From: " J. Bruce Fields" <bfields@citi.umich.edu>
To: Benny Halevy <bhalevy@panasas.com>
Cc: linux-nfs@vger.kernel.org, pnfs@linux-nfs.org
Subject: Re: [RFC 10/11] nfsd: close delegation only on last reference
Date: Wed, 16 Dec 2009 16:18:48 -0500 [thread overview]
Message-ID: <20091216211848.GD28822@fieldses.org> (raw)
In-Reply-To: <1260985349-21701-1-git-send-email-bhalevy@panasas.com>
On Wed, Dec 16, 2009 at 07:42:29PM +0200, Benny Halevy wrote:
> as dl_vfs_file may be used while the delegation structure is
> being referenced.
A callback to the client (which can last a while if the client is
unresponsive) can also hold a reference to the delegation. It doesn't
need a reference on the file.
I guess I can't see any bug in holding a reference on the file in that
case, but it seems better not to (if only to allow freeing it earlier).
Maybe the limited amount of information used by the callback should be
reference-counted separately. I'm not sure.
--b.
>
> Signed-off-by: Benny Halevy <bhalevy@panasas.com>
> ---
> fs/nfsd/nfs4state.c | 27 +++++++++++++--------------
> 1 files changed, 13 insertions(+), 14 deletions(-)
>
> diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c
> index 411226a..053fd2b 100644
> --- a/fs/nfsd/nfs4state.c
> +++ b/fs/nfsd/nfs4state.c
> @@ -203,29 +203,17 @@ alloc_init_deleg(struct nfs4_client *clp, struct nfs4_stateid *stp, struct svc_f
> return dp;
> }
>
> -void
> -nfs4_put_delegation(struct nfs4_delegation *dp)
> -{
> - if (atomic_dec_and_test(&dp->dl_count)) {
> - dprintk("NFSD: freeing dp %p\n",dp);
> - put_nfs4_file(dp->dl_file);
> - kmem_cache_free(deleg_slab, dp);
> - num_delegations--;
> - }
> -}
> -
> /* Remove the associated file_lock first, then remove the delegation.
> * lease_modify() is called to remove the FS_LEASE file_lock from
> * the i_flock list, eventually calling nfsd's lock_manager
> * fl_release_callback.
> */
> static void
> -nfs4_close_delegation(struct nfs4_delegation *dp)
> +__close_delegation(struct nfs4_delegation *dp)
> {
> struct file *filp = dp->dl_vfs_file;
>
> dprintk("NFSD: close_delegation dp %p\n",dp);
> - dp->dl_vfs_file = NULL;
> /* The following nfsd_close may not actually close the file,
> * but we want to remove the lease in any case. */
> if (dp->dl_flock)
> @@ -233,6 +221,18 @@ nfs4_close_delegation(struct nfs4_delegation *dp)
> nfsd_close(filp);
> }
>
> +void
> +nfs4_put_delegation(struct nfs4_delegation *dp)
> +{
> + if (atomic_dec_and_test(&dp->dl_count)) {
> + dprintk("NFSD: freeing dp %p\n",dp);
> + __close_delegation(dp);
> + put_nfs4_file(dp->dl_file);
> + kmem_cache_free(deleg_slab, dp);
> + num_delegations--;
> + }
> +}
> +
> /* Called under the state lock. */
> static void
> unhash_delegation(struct nfs4_delegation *dp)
> @@ -242,7 +242,6 @@ unhash_delegation(struct nfs4_delegation *dp)
> list_del_init(&dp->dl_perclnt);
> list_del_init(&dp->dl_recall_lru);
> spin_unlock(&deleg_lock);
> - nfs4_close_delegation(dp);
> nfs4_put_delegation(dp);
> }
>
> --
> 1.6.5.1
>
next prev parent reply other threads:[~2009-12-16 21:18 UTC|newest]
Thread overview: 24+ messages / expand[flat|nested] mbox.gz Atom feed top
2009-12-16 17:39 [RFC 0/11] state lock reduction prep Benny Halevy
2009-12-16 17:40 ` [RFC 01/11] nfsd: introduce nfs4_client state Benny Halevy
2009-12-16 22:33 ` [pnfs] " Peter Staubach
2009-12-16 23:17 ` Benny Halevy
2009-12-16 17:40 ` [RFC 02/11] nfsd: mark client state as expired in expire_client Benny Halevy
2009-12-16 17:40 ` [RFC 03/11] nfsd: introduce get_nfs4_client Benny Halevy
2009-12-16 17:41 ` [RFC 04/11] nfsd: mark client for renew Benny Halevy
2009-12-16 20:48 ` J. Bruce Fields
2009-12-16 20:53 ` J. Bruce Fields
2009-12-16 21:56 ` Benny Halevy
2009-12-16 22:06 ` J. Bruce Fields
2009-12-16 22:23 ` Benny Halevy
2009-12-17 20:19 ` J. Bruce Fields
2009-12-16 17:41 ` [RFC 05/11] nfsd: skip clients marked for renewal Benny Halevy
2009-12-16 20:58 ` J. Bruce Fields
2009-12-16 21:54 ` Benny Halevy
2009-12-16 17:41 ` [RFC 06/11] nfsd41: no need to hold the state lock around mark_client_for_renew Benny Halevy
2009-12-16 17:41 ` [RFC 07/11] nfsd: rename recall_lock to deleg_lock Benny Halevy
2009-12-16 17:42 ` [RFC 08/11] nfsd: use deleg_lock to protect dl_perfile and dl_perclnt Benny Halevy
2009-12-16 17:42 ` [RFC 09/11] nfsd: get a reference count on the delgation while outside of the deleg_lock Benny Halevy
2009-12-16 17:42 ` [RFC 10/11] nfsd: close delegation only on last reference Benny Halevy
2009-12-16 21:18 ` J. Bruce Fields [this message]
2009-12-16 22:04 ` Benny Halevy
2009-12-16 17:42 ` [RFC 11/11] nfsd: refactor unhash_delegation Benny Halevy
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20091216211848.GD28822@fieldses.org \
--to=bfields@citi.umich.edu \
--cc=bhalevy@panasas.com \
--cc=linux-nfs@vger.kernel.org \
--cc=pnfs@linux-nfs.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.