From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-5.3 required=3.0 tests=DKIM_SIGNED,DKIM_VALID, DKIM_VALID_AU,FREEMAIL_FORGED_FROMDOMAIN,FREEMAIL_FROM, HEADER_FROM_DIFFERENT_DOMAINS,INCLUDES_PATCH,MAILING_LIST_MULTI,SPF_HELO_NONE, SPF_PASS,USER_AGENT_SANE_1 autolearn=ham autolearn_force=no version=3.4.0 Received: from mail.kernel.org (mail.kernel.org [198.145.29.99]) by smtp.lore.kernel.org (Postfix) with ESMTP id B6004C432C1 for ; Tue, 24 Sep 2019 09:33:24 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.kernel.org (Postfix) with ESMTP id 6F1DA2146E for ; Tue, 24 Sep 2019 09:33:24 +0000 (UTC) Authentication-Results: mail.kernel.org; dkim=pass (2048-bit key) header.d=gmail.com header.i=@gmail.com header.b="bnSVC199" Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S2395176AbfIXJdY (ORCPT ); Tue, 24 Sep 2019 05:33:24 -0400 Received: from mail-wr1-f65.google.com ([209.85.221.65]:41017 "EHLO mail-wr1-f65.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S2390805AbfIXJdX (ORCPT ); Tue, 24 Sep 2019 05:33:23 -0400 Received: by mail-wr1-f65.google.com with SMTP id h7so1114940wrw.8; Tue, 24 Sep 2019 02:33:19 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20161025; h=to:cc:references:from:autocrypt:subject:message-id:date:user-agent :mime-version:in-reply-to; bh=C6yH7/gGjBhKqOMW5KC8HwPnJ4drmejlx8w5NBDl5W4=; b=bnSVC199YlDC8wYmEePJJvNzUHHIMEXXOTvB3FOxpe5uzgOHsOaPPwMFflcCws0GMd rguZKSOJEiIMufLFyFzTWmCKHwlVX6Q9CTihCF1XX1L7cA0KCFtgQ1vTg3ZKI3YVZcNb FPS9eihNt0nfhpwF7vA43+4QwtJVo+bhm/9r/92YaSSJ0URedpHnPHqpo6XAWKksJaIb v5I5nJgOrvMiBg3W/fv1bN2W7kiFXbz4HxAArO+uifdhf1pLpxXzLsUMrwSw1/FB4j9B dMPkf9vLk8k67maNrskglxmVI9zO8UHXq5Ij+giB0cQ1MrcoFGXz0oVhpHrQZr6gAkHW b35A== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:to:cc:references:from:autocrypt:subject :message-id:date:user-agent:mime-version:in-reply-to; bh=C6yH7/gGjBhKqOMW5KC8HwPnJ4drmejlx8w5NBDl5W4=; b=IeZkEQAV4RGo6k2Y9RG0qfyTdGNqaQ7+5vzr3BG7Npg7GzAh6jCIujeAa5MZeBUcIO yK00kftAp1neg/gKpkepXFX9DezJ+xXBmJv+LYmviJZoXOdxq7xzNrf/0vz3tZQv29Ux WHqjrY9KCR5pnxSgA8ABbDlOOE+pPJOH8T6rgCQ7NF6FTXwoGn/jI/fG/grokbzqZyxt OsTvD8l+1mXOA06kLGfGkvqLFRzo0fuqpulX8iYi04BDxWsrPkcp+pZuV9HsTqvd7iPS Hou8QznjYBq3iXpGoChnDkSNu+tD3yuCHIrgBWDNwp115ZqGedGCjK2HHof0rr0B/Ba3 n7Xw== X-Gm-Message-State: APjAAAXReYcAxhmNdiiGUJuw0sLSzRT0T4fttS+a6TRsN/2S67WpJw7E DWbbJWpO60ntg5zRd56Z7NCZAc2vzY1HUA== X-Google-Smtp-Source: APXvYqyUj08no09jND8EC7UUrse3ElPHO9PljGRmK0Riv8bJcshOgyIh0MFuVkkipnecUjvRbvCffw== X-Received: by 2002:a5d:5183:: with SMTP id k3mr1424427wrv.55.1569317598076; Tue, 24 Sep 2019 02:33:18 -0700 (PDT) Received: from [192.168.1.75] ([65.39.69.237]) by smtp.gmail.com with ESMTPSA id g185sm2994575wme.10.2019.09.24.02.33.17 (version=TLS1_3 cipher=TLS_AES_128_GCM_SHA256 bits=128/128); Tue, 24 Sep 2019 02:33:17 -0700 (PDT) To: Jens Axboe , Ingo Molnar Cc: Ingo Molnar , Peter Zijlstra , linux-block@vger.kernel.org, linux-kernel@vger.kernel.org References: <20190923083549.GA42487@gmail.com> <731b2087-7786-5374-68ff-8cba42f0cd68@kernel.dk> <759b9b48-1de3-1d43-3e39-9c530bfffaa0@kernel.dk> <43244626-9cfd-0c0b-e7a1-878363712ef3@gmail.com> From: Pavel Begunkov Autocrypt: addr=asml.silence@gmail.com; prefer-encrypt=mutual; keydata= mQINBFmKBOQBEAC76ZFxLAKpDw0bKQ8CEiYJRGn8MHTUhURL02/7n1t0HkKQx2K1fCXClbps bdwSHrhOWdW61pmfMbDYbTj6ZvGRvhoLWfGkzujB2wjNcbNTXIoOzJEGISHaPf6E2IQx1ik9 6uqVkK1OMb7qRvKH0i7HYP4WJzYbEWVyLiAxUj611mC9tgd73oqZ2pLYzGTqF2j6a/obaqha +hXuWTvpDQXqcOZJXIW43atprH03G1tQs7VwR21Q1eq6Yvy2ESLdc38EqCszBfQRMmKy+cfp W3U9Mb1w0L680pXrONcnlDBCN7/sghGeMHjGKfNANjPc+0hzz3rApPxpoE7HC1uRiwC4et83 CKnncH1l7zgeBT9Oa3qEiBlaa1ZCBqrA4dY+z5fWJYjMpwI1SNp37RtF8fKXbKQg+JuUjAa9 Y6oXeyEvDHMyJYMcinl6xCqCBAXPHnHmawkMMgjr3BBRzODmMr+CPVvnYe7BFYfoajzqzq+h EyXSl3aBf0IDPTqSUrhbmjj5OEOYgRW5p+mdYtY1cXeK8copmd+fd/eTkghok5li58AojCba jRjp7zVOLOjDlpxxiKhuFmpV4yWNh5JJaTbwCRSd04sCcDNlJj+TehTr+o1QiORzc2t+N5iJ NbILft19Izdn8U39T5oWiynqa1qCLgbuFtnYx1HlUq/HvAm+kwARAQABtDFQYXZlbCBCZWd1 bmtvdiAoc2lsZW5jZSkgPGFzbWwuc2lsZW5jZUBnbWFpbC5jb20+iQJOBBMBCAA4FiEE+6Ju PTjTbx479o3OWt5b1Glr+6UFAlmKBOQCGwMFCwkIBwIGFQgJCgsCBBYCAwECHgECF4AACgkQ Wt5b1Glr+6WxZA//QueaKHzgdnOikJ7NA/Vq8FmhRlwgtP0+E+w93kL+ZGLzS/cUCIjn2f4Q Mcutj2Neg0CcYPX3b2nJiKr5Vn0rjJ/suiaOa1h1KzyNTOmxnsqE5fmxOf6C6x+NKE18I5Jy xzLQoktbdDVA7JfB1itt6iWSNoOTVcvFyvfe5ggy6FSCcP+m1RlR58XxVLH+qlAvxxOeEr/e aQfUzrs7gqdSd9zQGEZo0jtuBiB7k98t9y0oC9Jz0PJdvaj1NZUgtXG9pEtww3LdeXP/TkFl HBSxVflzeoFaj4UAuy8+uve7ya/ECNCc8kk0VYaEjoVrzJcYdKP583iRhOLlZA6HEmn/+Gh9 4orG67HNiJlbFiW3whxGizWsrtFNLsSP1YrEReYk9j1SoUHHzsu+ZtNfKuHIhK0sU07G1OPN 2rDLlzUWR9Jc22INAkhVHOogOcc5ajMGhgWcBJMLCoi219HlX69LIDu3Y34uIg9QPZIC2jwr 24W0kxmK6avJr7+n4o8m6sOJvhlumSp5TSNhRiKvAHB1I2JB8Q1yZCIPzx+w1ALxuoWiCdwV M/azguU42R17IuBzK0S3hPjXpEi2sK/k4pEPnHVUv9Cu09HCNnd6BRfFGjo8M9kZvw360gC1 reeMdqGjwQ68o9x0R7NBRrtUOh48TDLXCANAg97wjPoy37dQE7e5Ag0EWYoE5AEQAMWS+aBV IJtCjwtfCOV98NamFpDEjBMrCAfLm7wZlmXy5I6o7nzzCxEw06P2rhzp1hIqkaab1kHySU7g dkpjmQ7Jjlrf6KdMP87mC/Hx4+zgVCkTQCKkIxNE76Ff3O9uTvkWCspSh9J0qPYyCaVta2D1 Sq5HZ8WFcap71iVO1f2/FEHKJNz/YTSOS/W7dxJdXl2eoj3gYX2UZNfoaVv8OXKaWslZlgqN jSg9wsTv1K73AnQKt4fFhscN9YFxhtgD/SQuOldE5Ws4UlJoaFX/yCoJL3ky2kC0WFngzwRF Yo6u/KON/o28yyP+alYRMBrN0Dm60FuVSIFafSqXoJTIjSZ6olbEoT0u17Rag8BxnxryMrgR dkccq272MaSS0eOC9K2rtvxzddohRFPcy/8bkX+t2iukTDz75KSTKO+chce62Xxdg62dpkZX xK+HeDCZ7gRNZvAbDETr6XI63hPKi891GeZqvqQVYR8e+V2725w+H1iv3THiB1tx4L2bXZDI DtMKQ5D2RvCHNdPNcZeldEoJwKoA60yg6tuUquvsLvfCwtrmVI2rL2djYxRfGNmFMrUDN1Xq F3xozA91q3iZd9OYi9G+M/OA01husBdcIzj1hu0aL+MGg4Gqk6XwjoSxVd4YT41kTU7Kk+/I 5/Nf+i88ULt6HanBYcY/+Daeo/XFABEBAAGJAjYEGAEIACAWIQT7om49ONNvHjv2jc5a3lvU aWv7pQUCWYoE5AIbDAAKCRBa3lvUaWv7pfmcEACKTRQ28b1y5ztKuLdLr79+T+LwZKHjX++P 4wKjEOECCcB6KCv3hP+J2GCXDOPZvdg/ZYZafqP68Yy8AZqkfa4qPYHmIdpODtRzZSL48kM8 LRzV8Rl7J3ItvzdBRxf4T/Zseu5U6ELiQdCUkPGsJcPIJkgPjO2ROG/ZtYa9DvnShNWPlp+R uPwPccEQPWO/NP4fJl2zwC6byjljZhW5kxYswGMLBwb5cDUZAisIukyAa8Xshdan6C2RZcNs rB3L7vsg/R8UCehxOH0C+NypG2GqjVejNZsc7bgV49EOVltS+GmGyY+moIzxsuLmT93rqyII 5rSbbcTLe6KBYcs24XEoo49Zm9oDA3jYvNpeYD8rDcnNbuZh9kTgBwFN41JHOPv0W2FEEWqe JsCwQdcOQ56rtezdCJUYmRAt3BsfjN3Jn3N6rpodi4Dkdli8HylM5iq4ooeb5VkQ7UZxbCWt UVMKkOCdFhutRmYp0mbv2e87IK4erwNHQRkHUkzbsuym8RVpAZbLzLPIYK/J3RTErL6Z99N2 m3J6pjwSJY/zNwuFPs9zGEnRO4g0BUbwGdbuvDzaq6/3OJLKohr5eLXNU3JkT+3HezydWm3W OPhauth7W0db74Qd49HXK0xe/aPrK+Cp+kU1HRactyNtF8jZQbhMCC8vMGukZtWaAwpjWiiH bA== Subject: Re: [PATCH v2 0/2] Optimise io_uring completion waiting Message-ID: <0fec66fb-4534-59f8-cd88-d8d2297779aa@gmail.com> Date: Tue, 24 Sep 2019 12:33:13 +0300 User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:68.0) Gecko/20100101 Thunderbird/68.1.0 MIME-Version: 1.0 In-Reply-To: Content-Type: multipart/signed; micalg=pgp-sha256; protocol="application/pgp-signature"; boundary="8PEOQlqvbByxt9af5p0EE8WiixlcvoZMA" Sender: linux-block-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-block@vger.kernel.org This is an OpenPGP/MIME signed message (RFC 4880 and 3156) --8PEOQlqvbByxt9af5p0EE8WiixlcvoZMA Content-Type: multipart/mixed; boundary="seaUA6HPaCZDrdidUIQSqHzQzewtApgsq"; protected-headers="v1" From: Pavel Begunkov To: Jens Axboe , Ingo Molnar Cc: Ingo Molnar , Peter Zijlstra , linux-block@vger.kernel.org, linux-kernel@vger.kernel.org Message-ID: <0fec66fb-4534-59f8-cd88-d8d2297779aa@gmail.com> Subject: Re: [PATCH v2 0/2] Optimise io_uring completion waiting References: <20190923083549.GA42487@gmail.com> <731b2087-7786-5374-68ff-8cba42f0cd68@kernel.dk> <759b9b48-1de3-1d43-3e39-9c530bfffaa0@kernel.dk> <43244626-9cfd-0c0b-e7a1-878363712ef3@gmail.com> In-Reply-To: --seaUA6HPaCZDrdidUIQSqHzQzewtApgsq Content-Type: text/plain; charset=utf-8 Content-Language: en-US Content-Transfer-Encoding: quoted-printable On 24/09/2019 11:36, Jens Axboe wrote: > On 9/24/19 2:27 AM, Jens Axboe wrote: >> On 9/24/19 2:02 AM, Jens Axboe wrote: >>> On 9/24/19 1:06 AM, Pavel Begunkov wrote: >>>> On 24/09/2019 02:00, Jens Axboe wrote: >>>>>> I think we can do the same thing, just wrapping the waitqueue in a= >>>>>> structure with a count in it, on the stack. Got some flight time >>>>>> coming up later today, let me try and cook up a patch. >>>>> >>>>> Totally untested, and sent out 5 min before departure... But someth= ing >>>>> like this. >>>> Hmm, reminds me my first version. Basically that's the same thing bu= t >>>> with macroses inlined. I wanted to make it reusable and self-contain= ed, >>>> though. >>>> >>>> If you don't think it could be useful in other places, sure, we coul= d do >>>> something like that. Is that so? >>> >>> I totally agree it could be useful in other places. Maybe formalized = and >>> used with wake_up_nr() instead of adding a new primitive? Haven't loo= ked >>> into that, I may be talking nonsense. >>> >>> In any case, I did get a chance to test it and it works for me. Here'= s >>> the "finished" version, slightly cleaned up and with a comment added >>> for good measure. >> >> Notes: >> >> This version gets the ordering right, you need exclusive waits to get >> fifo ordering on the waitqueue. >> >> Both versions (yours and mine) suffer from the problem of potentially >> waking too many. I don't think this is a real issue, as generally we >> don't do threaded access to the io_urings. But if you had the followin= g >> tasks wait on the cqring: >> >> [min_events =3D 32], [min_events =3D 8], [min_events =3D 8] >> >> and we reach the io_cqring_events() =3D=3D threshold, we'll wake all t= hree. >> I don't see a good solution to this, so I suspect we just live with >> until proven an issue. Both versions are much better than what we have= >> now. >=20 > Forgot an issue around signal handling, version below adds the > right check for that too. It seems to be a good reason to not keep reimplementing "prepare_to_wait*() + wait loop" every time, but keep it in sched :) >=20 > Curious what your test case was for this? You mean a performance test case? It's briefly described in a comment for the second patch. That's just rewritten io_uring-bench, with 1. a thread generating 1 request per call in a loop 2. and the second thread waiting for ~128 events. Both are pinned to the same core. >=20 >=20 > diff --git a/fs/io_uring.c b/fs/io_uring.c > index ca7570aca430..3fbab5692f14 100644 > --- a/fs/io_uring.c > +++ b/fs/io_uring.c > @@ -2768,6 +2768,42 @@ static int io_ring_submit(struct io_ring_ctx *ct= x, unsigned int to_submit, > return submit; > } > =20 > +struct io_wait_queue { > + struct wait_queue_entry wq; > + struct io_ring_ctx *ctx; > + struct task_struct *task; > + unsigned to_wait; > + unsigned nr_timeouts; > +}; > + > +static inline bool io_should_wake(struct io_wait_queue *iowq) > +{ > + struct io_ring_ctx *ctx =3D iowq->ctx; > + > + /* > + * Wake up if we have enough events, or if a timeout occured since we= > + * started waiting. For timeouts, we always want to return to userspa= ce, > + * regardless of event count. > + */ > + return io_cqring_events(ctx->rings) >=3D iowq->to_wait || > + atomic_read(&ctx->cq_timeouts) !=3D iowq->nr_timeouts; > +} > + > +static int io_wake_function(struct wait_queue_entry *curr, unsigned in= t mode, > + int wake_flags, void *key) > +{ > + struct io_wait_queue *iowq =3D container_of(curr, struct io_wait_queu= e, > + wq); > + > + if (io_should_wake(iowq)) { > + list_del_init(&curr->entry); > + wake_up_process(iowq->task); > + return 1; > + } > + > + return -1; > +} > + > /* > * Wait until events become available, if we don't already have some. = The > * application must reap them itself, as they reside on the shared cq = ring. > @@ -2775,8 +2811,16 @@ static int io_ring_submit(struct io_ring_ctx *ct= x, unsigned int to_submit, > static int io_cqring_wait(struct io_ring_ctx *ctx, int min_events, > const sigset_t __user *sig, size_t sigsz) > { > + struct io_wait_queue iowq =3D { > + .wq =3D { > + .func =3D io_wake_function, > + .entry =3D LIST_HEAD_INIT(iowq.wq.entry), > + }, > + .task =3D current, > + .ctx =3D ctx, > + .to_wait =3D min_events, > + }; > struct io_rings *rings =3D ctx->rings; > - unsigned nr_timeouts; > int ret; > =20 > if (io_cqring_events(rings) >=3D min_events) > @@ -2795,15 +2839,18 @@ static int io_cqring_wait(struct io_ring_ctx *c= tx, int min_events, > return ret; > } > =20 > - nr_timeouts =3D atomic_read(&ctx->cq_timeouts); > - /* > - * Return if we have enough events, or if a timeout occured since > - * we started waiting. For timeouts, we always want to return to > - * userspace. > - */ > - ret =3D wait_event_interruptible(ctx->wait, > - io_cqring_events(rings) >=3D min_events || > - atomic_read(&ctx->cq_timeouts) !=3D nr_timeouts); > + iowq.nr_timeouts =3D atomic_read(&ctx->cq_timeouts); > + prepare_to_wait_exclusive(&ctx->wait, &iowq.wq, TASK_INTERRUPTIBLE); > + do { > + if (io_should_wake(&iowq)) > + break; > + schedule(); > + if (signal_pending(current)) > + break; > + set_current_state(TASK_INTERRUPTIBLE); > + } while (1); > + finish_wait(&ctx->wait, &iowq.wq); > + > restore_saved_sigmask_unless(ret =3D=3D -ERESTARTSYS); > if (ret =3D=3D -ERESTARTSYS) > ret =3D -EINTR; >=20 --=20 Yours sincerely, Pavel Begunkov --seaUA6HPaCZDrdidUIQSqHzQzewtApgsq-- --8PEOQlqvbByxt9af5p0EE8WiixlcvoZMA Content-Type: application/pgp-signature; name="signature.asc" Content-Description: OpenPGP digital signature Content-Disposition: attachment; filename="signature.asc" -----BEGIN PGP SIGNATURE----- iQIzBAEBCAAdFiEE+6JuPTjTbx479o3OWt5b1Glr+6UFAl2J4tkACgkQWt5b1Glr +6VYdQ/+Mx7N6LNvYiwc6C1BuUEUmj5N9HHtez7ikgVX8bPO2uQc3+PmSr9l3hHc eEFsuPXWR9ClMemtRaFyX47BzOaQZX/JuffN1wr8MeKd8OXeSQLK0Ymrn3bMKdTP RVv8ROS8Qbb9EyCjC913DiUkxNmZz+Wm45+yg40OQ0doRyo2hcWmIsVvr1YoY/Y0 WxdRU7TKUNrD8NwST1grQLLBKg+cMKGR1XYy7SFJf+TJM3DJ5Awy2V2aqVJXx+qI WH0iHtReqYwTNCXEZJk5AA7KM3ukApM96Kzg8LbbL0Vjb94T2TQ8MBpQ9VfQbeT/ H/8DMpqauWqJqlVyzsaVuNfDOlkFY2YsA2ZD+EWP3E4VfF+oYFNV/xE+SmCvjXjh gux59p2wJyKMAMsZrE4/1PimE4NdMehjEVSYDJZ8epl0nOzUieUzdpdZxh69Yxy7 7eguBIDwtc942i+HYiwQmDM6czU3vk+I1/0t01Ua9RgO16+Whbfu33iSBhXdY4yc rQF1Dy5EyHjvKDU94jTDxB3xzUwCTbzHV6eB29Y+dGIkBGsoxf8EDTeaeKKy8O4W jPg/2527/e/LoM4halXbDOMIpIdVxZJykLm88MeJybgIY+b5qR5CmHygCEbqcYVr YW0+ds9FXcLro19WuTeU1aqkUfebclaT4bmBtDLoU7bfxATOZbY= =u/dF -----END PGP SIGNATURE----- --8PEOQlqvbByxt9af5p0EE8WiixlcvoZMA--