All of lore.kernel.org
 help / color / mirror / Atom feed
From: Amir Goldstein <amir73il@gmail.com>
To: Jan Kara <jack@suse.cz>, Jens Axboe <axboe@kernel.dk>
Cc: Christian Brauner <brauner@kernel.org>, linux-fsdevel@vger.kernel.org
Subject: [PATCH v3] fsnotify: optimize the case of no parent watcher
Date: Tue, 16 Jan 2024 13:32:47 +0200	[thread overview]
Message-ID: <20240116113247.758848-1-amir73il@gmail.com> (raw)

If parent inode is not watching, check for the event in masks of
sb/mount/inode masks early to optimize out most of the code in
__fsnotify_parent() and avoid calling fsnotify().

Jens has reported that this optimization improves BW and IOPS in an
io_uring benchmark by more than 10% and reduces perf reported CPU usage.

before:

+    4.51%  io_uring  [kernel.vmlinux]  [k] fsnotify
+    3.67%  io_uring  [kernel.vmlinux]  [k] __fsnotify_parent

after:

+    2.37%  io_uring  [kernel.vmlinux]  [k] __fsnotify_parent

Reported-and-tested-by: Jens Axboe <axboe@kernel.dk>
Link: https://lore.kernel.org/linux-fsdevel/b45bd8ff-5654-4e67-90a6-aad5e6759e0b@kernel.dk/
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
---

Jan,

Considering that this change looks like a clear win and it actually
the change that you suggested, I cleaned it up a bit and posting for
your consideration.

I've kept the wrappers fsnotify_path() and fsnotify_sb_has_watchers()
although they are not directly related to the optimization, because they
makes the code a bit nicer IMO.

Jens,

I took the liberty to add Reported-and-tested-by to attribute for your
help even though the patch that you tested was slightly different than
this cleaned up version, because logically it should be identical.

Thanks,
Amir.

Changes since v2:
- Add helper fsnotify_object_watched()
- Move the optimization from inlined fsnotify_path() to
  extern __fsnotify_parent()
- Optimize for any event - not only content events
- Drop the SB_I_CONTENT_WATCHED flag

 fs/notify/fsnotify.c     | 28 +++++++++++++++++-----------
 include/linux/fsnotify.h | 22 +++++++++++++++-------
 2 files changed, 32 insertions(+), 18 deletions(-)

diff --git a/fs/notify/fsnotify.c b/fs/notify/fsnotify.c
index 8bfd690e9f10..6582204cca70 100644
--- a/fs/notify/fsnotify.c
+++ b/fs/notify/fsnotify.c
@@ -141,7 +141,7 @@ void __fsnotify_update_child_dentry_flags(struct inode *inode)
 }
 
 /* Are inode/sb/mount interested in parent and name info with this event? */
-static bool fsnotify_event_needs_parent(struct inode *inode, struct mount *mnt,
+static bool fsnotify_event_needs_parent(struct inode *inode, __u32 mnt_mask,
 					__u32 mask)
 {
 	__u32 marks_mask = 0;
@@ -160,13 +160,22 @@ static bool fsnotify_event_needs_parent(struct inode *inode, struct mount *mnt,
 	/* Did either inode/sb/mount subscribe for events with parent/name? */
 	marks_mask |= fsnotify_parent_needed_mask(inode->i_fsnotify_mask);
 	marks_mask |= fsnotify_parent_needed_mask(inode->i_sb->s_fsnotify_mask);
-	if (mnt)
-		marks_mask |= fsnotify_parent_needed_mask(mnt->mnt_fsnotify_mask);
+	marks_mask |= fsnotify_parent_needed_mask(mnt_mask);
 
 	/* Did they subscribe for this event with parent/name info? */
 	return mask & marks_mask;
 }
 
+/* Are there any inode/mount/sb objects that are interested in this event? */
+static inline bool fsnotify_object_watched(struct inode *inode, __u32 mnt_mask,
+					   __u32 mask)
+{
+	__u32 marks_mask = inode->i_fsnotify_mask | mnt_mask |
+			   inode->i_sb->s_fsnotify_mask;
+
+	return mask & marks_mask & ALL_FSNOTIFY_EVENTS;
+}
+
 /*
  * Notify this dentry's parent about a child's events with child name info
  * if parent is watching or if inode/sb/mount are interested in events with
@@ -179,7 +188,7 @@ int __fsnotify_parent(struct dentry *dentry, __u32 mask, const void *data,
 		      int data_type)
 {
 	const struct path *path = fsnotify_data_path(data, data_type);
-	struct mount *mnt = path ? real_mount(path->mnt) : NULL;
+	__u32 mnt_mask = path ? real_mount(path->mnt)->mnt_fsnotify_mask : 0;
 	struct inode *inode = d_inode(dentry);
 	struct dentry *parent;
 	bool parent_watched = dentry->d_flags & DCACHE_FSNOTIFY_PARENT_WATCHED;
@@ -190,16 +199,13 @@ int __fsnotify_parent(struct dentry *dentry, __u32 mask, const void *data,
 	struct qstr *file_name = NULL;
 	int ret = 0;
 
-	/*
-	 * Do inode/sb/mount care about parent and name info on non-dir?
-	 * Do they care about any event at all?
-	 */
-	if (!inode->i_fsnotify_marks && !inode->i_sb->s_fsnotify_marks &&
-	    (!mnt || !mnt->mnt_fsnotify_marks) && !parent_watched)
+	/* Optimize the likely case of nobody watching this path */
+	if (likely(!parent_watched) &&
+	    likely(!fsnotify_object_watched(inode, mnt_mask, mask)))
 		return 0;
 
 	parent = NULL;
-	parent_needed = fsnotify_event_needs_parent(inode, mnt, mask);
+	parent_needed = fsnotify_event_needs_parent(inode, mnt_mask, mask);
 	if (!parent_watched && !parent_needed)
 		goto notify;
 
diff --git a/include/linux/fsnotify.h b/include/linux/fsnotify.h
index 8300a5286988..2ddfd7d1eca8 100644
--- a/include/linux/fsnotify.h
+++ b/include/linux/fsnotify.h
@@ -17,6 +17,12 @@
 #include <linux/slab.h>
 #include <linux/bug.h>
 
+/* Are there any inode/mount/sb objects that are being watched at all? */
+static inline bool fsnotify_sb_has_watchers(struct super_block *sb)
+{
+	return atomic_long_read(&sb->s_fsnotify_connectors);
+}
+
 /*
  * Notify this @dir inode about a change in a child directory entry.
  * The directory entry may have turned positive or negative or its inode may
@@ -30,7 +36,7 @@ static inline int fsnotify_name(__u32 mask, const void *data, int data_type,
 				struct inode *dir, const struct qstr *name,
 				u32 cookie)
 {
-	if (atomic_long_read(&dir->i_sb->s_fsnotify_connectors) == 0)
+	if (!fsnotify_sb_has_watchers(dir->i_sb))
 		return 0;
 
 	return fsnotify(mask, data, data_type, dir, name, NULL, cookie);
@@ -44,7 +50,7 @@ static inline void fsnotify_dirent(struct inode *dir, struct dentry *dentry,
 
 static inline void fsnotify_inode(struct inode *inode, __u32 mask)
 {
-	if (atomic_long_read(&inode->i_sb->s_fsnotify_connectors) == 0)
+	if (!fsnotify_sb_has_watchers(inode->i_sb))
 		return;
 
 	if (S_ISDIR(inode->i_mode))
@@ -59,7 +65,7 @@ static inline int fsnotify_parent(struct dentry *dentry, __u32 mask,
 {
 	struct inode *inode = d_inode(dentry);
 
-	if (atomic_long_read(&inode->i_sb->s_fsnotify_connectors) == 0)
+	if (!fsnotify_sb_has_watchers(inode->i_sb))
 		return 0;
 
 	if (S_ISDIR(inode->i_mode)) {
@@ -89,15 +95,17 @@ static inline void fsnotify_dentry(struct dentry *dentry, __u32 mask)
 	fsnotify_parent(dentry, mask, dentry, FSNOTIFY_EVENT_DENTRY);
 }
 
-static inline int fsnotify_file(struct file *file, __u32 mask)
+static inline int fsnotify_path(const struct path *path, __u32 mask)
 {
-	const struct path *path;
+	return fsnotify_parent(path->dentry, mask, path, FSNOTIFY_EVENT_PATH);
+}
 
+static inline int fsnotify_file(struct file *file, __u32 mask)
+{
 	if (file->f_mode & FMODE_NONOTIFY)
 		return 0;
 
-	path = &file->f_path;
-	return fsnotify_parent(path->dentry, mask, path, FSNOTIFY_EVENT_PATH);
+	return fsnotify_path(&file->f_path, mask);
 }
 
 #ifdef CONFIG_FANOTIFY_ACCESS_PERMISSIONS
-- 
2.34.1


             reply	other threads:[~2024-01-16 11:32 UTC|newest]

Thread overview: 13+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-01-16 11:32 Amir Goldstein [this message]
2024-01-16 12:04 ` [PATCH v3] fsnotify: optimize the case of no parent watcher Jan Kara
2024-01-16 12:53   ` Amir Goldstein
2024-01-24 16:07     ` Jan Kara
2024-01-24 16:20       ` Amir Goldstein
2024-02-13 19:45         ` Amir Goldstein
2024-02-14 11:23           ` Jan Kara
2024-02-14 13:40             ` Amir Goldstein
2024-02-15  8:36               ` Jan Kara
2024-03-06 14:51                 ` Amir Goldstein
2024-03-08 16:00                   ` Jan Kara
2024-03-11 13:51                     ` Christian Brauner
2024-02-15 15:07           ` Jens Axboe

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=20240116113247.758848-1-amir73il@gmail.com \
    --to=amir73il@gmail.com \
    --cc=axboe@kernel.dk \
    --cc=brauner@kernel.org \
    --cc=jack@suse.cz \
    --cc=linux-fsdevel@vger.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 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.