[<prev] [next>] [thread-next>] [day] [month] [year] [list]
Message-ID: <20210907142145.112096-1-Niklas.Cassel@wdc.com>
Date: Tue, 7 Sep 2021 14:21:54 +0000
From: Niklas Cassel <Niklas.Cassel@....com>
To: Jens Axboe <axboe@...nel.dk>
CC: Bart Van Assche <bvanassche@....org>,
Damien Le Moal <Damien.LeMoal@....com>,
Paolo Valente <paolo.valente@...aro.org>,
Ming Lei <ming.lei@...hat.com>,
Niklas Cassel <Niklas.Cassel@....com>,
"linux-block@...r.kernel.org" <linux-block@...r.kernel.org>,
"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: [PATCH 0/2] don't call io scheduler callbacks for passthrough
requests
From: Niklas Cassel <niklas.cassel@....com>
Recently, there were some extra "dummy" data added into mq-deadline
rq->elv.priv, that was added for the solve purpose of working around
quirky behavior in blk-mq (which called the scheduler callbacks even
for requests that bypassed the scheduler).
Fix blk-mq to not call the I/O scheduler callbacks for passthrough requests,
since they will later bypass the I/O scheduler anyway.
This way, we can remove unnecessary logic in mq-deadline that was added to
deal with this quirky behavior.
BFQ should be able to perform a similar cleanup, if they wish, and should
then be able to drop this comment from the code:
https://github.com/torvalds/linux/blob/v5.14/block/bfq-iosched.c#L6456-L6462
Considering that I'm quite unfamiliar with the BFQ code, I'd rather
someone who is a bit more familiar with BFQ performs that cleanup.
Kind regards,
Niklas
Niklas Cassel (2):
blk-mq: don't call callbacks for requests that bypassed the scheduler
Revert "mq-deadline: Fix request accounting"
block/blk-mq.c | 7 ++++++-
block/mq-deadline.c | 16 +++++-----------
2 files changed, 11 insertions(+), 12 deletions(-)
--
2.31.1
Powered by blists - more mailing lists