All of lore.kernel.org
 help / color / mirror / Atom feed
From: Jens Axboe <axboe@kernel.dk>
To: io-uring@vger.kernel.org, linux-block@vger.kernel.org
Cc: Jens Axboe <axboe@kernel.dk>
Subject: [PATCH 3/3] block: enable bio allocation cache for IRQ driven IO
Date: Wed, 15 Dec 2021 09:30:09 -0700	[thread overview]
Message-ID: <20211215163009.15269-4-axboe@kernel.dk> (raw)
In-Reply-To: <20211215163009.15269-1-axboe@kernel.dk>

We currently cannot use the bio recycling allocation cache for IRQ driven
IO, as the cache isn't IRQ safe (by design).

Add a way for the completion side to pass back a bio that needs freeing,
so we can do it from the io_uring side. io_uring completions always
run in task context.

This is good for about a 13% improvement in IRQ driven IO, taking us from
around 6.3M/core to 7.1M/core IOPS.

Signed-off-by: Jens Axboe <axboe@kernel.dk>
---
 block/fops.c       | 11 ++++++++---
 fs/io_uring.c      |  6 ++++++
 include/linux/fs.h |  4 ++++
 3 files changed, 18 insertions(+), 3 deletions(-)

diff --git a/block/fops.c b/block/fops.c
index bcf866b07edc..c7794d42be85 100644
--- a/block/fops.c
+++ b/block/fops.c
@@ -296,14 +296,19 @@ static void blkdev_bio_end_io_async(struct bio *bio)
 		ret = blk_status_to_errno(bio->bi_status);
 	}
 
-	iocb->ki_complete(iocb, ret);
-
 	if (dio->flags & DIO_SHOULD_DIRTY) {
 		bio_check_pages_dirty(bio);
 	} else {
 		bio_release_pages(bio, false);
-		bio_put(bio);
+		if (iocb->ki_flags & IOCB_BIO_PASSBACK) {
+			iocb->ki_flags |= IOCB_PRIV_IS_BIO;
+			iocb->private = bio;
+		} else {
+			bio_put(bio);
+		}
 	}
+
+	iocb->ki_complete(iocb, ret);
 }
 
 static ssize_t __blkdev_direct_IO_async(struct kiocb *iocb,
diff --git a/fs/io_uring.c b/fs/io_uring.c
index 1a647a6a5add..b0302c0407e6 100644
--- a/fs/io_uring.c
+++ b/fs/io_uring.c
@@ -2828,6 +2828,11 @@ static void io_req_task_complete(struct io_kiocb *req, bool *locked)
 	unsigned int cflags = io_put_kbuf(req);
 	int res = req->result;
 
+#ifdef CONFIG_BLOCK
+	if (req->rw.kiocb.ki_flags & IOCB_PRIV_IS_BIO)
+		bio_put(req->rw.kiocb.private);
+#endif
+
 	if (*locked) {
 		io_req_complete_state(req, res, cflags);
 		io_req_add_compl_list(req);
@@ -3024,6 +3029,7 @@ static int io_prep_rw(struct io_kiocb *req, const struct io_uring_sqe *sqe)
 	} else {
 		if (kiocb->ki_flags & IOCB_HIPRI)
 			return -EINVAL;
+		kiocb->ki_flags |= IOCB_ALLOC_CACHE | IOCB_BIO_PASSBACK;
 		kiocb->ki_complete = io_complete_rw;
 	}
 
diff --git a/include/linux/fs.h b/include/linux/fs.h
index 6b8dc1a78df6..bf7a76dfdc29 100644
--- a/include/linux/fs.h
+++ b/include/linux/fs.h
@@ -322,6 +322,10 @@ enum rw_hint {
 #define IOCB_NOIO		(1 << 20)
 /* can use bio alloc cache */
 #define IOCB_ALLOC_CACHE	(1 << 21)
+/* iocb supports bio passback */
+#define IOCB_BIO_PASSBACK	(1 << 22)
+/* iocb->private holds bio to put */
+#define IOCB_PRIV_IS_BIO	(1 << 23)
 
 struct kiocb {
 	struct file		*ki_filp;
-- 
2.34.1


  parent reply	other threads:[~2021-12-15 16:30 UTC|newest]

Thread overview: 8+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-12-15 16:30 [PATCHSET 0/3] Improve IRQ driven performance Jens Axboe
2021-12-15 16:30 ` [PATCH 1/3] block: add completion handler for fast path Jens Axboe
2021-12-16  9:10   ` Christoph Hellwig
2021-12-15 16:30 ` [PATCH 2/3] block: use singly linked list for bio cache Jens Axboe
2021-12-16  9:11   ` Christoph Hellwig
2021-12-15 16:30 ` Jens Axboe [this message]
2021-12-16 14:33   ` [PATCH 3/3] block: enable bio allocation cache for IRQ driven IO Christoph Hellwig
2021-12-16 15:41     ` 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=20211215163009.15269-4-axboe@kernel.dk \
    --to=axboe@kernel.dk \
    --cc=io-uring@vger.kernel.org \
    --cc=linux-block@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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.