Merge tag 'md-3.3' of git://neil.brown.name/md

md update for 3.3

Big change is new hot-replacement.
A slot in an array can hold 2 devices - one that
wants-replacement and one that is the replacement.
Once the replacement is built - either from the
original or (in the case of errors) from elsewhere,
the wants-replacement device will be removed.

* tag 'md-3.3' of git://neil.brown.name/md: (36 commits)
  md/raid1: Mark device want_replacement when we see a write error.
  md/raid1: If there is a spare and a want_replacement device, start replacement.
  md/raid1: recognise replacements when assembling arrays.
  md/raid1: handle activation of replacement device when recovery completes.
  md/raid1: Allow a failed replacement device to be removed.
  md/raid1: Allocate spare to store replacement devices and their bios.
  md/raid1:  Replace use of mddev->raid_disks with conf->raid_disks.
  md/raid10: If there is a spare and a want_replacement device, start replacement.
  md/raid10: recognise replacements when assembling array.
  md/raid10: Allow replacement device to be replace old drive.
  md/raid10: handle recovery of replacement devices.
  md/raid10:  Handle replacement devices during resync.
  md/raid10: writes should get directed to replacement as well as original.
  md/raid10: allow removal of failed replacement devices.
  md/raid10: preferentially read from replacement device if possible.
  md/raid10:  change read_balance to return an rdev
  md/raid10: prepare data structures for handling replacement.
  md/raid5: Mark device want_replacement when we see a write error.
  md/raid5: If there is a spare and a want_replacement device, start replacement.
  md/raid5: recognise replacements when assembling array.
  ...

13 files changed, 1280 insertions, 438 deletions

diff --git a/Documentation/md.txt b/Documentation/md.txt
index fc94770f44a..993fba37b7d 100644
--- a/Documentation/md.txt
+++ b/Documentation/md.txt
@@ -357,14 +357,14 @@ Each directory contains:
         written to, that device.
 
       state
-        A file recording the current state of the device in the array
+	A file recording the current state of the device in the array
 	which can be a comma separated list of
 	      faulty   - device has been kicked from active use due to
-                         a detected fault or it has unacknowledged bad
-                         blocks
+			 a detected fault, or it has unacknowledged bad
+			 blocks
 	      in_sync  - device is a fully in-sync member of the array
 	      writemostly - device will only be subject to read
-		         requests if there are no other options.
+			 requests if there are no other options.
 			 This applies only to raid1 arrays.
 	      blocked  - device has failed, and the failure hasn't been
 			 acknowledged yet by the metadata handler.
@@ -374,6 +374,13 @@ Each directory contains:
 			 This includes spares that are in the process
 			 of being recovered to
 	      write_error - device has ever seen a write error.
+	      want_replacement - device is (mostly) working but probably
+			 should be replaced, either due to errors or
+			 due to user request.
+	      replacement - device is a replacement for another active
+			 device with same raid_disk.
+
+
 	This list may grow in future.
 	This can be written to.
 	Writing "faulty"  simulates a failure on the device.
@@ -386,6 +393,13 @@ Each directory contains:
 	Writing "in_sync" sets the in_sync flag.
 	Writing "write_error" sets writeerrorseen flag.
 	Writing "-write_error" clears writeerrorseen flag.
+	Writing "want_replacement" is allowed at any time except to a
+		replacement device or a spare.  It sets the flag.
+	Writing "-want_replacement" is allowed at any time.  It clears
+		the flag.
+	Writing "replacement" or "-replacement" is only allowed before
+		starting the array.  It sets or clears the flag.
+
 
 	This file responds to select/poll. Any change to 'faulty'
 	or 'blocked' causes an event.
diff --git a/drivers/md/bitmap.c b/drivers/md/bitmap.c
index 6d03774b176..cdf36b1e9aa 100644
--- a/drivers/md/bitmap.c
+++ b/drivers/md/bitmap.c
@@ -1149,12 +1149,12 @@ void bitmap_daemon_work(struct mddev *mddev)
 		return;
 	}
 	if (time_before(jiffies, bitmap->daemon_lastrun
-			+ bitmap->mddev->bitmap_info.daemon_sleep))
+			+ mddev->bitmap_info.daemon_sleep))
 		goto done;
 
 	bitmap->daemon_lastrun = jiffies;
 	if (bitmap->allclean) {
-		bitmap->mddev->thread->timeout = MAX_SCHEDULE_TIMEOUT;
+		mddev->thread->timeout = MAX_SCHEDULE_TIMEOUT;
 		goto done;
 	}
 	bitmap->allclean = 1;
@@ -1206,7 +1206,7 @@ void bitmap_daemon_work(struct mddev *mddev)
 			 * sure that events_cleared is up-to-date.
 			 */
 			if (bitmap->need_sync &&
-			    bitmap->mddev->bitmap_info.external == 0) {
+			    mddev->bitmap_info.external == 0) {
 				bitmap_super_t *sb;
 				bitmap->need_sync = 0;
 				sb = kmap_atomic(bitmap->sb_page, KM_USER0);
@@ -1270,8 +1270,8 @@ void bitmap_daemon_work(struct mddev *mddev)
 
  done:
 	if (bitmap->allclean == 0)
-		bitmap->mddev->thread->timeout =
-			bitmap->mddev->bitmap_info.daemon_sleep;
+		mddev->thread->timeout =
+			mddev->bitmap_info.daemon_sleep;
 	mutex_unlock(&mddev->bitmap_info.mutex);
 }
 
@@ -1587,7 +1587,7 @@ static void bitmap_set_memory_bits(struct bitmap *bitmap, sector_t offset, int n
 	}
 	if (!*bmc) {
 		struct page *page;
-		*bmc = 1 | (needed ? NEEDED_MASK : 0);
+		*bmc = 2 | (needed ? NEEDED_MASK : 0);
 		bitmap_count_page(bitmap, offset, 1);
 		page = filemap_get_page(bitmap, offset >> CHUNK_BLOCK_SHIFT(bitmap));
 		set_page_attr(bitmap, page, BITMAP_PAGE_PENDING);
diff --git a/drivers/md/md.c b/drivers/md/md.c
index 5d1b6762f10..ca8527fe77e 100644
--- a/drivers/md/md.c
+++ b/drivers/md/md.c
@@ -1713,6 +1713,8 @@ static int super_1_validate(struct mddev *mddev, struct md_rdev *rdev)
 		}
 		if (sb->devflags & WriteMostly1)
 			set_bit(WriteMostly, &rdev->flags);
+		if (le32_to_cpu(sb->feature_map) & MD_FEATURE_REPLACEMENT)
+			set_bit(Replacement, &rdev->flags);
 	} else /* MULTIPATH are always insync */
 		set_bit(In_sync, &rdev->flags);
 
@@ -1766,6 +1768,9 @@ static void super_1_sync(struct mddev *mddev, struct md_rdev *rdev)
 		sb->recovery_offset =
 			cpu_to_le64(rdev->recovery_offset);
 	}
+	if (test_bit(Replacement, &rdev->flags))
+		sb->feature_map |=
+			cpu_to_le32(MD_FEATURE_REPLACEMENT);
 
 	if (mddev->reshape_position != MaxSector) {
 		sb->feature_map |= cpu_to_le32(MD_FEATURE_RESHAPE_ACTIVE);
@@ -2559,6 +2564,15 @@ state_show(struct md_rdev *rdev, char *page)
 		len += sprintf(page+len, "%swrite_error", sep);
 		sep = ",";
 	}
+	if (test_bit(WantReplacement, &rdev->flags)) {
+		len += sprintf(page+len, "%swant_replacement", sep);
+		sep = ",";
+	}
+	if (test_bit(Replacement, &rdev->flags)) {
+		len += sprintf(page+len, "%sreplacement", sep);
+		sep = ",";
+	}
+
 	return len+sprintf(page+len, "\n");
 }
 
@@ -2627,6 +2641,42 @@ state_store(struct md_rdev *rdev, const char *buf, size_t len)
 	} else if (cmd_match(buf, "-write_error")) {
 		clear_bit(WriteErrorSeen, &rdev->flags);
 		err = 0;
+	} else if (cmd_match(buf, "want_replacement")) {
+		/* Any non-spare device that is not a replacement can
+		 * become want_replacement at any time, but we then need to
+		 * check if recovery is needed.
+		 */
+		if (rdev->raid_disk >= 0 &&
+		    !test_bit(Replacement, &rdev->flags))
+			set_bit(WantReplacement, &rdev->flags);
+		set_bit(MD_RECOVERY_NEEDED, &rdev->mddev->recovery);
+		md_wakeup_thread(rdev->mddev->thread);
+		err = 0;
+	} else if (cmd_match(buf, "-want_replacement")) {
+		/* Clearing 'want_replacement' is always allowed.
+		 * Once replacements starts it is too late though.
+		 */
+		err = 0;
+		clear_bit(WantReplacement, &rdev->flags);
+	} else if (cmd_match(buf, "replacement")) {
+		/* Can only set a device as a replacement when array has not
+		 * yet been started.  Once running, replacement is automatic
+		 * from spares, or by assigning 'slot'.
+		 */
+		if (rdev->mddev->pers)
+			err = -EBUSY;
+		else {
+			set_bit(Replacement, &rdev->flags);
+			err = 0;
+		}
+	} else if (cmd_match(buf, "-replacement")) {
+		/* Similarly, can only clear Replacement before start */
+		if (rdev->mddev->pers)
+			err = -EBUSY;
+		else {
+			clear_bit(Replacement, &rdev->flags);
+			err = 0;
+		}
 	}
 	if (!err)
 		sysfs_notify_dirent_safe(rdev->sysfs_state);
@@ -2688,7 +2738,7 @@ slot_store(struct md_rdev *rdev, const char *buf, size_t len)
 		if (rdev->mddev->pers->hot_remove_disk == NULL)
 			return -EINVAL;
 		err = rdev->mddev->pers->
-			hot_remove_disk(rdev->mddev, rdev->raid_disk);
+			hot_remove_disk(rdev->mddev, rdev);
 		if (err)
 			return err;
 		sysfs_unlink_rdev(rdev->mddev, rdev);
@@ -2696,7 +2746,6 @@ slot_store(struct md_rdev *rdev, const char *buf, size_t len)
 		set_bit(MD_RECOVERY_NEEDED, &rdev->mddev->recovery);
 		md_wakeup_thread(rdev->mddev->thread);
 	} else if (rdev->mddev->pers) {
-		struct md_rdev *rdev2;
 		/* Activating a spare .. or possibly reactivating
 		 * if we ever get bitmaps working here.
 		 */
@@ -2710,10 +2759,6 @@ slot_store(struct md_rdev *rdev, const char *buf, size_t len)
 		if (rdev->mddev->pers->hot_add_disk == NULL)
 			return -EINVAL;
 
-		list_for_each_entry(rdev2, &rdev->mddev->disks, same_set)
-			if (rdev2->raid_disk == slot)
-				return -EEXIST;
-
 		if (slot >= rdev->mddev->raid_disks &&
 		    slot >= rdev->mddev->raid_disks + rdev->mddev->delta_disks)
 			return -ENOSPC;
@@ -6053,8 +6098,15 @@ static int md_ioctl(struct block_device *bdev, fmode_t mode,
 	struct mddev *mddev = NULL;
 	int ro;
 
-	if (!capable(CAP_SYS_ADMIN))
-		return -EACCES;
+	switch (cmd) {
+	case RAID_VERSION:
+	case GET_ARRAY_INFO:
+	case GET_DISK_INFO:
+		break;
+	default:
+		if (!capable(CAP_SYS_ADMIN))
+			return -EACCES;
+	}
 
 	/*
 	 * Commands dealing with the RAID driver but not any
@@ -6714,8 +6766,11 @@ static int md_seq_show(struct seq_file *seq, void *v)
 			if (test_bit(Faulty, &rdev->flags)) {
 				seq_printf(seq, "(F)");
 				continue;
-			} else if (rdev->raid_disk < 0)
+			}
+			if (rdev->raid_disk < 0)
 				seq_printf(seq, "(S)"); /* spare */
+			if (test_bit(Replacement, &rdev->flags))
+				seq_printf(seq, "(R)");
 			sectors += rdev->sectors;
 		}
 
@@ -7337,29 +7392,27 @@ static int remove_and_add_spares(struct mddev *mddev)
 		     ! test_bit(In_sync, &rdev->flags)) &&
 		    atomic_read(&rdev->nr_pending)==0) {
 			if (mddev->pers->hot_remove_disk(
-				    mddev, rdev->raid_disk)==0) {
+				    mddev, rdev) == 0) {
 				sysfs_unlink_rdev(mddev, rdev);
 				rdev->raid_disk = -1;
 			}
 		}
 
-	if (mddev->degraded) {
-		list_for_each_entry(rdev, &mddev->disks, same_set) {
-			if (rdev->raid_disk >= 0 &&
-			    !test_bit(In_sync, &rdev->flags) &&
-			    !test_bit(Faulty, &rdev->flags))
+	list_for_each_entry(rdev, &mddev->disks, same_set) {
+		if (rdev->raid_disk >= 0 &&
+		    !test_bit(In_sync, &rdev->flags) &&
+		    !test_bit(Faulty, &rdev->flags))
+			spares++;
+		if (rdev->raid_disk < 0
+		    && !test_bit(Faulty, &rdev->flags)) {
+			rdev->recovery_offset = 0;
+			if (mddev->pers->
+			    hot_add_disk(mddev, rdev) == 0) {
+				if (sysfs_link_rdev(mddev, rdev))
+					/* failure here is OK */;
 				spares++;
-			if (rdev->raid_disk < 0
-			    && !test_bit(Faulty, &rdev->flags)) {
-				rdev->recovery_offset = 0;
-				if (mddev->pers->
-				    hot_add_disk(mddev, rdev) == 0) {
-					if (sysfs_link_rdev(mddev, rdev))
-						/* failure here is OK */;
-					spares++;
-					md_new_event(mddev);
-					set_bit(MD_CHANGE_DEVS, &mddev->flags);
-				}
+				md_new_event(mddev);
+				set_bit(MD_CHANGE_DEVS, &mddev->flags);
 			}
 		}
 	}
@@ -7474,7 +7527,7 @@ void md_check_recovery(struct mddev *mddev)
 				    test_bit(Faulty, &rdev->flags) &&
 				    atomic_read(&rdev->nr_pending)==0) {
 					if (mddev->pers->hot_remove_disk(
-						    mddev, rdev->raid_disk)==0) {
+						    mddev, rdev) == 0) {
 						sysfs_unlink_rdev(mddev, rdev);
 						rdev->raid_disk = -1;
 					}
diff --git a/drivers/md/md.h b/drivers/md/md.h
index cf742d9306e..44c63dfeeb2 100644
--- a/drivers/md/md.h
+++ b/drivers/md/md.h
@@ -72,34 +72,7 @@ struct md_rdev {
 	 * This reduces the burden of testing multiple flags in many cases
 	 */
 
-	unsigned long	flags;
-#define	Faulty		1		/* device is known to have a fault */
-#define	In_sync		2		/* device is in_sync with rest of array */
-#define	WriteMostly	4		/* Avoid reading if at all possible */
-#define	AutoDetected	7		/* added by auto-detect */
-#define Blocked		8		/* An error occurred but has not yet
-					 * been acknowledged by the metadata
-					 * handler, so don't allow writes
-					 * until it is cleared */
-#define WriteErrorSeen	9		/* A write error has been seen on this
-					 * device
-					 */
-#define FaultRecorded	10		/* Intermediate state for clearing
-					 * Blocked.  The Fault is/will-be
-					 * recorded in the metadata, but that
-					 * metadata hasn't been stored safely
-					 * on disk yet.
-					 */
-#define BlockedBadBlocks 11		/* A writer is blocked because they
-					 * found an unacknowledged bad-block.
-					 * This can safely be cleared at any
-					 * time, and the writer will re-check.
-					 * It may be set at any time, and at
-					 * worst the writer will timeout and
-					 * re-check.  So setting it as
-					 * accurately as possible is good, but
-					 * not absolutely critical.
-					 */
+	unsigned long	flags;	/* bit set of 'enum flag_bits' bits. */
 	wait_queue_head_t blocked_wait;
 
 	int desc_nr;			/* descriptor index in the superblock */
@@ -152,6 +125,44 @@ struct md_rdev {
 		sector_t size;		/* in sectors */
 	} badblocks;
 };
+enum flag_bits {
+	Faulty,			/* device is known to have a fault */
+	In_sync,		/* device is in_sync with rest of array */
+	WriteMostly,		/* Avoid reading if at all possible */
+	AutoDetected,		/* added by auto-detect */
+	Blocked,		/* An error occurred but has not yet
+				 * been acknowledged by the metadata
+				 * handler, so don't allow writes
+				 * until it is cleared */
+	WriteErrorSeen,		/* A write error has been seen on this
+				 * device
+				 */
+	FaultRecorded,		/* Intermediate state for clearing
+				 * Blocked.  The Fault is/will-be
+				 * recorded in the metadata, but that
+				 * metadata hasn't been stored safely
+				 * on disk yet.
+				 */
+	BlockedBadBlocks,	/* A writer is blocked because they
+				 * found an unacknowledged bad-block.
+				 * This can safely be cleared at any
+				 * time, and the writer will re-check.
+				 * It may be set at any time, and at
+				 * worst the writer will timeout and
+				 * re-check.  So setting it as
+				 * accurately as possible is good, but
+				 * not absolutely critical.
+				 */
+	WantReplacement,	/* This device is a candidate to be
+				 * hot-replaced, either because it has
+				 * reported some faults, or because
+				 * of explicit request.
+				 */
+	Replacement,		/* This device is a replacement for
+				 * a want_replacement device with same
+				 * raid_disk number.
+				 */
+};
 
 #define BB_LEN_MASK	(0x00000000000001FFULL)
 #define BB_OFFSET_MASK	(0x7FFFFFFFFFFFFE00ULL)
@@ -428,7 +439,7 @@ struct md_personality
 	 */
 	void (*error_handler)(struct mddev *mddev, struct md_rdev *rdev);
 	int (*hot_add_disk) (struct mddev *mddev, struct md_rdev *rdev);
-	int (*hot_remove_disk) (struct mddev *mddev, int number);
+	int (*hot_remove_disk) (struct mddev *mddev, struct md_rdev *rdev);
 	int (*spare_active) (struct mddev *mddev);
 	sector_t (*sync_request)(struct mddev *mddev, sector_t sector_nr, int *skipped, int go_faster);
 	int (*resize) (struct mddev *mddev, sector_t sectors);
@@ -482,15 +493,20 @@ static inline char * mdname (struct mddev * mddev)
 static inline int sysfs_link_rdev(struct mddev *mddev, struct md_rdev *rdev)
 {
 	char nm[20];
-	sprintf(nm, "rd%d", rdev->raid_disk);
-	return sysfs_create_link(&mddev->kobj, &rdev->kobj, nm);
+	if (!test_bit(Replacement, &rdev->flags)) {
+		sprintf(nm, "rd%d", rdev->raid_disk);
+		return sysfs_create_link(&mddev->kobj, &rdev->kobj, nm);
+	} else
+		return 0;
 }
 
 static inline void sysfs_unlink_rdev(struct mddev *mddev, struct md_rdev *rdev)
 {
 	char nm[20];
-	sprintf(nm, "rd%d", rdev->raid_disk);
-	sysfs_remove_link(&mddev->kobj, nm);
+	if (!test_bit(Replacement, &rdev->flags)) {
+		sprintf(nm, "rd%d", rdev->raid_disk);
+		sysfs_remove_link(&mddev->kobj, nm);
+	}
 }
 
 /*
diff --git a/drivers/md/multipath.c b/drivers/md/multipath.c
index 5899246fa37..a222f516660 100644
--- a/drivers/md/multipath.c
+++ b/drivers/md/multipath.c
@@ -292,17 +292,16 @@ static int multipath_add_disk(struct mddev *mddev, struct md_rdev *rdev)
 	return err;
 }
 
-static int multipath_remove_disk(struct mddev *mddev, int number)
+static int multipath_remove_disk(struct mddev *mddev, struct md_rdev *rdev)
 {
 	struct mpconf *conf = mddev->private;
 	int err = 0;
-	struct md_rdev *rdev;
+	int number = rdev->raid_disk;
 	struct multipath_info *p = conf->multipaths + number;
 
 	print_multipath_conf(conf);
 
-	rdev = p->rdev;
-	if (rdev) {
+	if (rdev == p->rdev) {
 		if (test_bit(In_sync, &rdev->flags) ||
 		    atomic_read(&rdev->nr_pending)) {
 			printk(KERN_ERR "hot-remove-disk, slot %d is identified"
diff --git a/drivers/md/raid1.c b/drivers/md/raid1.c
index ede2461e79c..cc24f0cb7ee 100644
--- a/drivers/md/raid1.c
+++ b/drivers/md/raid1.c
@@ -135,7 +135,7 @@ out_free_pages:
 			put_page(r1_bio->bios[j]->bi_io_vec[i].bv_page);
 	j = -1;
 out_free_bio:
-	while ( ++j < pi->raid_disks )
+	while (++j < pi->raid_disks)
 		bio_put(r1_bio->bios[j]);
 	r1bio_pool_free(r1_bio, data);
 	return NULL;
@@ -164,7 +164,7 @@ static void put_all_bios(struct r1conf *conf, struct r1bio *r1_bio)
 {
 	int i;
 
-	for (i = 0; i < conf->raid_disks; i++) {
+	for (i = 0; i < conf->raid_disks * 2; i++) {
 		struct bio **bio = r1_bio->bios + i;
 		if (!BIO_SPECIAL(*bio))
 			bio_put(*bio);
@@ -185,7 +185,7 @@ static void put_buf(struct r1bio *r1_bio)
 	struct r1conf *conf = r1_bio->mddev->private;
 	int i;
 
-	for (i=0; i<conf->raid_disks; i++) {
+	for (i = 0; i < conf->raid_disks * 2; i++) {
 		struct bio *bio = r1_bio->bios[i];
 		if (bio->bi_end_io)
 			rdev_dec_pending(conf->mirrors[i].rdev, r1_bio->mddev);
@@ -277,13 +277,14 @@ static inline void update_head_pos(int disk, struct r1bio *r1_bio)
 static int find_bio_disk(struct r1bio *r1_bio, struct bio *bio)
 {
 	int mirror;
-	int raid_disks = r1_bio->mddev->raid_disks;
+	struct r1conf *conf = r1_bio->mddev->private;
+	int raid_disks = conf->raid_disks;
 
-	for (mirror = 0; mirror < raid_disks; mirror++)
+	for (mirror = 0; mirror < raid_disks * 2; mirror++)
 		if (r1_bio->bios[mirror] == bio)
 			break;
 
-	BUG_ON(mirror == raid_disks);
+	BUG_ON(mirror == raid_disks * 2);
 	update_head_pos(mirror, r1_bio);
 
 	return mirror;
@@ -390,6 +391,11 @@ static void raid1_end_write_request(struct bio *bio, int error)
 	if (!uptodate) {
 		set_bit(WriteErrorSeen,
 			&conf->mirrors[mirror].rdev->flags);
+		if (!test_and_set_bit(WantReplacement,
+				      &conf->mirrors[mirror].rdev->flags))
+			set_bit(MD_RECOVERY_NEEDED, &
+				conf->mddev->recovery);
+
 		set_bit(R1BIO_WriteError, &r1_bio->state);
 	} else {
 		/*
@@ -505,7 +511,7 @@ static int read_balance(struct r1conf *conf, struct r1bio *r1_bio, int *max_sect
 		start_disk = conf->last_used;
 	}
 
-	for (i = 0 ; i < conf->raid_disks ; i++) {
+	for (i = 0 ; i < conf->raid_disks * 2 ; i++) {
 		sector_t dist;
 		sector_t first_bad;
 		int bad_sectors;
@@ -609,7 +615,7 @@ int md_raid1_congested(struct mddev *mddev, int bits)
 		return 1;
 
 	rcu_read_lock();
-	for (i = 0; i < mddev->raid_disks; i++) {
+	for (i = 0; i < conf->raid_disks; i++) {
 		struct md_rdev *rdev = rcu_dereference(conf->mirrors[i].rdev);
 		if (rdev && !test_bit(Faulty, &rdev->flags)) {
 			struct request_queue *q = bdev_get_queue(rdev->bdev);
@@ -974,7 +980,7 @@ read_again:
 	 */
 	plugged = mddev_check_plugged(mddev);
 
-	disks = conf->raid_disks;
+	disks = conf->raid_disks * 2;
  retry_write:
 	blocked_rdev = NULL;
 	rcu_read_lock();
@@ -988,7 +994,8 @@ read_again:
 		}
 		r1_bio->bios[i] = NULL;
 		if (!rdev || test_bit(Faulty, &rdev->flags)) {
-			set_bit(R1BIO_Degraded, &r1_bio->state);
+			if (i < conf->raid_disks)
+				set_bit(R1BIO_Degraded, &r1_bio->state);
 			continue;
 		}
 
@@ -1263,6 +1270,25 @@ static int raid1_spare_active(struct mddev *mddev)
 	 */
 	for (i = 0; i < conf->raid_disks; i++) {
 		struct md_rdev *rdev = conf->mirrors[i].rdev;
+		struct md_rdev *repl = conf->mirrors[conf->raid_disks + i].rdev;
+		if (repl
+		    && repl->recovery_offset == MaxSector
+		    && !test_bit(Faulty, &repl->flags)
+		    && !test_and_set_bit(In_sync, &repl->flags)) {
+			/* replacement has just become active */
+			if (!rdev ||
+			    !test_and_clear_bit(In_sync, &rdev->flags))
+				count++;
+			if (rdev) {
+				/* Replaced device not technically
+				 * faulty, but we need to be sure
+				 * it gets removed and never re-added
+				 */
+				set_bit(Faulty, &rdev->flags);
+				sysfs_notify_dirent_safe(
+					rdev->sysfs_state);
+			}
+		}
 		if (rdev
 		    && !test_bit(Faulty, &rdev->flags)
 		    && !test_and_set_bit(In_sync, &rdev->flags)) {
@@ -1286,7 +1312,7 @@ static int raid1_add_disk(struct mddev *mddev, struct md_rdev *rdev)
 	int mirror = 0;
 	struct mirror_info *p;
 	int first = 0;
-	int last = mddev->raid_disks - 1;
+	int last = conf->raid_disks - 1;
 
 	if (mddev->recovery_disabled == conf->recovery_disabled)
 		return -EBUSY;
@@ -1294,8 +1320,9 @@ static int raid1_add_disk(struct mddev *mddev, struct md_rdev *rdev)
 	if (rdev->raid_disk >= 0)
 		first = last = rdev->raid_disk;
 
-	for (mirror = first; mirror <= last; mirror++)
-		if ( !(p=conf->mirrors+mirror)->rdev) {
+	for (mirror = first; mirror <= last; mirror++) {
+		p = conf->mirrors+mirror;
+		if (!p->rdev) {
 
 			disk_stack_limits(mddev->gendisk, rdev->bdev,
 					  rdev->data_offset << 9);
@@ -1322,21 +1349,35 @@ static int raid1_add_disk(struct mddev *mddev, struct md_rdev *rdev)
 			rcu_assign_pointer(p->rdev, rdev);
 			break;
 		}
+		if (test_bit(WantReplacement, &p->rdev->flags) &&
+		    p[conf->raid_disks].rdev == NULL) {
+			/* Add this device as a replacement */
+			clear_bit(In_sync, &rdev->flags);
+			set_bit(Replacement, &rdev->flags);
+			rdev->raid_disk = mirror;
+			err = 0;
+			conf->fullsync = 1;
+			rcu_assign_pointer(p[conf->raid_disks].rdev, rdev);
+			break;
+		}
+	}
 	md_integrity_add_rdev(rdev, mddev);
 	print_conf(conf);
 	return err;
 }
 
-static int raid1_remove_disk(struct mddev *mddev, int number)
+static int raid1_remove_disk(struct mddev *mddev, struct md_rdev *rdev)
 {
 	struct r1conf *conf = mddev->private;
 	int err = 0;
-	struct md_rdev *rdev;
+	int number = rdev->raid_disk;
 	struct mirror_info *p = conf->mirrors+ number;
 
+	if (rdev != p->rdev)
+		p = conf->mirrors + conf->raid_disks + number;
+
 	print_conf(conf);
-	rdev = p->rdev;
-	if (rdev) {
+	if (rdev == p->rdev) {
 		if (test_bit(In_sync, &rdev->flags) ||
 		    atomic_read(&rdev->nr_pending)) {
 			err = -EBUSY;
@@ -1358,7 +1399,21 @@ static int raid1_remove_disk(struct mddev *mddev, int number)
 			err = -EBUSY;
 			p->rdev = rdev;
 			goto abort;
-		}
+		} else if (conf->mirrors[conf->raid_disks + number].rdev) {
+			/* We just removed a device that is being replaced.
+			 * Move down the replacement.  We drain all IO before
+			 * doing this to avoid confusion.
+			 */
+			struct md_rdev *repl =
+				conf->mirrors[conf->raid_disks + number].rdev;
+			raise_barrier(conf);
+			clear_bit(Replacement, &repl->flags);
+			p->rdev = repl;
+			conf->mirrors[conf->raid_disks + number].rdev = NULL;
+			lower_barrier(conf);
+			clear_bit(WantReplacement, &rdev->flags);
+		} else
+			clear_bit(WantReplacement, &rdev->flags);
 		err = md_integrity_register(mddev);
 	}
 abort:
@@ -1411,6 +1466,10 @@ static void end_sync_write(struct bio *bio, int error)
 		} while (sectors_to_go > 0);
 		set_bit(WriteErrorSeen,
 			&conf->mirrors[mirror].rdev->flags);
+		if (!test_and_set_bit(WantReplacement,
+				      &conf->mirrors[mirror].rdev->flags))
+			set_bit(MD_RECOVERY_NEEDED, &
+				mddev->recovery);
 		set_bit(R1BIO_WriteError, &r1_bio->state);
 	} else if (is_badblock(conf->mirrors[mirror].rdev,
 			       r1_bio->sector,
@@ -1441,8 +1500,13 @@ static int r1_sync_page_io(struct md_rdev *rdev, sector_t sector,
 	if (sync_page_io(rdev, sector, sectors << 9, page, rw, false))
 		/* success */
 		return 1;
-	if (rw == WRITE)
+	if (rw == WRITE) {
 		set_bit(WriteErrorSeen, &rdev->flags);
+		if (!test_and_set_bit(WantReplacement,
+				      &rdev->flags))
+			set_bit(MD_RECOVERY_NEEDED, &
+				rdev->mddev->recovery);
+	}
 	/* need to record an error - either for the block or the device */
 	if (!rdev_set_badblocks(rdev, sector, sectors, 0))
 		md_error(rdev->mddev, rdev);
@@ -1493,7 +1557,7 @@ static int fix_sync_read_error(struct r1bio *r1_bio)
 				}
 			}
 			d++;
-			if (d == conf->raid_disks)
+			if (d == conf->raid_disks * 2)
 				d = 0;
 		} while (!success && d != r1_bio->read_disk);
 
@@ -1510,7 +1574,7 @@ static int fix_sync_read_error(struct r1bio *r1_bio)
 			       mdname(mddev),
 			       bdevname(bio->bi_bdev, b),
 			       (unsigned long long)r1_bio->sector);
-			for (d = 0; d < conf->raid_disks; d++) {
+			for (d = 0; d < conf->raid_disks * 2; d++) {
 				rdev = conf->mirrors[d].rdev;
 				if (!rdev || test_bit(Faulty, &rdev->flags))
 					continue;
@@ -1536,7 +1600,7 @@ static int fix_sync_read_error(struct r1bio *r1_bio)
 		/* write it back and re-read */
 		while (d != r1_bio->read_disk) {
 			if (d == 0)
-				d = conf->raid_disks;
+				d = conf->raid_disks * 2;
 			d--;
 			if (r1_bio->bios[d]->bi_end_io != end_sync_read)
 				continue;
@@ -1551,7 +1615,7 @@ static int fix_sync_read_error(struct r1bio *r1_bio)
 		d = start;
 		while (d != r1_bio->read_disk) {
 			if (d == 0)
-				d = conf->raid_disks;
+				d = conf->raid_disks * 2;
 			d--;
 			if (r1_bio->bios[d]->bi_end_io != end_sync_read)
 				continue;
@@ -1584,7 +1648,7 @@ static int process_checks(struct r1bio *r1_bio)
 	int primary;
 	int i;
 
-	for (primary = 0; primary < conf->raid_disks; primary++)
+	for (primary = 0; primary < conf->raid_disks * 2; primary++)
 		if (r1_bio->bios[primary]->bi_end_io == end_sync_read &&
 		    test_bit(BIO_UPTODATE, &r1_bio->bios[primary]->bi_flags)) {
 			r1_bio->bios[primary]->bi_end_io = NULL;
@@ -1592,7 +1656,7 @@ static int process_checks(struct r1bio *r1_bio)
 			break;
 		}
 	r1_bio->read_disk = primary;
-	for (i = 0; i < conf->raid_disks; i++) {
+	for (i = 0; i < conf->raid_disks * 2; i++) {
 		int j;
 		int vcnt = r1_bio->sectors >> (PAGE_SHIFT- 9);
 		struct bio *pbio = r1_bio->bios[primary];
@@ -1656,7 +1720,7 @@ static void sync_request_write(struct mddev *mddev, struct r1bio *r1_bio)
 {
 	struct r1conf *conf = mddev->private;
 	int i;
-	int disks = conf->raid_disks;
+	int disks = conf->raid_disks * 2;
 	struct bio *bio, *wbio;
 
 	bio = r1_bio->bios[r1_bio->read_disk];
@@ -1737,7 +1801,7 @@ static void fix_read_error(struct r1conf *conf, int read_disk,
 				success = 1;
 			else {
 				d++;
-				if (d == conf->raid_disks)
+				if (d == conf->raid_disks * 2)
 					d = 0;
 			}
 		} while (!success && d != read_disk);
@@ -1753,7 +1817,7 @@ static void fix_read_error(struct r1conf *conf, int read_disk,
 		start = d;
 		while (d != read_disk) {
 			if (d==0)
-				d = conf->raid_disks;
+				d = conf->raid_disks * 2;
 			d--;
 			rdev = conf->mirrors[d].rdev;
 			if (rdev &&
@@ -1765,7 +1829,7 @@ static void fix_read_error(struct r1conf *conf, int read_disk,
 		while (d != read_disk) {
 			char b[BDEVNAME_SIZE];
 			if (d==0)
-				d = conf->raid_disks;
+				d = conf->raid_disks * 2;
 			d--;
 			rdev = conf->mirrors[d].rdev;
 			if (rdev &&
@@ -1887,7 +1951,7 @@ static void handle_sync_write_finished(struct r1conf *conf, struct r1bio *r1_bio
 {
 	int m;
 	int s = r1_bio->sectors;
-	for (m = 0; m < conf->raid_disks ; m++) {
+	for (m = 0; m < conf->raid_disks * 2 ; m++) {
 		struct md_rdev *rdev = conf->mirrors[m].rdev;
 		struct bio *bio = r1_bio->bios[m];
 		if (bio->bi_end_io == NULL)
@@ -1909,7 +1973,7 @@ static void handle_sync_write_finished(struct r1conf *conf, struct r1bio *r1_bio
 static void handle_write_finished(struct r1conf *conf, struct r1bio *r1_bio)
 {
 	int m;
-	for (m = 0; m < conf->raid_disks ; m++)
+	for (m = 0; m < conf->raid_disks * 2 ; m++)
 		if (r1_bio->bios[m] == IO_MADE_GOOD) {
 			struct md_rdev *rdev = conf->mirrors[m].rdev;
 			rdev_clear_badblocks(rdev,
@@ -2184,7 +2248,7 @@ static sector_t sync_request(struct mddev *mddev, sector_t sector_nr, int *skipp
 	r1_bio->state = 0;
 	set_bit(R1BIO_IsSync, &r1_bio->state);
 
-	for (i=0; i < conf->raid_disks; i++) {
+	for (i = 0; i < conf->raid_disks * 2; i++) {
 		struct md_rdev *rdev;
 		bio = r1_bio->bios[i];
 
@@ -2203,7 +2267,8 @@ static sector_t sync_request(struct mddev *mddev, sector_t sector_nr, int *skipp
 		rdev = rcu_dereference(conf->mirrors[i].rdev);
 		if (rdev == NULL ||
 		    test_bit(Faulty, &rdev->flags)) {
-			still_degraded = 1;
+			if (i < conf->raid_disks)
+				still_degraded = 1;
 		} else if (!test_bit(In_sync, &rdev->flags)) {
 			bio->bi_rw = WRITE;
 			bio->bi_end_io = end_sync_write;
@@ -2254,7 +2319,7 @@ static sector_t sync_request(struct mddev *mddev, sector_t sector_nr, int *skipp
 		 * need to mark them bad on all write targets
 		 */
 		int ok = 1;
-		for (i = 0 ; i < conf->raid_disks ; i++)
+		for (i = 0 ; i < conf->raid_disks * 2 ; i++)
 			if (r1_bio->bios[i]->bi_end_io == end_sync_write) {
 				struct md_rdev *rdev =
 					rcu_dereference(conf->mirrors[i].rdev);
@@ -2323,7 +2388,7 @@ static sector_t sync_request(struct mddev *mddev, sector_t sector_nr, int *skipp
 				len = sync_blocks<<9;
 		}
 
-		for (i=0 ; i < conf->raid_disks; i++) {
+		for (i = 0 ; i < conf->raid_disks * 2; i++) {
 			bio = r1_bio->bios[i];
 			if (bio->bi_end_io) {
 				page = bio->bi_io_vec[bio->bi_vcnt].bv_page;
@@ -2356,7 +2421,7 @@ static sector_t sync_request(struct mddev *mddev, sector_t sector_nr, int *skipp
 	 */
 	if (test_bit(MD_RECOVERY_REQUESTED, &mddev->recovery)) {
 		atomic_set(&r1_bio->remaining, read_targets);
-		for (i=0; i<conf->raid_disks; i++) {
+		for (i = 0; i < conf->raid_disks * 2; i++) {
 			bio = r1_bio->bios[i];
 			if (bio->bi_end_io == end_sync_read) {
 				md_sync_acct(bio->bi_bdev, nr_sectors);
@@ -2393,7 +2458,8 @@ static struct r1conf *setup_conf(struct mddev *mddev)
 	if (!conf)
 		goto abort;
 
-	conf->mirrors = kzalloc(sizeof(struct mirror_info)*mddev->raid_disks,
+	conf->mirrors = kzalloc(sizeof(struct mirror_info)
+				* mddev->raid_disks * 2,
 				 GFP_KERNEL);
 	if (!conf->mirrors)
 		goto abort;
@@ -2405,7 +2471,7 @@ static struct r1conf *setup_conf(struct mddev *mddev)
 	conf->poolinfo = kzalloc(sizeof(*conf->poolinfo), GFP_KERNEL);
 	if (!conf->poolinfo)
 		goto abort;
-	conf->poolinfo->raid_disks = mddev->raid_disks;
+	conf->poolinfo->raid_disks = mddev->raid_disks * 2;
 	conf->r1bio_pool = mempool_create(NR_RAID1_BIOS, r1bio_pool_alloc,
 					  r1bio_pool_free,
 					  conf->poolinfo);
@@ -2414,14 +2480,20 @@ static struct r1conf *setup_conf(struct mddev *mddev)
 
 	conf->poolinfo->mddev = mddev;
 
+	err = -EINVAL;
 	spin_lock_init(&conf->device_lock);
 	list_for_each_entry(rdev, &mddev->disks, same_set) {
 		int disk_idx = rdev->raid_disk;
 		if (disk_idx >= mddev->raid_disks
 		    || disk_idx < 0)
 			continue;
-		disk = conf->mirrors + disk_idx;
+		if (test_bit(Replacement, &rdev->flags))
+			disk = conf->mirrors + conf->raid_disks + disk_idx;
+		else
+			disk = conf->mirrors + disk_idx;
 
+		if (disk->rdev)
+			goto abort;
 		disk->rdev = rdev;
 
 		disk->head_position = 0;
@@ -2437,11 +2509,27 @@ static struct r1conf *setup_conf(struct mddev *mddev)
 	conf->pending_count = 0;
 	conf->recovery_disabled = mddev->recovery_disabled - 1;
 
+	err = -EIO;
 	conf->last_used = -1;
-	for (i = 0; i < conf->raid_disks; i++) {
+	for (i = 0; i < conf->raid_disks * 2; i++) {
 
 		disk = conf->mirrors + i;
 
+		if (i < conf->raid_disks &&
+		    disk[conf->raid_disks].rdev) {
+			/* This slot has a replacement. */
+			if (!disk->rdev) {
+				/* No original, just make the replacement
+				 * a recovering spare
+				 */
+				disk->rdev =
+					disk[conf->raid_disks].rdev;
+