[PATCHSET,0/8] block plugging improvements

Message ID	20181126163556.5181-1-axboe@kernel.dk (mailing list archive)
Headers	show Return-Path: <linux-block-owner@kernel.org> From: Jens Axboe <axboe@kernel.dk> To: linux-block@vger.kernel.org, linux-nvme@lists.infradead.org Subject: [PATCHSET 0/8] block plugging improvements Date: Mon, 26 Nov 2018 09:35:48 -0700 Message-Id: <20181126163556.5181-1-axboe@kernel.dk> Sender: linux-block-owner@vger.kernel.org Precedence: bulk
Series	block plugging improvements \| expand [PATCHSET,0/8] block plugging improvements [1/8] block: sum requests in the plug structure [2/8] block: improve logic around when to sort a plug list [3/8] blk-mq: add mq_ops->commit_rqs() [4/8] nvme: implement mq_ops->commit_rqs() hook [5/8] virtio_blk: implement mq_ops->commit_rqs() hook [6/8] ataflop: implement mq_ops->commit_rqs() hook [7/8] blk-mq: use bd->last == true for list inserts [8/8] blk-mq: add plug case for devices that implement ->commits_rqs()

Message ID

20181126163556.5181-1-axboe@kernel.dk (mailing list archive)

Headers

From: Jens Axboe <axboe@kernel.dk>
To: linux-block@vger.kernel.org, linux-nvme@lists.infradead.org
Subject: [PATCHSET 0/8] block plugging improvements
Date: Mon, 26 Nov 2018 09:35:48 -0700
Message-Id: <20181126163556.5181-1-axboe@kernel.dk>
Sender: linux-block-owner@vger.kernel.org
Precedence: bulk

Series

block plugging improvements | expand

Message

Jens Axboe Nov. 26, 2018, 4:35 p.m. UTC

Series improving plugging for fast devices, but some fixes in here too.

1-2 are improvements around plugging accounting. Changes the behavior
a bit, but works fine for me.

3-6 add a ->commit_rqs() hook and implement it in drivers that use (or
will use) bd->last to optimize IO submission. If a driver currently uses
bd->last to know if it's needed to kick the hardware into action, there
are cases where we flag bd->last == false, but then fail to submit any
further IO due to other resource constraints. We probably get saved by
the fact that this happens for the case where we have pending IO and
that will eventually guarantee forward progress, but we really should
kick IO into gear at that point.

7-8 improve plugging for blk-mq.

In terms of improvements, for virtualized nvme, I've seen a 2-x IOPS
improvement with proper handling of bd->last with this series.