public inbox for [email protected]
 help / color / mirror / Atom feed
From: Pavel Begunkov <[email protected]>
To: Marcelo Diop-Gonzalez <[email protected]>, [email protected]
Cc: [email protected]
Subject: Re: [PATCH v3 1/1] io_uring: flush timeouts that should already have expired
Date: Thu, 14 Jan 2021 21:40:01 +0000	[thread overview]
Message-ID: <[email protected]> (raw)
In-Reply-To: <[email protected]>

On 14/01/2021 15:50, Marcelo Diop-Gonzalez wrote:
> Right now io_flush_timeouts() checks if the current number of events
> is equal to ->timeout.target_seq, but this will miss some timeouts if
> there have been more than 1 event added since the last time they were
> flushed (possible in io_submit_flush_completions(), for example). Fix
> it by recording the last sequence at which timeouts were flushed so
> that the number of events seen can be compared to the number of events
> needed without overflow.

Looks good, but there is a little change I'll ask you to make (see
below). In a meanwhile I'll test it, so the patch on the fast track.

> 
> Signed-off-by: Marcelo Diop-Gonzalez <[email protected]>
> ---
>  fs/io_uring.c | 29 +++++++++++++++++++++++++----
>  1 file changed, 25 insertions(+), 4 deletions(-)
> 
> diff --git a/fs/io_uring.c b/fs/io_uring.c
> index 372be9caf340..71d8fa0733ad 100644
> --- a/fs/io_uring.c
> +++ b/fs/io_uring.c
> @@ -354,6 +354,7 @@ struct io_ring_ctx {
>  		unsigned		cq_entries;
>  		unsigned		cq_mask;
>  		atomic_t		cq_timeouts;
> +		unsigned		cq_last_tm_flush;
>  		unsigned long		cq_check_overflow;
>  		struct wait_queue_head	cq_wait;
>  		struct fasync_struct	*cq_fasync;
> @@ -1639,19 +1640,36 @@ static void __io_queue_deferred(struct io_ring_ctx *ctx)
>  
>  static void io_flush_timeouts(struct io_ring_ctx *ctx)
>  {
> -	while (!list_empty(&ctx->timeout_list)) {
> +	u32 seq = ctx->cached_cq_tail - atomic_read(&ctx->cq_timeouts);

This assignment should go after list_empty() -- because of the atomic part
my compiler can't reshuffle them itself.

> +
> +	if (list_empty(&ctx->timeout_list))
> +		return;
[...]
>  static void io_commit_cqring(struct io_ring_ctx *ctx)
> @@ -5837,6 +5855,9 @@ static int io_timeout(struct io_kiocb *req)
>  	tail = ctx->cached_cq_tail - atomic_read(&ctx->cq_timeouts);
>  	req->timeout.target_seq = tail + off;
>  
> +	/* Update the last seq here in case io_flush_timeouts() hasn't */
> +	ctx->cq_last_tm_flush = tail;

Have to note that it's ok to do because we don't mix submissions and
completions, so io_timeout should never fall under same completion_lock
section as cq commit,

but otherwise some future locked version of io_timeout would be cutting
off a part of the current flush window (i.e. this [last, cur] thing).

-- 
Pavel Begunkov

  reply	other threads:[~2021-01-14 21:44 UTC|newest]

Thread overview: 6+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-01-14 15:50 [PATCH v3 0/1] io_uring: fix skipping of old timeout events Marcelo Diop-Gonzalez
2021-01-14 15:50 ` [PATCH v3 1/1] io_uring: flush timeouts that should already have expired Marcelo Diop-Gonzalez
2021-01-14 21:40   ` Pavel Begunkov [this message]
2021-01-15 14:45     ` Pavel Begunkov
2021-01-14 21:42 ` [PATCH v3 0/1] io_uring: fix skipping of old timeout events Pavel Begunkov
2021-01-15 16:37   ` Marcelo Diop-Gonzalez

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 \
    [email protected] \
    [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