md updates for 3.9

mostly little bugfixes.
 Only "feature" is a new RAID10 layout which slightly
 improves the number of sets of devices that can concurrently
 fail, without data loss.
 -----BEGIN PGP SIGNATURE-----
 Version: GnuPG v2.0.19 (GNU/Linux)
 
 iQIVAwUAUTPm+znsnt1WYoG5AQLLsw/+PMqr8roC4twgxTWV1NRbU8NtOcRi9Rj9
 uvBS63uYAaLdi/D3UBKFYczmNCu9knuXbcp9SgFDxH7LlthQsWN/GYnif06pPo3w
 9Agu5M8c062TJEG1vrnX6FhPO6pNgrWFr3h+CKkTiD3179i9DoQpP8LXQToeyMtI
 YRMQf/zCkxYtDvWAP0iwsEWtw8cf+q9I/uGPhQ1L+DnZapXYdbtnqWBRz9q6mrDt
 orcGrP41aZHvnOHUaTbwmaorCKkf/Ys4SMaGenrSFpnpQMypt7VgNuwHC59LxvJT
 5eiFG/26zIsv7Wk0jv/TvFP5qzUPo0/PFkd5ug0ArvbVRiXS2cMJDwQvMdO1toxD
 i5Bb+P9DptadvoWhOTgIpxnG77yRH45wJvyJOk+ZfS1/IO87nCRa3d0yiNOU5e2/
 o0VdXPZRr72sdKKTK6kQuYfwCPb+Z2Pz6Q8BJdk6GxlmTXyP6sKhIgwUX86534fE
 LrOxfK8qV+GetVu3X02RoX2CyJJRQHXyXmbHuSzXuo/JiOYtDigAydwNZChvf+tf
 OoMY9K8vgNbhnGsUG6la7XPvZ+6dZMjdnxp2HB99Ml5A3PWZd75i5T6IHHxIQFbD
 C3z9PWTWP+hK4k15DEyjlELtsE9WduGTXG4kUcf328xJ/7lj4VIImVugdCz+1B6z
 +HlI6BiLwzY=
 =YdVD
 -----END PGP SIGNATURE-----

Merge tag 'md-3.9' of git://neil.brown.name/md

Pull md updates from NeilBrown:
 "Mostly little bugfixes.

  Only "feature" is a new RAID10 layout which slightly improves the
  number of sets of devices that can concurrently fail, without data
  loss."

* tag 'md-3.9' of git://neil.brown.name/md:
  md: expedite metadata update when switching  read-auto -> active
  md: remove CONFIG_MULTICORE_RAID456
  md/raid1,raid10: fix deadlock with freeze_array()
  md/raid0: improve error message when converting RAID4-with-spares to RAID0
  md: raid0: fix error return from create_stripe_zones.
  md: fix two bugs when attempting to resize RAID0 array.
  DM RAID: Add support for MD's RAID10 "far" and "offset" algorithms
  MD RAID10: Improve redundancy for 'far' and 'offset' algorithms (part 2)
  MD RAID10: Improve redundancy for 'far' and 'offset' algorithms (part 1)
  MD RAID10: Minor non-functional code changes
  md: raid1,10: Handle REQ_WRITE_SAME flag in write bios
  md: protect against crash upon fsync on ro array

Documentation/device-mapper/dm-raid.txt

@@ -30,6 +30,7 @@ The target is named "raid" and it accepts the following parameters:
   raid10        Various RAID10 inspired algorithms chosen by additional params
 		- RAID10: Striped Mirrors (aka 'Striping on top of mirrors')
 		- RAID1E: Integrated Adjacent Stripe Mirroring
+		- RAID1E: Integrated Offset Stripe Mirroring
 		-  and other similar RAID10 variants
 
   Reference: Chapter 4 of
@@ -64,15 +65,15 @@ The target is named "raid" and it accepts the following parameters:
 		synchronisation state for each region.
 
         [raid10_copies   <# copies>]
-        [raid10_format   near]
+        [raid10_format   <near|far|offset>]
 		These two options are used to alter the default layout of
 		a RAID10 configuration.  The number of copies is can be
-		specified, but the default is 2.  There are other variations
+		specified, but the default is 2.  There are also three
-		to how the copies are laid down - the default and only current
+		variations to how the copies are laid down - the default
-		option is "near".  Near copies are what most people think of
+		is "near".  Near copies are what most people think of with
-		with respect to mirroring.  If these options are left
+		respect to mirroring.  If these options are left unspecified,
-		unspecified, or 'raid10_copies 2' and/or 'raid10_format near'
+		or 'raid10_copies 2' and/or 'raid10_format near' are given,
-		are given, then the layouts for 2, 3 and 4 devices are:
+		then the layouts for 2, 3 and 4 devices	are:
 		2 drives         3 drives          4 drives
 		--------         ----------        --------------
 		A1  A1           A1  A1  A2        A1  A1  A2  A2
@@ -85,6 +86,33 @@ The target is named "raid" and it accepts the following parameters:
 		3-device layout is what might be called a 'RAID1E - Integrated
 		Adjacent Stripe Mirroring'.
 
+		If 'raid10_copies 2' and 'raid10_format far', then the layouts
+		for 2, 3 and 4 devices are:
+		2 drives             3 drives             4 drives
+		--------             --------------       --------------------
+		A1  A2               A1   A2   A3         A1   A2   A3   A4
+		A3  A4               A4   A5   A6         A5   A6   A7   A8
+		A5  A6               A7   A8   A9         A9   A10  A11  A12
+		..  ..               ..   ..   ..         ..   ..   ..   ..
+		A2  A1               A3   A1   A2         A2   A1   A4   A3
+		A4  A3               A6   A4   A5         A6   A5   A8   A7
+		A6  A5               A9   A7   A8         A10  A9   A12  A11
+		..  ..               ..   ..   ..         ..   ..   ..   ..
+
+		If 'raid10_copies 2' and 'raid10_format offset', then the
+		layouts for 2, 3 and 4 devices are:
+		2 drives       3 drives           4 drives
+		--------       ------------       -----------------
+		A1  A2         A1  A2  A3         A1  A2  A3  A4
+		A2  A1         A3  A1  A2         A2  A1  A4  A3
+		A3  A4         A4  A5  A6         A5  A6  A7  A8
+		A4  A3         A6  A4  A5         A6  A5  A8  A7
+		A5  A6         A7  A8  A9         A9  A10 A11 A12
+		A6  A5         A9  A7  A8         A10 A9  A12 A11
+		..  ..         ..  ..  ..         ..  ..  ..  ..
+		Here we see layouts closely akin to 'RAID1E - Integrated
+		Offset Stripe Mirroring'.
+
 <#raid_devs>: The number of devices composing the array.
 	Each device consists of two entries.  The first is the device
 	containing the metadata (if any); the second is the one containing the
@@ -142,3 +170,5 @@ Version History
 1.3.0	Added support for RAID 10
 1.3.1	Allow device replacement/rebuild for RAID 10
 1.3.2   Fix/improve redundancy checking for RAID10
+1.4.0	Non-functional change.  Removes arg from mapping function.
+1.4.1   Add RAID10 "far" and "offset" algorithm support.

drivers/md/Kconfig

@@ -154,17 +154,6 @@ config MD_RAID456
 
 	  If unsure, say Y.
 
-config MULTICORE_RAID456
-	bool "RAID-4/RAID-5/RAID-6 Multicore processing (EXPERIMENTAL)"
-	depends on MD_RAID456
-	depends on SMP
-	depends on EXPERIMENTAL
-	---help---
-	  Enable the raid456 module to dispatch per-stripe raid operations to a
-	  thread pool.
-
-	  If unsure, say N.
-
 config MD_MULTIPATH
 	tristate "Multipath I/O support"
 	depends on BLK_DEV_MD

drivers/md/dm-raid.c

@@ -91,15 +91,44 @@ static struct raid_type {
 	{"raid6_nc", "RAID6 (N continue)",		2, 4, 6, ALGORITHM_ROTATING_N_CONTINUE}
 };
 
+static char *raid10_md_layout_to_format(int layout)
+{
+	/*
+	 * Bit 16 and 17 stand for "offset" and "use_far_sets"
+	 * Refer to MD's raid10.c for details
+	 */
+	if ((layout & 0x10000) && (layout & 0x20000))
+		return "offset";
+
+	if ((layout & 0xFF) > 1)
+		return "near";
+
+	return "far";
+}
+
 static unsigned raid10_md_layout_to_copies(int layout)
 {
-	return layout & 0xFF;
+	if ((layout & 0xFF) > 1)
+		return layout & 0xFF;
+	return (layout >> 8) & 0xFF;
 }
 
 static int raid10_format_to_md_layout(char *format, unsigned copies)
 {
-	/* 1 "far" copy, and 'copies' "near" copies */
+	unsigned n = 1, f = 1;
-	return (1 << 8) | (copies & 0xFF);
+
+	if (!strcmp("near", format))
+		n = copies;
+	else
+		f = copies;
+
+	if (!strcmp("offset", format))
+		return 0x30000 | (f << 8) | n;
+
+	if (!strcmp("far", format))
+		return 0x20000 | (f << 8) | n;
+
+	return (f << 8) | n;
 }
 
 static struct raid_type *get_raid_type(char *name)
@@ -352,6 +381,7 @@ static int validate_raid_redundancy(struct raid_set *rs)
 {
 	unsigned i, rebuild_cnt = 0;
 	unsigned rebuilds_per_group, copies, d;
+	unsigned group_size, last_group_start;
 
 	for (i = 0; i < rs->md.raid_disks; i++)
 		if (!test_bit(In_sync, &rs->dev[i].rdev.flags) ||
@@ -379,9 +409,6 @@ static int validate_raid_redundancy(struct raid_set *rs)
 		 * as long as the failed devices occur in different mirror
 		 * groups (i.e. different stripes).
 		 *
-		 * Right now, we only allow for "near" copies.  When other
-		 * formats are added, we will have to check those too.
-		 *
 		 * When checking "near" format, make sure no adjacent devices
 		 * have failed beyond what can be handled.  In addition to the
 		 * simple case where the number of devices is a multiple of the
@@ -391,14 +418,41 @@ static int validate_raid_redundancy(struct raid_set *rs)
 		 *          A    A    B    B    C
 		 *          C    D    D    E    E
 		 */
-		for (i = 0; i < rs->md.raid_disks * copies; i++) {
+		if (!strcmp("near", raid10_md_layout_to_format(rs->md.layout))) {
-			if (!(i % copies))
+			for (i = 0; i < rs->md.raid_disks * copies; i++) {
+				if (!(i % copies))
+					rebuilds_per_group = 0;
+				d = i % rs->md.raid_disks;
+				if ((!rs->dev[d].rdev.sb_page ||
+				     !test_bit(In_sync, &rs->dev[d].rdev.flags)) &&
+				    (++rebuilds_per_group >= copies))
+					goto too_many;
+			}
+			break;
+		}
+
+		/*
+		 * When checking "far" and "offset" formats, we need to ensure
+		 * that the device that holds its copy is not also dead or
+		 * being rebuilt.  (Note that "far" and "offset" formats only
+		 * support two copies right now.  These formats also only ever
+		 * use the 'use_far_sets' variant.)
+		 *
+		 * This check is somewhat complicated by the need to account
+		 * for arrays that are not a multiple of (far) copies.  This
+		 * results in the need to treat the last (potentially larger)
+		 * set differently.
+		 */
+		group_size = (rs->md.raid_disks / copies);
+		last_group_start = (rs->md.raid_disks / group_size) - 1;
+		last_group_start *= group_size;
+		for (i = 0; i < rs->md.raid_disks; i++) {
+			if (!(i % copies) && !(i > last_group_start))
 				rebuilds_per_group = 0;
-			d = i % rs->md.raid_disks;
+			if ((!rs->dev[i].rdev.sb_page ||
-			if ((!rs->dev[d].rdev.sb_page ||
+			     !test_bit(In_sync, &rs->dev[i].rdev.flags)) &&
-			     !test_bit(In_sync, &rs->dev[d].rdev.flags)) &&
 			    (++rebuilds_per_group >= copies))
-				goto too_many;
+					goto too_many;
 		}
 		break;
 	default:
@@ -433,7 +487,7 @@ too_many:
  *
  * RAID10-only options:
  *    [raid10_copies <# copies>]        Number of copies.  (Default: 2)
- *    [raid10_format <near>]            Layout algorithm.  (Default: near)
+ *    [raid10_format <near|far|offset>] Layout algorithm.  (Default: near)
  */
 static int parse_raid_params(struct raid_set *rs, char **argv,
 			     unsigned num_raid_params)
@@ -520,7 +574,9 @@ static int parse_raid_params(struct raid_set *rs, char **argv,
 				rs->ti->error = "'raid10_format' is an invalid parameter for this RAID type";
 				return -EINVAL;
 			}
-			if (strcmp("near", argv[i])) {
+			if (strcmp("near", argv[i]) &&
+			    strcmp("far", argv[i]) &&
+			    strcmp("offset", argv[i])) {
 				rs->ti->error = "Invalid 'raid10_format' value given";
 				return -EINVAL;
 			}
@@ -644,6 +700,15 @@ static int parse_raid_params(struct raid_set *rs, char **argv,
 			return -EINVAL;
 		}
 
+		/*
+		 * If the format is not "near", we only support
+		 * two copies at the moment.
+		 */
+		if (strcmp("near", raid10_format) && (raid10_copies > 2)) {
+			rs->ti->error = "Too many copies for given RAID10 format.";
+			return -EINVAL;
+		}
+
 		/* (Len * #mirrors) / #devices */
 		sectors_per_dev = rs->ti->len * raid10_copies;
 		sector_div(sectors_per_dev, rs->md.raid_disks);
@@ -854,17 +919,30 @@ static int super_init_validation(struct mddev *mddev, struct md_rdev *rdev)
 	/*
 	 * Reshaping is not currently allowed
 	 */
-	if ((le32_to_cpu(sb->level) != mddev->level) ||
+	if (le32_to_cpu(sb->level) != mddev->level) {
-	    (le32_to_cpu(sb->layout) != mddev->layout) ||
+		DMERR("Reshaping arrays not yet supported. (RAID level change)");
-	    (le32_to_cpu(sb->stripe_sectors) != mddev->chunk_sectors)) {
+		return -EINVAL;
-		DMERR("Reshaping arrays not yet supported.");
+	}
+	if (le32_to_cpu(sb->layout) != mddev->layout) {
+		DMERR("Reshaping arrays not yet supported. (RAID layout change)");
+		DMERR("  0x%X vs 0x%X", le32_to_cpu(sb->layout), mddev->layout);
+		DMERR("  Old layout: %s w/ %d copies",
+		      raid10_md_layout_to_format(le32_to_cpu(sb->layout)),
+		      raid10_md_layout_to_copies(le32_to_cpu(sb->layout)));
+		DMERR("  New layout: %s w/ %d copies",
+		      raid10_md_layout_to_format(mddev->layout),
+		      raid10_md_layout_to_copies(mddev->layout));
+		return -EINVAL;
+	}
+	if (le32_to_cpu(sb->stripe_sectors) != mddev->chunk_sectors) {
+		DMERR("Reshaping arrays not yet supported. (stripe sectors change)");
 		return -EINVAL;
 	}
 
 	/* We can only change the number of devices in RAID1 right now */
 	if ((rs->raid_type->level != 1) &&
 	    (le32_to_cpu(sb->num_devices) != mddev->raid_disks)) {
-		DMERR("Reshaping arrays not yet supported.");
+		DMERR("Reshaping arrays not yet supported. (device count change)");
 		return -EINVAL;
 	}
 
@@ -1329,7 +1407,8 @@ static void raid_status(struct dm_target *ti, status_type_t type,
 			       raid10_md_layout_to_copies(rs->md.layout));
 
 		if (rs->print_flags & DMPF_RAID10_FORMAT)
-			DMEMIT(" raid10_format near");
+			DMEMIT(" raid10_format %s",
+			       raid10_md_layout_to_format(rs->md.layout));
 
 		DMEMIT(" %d", rs->md.raid_disks);
 		for (i = 0; i < rs->md.raid_disks; i++) {
@@ -1418,6 +1497,10 @@ static struct target_type raid_target = {
 
 static int __init dm_raid_init(void)
 {
+	DMINFO("Loading target version %u.%u.%u",
+	       raid_target.version[0],
+	       raid_target.version[1],
+	       raid_target.version[2]);
 	return dm_register_target(&raid_target);
 }
 

drivers/md/md.c

@@ -307,6 +307,10 @@ static void md_make_request(struct request_queue *q, struct bio *bio)
 		bio_io_error(bio);
 		return;
 	}
+	if (mddev->ro == 1 && unlikely(rw == WRITE)) {
+		bio_endio(bio, bio_sectors(bio) == 0 ? 0 : -EROFS);
+		return;
+	}
 	smp_rmb(); /* Ensure implications of  'active' are visible */
 	rcu_read_lock();
 	if (mddev->suspended) {
@@ -2994,6 +2998,9 @@ rdev_size_store(struct md_rdev *rdev, const char *buf, size_t len)
 		} else if (!sectors)
 			sectors = (i_size_read(rdev->bdev->bd_inode) >> 9) -
 				rdev->data_offset;
+		if (!my_mddev->pers->resize)
+			/* Cannot change size for RAID0 or Linear etc */
+			return -EINVAL;
 	}
 	if (sectors < my_mddev->dev_sectors)
 		return -EINVAL; /* component must fit device */
@@ -6525,7 +6532,17 @@ static int md_ioctl(struct block_device *bdev, fmode_t mode,
 			mddev->ro = 0;
 			sysfs_notify_dirent_safe(mddev->sysfs_state);
 			set_bit(MD_RECOVERY_NEEDED, &mddev->recovery);
-			md_wakeup_thread(mddev->thread);
+			/* mddev_unlock will wake thread */
+			/* If a device failed while we were read-only, we
+			 * need to make sure the metadata is updated now.
+			 */
+			if (test_bit(MD_CHANGE_DEVS, &mddev->flags)) {
+				mddev_unlock(mddev);
+				wait_event(mddev->sb_wait,
+					   !test_bit(MD_CHANGE_DEVS, &mddev->flags) &&
+					   !test_bit(MD_CHANGE_PENDING, &mddev->flags));
+				mddev_lock(mddev);
+			}
 		} else {
 			err = -EROFS;
 			goto abort_unlock;

drivers/md/raid0.c

@@ -175,7 +175,13 @@ static int create_strip_zones(struct mddev *mddev, struct r0conf **private_conf)
 			rdev1->new_raid_disk = j;
 		}
 
-		if (j < 0 || j >= mddev->raid_disks) {
+		if (j < 0) {
+			printk(KERN_ERR
+			       "md/raid0:%s: remove inactive devices before converting to RAID0\n",
+			       mdname(mddev));
+			goto abort;
+		}
+		if (j >= mddev->raid_disks) {
 			printk(KERN_ERR "md/raid0:%s: bad disk number %d - "
 			       "aborting!\n", mdname(mddev), j);
 			goto abort;
@@ -289,7 +295,7 @@ abort:
 	kfree(conf->strip_zone);
 	kfree(conf->devlist);
 	kfree(conf);
-	*private_conf = NULL;
+	*private_conf = ERR_PTR(err);
 	return err;
 }
 
@@ -411,7 +417,8 @@ static sector_t raid0_size(struct mddev *mddev, sector_t sectors, int raid_disks
 		  "%s does not support generic reshape\n", __func__);
 
 	rdev_for_each(rdev, mddev)
-		array_sectors += rdev->sectors;
+		array_sectors += (rdev->sectors &
+				  ~(sector_t)(mddev->chunk_sectors-1));
 
 	return array_sectors;
 }

drivers/md/raid1.c

@@ -967,6 +967,7 @@ static void raid1_unplug(struct blk_plug_cb *cb, bool from_schedule)
 		bio_list_merge(&conf->pending_bio_list, &plug->pending);
 		conf->pending_count += plug->pending_cnt;
 		spin_unlock_irq(&conf->device_lock);
+		wake_up(&conf->wait_barrier);
 		md_wakeup_thread(mddev->thread);
 		kfree(plug);
 		return;
@@ -1000,6 +1001,7 @@ static void make_request(struct mddev *mddev, struct bio * bio)
 	const unsigned long do_flush_fua = (bio->bi_rw & (REQ_FLUSH | REQ_FUA));
 	const unsigned long do_discard = (bio->bi_rw
 					  & (REQ_DISCARD | REQ_SECURE));
+	const unsigned long do_same = (bio->bi_rw & REQ_WRITE_SAME);
 	struct md_rdev *blocked_rdev;
 	struct blk_plug_cb *cb;
 	struct raid1_plug_cb *plug = NULL;
@@ -1301,7 +1303,8 @@ read_again:
 				   conf->mirrors[i].rdev->data_offset);
 		mbio->bi_bdev = conf->mirrors[i].rdev->bdev;
 		mbio->bi_end_io	= raid1_end_write_request;
-		mbio->bi_rw = WRITE | do_flush_fua | do_sync | do_discard;
+		mbio->bi_rw =
+			WRITE | do_flush_fua | do_sync | do_discard | do_same;
 		mbio->bi_private = r1_bio;
 
 		atomic_inc(&r1_bio->remaining);
@@ -2818,6 +2821,9 @@ static int run(struct mddev *mddev)
 	if (IS_ERR(conf))
 		return PTR_ERR(conf);
 
+	if (mddev->queue)
+		blk_queue_max_write_same_sectors(mddev->queue,
+						 mddev->chunk_sectors);
 	rdev_for_each(rdev, mddev) {
 		if (!mddev->gendisk)
 			continue;

drivers/md/raid10.c

@@ -38,21 +38,36 @@
  *    near_copies (stored in low byte of layout)
  *    far_copies (stored in second byte of layout)
  *    far_offset (stored in bit 16 of layout )
+ *    use_far_sets (stored in bit 17 of layout )
  *
- * The data to be stored is divided into chunks using chunksize.
+ * The data to be stored is divided into chunks using chunksize.  Each device
- * Each device is divided into far_copies sections.
+ * is divided into far_copies sections.   In each section, chunks are laid out
- * In each section, chunks are laid out in a style similar to raid0, but
+ * in a style similar to raid0, but near_copies copies of each chunk is stored
- * near_copies copies of each chunk is stored (each on a different drive).
+ * (each on a different drive).  The starting device for each section is offset
- * The starting device for each section is offset near_copies from the starting
+ * near_copies from the starting device of the previous section.  Thus there
- * device of the previous section.
+ * are (near_copies * far_copies) of each chunk, and each is on a different
- * Thus they are (near_copies*far_copies) of each chunk, and each is on a different
+ * drive.  near_copies and far_copies must be at least one, and their product
- * drive.
+ * is at most raid_disks.
- * near_copies and far_copies must be at least one, and their product is at most
- * raid_disks.
  *
  * If far_offset is true, then the far_copies are handled a bit differently.
- * The copies are still in different stripes, but instead of be very far apart
+ * The copies are still in different stripes, but instead of being very far
- * on disk, there are adjacent stripes.
+ * apart on disk, there are adjacent stripes.
+ *
+ * The far and offset algorithms are handled slightly differently if
+ * 'use_far_sets' is true.  In this case, the array's devices are grouped into
+ * sets that are (near_copies * far_copies) in size.  The far copied stripes
+ * are still shifted by 'near_copies' devices, but this shifting stays confined
+ * to the set rather than the entire array.  This is done to improve the number
+ * of device combinations that can fail without causing the array to fail.
+ * Example 'far' algorithm w/o 'use_far_sets' (each letter represents a chunk
+ * on a device):
+ *    A B C D    A B C D E
+ *      ...         ...
+ *    D A B C    E A B C D
+ * Example 'far' algorithm w/ 'use_far_sets' enabled (sets illustrated w/ []'s):
+ *    [A B] [C D]    [A B] [C D E]
+ *    |...| |...|    |...| | ... |
+ *    [B A] [D C]    [B A] [E C D]
  */
 
 /*
@@ -535,6 +550,13 @@ static void __raid10_find_phys(struct geom *geo, struct r10bio *r10bio)
 	sector_t stripe;
 	int dev;
 	int slot = 0;
+	int last_far_set_start, last_far_set_size;
+
+	last_far_set_start = (geo->raid_disks / geo->far_set_size) - 1;
+	last_far_set_start *= geo->far_set_size;
+
+	last_far_set_size = geo->far_set_size;
+	last_far_set_size += (geo->raid_disks % geo->far_set_size);
 
 	/* now calculate first sector/dev */
 	chunk = r10bio->sector >> geo->chunk_shift;
@@ -551,15 +573,25 @@ static void __raid10_find_phys(struct geom *geo, struct r10bio *r10bio)
 	/* and calculate all the others */
 	for (n = 0; n < geo->near_copies; n++) {
 		int d = dev;
+		int set;
 		sector_t s = sector;
-		r10bio->devs[slot].addr = sector;
 		r10bio->devs[slot].devnum = d;
+		r10bio->devs[slot].addr = s;
 		slot++;
 
 		for (f = 1; f < geo->far_copies; f++) {
+			set = d / geo->far_set_size;
 			d += geo->near_copies;
-			if (d >= geo->raid_disks)
+
-				d -= geo->raid_disks;
+			if ((geo->raid_disks % geo->far_set_size) &&
+			    (d > last_far_set_start)) {
+				d -= last_far_set_start;
+				d %= last_far_set_size;
+				d += last_far_set_start;
+			} else {
+				d %= geo->far_set_size;
+				d += geo->far_set_size * set;
+			}
 			s += geo->stride;
 			r10bio->devs[slot].devnum = d;
 			r10bio->devs[slot].addr = s;
@@ -595,6 +627,20 @@ static sector_t raid10_find_virt(struct r10conf *conf, sector_t sector, int dev)
 	 * or recovery, so reshape isn't happening
 	 */
 	struct geom *geo = &conf->geo;
+	int far_set_start = (dev / geo->far_set_size) * geo->far_set_size;
+	int far_set_size = geo->far_set_size;
+	int last_far_set_start;
+
+	if (geo->raid_disks % geo->far_set_size) {
+		last_far_set_start = (geo->raid_disks / geo->far_set_size) - 1;
+		last_far_set_start *= geo->far_set_size;
+
+		if (dev >= last_far_set_start) {
+			far_set_size = geo->far_set_size;
+			far_set_size += (geo->raid_disks % geo->far_set_size);
+			far_set_start = last_far_set_start;
+		}
+	}
 
 	offset = sector & geo->chunk_mask;
 	if (geo->far_offset) {
@@ -602,13 +648,13 @@ static sector_t raid10_find_virt(struct r10conf *conf, sector_t sector, int dev)
 		chunk = sector >> geo->chunk_shift;
 		fc = sector_div(chunk, geo->far_copies);
 		dev -= fc * geo->near_copies;
-		if (dev < 0)
+		if (dev < far_set_start)
-			dev += geo->raid_disks;
+			dev += far_set_size;
 	} else {
 		while (sector >= geo->stride) {
 			sector -= geo->stride;
-			if (dev < geo->near_copies)
+			if (dev < (geo->near_copies + far_set_start))
-				dev += geo->raid_disks - geo->near_copies;
+				dev += far_set_size - geo->near_copies;
 			else
 				dev -= geo->near_copies;
 		}
@@ -1073,6 +1119,7 @@ static void raid10_unplug(struct blk_plug_cb *cb, bool from_schedule)
 		bio_list_merge(&conf->pending_bio_list, &plug->pending);
 		conf->pending_count += plug->pending_cnt;
 		spin_unlock_irq(&conf->device_lock);
+		wake_up(&conf->wait_barrier);
 		md_wakeup_thread(mddev->thread);
 		kfree(plug);
 		return;
@@ -1105,6 +1152,7 @@ static void make_request(struct mddev *mddev, struct bio * bio)
 	const unsigned long do_fua = (bio->bi_rw & REQ_FUA);
 	const unsigned long do_discard = (bio->bi_rw
 					  & (REQ_DISCARD | REQ_SECURE));
+	const unsigned long do_same = (bio->bi_rw & REQ_WRITE_SAME);
 	unsigned long flags;
 	struct md_rdev *blocked_rdev;
 	struct blk_plug_cb *cb;
@@ -1460,7 +1508,8 @@ retry_write:
 							      rdev));
 			mbio->bi_bdev = rdev->bdev;
 			mbio->bi_end_io	= raid10_end_write_request;
-			mbio->bi_rw = WRITE | do_sync | do_fua | do_discard;
+			mbio->bi_rw =
+				WRITE | do_sync | do_fua | do_discard | do_same;
 			mbio->bi_private = r10_bio;
 
 			atomic_inc(&r10_bio->remaining);
@@ -1502,7 +1551,8 @@ retry_write:
 						   r10_bio, rdev));
 			mbio->bi_bdev = rdev->bdev;
 			mbio->bi_end_io	= raid10_end_write_request;
-			mbio->bi_rw = WRITE | do_sync | do_fua | do_discard;
+			mbio->bi_rw =
+				WRITE | do_sync | do_fua | do_discard | do_same;
 			mbio->bi_private = r10_bio;
 
 			atomic_inc(&r10_bio->remaining);
@@ -3436,7 +3486,7 @@ static int setup_geo(struct geom *geo, struct mddev *mddev, enum geo_type new)
 		disks = mddev->raid_disks + mddev->delta_disks;
 		break;
 	}
-	if (layout >> 17)
+	if (layout >> 18)
 		return -1;
 	if (chunk < (PAGE_SIZE >> 9) ||
 	    !is_power_of_2(chunk))
@@ -3448,6 +3498,7 @@ static int setup_geo(struct geom *geo, struct mddev *mddev, enum geo_type new)
 	geo->near_copies = nc;
 	geo->far_copies = fc;
 	geo->far_offset = fo;
+	geo->far_set_size = (layout & (1<<17)) ? disks / fc : disks;
 	geo->chunk_mask = chunk - 1;
 	geo->chunk_shift = ffz(~chunk);
 	return nc*fc;
@@ -3569,6 +3620,8 @@ static int run(struct mddev *mddev)
 	if (mddev->queue) {
 		blk_queue_max_discard_sectors(mddev->queue,
 					      mddev->chunk_sectors);
+		blk_queue_max_write_same_sectors(mddev->queue,
+						 mddev->chunk_sectors);
 		blk_queue_io_min(mddev->queue, chunk_size);
 		if (conf->geo.raid_disks % conf->geo.near_copies)
 			blk_queue_io_opt(mddev->queue, chunk_size * conf->geo.raid_disks);

drivers/md/raid10.h

@@ -33,6 +33,11 @@ struct r10conf {
 					       * far_offset, in which case it is
 					       * 1 stripe.
 					       */
+		int             far_set_size; /* The number of devices in a set,
+					       * where a 'set' are devices that
+					       * contain far/offset copies of
+					       * each other.
+					       */
 		int		chunk_shift; /* shift from chunks to sectors */
 		sector_t	chunk_mask;
 	} prev, geo;

drivers/md/raid5.c

@@ -1403,7 +1403,7 @@ static void ops_run_check_pq(struct stripe_head *sh, struct raid5_percpu *percpu
 			   &sh->ops.zero_sum_result, percpu->spare_page, &submit);
 }
 
-static void __raid_run_ops(struct stripe_head *sh, unsigned long ops_request)
+static void raid_run_ops(struct stripe_head *sh, unsigned long ops_request)
 {
 	int overlap_clear = 0, i, disks = sh->disks;
 	struct dma_async_tx_descriptor *tx = NULL;
@@ -1468,36 +1468,6 @@ static void __raid_run_ops(struct stripe_head *sh, unsigned long ops_request)
 	put_cpu();
 }
 
-#ifdef CONFIG_MULTICORE_RAID456
-static void async_run_ops(void *param, async_cookie_t cookie)
-{
-	struct stripe_head *sh = param;
-	unsigned long ops_request = sh->ops.request;
-
-	clear_bit_unlock(STRIPE_OPS_REQ_PENDING, &sh->state);
-	wake_up(&sh->ops.wait_for_ops);
-
-	__raid_run_ops(sh, ops_request);
-	release_stripe(sh);
-}
-
-static void raid_run_ops(struct stripe_head *sh, unsigned long ops_request)
-{
-	/* since handle_stripe can be called outside of raid5d context
-	 * we need to ensure sh->ops.request is de-staged before another
-	 * request arrives
-	 */
-	wait_event(sh->ops.wait_for_ops,
-		   !test_and_set_bit_lock(STRIPE_OPS_REQ_PENDING, &sh->state));
-	sh->ops.request = ops_request;
-
-	atomic_inc(&sh->count);
-	async_schedule(async_run_ops, sh);
-}
-#else
-#define raid_run_ops __raid_run_ops
-#endif
-
 static int grow_one_stripe(struct r5conf *conf)
 {
 	struct stripe_head *sh;
@@ -1506,9 +1476,6 @@ static int grow_one_stripe(struct r5conf *conf)
 		return 0;
 
 	sh->raid_conf = conf;
-	#ifdef CONFIG_MULTICORE_RAID456
-	init_waitqueue_head(&sh->ops.wait_for_ops);
-	#endif
 
 	spin_lock_init(&sh->stripe_lock);
 
@@ -1627,9 +1594,6 @@ static int resize_stripes(struct r5conf *conf, int newsize)
 			break;
 
 		nsh->raid_conf = conf;
-		#ifdef CONFIG_MULTICORE_RAID456
-		init_waitqueue_head(&nsh->ops.wait_for_ops);
-		#endif
 		spin_lock_init(&nsh->stripe_lock);
 
 		list_add(&nsh->lru, &newstripes);