linux-block.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Paolo Valente <paolo.valente@linaro.org>
To: Jens Axboe <axboe@kernel.dk>
Cc: linux-block@vger.kernel.org, linux-kernel@vger.kernel.org,
	mariottiluca1@hotmail.it, holger@applied-asynchrony.com,
	pedroni.pietro.96@gmail.com,
	Paolo Valente <paolo.valente@linaro.org>
Subject: [PATCH FIXES/IMPROVEMENTS 6/7] block, bfq: check waker only for queues with no in-flight I/O
Date: Sat, 19 Jun 2021 16:09:47 +0200	[thread overview]
Message-ID: <20210619140948.98712-7-paolo.valente@linaro.org> (raw)
In-Reply-To: <20210619140948.98712-1-paolo.valente@linaro.org>

Consider two bfq_queues, say Q1 and Q2, with Q2 empty. If a request of
Q1 gets completed shortly before a new request arrives for Q2, then
BFQ flags Q1 as a candidate waker for Q2. Yet, the arrival of this new
request may have a different cause, in the following case. If also Q2
has requests in flight while waiting for the arrival of a new request,
then the completion of its own requests may be the actual cause of the
awakening of the process that sends I/O to Q2. So Q1 may be flagged
wrongly as a candidate waker.

This commit avoids this deceptive flagging, by disabling
candidate-waker flagging for Q2, if Q2 has in-flight I/O.

Signed-off-by: Paolo Valente <paolo.valente@linaro.org>
---
 block/bfq-iosched.c | 21 +++++++++++++--------
 1 file changed, 13 insertions(+), 8 deletions(-)

diff --git a/block/bfq-iosched.c b/block/bfq-iosched.c
index 7bf073ef9443..a273b2bcea2a 100644
--- a/block/bfq-iosched.c
+++ b/block/bfq-iosched.c
@@ -1985,14 +1985,18 @@ static void bfq_update_io_intensity(struct bfq_queue *bfqq, u64 now_ns)
  * Turning back to the detection of a waker queue, a queue Q is deemed
  * as a waker queue for bfqq if, for three consecutive times, bfqq
  * happens to become non empty right after a request of Q has been
- * completed. In particular, on the first time, Q is tentatively set
- * as a candidate waker queue, while on the third consecutive time
- * that Q is detected, the field waker_bfqq is set to Q, to confirm
- * that Q is a waker queue for bfqq. These detection steps are
- * performed only if bfqq has a long think time, so as to make it more
- * likely that bfqq's I/O is actually being blocked by a
- * synchronization. This last filter, plus the above three-times
- * requirement, make false positives less likely.
+ * completed. In this respect, even if bfqq is empty, we do not check
+ * for a waker if it still has some in-flight I/O. In fact, in this
+ * case bfqq is actually still being served by the drive, and may
+ * receive new I/O on the completion of some of the in-flight
+ * requests. In particular, on the first time, Q is tentatively set as
+ * a candidate waker queue, while on the third consecutive time that Q
+ * is detected, the field waker_bfqq is set to Q, to confirm that Q is
+ * a waker queue for bfqq. These detection steps are performed only if
+ * bfqq has a long think time, so as to make it more likely that
+ * bfqq's I/O is actually being blocked by a synchronization. This
+ * last filter, plus the above three-times requirement, make false
+ * positives less likely.
  *
  * NOTE
  *
@@ -2018,6 +2022,7 @@ static void bfq_check_waker(struct bfq_data *bfqd, struct bfq_queue *bfqq,
 	if (!bfqd->last_completed_rq_bfqq ||
 	    bfqd->last_completed_rq_bfqq == bfqq ||
 	    bfq_bfqq_has_short_ttime(bfqq) ||
+	    bfqq->dispatched > 0 ||
 	    now_ns - bfqd->last_completion >= 4 * NSEC_PER_MSEC ||
 	    bfqd->last_completed_rq_bfqq == bfqq->waker_bfqq)
 		return;
-- 
2.20.1


  parent reply	other threads:[~2021-06-19 14:10 UTC|newest]

Thread overview: 18+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-06-19 14:09 [PATCH FIXES/IMPROVEMENTS 0/7] block, bfq: preserve control, boost throughput, fix bugs Paolo Valente
2021-06-19 14:09 ` [PATCH FIXES/IMPROVEMENTS 1/7] block, bfq: let also stably merged queues enjoy weight raising Paolo Valente
2021-06-19 14:09 ` [PATCH FIXES/IMPROVEMENTS 2/7] block, bfq: fix delayed stable merge check Paolo Valente
2021-06-19 14:09 ` [PATCH FIXES/IMPROVEMENTS 3/7] block, bfq: consider also creation time in delayed stable merge Paolo Valente
2021-06-19 14:09 ` [PATCH FIXES/IMPROVEMENTS 4/7] block, bfq: boost throughput by extending queue-merging times Paolo Valente
2021-06-19 14:09 ` [PATCH FIXES/IMPROVEMENTS 5/7] block, bfq: avoid delayed merge of async queues Paolo Valente
2021-06-19 14:09 ` Paolo Valente [this message]
2021-06-19 14:09 ` [PATCH FIXES/IMPROVEMENTS 7/7] block, bfq: reset waker pointer with shared queues Paolo Valente
2021-06-21 16:08 ` [PATCH FIXES/IMPROVEMENTS 0/7] block, bfq: preserve control, boost throughput, fix bugs Jens Axboe
2021-06-21 19:55 ` Oleksandr Natalenko
2021-06-21 20:03   ` Piotr Górski
2021-06-22  7:08   ` Paolo Valente
2021-06-22  7:35     ` Oleksandr Natalenko
2021-06-22 16:29       ` Jan Kara
2021-06-22 17:26         ` Oleksandr Natalenko
2021-07-02 22:07         ` Oleksandr Natalenko
2021-08-02 20:40           ` Oleksandr Natalenko
2021-08-03 10:45             ` Jan Kara

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=20210619140948.98712-7-paolo.valente@linaro.org \
    --to=paolo.valente@linaro.org \
    --cc=axboe@kernel.dk \
    --cc=holger@applied-asynchrony.com \
    --cc=linux-block@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mariottiluca1@hotmail.it \
    --cc=pedroni.pietro.96@gmail.com \
    --subject='Re: [PATCH FIXES/IMPROVEMENTS 6/7] block, bfq: check waker only for queues with no in-flight I/O' \
    /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

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).