From: Xiubo Li <xiubli@redhat.com>
To: Al Viro <viro@zeniv.linux.org.uk>, linux-fsdevel@vger.kernel.org
Cc: ceph-devel@vger.kernel.org
Subject: Re: [PATCH 17/22] get rid of passing callbacks to ceph __dentry_leases_walk()
Date: Thu, 21 Dec 2023 08:45:18 +0800 [thread overview]
Message-ID: <446cf570-4d3d-4bdb-978c-a61d801a8c32@redhat.com> (raw)
In-Reply-To: <20231220052925.GP1674809@ZenIV>
On 12/20/23 13:29, Al Viro wrote:
> __dentry_leases_walk() is gets a callback and calls it for
> a bunch of denties; there are exactly two callers and
> we already have a flag telling them apart - lwc->dir_lease.
>
> Seeing that indirect calls are costly these days, let's
> get rid of the callback and just call the right function
> directly. Has a side benefit of saner signatures...
>
> Signed-off-by Al Viro <viro@zeniv.linux.org.uk>
> ---
> fs/ceph/dir.c | 21 +++++++++++++--------
> 1 file changed, 13 insertions(+), 8 deletions(-)
>
> diff --git a/fs/ceph/dir.c b/fs/ceph/dir.c
> index 91709934c8b1..768158743750 100644
> --- a/fs/ceph/dir.c
> +++ b/fs/ceph/dir.c
> @@ -1593,10 +1593,12 @@ struct ceph_lease_walk_control {
> unsigned long dir_lease_ttl;
> };
>
> +static int __dir_lease_check(const struct dentry *, struct ceph_lease_walk_control *);
> +static int __dentry_lease_check(const struct dentry *);
> +
> static unsigned long
> __dentry_leases_walk(struct ceph_mds_client *mdsc,
> - struct ceph_lease_walk_control *lwc,
> - int (*check)(struct dentry*, void*))
> + struct ceph_lease_walk_control *lwc)
> {
> struct ceph_dentry_info *di, *tmp;
> struct dentry *dentry, *last = NULL;
> @@ -1624,7 +1626,10 @@ __dentry_leases_walk(struct ceph_mds_client *mdsc,
> goto next;
> }
>
> - ret = check(dentry, lwc);
> + if (lwc->dir_lease)
> + ret = __dir_lease_check(dentry, lwc);
> + else
> + ret = __dentry_lease_check(dentry);
> if (ret & TOUCH) {
> /* move it into tail of dir lease list */
> __dentry_dir_lease_touch(mdsc, di);
> @@ -1681,7 +1686,7 @@ __dentry_leases_walk(struct ceph_mds_client *mdsc,
> return freed;
> }
>
> -static int __dentry_lease_check(struct dentry *dentry, void *arg)
> +static int __dentry_lease_check(const struct dentry *dentry)
> {
> struct ceph_dentry_info *di = ceph_dentry(dentry);
> int ret;
> @@ -1696,9 +1701,9 @@ static int __dentry_lease_check(struct dentry *dentry, void *arg)
> return DELETE;
> }
>
> -static int __dir_lease_check(struct dentry *dentry, void *arg)
> +static int __dir_lease_check(const struct dentry *dentry,
> + struct ceph_lease_walk_control *lwc)
> {
> - struct ceph_lease_walk_control *lwc = arg;
> struct ceph_dentry_info *di = ceph_dentry(dentry);
>
> int ret = __dir_lease_try_check(dentry);
> @@ -1737,7 +1742,7 @@ int ceph_trim_dentries(struct ceph_mds_client *mdsc)
>
> lwc.dir_lease = false;
> lwc.nr_to_scan = CEPH_CAPS_PER_RELEASE * 2;
> - freed = __dentry_leases_walk(mdsc, &lwc, __dentry_lease_check);
> + freed = __dentry_leases_walk(mdsc, &lwc);
> if (!lwc.nr_to_scan) /* more invalid leases */
> return -EAGAIN;
>
> @@ -1747,7 +1752,7 @@ int ceph_trim_dentries(struct ceph_mds_client *mdsc)
> lwc.dir_lease = true;
> lwc.expire_dir_lease = freed < count;
> lwc.dir_lease_ttl = mdsc->fsc->mount_options->caps_wanted_delay_max * HZ;
> - freed +=__dentry_leases_walk(mdsc, &lwc, __dir_lease_check);
> + freed +=__dentry_leases_walk(mdsc, &lwc);
> if (!lwc.nr_to_scan) /* more to check */
> return -EAGAIN;
>
Reviewed-by: Xiubo Li <xiubli@redhat.com>
Al,
I think these two ceph patches won't be dependent by your following
patches, right ?
If so we can apply them to ceph-client tree and run more tests.
Thanks!
next prev parent reply other threads:[~2023-12-21 0:45 UTC|newest]
Thread overview: 38+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-12-20 5:13 [PATCHES] assorted fs cleanups Al Viro
2023-12-20 5:15 ` PATCH 01/22] hostfs: use d_splice_alias() calling conventions to simplify failure exits Al Viro
2023-12-20 5:16 ` [PATCH 02/22] /proc/sys: " Al Viro
2023-12-20 5:17 ` [PATCH 03/22] zonefs: d_splice_alias() will do the right thing on ERR_PTR() inode Al Viro
2023-12-20 7:09 ` Damien Le Moal
2023-12-20 5:18 ` [PATCH 04/22] udf: " Al Viro
2023-12-21 12:19 ` Jan Kara
2023-12-20 5:19 ` [PATCH 05/22] affs: d_obtain_alias(ERR_PTR(...)) will do the right thing Al Viro
2023-12-20 5:19 ` [PATCH 06/22] befs: " Al Viro
2023-12-20 5:20 ` [PATCH 07/22] ceph: d_obtain_{alias,root}(ERR_PTR(...)) " Al Viro
2023-12-21 0:35 ` Xiubo Li
2023-12-20 5:21 ` [PATCH 08/22] gfs2: d_obtain_alias(ERR_PTR(...)) " Al Viro
2023-12-20 5:22 ` [PATCH 09/22] kernfs: d_obtain_alias(NULL) " Al Viro
2023-12-20 5:23 ` nilfs2: d_obtain_alias(ERR_PTR(...)) " Al Viro
2023-12-20 14:46 ` Ryusuke Konishi
2023-12-20 5:23 ` [PATCH 11/22] udf: " Al Viro
2023-12-21 12:20 ` Jan Kara
2023-12-20 5:24 ` [PATCH 12/22] ext4_add_entry(): ->d_name.len is never 0 Al Viro
2023-12-20 5:25 ` [PATCH 13/22] bfs_add_entry(): get rid of pointless ->d_name.len checks Al Viro
2023-12-20 11:03 ` Tigran Aivazian
2023-12-20 5:26 ` [PATCH 14/22] __ocfs2_add_entry(), ocfs2_prepare_dir_for_insert(): namelen checks Al Viro
2023-12-20 5:27 ` [PATCH 15/22] reiserfs_add_entry(): get rid of pointless " Al Viro
2023-12-20 5:28 ` [PATCH 16/22] udf_fiiter_add_entry(): check for zero ->d_name.len is bogus Al Viro
2023-12-21 12:22 ` Jan Kara
2023-12-20 5:29 ` [PATCH 17/22] get rid of passing callbacks to ceph __dentry_leases_walk() Al Viro
2023-12-21 0:45 ` Xiubo Li [this message]
2023-12-21 1:12 ` Al Viro
2023-12-21 1:16 ` Xiubo Li
2023-12-20 5:30 ` [PATCH 18/22] nfsd: kill stale comment about simple_fill_super() requirements Al Viro
2023-12-20 14:22 ` Chuck Lever
2023-12-20 5:31 ` [PATCH 19/22] ocfs2_find_match(): there's no such thing as NULL or negative ->d_parent Al Viro
2023-12-20 5:31 ` [PATCH 20/22] gfs2: use is_subdir() Al Viro
2023-12-20 5:32 ` [PATCH 21/22] orangefs: saner arguments passing in readdir guts Al Viro
2023-12-27 12:05 ` Mike Marshall
2024-01-02 16:25 ` Mike Marshall
2023-12-20 5:33 ` [PATCH 22/22] apparmorfs: don't duplicate kfree_link() Al Viro
2023-12-20 12:16 ` [PATCHES] assorted fs cleanups Andreas Grünbacher
2023-12-20 17:15 ` Al Viro
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=446cf570-4d3d-4bdb-978c-a61d801a8c32@redhat.com \
--to=xiubli@redhat.com \
--cc=ceph-devel@vger.kernel.org \
--cc=linux-fsdevel@vger.kernel.org \
--cc=viro@zeniv.linux.org.uk \
/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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).