From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from mail-wm2-f12.google.com (mail-wm2-f12.google.com [74.125.225.140]) (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 E9A58430CCA for ; Thu, 10 Sep 2026 08:11:27 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=74.125.225.140 ARC-Seal:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1789027893; cv=none; b=SZ7tt+UVzQx9G/EsIEckU1UD/qPfl51UORWIRwLY4mGJiPF+2hyMIUS3n/wTjNQWZZpaD+1VWOYecarDO1GTnge5hNXip0gVj0gaVProMQVv0edYlHmyrwurv9s4FOaCjBU2Pdbhq/RmjLCUkpCqFn7MuORdknXOivB2FgbRY44= ARC-Message-Signature:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1789027893; c=relaxed/simple; bh=N4W4/1Nb4xxCQW/b/yh4LAIUt+gjMzmX7Ei/zLa6UWE=; h=From:To:Cc:Subject:Date:Message-ID:In-Reply-To:References: MIME-Version; b=gA/3FxIbMig3/XuPgBuXoYW9PlvdGTiJ0lNNkaAjaTeTq/BPeXNrEZKzCGU5i7iqmglA3SLKeondFOzA1DKYo2YmpyvyjoWTh7VhsvIEFO1TSBqkB/CummHNYRdr0C/fGgVjahbtOKotfs6ZN1P/eOnZdmsbu300F9EW+aqkvNY= ARC-Authentication-Results:i=1; smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=ionos.com; spf=pass smtp.mailfrom=ionos.com; dkim=pass (2048-bit key) header.d=ionos.com header.i=@ionos.com header.b=T0cUXcFR; arc=none smtp.client-ip=74.125.225.140 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=ionos.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=ionos.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=ionos.com header.i=@ionos.com header.b="T0cUXcFR" Received: by mail-wm2-f12.google.com with SMTP id 5b1f17b1804b1-49d1ddd5d0aso1916455e9.3 for ; Thu, 10 Sep 2026 01:11:27 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=ionos.com; s=google; t=1789027886; x=1789632686; darn=vger.kernel.org; h=content-transfer-encoding:mime-version:references:in-reply-to :message-id:date:subject:cc:to:from:from:to:cc:subject:date :message-id:reply-to:content-type; bh=+GHgzzQ8rfbRP2Put64Cq82EilqCh8jYa5rghgFsni8=; b=T0cUXcFRSTTCOZWzfmdw5HeKCYRLaYu12TxzP2/SIWE6PyPMHPyt89TkO5VQE+snSf tx9HUy0BVsGsBuWC1/WpnSEMPRS06BxJ/lEp+/+1cGEhmp9IoOus4x6O+MD7GvxOVJ+Q pdiZJpfEgsWdkODpzHVo4b9cj9pwbbrxAfGt5KNagkjAcD63Ri4Khqa40rVXueVjsNlt FI9x16Nzy7fhPNZsDYqef2jzxrrJb1oenLhvORzggAzha/aga88CuPzIoSdFQNAcK36K +NOjZZeR+vkm8H0MamnFkPIpW1svySdoLdnBcIN3lfkXS89CklcY9P06Oy/hNiu+Vtan ++vQ== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20251104; t=1789027886; x=1789632686; h=content-transfer-encoding:mime-version:references:in-reply-to :message-id:date:subject:cc:to:from:x-gm-gg:x-gm-message-state:from :to:cc:subject:date:message-id:reply-to:content-type; bh=+GHgzzQ8rfbRP2Put64Cq82EilqCh8jYa5rghgFsni8=; b=VPjz4+qWjGXLMaoCpwFDKL6mcegWsrt2uqx8yut8U5bsah10T1qZyQ1mn2fl+nQ7w5 RKNIpzfnih6jvH4P3vjJGVgVTupfGsrTLH+VZVumAFdQe7LOPyCLC677uT6Y6kThKUy9 a4D49wbIONIMbhhdm9SmqHe8Oys1ndIolpHrMRoan2FnWdP7biQiDungDxpxq8vOCQNQ PZqEA7nC8xTDp75FCKv/XYoc2UZq7eRFsAhX/YWA1q0CIfUMhJGTEFSz+sdvG+CidB2v MLTmksVgriCmeE9iJhA/hQ83Kka4LaTMTmkTB7/hGV1C9IhudUtlUuoXTVQouDCCA3wd LtgQ== X-Forwarded-Encrypted: i=1; AKwUvBy3Q40RUyXb8gMG97Ggtt1lWU7tM8fPmRo05WVNL44sk9epd1Ageqjv/q8rx1LwGGPAQneG66sA7al4@vger.kernel.org X-Gm-Message-State: AFuF++lhAgw7mOh54/cye5N6OtfumdmSfkK/mplQSEuGy3J0trnZSemj ATleTuNv1rZOdIhdbtDH7ulQpUG8M65vHfRdxKkYq0kZIhTSNOzdVnKwvkSSNtbeWCw= X-Gm-Gg: AYBFou1b22YZPzBFnUKrobS6Cflpb4JgRwg+jZmLhI5sNGR3l/5rVk5ezcSSVtHbJg8 VzWQ2YZT0GSOT1RP2B1l6TUCroHaSeNDckO66AXgRkT4AU0zAt/0TukBuaEKo010B/ZniK9+BVE pwS+9yh6lfbfv3i97O0MWFWf7Y72B4MCDLgNWOZhq5fGyli67YM5LhrZ22aJrpbIKWeAAar8v2x LRpuFlDyV7pXM5ashCnZ++ig8dQ6RvC9ntuAoLW2y9CBzfv9X0PdzshT/bP/xNnTsqcycBGCMPH xfcD1Oq27Fo3WX1mHq3M41j9fl10Ak3WPQHG2rIx9kJ2k+IiCh58mcvkp6ro7WUCcyKwFXP+4f4 efdWianuR175SAqDWPM4q7R8OhNvE86z939IQt+c7N8lcK9U0La+X5D+mdz7JVIn2ldFi539u1u HCiLQye3aDJxFQhRjgx479+XH9JUlLLFiQ7b7ff2LWIFSU7x4vNnacNysQeep+8o4DmLiLinjEn /6cNw6J9yFpVXvmxmFG0H9dplxx6spM25hp2KxuOJ9v X-Received: by 2002:a05:600c:c48f:b0:49b:9241:7ff0 with SMTP id 5b1f17b1804b1-49d010c4f64mr288757255e9.0.1789027885618; Thu, 10 Sep 2026 01:11:25 -0700 (PDT) Received: from jwang-ThinkPad-T14-Gen-6.fkb.profitbricks.net ([212.227.34.98]) by smtp.gmail.com with ESMTPSA id 5b1f17b1804b1-49d20fc1be3sm55261135e9.4.2026.09.10.01.11.24 (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Thu, 10 Sep 2026 01:11:25 -0700 (PDT) From: Jack Wang To: Song Liu , Yu Kuai , linux-raid@vger.kernel.org, Nilay Shroff , abd.masalkhi@gmail.com Cc: linux-block@vger.kernel.org, Jens Axboe , Christoph Hellwig , Damien Le Moal , Ming Lei , Xiao Ni , Li Nan , Mike Snitzer , Mikulas Patocka , dm-devel@lists.linux.dev, linux-kernel@vger.kernel.org, Jack Wang Subject: [PATCH v2 7/8] md: open new legs before locking the array Date: Thu, 10 Sep 2026 10:11:12 +0200 Message-ID: <20260910081114.1605746-8-jinpu.wang@ionos.com> X-Mailer: git-send-email 2.43.0 In-Reply-To: <20260910081114.1605746-1-jinpu.wang@ionos.com> References: <20260910081114.1605746-1-jinpu.wang@ionos.com> Precedence: bulk X-Mailing-List: linux-raid@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Content-Transfer-Encoding: 8bit From: Jack Wang Opening a leg takes disk->open_mutex, and the scsi disk probe path nests q->limits_lock inside it (sd_open() -> sd_revalidate_disk()). md opens legs under reconfig_mutex, which this series makes q->limits_lock nest outside, closing a cycle. Booting with lockdep on an md root reports it during assembly: -> #2 (&q->limits_lock): sd_revalidate_disk / sd_open -> #1 (&disk->open_mutex): md_import_device md_add_new_disk md_ioctl <- ADD_NEW_DISK -> #0 (&mddev->reconfig_mutex): md_ioctl <- RUN_ARRAY Move every open out from under the lock. md_import_new_disk() mirrors md_add_new_disk()'s branch selection so all three of its branches take a pre-opened leg, and hot_add_disk(), new_dev_store() and md_setup_drive() open before they lock as well. The mddev fields the open depends on are read without reconfig_mutex, so each caller rechecks them once the array is locked and rejects the add with -EBUSY if the branch or the superblock format would have changed. md_autostart_arrays() needs no change: it opens under detected_devices_mutex. Assisted-by: LLM Signed-off-by: Jack Wang --- drivers/md/md-autodetect.c | 17 ++- drivers/md/md.c | 238 ++++++++++++++++++++++++++----------- drivers/md/md.h | 21 +++- 3 files changed, 203 insertions(+), 73 deletions(-) diff --git a/drivers/md/md-autodetect.c b/drivers/md/md-autodetect.c index e15ae2fb58a2..e592577356ad 100644 --- a/drivers/md/md-autodetect.c +++ b/drivers/md/md-autodetect.c @@ -208,6 +208,7 @@ static void __init md_setup_drive(struct md_setup_args *args) .major = MAJOR(devices[i]), .minor = MINOR(devices[i]), }; + struct md_new_disk nd; if (args->level != LEVEL_NONE) { dinfo.number = i; @@ -216,7 +217,21 @@ static void __init md_setup_drive(struct md_setup_args *args) (1 << MD_DISK_ACTIVE) | (1 << MD_DISK_SYNC); } - md_add_new_disk(mddev, &dinfo, NULL); + /* + * Opening a leg takes disk->open_mutex, which must not nest + * inside reconfig_mutex, see md_import_new_disk(). Drop the + * array lock around it; this is __init and the array is not + * reachable yet, so nothing else can touch it in between. + */ + mddev_unlock(mddev); + if (md_import_new_disk(mddev, &dinfo, &nd)) { + mddev_lock_nointr(mddev); + continue; + } + mddev_lock_nointr(mddev); + + md_add_new_disk(mddev, &dinfo, &nd, NULL); + md_put_new_disk(&nd); } /* diff --git a/drivers/md/md.c b/drivers/md/md.c index 5be956e80563..fa033d7d3831 100644 --- a/drivers/md/md.c +++ b/drivers/md/md.c @@ -4942,6 +4942,7 @@ new_dev_store(struct mddev *mddev, const char *buf, size_t len) struct md_rdev *rdev; unsigned int noio_flags; int err; + int persistent, external, major_version, minor_version; if (!*buf || *e != ':' || !e[1] || e[1] == '\n') return -EINVAL; @@ -4953,32 +4954,52 @@ new_dev_store(struct mddev *mddev, const char *buf, size_t len) minor != MINOR(dev)) return -EOVERFLOW; - err = mddev_suspend_and_lock(mddev); - if (err) - return err; - noio_flags = memalloc_noio_save(); - if (mddev->persistent) { - rdev = md_import_device(dev, mddev->major_version, - mddev->minor_version); - if (!IS_ERR(rdev) && !list_empty(&mddev->disks)) { - struct md_rdev *rdev0 - = list_entry(mddev->disks.next, - struct md_rdev, same_set); - err = super_types[mddev->major_version] - .load_super(rdev, rdev0, mddev->minor_version); - if (err < 0) - goto out; - } - } else if (mddev->external) + /* + * Open before locking the array: bdev_open() takes disk->open_mutex, + * which must not nest inside reconfig_mutex, see md_import_new_disk(). + * The fields below are read without the lock and rechecked under it. + */ + persistent = READ_ONCE(mddev->persistent); + external = READ_ONCE(mddev->external); + major_version = READ_ONCE(mddev->major_version); + minor_version = READ_ONCE(mddev->minor_version); + + if (persistent) + rdev = md_import_device(dev, major_version, minor_version); + else if (external) rdev = md_import_device(dev, -2, -1); else rdev = md_import_device(dev, -1, -1); - if (IS_ERR(rdev)) { - memalloc_noio_restore(noio_flags); - mddev_unlock_and_resume(mddev); + if (IS_ERR(rdev)) return PTR_ERR(rdev); + + err = mddev_suspend_and_lock(mddev); + if (err) { + export_rdev(rdev); + return err; } + noio_flags = memalloc_noio_save(); + + if (persistent != mddev->persistent || external != mddev->external || + major_version != mddev->major_version || + minor_version != mddev->minor_version) { + pr_warn("%s: array reconfigured while opening %pg\n", + mdname(mddev), rdev->bdev); + err = -EBUSY; + goto out; + } + + if (mddev->persistent && !list_empty(&mddev->disks)) { + struct md_rdev *rdev0 + = list_entry(mddev->disks.next, + struct md_rdev, same_set); + err = super_types[mddev->major_version] + .load_super(rdev, rdev0, mddev->minor_version); + if (err < 0) + goto out; + } + err = bind_rdev_to_array(rdev, mddev); out: if (err) @@ -7685,12 +7706,35 @@ static int get_disk_info(struct mddev *mddev, void __user * arg) return 0; } +/* + * @nd carries an rdev the caller opened before locking the array, for the + * branch its snapshot selected. Every caller must open first; doing it + * here would nest disk->open_mutex inside reconfig_mutex. + */ int md_add_new_disk(struct mddev *mddev, struct mdu_disk_info_s *info, - struct queue_limits *lim) + struct md_new_disk *nd, struct queue_limits *lim) { struct md_rdev *rdev; dev_t dev = MKDEV(info->major,info->minor); + /* + * The open ran unlocked, so anything that selects a different branch + * below, or a different superblock format, means it was done against + * an array that no longer looks like this one. + */ + if (nd && nd->rdev && + (nd->have_raid_disks != (mddev->raid_disks != 0) || + nd->have_pers != !!mddev->pers || + nd->persistent != mddev->persistent || + nd->major_version != mddev->major_version || + nd->minor_version != mddev->minor_version)) { + pr_warn("%s: array reconfigured while opening %pg\n", + mdname(mddev), nd->rdev->bdev); + export_rdev(nd->rdev); + nd->rdev = NULL; + return -EBUSY; + } + if (mddev_is_clustered(mddev) && !(info->state & ((1 << MD_DISK_CLUSTER_ADD) | (1 << MD_DISK_CANDIDATE)))) { pr_warn("%s: Cannot add to clustered mddev.\n", @@ -7703,13 +7747,12 @@ int md_add_new_disk(struct mddev *mddev, struct mdu_disk_info_s *info, if (!mddev->raid_disks) { int err; + /* expecting a device which has a superblock */ - rdev = md_import_device(dev, mddev->major_version, mddev->minor_version); - if (IS_ERR(rdev)) { - pr_warn("md: md_import_device returned %ld\n", - PTR_ERR(rdev)); - return PTR_ERR(rdev); - } + if (WARN_ON_ONCE(!nd || !nd->rdev)) + return -EINVAL; + rdev = nd->rdev; + nd->rdev = NULL; if (!list_empty(&mddev->disks)) { struct md_rdev *rdev0 = list_entry(mddev->disks.next, @@ -7742,16 +7785,10 @@ int md_add_new_disk(struct mddev *mddev, struct mdu_disk_info_s *info, mdname(mddev)); return -EINVAL; } - if (mddev->persistent) - rdev = md_import_device(dev, mddev->major_version, - mddev->minor_version); - else - rdev = md_import_device(dev, -1, -1); - if (IS_ERR(rdev)) { - pr_warn("md: md_import_device returned %ld\n", - PTR_ERR(rdev)); - return PTR_ERR(rdev); - } + if (WARN_ON_ONCE(!nd || !nd->rdev)) + return -EINVAL; + rdev = nd->rdev; + nd->rdev = NULL; /* set saved_raid_disk if appropriate */ if (!mddev->persistent) { if (info->state & (1<state & (1<rdev)) + return -EINVAL; + rdev = nd->rdev; + nd->rdev = NULL; rdev->desc_nr = info->number; if (info->raid_disk < mddev->raid_disks) rdev->raid_disk = info->raid_disk; @@ -7930,7 +7966,8 @@ static int hot_remove_disk(struct mddev *mddev, dev_t dev, return -EBUSY; } -static int hot_add_disk(struct mddev *mddev, dev_t dev) +/* @nd carries a leg the caller opened before the array was locked */ +static int hot_add_disk(struct mddev *mddev, struct md_new_disk *nd) { int err; struct md_rdev *rdev; @@ -7949,12 +7986,10 @@ static int hot_add_disk(struct mddev *mddev, dev_t dev) return -EINVAL; } - rdev = md_import_device(dev, -1, 0); - if (IS_ERR(rdev)) { - pr_warn("md: error, md_import_device() returned %ld\n", - PTR_ERR(rdev)); + if (WARN_ON_ONCE(!nd->rdev)) return -EINVAL; - } + rdev = nd->rdev; + nd->rdev = NULL; if (mddev->persistent) rdev->sb_start = calc_dev_sboffset(rdev); @@ -8497,14 +8532,61 @@ static inline int md_ioctl_valid(unsigned int cmd) * journal device or a personality without ->hot_remove_disk, but that depends * on disk info still in user memory here, so it is included as a whole. */ -static bool md_ioctl_may_add_disk(unsigned int cmd) + +/* + * Open the leg before the array is locked; bdev_open() takes + * disk->open_mutex, which must not nest inside reconfig_mutex. mddev is + * read unlocked on purpose, and md_add_new_disk() rechecks the snapshot. + */ +int md_import_new_disk(struct mddev *mddev, struct mdu_disk_info_s *info, + struct md_new_disk *nd) { - switch (cmd) { - case ADD_NEW_DISK: - case HOT_REMOVE_DISK: - return true; - default: - return false; + dev_t dev = MKDEV(info->major, info->minor); + struct md_rdev *rdev; + + memset(nd, 0, sizeof(*nd)); + nd->have_raid_disks = READ_ONCE(mddev->raid_disks) != 0; + nd->have_pers = !!READ_ONCE(mddev->pers); + nd->persistent = READ_ONCE(mddev->persistent); + nd->major_version = READ_ONCE(mddev->major_version); + nd->minor_version = READ_ONCE(mddev->minor_version); + + if (!nd->have_raid_disks) { + /* a device with a superblock, for an array being assembled */ + rdev = md_import_device(dev, nd->major_version, + nd->minor_version); + } else if (nd->have_pers) { + /* a hot spare; this is the branch that stacks limits */ + nd->stacks = true; + if (nd->persistent) + rdev = md_import_device(dev, nd->major_version, + nd->minor_version); + else + rdev = md_import_device(dev, -1, -1); + } else if (nd->major_version == 0) { + rdev = md_import_device(dev, -1, 0); + } else { + /* md_add_new_disk() rejects this, nothing to open */ + return 0; + } + + if (IS_ERR(rdev)) { + int err = PTR_ERR(rdev); + + pr_warn("md: md_import_device returned %d\n", err); + return err; + } + + nd->rdev = rdev; + return 0; +} + +/* release a leg md_add_new_disk() did not take ownership of */ +void md_put_new_disk(struct md_new_disk *nd) +{ + if (nd->rdev) { + export_rdev(nd->rdev); + nd->rdev = NULL; } } @@ -8566,6 +8648,8 @@ static int md_ioctl(struct block_device *bdev, blk_mode_t mode, struct request_queue *q = NULL; struct queue_limits lim; struct queue_limits *limp = NULL; + struct md_new_disk nd = { }; + mdu_disk_info_t info; bool suspend; err = md_ioctl_valid(cmd); @@ -8616,8 +8700,27 @@ static int md_ioctl(struct block_device *bdev, blk_mode_t mode, if (!md_is_rdwr(mddev)) flush_work(&mddev->sync_work); + if (cmd == ADD_NEW_DISK) { + if (copy_from_user(&info, argp, sizeof(info))) { + err = -EFAULT; + goto out; + } + err = md_import_new_disk(mddev, &info, &nd); + if (err) + goto out; + } else if (cmd == HOT_ADD_DISK) { + nd.rdev = md_import_device(new_decode_dev(arg), -1, 0); + if (IS_ERR(nd.rdev)) { + pr_warn("md: error, md_import_device() returned %ld\n", + PTR_ERR(nd.rdev)); + nd.rdev = NULL; + err = -EINVAL; + goto out; + } + } + /* q->limits_lock nests outside both, see md_start_sync() */ - if ((md_ioctl_may_add_disk(cmd) || cmd == RUN_ARRAY) && + if ((nd.stacks || cmd == HOT_REMOVE_DISK || cmd == RUN_ARRAY) && !mddev_is_dm(mddev)) { q = mddev->gendisk->queue; lim = queue_limits_start_update(q); @@ -8681,14 +8784,10 @@ static int md_ioctl(struct block_device *bdev, blk_mode_t mode, * So require mddev->pers and MD_DISK_SYNC. */ if (mddev->pers) { - mdu_disk_info_t info; - if (copy_from_user(&info, argp, sizeof(info))) - err = -EFAULT; - else if (!(info.state & (1<flags); return err; diff --git a/drivers/md/md.h b/drivers/md/md.h index 73f6ef20f266..73a27d83d65a 100644 --- a/drivers/md/md.h +++ b/drivers/md/md.h @@ -1051,8 +1051,27 @@ struct mdu_disk_info_s; extern int mdp_major; void md_autostart_arrays(int part); int md_set_array_info(struct mddev *mddev, struct mdu_array_info_s *info); +/* + * A leg opened before the array was locked, with the mddev fields that + * selected the branch and the superblock format. Opening takes + * disk->open_mutex, which must not nest inside reconfig_mutex; the fields + * are read unlocked and md_add_new_disk() rechecks them. + */ +struct md_new_disk { + struct md_rdev *rdev; + bool stacks; /* the add can reach ->hot_add_disk() */ + bool have_pers; + bool have_raid_disks; + int persistent; + int major_version; + int minor_version; +}; + +int md_import_new_disk(struct mddev *mddev, struct mdu_disk_info_s *info, + struct md_new_disk *nd); +void md_put_new_disk(struct md_new_disk *nd); int md_add_new_disk(struct mddev *mddev, struct mdu_disk_info_s *info, - struct queue_limits *lim); + struct md_new_disk *nd, struct queue_limits *lim); int do_md_run(struct mddev *mddev, struct queue_limits *lim); #define MDDEV_STACK_INTEGRITY (1u << 0) int mddev_stack_rdev_limits(struct mddev *mddev, struct queue_limits *lim, -- 2.43.0