From 8811b5968f6216e97ccb9fe7b9883af39e339921 Mon Sep 17 00:00:00 2001 From: Shaohua Li Date: Thu, 2 Aug 2012 08:33:00 +1000 Subject: raid5: make_request use batch stripe release make_request() does stripe release for every stripe and the stripe usually has count 1, which makes previous release_stripe() optimization not work. In my test, this release_stripe() becomes the heaviest pleace to take conf->device_lock after previous patches applied. Below patch makes stripe release batch. All the stripes will be released in unplug. The STRIPE_ON_UNPLUG_LIST bit is to protect concurrent access stripe lru. Signed-off-by: Shaohua Li Signed-off-by: NeilBrown --- drivers/md/raid5.c | 62 +++++++++++++++++++++++++++++++++++++++++++++++++++--- drivers/md/raid5.h | 1 + 2 files changed, 60 insertions(+), 3 deletions(-) (limited to 'drivers/md') diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c index bde9da2baa39..978ba9b7a3c4 100644 --- a/drivers/md/raid5.c +++ b/drivers/md/raid5.c @@ -471,7 +471,8 @@ get_active_stripe(struct r5conf *conf, sector_t sector, } else { if (atomic_read(&sh->count)) { BUG_ON(!list_empty(&sh->lru) - && !test_bit(STRIPE_EXPANDING, &sh->state)); + && !test_bit(STRIPE_EXPANDING, &sh->state) + && !test_bit(STRIPE_ON_UNPLUG_LIST, &sh->state)); } else { if (!test_bit(STRIPE_HANDLE, &sh->state)) atomic_inc(&conf->active_stripes); @@ -3988,6 +3989,62 @@ static struct stripe_head *__get_priority_stripe(struct r5conf *conf) return sh; } +struct raid5_plug_cb { + struct blk_plug_cb cb; + struct list_head list; +}; + +static void raid5_unplug(struct blk_plug_cb *blk_cb, bool from_schedule) +{ + struct raid5_plug_cb *cb = container_of( + blk_cb, struct raid5_plug_cb, cb); + struct stripe_head *sh; + struct mddev *mddev = cb->cb.data; + struct r5conf *conf = mddev->private; + + if (cb->list.next && !list_empty(&cb->list)) { + spin_lock_irq(&conf->device_lock); + while (!list_empty(&cb->list)) { + sh = list_first_entry(&cb->list, struct stripe_head, lru); + list_del_init(&sh->lru); + /* + * avoid race release_stripe_plug() sees + * STRIPE_ON_UNPLUG_LIST clear but the stripe + * is still in our list + */ + smp_mb__before_clear_bit(); + clear_bit(STRIPE_ON_UNPLUG_LIST, &sh->state); + __release_stripe(conf, sh); + } + spin_unlock_irq(&conf->device_lock); + } + kfree(cb); +} + +static void release_stripe_plug(struct mddev *mddev, + struct stripe_head *sh) +{ + struct blk_plug_cb *blk_cb = blk_check_plugged( + raid5_unplug, mddev, + sizeof(struct raid5_plug_cb)); + struct raid5_plug_cb *cb; + + if (!blk_cb) { + release_stripe(sh); + return; + } + + cb = container_of(blk_cb, struct raid5_plug_cb, cb); + + if (cb->list.next == NULL) + INIT_LIST_HEAD(&cb->list); + + if (!test_and_set_bit(STRIPE_ON_UNPLUG_LIST, &sh->state)) + list_add_tail(&sh->lru, &cb->list); + else + release_stripe(sh); +} + static void make_request(struct mddev *mddev, struct bio * bi) { struct r5conf *conf = mddev->private; @@ -4116,8 +4173,7 @@ static void make_request(struct mddev *mddev, struct bio * bi) if ((bi->bi_rw & REQ_SYNC) && !test_and_set_bit(STRIPE_PREREAD_ACTIVE, &sh->state)) atomic_inc(&conf->preread_active_stripes); - mddev_check_plugged(mddev); - release_stripe(sh); + release_stripe_plug(mddev, sh); } else { /* cannot get stripe for read-ahead, just give-up */ clear_bit(BIO_UPTODATE, &bi->bi_flags); diff --git a/drivers/md/raid5.h b/drivers/md/raid5.h index 2164021f3b5f..9a7b36f0a425 100644 --- a/drivers/md/raid5.h +++ b/drivers/md/raid5.h @@ -319,6 +319,7 @@ enum { STRIPE_BIOFILL_RUN, STRIPE_COMPUTE_RUN, STRIPE_OPS_REQ_PENDING, + STRIPE_ON_UNPLUG_LIST, }; /* -- cgit v1.2.3 From 46a06401f6ba13e59d24746fa9ffa6773b69eee3 Mon Sep 17 00:00:00 2001 From: Shaohua Li Date: Thu, 2 Aug 2012 08:33:15 +1000 Subject: raid5: raid5d handle stripe in batch way Let raid5d handle stripe in batch way to reduce conf->device_lock locking. Signed-off-by: Shaohua Li Signed-off-by: NeilBrown --- drivers/md/raid5.c | 45 ++++++++++++++++++++++++++++++++------------- 1 file changed, 32 insertions(+), 13 deletions(-) (limited to 'drivers/md') diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c index 978ba9b7a3c4..9e41ae37bd40 100644 --- a/drivers/md/raid5.c +++ b/drivers/md/raid5.c @@ -4574,6 +4574,30 @@ static int retry_aligned_read(struct r5conf *conf, struct bio *raid_bio) return handled; } +#define MAX_STRIPE_BATCH 8 +static int handle_active_stripes(struct r5conf *conf) +{ + struct stripe_head *batch[MAX_STRIPE_BATCH], *sh; + int i, batch_size = 0; + + while (batch_size < MAX_STRIPE_BATCH && + (sh = __get_priority_stripe(conf)) != NULL) + batch[batch_size++] = sh; + + if (batch_size == 0) + return batch_size; + spin_unlock_irq(&conf->device_lock); + + for (i = 0; i < batch_size; i++) + handle_stripe(batch[i]); + + cond_resched(); + + spin_lock_irq(&conf->device_lock); + for (i = 0; i < batch_size; i++) + __release_stripe(conf, batch[i]); + return batch_size; +} /* * This is our raid5 kernel thread. @@ -4584,7 +4608,6 @@ static int retry_aligned_read(struct r5conf *conf, struct bio *raid_bio) */ static void raid5d(struct mddev *mddev) { - struct stripe_head *sh; struct r5conf *conf = mddev->private; int handled; struct blk_plug plug; @@ -4598,6 +4621,7 @@ static void raid5d(struct mddev *mddev) spin_lock_irq(&conf->device_lock); while (1) { struct bio *bio; + int batch_size; if ( !list_empty(&conf->bitmap_list)) { @@ -4621,21 +4645,16 @@ static void raid5d(struct mddev *mddev) handled++; } - sh = __get_priority_stripe(conf); - - if (!sh) + batch_size = handle_active_stripes(conf); + if (!batch_size) break; - spin_unlock_irq(&conf->device_lock); - - handled++; - handle_stripe(sh); - release_stripe(sh); - cond_resched(); + handled += batch_size; - if (mddev->flags & ~(1<flags & ~(1<device_lock); md_check_recovery(mddev); - - spin_lock_irq(&conf->device_lock); + spin_lock_irq(&conf->device_lock); + } } pr_debug("%d stripes handled\n", handled); -- cgit v1.2.3 From f54a9d0e59c4bea3db733921ca9147612a6f292c Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Thu, 2 Aug 2012 08:33:20 +1000 Subject: md/raid1: submit IO from originating thread instead of md thread. queuing writes to the md thread means that all requests go through the one processor which may not be able to keep up with very high request rates. So use the plugging infrastructure to submit all requests on unplug. If a 'schedule' is needed, we fall back on the old approach of handing the requests to the thread for it to handle. Signed-off-by: NeilBrown --- drivers/md/bitmap.c | 2 +- drivers/md/raid1.c | 57 ++++++++++++++++++++++++++++++++++++++++++++++++++--- 2 files changed, 55 insertions(+), 4 deletions(-) (limited to 'drivers/md') diff --git a/drivers/md/bitmap.c b/drivers/md/bitmap.c index 15dbe03117e4..94e7f6ba2e11 100644 --- a/drivers/md/bitmap.c +++ b/drivers/md/bitmap.c @@ -1305,7 +1305,7 @@ int bitmap_startwrite(struct bitmap *bitmap, sector_t offset, unsigned long sect prepare_to_wait(&bitmap->overflow_wait, &__wait, TASK_UNINTERRUPTIBLE); spin_unlock_irq(&bitmap->counts.lock); - io_schedule(); + schedule(); finish_wait(&bitmap->overflow_wait, &__wait); continue; } diff --git a/drivers/md/raid1.c b/drivers/md/raid1.c index 36a8fc059ac3..9f01870d031c 100644 --- a/drivers/md/raid1.c +++ b/drivers/md/raid1.c @@ -870,6 +870,44 @@ do_sync_io: pr_debug("%dB behind alloc failed, doing sync I/O\n", bio->bi_size); } +struct raid1_plug_cb { + struct blk_plug_cb cb; + struct bio_list pending; + int pending_cnt; +}; + +static void raid1_unplug(struct blk_plug_cb *cb, bool from_schedule) +{ + struct raid1_plug_cb *plug = container_of(cb, struct raid1_plug_cb, + cb); + struct mddev *mddev = plug->cb.data; + struct r1conf *conf = mddev->private; + struct bio *bio; + + if (from_schedule) { + spin_lock_irq(&conf->device_lock); + bio_list_merge(&conf->pending_bio_list, &plug->pending); + conf->pending_count += plug->pending_cnt; + spin_unlock_irq(&conf->device_lock); + md_wakeup_thread(mddev->thread); + kfree(plug); + return; + } + + /* we aren't scheduling, so we can do the write-out directly. */ + bio = bio_list_get(&plug->pending); + bitmap_unplug(mddev->bitmap); + wake_up(&conf->wait_barrier); + + while (bio) { /* submit pending writes */ + struct bio *next = bio->bi_next; + bio->bi_next = NULL; + generic_make_request(bio); + bio = next; + } + kfree(plug); +} + static void make_request(struct mddev *mddev, struct bio * bio) { struct r1conf *conf = mddev->private; @@ -883,6 +921,8 @@ static void make_request(struct mddev *mddev, struct bio * bio) const unsigned long do_sync = (bio->bi_rw & REQ_SYNC); const unsigned long do_flush_fua = (bio->bi_rw & (REQ_FLUSH | REQ_FUA)); struct md_rdev *blocked_rdev; + struct blk_plug_cb *cb; + struct raid1_plug_cb *plug = NULL; int first_clone; int sectors_handled; int max_sectors; @@ -1185,11 +1225,22 @@ read_again: mbio->bi_private = r1_bio; atomic_inc(&r1_bio->remaining); + + cb = blk_check_plugged(raid1_unplug, mddev, sizeof(*plug)); + if (cb) + plug = container_of(cb, struct raid1_plug_cb, cb); + else + plug = NULL; spin_lock_irqsave(&conf->device_lock, flags); - bio_list_add(&conf->pending_bio_list, mbio); - conf->pending_count++; + if (plug) { + bio_list_add(&plug->pending, mbio); + plug->pending_cnt++; + } else { + bio_list_add(&conf->pending_bio_list, mbio); + conf->pending_count++; + } spin_unlock_irqrestore(&conf->device_lock, flags); - if (!mddev_check_plugged(mddev)) + if (!plug) md_wakeup_thread(mddev->thread); } /* Mustn't call r1_bio_write_done before this next test, -- cgit v1.2.3 From d9f691c365a83ce2530f0e46b947365c2db44ea0 Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Thu, 2 Aug 2012 08:35:43 +1000 Subject: md/dm-raid: DM_RAID should select MD_RAID10 Now that DM_RAID supports raid10, it needs to select that code to ensure it is included. Cc: Jonathan Brassow Reported-by: Fengguang Wu Signed-off-by: NeilBrown --- drivers/md/Kconfig | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) (limited to 'drivers/md') diff --git a/drivers/md/Kconfig b/drivers/md/Kconfig index 10f122a3a856..1de441a6c55f 100644 --- a/drivers/md/Kconfig +++ b/drivers/md/Kconfig @@ -277,13 +277,14 @@ config DM_MIRROR needed for live data migration tools such as 'pvmove'. config DM_RAID - tristate "RAID 1/4/5/6 target" + tristate "RAID 1/4/5/6/10 target" depends on BLK_DEV_DM select MD_RAID1 + select MD_RAID10 select MD_RAID456 select BLK_DEV_MD ---help--- - A dm target that supports RAID1, RAID4, RAID5 and RAID6 mappings + A dm target that supports RAID1, RAID10, RAID4, RAID5 and RAID6 mappings A RAID-5 set of N drives with a capacity of C MB per drive provides the capacity of C * (N - 1) MB, and protects against a failure -- cgit v1.2.3 From 667a5313ecd7308d79629c0738b0db588b0b0a4e Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Thu, 16 Aug 2012 16:46:12 +1000 Subject: md: Don't truncate size at 4TB for RAID0 and Linear commit 27a7b260f71439c40546b43588448faac01adb93 md: Fix handling for devices from 2TB to 4TB in 0.90 metadata. changed 0.90 metadata handling to truncated size to 4TB as that is all that 0.90 can record. However for RAID0 and Linear, 0.90 doesn't need to record the size, so this truncation is not needed and causes working arrays to become too small. So avoid the truncation for RAID0 and Linear This bug was introduced in 3.1 and is suitable for any stable kernels from then onwards. As the offending commit was tagged for 'stable', any stable kernel that it was applied to should also get this patch. That includes at least 2.6.32, 2.6.33 and 3.0. (Thanks to Ben Hutchings for providing that list). Cc: stable@vger.kernel.org Signed-off-by: Neil Brown --- drivers/md/md.c | 9 ++++++--- 1 file changed, 6 insertions(+), 3 deletions(-) (limited to 'drivers/md') diff --git a/drivers/md/md.c b/drivers/md/md.c index fcd098794d37..3f6203a4c7ea 100644 --- a/drivers/md/md.c +++ b/drivers/md/md.c @@ -1108,8 +1108,11 @@ static int super_90_load(struct md_rdev *rdev, struct md_rdev *refdev, int minor ret = 0; } rdev->sectors = rdev->sb_start; - /* Limit to 4TB as metadata cannot record more than that */ - if (rdev->sectors >= (2ULL << 32)) + /* Limit to 4TB as metadata cannot record more than that. + * (not needed for Linear and RAID0 as metadata doesn't + * record this size) + */ + if (rdev->sectors >= (2ULL << 32) && sb->level >= 1) rdev->sectors = (2ULL << 32) - 2; if (rdev->sectors < ((sector_t)sb->size) * 2 && sb->level >= 1) @@ -1400,7 +1403,7 @@ super_90_rdev_size_change(struct md_rdev *rdev, sector_t num_sectors) /* Limit to 4TB as metadata cannot record more than that. * 4TB == 2^32 KB, or 2*2^32 sectors. */ - if (num_sectors >= (2ULL << 32)) + if (num_sectors >= (2ULL << 32) && rdev->mddev->level >= 1) num_sectors = (2ULL << 32) - 2; md_super_write(rdev->mddev, rdev, rdev->sb_start, rdev->sb_size, rdev->sb_page); -- cgit v1.2.3 From e0ee778528bbaad28a5c69d2e219269a3a096607 Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Sat, 18 Aug 2012 09:51:42 +1000 Subject: md/raid10: fix problem with on-stack allocation of r10bio structure. A 'struct r10bio' has an array of per-copy information at the end. This array is declared with size [0] and r10bio_pool_alloc allocates enough extra space to store the per-copy information depending on the number of copies needed. So declaring a 'struct r10bio on the stack isn't going to work. It won't allocate enough space, and memory corruption will ensue. So in the two places where this is done, declare a sufficiently large structure and use that instead. The two call-sites of this bug were introduced in 3.4 and 3.5 so this is suitable for both those kernels. The patch will have to be modified for 3.4 as it only has one bug. Cc: stable@vger.kernel.org Reported-by: Ivan Vasilyev Tested-by: Ivan Vasilyev Signed-off-by: NeilBrown --- drivers/md/raid10.c | 30 +++++++++++++++++++----------- drivers/md/raid10.h | 2 +- 2 files changed, 20 insertions(+), 12 deletions(-) (limited to 'drivers/md') diff --git a/drivers/md/raid10.c b/drivers/md/raid10.c index de5ed6fd8806..1c2eb38f3c51 100644 --- a/drivers/md/raid10.c +++ b/drivers/md/raid10.c @@ -659,7 +659,11 @@ static int raid10_mergeable_bvec(struct request_queue *q, max = biovec->bv_len; if (mddev->merge_check_needed) { - struct r10bio r10_bio; + struct { + struct r10bio r10_bio; + struct r10dev devs[conf->copies]; + } on_stack; + struct r10bio *r10_bio = &on_stack.r10_bio; int s; if (conf->reshape_progress != MaxSector) { /* Cannot give any guidance during reshape */ @@ -667,18 +671,18 @@ static int raid10_mergeable_bvec(struct request_queue *q, return biovec->bv_len; return 0; } - r10_bio.sector = sector; - raid10_find_phys(conf, &r10_bio); + r10_bio->sector = sector; + raid10_find_phys(conf, r10_bio); rcu_read_lock(); for (s = 0; s < conf->copies; s++) { - int disk = r10_bio.devs[s].devnum; + int disk = r10_bio->devs[s].devnum; struct md_rdev *rdev = rcu_dereference( conf->mirrors[disk].rdev); if (rdev && !test_bit(Faulty, &rdev->flags)) { struct request_queue *q = bdev_get_queue(rdev->bdev); if (q->merge_bvec_fn) { - bvm->bi_sector = r10_bio.devs[s].addr + bvm->bi_sector = r10_bio->devs[s].addr + rdev->data_offset; bvm->bi_bdev = rdev->bdev; max = min(max, q->merge_bvec_fn( @@ -690,7 +694,7 @@ static int raid10_mergeable_bvec(struct request_queue *q, struct request_queue *q = bdev_get_queue(rdev->bdev); if (q->merge_bvec_fn) { - bvm->bi_sector = r10_bio.devs[s].addr + bvm->bi_sector = r10_bio->devs[s].addr + rdev->data_offset; bvm->bi_bdev = rdev->bdev; max = min(max, q->merge_bvec_fn( @@ -4414,14 +4418,18 @@ static int handle_reshape_read_error(struct mddev *mddev, { /* Use sync reads to get the blocks from somewhere else */ int sectors = r10_bio->sectors; - struct r10bio r10b; struct r10conf *conf = mddev->private; + struct { + struct r10bio r10_bio; + struct r10dev devs[conf->copies]; + } on_stack; + struct r10bio *r10b = &on_stack.r10_bio; int slot = 0; int idx = 0; struct bio_vec *bvec = r10_bio->master_bio->bi_io_vec; - r10b.sector = r10_bio->sector; - __raid10_find_phys(&conf->prev, &r10b); + r10b->sector = r10_bio->sector; + __raid10_find_phys(&conf->prev, r10b); while (sectors) { int s = sectors; @@ -4432,7 +4440,7 @@ static int handle_reshape_read_error(struct mddev *mddev, s = PAGE_SIZE >> 9; while (!success) { - int d = r10b.devs[slot].devnum; + int d = r10b->devs[slot].devnum; struct md_rdev *rdev = conf->mirrors[d].rdev; sector_t addr; if (rdev == NULL || @@ -4440,7 +4448,7 @@ static int handle_reshape_read_error(struct mddev *mddev, !test_bit(In_sync, &rdev->flags)) goto failed; - addr = r10b.devs[slot].addr + idx * PAGE_SIZE; + addr = r10b->devs[slot].addr + idx * PAGE_SIZE; success = sync_page_io(rdev, addr, s << 9, diff --git a/drivers/md/raid10.h b/drivers/md/raid10.h index 007c2c68dd83..1054cf602345 100644 --- a/drivers/md/raid10.h +++ b/drivers/md/raid10.h @@ -110,7 +110,7 @@ struct r10bio { * We choose the number when they are allocated. * We sometimes need an extra bio to write to the replacement. */ - struct { + struct r10dev { struct bio *bio; union { struct bio *repl_bio; /* used for resync and -- cgit v1.2.3