From patchwork Thu Apr 18 05:44:15 2024 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: tada keisuke X-Patchwork-Id: 13634215 Received: from mo-csw-fb.securemx.jp (mo-csw-fb1120.securemx.jp [210.130.202.128]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 13E2D2CCC2; Thu, 18 Apr 2024 06:45:00 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=210.130.202.128 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1713422703; cv=none; b=EfTehW8PZSY1KZkcPgA0OLR3wUCflQ+Ji3EmK+aAIZ904n4P8+9b+fyb3bX8geYlg9u3NELXt7voarbjYLK4tda5raNcnXK4LN3+bkDkAFrB4mksN2ov9ZngBYu1vyFcUPkwiFlKdgOKyWG1cY7V/W6htHb59X+taol06UzPlNQ= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1713422703; c=relaxed/simple; bh=L+/4oTFeLIWsXRBNEPceBnHFBU9eV7Xxg2EuFA3urAw=; h=From:To:CC:Subject:Date:Message-ID:Content-Type:MIME-Version; b=DWOMfFzA0QRpTpaMLWAnJROIqGVPF0V87o6tzdzfM+NnbacaD1WEaqvCzSgZMxsLHR/6yl7YqmI7xTSwO/MsuDOjb1u9ktPSu5ZV9rDoBsOsvNQ3KXPVTp8+BlqaUwef6wibl9wgX0JnqQSEQeVq83pjSWvCQe3zlTmB2DsLPk4= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=kioxia.com; spf=pass smtp.mailfrom=kioxia.com; arc=none smtp.client-ip=210.130.202.128 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=kioxia.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=kioxia.com Received: by mo-csw-fb.securemx.jp (mx-mo-csw-fb1120) id 43I5iUat2019029; Thu, 18 Apr 2024 14:44:30 +0900 Received: by mo-csw.securemx.jp (mx-mo-csw1120) id 43I5iHEO1974788; Thu, 18 Apr 2024 14:44:17 +0900 X-Iguazu-Qid: 2rWhTy7tBPr0YaskhI X-Iguazu-QSIG: v=2; s=0; t=1713419056; q=2rWhTy7tBPr0YaskhI; m=/bUuy0ljH1qi2LvfcftYTdnhRM+6KW6sOhz7M0Nb/fo= Received: from CNN1EMTA03.test.kioxia.com ([202.248.33.144]) by relay.securemx.jp (mx-mr1122) id 43I5iGfT2852290 (version=TLSv1.2 cipher=ECDHE-RSA-AES128-GCM-SHA256 bits=128 verify=NOT); Thu, 18 Apr 2024 14:44:16 +0900 Received: from Switcher-Post_Send (gateway [10.232.20.1]) by CNN1EMTA03.test.kioxia.com (Postfix) with ESMTP id 120DC2FE0F; Thu, 18 Apr 2024 14:44:16 +0900 (JST) Received: from CNN1ESTR04.kioxia.com (localhost [127.0.0.1]) by Switcher-Post_Send (Postfix) with ESMTP id 5B6331900001EB; Thu, 18 Apr 2024 14:31:33 +0900 (JST) Received: from localhost [127.0.0.1] by CNN1ESTR04.kioxia.com with ESMTP id 0003QAAAAAA02216; Thu, 18 Apr 2024 14:31:33 +0900 Received: from CNN1EXMB04.r1.kioxia.com (CNN1EXMB04.r1.kioxia.com [10.232.20.153]) by Switcher-Pre_Send (Postfix) with ESMTP id 4FF53A0640E18; Thu, 18 Apr 2024 14:31:33 +0900 (JST) Received: from CNN1EXMB04.r1.kioxia.com (10.232.20.153) by CNN1EXMB04.r1.kioxia.com (10.232.20.153) with Microsoft SMTP Server (version=TLS1_2, cipher=TLS_ECDHE_RSA_WITH_AES_128_GCM_SHA256) id 15.1.2507.35; Thu, 18 Apr 2024 14:44:15 +0900 Received: from CNN1EXMB04.r1.kioxia.com ([10.13.100.23]) by CNN1EXMB04.r1.kioxia.com ([10.13.100.23]) with mapi id 15.01.2507.035; Thu, 18 Apr 2024 14:44:15 +0900 From: tada keisuke To: "song@kernel.org" , "yukuai3@huawei.com" CC: "linux-raid@vger.kernel.org" , "linux-kernel@vger.kernel.org" Subject: [PATCH v2 06/11] md: avoid conflicts in active_aligned_reads operations Thread-Topic: [PATCH v2 06/11] md: avoid conflicts in active_aligned_reads operations Thread-Index: AdqLx+KRoW2QXRHySCi80jeCEydOoA== Date: Thu, 18 Apr 2024 05:44:15 +0000 Message-ID: <2732579d7512407c89a4cc9db1c63934@kioxia.com> Accept-Language: ja-JP, en-US Content-Language: ja-JP X-MS-Has-Attach: X-MS-TNEF-Correlator: x-tm-as-product-ver: ISME-14.0.0.2080-9.0.1002-28328.001 x-tm-as-result: No-10--10.964700-8.000000 x-tmase-matchedrid: QprU7jbCnXWmagT1k9kBpgrcxrzwsv5uXPK9y3z82GsPdZmyx8xAa4Ul JlxlTIT+SMhOCdNodgNTbKunqk1vfbVQ6XPWwtdyEXjPIvKd74BMkOX0UoduufLIj08N6FsEuJE UmdYPGSVzy5czsqHK1z4BLF18v1Ji23JrSYCBAcfN+qWlu2ZxaFr2hZxjCLzqMoh6scCF9jF96t 9wDfdERVvLgfDdKWEFEn91NheE+HWip6xhGipIWazGfgakLdja0NaTZUEzRZzC9DxHKrnk8luWw Soz2HVyH7iboQD+7I4VaGy/EyU2eafccBtB84hSnXdphQTSK/K/yN2q8U674odzPqBXhP/S4Mmu En4nwqo1W+3DxdRobxDflJk00ivp0FQxfjRHOzi4u3nS+3EEDn0tCKdnhB581kTfEkyaZdz6C0e Ps7A07YVH0dq7wY7up8Odl1VwpCQwXEs0RQd4EzS1UmI+/vnAHxGeptjWSXBAfvnw3P2gkZlQDd ZWm5Nd x-tm-as-user-approved-sender: No x-tm-as-user-blocked-sender: No x-tmase-result: 10--10.964700-8.000000 x-tmase-version: ISME-14.0.0.2080-9.0.1002-28328.001 x-tm-snts-smtp: 1EA07DDBD178E50F0C09B170C34B3B70AA030D08F4DE65BD5931C7C15CE0C6512000:8 Precedence: bulk X-Mailing-List: linux-raid@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 X-CrossPremisesHeadersFilteredBySendConnector: CNN1EXMB04.r1.kioxia.com X-OrganizationHeadersPreserved: CNN1EXMB04.r1.kioxia.com This patch depends on patch 03. Change the type of active_aligned_reads from atomic_t to percpu_ref in percpu mode. Signed-off-by: Keisuke TADA Signed-off-by: Toshifumi OHTAKE --- drivers/md/raid5.c | 27 +++++++++++++++++++-------- drivers/md/raid5.h | 13 ++++--------- 2 files changed, 23 insertions(+), 17 deletions(-) diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c index b7056854375d..f267856efb48 100644 --- a/drivers/md/raid5.c +++ b/drivers/md/raid5.c @@ -5435,8 +5435,7 @@ static void raid5_align_endio(struct bio *bi) if (!error) { bio_endio(raid_bi); - if (active_aligned_reads_dec_and_test(conf)) - wake_up(&conf->wait_for_quiescent); + active_aligned_reads_dec(conf); return; } @@ -5507,8 +5506,8 @@ static int raid5_read_one_chunk(struct mddev *mddev, struct bio *raid_bio) /* quiesce is in progress, so we need to undo io activation and wait * for it to finish */ - if (did_inc && active_aligned_reads_dec_and_test(conf)) - wake_up(&conf->wait_for_quiescent); + if (did_inc) + active_aligned_reads_dec(conf); spin_lock_irq(&conf->device_lock); wait_event_lock_irq(conf->wait_for_quiescent, conf->quiesce == 0, conf->device_lock); @@ -6608,8 +6607,7 @@ static int retry_aligned_read(struct r5conf *conf, struct bio *raid_bio, bio_endio(raid_bio); - if (active_aligned_reads_dec_and_test(conf)) - wake_up(&conf->wait_for_quiescent); + active_aligned_reads_dec(conf); return handled; } @@ -7315,6 +7313,7 @@ static void free_conf(struct r5conf *conf) { int i; + percpu_ref_exit(&conf->active_aligned_reads); log_exit(conf); shrinker_free(conf->shrinker); @@ -7396,6 +7395,12 @@ static unsigned long raid5_cache_count(struct shrinker *shrink, return max_stripes - min_stripes; } +static void percpu_wakeup_handle_req_active(struct percpu_ref *r) +{ + struct r5conf *conf = container_of(r, struct r5conf, active_aligned_reads); + + wake_up(&conf->wait_for_quiescent); +} static struct r5conf *setup_conf(struct mddev *mddev, bool quiesce) { @@ -7483,7 +7488,11 @@ static struct r5conf *setup_conf(struct mddev *mddev, bool quiesce) percpu_ref_init_flags = PERCPU_REF_ALLOW_REINIT | PERCPU_REF_INIT_DEAD; else percpu_ref_init_flags = PERCPU_REF_ALLOW_REINIT; - atomic_set(&conf->active_aligned_reads, 0); + ret = percpu_ref_init(&conf->active_aligned_reads, percpu_wakeup_handle_req_active, + percpu_ref_init_flags, GFP_KERNEL); + if (ret) + goto abort; + ret = -ENOMEM; spin_lock_init(&conf->pending_bios_lock); conf->batch_bio_dispatch = true; rdev_for_each(rdev, mddev) { @@ -7675,7 +7684,7 @@ static struct r5conf *setup_conf_for_takeover(struct mddev *mddev) if (mddev->level == 4 || mddev->level == 5 || mddev->level == 6) { conf = mddev->private; - quiesce = false; + quiesce = percpu_ref_is_dying(&conf->active_aligned_reads); } return setup_conf(mddev, quiesce); } @@ -8607,6 +8616,7 @@ static void raid5_quiesce(struct mddev *mddev, int quiesce) * quiesce started and reverts to slow (locked) path. */ smp_store_release(&conf->quiesce, 2); + percpu_ref_kill(&conf->active_aligned_reads); wait_event_cmd(conf->wait_for_quiescent, atomic_read(&conf->active_stripes) == 0 && active_aligned_reads_is_zero(conf), @@ -8619,6 +8629,7 @@ static void raid5_quiesce(struct mddev *mddev, int quiesce) } else { /* re-enable writes */ lock_all_device_hash_locks_irq(conf); + percpu_ref_reinit(&conf->active_aligned_reads); conf->quiesce = 0; wake_up(&conf->wait_for_quiescent); wake_up(&conf->wait_for_overlap); diff --git a/drivers/md/raid5.h b/drivers/md/raid5.h index 5bd6bb3540c5..c4ab418e2084 100644 --- a/drivers/md/raid5.h +++ b/drivers/md/raid5.h @@ -621,7 +621,7 @@ struct r5conf { unsigned int retry_read_offset; /* sector offset into retry_read_aligned */ struct bio *retry_read_aligned_list; /* aligned bios retry list */ atomic_t preread_active_stripes; /* stripes with scheduled io */ - atomic_t active_aligned_reads; + struct percpu_ref active_aligned_reads; atomic_t pending_full_writes; /* full write backlog */ int bypass_count; /* bypassed prereads */ int bypass_threshold; /* preread nice */ @@ -698,22 +698,17 @@ struct r5conf { static inline void active_aligned_reads_inc(struct r5conf *conf) { - atomic_inc(&conf->active_aligned_reads); + percpu_ref_get(&conf->active_aligned_reads); } static inline void active_aligned_reads_dec(struct r5conf *conf) { - atomic_dec(&conf->active_aligned_reads); + percpu_ref_put(&conf->active_aligned_reads); } static inline bool active_aligned_reads_is_zero(struct r5conf *conf) { - return atomic_read(&conf->active_aligned_reads) == 0; -} - -static inline bool active_aligned_reads_dec_and_test(struct r5conf *conf) -{ - return atomic_dec_and_test(&conf->active_aligned_reads); + return percpu_ref_is_zero(&conf->active_aligned_reads); } #if PAGE_SIZE == DEFAULT_STRIPE_SIZE