From: Pavel Begunkov <asml.silence@gmail.com>
To: io-uring@vger.kernel.org
Cc: asml.silence@gmail.com
Subject: [PATCH 2/3] io_uring/eventfd: clean up rcu locking
Date: Thu, 24 Apr 2025 12:31:17 +0100 [thread overview]
Message-ID: <91a925e708ca8a5aa7fee61f96d29b24ea9adeaf.1745493845.git.asml.silence@gmail.com> (raw)
In-Reply-To: <cover.1745493845.git.asml.silence@gmail.com>
Conditional locking is never welcome if there are better options. Move
rcu locking into io_eventfd_signal(), make it unconditional and use
guards. It also helps with sparse warnings.
Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
---
io_uring/eventfd.c | 24 +++++++-----------------
1 file changed, 7 insertions(+), 17 deletions(-)
diff --git a/io_uring/eventfd.c b/io_uring/eventfd.c
index a9da2d0d7510..8c2835ac17a0 100644
--- a/io_uring/eventfd.c
+++ b/io_uring/eventfd.c
@@ -47,13 +47,6 @@ static void io_eventfd_do_signal(struct rcu_head *rcu)
io_eventfd_put(ev_fd);
}
-static void io_eventfd_release(struct io_ev_fd *ev_fd, bool put_ref)
-{
- if (put_ref)
- io_eventfd_put(ev_fd);
- rcu_read_unlock();
-}
-
/*
* Returns true if the caller should put the ev_fd reference, false if not.
*/
@@ -89,11 +82,6 @@ static struct io_ev_fd *io_eventfd_grab(struct io_ring_ctx *ctx)
{
struct io_ev_fd *ev_fd;
- if (READ_ONCE(ctx->rings->cq_flags) & IORING_CQ_EVENTFD_DISABLED)
- return NULL;
-
- rcu_read_lock();
-
/*
* rcu_dereference ctx->io_ev_fd once and use it for both for checking
* and eventfd_signal
@@ -108,15 +96,18 @@ static struct io_ev_fd *io_eventfd_grab(struct io_ring_ctx *ctx)
if (io_eventfd_trigger(ev_fd) && refcount_inc_not_zero(&ev_fd->refs))
return ev_fd;
- rcu_read_unlock();
return NULL;
}
void io_eventfd_signal(struct io_ring_ctx *ctx, bool cqe_event)
{
- bool skip = false, put_ref = true;
+ bool skip = false;
struct io_ev_fd *ev_fd;
+ if (READ_ONCE(ctx->rings->cq_flags) & IORING_CQ_EVENTFD_DISABLED)
+ return;
+
+ guard(rcu)();
ev_fd = io_eventfd_grab(ctx);
if (!ev_fd)
return;
@@ -137,9 +128,8 @@ void io_eventfd_signal(struct io_ring_ctx *ctx, bool cqe_event)
spin_unlock(&ctx->completion_lock);
}
- if (!skip)
- put_ref = __io_eventfd_signal(ev_fd);
- io_eventfd_release(ev_fd, put_ref);
+ if (skip || __io_eventfd_signal(ev_fd))
+ io_eventfd_put(ev_fd);
}
int io_eventfd_register(struct io_ring_ctx *ctx, void __user *arg,
--
2.48.1
next prev parent reply other threads:[~2025-04-24 11:30 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-04-24 11:31 [PATCH 0/3] eventfd signalling cleanup Pavel Begunkov
2025-04-24 11:31 ` [PATCH 1/3] io_uring/eventfd: dedup signalling helpers Pavel Begunkov
2025-04-24 11:31 ` Pavel Begunkov [this message]
2025-04-24 11:31 ` [PATCH 3/3] io_uring/eventfd: open code io_eventfd_grab() Pavel Begunkov
2025-04-24 13:16 ` [PATCH 0/3] eventfd signalling cleanup 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=91a925e708ca8a5aa7fee61f96d29b24ea9adeaf.1745493845.git.asml.silence@gmail.com \
--to=asml.silence@gmail.com \
--cc=io-uring@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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox