D3D513 count=3D3D1
> >> >	dd: /dev/ad1: Invalid argument
> >> >	0+0 records in
> >> >	0+0 records out
> >> >	0 bytes transferred in 0.000254 secs (0 bytes/sec)
> >> >
> >> >Why this restriction?
> >>=20
> >> That is how disks work.
> >
> >I should reformulate my question.
> >
> >I've found some linux code which reads from a raw cd device in
> >non-multiples of the sector size; I imagine that this code works on
> >linux. This leads me to think that it is possible to overcome the
> >hardware limitation by software means. Am I right?
>=20
> Yes, the linux kernel does not give access to the raw disks, you
> always have to pass though their buffer cache.
>=20
> FreeBSD does never give you access to the buffer cache but always
> sends you directly to the disk.
>=20
> The userland application will have to read an integral number of
> sectors at their physical boundaries on FreeBSD.

Do you think it would be feasible (from a performance and/or
complexity standpoint) to add the necessary magic to GEOM so that the
restriction is lifted?

--=20
Jean-Yves Lefort

jylefort@FreeBSD.org
http://lefort.be.eu.org/

--Signature=_Sun__23_Apr_2006_18_30_49_+0200_FQ_43J5mh1sDLPJD
Content-Type: application/pgp-signature

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.2.2 (FreeBSD)

iD8DBQFES6u5yzD7UaO4AGoRAqNyAJ0YY3da1eOaqUYUgbzxq0GxRzHHNgCfdOUb
0ttSevgbfbCqufsLvet6low=
=By/K
-----END PGP SIGNATURE-----

--Signature=_Sun__23_Apr_2006_18_30_49_+0200_FQ_43J5mh1sDLPJD--




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20060423183049.0e90ee1e.jylefort>