From: "Serge E. Hallyn" <serue@us.ibm.com>
To: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: Greg Kroah-Hartman <gregkh@suse.de>,
Kay Sievers <kay.sievers@vrfy.org>,
linux-kernel@vger.kernel.org, Tejun Heo <tj@kernel.org>,
Cornelia Huck <cornelia.huck@de.ibm.com>,
linux-fsdevel@vger.kernel.org,
Eric Dumazet <eric.dumazet@gmail.com>,
Benjamin LaHaise <bcrl@lhnet.ca>,
"Eric W. Biederman" <ebiederm@aristanetworks.com>
Subject: Re: [PATCH 3/7] sysfs: Keep an nlink count on sysfs directories.
Date: Mon, 11 Jan 2010 23:56:41 -0600 [thread overview]
Message-ID: <20100112055641.GB10756@us.ibm.com> (raw)
In-Reply-To: <1263241315-19499-3-git-send-email-ebiederm@xmission.com>
Quoting Eric W. Biederman (ebiederm@xmission.com):
> From: Eric W. Biederman <ebiederm@aristanetworks.com>
>
> On large directories sysfs_count_nlinks can be a significant
> bottleneck, so keep a count in sysfs_dirent. If we exceed
> the maximum number of directory entries we can store return
> nlink of 1. An nlink of 1 matches what reiserfs does in
> this case, and it let's find and similar utlities know that
> we have a the directory nlink can not be used for optimization
> purposes.
>
> Signed-off-by: Eric W. Biederman <ebiederm@aristanetworks.com>
> ---
> fs/sysfs/dir.c | 20 ++++++++++++++++++++
> fs/sysfs/inode.c | 15 +--------------
> fs/sysfs/mount.c | 1 +
> fs/sysfs/sysfs.h | 1 +
> 4 files changed, 23 insertions(+), 14 deletions(-)
>
> diff --git a/fs/sysfs/dir.c b/fs/sysfs/dir.c
> index 5c4703d..1c364be 100644
> --- a/fs/sysfs/dir.c
> +++ b/fs/sysfs/dir.c
> @@ -359,6 +359,7 @@ struct sysfs_dirent *sysfs_new_dirent(const char *name, umode_t mode, int type)
> sd->s_name = name;
> sd->s_mode = mode;
> sd->s_flags = type;
> + sd->s_nlink = type == SYSFS_DIR? 2:1;
>
> return sd;
>
> @@ -392,6 +393,23 @@ void sysfs_addrm_start(struct sysfs_addrm_cxt *acxt,
> mutex_lock(&sysfs_mutex);
> }
>
> +static void sysfs_dir_inc_nlink(struct sysfs_dirent *sd)
> +{
> + /* If we overflow force nlink to be 1 */
> + if (sd->s_nlink > 1) {
> + sd->s_nlink++;
> + if (sd->s_nlink == 0)
> + sd->s_nlink = 1;
> + }
> +}
Now, the original code only ups nlink on parent if child is a dir.
IIUC your code will now inc nlink for files as well. Is any userspace
going to be confused by that?
> +
> +static void sysfs_dir_dec_nlink(struct sysfs_dirent *sd)
> +{
> + /* Don't decrement if we overflowed an increment */
> + if (sd->s_nlink != 1)
> + sd->s_nlink--;
> +}
> +
> /**
> * __sysfs_add_one - add sysfs_dirent to parent without warning
> * @acxt: addrm context to use
> @@ -420,6 +438,7 @@ int __sysfs_add_one(struct sysfs_addrm_cxt *acxt, struct sysfs_dirent *sd)
> return -EEXIST;
>
> sd->s_parent = sysfs_get(acxt->parent_sd);
> + sysfs_dir_inc_nlink(sd->s_parent);
>
> sysfs_link_sibling(sd);
>
> @@ -512,6 +531,7 @@ void sysfs_remove_one(struct sysfs_addrm_cxt *acxt, struct sysfs_dirent *sd)
>
> BUG_ON(sd->s_flags & SYSFS_FLAG_REMOVED);
>
> + sysfs_dir_dec_nlink(sd->s_parent);
> sysfs_unlink_sibling(sd);
>
> /* Update timestamps on the parent */
> diff --git a/fs/sysfs/inode.c b/fs/sysfs/inode.c
> index 104cbc1..f0172b3 100644
> --- a/fs/sysfs/inode.c
> +++ b/fs/sysfs/inode.c
> @@ -201,18 +201,6 @@ static inline void set_inode_attr(struct inode * inode, struct iattr * iattr)
> inode->i_ctime = iattr->ia_ctime;
> }
>
> -static int sysfs_count_nlink(struct sysfs_dirent *sd)
> -{
> - struct sysfs_dirent *child;
> - int nr = 0;
> -
> - for (child = sd->s_dir.children; child; child = child->s_sibling)
> - if (sysfs_type(child) == SYSFS_DIR)
> - nr++;
> -
> - return nr + 2;
> -}
> -
> static void sysfs_refresh_inode(struct sysfs_dirent *sd, struct inode *inode)
> {
> struct sysfs_inode_attrs *iattrs = sd->s_iattr;
> @@ -228,8 +216,7 @@ static void sysfs_refresh_inode(struct sysfs_dirent *sd, struct inode *inode)
> iattrs->ia_secdata_len);
> }
>
> - if (sysfs_type(sd) == SYSFS_DIR)
> - inode->i_nlink = sysfs_count_nlink(sd);
> + inode->i_nlink = sd->s_nlink;
> }
>
> int sysfs_getattr(struct vfsmount *mnt, struct dentry *dentry, struct kstat *stat)
> diff --git a/fs/sysfs/mount.c b/fs/sysfs/mount.c
> index 4974995..372a32c 100644
> --- a/fs/sysfs/mount.c
> +++ b/fs/sysfs/mount.c
> @@ -37,6 +37,7 @@ struct sysfs_dirent sysfs_root = {
> .s_count = ATOMIC_INIT(1),
> .s_flags = SYSFS_DIR,
> .s_mode = S_IFDIR | S_IRWXU | S_IRUGO | S_IXUGO,
> + .s_nlink = 2,
> .s_ino = 1,
> };
>
> diff --git a/fs/sysfs/sysfs.h b/fs/sysfs/sysfs.h
> index 0417805..85d5c6c 100644
> --- a/fs/sysfs/sysfs.h
> +++ b/fs/sysfs/sysfs.h
> @@ -67,6 +67,7 @@ struct sysfs_dirent {
>
> unsigned int s_flags;
> unsigned short s_mode;
> + unsigned short s_nlink;
> ino_t s_ino;
> struct sysfs_inode_attrs *s_iattr;
> };
> --
> 1.6.5.2.143.g8cc62
next prev parent reply other threads:[~2010-01-12 5:56 UTC|newest]
Thread overview: 36+ messages / expand[flat|nested] mbox.gz Atom feed top
2010-01-11 20:16 [PATCH 0/7] General sysfs enhancements Eric W. Biederman
2010-01-11 20:21 ` [PATCH 1/7] sysfs: Serialize updates to the vfs inode Eric W. Biederman
2010-01-12 5:41 ` Serge E. Hallyn
2010-01-11 20:21 ` [PATCH 2/7] sysfs: Pack sysfs_dirent more tightly Eric W. Biederman
2010-01-12 0:41 ` Tejun Heo
2010-01-11 20:21 ` [PATCH 3/7] sysfs: Keep an nlink count on sysfs directories Eric W. Biederman
2010-01-12 0:46 ` Tejun Heo
2010-01-12 0:53 ` Benjamin LaHaise
2010-01-12 1:06 ` Eric W. Biederman
2010-01-12 1:12 ` Benjamin LaHaise
2010-01-12 1:23 ` Eric W. Biederman
2010-01-12 6:22 ` Tejun Heo
2010-01-12 15:49 ` Valdis.Kletnieks
2010-01-12 1:02 ` Eric W. Biederman
2010-01-12 5:56 ` Serge E. Hallyn [this message]
2010-01-12 8:30 ` Eric W. Biederman
2010-01-12 12:41 ` Cornelia Huck
2010-01-12 15:34 ` Eric W. Biederman
2010-01-11 20:21 ` [PATCH 4/7] sysfs: Implement sysfs_rename_link Eric W. Biederman
2010-01-12 6:24 ` Tejun Heo
2010-01-12 17:30 ` Serge E. Hallyn
2010-01-11 20:21 ` [PATCH 5/7] driver core: Use sysfs_rename_link in device_rename Eric W. Biederman
2010-01-12 6:25 ` Tejun Heo
2010-01-12 17:34 ` Serge E. Hallyn
2010-01-11 20:21 ` [PATCH 6/7] sysfs: Pass super_block to sysfs_get_inode Eric W. Biederman
2010-01-12 17:43 ` Serge E. Hallyn
2010-01-11 20:21 ` [PATCH 7/7] sysfs: Kill unused sysfs_sb variable Eric W. Biederman
2010-01-12 17:43 ` Serge E. Hallyn
2010-01-15 21:37 ` [PATCH 0/7] General sysfs enhancements Greg KH
2010-02-13 3:20 ` [PATCH 0/6] " Eric W. Biederman
2010-02-13 3:22 ` [PATCH 1/6] sysfs: Serialize updates to the vfs inode Eric W. Biederman
2010-02-13 3:22 ` [PATCH 2/6] sysfs: Pack sysfs_dirent more tightly Eric W. Biederman
2010-02-13 3:22 ` [PATCH 3/6] sysfs: Implement sysfs_rename_link Eric W. Biederman
2010-02-13 3:22 ` [PATCH 4/6] driver core: Use sysfs_rename_link in device_rename Eric W. Biederman
2010-02-13 3:22 ` [PATCH 5/6] sysfs: Pass super_block to sysfs_get_inode Eric W. Biederman
2010-02-13 3:22 ` [PATCH 6/6] sysfs: Kill unused sysfs_sb variable Eric W. Biederman
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=20100112055641.GB10756@us.ibm.com \
--to=serue@us.ibm.com \
--cc=bcrl@lhnet.ca \
--cc=cornelia.huck@de.ibm.com \
--cc=ebiederm@aristanetworks.com \
--cc=ebiederm@xmission.com \
--cc=eric.dumazet@gmail.com \
--cc=gregkh@suse.de \
--cc=kay.sievers@vrfy.org \
--cc=linux-fsdevel@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=tj@kernel.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 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).