From: Dmitry Monakhov <dmtrmonakhov@yandex-team.ru>
To: "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>
Cc: "linux-fsdevel@vger.kernel.org" <linux-fsdevel@vger.kernel.org>,
"axboe@kernel.dk" <axboe@kernel.dk>
Subject: Re: [PATCH] io_uring: Try to merge io requests only for regular files
Date: Fri, 19 Mar 2021 10:45:46 +0300 [thread overview]
Message-ID: <61611616138730@mail.yandex-team.ru> (raw)
In-Reply-To: <20210319052859.3220-1-dmtrmonakhov@yandex-team.ru>
[-- Attachment #1: Type: text/plain, Size: 1311 bytes --]
- stable@
19.03.2021, 08:29, "Dmitry Monakhov" <dmtrmonakhov@yandex-team.ru>:
> Otherwise we may endup blocking on pipe or socket.
>
> Fixes: 6d5d5ac ("io_uring: extend async work merge")
> Testcase: https://github.com/dmonakhov/liburing/commit/16d171b6ef9d68e6db66650a83d98c5c721d01f6
> Signed-off-by: Dmitry Monakhov <dmtrmonakhov@yandex-team.ru>
> ---
> fs/io_uring.c | 3 +++
> 1 file changed, 3 insertions(+)
>
> diff --git a/fs/io_uring.c b/fs/io_uring.c
> index 478df7e..848657c 100644
> --- a/fs/io_uring.c
> +++ b/fs/io_uring.c
> @@ -2183,6 +2183,9 @@ static int __io_submit_sqe(struct io_ring_ctx *ctx, struct io_kiocb *req,
> static struct async_list *io_async_list_from_req(struct io_ring_ctx *ctx,
> struct io_kiocb *req)
> {
> + if (!(req->flags & REQ_F_ISREG))
> + return NULL;
> +
IMHO it is reasonable to completely disable io_should_merge logic because
even with the this fix it still affected by latency spikes like follows:
->submit_read: req1[slow_hdd, sector=z]
->submit_read: req2[nvme, sector=X]
-> wait(req2) -> fast
->submit_read: req3[nvme, sector=Y]
--> wait(req3) ->slow
if completes if X and Y belongs to same page merge logic will wait for req1 to complete
[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: 0001-io_uring-completely-disable-io_should_merge-logic.patch --]
[-- Type: text/x-diff; name="0001-io_uring-completely-disable-io_should_merge-logic.patch", Size: 780 bytes --]
From c24cda9aa902775e4e23575b758f009e9fae4640 Mon Sep 17 00:00:00 2001
From: Dmitry Monakhov <dmtrmonakhov@yandex-team.ru>
Date: Fri, 19 Mar 2021 08:48:18 +0300
Subject: [PATCH] io_uring: completely disable io_should_merge logic
io_should_merge logic is wierd and prone to latency spikes because of slow neighbors.
Signed-off-by: Dmitry Monakhov <dmtrmonakhov@yandex-team.ru>
diff --git a/fs/io_uring.c b/fs/io_uring.c
index 478df7e..a4fb3a7 100644
--- a/fs/io_uring.c
+++ b/fs/io_uring.c
@@ -1288,7 +1288,7 @@ static ssize_t io_import_iovec(struct io_ring_ctx *ctx, int rw,
static inline bool io_should_merge(struct async_list *al, struct kiocb *kiocb)
{
- if (al->file == kiocb->ki_filp) {
+ if (al->file == kiocb->ki_filp && 0) {
off_t start, end;
/*
--
2.7.4
next prev parent reply other threads:[~2021-03-19 7:48 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-03-19 5:28 [PATCH] io_uring: Try to merge io requests only for regular files Dmitry Monakhov
2021-03-19 6:36 ` Greg KH
2021-03-19 7:45 ` Dmitry Monakhov [this message]
2021-03-19 10:23 ` Pavel Begunkov
2021-03-19 14:06 ` Matthew Wilcox
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=61611616138730@mail.yandex-team.ru \
--to=dmtrmonakhov@yandex-team.ru \
--cc=axboe@kernel.dk \
--cc=linux-fsdevel@vger.kernel.org \
--cc=linux-kernel@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;
as well as URLs for NNTP newsgroup(s).