From: Sagi Grimberg <sagi@grimberg.me>
To: Jens Axboe <axboe@kernel.dk>, Christoph Hellwig <hch@lst.de>
Cc: Jeffle Xu <jefflexu@linux.alibaba.com>,
Ming Lei <ming.lei@redhat.com>,
Damien Le Moal <Damien.LeMoal@wdc.com>,
Keith Busch <kbusch@kernel.org>,
"Wunderlich, Mark" <mark.wunderlich@intel.com>,
"Vasudevan, Anil" <anil.vasudevan@intel.com>,
linux-block@vger.kernel.org, linux-fsdevel@vger.kernel.org,
linux-nvme@lists.infradead.org
Subject: Re: switch block layer polling to a bio based model v4
Date: Tue, 12 Oct 2021 17:57:43 +0300 [thread overview]
Message-ID: <040104f6-720d-35ed-7e15-a704e6488fd4@grimberg.me> (raw)
In-Reply-To: <07f31547-5570-4150-7a4b-1d773fb9fa87@kernel.dk>
>> Hi all,
>>
>> This series clean up the block polling code a bit and changes the interface
>> to poll for a specific bio instead of a request_queue and cookie pair.
>>
>> Polling for the bio itself leads to a few advantages:
>>
>> - the cookie construction can made entirely private in blk-mq.c
>> - the caller does not need to remember the request_queue and cookie
>> separately and thus sidesteps their lifetime issues
>> - keeping the device and the cookie inside the bio allows to trivially
>> support polling BIOs remapping by stacking drivers
>> - a lot of code to propagate the cookie back up the submission path can
>> removed entirely
>>
>> The one major caveat is that this requires RCU freeing polled BIOs to make
>> sure the bio that contains the polling information is still alive when
>> io_uring tries to poll it through the iocb. For synchronous polling all the
>> callers have a bio reference anyway, so this is not an issue.
>
> I ran this through the usual peak testing, and it doesn't seem to regress
> anything for me. We're still at around ~7.4M polled IOPS on a single CPU
> core:
>
> taskset -c 0,16 t/io_uring -d128 -b512 -s32 -c32 -p1 -F1 -B1 -D1 -n2 /dev/nvme1n1 /dev/nvme2n1
> Added file /dev/nvme1n1 (submitter 0)
> Added file /dev/nvme2n1 (submitter 1)
> polled=1, fixedbufs=1, register_files=1, buffered=0, QD=128
> Engine=io_uring, sq_ring=128, cq_ring=256
> submitter=0, tid=1199
> submitter=1, tid=1200
> IOPS=7322112, BW=3575MiB/s, IOS/call=32/31, inflight=(110 71)
> IOPS=7452736, BW=3639MiB/s, IOS/call=32/31, inflight=(52 80)
> IOPS=7419904, BW=3623MiB/s, IOS/call=32/31, inflight=(78 104)
> IOPS=7392576, BW=3609MiB/s, IOS/call=32/32, inflight=(75 102)
Jens, is that with nvme_core.multipath=Y ?
next prev parent reply other threads:[~2021-10-12 14:57 UTC|newest]
Thread overview: 38+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-10-12 11:12 switch block layer polling to a bio based model v4 Christoph Hellwig
2021-10-12 11:12 ` [PATCH 01/16] direct-io: remove blk_poll support Christoph Hellwig
2021-10-13 10:42 ` Sagi Grimberg
2021-10-28 1:26 ` Chaitanya Kulkarni
2021-10-12 11:12 ` [PATCH 02/16] block: don't try to poll multi-bio I/Os in __blkdev_direct_IO Christoph Hellwig
2021-10-12 11:12 ` [PATCH 03/16] iomap: don't try to poll multi-bio I/Os in __iomap_dio_rw Christoph Hellwig
2021-10-12 11:12 ` [PATCH 04/16] io_uring: fix a layering violation in io_iopoll_req_issued Christoph Hellwig
2021-10-12 11:12 ` [PATCH 05/16] blk-mq: factor out a blk_qc_to_hctx helper Christoph Hellwig
2021-10-12 11:12 ` [PATCH 06/16] blk-mq: factor out a "classic" poll helper Christoph Hellwig
2021-10-12 11:12 ` [PATCH 07/16] blk-mq: remove blk_qc_t_to_tag and blk_qc_t_is_internal Christoph Hellwig
2021-10-12 11:12 ` [PATCH 08/16] blk-mq: remove blk_qc_t_valid Christoph Hellwig
2021-10-12 11:12 ` [PATCH 09/16] block: replace the spin argument to blk_iopoll with a flags argument Christoph Hellwig
2021-10-13 10:44 ` Sagi Grimberg
2021-10-12 11:12 ` [PATCH 10/16] io_uring: don't sleep when polling for I/O Christoph Hellwig
2021-10-13 10:45 ` Sagi Grimberg
2021-10-12 11:12 ` [PATCH 11/16] block: rename REQ_HIPRI to REQ_POLLED Christoph Hellwig
2021-10-13 10:45 ` Sagi Grimberg
2021-10-12 11:12 ` [PATCH 12/16] block: use SLAB_TYPESAFE_BY_RCU for the bio slab Christoph Hellwig
2021-10-12 11:12 ` [PATCH 13/16] block: define 'struct bvec_iter' as packed Christoph Hellwig
2021-10-12 11:12 ` [PATCH 14/16] block: switch polling to be bio based Christoph Hellwig
2021-10-13 9:59 ` Ming Lei
2021-10-13 10:45 ` Sagi Grimberg
2021-10-15 8:30 ` Pankaj Raghav
2021-10-15 13:24 ` Christoph Hellwig
2021-11-03 7:11 ` chenxiang (M)
2021-11-03 7:22 ` Christoph Hellwig
2021-11-03 8:05 ` chenxiang (M)
2021-10-12 11:12 ` [PATCH 15/16] block: don't allow writing to the poll queue attribute Christoph Hellwig
2021-10-12 11:12 ` [PATCH 16/16] nvme-multipath: enable polled I/O Christoph Hellwig
2021-10-13 10:46 ` Sagi Grimberg
2021-10-12 14:47 ` switch block layer polling to a bio based model v4 Jens Axboe
2021-10-12 14:57 ` Sagi Grimberg [this message]
2021-10-12 14:58 ` Jens Axboe
2021-10-12 15:09 ` Sagi Grimberg
2021-10-12 15:06 ` Christoph Hellwig
-- strict thread matches above, loose matches on Subject: below --
2021-06-15 13:10 Christoph Hellwig
2021-06-15 14:37 ` Keith Busch
2021-06-15 16:06 ` Christoph Hellwig
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=040104f6-720d-35ed-7e15-a704e6488fd4@grimberg.me \
--to=sagi@grimberg.me \
--cc=Damien.LeMoal@wdc.com \
--cc=anil.vasudevan@intel.com \
--cc=axboe@kernel.dk \
--cc=hch@lst.de \
--cc=jefflexu@linux.alibaba.com \
--cc=kbusch@kernel.org \
--cc=linux-block@vger.kernel.org \
--cc=linux-fsdevel@vger.kernel.org \
--cc=linux-nvme@lists.infradead.org \
--cc=mark.wunderlich@intel.com \
--cc=ming.lei@redhat.com \
/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 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).