From: Jes Sorensen <jes.sorensen@gmail.com>
To: linux-kernel@vger.kernel.org
Cc: john@johnmccutchan.com, rlove@rlove.org, eparis@parisplace.org,
josef@toxicpanda.com, kernel-team@fb.com,
Jes Sorensen <jsorensen@fb.com>
Subject: [PATCH 5/5] inotify: Avoid taking spinlock for each event processed in read()
Date: Fri, 24 Mar 2017 17:36:38 -0400 [thread overview]
Message-ID: <20170324213638.9114-6-jsorensen@fb.com> (raw)
In-Reply-To: <20170324213638.9114-1-jsorensen@fb.com>
Splice off the notification list while processing read events, which allows
it to be processed without taking and releasing the spinlock for each
event.
Signed-off-by: Jes Sorensen <jsorensen@fb.com>
Reviewed-by: Josef Bacik <jbacik@fb.com>
---
fs/notify/inotify/inotify_user.c | 28 ++++++++++++++++++++++++----
1 file changed, 24 insertions(+), 4 deletions(-)
diff --git a/fs/notify/inotify/inotify_user.c b/fs/notify/inotify/inotify_user.c
index 4a93750..45b1f69 100644
--- a/fs/notify/inotify/inotify_user.c
+++ b/fs/notify/inotify/inotify_user.c
@@ -224,22 +224,27 @@ static ssize_t inotify_read(struct file *file, char __user *buf,
struct fsnotify_group *group;
struct fsnotify_event *kevent;
char __user *start;
- int ret;
+ int ret, consumed = 0;
DEFINE_WAIT_FUNC(wait, woken_wake_function);
+ LIST_HEAD(tmp_list);
start = buf;
group = file->private_data;
mutex_lock(&group->inotify_data.consumer_mutex);
add_wait_queue(&group->notification_waitq, &wait);
+
+ spin_lock(&group->notification_lock);
+ list_splice_init(&group->notification_list, &tmp_list);
+ spin_unlock(&group->notification_lock);
+
while (1) {
- spin_lock(&group->notification_lock);
- kevent = get_one_event(&group->notification_list, count);
- spin_unlock(&group->notification_lock);
+ kevent = get_one_event(&tmp_list, count);
pr_debug("%s: group=%p kevent=%p\n", __func__, group, kevent);
if (kevent) {
+ consumed++;
ret = PTR_ERR(kevent);
if (IS_ERR(kevent))
break;
@@ -262,8 +267,23 @@ static ssize_t inotify_read(struct file *file, char __user *buf,
if (start != buf)
break;
+ spin_lock(&group->notification_lock);
+ list_splice_init(&tmp_list, &group->notification_list);
+ group->q_len -= consumed;
+ consumed = 0;
+ spin_unlock(&group->notification_lock);
+
wait_woken(&wait, TASK_INTERRUPTIBLE, MAX_SCHEDULE_TIMEOUT);
+
+ spin_lock(&group->notification_lock);
+ list_splice_init(&group->notification_list, &tmp_list);
+ spin_unlock(&group->notification_lock);
}
+ spin_lock(&group->notification_lock);
+ list_splice(&tmp_list, &group->notification_list);
+ group->q_len -= consumed;
+ spin_unlock(&group->notification_lock);
+
remove_wait_queue(&group->notification_waitq, &wait);
mutex_unlock(&group->inotify_data.consumer_mutex);
--
2.9.3
prev parent reply other threads:[~2017-03-24 21:37 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-03-24 21:36 [PATCH 0/5] inofify: Reduce lock contention on read() Jes Sorensen
2017-03-24 21:36 ` [PATCH 1/5] notify: Call mutex_destroy() before freeing mutex memory Jes Sorensen
2017-03-24 21:36 ` [PATCH 2/5] inotify: Use mutex to prevent threaded clients reading events out of order Jes Sorensen
2017-03-24 21:36 ` [PATCH 3/5] notify: Split up some fsnotify functions Jes Sorensen
2017-03-24 21:36 ` [PATCH 4/5] inotify: switch get_one_event() to use fsnotify_list_*() helpers Jes Sorensen
2017-03-24 21:36 ` Jes Sorensen [this message]
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=20170324213638.9114-6-jsorensen@fb.com \
--to=jes.sorensen@gmail.com \
--cc=eparis@parisplace.org \
--cc=john@johnmccutchan.com \
--cc=josef@toxicpanda.com \
--cc=jsorensen@fb.com \
--cc=kernel-team@fb.com \
--cc=linux-kernel@vger.kernel.org \
--cc=rlove@rlove.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
all inboxes | Powered by JetHome®