Subject: Re: VOP_READ with IO_EXT versus VOP_GETEXTATTR
To: None <tech-kern@NetBSD.org>
From: Bill Studenmund <wrstuden@netbsd.org>
List: tech-kern
Date: 09/18/2006 10:45:48
--dDRMvlgZJXvWKvBx
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

On Sun, Sep 17, 2006 at 01:20:23AM +0200, Reinoud Zandijk wrote:
> Dear folks,
>=20
> i'm puzzled about the IO_EXT flag in the ioflags passed to VOP_READ and=
=20
> VOP_WRITE. It states that is an operation on extended attributes. But isn=
't =20
> VOP_GETEXTATTR defined for?

Ask Jason. As cvs blame indicates, he added both at around the same time,=
=20
nearly two years ago.

It looks like FreeBSD plans to eventually have VOP_READ() with IO_EXT get=
=20
redirected to a getattr call, however the latest code is:

	if (ap->a_ioflag & IO_EXT)
#ifdef notyet
		return (ffs_extread(vp, uio, ioflag));
#else
		panic("ffs_read+IO_EXT");
#endif

> Or should i read the IO_EXT flag as operating on sub-files? but then=20
> VOP_READDIR should have a flag too... and i can't find that one.

No, I don't think VOP_READDIR should have such a flag. If you're reading a=
=20
directory, you are reading the directory. Don't add subfiles to=20
directories, until we find a semantic that really can't be handled=20
otherwise.

Take care,

Bill

--dDRMvlgZJXvWKvBx
Content-Type: application/pgp-signature
Content-Disposition: inline

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.3 (NetBSD)

iD8DBQFFDttMWz+3JHUci9cRAldAAKCHO1pK8jeF2Clg6VGEY8XEO3Z/mQCfa0Y4
d27EGjZHaczA5Qfm3l4id90=
=StRh
-----END PGP SIGNATURE-----

--dDRMvlgZJXvWKvBx--