From 4125758074279115346c1f17c986d24cd1ab799c Mon Sep 17 00:00:00 2001 From: Song Liu Date: Mon, 23 May 2016 17:25:06 -0700 Subject: right meaning of PARITY_ENABLE_RMW and PARITY_PREFER_RMW In current handle_stripe_dirtying, the code prefers rmw with PARITY_ENABLE_RMW; while prefers rcw with PARITY_PREFER_RMW. This patch reverses this behavior. Signed-off-by: Song Liu Signed-off-by: Shaohua Li --- drivers/md/raid5.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'drivers/md/raid5.c') diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c index 8959e6dd31dd..ad9e15a3ef51 100644 --- a/drivers/md/raid5.c +++ b/drivers/md/raid5.c @@ -3600,7 +3600,7 @@ static void handle_stripe_dirtying(struct r5conf *conf, pr_debug("for sector %llu, rmw=%d rcw=%d\n", (unsigned long long)sh->sector, rmw, rcw); set_bit(STRIPE_HANDLE, &sh->state); - if ((rmw < rcw || (rmw == rcw && conf->rmw_level == PARITY_ENABLE_RMW)) && rmw > 0) { + if ((rmw < rcw || (rmw == rcw && conf->rmw_level == PARITY_PREFER_RMW)) && rmw > 0) { /* prefer read-modify-write, but need to get some data */ if (conf->mddev->queue) blk_add_trace_msg(conf->mddev->queue, @@ -3627,7 +3627,7 @@ static void handle_stripe_dirtying(struct r5conf *conf, } } } - if ((rcw < rmw || (rcw == rmw && conf->rmw_level != PARITY_ENABLE_RMW)) && rcw > 0) { + if ((rcw < rmw || (rcw == rmw && conf->rmw_level != PARITY_PREFER_RMW)) && rcw > 0) { /* want reconstruct write, but need to get some data */ int qread =0; rcw = 0; -- cgit v1.2.3 From e50d3992328320acf2454c1e7301a094cd90aebc Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Thu, 2 Jun 2016 16:19:52 +1000 Subject: md/raid5: add rcu protection to rdev accesses in handle_failed_sync. The rdev could be freed while handle_failed_sync is running, so rcu protection is needed. Signed-off-by: NeilBrown Signed-off-by: Shaohua Li --- drivers/md/raid5.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) (limited to 'drivers/md/raid5.c') diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c index ad9e15a3ef51..e56c7e0627fa 100644 --- a/drivers/md/raid5.c +++ b/drivers/md/raid5.c @@ -3210,15 +3210,16 @@ handle_failed_sync(struct r5conf *conf, struct stripe_head *sh, /* During recovery devices cannot be removed, so * locking and refcounting of rdevs is not needed */ + rcu_read_lock(); for (i = 0; i < conf->raid_disks; i++) { - struct md_rdev *rdev = conf->disks[i].rdev; + struct md_rdev *rdev = rcu_dereference(conf->disks[i].rdev); if (rdev && !test_bit(Faulty, &rdev->flags) && !test_bit(In_sync, &rdev->flags) && !rdev_set_badblocks(rdev, sh->sector, STRIPE_SECTORS, 0)) abort = 1; - rdev = conf->disks[i].replacement; + rdev = rcu_dereference(conf->disks[i].replacement); if (rdev && !test_bit(Faulty, &rdev->flags) && !test_bit(In_sync, &rdev->flags) @@ -3226,6 +3227,7 @@ handle_failed_sync(struct r5conf *conf, struct stripe_head *sh, STRIPE_SECTORS, 0)) abort = 1; } + rcu_read_unlock(); if (abort) conf->recovery_disabled = conf->mddev->recovery_disabled; -- cgit v1.2.3 From 3f232d6a95fc0fe539e0855ee7f3854b03c8104e Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Thu, 2 Jun 2016 16:19:52 +1000 Subject: md/raid5: add rcu protection to rdev accesses in want_replace Being in the middle of resync is no longer protection against failed rdevs disappearing. So add rcu protection. Signed-off-by: NeilBrown Signed-off-by: Shaohua Li --- drivers/md/raid5.c | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-) (limited to 'drivers/md/raid5.c') diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c index e56c7e0627fa..198b2a9eef98 100644 --- a/drivers/md/raid5.c +++ b/drivers/md/raid5.c @@ -3239,15 +3239,16 @@ static int want_replace(struct stripe_head *sh, int disk_idx) { struct md_rdev *rdev; int rv = 0; - /* Doing recovery so rcu locking not required */ - rdev = sh->raid_conf->disks[disk_idx].replacement; + + rcu_read_lock(); + rdev = rcu_dereference(sh->raid_conf->disks[disk_idx].replacement); if (rdev && !test_bit(Faulty, &rdev->flags) && !test_bit(In_sync, &rdev->flags) && (rdev->recovery_offset <= sh->sector || rdev->mddev->recovery_cp <= sh->sector)) rv = 1; - + rcu_read_unlock(); return rv; } -- cgit v1.2.3 From 5fd133511d5f2a9167573d100e14fddd9c621581 Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Thu, 2 Jun 2016 16:19:52 +1000 Subject: md/raid5: add rcu protection to rdev accesses in raid5_status. Signed-off-by: NeilBrown Signed-off-by: Shaohua Li --- drivers/md/raid5.c | 10 ++++++---- 1 file changed, 6 insertions(+), 4 deletions(-) (limited to 'drivers/md/raid5.c') diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c index 198b2a9eef98..c07b22e8d946 100644 --- a/drivers/md/raid5.c +++ b/drivers/md/raid5.c @@ -7069,10 +7069,12 @@ static void raid5_status(struct seq_file *seq, struct mddev *mddev) seq_printf(seq, " level %d, %dk chunk, algorithm %d", mddev->level, conf->chunk_sectors / 2, mddev->layout); seq_printf (seq, " [%d/%d] [", conf->raid_disks, conf->raid_disks - mddev->degraded); - for (i = 0; i < conf->raid_disks; i++) - seq_printf (seq, "%s", - conf->disks[i].rdev && - test_bit(In_sync, &conf->disks[i].rdev->flags) ? "U" : "_"); + rcu_read_lock(); + for (i = 0; i < conf->raid_disks; i++) { + struct md_rdev *rdev = rcu_dereference(conf->disks[i].rdev); + seq_printf (seq, "%s", rdev && test_bit(In_sync, &rdev->flags) ? "U" : "_"); + } + rcu_read_unlock(); seq_printf (seq, "]"); } -- cgit v1.2.3 From f5b67ae86ee317db20c0e10d54f16a0bbbd3207d Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Thu, 2 Jun 2016 16:19:53 +1000 Subject: md: be extra careful not to take a reference to a Faulty device. It is important that we never increment rdev->nr_pending on a Faulty device as ->hot_remove_disk() assumes that once the Faulty flag is visible no code will take a new reference. Some places take a new reference after only check In_sync. This should be safe as the two are changed together. However to make the code more obviously safe, add checks for 'Faulty' as well. Note: the actual rule is: Never increment nr_pending if Faulty is set and Blocked is clear, never clear Faulty, and never set Blocked without holding a reference through nr_pending. fix build error (Shaohua) Signed-off-by: NeilBrown Signed-off-by: Shaohua Li --- drivers/md/raid5.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'drivers/md/raid5.c') diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c index c07b22e8d946..f6a191aaaa91 100644 --- a/drivers/md/raid5.c +++ b/drivers/md/raid5.c @@ -3080,7 +3080,8 @@ handle_failed_stripe(struct r5conf *conf, struct stripe_head *sh, struct md_rdev *rdev; rcu_read_lock(); rdev = rcu_dereference(conf->disks[i].rdev); - if (rdev && test_bit(In_sync, &rdev->flags)) + if (rdev && test_bit(In_sync, &rdev->flags) && + !test_bit(Faulty, &rdev->flags)) atomic_inc(&rdev->nr_pending); else rdev = NULL; -- cgit v1.2.3 From d787be4092e27728cb4c012bee9762098ef3c662 Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Thu, 2 Jun 2016 16:19:53 +1000 Subject: md: reduce the number of synchronize_rcu() calls when multiple devices fail. Every time a device is removed with ->hot_remove_disk() a synchronize_rcu() call is made which can delay several milliseconds in some case. If lots of devices fail at once - as could happen with a large RAID10 where one set of devices are removed all at once - these delays can add up to be very inconcenient. As failure is not reversible we can check for that first, setting a separate flag if it is found, and then all synchronize_rcu() once for all the flagged devices. Then ->hot_remove_disk() function can skip the synchronize_rcu() step if the flag is set. fix build error(Shaohua) Signed-off-by: NeilBrown Signed-off-by: Shaohua Li --- drivers/md/raid5.c | 15 +++++++++------ 1 file changed, 9 insertions(+), 6 deletions(-) (limited to 'drivers/md/raid5.c') diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c index f6a191aaaa91..413cc7d847da 100644 --- a/drivers/md/raid5.c +++ b/drivers/md/raid5.c @@ -7197,12 +7197,15 @@ static int raid5_remove_disk(struct mddev *mddev, struct md_rdev *rdev) goto abort; } *rdevp = NULL; - synchronize_rcu(); - if (atomic_read(&rdev->nr_pending)) { - /* lost the race, try later */ - err = -EBUSY; - *rdevp = rdev; - } else if (p->replacement) { + if (!test_bit(RemoveSynchronized, &rdev->flags)) { + synchronize_rcu(); + if (atomic_read(&rdev->nr_pending)) { + /* lost the race, try later */ + err = -EBUSY; + *rdevp = rdev; + } + } + if (p->replacement) { /* We must have just cleared 'rdev' */ p->rdev = p->replacement; clear_bit(Replacement, &p->replacement->flags); -- cgit v1.2.3