--jq0ap7NbKX2Kqbes
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

On Thu, Feb 19, 2004 at 06:46:04PM +0900, Bang Jun-Young wrote:
> > anyone got pagelocks recently ?
>=20
> Are you using lfs? The deadlock occurred to me when I tried to extract a
> very large tarball on a lfs volume. Kernels between 2/14 and yesterday
> have the same problem. What's odd is, the same thing was okay with a ffs
> (+softdep) partition.

I don't know LFS, but I understand it doesn't use the uvm page cache,
and all data is still done using the buffercache (which has now
changed under it).  As a result, it could easily trigger the same kind
of tuning problems and assumptions with a buffercache full of dirty
data as a big burst of softdep metadata updates.

--
Dan.

--jq0ap7NbKX2Kqbes
Content-Type: application/pgp-signature
Content-Disposition: inline

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (NetBSD)

iD8DBQFANGN9EAVxvV4N66cRAntHAKCMe5krr6lPih+PNH0EQAyPNHdyzgCgkZi2
a5MMOHaGpR8iLbfqc4EAX4g=
=DMJv
-----END PGP SIGNATURE-----

--jq0ap7NbKX2Kqbes--