Date: Wed, 2 Jun 2010 12:45:04 +0300 From: Kostik Belousov <kostikbel@gmail.com> To: Mikolaj Golub <to.my.trociny@gmail.com> Cc: freebsd-fs@freebsd.org Subject: Re: nullfs: vop_rename: fdvp is locked but should not be Message-ID: <20100602094504.GN83316@deviant.kiev.zoral.com.ua> In-Reply-To: <86bpbtonma.fsf@zhuzha.ua1> References: <86fx16onx6.fsf@zhuzha.ua1> <20100601140721.GH83316@deviant.kiev.zoral.com.ua> <86bpbtonma.fsf@zhuzha.ua1>
next in thread | previous in thread | raw e-mail | index | archive | help
--b385GY4hqirqljNZ Content-Type: text/plain; charset=us-ascii Content-Disposition: inline Content-Transfer-Encoding: quoted-printable On Wed, Jun 02, 2010 at 11:16:45AM +0300, Mikolaj Golub wrote: >=20 > KB> I am curious to look at the final patch. Note that you proposing to = add > KB> fs-specific check to vfs_subr.c. Checking that the the vnode locks a= re > KB> different instead of that vnodes itself are different might be better > KB> approach for vop_rename_pre. >=20 > Ok. Looks a bit tricky :-). Then may be it is safe just to skip ASSERT() = if > the "target" and "from" vnodes are on different fs, like in the patch bel= ow? >=20 > I have tried the patch on CURRENT. It looks like it works... I do not think that the attached patch is the right solution Again, try. comparing the pointers to the vnode locks instead of the vnode pointers. Something like if (a->a_tdvp->v_vnlock !=3D a->a_fdvp->v_vnlock ... --b385GY4hqirqljNZ Content-Type: application/pgp-signature Content-Disposition: inline -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.10 (FreeBSD) iEYEARECAAYFAkwGKCAACgkQC3+MBN1Mb4gIzgCgkt+FR7+AVpfA5R6eVv9lDLR+ KlcAoOcSTkG0ircVSI5sFDcc3h9BgFXJ =5xvT -----END PGP SIGNATURE----- --b385GY4hqirqljNZ--
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20100602094504.GN83316>