Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/shli/md

Pull MD updates from Shaohua Li:

 - Add Partial Parity Log (ppl) feature found in Intel IMSM raid array
   by Artur Paszkiewicz. This feature is another way to close RAID5
   writehole. The Linux implementation is also available for normal
   RAID5 array if specific superblock bit is set.

 - A number of md-cluser fixes and enabling md-cluster array resize from
   Guoqing Jiang

 - A bunch of patches from Ming Lei and Neil Brown to rewrite MD bio
   handling related code. Now MD doesn't directly access bio bvec,
   bi_phys_segments and uses modern bio API for bio split.

 - Improve RAID5 IO pattern to improve performance for hard disk based
   RAID5/6 from me.

 - Several patches from Song Liu to speed up raid5-cache recovery and
   allow raid5 cache feature disabling in runtime.

 - Fix a performance regression in raid1 resync from Xiao Ni.

 - Other cleanup and fixes from various people.

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/shli/md: (84 commits)
  md/raid10: skip spare disk as 'first' disk
  md/raid1: Use a new variable to count flighting sync requests
  md: clear WantReplacement once disk is removed
  md/raid1/10: remove unused queue
  md: handle read-only member devices better.
  md/raid10: wait up frozen array in handle_write_completed
  uapi: fix linux/raid/md_p.h userspace compilation error
  md-cluster: Fix a memleak in an error handling path
  md: support disabling of create-on-open semantics.
  md: allow creation of mdNNN arrays via md_mod/parameters/new_array
  raid5-ppl: use a single mempool for ppl_io_unit and header_page
  md/raid0: fix up bio splitting.
  md/linear: improve bio splitting.
  md/raid5: make chunk_aligned_read() split bios more cleanly.
  md/raid10: simplify handle_read_error()
  md/raid10: simplify the splitting of requests.
  md/raid1: factor out flush_bio_list()
  md/raid1: simplify handle_read_error().
  Revert "block: introduce bio_copy_data_partial"
  md/raid1: simplify alloc_behind_master_bio()
  ...

18 files changed, 3434 insertions, 1418 deletions

diff --git a/drivers/md/Makefile b/drivers/md/Makefile
index 3cbda1af87a0..4d48714ccc6b 100644
--- a/drivers/md/Makefile
+++ b/drivers/md/Makefile
@@ -18,7 +18,7 @@ dm-cache-cleaner-y += dm-cache-policy-cleaner.o
 dm-era-y	+= dm-era-target.o
 dm-verity-y	+= dm-verity-target.o
 md-mod-y	+= md.o bitmap.o
-raid456-y	+= raid5.o raid5-cache.o
+raid456-y	+= raid5.o raid5-cache.o raid5-ppl.o
 
 # Note: link order is important.  All raid personalities
 # and must come before md.o, as they each initialise 
diff --git a/drivers/md/bitmap.c b/drivers/md/bitmap.c
index 9fb2ccac958a..bf7419a56454 100644
--- a/drivers/md/bitmap.c
+++ b/drivers/md/bitmap.c
@@ -471,6 +471,7 @@ void bitmap_update_sb(struct bitmap *bitmap)
 	kunmap_atomic(sb);
 	write_page(bitmap, bitmap->storage.sb_page, 1);
 }
+EXPORT_SYMBOL(bitmap_update_sb);
 
 /* print out the bitmap file superblock */
 void bitmap_print_sb(struct bitmap *bitmap)
@@ -696,7 +697,7 @@ re_read:
 
 out:
 	kunmap_atomic(sb);
-	/* Assiging chunksize is required for "re_read" */
+	/* Assigning chunksize is required for "re_read" */
 	bitmap->mddev->bitmap_info.chunksize = chunksize;
 	if (err == 0 && nodes && (bitmap->cluster_slot < 0)) {
 		err = md_setup_cluster(bitmap->mddev, nodes);
@@ -1727,7 +1728,7 @@ void bitmap_flush(struct mddev *mddev)
 /*
  * free memory that was allocated
  */
-static void bitmap_free(struct bitmap *bitmap)
+void bitmap_free(struct bitmap *bitmap)
 {
 	unsigned long k, pages;
 	struct bitmap_page *bp;
@@ -1761,6 +1762,21 @@ static void bitmap_free(struct bitmap *bitmap)
 	kfree(bp);
 	kfree(bitmap);
 }
+EXPORT_SYMBOL(bitmap_free);
+
+void bitmap_wait_behind_writes(struct mddev *mddev)
+{
+	struct bitmap *bitmap = mddev->bitmap;
+
+	/* wait for behind writes to complete */
+	if (bitmap && atomic_read(&bitmap->behind_writes) > 0) {
+		pr_debug("md:%s: behind writes in progress - waiting to stop.\n",
+			 mdname(mddev));
+		/* need to kick something here to make sure I/O goes? */
+		wait_event(bitmap->behind_wait,
+			   atomic_read(&bitmap->behind_writes) == 0);
+	}
+}
 
 void bitmap_destroy(struct mddev *mddev)
 {
@@ -1769,6 +1785,8 @@ void bitmap_destroy(struct mddev *mddev)
 	if (!bitmap) /* there was no bitmap */
 		return;
 
+	bitmap_wait_behind_writes(mddev);
+
 	mutex_lock(&mddev->bitmap_info.mutex);
 	spin_lock(&mddev->lock);
 	mddev->bitmap = NULL; /* disconnect from the md device */
@@ -1920,6 +1938,27 @@ out:
 }
 EXPORT_SYMBOL_GPL(bitmap_load);
 
+struct bitmap *get_bitmap_from_slot(struct mddev *mddev, int slot)
+{
+	int rv = 0;
+	struct bitmap *bitmap;
+
+	bitmap = bitmap_create(mddev, slot);
+	if (IS_ERR(bitmap)) {
+		rv = PTR_ERR(bitmap);
+		return ERR_PTR(rv);
+	}
+
+	rv = bitmap_init_from_disk(bitmap, 0);
+	if (rv) {
+		bitmap_free(bitmap);
+		return ERR_PTR(rv);
+	}
+
+	return bitmap;
+}
+EXPORT_SYMBOL(get_bitmap_from_slot);
+
 /* Loads the bitmap associated with slot and copies the resync information
  * to our bitmap
  */
@@ -1929,14 +1968,13 @@ int bitmap_copy_from_slot(struct mddev *mddev, int slot,
 	int rv = 0, i, j;
 	sector_t block, lo = 0, hi = 0;
 	struct bitmap_counts *counts;
-	struct bitmap *bitmap = bitmap_create(mddev, slot);
-
-	if (IS_ERR(bitmap))
-		return PTR_ERR(bitmap);
+	struct bitmap *bitmap;
 
-	rv = bitmap_init_from_disk(bitmap, 0);
-	if (rv)
-		goto err;
+	bitmap = get_bitmap_from_slot(mddev, slot);
+	if (IS_ERR(bitmap)) {
+		pr_err("%s can't get bitmap from slot %d\n", __func__, slot);
+		return -1;
+	}
 
 	counts = &bitmap->counts;
 	for (j = 0; j < counts->chunks; j++) {
@@ -1963,8 +2001,7 @@ int bitmap_copy_from_slot(struct mddev *mddev, int slot,
 	bitmap_unplug(mddev->bitmap);
 	*low = lo;
 	*high = hi;
-err:
-	bitmap_free(bitmap);
+
 	return rv;
 }
 EXPORT_SYMBOL_GPL(bitmap_copy_from_slot);
diff --git a/drivers/md/bitmap.h b/drivers/md/bitmap.h
index 5b6dd63dda91..d15721ac07a6 100644
--- a/drivers/md/bitmap.h
+++ b/drivers/md/bitmap.h
@@ -267,8 +267,11 @@ void bitmap_daemon_work(struct mddev *mddev);
 
 int bitmap_resize(struct bitmap *bitmap, sector_t blocks,
 		  int chunksize, int init);
+struct bitmap *get_bitmap_from_slot(struct mddev *mddev, int slot);
 int bitmap_copy_from_slot(struct mddev *mddev, int slot,
 				sector_t *lo, sector_t *hi, bool clear_bits);
+void bitmap_free(struct bitmap *bitmap);
+void bitmap_wait_behind_writes(struct mddev *mddev);
 #endif
 
 #endif
diff --git a/drivers/md/linear.c b/drivers/md/linear.c
index 377a8a3672e3..df6f2c98eca7 100644
--- a/drivers/md/linear.c
+++ b/drivers/md/linear.c
@@ -249,54 +249,49 @@ static void linear_make_request(struct mddev *mddev, struct bio *bio)
 {
 	char b[BDEVNAME_SIZE];
 	struct dev_info *tmp_dev;
-	struct bio *split;
 	sector_t start_sector, end_sector, data_offset;
+	sector_t bio_sector = bio->bi_iter.bi_sector;
 
 	if (unlikely(bio->bi_opf & REQ_PREFLUSH)) {
 		md_flush_request(mddev, bio);
 		return;
 	}
 
-	do {
-		sector_t bio_sector = bio->bi_iter.bi_sector;
-		tmp_dev = which_dev(mddev, bio_sector);
-		start_sector = tmp_dev->end_sector - tmp_dev->rdev->sectors;
-		end_sector = tmp_dev->end_sector;
-		data_offset = tmp_dev->rdev->data_offset;
-		bio->bi_bdev = tmp_dev->rdev->bdev;
-
-		if (unlikely(bio_sector >= end_sector ||
-			     bio_sector < start_sector))
-			goto out_of_bounds;
-
-		if (unlikely(bio_end_sector(bio) > end_sector)) {
-			/* This bio crosses a device boundary, so we have to
-			 * split it.
-			 */
-			split = bio_split(bio, end_sector - bio_sector,
-					  GFP_NOIO, fs_bio_set);
-			bio_chain(split, bio);
-		} else {
-			split = bio;
-		}
+	tmp_dev = which_dev(mddev, bio_sector);
+	start_sector = tmp_dev->end_sector - tmp_dev->rdev->sectors;
+	end_sector = tmp_dev->end_sector;
+	data_offset = tmp_dev->rdev->data_offset;
+
+	if (unlikely(bio_sector >= end_sector ||
+		     bio_sector < start_sector))
+		goto out_of_bounds;
+
+	if (unlikely(bio_end_sector(bio) > end_sector)) {
+		/* This bio crosses a device boundary, so we have to split it */
+		struct bio *split = bio_split(bio, end_sector - bio_sector,
+					      GFP_NOIO, mddev->bio_set);
+		bio_chain(split, bio);
+		generic_make_request(bio);
+		bio = split;
+	}
 
-		split->bi_iter.bi_sector = split->bi_iter.bi_sector -
-			start_sector + data_offset;
-
-		if (unlikely((bio_op(split) == REQ_OP_DISCARD) &&
-			 !blk_queue_discard(bdev_get_queue(split->bi_bdev)))) {
-			/* Just ignore it */
-			bio_endio(split);
-		} else {
-			if (mddev->gendisk)
-				trace_block_bio_remap(bdev_get_queue(split->bi_bdev),
-						      split, disk_devt(mddev->gendisk),
-						      bio_sector);
-			mddev_check_writesame(mddev, split);
-			mddev_check_write_zeroes(mddev, split);
-			generic_make_request(split);
-		}
-	} while (split != bio);
+	bio->bi_bdev = tmp_dev->rdev->bdev;
+	bio->bi_iter.bi_sector = bio->bi_iter.bi_sector -
+		start_sector + data_offset;
+
+	if (unlikely((bio_op(bio) == REQ_OP_DISCARD) &&
+		     !blk_queue_discard(bdev_get_queue(bio->bi_bdev)))) {
+		/* Just ignore it */
+		bio_endio(bio);
+	} else {
+		if (mddev->gendisk)
+			trace_block_bio_remap(bdev_get_queue(bio->bi_bdev),
+					      bio, disk_devt(mddev->gendisk),
+					      bio_sector);
+		mddev_check_writesame(mddev, bio);
+		mddev_check_write_zeroes(mddev, bio);
+		generic_make_request(bio);
+	}
 	return;
 
 out_of_bounds:
diff --git a/drivers/md/md-cluster.c b/drivers/md/md-cluster.c
index 321ecac23027..7299ce2f08a8 100644
--- a/drivers/md/md-cluster.c
+++ b/drivers/md/md-cluster.c
@@ -67,9 +67,10 @@ struct resync_info {
  * set up all the related infos such as bitmap and personality */
 #define		MD_CLUSTER_ALREADY_IN_CLUSTER		6
 #define		MD_CLUSTER_PENDING_RECV_EVENT		7
-
+#define 	MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD		8
 
 struct md_cluster_info {
+	struct mddev *mddev; /* the md device which md_cluster_info belongs to */
 	/* dlm lock space and resources for clustered raid. */
 	dlm_lockspace_t *lockspace;
 	int slot_number;
@@ -103,6 +104,7 @@ enum msg_type {
 	REMOVE,
 	RE_ADD,
 	BITMAP_NEEDS_SYNC,
+	CHANGE_CAPACITY,
 };
 
 struct cluster_msg {
@@ -523,11 +525,17 @@ static void process_add_new_disk(struct mddev *mddev, struct cluster_msg *cmsg)
 
 static void process_metadata_update(struct mddev *mddev, struct cluster_msg *msg)
 {
+	int got_lock = 0;
 	struct md_cluster_info *cinfo = mddev->cluster_info;
 	mddev->good_device_nr = le32_to_cpu(msg->raid_slot);
-	set_bit(MD_RELOAD_SB, &mddev->flags);
+
 	dlm_lock_sync(cinfo->no_new_dev_lockres, DLM_LOCK_CR);
-	md_wakeup_thread(mddev->thread);
+	wait_event(mddev->thread->wqueue,
+		   (got_lock = mddev_trylock(mddev)) ||
+		    test_bit(MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD, &cinfo->state));
+	md_reload_sb(mddev, mddev->good_device_nr);
+	if (got_lock)
+		mddev_unlock(mddev);
 }
 
 static void process_remove_disk(struct mddev *mddev, struct cluster_msg *msg)
@@ -572,6 +580,10 @@ static int process_recvd_msg(struct mddev *mddev, struct cluster_msg *msg)
 	case METADATA_UPDATED:
 		process_metadata_update(mddev, msg);
 		break;
+	case CHANGE_CAPACITY:
+		set_capacity(mddev->gendisk, mddev->array_sectors);
+		revalidate_disk(mddev->gendisk);
+		break;
 	case RESYNCING:
 		process_suspend_info(mddev, le32_to_cpu(msg->slot),
 				     le64_to_cpu(msg->low),
@@ -646,11 +658,29 @@ out:
  * Takes the lock on the TOKEN lock resource so no other
  * node can communicate while the operation is underway.
  */
-static int lock_token(struct md_cluster_info *cinfo)
+static int lock_token(struct md_cluster_info *cinfo, bool mddev_locked)
 {
-	int error;
+	int error, set_bit = 0;
+	struct mddev *mddev = cinfo->mddev;
 
+	/*
+	 * If resync thread run after raid1d thread, then process_metadata_update
+	 * could not continue if raid1d held reconfig_mutex (and raid1d is blocked
+	 * since another node already got EX on Token and waitting the EX of Ack),
+	 * so let resync wake up thread in case flag is set.
+	 */
+	if (mddev_locked && !test_bit(MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD,
+				      &cinfo->state)) {
+		error = test_and_set_bit_lock(MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD,
+					      &cinfo->state);
+		WARN_ON_ONCE(error);
+		md_wakeup_thread(mddev->thread);
+		set_bit = 1;
+	}
 	error = dlm_lock_sync(cinfo->token_lockres, DLM_LOCK_EX);
+	if (set_bit)
+		clear_bit_unlock(MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD, &cinfo->state);
+
 	if (error)
 		pr_err("md-cluster(%s:%d): failed to get EX on TOKEN (%d)\n",
 				__func__, __LINE__, error);
@@ -663,12 +693,12 @@ static int lock_token(struct md_cluster_info *cinfo)
 /* lock_comm()
  * Sets the MD_CLUSTER_SEND_LOCK bit to lock the send channel.
  */
-static int lock_comm(struct md_cluster_info *cinfo)
+static int lock_comm(struct md_cluster_info *cinfo, bool mddev_locked)
 {
 	wait_event(cinfo->wait,
 		   !test_and_set_bit(MD_CLUSTER_SEND_LOCK, &cinfo->state));
 
-	return lock_token(cinfo);
+	return lock_token(cinfo, mddev_locked);
 }
 
 static void unlock_comm(struct md_cluster_info *cinfo)
@@ -743,11 +773,12 @@ failed_message:
 	return error;
 }
 
-static int sendmsg(struct md_cluster_info *cinfo, struct cluster_msg *cmsg)
+static int sendmsg(struct md_cluster_info *cinfo, struct cluster_msg *cmsg,
+		   bool mddev_locked)
 {
 	int ret;
 
-	lock_comm(cinfo);
+	lock_comm(cinfo, mddev_locked);
 	ret = __sendmsg(cinfo, cmsg);
 	unlock_comm(cinfo);
 	return ret;
@@ -834,6 +865,7 @@ static int join(struct mddev *mddev, int nodes)
 	mutex_init(&cinfo->recv_mutex);
 
 	mddev->cluster_info = cinfo;
+	cinfo->mddev = mddev;
 
 	memset(str, 0, 64);
 	sprintf(str, "%pU", mddev->uuid);
@@ -908,6 +940,7 @@ static int join(struct mddev *mddev, int nodes)
 
 	return 0;
 err:
+	set_bit(MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD, &cinfo->state);
 	md_unregister_thread(&cinfo->recovery_thread);
 	md_unregister_thread(&cinfo->recv_thread);
 	lockres_free(cinfo->message_lockres);
@@ -943,7 +976,7 @@ static void resync_bitmap(struct mddev *mddev)
 	int err;
 
 	cmsg.type = cpu_to_le32(BITMAP_NEEDS_SYNC);
-	err = sendmsg(cinfo, &cmsg);
+	err = sendmsg(cinfo, &cmsg, 1);
 	if (err)
 		pr_err("%s:%d: failed to send BITMAP_NEEDS_SYNC message (%d)\n",
 			__func__, __LINE__, err);
@@ -963,6 +996,7 @@ static int leave(struct mddev *mddev)
 	if (cinfo->slot_number > 0 && mddev->recovery_cp != MaxSector)
 		resync_bitmap(mddev);
 
+	set_bit(MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD, &cinfo->state);
 	md_unregister_thread(&cinfo->recovery_thread);
 	md_unregister_thread(&cinfo->recv_thread);
 	lockres_free(cinfo->message_lockres);
@@ -997,16 +1031,30 @@ static int slot_number(struct mddev *mddev)
 static int metadata_update_start(struct mddev *mddev)
 {
 	struct md_cluster_info *cinfo = mddev->cluster_info;
+	int ret;
+
+	/*
+	 * metadata_update_start is always called with the protection of
+	 * reconfig_mutex, so set WAITING_FOR_TOKEN here.
+	 */
+	ret = test_and_set_bit_lock(MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD,
+				    &cinfo->state);
+	WARN_ON_ONCE(ret);
+	md_wakeup_thread(mddev->thread);
 
 	wait_event(cinfo->wait,
 		   !test_and_set_bit(MD_CLUSTER_SEND_LOCK, &cinfo->state) ||
 		   test_and_clear_bit(MD_CLUSTER_SEND_LOCKED_ALREADY, &cinfo->state));
 
 	/* If token is already locked, return 0 */
-	if (cinfo->token_lockres->mode == DLM_LOCK_EX)
+	if (cinfo->token_lockres->mode == DLM_LOCK_EX) {
+		clear_bit_unlock(MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD, &cinfo->state);
 		return 0;
+	}
 
-	return lock_token(cinfo);
+	ret = lock_token(cinfo, 1);
+	clear_bit_unlock(MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD, &cinfo->state);
+	return ret;
 }
 
 static int metadata_update_finish(struct mddev *mddev)
@@ -1043,6 +1091,141 @@ static void metadata_update_cancel(struct mddev *mddev)
 	unlock_comm(cinfo);
 }
 
+/*
+ * return 0 if all the bitmaps have the same sync_size
+ */
+int cluster_check_sync_size(struct mddev *mddev)
+{
+	int i, rv;
+	bitmap_super_t *sb;
+	unsigned long my_sync_size, sync_size = 0;
+	int node_num = mddev->bitmap_info.nodes;
+	int current_slot = md_cluster_ops->slot_number(mddev);
+	struct bitmap *bitmap = mddev->bitmap;
+	char str[64];
+	struct dlm_lock_resource *bm_lockres;
+
+	sb = kmap_atomic(bitmap->storage.sb_page);
+	my_sync_size = sb->sync_size;
+	kunmap_atomic(sb);
+
+	for (i = 0; i < node_num; i++) {
+		if (i == current_slot)
+			continue;
+
+		bitmap = get_bitmap_from_slot(mddev, i);
+		if (IS_ERR(bitmap)) {
+			pr_err("can't get bitmap from slot %d\n", i);
+			return -1;
+		}
+
+		/*
+		 * If we can hold the bitmap lock of one node then
+		 * the slot is not occupied, update the sb.
+		 */
+		snprintf(str, 64, "bitmap%04d", i);
+		bm_lockres = lockres_init(mddev, str, NULL, 1);
+		if (!bm_lockres) {
+			pr_err("md-cluster: Cannot initialize %s\n", str);
+			bitmap_free(bitmap);
+			return -1;
+		}
+		bm_lockres->flags |= DLM_LKF_NOQUEUE;
+		rv = dlm_lock_sync(bm_lockres, DLM_LOCK_PW);
+		if (!rv)
+			bitmap_update_sb(bitmap);
+		lockres_free(bm_lockres);
+
+		sb = kmap_atomic(bitmap->storage.sb_page);
+		if (sync_size == 0)
+			sync_size = sb->sync_size;
+		else if (sync_size != sb->sync_size) {
+			kunmap_atomic(sb);
+			bitmap_free(bitmap);
+			return -1;
+		}
+		kunmap_atomic(sb);
+		bitmap_free(bitmap);
+	}
+
+	return (my_sync_size == sync_size) ? 0 : -1;
+}
+
+/*
+ * Update the size for cluster raid is a little more complex, we perform it
+ * by the steps:
+ * 1. hold token lock and update superblock in initiator node.
+ * 2. send METADATA_UPDATED msg to other nodes.
+ * 3. The initiator node continues to check each bitmap's sync_size, if all
+ *    bitmaps have the same value of sync_size, then we can set capacity and
+ *    let other nodes to perform it. If one node can't update sync_size
+ *    accordingly, we need to revert to previous value.
+ */
+static void update_size(struct mddev *mddev, sector_t old_dev_sectors)
+{
+	struct md_cluster_info *cinfo = mddev->cluster_info;
+	struct cluster_msg cmsg;
+	struct md_rdev *rdev;
+	int ret = 0;
+	int raid_slot = -1;
+
+	md_update_sb(mddev, 1);
+	lock_comm(cinfo, 1);
+
+	memset(&cmsg, 0, sizeof(cmsg));
+	cmsg.type = cpu_to_le32(METADATA_UPDATED);
+	rdev_for_each(rdev, mddev)
+		if (rdev->raid_disk >= 0 && !test_bit(Faulty, &rdev->flags)) {
+			raid_slot = rdev->desc_nr;
+			break;
+		}
+	if (raid_slot >= 0) {
+		cmsg.raid_slot = cpu_to_le32(raid_slot);
+		/*
+		 * We can only change capiticy after all the nodes can do it,
+		 * so need to wait after other nodes already received the msg
+		 * and handled the change
+		 */
+		ret = __sendmsg(cinfo, &cmsg);
+		if (ret) {
+			pr_err("%s:%d: failed to send METADATA_UPDATED msg\n",
+			       __func__, __LINE__);
+			unlock_comm(cinfo);
+			return;
+		}
+	} else {
+		pr_err("md-cluster: No good device id found to send\n");
+		unlock_comm(cinfo);
+		return;
+	}
+
+	/*
+	 * check the sync_size from other node's bitmap, if sync_size
+	 * have already updated in other nodes as expected, send an
+	 * empty metadata msg to permit the change of capacity
+	 */
+	if (cluster_check_sync_size(mddev) == 0) {
+		memset(&cmsg, 0, sizeof(cmsg));
+		cmsg.type = cpu_to_le32(CHANGE_CAPACITY);
+		ret = __sendmsg(cinfo, &cmsg);
+		if (ret)
+			pr_err("%s:%d: failed to send CHANGE_CAPACITY msg\n",
+			       __func__, __LINE__);
+		set_capacity(mddev->gendisk, mddev->array_sectors);
+		revalidate_disk(mddev->gendisk);
+	} else {
+		/* revert to previous sectors */
+		ret = mddev->pers->resize(mddev, old_dev_sectors);
+		if (!ret)
+			revalidate_disk(mddev->gendisk);
+		ret = __sendmsg(cinfo, &cmsg);
+		if (ret)
+			pr_err("%s:%d: failed to send METADATA_UPDATED msg\n",
+			       __func__, __LINE__);
+	}
+	unlock_comm(cinfo);
+}
+
 static int resync_start(struct mddev *mddev)
 {
 	struct md_cluster_info *cinfo = mddev->cluster_info;
@@ -1069,7 +1252,14 @@ static int resync_info_update(struct mddev *mddev, sector_t lo, sector_t hi)
 	cmsg.low = cpu_to_le64(lo);
 	cmsg.high = cpu_to_le64(hi);
 
-	return sendmsg(cinfo, &cmsg);
+	/*
+	 * mddev_lock is held if resync_info_update is called from
+	 * resync_finish (md_reap_sync_thread -> resync_finish)
+	 */
+	if (lo == 0 && hi == 0)
+		return sendmsg(cinfo, &cmsg, 1);
+	else
+		return sendmsg(cinfo, &cmsg, 0);
 }
 
 static int resync_finish(struct mddev *mddev)
@@ -1119,7 +1309,7 @@ static int add_new_disk(struct mddev *mddev, struct md_rdev *rdev)
 	cmsg.type = cpu_to_le32(NEWDISK);
 	memcpy(cmsg.uuid, uuid, 16);
 	cmsg.raid_slot = cpu_to_le32(rdev->desc_nr);
-	lock_comm(cinfo);
+	lock_comm(cinfo, 1);
 	ret = __sendmsg(cinfo, &cmsg);
 	if (ret)
 		return ret;
@@ -1179,7 +1369,7 @@ static int remove_disk(struct mddev *mddev, struct md_rdev *rdev)
 	struct md_cluster_info *cinfo = mddev->cluster_info;
 	cmsg.type = cpu_to_le32(REMOVE);
 	cmsg.raid_slot = cpu_to_le32(rdev->desc_nr);
-	return sendmsg(cinfo, &cmsg);
+	return sendmsg(cinfo, &cmsg, 1);
 }
 
 static int lock_all_bitmaps(struct mddev *mddev)
@@ -1243,7 +1433,7 @@ static int gather_bitmaps(struct md_rdev *rdev)
 
 	cmsg.type = cpu_to_le32(RE_ADD);
 	cmsg.raid_slot = cpu_to_le32(rdev->desc_nr);
-	err = sendmsg(cinfo, &cmsg);
+	err = sendmsg(cinfo, &cmsg, 1);
 	if (err)
 		goto out;
 
@@ -1281,6 +1471,7 @@ static struct md_cluster_operations cluster_ops = {
 	.gather_bitmaps = gather_bitmaps,
 	.lock_all_bitmaps = lock_all_bitmaps,
 	.unlock_all_bitmaps = unlock_all_bitmaps,
+	.update_size = update_size,
 };
 
 static int __init cluster_init(void)
diff --git a/drivers/md/md-cluster.h b/drivers/md/md-cluster.h
index e765499ba591..274016177983 100644
--- a/drivers/md/md-cluster.h
+++ b/drivers/md/md-cluster.h
@@ -27,6 +27,7 @@ struct md_cluster_operations {
 	int (*gather_bitmaps)(struct md_rdev *rdev);
 	int (*lock_all_bitmaps)(struct mddev *mddev);
 	void (*unlock_all_bitmaps)(struct mddev *mddev);
+	void (*update_size)(struct mddev *mddev, sector_t old_dev_sectors);
 };
 
 #endif /* _MD_CLUSTER_H */
diff --git a/drivers/md/md.c b/drivers/md/md.c
index f6ae1d67bcd0..82f798be964f 100644
--- a/drivers/md/md.c
+++ b/drivers/md/md.c
@@ -65,6 +65,8 @@
 #include <linux/raid/md_p.h>
 #include <linux/raid/md_u.h>
 #include <linux/slab.h>
+#include <linux/percpu-refcount.h>
+
 #include <trace/events/block.h>
 #include "md.h"
 #include "bitmap.h"
@@ -172,6 +174,16 @@ static const struct block_device_operations md_fops;
 
 static int start_readonly;
 
+/*
+ * The original mechanism for creating an md device is to create
+ * a device node in /dev and to open it.  This causes races with device-close.
+ * The preferred method is to write to the "new_array" module parameter.
+ * This can avoid races.
+ * Setting create_on_open to false disables the original mechanism
+ * so all the races disappear.
+ */
+static bool create_on_open = true;
+
 /* bio_clone_mddev
  * like bio_clone, but with a local bio set
  */
@@ -1507,6 +1519,12 @@ static int super_1_load(struct md_rdev *rdev, struct md_rdev *refdev, int minor_
 	} else if (sb->bblog_offset != 0)
 		rdev->badblocks.shift = 0;
 
+	if (le32_to_cpu(sb->feature_map) & MD_FEATURE_PPL) {
+		rdev->ppl.offset = (__s16)le16_to_cpu(sb->ppl.offset);
+		rdev->ppl.size = le16_to_cpu(sb->ppl.size);
+		rdev->ppl.sector = rdev->sb_start + rdev->ppl.offset;
+	}
+
 	if (!refdev) {
 		ret = 1;
 	} else {
@@ -1619,6 +1637,13 @@ static int super_1_validate(struct mddev *mddev, struct md_rdev *rdev)
 
 		if (le32_to_cpu(sb->feature_map) & MD_FEATURE_JOURNAL)
 			set_bit(MD_HAS_JOURNAL, &mddev->flags);
+
+		if (le32_to_cpu(sb->feature_map) & MD_FEATURE_PPL) {
+			if (le32_to_cpu(sb->feature_map) &
+			    (MD_FEATURE_BITMAP_OFFSET | MD_FEATURE_JOURNAL))
+				return -EINVAL;
+			set_bit(MD_HAS_PPL, &mddev->flags);
+		}
 	} else if (mddev->pers == NULL) {
 		/* Insist of good event counter while assembling, except for
 		 * spares (which don't need an event count) */
@@ -1832,6 +1857,12 @@ retry:
 	if (test_bit(MD_HAS_JOURNAL, &mddev->flags))
 		sb->feature_map |= cpu_to_le32(MD_FEATURE_JOURNAL);
 
+	if (test_bit(MD_HAS_PPL, &mddev->flags)) {
+		sb->feature_map |= cpu_to_le32(MD_FEATURE_PPL);
+		sb->ppl.offset = cpu_to_le16(rdev->ppl.offset);
+		sb->ppl.size = cpu_to_le16(rdev->ppl.size);
+	}
+
 	rdev_for_each(rdev2, mddev) {
 		i = rdev2->desc_nr;
 		if (test_bit(Faulty, &rdev2->flags))
@@ -2072,6 +2103,10 @@ static int bind_rdev_to_array(struct md_rdev *rdev, struct mddev *mddev)
 	if (find_rdev(mddev, rdev->bdev->bd_dev))
 		return -EEXIST;
 
+	if ((bdev_read_only(rdev->bdev) || bdev_read_only(rdev->meta_bdev)) &&
+	    mddev->pers)
+		return -EROFS;
+
 	/* make sure rdev->sectors exceeds mddev->dev_sectors */
 	if (!test_bit(Journal, &rdev->flags) &&
 	    rdev->sectors &&
@@ -2233,6 +2268,33 @@ static void export_array(struct mddev *mddev)
 	mddev->major_version = 0;
 }
 
+static bool set_in_sync(struct mddev *mddev)
+{
+	WARN_ON_ONCE(!spin_is_locked(&mddev->lock));
+	if (!mddev->in_sync) {
+		mddev->sync_checkers++;
+		spin_unlock(&mddev->lock);
+		percpu_ref_switch_to_atomic_sync(&mddev->writes_pending);
+		spin_lock(&mddev->lock);
+		if (!mddev->in_sync &&
+		    percpu_ref_is_zero(&mddev->writes_pending)) {
+			mddev->in_sync = 1;
+			/*
+			 * Ensure ->in_sync is visible before we clear
+			 * ->sync_checkers.
+			 */
+			smp_mb();
+			set_bit(MD_SB_CHANGE_CLEAN, &mddev->sb_flags);
+			sysfs_notify_dirent_safe(mddev->sysfs_state);
+		}
+		if (--mddev->sync_checkers == 0)
+			percpu_ref_switch_to_percpu(&mddev->writes_pending);
+	}
+	if (mddev->safemode == 1)
+		mddev->safemode = 0;
+	return mddev->in_sync;
+}
+
 static void sync_sbs(struct mddev *mddev, int nospares)
 {
 	/* Update each superblock (in-memory image), but
@@ -3131,6 +3193,78 @@ static ssize_t ubb_store(struct md_rdev *rdev, const char *page, size_t len)
 static struct rdev_sysfs_entry rdev_unack_bad_blocks =
 __ATTR(unacknowledged_bad_blocks, S_IRUGO|S_IWUSR, ubb_show, ubb_store);
 
+static ssize_t
+ppl_sector_show(struct md_rdev *rdev, char *page)
+{
+	return sprintf(page, "%llu\n", (unsigned long long)rdev->ppl.sector);
+}
+
+static ssize_t
+ppl_sector_store(struct md_rdev *rdev, const char *buf, size_t len)
+{
+	unsigned long long sector;
+
+	if (kstrtoull(buf, 10, &sector) < 0)
+		return -EINVAL;
+	if (sector != (sector_t)sector)
+		return -EINVAL;
+
+	if (rdev->mddev->pers && test_bit(MD_HAS_PPL, &rdev->mddev->flags) &&
+	    rdev->raid_disk >= 0)
+		return -EBUSY;
+
+	if (rdev->mddev->persistent) {
+		if (rdev->mddev->major_version == 0)
+			return -EINVAL;
+		if ((sector > rdev->sb_start &&
+		     sector - rdev->sb_start > S16_MAX) ||
+		    (sector < rdev->sb_start &&
+		     rdev->sb_start - sector > -S16_MIN))
+			return -EINVAL;
+		rdev->ppl.offset = sector - rdev->sb_start;
+	} else if (!rdev->mddev->external) {
+		return -EBUSY;
+	}
+	rdev->ppl.sector = sector;
+	return len;
+}
+
+static struct rdev_sysfs_entry rdev_ppl_sector =
+__ATTR(ppl_sector, S_IRUGO|S_IWUSR, ppl_sector_show, ppl_sector_store);
+
+static ssize_t
+ppl_size_show(struct md_rdev *rdev, char *page)
+{
+	return sprintf(page, "%u\n", rdev->ppl.size);
+}
+
+static ssize_t
+ppl_size_store(struct md_rdev *rdev, const char *buf, size_t len)
+{
+	unsigned int size;
+
+	if (kstrtouint(buf, 10, &size) < 0)
+		return -EINVAL;
+
+	if (rdev->mddev->pers && test_bit(MD_HAS_PPL, &rdev->mddev->flags) &&
+	    rdev->raid_disk >= 0)
+		return -EBUSY;
+
+	if (rdev->mddev->persistent) {
+		if (rdev->mddev->major_version == 0)
+			return -EINVAL;
+		if (size > U16_MAX)
+			return -EINVAL;
+	} else if (!rdev->mddev->external) {
+		return -EBUSY;
+	}
+	rdev->ppl.size = size;
+	return len;
+}
+
+static struct rdev_sysfs_entry rdev_ppl_size =
+__ATTR(ppl_size, S_IRUGO|S_IWUSR, ppl_size_show, ppl_size_store);
+
 static struct attribute *rdev_default_attrs[] = {
 	&rdev_state.attr,
 	&rdev_errors.attr,
@@ -3141,6 +3275,8 @@ static struct attribute *rdev_default_attrs[] = {
 	&rdev_recovery_start.attr,
 	&rdev_bad_blocks.attr,
 	&rdev_unack_bad_blocks.attr,
+	&rdev_ppl_sector.attr,
+	&rdev_ppl_size.attr,
 	NULL,
 };
 static ssize_t
@@ -3903,6 +4039,7 @@ array_state_show(struct mddev *mddev, char *page)
 			st = read_auto;
 			break;
 		case 0:
+			spin_lock(&mddev->lock);
 			if (test_bit(MD_SB_CHANGE_PENDING, &mddev->sb_flags))
 				st = write_pending;
 			else if (mddev->in_sync)
@@ -3911,6 +4048,7 @@ array_state_show(struct mddev *mddev, char *page)
 				st = active_idle;
 			else
 				st = active;
+			spin_unlock(&mddev->lock);
 		}
 	else {
 		if (list_empty(&mddev->disks) &&
@@ -3931,7 +4069,7 @@ static int restart_array(struct mddev *mddev);
 static ssize_t
 array_state_store(struct mddev *mddev, const char *buf, size_t len)
 {
-	int err;
+	int err = 0;
 	enum array_state st = match_word(buf, array_states);
 
 	if (mddev->pers && (st == active || st == clean) && mddev->ro != 1) {
@@ -3944,18 +4082,9 @@ array_state_store(struct mddev *mddev, const char *buf, size_t len)
 			clear_bit(MD_SB_CHANGE_PENDING, &mddev->sb_flags);
 			md_wakeup_thread(mddev->thread);
 			wake_up(&mddev->sb_wait);
-			err = 0;
 		} else /* st == clean */ {
 			restart_array(mddev);
-			if (atomic_read(&mddev->writes_pending) == 0) {
-				if (mddev->in_sync == 0) {
-					mddev->in_sync = 1;
-					if (mddev->safemode == 1)
-						mddev->safemode = 0;
-					set_bit(MD_SB_CHANGE_CLEAN, &mddev->sb_flags);
-				}
-				err = 0;
-			} else
+			if (!set_in_sync(mddev))
 				err = -EBUSY;
 		}