public inbox for [email protected]
 help / color / mirror / Atom feed
From: Pavel Begunkov <[email protected]>
To: [email protected]
Cc: Jens Axboe <[email protected]>, [email protected]
Subject: [PATCH v3 09/11] io_uring: waitqueue-less cq waiting
Date: Mon,  9 Jan 2023 14:46:11 +0000	[thread overview]
Message-ID: <103d174d35d919d4cb0922d8a9c93a8f0c35f74a.1673274244.git.asml.silence@gmail.com> (raw)
In-Reply-To: <[email protected]>

With DEFER_TASKRUN only ctx->submitter_task might be waiting for CQEs,
we can use this to optimise io_cqring_wait(). Replace ->cq_wait
waitqueue with waking the task directly.

It works but misses an important optimisation covered by the following
patch, so this patch without follow ups might hurt performance.

Signed-off-by: Pavel Begunkov <[email protected]>
---
 io_uring/io_uring.c | 19 ++++++++++++-------
 1 file changed, 12 insertions(+), 7 deletions(-)

diff --git a/io_uring/io_uring.c b/io_uring/io_uring.c
index 0dabd0f3271f..62d879b14873 100644
--- a/io_uring/io_uring.c
+++ b/io_uring/io_uring.c
@@ -1263,7 +1263,7 @@ static void io_req_local_work_add(struct io_kiocb *req)
 		percpu_ref_put(&ctx->refs);
 		return;
 	}
-	/* need it for the following io_cqring_wake() */
+	/* needed for the following wake up */
 	smp_mb__after_atomic();
 
 	if (unlikely(atomic_read(&req->task->io_uring->in_idle))) {
@@ -1274,10 +1274,9 @@ static void io_req_local_work_add(struct io_kiocb *req)
 
 	if (ctx->flags & IORING_SETUP_TASKRUN_FLAG)
 		atomic_or(IORING_SQ_TASKRUN, &ctx->rings->sq_flags);
-
 	if (ctx->has_evfd)
 		io_eventfd_signal(ctx);
-	__io_cqring_wake(ctx);
+	wake_up_state(ctx->submitter_task, TASK_INTERRUPTIBLE);
 	percpu_ref_put(&ctx->refs);
 }
 
@@ -2578,12 +2577,17 @@ static int io_cqring_wait(struct io_ring_ctx *ctx, int min_events,
 	do {
 		unsigned long check_cq;
 
-		prepare_to_wait_exclusive(&ctx->cq_wait, &iowq.wq,
-						TASK_INTERRUPTIBLE);
+		if (ctx->flags & IORING_SETUP_DEFER_TASKRUN) {
+			set_current_state(TASK_INTERRUPTIBLE);
+		} else {
+			prepare_to_wait_exclusive(&ctx->cq_wait, &iowq.wq,
+							TASK_INTERRUPTIBLE);
+		}
+
 		ret = io_cqring_wait_schedule(ctx, &iowq);
+		__set_current_state(TASK_RUNNING);
 		if (ret < 0)
 			break;
-		__set_current_state(TASK_RUNNING);
 		/*
 		 * Run task_work after scheduling and before io_should_wake().
 		 * If we got woken because of task_work being processed, run it
@@ -2611,7 +2615,8 @@ static int io_cqring_wait(struct io_ring_ctx *ctx, int min_events,
 		cond_resched();
 	} while (1);
 
-	finish_wait(&ctx->cq_wait, &iowq.wq);
+	if (!(ctx->flags & IORING_SETUP_DEFER_TASKRUN))
+		finish_wait(&ctx->cq_wait, &iowq.wq);
 	restore_saved_sigmask_unless(ret == -EINTR);
 
 	return READ_ONCE(rings->cq.head) == READ_ONCE(rings->cq.tail) ? ret : 0;
-- 
2.38.1


  parent reply	other threads:[~2023-01-09 14:51 UTC|newest]

Thread overview: 13+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-01-09 14:46 [PATCH v3 00/11] CQ waiting / task_work optimisations Pavel Begunkov
2023-01-09 14:46 ` [PATCH v3 01/11] io_uring: move submitter_task out of cold cacheline Pavel Begunkov
2023-01-09 14:46 ` [PATCH v3 02/11] io_uring: refactor io_wake_function Pavel Begunkov
2023-01-09 14:46 ` [PATCH v3 03/11] io_uring: don't set TASK_RUNNING in local tw runner Pavel Begunkov
2023-01-09 14:46 ` [PATCH v3 04/11] io_uring: mark io_run_local_work static Pavel Begunkov
2023-01-09 14:46 ` [PATCH v3 05/11] io_uring: move io_run_local_work_locked Pavel Begunkov
2023-01-09 14:46 ` [PATCH v3 06/11] io_uring: separate wq for ring polling Pavel Begunkov
2023-01-09 14:46 ` [PATCH v3 07/11] io_uring: add lazy poll_wq activation Pavel Begunkov
2023-01-09 14:46 ` [PATCH v3 08/11] io_uring: wake up optimisations Pavel Begunkov
2023-01-09 14:46 ` Pavel Begunkov [this message]
2023-01-09 14:46 ` [PATCH v3 10/11] io_uring: add io_req_local_work_add wake fast path Pavel Begunkov
2023-01-09 14:46 ` [PATCH v3 11/11] io_uring: optimise deferred tw execution Pavel Begunkov
2023-01-11 18:00 ` [PATCH v3 00/11] CQ waiting / task_work optimisations 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=103d174d35d919d4cb0922d8a9c93a8f0c35f74a.1673274244.git.asml.silence@gmail.com \
    [email protected] \
    [email protected] \
    [email protected] \
    /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