From patchwork Tue Nov 6 21:35:02 2018 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Mikulas Patocka X-Patchwork-Id: 10671527 X-Patchwork-Delegate: snitzer@redhat.com Return-Path: Received: from mail.wl.linuxfoundation.org (pdx-wl-mail.web.codeaurora.org [172.30.200.125]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id 868AF109C for ; Tue, 6 Nov 2018 21:55:18 +0000 (UTC) Received: from mail.wl.linuxfoundation.org (localhost [127.0.0.1]) by mail.wl.linuxfoundation.org (Postfix) with ESMTP id 7662E28FE3 for ; Tue, 6 Nov 2018 21:55:18 +0000 (UTC) Received: by mail.wl.linuxfoundation.org (Postfix, from userid 486) id 6A58E29143; Tue, 6 Nov 2018 21:55:18 +0000 (UTC) X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on pdx-wl-mail.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-7.9 required=2.0 tests=BAYES_00,MAILING_LIST_MULTI, RCVD_IN_DNSWL_HI autolearn=ham version=3.3.1 Received: from mx1.redhat.com (mx1.redhat.com [209.132.183.28]) (using TLSv1.2 with cipher DHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by mail.wl.linuxfoundation.org (Postfix) with ESMTPS id E1F5828FE3 for ; Tue, 6 Nov 2018 21:55:17 +0000 (UTC) Received: from smtp.corp.redhat.com (int-mx03.intmail.prod.int.phx2.redhat.com [10.5.11.13]) (using TLSv1.2 with cipher AECDH-AES256-SHA (256/256 bits)) (No client certificate requested) by mx1.redhat.com (Postfix) with ESMTPS id 131402F3D; Tue, 6 Nov 2018 21:55:17 +0000 (UTC) Received: from colo-mx.corp.redhat.com (colo-mx01.intmail.prod.int.phx2.redhat.com [10.5.11.20]) by smtp.corp.redhat.com (Postfix) with ESMTPS id CFA11AF8D9; Tue, 6 Nov 2018 21:55:16 +0000 (UTC) Received: from lists01.pubmisc.prod.ext.phx2.redhat.com (lists01.pubmisc.prod.ext.phx2.redhat.com [10.5.19.33]) by colo-mx.corp.redhat.com (Postfix) with ESMTP id 8CDB3180B5B7; Tue, 6 Nov 2018 21:55:16 +0000 (UTC) Received: from smtp.corp.redhat.com (int-mx04.intmail.prod.int.phx2.redhat.com [10.5.11.14]) by lists01.pubmisc.prod.ext.phx2.redhat.com (8.13.8/8.13.8) with ESMTP id wA6Ld7xs004262 for ; Tue, 6 Nov 2018 16:39:07 -0500 Received: by smtp.corp.redhat.com (Postfix) id 3CBFB607DE; Tue, 6 Nov 2018 21:39:07 +0000 (UTC) Delivered-To: dm-devel@redhat.com Received: from leontynka.twibright.com (ovpn-204-55.brq.redhat.com [10.40.204.55]) by smtp.corp.redhat.com (Postfix) with ESMTPS id A4E6660C4D; Tue, 6 Nov 2018 21:39:04 +0000 (UTC) Received: from debian-a64.vm ([192.168.208.2]) by leontynka.twibright.com with smtp (Exim 4.89) (envelope-from ) id 1gK93u-0005Ne-1P; Tue, 06 Nov 2018 22:39:03 +0100 Received: by debian-a64.vm (sSMTP sendmail emulation); Tue, 06 Nov 2018 22:38:57 +0100 Message-Id: <20181106213857.112447880@debian-a64.vm> User-Agent: quilt/0.65 Date: Tue, 06 Nov 2018 22:35:02 +0100 From: Mikulas Patocka To: Mike Snitzer , "Alasdair G. Kergon" MIME-Version: 1.0 Content-Disposition: inline; filename=dm-percpu.patch X-Scanned-By: MIMEDefang 2.79 on 10.5.11.14 X-loop: dm-devel@redhat.com Cc: dm-devel@redhat.com, Mikulas Patocka Subject: [dm-devel] [patch 4/5] dm: use percpu counters X-BeenThere: dm-devel@redhat.com X-Mailman-Version: 2.1.12 Precedence: junk List-Id: device-mapper development List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Sender: dm-devel-bounces@redhat.com Errors-To: dm-devel-bounces@redhat.com X-Scanned-By: MIMEDefang 2.79 on 10.5.11.13 X-Greylist: Sender IP whitelisted, not delayed by milter-greylist-4.5.16 (mx1.redhat.com [10.5.110.28]); Tue, 06 Nov 2018 21:55:17 +0000 (UTC) X-Virus-Scanned: ClamAV using ClamSMTP Use percpu inflight counters to avoid cache line bouncing and improve performance. Signed-off-by: Mikulas Patocka --- drivers/md/dm-core.h | 5 ++++ drivers/md/dm-rq.c | 10 +++++++-- drivers/md/dm.c | 52 +++++++++++++++++++++++++++++++++++---------------- 3 files changed, 49 insertions(+), 18 deletions(-) -- dm-devel mailing list dm-devel@redhat.com https://www.redhat.com/mailman/listinfo/dm-devel Index: linux-2.6/drivers/md/dm-core.h =================================================================== --- linux-2.6.orig/drivers/md/dm-core.h 2018-11-06 00:45:06.320000000 +0100 +++ linux-2.6/drivers/md/dm-core.h 2018-11-06 00:45:35.640000000 +0100 @@ -24,6 +24,10 @@ struct dm_kobject_holder { struct completion completion; }; +struct dm_percpu { + unsigned inflight[2]; +}; + /* * DM core internal structure that used directly by dm.c and dm-rq.c * DM targets must _not_ deference a mapped_device to directly access its members! @@ -63,6 +67,7 @@ struct mapped_device { /* * A list of ios that arrived while we were suspended. */ + struct dm_percpu __percpu *counters; struct work_struct work; wait_queue_head_t wait; spinlock_t deferred_lock; Index: linux-2.6/drivers/md/dm.c =================================================================== --- linux-2.6.orig/drivers/md/dm.c 2018-11-06 00:45:06.320000000 +0100 +++ linux-2.6/drivers/md/dm.c 2018-11-06 22:31:33.980000000 +0100 @@ -597,19 +597,33 @@ static void free_tio(struct dm_target_io int md_in_flight(struct mapped_device *md) { - return atomic_read(&dm_disk(md)->part0.in_flight[READ]) + - atomic_read(&dm_disk(md)->part0.in_flight[WRITE]); + int cpu; + unsigned sum = 0; + for_each_possible_cpu(cpu) { + struct dm_percpu *p = per_cpu_ptr(md->counters, cpu); + sum += p->inflight[READ] + p->inflight[WRITE]; + } + return (int)sum; } static void start_io_acct(struct dm_io *io) { struct mapped_device *md = io->md; struct bio *bio = io->orig_bio; + struct hd_struct *part; + int sgrp, cpu; io->start_time = jiffies; - generic_start_io_acct(md->queue, bio_op(bio), bio_sectors(bio), - &dm_disk(md)->part0); + part = &dm_disk(md)->part0; + sgrp = op_stat_group(bio_op(bio)); + cpu = part_stat_lock(); + part_round_stats(md->queue, cpu, part); + part_stat_inc(cpu, part, ios[sgrp]); + part_stat_add(cpu, part, sectors[sgrp], bio_sectors(bio)); + part_stat_unlock(); + + this_cpu_inc(md->counters->inflight[bio_data_dir(bio)]); if (unlikely(dm_stats_used(&md->stats))) dm_stats_account_io(&md->stats, bio_data_dir(bio), @@ -622,25 +636,25 @@ static void end_io_acct(struct dm_io *io struct mapped_device *md = io->md; struct bio *bio = io->orig_bio; unsigned long duration = jiffies - io->start_time; + struct hd_struct *part; + int sgrp, cpu; - /* - * make sure that atomic_dec in generic_end_io_acct is not reordered - * with previous writes - */ - smp_mb__before_atomic(); - generic_end_io_acct(md->queue, bio_op(bio), &dm_disk(md)->part0, - io->start_time); - /* - * generic_end_io_acct does atomic_dec, this barrier makes sure that - * atomic_dec is not reordered with waitqueue_active - */ - smp_mb__after_atomic(); + part = &dm_disk(md)->part0; + sgrp = op_stat_group(bio_op(bio)); + cpu = part_stat_lock(); + part_stat_add(cpu, part, nsecs[sgrp], jiffies_to_nsecs(duration)); + part_round_stats(md->queue, cpu, part); + part_stat_unlock(); + + smp_wmb(); + this_cpu_dec(md->counters->inflight[bio_data_dir(bio)]); if (unlikely(dm_stats_used(&md->stats))) dm_stats_account_io(&md->stats, bio_data_dir(bio), bio->bi_iter.bi_sector, bio_sectors(bio), true, duration, &io->stats_aux); + smp_mb(); /* nudge anyone waiting on suspend queue */ if (unlikely(waitqueue_active(&md->wait))) { if (!md_in_flight(md)) @@ -1828,6 +1842,8 @@ static void cleanup_mapped_device(struct if (md->queue) blk_cleanup_queue(md->queue); + free_percpu(md->counters); + cleanup_srcu_struct(&md->io_barrier); if (md->bdev) { @@ -1899,6 +1915,10 @@ static struct mapped_device *alloc_dev(i if (!md->disk) goto bad; + md->counters = alloc_percpu(struct dm_percpu); + if (!md->counters) + goto bad; + init_waitqueue_head(&md->wait); INIT_WORK(&md->work, dm_wq_work); init_waitqueue_head(&md->eventq); Index: linux-2.6/drivers/md/dm-rq.c =================================================================== --- linux-2.6.orig/drivers/md/dm-rq.c 2018-10-30 15:09:45.770000000 +0100 +++ linux-2.6/drivers/md/dm-rq.c 2018-11-06 00:53:13.870000000 +0100 @@ -172,6 +172,12 @@ static void rq_end_stats(struct mapped_d } } +static unsigned rq_md_in_flight(struct mapped_device *md) +{ + return atomic_read(&dm_disk(md)->part0.in_flight[READ]) + + atomic_read(&dm_disk(md)->part0.in_flight[WRITE]); +} + /* * Don't touch any member of the md after calling this function because * the md may be freed in dm_put() at the end of this function. @@ -185,7 +191,7 @@ static void rq_completed(struct mapped_d atomic_dec(&dm_disk(md)->part0.in_flight[rw]); /* nudge anyone waiting on suspend queue */ - if (!md_in_flight(md)) + if (!rq_md_in_flight(md)) wake_up(&md->wait); /* @@ -674,7 +680,7 @@ static void dm_old_request_fn(struct req pos = blk_rq_pos(rq); if ((dm_old_request_peeked_before_merge_deadline(md) && - md_in_flight(md) && rq->bio && !bio_multiple_segments(rq->bio) && + rq_md_in_flight(md) && rq->bio && !bio_multiple_segments(rq->bio) && md->last_rq_pos == pos && md->last_rq_rw == rq_data_dir(rq)) || (ti->type->busy && ti->type->busy(ti))) { blk_delay_queue(q, 10);