[PATCHSET,0/9] Batched completions

Message ID	20211012181742.672391-1-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 Subject: [PATCHSET 0/9] Batched completions Date: Tue, 12 Oct 2021 12:17:33 -0600 Message-Id: <20211012181742.672391-1-axboe@kernel.dk> MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Precedence: bulk
Series	Batched completions \| expand [PATCHSET,0/9] Batched completions [1/9] block: add a struct io_batch argument to fops->iopoll() [2/9] sbitmap: add helper to clear a batch of tags [3/9] sbitmap: test bit before calling test_and_set_bit() [4/9] block: add support for blk_mq_end_request_batch() [5/9] nvme: move the fast path nvme error and disposition helpers [6/9] nvme: add support for batched completion of polled IO [7/9] block: assign batch completion handler in blk_poll() [8/9] io_uring: utilize the io_batch infrastructure for more efficient polled IO [9/9] nvme: wire up completion batching for the IRQ path

Message ID

20211012181742.672391-1-axboe@kernel.dk (mailing list archive)

Headers

From: Jens Axboe <axboe@kernel.dk>
To: linux-block@vger.kernel.org
Subject: [PATCHSET 0/9] Batched completions
Date: Tue, 12 Oct 2021 12:17:33 -0600
Message-Id: <20211012181742.672391-1-axboe@kernel.dk>
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Precedence: bulk

Series

Batched completions | expand

Message

Jens Axboe Oct. 12, 2021, 6:17 p.m. UTC

Hi,

We now do decent batching of allocations for submit, but we still
complete requests individually. This costs a lot of CPU cycles.

This patchset adds support for collecting requests for completion,
and then completing them as a batch. This includes things like freeing
a batch of tags.

A few rough edges in this patchset, but sending it out for comments
so I can whip it into an upstreamable thing. As per patch 8, the
wins here are pretty massive - 8-10% improvement in the peak IOPS.