From: Hao Xu <[email protected]>
To: Pavel Begunkov <[email protected]>,
Jens Axboe <[email protected]>,
[email protected]
Subject: Re: [PATCH 3/3] io_uring: don't spinlock when not posting CQEs
Date: Sun, 12 Sep 2021 04:12:35 +0800 [thread overview]
Message-ID: <[email protected]> (raw)
In-Reply-To: <3a5f0436099b84f71fdc8c9bd9f21842581feaf9.1631367587.git.asml.silence@gmail.com>
在 2021/9/11 下午9:52, Pavel Begunkov 写道:
> When no of queued for the batch completion requests need to post an CQE,
> see IOSQE_CQE_SKIP_SUCCESS, avoid grabbing ->completion_lock and other
> commit/post.
>
> Signed-off-by: Pavel Begunkov <[email protected]>
> ---
> fs/io_uring.c | 24 +++++++++++++++---------
> 1 file changed, 15 insertions(+), 9 deletions(-)
>
> diff --git a/fs/io_uring.c b/fs/io_uring.c
> index 172c857e8b3f..8983a5a6851a 100644
> --- a/fs/io_uring.c
> +++ b/fs/io_uring.c
> @@ -317,6 +317,7 @@ struct io_submit_state {
>
> bool plug_started;
> bool need_plug;
> + bool flush_cqes;
>
> /*
> * Batch completion logic
> @@ -1858,6 +1859,8 @@ static void io_req_complete_state(struct io_kiocb *req, long res,
> req->result = res;
> req->compl.cflags = cflags;
> req->flags |= REQ_F_COMPLETE_INLINE;
> + if (!(req->flags & IOSQE_CQE_SKIP_SUCCESS))
Haven't look into this patchset deeply, but this looks
like should be REQ_F_CQE_SKIP ?
> + req->ctx->submit_state.flush_cqes = true;
> }
>
> static inline void __io_req_complete(struct io_kiocb *req, unsigned issue_flags,
> @@ -2354,17 +2357,19 @@ static void __io_submit_flush_completions(struct io_ring_ctx *ctx)
> int i, nr = state->compl_nr;
> struct req_batch rb;
>
> - spin_lock(&ctx->completion_lock);
> - for (i = 0; i < nr; i++) {
> - struct io_kiocb *req = state->compl_reqs[i];
> + if (state->flush_cqes) {
> + spin_lock(&ctx->completion_lock);
> + for (i = 0; i < nr; i++) {
> + struct io_kiocb *req = state->compl_reqs[i];
>
> - if (!(req->flags & REQ_F_CQE_SKIP))
> - __io_fill_cqe(ctx, req->user_data, req->result,
> - req->compl.cflags);
> + if (!(req->flags & REQ_F_CQE_SKIP))
> + __io_fill_cqe(ctx, req->user_data, req->result,
> + req->compl.cflags);
> + }
> + io_commit_cqring(ctx);
> + spin_unlock(&ctx->completion_lock);
> + io_cqring_ev_posted(ctx);
> }
> - io_commit_cqring(ctx);
> - spin_unlock(&ctx->completion_lock);
> - io_cqring_ev_posted(ctx);
>
> io_init_req_batch(&rb);
> for (i = 0; i < nr; i++) {
> @@ -2376,6 +2381,7 @@ static void __io_submit_flush_completions(struct io_ring_ctx *ctx)
>
> io_req_free_batch_finish(ctx, &rb);
> state->compl_nr = 0;
> + state->flush_cqes = false;
> }
>
> /*
>
next prev parent reply other threads:[~2021-09-11 20:12 UTC|newest]
Thread overview: 11+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-09-11 13:51 [RFC][PATCH 0/3] allow to skip CQE posting Pavel Begunkov
2021-09-11 13:52 ` [PATCH 1/3] io_uring: clean cqe filling functions Pavel Begunkov
2021-09-12 18:24 ` Hao Xu
2021-09-12 23:20 ` Pavel Begunkov
2021-09-11 13:52 ` [PATCH 2/3] io_uring: add option to skip CQE posting Pavel Begunkov
2021-09-11 13:52 ` [PATCH 3/3] io_uring: don't spinlock when not posting CQEs Pavel Begunkov
2021-09-11 20:12 ` Hao Xu [this message]
2021-09-11 21:10 ` Pavel Begunkov
2021-09-12 18:14 ` Hao Xu
2021-09-12 22:16 ` Pavel Begunkov
2021-09-12 23:49 ` [RFC][PATCH 0/3] allow to skip CQE posting Pavel Begunkov
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=1cc2816e-bf18-fbb9-b5ed-e8786babc4fc@linux.alibaba.com \
[email protected] \
[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