Merge tag 'md-3.7' of git://neil.brown.name/md

Pull md updates from NeilBrown:
 - "discard" support, some dm-raid improvements and other assorted bits
   and pieces.

* tag 'md-3.7' of git://neil.brown.name/md: (29 commits)
  md: refine reporting of resync/reshape delays.
  md/raid5: be careful not to resize_stripes too big.
  md: make sure manual changes to recovery checkpoint are saved.
  md/raid10: use correct limit variable
  md: writing to sync_action should clear the read-auto state.
  Subject: [PATCH] md:change resync_mismatches to atomic64_t to avoid races
  md/raid5: make sure to_read and to_write never go negative.
  md: When RAID5 is dirty, force reconstruct-write instead of read-modify-write.
  md/raid5: protect debug message against NULL derefernce.
  md/raid5: add some missing locking in handle_failed_stripe.
  MD: raid5 avoid unnecessary zero page for trim
  MD: raid5 trim support
  md/bitmap:Don't use IS_ERR to judge alloc_page().
  md/raid1: Don't release reference to device while handling read error.
  raid: replace list_for_each_continue_rcu with new interface
  add further __init annotations to crypto/xor.c
  DM RAID: Fix for "sync" directive ineffectiveness
  DM RAID: Fix comparison of index and quantity for "rebuild" parameter
  DM RAID: Add rebuild capability for RAID10
  DM RAID: Move 'rebuild' checking code to its own function
  ...

Documentation/device-mapper/dm-raid.txt

@@ -132,3 +132,12 @@ Here we can see the RAID type is raid4, there are 5 devices - all of
 which are 'A'live, and the array is 2/490221568 complete with recovery.
 Faulty or missing devices are marked 'D'.  Devices that are out-of-sync
 are marked 'a'.
+
+
+Version History
+---------------
+1.0.0	Initial version.  Support for RAID 4/5/6
+1.1.0	Added support for RAID 1
+1.2.0	Handle creation of arrays that contain failed devices.
+1.3.0	Added support for RAID 10
+1.3.1	Allow device replacement/rebuild for RAID 10

crypto/xor.c

@@ -56,11 +56,11 @@ xor_blocks(unsigned int src_count, unsigned int bytes, void *dest, void **srcs)
 EXPORT_SYMBOL(xor_blocks);
 
 /* Set of all registered templates.  */
-static struct xor_block_template *template_list;
+static struct xor_block_template *__initdata template_list;
 
 #define BENCH_SIZE (PAGE_SIZE)
 
-static void
+static void __init
 do_xor_speed(struct xor_block_template *tmpl, void *b1, void *b2)
 {
 	int speed;

drivers/md/bitmap.c

@@ -163,20 +163,17 @@ static struct md_rdev *next_active_rdev(struct md_rdev *rdev, struct mddev *mdde
 	 * As devices are only added or removed when raid_disk is < 0 and
 	 * nr_pending is 0 and In_sync is clear, the entries we return will
 	 * still be in the same position on the list when we re-enter
-	 * list_for_each_continue_rcu.
+	 * list_for_each_entry_continue_rcu.
 	 */
-	struct list_head *pos;
 	rcu_read_lock();
 	if (rdev == NULL)
 		/* start at the beginning */
-		pos = &mddev->disks;
+		rdev = list_entry_rcu(&mddev->disks, struct md_rdev, same_set);
 	else {
 		/* release the previous rdev and start from there. */
 		rdev_dec_pending(rdev, mddev);
-		pos = &rdev->same_set;
 	}
-	list_for_each_continue_rcu(pos, &mddev->disks) {
-		rdev = list_entry(pos, struct md_rdev, same_set);
+	list_for_each_entry_continue_rcu(rdev, &mddev->disks, same_set) {
 		if (rdev->raid_disk >= 0 &&
 		    !test_bit(Faulty, &rdev->flags)) {
 			/* this is a usable devices */
@@ -473,14 +470,10 @@ static int bitmap_new_disk_sb(struct bitmap *bitmap)
 {
 	bitmap_super_t *sb;
 	unsigned long chunksize, daemon_sleep, write_behind;
-	int err = -EINVAL;
 
 	bitmap->storage.sb_page = alloc_page(GFP_KERNEL);
-	if (IS_ERR(bitmap->storage.sb_page)) {
-		err = PTR_ERR(bitmap->storage.sb_page);
-		bitmap->storage.sb_page = NULL;
-		return err;
-	}
+	if (bitmap->storage.sb_page == NULL)
+		return -ENOMEM;
 	bitmap->storage.sb_page->index = 0;
 
 	sb = kmap_atomic(bitmap->storage.sb_page);

drivers/md/dm-raid.c

@@ -337,6 +337,84 @@ static int validate_region_size(struct raid_set *rs, unsigned long region_size)
 	return 0;
 }
 
+/*
+ * validate_rebuild_devices
+ * @rs
+ *
+ * Determine if the devices specified for rebuild can result in a valid
+ * usable array that is capable of rebuilding the given devices.
+ *
+ * Returns: 0 on success, -EINVAL on failure.
+ */
+static int validate_rebuild_devices(struct raid_set *rs)
+{
+	unsigned i, rebuild_cnt = 0;
+	unsigned rebuilds_per_group, copies, d;
+
+	if (!(rs->print_flags & DMPF_REBUILD))
+		return 0;
+
+	for (i = 0; i < rs->md.raid_disks; i++)
+		if (!test_bit(In_sync, &rs->dev[i].rdev.flags))
+			rebuild_cnt++;
+
+	switch (rs->raid_type->level) {
+	case 1:
+		if (rebuild_cnt >= rs->md.raid_disks)
+			goto too_many;
+		break;
+	case 4:
+	case 5:
+	case 6:
+		if (rebuild_cnt > rs->raid_type->parity_devs)
+			goto too_many;
+		break;
+	case 10:
+		copies = raid10_md_layout_to_copies(rs->md.layout);
+		if (rebuild_cnt < copies)
+			break;
+
+		/*
+		 * It is possible to have a higher rebuild count for RAID10,
+		 * 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
+		 * number of copies, we must also handle cases where the number
+		 * of devices is not a multiple of the number of copies.
+		 * E.g.    dev1 dev2 dev3 dev4 dev5
+		 *          A    A    B    B    C
+		 *          C    D    D    E    E
+		 */
+		rebuilds_per_group = 0;
+		for (i = 0; i < rs->md.raid_disks * copies; i++) {
+			d = i % rs->md.raid_disks;
+			if (!test_bit(In_sync, &rs->dev[d].rdev.flags) &&
+			    (++rebuilds_per_group >= copies))
+				goto too_many;
+			if (!((i + 1) % copies))
+				rebuilds_per_group = 0;
+		}
+		break;
+	default:
+		DMERR("The rebuild parameter is not supported for %s",
+		      rs->raid_type->name);
+		rs->ti->error = "Rebuild not supported for this RAID type";
+		return -EINVAL;
+	}
+
+	return 0;
+
+too_many:
+	rs->ti->error = "Too many rebuild devices specified";
+	return -EINVAL;
+}
+
 /*
  * Possible arguments are...
  *	<chunk_size> [optional_args]
@@ -365,7 +443,7 @@ static int parse_raid_params(struct raid_set *rs, char **argv,
 {
 	char *raid10_format = "near";
 	unsigned raid10_copies = 2;
-	unsigned i, rebuild_cnt = 0;
+	unsigned i;
 	unsigned long value, region_size = 0;
 	sector_t sectors_per_dev = rs->ti->len;
 	sector_t max_io_len;
@@ -461,31 +539,7 @@ static int parse_raid_params(struct raid_set *rs, char **argv,
 
 		/* Parameters that take a numeric value are checked here */
 		if (!strcasecmp(key, "rebuild")) {
-			rebuild_cnt++;
-
-			switch (rs->raid_type->level) {
-			case 1:
-				if (rebuild_cnt >= rs->md.raid_disks) {
-					rs->ti->error = "Too many rebuild devices specified";
-					return -EINVAL;
-				}
-				break;
-			case 4:
-			case 5:
-			case 6:
-				if (rebuild_cnt > rs->raid_type->parity_devs) {
-					rs->ti->error = "Too many rebuild devices specified for given RAID type";
-					return -EINVAL;
-				}
-				break;
-			case 10:
-			default:
-				DMERR("The rebuild parameter is not supported for %s", rs->raid_type->name);
-				rs->ti->error = "Rebuild not supported for this RAID type";
-				return -EINVAL;
-			}
-
-			if (value > rs->md.raid_disks) {
+			if (value >= rs->md.raid_disks) {
 				rs->ti->error = "Invalid rebuild index given";
 				return -EINVAL;
 			}
@@ -608,6 +662,9 @@ static int parse_raid_params(struct raid_set *rs, char **argv,
 	}
 	rs->md.dev_sectors = sectors_per_dev;
 
+	if (validate_rebuild_devices(rs))
+		return -EINVAL;
+
 	/* Assume there are no metadata devices until the drives are parsed */
 	rs->md.persistent = 0;
 	rs->md.external = 1;
@@ -960,6 +1017,19 @@ static int analyse_superblocks(struct dm_target *ti, struct raid_set *rs)
 
 	freshest = NULL;
 	rdev_for_each_safe(rdev, tmp, mddev) {
+		/*
+		 * Skipping super_load due to DMPF_SYNC will cause
+		 * the array to undergo initialization again as
+		 * though it were new.  This is the intended effect
+		 * of the "sync" directive.
+		 *
+		 * When reshaping capability is added, we must ensure
+		 * that the "sync" directive is disallowed during the
+		 * reshape.
+		 */
+		if (rs->print_flags & DMPF_SYNC)
+			continue;
+
 		if (!rdev->meta_bdev)
 			continue;
 
@@ -1360,7 +1430,7 @@ static void raid_resume(struct dm_target *ti)
 
 static struct target_type raid_target = {
 	.name = "raid",
-	.version = {1, 3, 0},
+	.version = {1, 3, 1},
 	.module = THIS_MODULE,
 	.ctr = raid_ctr,
 	.dtr = raid_dtr,

drivers/md/linear.c

@@ -138,6 +138,7 @@ static struct linear_conf *linear_conf(struct mddev *mddev, int raid_disks)
 	struct linear_conf *conf;
 	struct md_rdev *rdev;
 	int i, cnt;
+	bool discard_supported = false;
 
 	conf = kzalloc (sizeof (*conf) + raid_disks*sizeof(struct dev_info),
 			GFP_KERNEL);
@@ -171,6 +172,8 @@ static struct linear_conf *linear_conf(struct mddev *mddev, int raid_disks)
 		conf->array_sectors += rdev->sectors;
 		cnt++;
 
+		if (blk_queue_discard(bdev_get_queue(rdev->bdev)))
+			discard_supported = true;
 	}
 	if (cnt != raid_disks) {
 		printk(KERN_ERR "md/linear:%s: not enough drives present. Aborting!\n",
@@ -178,6 +181,11 @@ static struct linear_conf *linear_conf(struct mddev *mddev, int raid_disks)
 		goto out;
 	}
 
+	if (!discard_supported)
+		queue_flag_clear_unlocked(QUEUE_FLAG_DISCARD, mddev->queue);
+	else
+		queue_flag_set_unlocked(QUEUE_FLAG_DISCARD, mddev->queue);
+
 	/*
 	 * Here we calculate the device offsets.
 	 */
@@ -244,7 +252,9 @@ static int linear_add(struct mddev *mddev, struct md_rdev *rdev)
 	if (!newconf)
 		return -ENOMEM;
 
-	oldconf = rcu_dereference(mddev->private);
+	oldconf = rcu_dereference_protected(mddev->private,
+					    lockdep_is_held(
+						    &mddev->reconfig_mutex));
 	mddev->raid_disks++;
 	rcu_assign_pointer(mddev->private, newconf);
 	md_set_array_sectors(mddev, linear_size(mddev, 0, 0));
@@ -256,7 +266,10 @@ static int linear_add(struct mddev *mddev, struct md_rdev *rdev)
 
 static int linear_stop (struct mddev *mddev)
 {
-	struct linear_conf *conf = mddev->private;
+	struct linear_conf *conf =
+		rcu_dereference_protected(mddev->private,
+					  lockdep_is_held(
+						  &mddev->reconfig_mutex));
 
 	/*
 	 * We do not require rcu protection here since
@@ -326,6 +339,14 @@ static void linear_make_request(struct mddev *mddev, struct bio *bio)
 	bio->bi_sector = bio->bi_sector - start_sector
 		+ tmp_dev->rdev->data_offset;
 	rcu_read_unlock();
+
+	if (unlikely((bio->bi_rw & REQ_DISCARD) &&
+		     !blk_queue_discard(bdev_get_queue(bio->bi_bdev)))) {
+		/* Just ignore it */
+		bio_endio(bio, 0);
+		return;
+	}
+
 	generic_make_request(bio);
 }
 

drivers/md/md.c

@@ -674,7 +674,18 @@ static struct md_rdev * find_rdev_nr(struct mddev *mddev, int nr)
 	return NULL;
 }
 
-static struct md_rdev * find_rdev(struct mddev * mddev, dev_t dev)
+static struct md_rdev *find_rdev_nr_rcu(struct mddev *mddev, int nr)
+{
+	struct md_rdev *rdev;
+
+	rdev_for_each_rcu(rdev, mddev)
+		if (rdev->desc_nr == nr)
+			return rdev;
+
+	return NULL;
+}
+
+static struct md_rdev *find_rdev(struct mddev *mddev, dev_t dev)
 {
 	struct md_rdev *rdev;
 
@@ -685,6 +696,17 @@ static struct md_rdev * find_rdev(struct mddev * mddev, dev_t dev)
 	return NULL;
 }
 
+static struct md_rdev *find_rdev_rcu(struct mddev *mddev, dev_t dev)
+{
+	struct md_rdev *rdev;
+
+	rdev_for_each_rcu(rdev, mddev)
+		if (rdev->bdev->bd_dev == dev)
+			return rdev;
+
+	return NULL;
+}
+
 static struct md_personality *find_pers(int level, char *clevel)
 {
 	struct md_personality *pers;
@@ -2022,8 +2044,14 @@ EXPORT_SYMBOL(md_integrity_register);
 /* Disable data integrity if non-capable/non-matching disk is being added */
 void md_integrity_add_rdev(struct md_rdev *rdev, struct mddev *mddev)
 {
-	struct blk_integrity *bi_rdev = bdev_get_integrity(rdev->bdev);
-	struct blk_integrity *bi_mddev = blk_get_integrity(mddev->gendisk);
+	struct blk_integrity *bi_rdev;
+	struct blk_integrity *bi_mddev;
+
+	if (!mddev->gendisk)
+		return;
+
+	bi_rdev = bdev_get_integrity(rdev->bdev);
+	bi_mddev = blk_get_integrity(mddev->gendisk);
 
 	if (!bi_mddev) /* nothing to do */
 		return;
@@ -3754,6 +3782,8 @@ resync_start_store(struct mddev *mddev, const char *buf, size_t len)
 		return -EINVAL;
 
 	mddev->recovery_cp = n;
+	if (mddev->pers)
+		set_bit(MD_CHANGE_CLEAN, &mddev->flags);
 	return len;
 }
 static struct md_sysfs_entry md_resync_start =
@@ -4231,6 +4261,13 @@ action_store(struct mddev *mddev, const char *page, size_t len)
 		set_bit(MD_RECOVERY_REQUESTED, &mddev->recovery);
 		set_bit(MD_RECOVERY_SYNC, &mddev->recovery);
 	}
+	if (mddev->ro == 2) {
+		/* A write to sync_action is enough to justify
+		 * canceling read-auto mode
+		 */
+		mddev->ro = 0;
+		md_wakeup_thread(mddev->sync_thread);
+	}
 	set_bit(MD_RECOVERY_NEEDED, &mddev->recovery);
 	md_wakeup_thread(mddev->thread);
 	sysfs_notify_dirent_safe(mddev->sysfs_action);
@@ -4241,7 +4278,8 @@ static ssize_t
 mismatch_cnt_show(struct mddev *mddev, char *page)
 {
 	return sprintf(page, "%llu\n",
-		       (unsigned long long) mddev->resync_mismatches);
+		       (unsigned long long)
+		       atomic64_read(&mddev->resync_mismatches));
 }
 
 static struct md_sysfs_entry md_scan_mode =
@@ -4362,6 +4400,10 @@ sync_completed_show(struct mddev *mddev, char *page)
 	if (!test_bit(MD_RECOVERY_RUNNING, &mddev->recovery))
 		return sprintf(page, "none\n");
 
+	if (mddev->curr_resync == 1 ||
+	    mddev->curr_resync == 2)
+		return sprintf(page, "delayed\n");
+
 	if (test_bit(MD_RECOVERY_SYNC, &mddev->recovery) ||
 	    test_bit(MD_RECOVERY_RESHAPE, &mddev->recovery))
 		max_sectors = mddev->resync_max_sectors;
@@ -5207,7 +5249,7 @@ static void md_clean(struct mddev *mddev)
 	mddev->new_layout = 0;
 	mddev->new_chunk_sectors = 0;
 	mddev->curr_resync = 0;
-	mddev->resync_mismatches = 0;
+	atomic64_set(&mddev->resync_mismatches, 0);
 	mddev->suspend_lo = mddev->suspend_hi = 0;
 	mddev->sync_speed_min = mddev->sync_speed_max = 0;
 	mddev->recovery = 0;
@@ -5509,8 +5551,9 @@ static int get_array_info(struct mddev * mddev, void __user * arg)
 	int nr,working,insync,failed,spare;
 	struct md_rdev *rdev;
 
-	nr=working=insync=failed=spare=0;
-	rdev_for_each(rdev, mddev) {
+	nr = working = insync = failed = spare = 0;
+	rcu_read_lock();
+	rdev_for_each_rcu(rdev, mddev) {
 		nr++;
 		if (test_bit(Faulty, &rdev->flags))
 			failed++;
@@ -5522,6 +5565,7 @@ static int get_array_info(struct mddev * mddev, void __user * arg)
 				spare++;
 		}
 	}
+	rcu_read_unlock();
 
 	info.major_version = mddev->major_version;
 	info.minor_version = mddev->minor_version;
@@ -5605,7 +5649,8 @@ static int get_disk_info(struct mddev * mddev, void __user * arg)
 	if (copy_from_user(&info, arg, sizeof(info)))
 		return -EFAULT;
 
-	rdev = find_rdev_nr(mddev, info.number);
+	rcu_read_lock();
+	rdev = find_rdev_nr_rcu(mddev, info.number);
 	if (rdev) {
 		info.major = MAJOR(rdev->bdev->bd_dev);
 		info.minor = MINOR(rdev->bdev->bd_dev);
@@ -5624,6 +5669,7 @@ static int get_disk_info(struct mddev * mddev, void __user * arg)
 		info.raid_disk = -1;
 		info.state = (1<<MD_DISK_REMOVED);
 	}
+	rcu_read_unlock();
 
 	if (copy_to_user(arg, &info, sizeof(info)))
 		return -EFAULT;
@@ -6232,18 +6278,22 @@ static int update_array_info(struct mddev *mddev, mdu_array_info_t *info)
 static int set_disk_faulty(struct mddev *mddev, dev_t dev)
 {
 	struct md_rdev *rdev;
+	int err = 0;
 
 	if (mddev->pers == NULL)
 		return -ENODEV;
 
-	rdev = find_rdev(mddev, dev);
+	rcu_read_lock();
+	rdev = find_rdev_rcu(mddev, dev);
 	if (!rdev)
-		return -ENODEV;
-
-	md_error(mddev, rdev);
-	if (!test_bit(Faulty, &rdev->flags))
-		return -EBUSY;
-	return 0;
+		err =  -ENODEV;
+	else {
+		md_error(mddev, rdev);
+		if (!test_bit(Faulty, &rdev->flags))
+			err = -EBUSY;
+	}
+	rcu_read_unlock();
+	return err;
 }
 
 /*
@@ -6315,6 +6365,27 @@ static int md_ioctl(struct block_device *bdev, fmode_t mode,
 		goto abort;
 	}
 
+	/* Some actions do not requires the mutex */
+	switch (cmd) {
+	case GET_ARRAY_INFO:
+		if (!mddev->raid_disks && !mddev->external)
+			err = -ENODEV;
+		else
+			err = get_array_info(mddev, argp);
+		goto abort;
+
+	case GET_DISK_INFO:
+		if (!mddev->raid_disks && !mddev->external)
+			err = -ENODEV;
+		else
+			err = get_disk_info(mddev, argp);
+		goto abort;
+
+	case SET_DISK_FAULTY:
+		err = set_disk_faulty(mddev, new_decode_dev(arg));
+		goto abort;
+	}
+
 	err = mddev_lock(mddev);
 	if (err) {
 		printk(KERN_INFO 
@@ -6387,18 +6458,10 @@ static int md_ioctl(struct block_device *bdev, fmode_t mode,
 	 */
 	switch (cmd)
 	{
-		case GET_ARRAY_INFO:
-			err = get_array_info(mddev, argp);
-			goto done_unlock;
-
 		case GET_BITMAP_FILE:
 			err = get_bitmap_file(mddev, argp);
 			goto done_unlock;
 
-		case GET_DISK_INFO:
-			err = get_disk_info(mddev, argp);
-			goto done_unlock;
-
 		case RESTART_ARRAY_RW:
 			err = restart_array(mddev);
 			goto done_unlock;
@@ -6480,10 +6543,6 @@ static int md_ioctl(struct block_device *bdev, fmode_t mode,
 			err = hot_add_disk(mddev, new_decode_dev(arg));
 			goto done_unlock;
 
-		case SET_DISK_FAULTY:
-			err = set_disk_faulty(mddev, new_decode_dev(arg));
-			goto done_unlock;
-
 		case RUN_ARRAY:
 			err = do_md_run(mddev);
 			goto done_unlock;
@@ -6641,7 +6700,7 @@ static int md_thread(void * arg)
 
 		clear_bit(THREAD_WAKEUP, &thread->flags);
 		if (!kthread_should_stop())
-			thread->run(thread->mddev);
+			thread->run(thread);
 	}
 
 	return 0;
@@ -6656,8 +6715,8 @@ void md_wakeup_thread(struct md_thread *thread)
 	}
 }
 
-struct md_thread *md_register_thread(void (*run) (struct mddev *), struct mddev *mddev,
-				 const char *name)
+struct md_thread *md_register_thread(void (*run) (struct md_thread *),
+		struct mddev *mddev, const char *name)
 {
 	struct md_thread *thread;
 
@@ -6752,7 +6811,11 @@ static void status_resync(struct seq_file *seq, struct mddev * mddev)
 	int scale;
 	unsigned int per_milli;
 
-	resync = mddev->curr_resync - atomic_read(&mddev->recovery_active);
+	if (mddev->curr_resync <= 3)
+		resync = 0;
+	else
+		resync = mddev->curr_resync
+			- atomic_read(&mddev->recovery_active);
 
 	if (test_bit(MD_RECOVERY_SYNC, &mddev->recovery) ||
 	    test_bit(MD_RECOVERY_RESHAPE, &mddev->recovery))
@@ -6978,7 +7041,7 @@ static int md_seq_show(struct seq_file *seq, void *v)
 				if (mddev->curr_resync > 2) {
 					status_resync(seq, mddev);
 					seq_printf(seq, "\n      ");
-				} else if (mddev->curr_resync == 1 || mddev->curr_resync == 2)
+				} else if (mddev->curr_resync >= 1)
 					seq_printf(seq, "\tresync=DELAYED\n      ");
 				else if (mddev->recovery_cp < MaxSector)
 					seq_printf(seq, "\tresync=PENDING\n      ");
@@ -7206,8 +7269,9 @@ EXPORT_SYMBOL_GPL(md_allow_write);
 
 #define SYNC_MARKS	10
 #define	SYNC_MARK_STEP	(3*HZ)
-void md_do_sync(struct mddev *mddev)
+void md_do_sync(struct md_thread *thread)
 {
+	struct mddev *mddev = thread->mddev;
 	struct mddev *mddev2;
 	unsigned int currspeed = 0,
 		 window;
@@ -7311,7 +7375,7 @@ void md_do_sync(struct mddev *mddev)
 		 * which defaults to physical size, but can be virtual size
 		 */
 		max_sectors = mddev->resync_max_sectors;
-		mddev->resync_mismatches = 0;
+		atomic64_set(&mddev->resync_mismatches, 0);
 		/* we don't use the checkpoint if there's a bitmap */
 		if (test_bit(MD_RECOVERY_REQUESTED, &mddev->recovery))
 			j = mddev->resync_min;
@@ -7367,8 +7431,11 @@ void md_do_sync(struct mddev *mddev)
 		       "md: resuming %s of %s from checkpoint.\n",
 		       desc, mdname(mddev));
 		mddev->curr_resync = j;
-	}
+	} else
+		mddev->curr_resync = 3; /* no longer delayed */
 	mddev->curr_resync_completed = j;
+	sysfs_notify(&mddev->kobj, NULL, "sync_completed");
+	md_new_event(mddev);
 
 	blk_start_plug(&plug);
 	while (j < max_sectors) {
@@ -7421,7 +7488,8 @@ void md_do_sync(struct mddev *mddev)
 			break;
 
 		j += sectors;
-		if (j>1) mddev->curr_resync = j;
+		if (j > 2)
+			mddev->curr_resync = j;
 		mddev->curr_mark_cnt = io_sectors;
 		if (last_check == 0)
 			/* this is the earliest that rebuild will be
@@ -7543,8 +7611,6 @@ static int remove_and_add_spares(struct mddev *mddev)
 	int spares = 0;
 	int removed = 0;
 
-	mddev->curr_resync_completed = 0;
-
 	rdev_for_each(rdev, mddev)
 		if (rdev->raid_disk >= 0 &&
 		    !test_bit(Blocked, &rdev->flags) &&
@@ -7739,6 +7805,7 @@ void md_check_recovery(struct mddev *mddev)
 		/* Set RUNNING before clearing NEEDED to avoid
 		 * any transients in the value of "sync_action".
 		 */
+		mddev->curr_resync_completed = 0;
 		set_bit(MD_RECOVERY_RUNNING, &mddev->recovery);
 		/* Clear some bits that don't mean anything, but
 		 * might be left set
@@ -7752,7 +7819,7 @@ void md_check_recovery(struct mddev *mddev)
 		/* no recovery is running.
 		 * remove any failed drives, then
 		 * add spares if possible.
-		 * Spare are also removed and re-added, to allow
+		 * Spares are also removed and re-added, to allow
 		 * the personality to fail the re-add.
 		 */
 

drivers/md/md.h

@@ -282,7 +282,7 @@ struct mddev {
 
 	sector_t			resync_max_sectors; /* may be set by personality */
 
-	sector_t			resync_mismatches; /* count of sectors where
+	atomic64_t			resync_mismatches; /* count of sectors where
 							    * parity/replica mismatch found
 							    */
 
@@ -540,12 +540,13 @@ static inline void sysfs_unlink_rdev(struct mddev *mddev, struct md_rdev *rdev)
 	list_for_each_entry_rcu(rdev, &((mddev)->disks), same_set)
 
 struct md_thread {
-	void			(*run) (struct mddev *mddev);
+	void			(*run) (struct md_thread *thread);
 	struct mddev		*mddev;
 	wait_queue_head_t	wqueue;
 	unsigned long           flags;
 	struct task_struct	*tsk;
 	unsigned long		timeout;
+	void			*private;
 };
 
 #define THREAD_WAKEUP  0
@@ -584,7 +585,7 @@ static inline void safe_put_page(struct page *p)
 extern int register_md_personality(struct md_personality *p);
 extern int unregister_md_personality(struct md_personality *p);
 extern struct md_thread *md_register_thread(
-	void (*run)(struct mddev *mddev),
+	void (*run)(struct md_thread *thread),
 	struct mddev *mddev,
 	const char *name);
 extern void md_unregister_thread(struct md_thread **threadp);
@@ -603,7 +604,7 @@ extern void md_super_write(struct mddev *mddev, struct md_rdev *rdev,
 extern void md_super_wait(struct mddev *mddev);
 extern int sync_page_io(struct md_rdev *rdev, sector_t sector, int size, 
 			struct page *page, int rw, bool metadata_op);
-extern void md_do_sync(struct mddev *mddev);
+extern void md_do_sync(struct md_thread *thread);
 extern void md_new_event(struct mddev *mddev);
 extern int md_allow_write(struct mddev *mddev);
 extern void md_wait_for_blocked_rdev(struct md_rdev *rdev, struct mddev *mddev);

drivers/md/multipath.c

@@ -335,8 +335,9 @@ static int multipath_remove_disk(struct mddev *mddev, struct md_rdev *rdev)
  *	3.	Performs writes following reads for array syncronising.
  */
 
-static void multipathd (struct mddev *mddev)
+static void multipathd(struct md_thread *thread)
 {
+	struct mddev *mddev = thread->mddev;
 	struct multipath_bh *mp_bh;
 	struct bio *bio;
 	unsigned long flags;

drivers/md/raid0.c

@@ -88,6 +88,7 @@ static int create_strip_zones(struct mddev *mddev, struct r0conf **private_conf)
 	char b[BDEVNAME_SIZE];
 	char b2[BDEVNAME_SIZE];
 	struct r0conf *conf = kzalloc(sizeof(*conf), GFP_KERNEL);
+	bool discard_supported = false;
 
 	if (!conf)
 		return -ENOMEM;
@@ -195,6 +196,9 @@ static int create_strip_zones(struct mddev *mddev, struct r0conf **private_conf)
 		if (!smallest || (rdev1->sectors < smallest->sectors))
 			smallest = rdev1;
 		cnt++;
+
+		if (blk_queue_discard(bdev_get_queue(rdev1->bdev)))
+			discard_supported = true;
 	}
 	if (cnt != mddev->raid_disks) {
 		printk(KERN_ERR "md/raid0:%s: too few disks (%d of %d) - "
@@ -272,6 +276,11 @@ static int create_strip_zones(struct mddev *mddev, struct r0conf **private_conf)
 	blk_queue_io_opt(mddev->queue,
 			 (mddev->chunk_sectors << 9) * mddev->raid_disks);
 
+	if (!discard_supported)
+		queue_flag_clear_unlocked(QUEUE_FLAG_DISCARD, mddev->queue);
+	else
+		queue_flag_set_unlocked(QUEUE_FLAG_DISCARD, mddev->queue);
+
 	pr_debug("md/raid0:%s: done.\n", mdname(mddev));
 	*private_conf = conf;
 
@@ -423,6 +432,7 @@ static int raid0_run(struct mddev *mddev)
 		return -EINVAL;
 	blk_queue_max_hw_sectors(mddev->queue, mddev->chunk_sectors);
 	blk_queue_max_write_same_sectors(mddev->queue, mddev->chunk_sectors);
+	blk_queue_max_discard_sectors(mddev->queue, mddev->chunk_sectors);
 
 	/* if private is not null, we are here after takeover */
 	if (mddev->private == NULL) {
@@ -510,7 +520,7 @@ static void raid0_make_request(struct mddev *mddev, struct bio *bio)
 		sector_t sector = bio->bi_sector;
 		struct bio_pair *bp;
 		/* Sanity check -- queue functions should prevent this happening */
-		if (bio->bi_vcnt != 1 ||
+		if ((bio->bi_vcnt != 1 && bio->bi_vcnt != 0) ||
 		    bio->bi_idx != 0)
 			goto bad_map;
 		/* This is a one page bio that upper layers
@@ -536,6 +546,13 @@ static void raid0_make_request(struct mddev *mddev, struct bio *bio)
 	bio->bi_sector = sector_offset + zone->dev_start +
 		tmp_dev->data_offset;
 
+	if (unlikely((bio->bi_rw & REQ_DISCARD) &&
+		     !blk_queue_discard(bdev_get_queue(bio->bi_bdev)))) {
+		/* Just ignore it */
+		bio_endio(bio, 0);
+		return;
+	}
+
 	generic_make_request(bio);
 	return;
 

drivers/md/raid1.c

@@ -333,9 +333,10 @@ static void raid1_end_read_request(struct bio *bio, int error)
 		spin_unlock_irqrestore(&conf->device_lock, flags);
 	}
 
-	if (uptodate)
+	if (uptodate) {
 		raid_end_bio_io(r1_bio);
-	else {
+		rdev_dec_pending(conf->mirrors[mirror].rdev, conf->mddev);
+	} else {
 		/*
 		 * oops, read error:
 		 */
@@ -349,9 +350,8 @@ static void raid1_end_read_request(struct bio *bio, int error)
 			(unsigned long long)r1_bio->sector);
 		set_bit(R1BIO_ReadError, &r1_bio->state);
 		reschedule_retry(r1_bio);
+		/* don't drop the reference on read_disk yet */
 	}
-
-	rdev_dec_pending(conf->mirrors[mirror].rdev, conf->mddev);
 }
 
 static void close_write(struct r1bio *r1_bio)
@@ -781,7 +781,12 @@ static void flush_pending_writes(struct r1conf *conf)
 		while (bio) { /* submit pending writes */
 			struct bio *next = bio->bi_next;
 			bio->bi_next = NULL;
-			generic_make_request(bio);
+			if (unlikely((bio->bi_rw & REQ_DISCARD) &&
+			    !blk_queue_discard(bdev_get_queue(bio->bi_bdev))))
+				/* Just ignore it */
+				bio_endio(bio, 0);
+			else
+				generic_make_request(bio);
 			bio = next;
 		}
 	} else
@@ -994,6 +999,8 @@ static void make_request(struct mddev *mddev, struct bio * bio)
 	const int rw = bio_data_dir(bio);
 	const unsigned long do_sync = (bio->bi_rw & REQ_SYNC);
 	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));
 	struct md_rdev *blocked_rdev;
 	struct blk_plug_cb *cb;
 	struct raid1_plug_cb *plug = NULL;
@@ -1295,7 +1302,7 @@ static void make_request(struct mddev *mddev, struct bio * bio)
 				   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;
+		mbio->bi_rw = WRITE | do_flush_fua | do_sync | do_discard;
 		mbio->bi_private = r1_bio;
 
 		atomic_inc(&r1_bio->remaining);
@@ -1549,6 +1556,8 @@ static int raid1_add_disk(struct mddev *mddev, struct md_rdev *rdev)
 		clear_bit(Unmerged, &rdev->flags);
 	}
 	md_integrity_add_rdev(rdev, mddev);
+	if (blk_queue_discard(bdev_get_queue(rdev->bdev)))
+		queue_flag_set_unlocked(QUEUE_FLAG_DISCARD, mddev->queue);
 	print_conf(conf);
 	return err;
 }
@@ -1867,7 +1876,7 @@ static int process_checks(struct r1bio *r1_bio)
 		} else
 			j = 0;
 		if (j >= 0)
-			mddev->resync_mismatches += r1_bio->sectors;
+			atomic64_add(r1_bio->sectors, &mddev->resync_mismatches);
 		if (j < 0 || (test_bit(MD_RECOVERY_CHECK, &mddev->recovery)
 			      && test_bit(BIO_UPTODATE, &sbio->bi_flags))) {
 			/* No need to write to this device. */
@@ -2220,6 +2229,7 @@ static void handle_read_error(struct r1conf *conf, struct r1bio *r1_bio)
 		unfreeze_array(conf);
 	} else
 		md_error(mddev, conf->mirrors[r1_bio->read_disk].rdev);
+	rdev_dec_pending(conf->mirrors[r1_bio->read_disk].rdev, conf->mddev);
 
 	bio = r1_bio->bios[r1_bio->read_disk];
 	bdevname(bio->bi_bdev, b);
@@ -2285,8 +2295,9 @@ static void handle_read_error(struct r1conf *conf, struct r1bio *r1_bio)
 	}
 }
 
-static void raid1d(struct mddev *mddev)
+static void raid1d(struct md_thread *thread)
 {
+	struct mddev *mddev = thread->mddev;
 	struct r1bio *r1_bio;
 	unsigned long flags;
 	struct r1conf *conf = mddev->private;
@@ -2783,6 +2794,7 @@ static int run(struct mddev *mddev)
 	int i;
 	struct md_rdev *rdev;
 	int ret;
+	bool discard_supported = false;
 
 	if (mddev->level != 1) {
 		printk(KERN_ERR "md/raid1:%s: raid level not set to mirroring (%d)\n",
@@ -2812,6 +2824,8 @@ static int run(struct mddev *mddev)
 			continue;
 		disk_stack_limits(mddev->gendisk, rdev->bdev,
 				  rdev->data_offset << 9);
+		if (blk_queue_discard(bdev_get_queue(rdev->bdev)))
+			discard_supported = true;
 	}
 
 	mddev->degraded = 0;
@@ -2846,6 +2860,13 @@ static int run(struct mddev *mddev)
 		mddev->queue->backing_dev_info.congested_fn = raid1_congested;
 		mddev->queue->backing_dev_info.congested_data = mddev;
 		blk_queue_merge_bvec(mddev->queue, raid1_mergeable_bvec);
+
+		if (discard_supported)
+			queue_flag_set_unlocked(QUEUE_FLAG_DISCARD,
+						mddev->queue);
+		else
+			queue_flag_clear_unlocked(QUEUE_FLAG_DISCARD,
+						  mddev->queue);
 	}
 
 	ret =  md_integrity_register(mddev);

drivers/md/raid10.c

@@ -911,7 +911,12 @@ static void flush_pending_writes(struct r10conf *conf)
 		while (bio) { /* submit pending writes */
 			struct bio *next = bio->bi_next;
 			bio->bi_next = NULL;
-			generic_make_request(bio);
+			if (unlikely((bio->bi_rw & REQ_DISCARD) &&
+			    !blk_queue_discard(bdev_get_queue(bio->bi_bdev))))
+				/* Just ignore it */
+				bio_endio(bio, 0);
+			else
+				generic_make_request(bio);
 			bio = next;
 		}
 	} else
@@ -1050,6 +1055,44 @@ static sector_t choose_data_offset(struct r10bio *r10_bio,
 		return rdev->new_data_offset;
 }
 
+struct raid10_plug_cb {
+	struct blk_plug_cb	cb;
+	struct bio_list		pending;
+	int			pending_cnt;
+};
+
+static void raid10_unplug(struct blk_plug_cb *cb, bool from_schedule)
+{
+	struct raid10_plug_cb *plug = container_of(cb, struct raid10_plug_cb,
+						   cb);
+	struct mddev *mddev = plug->cb.data;
+	struct r10conf *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 r10conf *conf = mddev->private;
@@ -1061,8 +1104,12 @@ static void make_request(struct mddev *mddev, struct bio * bio)
 	const int rw = bio_data_dir(bio);
 	const unsigned long do_sync = (bio->bi_rw & REQ_SYNC);
 	const unsigned long do_fua = (bio->bi_rw & REQ_FUA);
+	const unsigned long do_discard = (bio->bi_rw
+					  & (REQ_DISCARD | REQ_SECURE));
 	unsigned long flags;
 	struct md_rdev *blocked_rdev;
+	struct blk_plug_cb *cb;
+	struct raid10_plug_cb *plug = NULL;
 	int sectors_handled;
 	int max_sectors;
 	int sectors;
@@ -1081,7 +1128,7 @@ static void make_request(struct mddev *mddev, struct bio * bio)
 			 || conf->prev.near_copies < conf->prev.raid_disks))) {
 		struct bio_pair *bp;
 		/* Sanity check -- queue functions should prevent this happening */
-		if (bio->bi_vcnt != 1 ||
+		if ((bio->bi_vcnt != 1 && bio->bi_vcnt != 0) ||
 		    bio->bi_idx != 0)
 			goto bad_map;
 		/* This is a one page bio that upper layers
@@ -1410,15 +1457,26 @@ static void make_request(struct mddev *mddev, struct bio * bio)
 						      conf->mirrors[d].rdev));
 		mbio->bi_bdev = conf->mirrors[d].rdev->bdev;
 		mbio->bi_end_io	= raid10_end_write_request;
-		mbio->bi_rw = WRITE | do_sync | do_fua;
+		mbio->bi_rw = WRITE | do_sync | do_fua | do_discard;
 		mbio->bi_private = r10_bio;
 
 		atomic_inc(&r10_bio->remaining);
+
+		cb = blk_check_plugged(raid10_unplug, mddev, sizeof(*plug));
+		if (cb)
+			plug = container_of(cb, struct raid10_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);
 
 		if (!r10_bio->devs[i].repl_bio)
@@ -1439,7 +1497,7 @@ static void make_request(struct mddev *mddev, struct bio * bio)
 					   conf->mirrors[d].replacement));
 		mbio->bi_bdev = conf->mirrors[d].replacement->bdev;
 		mbio->bi_end_io	= raid10_end_write_request;
-		mbio->bi_rw = WRITE | do_sync | do_fua;
+		mbio->bi_rw = WRITE | do_sync | do_fua | do_discard;
 		mbio->bi_private = r10_bio;
 
 		atomic_inc(&r10_bio->remaining);
@@ -1638,7 +1696,7 @@ static int raid10_spare_active(struct mddev *mddev)
 			   && !test_bit(Faulty, &tmp->rdev->flags)
 			   && !test_and_set_bit(In_sync, &tmp->rdev->flags)) {
 			count++;
-			sysfs_notify_dirent(tmp->rdev->sysfs_state);
+			sysfs_notify_dirent_safe(tmp->rdev->sysfs_state);
 		}
 	}
 	spin_lock_irqsave(&conf->device_lock, flags);
@@ -1725,6 +1783,9 @@ static int raid10_add_disk(struct mddev *mddev, struct md_rdev *rdev)
 		clear_bit(Unmerged, &rdev->flags);
 	}
 	md_integrity_add_rdev(rdev, mddev);
+	if (blk_queue_discard(bdev_get_queue(rdev->bdev)))
+		queue_flag_set_unlocked(QUEUE_FLAG_DISCARD, mddev->queue);
+
 	print_conf(conf);
 	return err;
 }
@@ -1952,7 +2013,7 @@ static void sync_request_write(struct mddev *mddev, struct r10bio *r10_bio)
 					break;
 			if (j == vcnt)
 				continue;
-			mddev->resync_mismatches += r10_bio->sectors;
+			atomic64_add(r10_bio->sectors, &mddev->resync_mismatches);
 			if (test_bit(MD_RECOVERY_CHECK, &mddev->recovery))
 				/* Don't fix anything. */
 				continue;
@@ -2673,8 +2734,9 @@ static void handle_write_completed(struct r10conf *conf, struct r10bio *r10_bio)
 	}
 }
 
-static void raid10d(struct mddev *mddev)
+static void raid10d(struct md_thread *thread)
 {
+	struct mddev *mddev = thread->mddev;
 	struct r10bio *r10_bio;
 	unsigned long flags;
 	struct r10conf *conf = mddev->private;
@@ -3158,7 +3220,7 @@ static sector_t sync_request(struct mddev *mddev, sector_t sector_nr,
 				else {
 					bad_sectors -= (sector - first_bad);
 					if (max_sync > bad_sectors)
-						max_sync = max_sync;
+						max_sync = bad_sectors;
 					continue;
 				}
 			}
@@ -3482,6 +3544,7 @@ static int run(struct mddev *mddev)
 	sector_t size;
 	sector_t min_offset_diff = 0;
 	int first = 1;
+	bool discard_supported = false;
 
 	if (mddev->private == NULL) {
 		conf = setup_conf(mddev);
@@ -3498,6 +3561,8 @@ static int run(struct mddev *mddev)
 
 	chunk_size = mddev->chunk_sectors << 9;
 	if (mddev->queue) {
+		blk_queue_max_discard_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);
@@ -3543,8 +3608,16 @@ static int run(struct mddev *mddev)
 					  rdev->data_offset << 9);
 
 		disk->head_position = 0;
+
+		if (blk_queue_discard(bdev_get_queue(rdev->bdev)))
+			discard_supported = true;
 	}
 
+	if (discard_supported)
+		queue_flag_set_unlocked(QUEUE_FLAG_DISCARD, mddev->queue);
+	else
+		queue_flag_clear_unlocked(QUEUE_FLAG_DISCARD, mddev->queue);
+
 	/* need to check that every block has at least one working mirror */
 	if (!enough(conf, -1)) {
 		printk(KERN_ERR "md/raid10:%s: not enough operational mirrors.\n",

drivers/md/raid5.c

@@ -551,6 +551,8 @@ static void ops_run_io(struct stripe_head *sh, struct stripe_head_state *s)
 				rw = WRITE_FUA;
 			else
 				rw = WRITE;
+			if (test_bit(R5_Discard, &sh->dev[i].flags))
+				rw |= REQ_DISCARD;
 		} else if (test_and_clear_bit(R5_Wantread, &sh->dev[i].flags))
 			rw = READ;
 		else if (test_and_clear_bit(R5_WantReplace,
@@ -1174,8 +1176,11 @@ ops_run_biodrain(struct stripe_head *sh, struct dma_async_tx_descriptor *tx)
 					set_bit(R5_WantFUA, &dev->flags);
 				if (wbi->bi_rw & REQ_SYNC)
 					set_bit(R5_SyncIO, &dev->flags);
-				tx = async_copy_data(1, wbi, dev->page,
-					dev->sector, tx);
+				if (wbi->bi_rw & REQ_DISCARD)
+					set_bit(R5_Discard, &dev->flags);
+				else
+					tx = async_copy_data(1, wbi, dev->page,
+						dev->sector, tx);
 				wbi = r5_next_bio(wbi, dev->sector);
 			}
 		}
@@ -1191,7 +1196,7 @@ static void ops_complete_reconstruct(void *stripe_head_ref)
 	int pd_idx = sh->pd_idx;
 	int qd_idx = sh->qd_idx;
 	int i;
-	bool fua = false, sync = false;
+	bool fua = false, sync = false, discard = false;
 
 	pr_debug("%s: stripe %llu\n", __func__,
 		(unsigned long long)sh->sector);
@@ -1199,13 +1204,15 @@ static void ops_complete_reconstruct(void *stripe_head_ref)
 	for (i = disks; i--; ) {
 		fua |= test_bit(R5_WantFUA, &sh->dev[i].flags);
 		sync |= test_bit(R5_SyncIO, &sh->dev[i].flags);
+		discard |= test_bit(R5_Discard, &sh->dev[i].flags);
 	}
 
 	for (i = disks; i--; ) {
 		struct r5dev *dev = &sh->dev[i];
 
 		if (dev->written || i == pd_idx || i == qd_idx) {
-			set_bit(R5_UPTODATE, &dev->flags);
+			if (!discard)
+				set_bit(R5_UPTODATE, &dev->flags);
 			if (fua)
 				set_bit(R5_WantFUA, &dev->flags);
 			if (sync)
@@ -1241,6 +1248,18 @@ ops_run_reconstruct5(struct stripe_head *sh, struct raid5_percpu *percpu,
 	pr_debug("%s: stripe %llu\n", __func__,
 		(unsigned long long)sh->sector);
 
+	for (i = 0; i < sh->disks; i++) {
+		if (pd_idx == i)
+			continue;
+		if (!test_bit(R5_Discard, &sh->dev[i].flags))
+			break;
+	}
+	if (i >= sh->disks) {
+		atomic_inc(&sh->count);
+		set_bit(R5_Discard, &sh->dev[pd_idx].flags);
+		ops_complete_reconstruct(sh);
+		return;
+	}
 	/* check if prexor is active which means only process blocks
 	 * that are part of a read-modify-write (written)
 	 */
@@ -1285,10 +1304,24 @@ ops_run_reconstruct6(struct stripe_head *sh, struct raid5_percpu *percpu,
 {
 	struct async_submit_ctl submit;
 	struct page **blocks = percpu->scribble;
-	int count;
+	int count, i;
 
 	pr_debug("%s: stripe %llu\n", __func__, (unsigned long long)sh->sector);
 
+	for (i = 0; i < sh->disks; i++) {
+		if (sh->pd_idx == i || sh->qd_idx == i)
+			continue;
+		if (!test_bit(R5_Discard, &sh->dev[i].flags))
+			break;
+	}
+	if (i >= sh->disks) {
+		atomic_inc(&sh->count);
+		set_bit(R5_Discard, &sh->dev[sh->pd_idx].flags);
+		set_bit(R5_Discard, &sh->dev[sh->qd_idx].flags);
+		ops_complete_reconstruct(sh);
+		return;
+	}
+
 	count = set_syndrome_sources(blocks, sh);
 
 	atomic_inc(&sh->count);
@@ -2408,11 +2441,11 @@ static int add_stripe_bio(struct stripe_head *sh, struct bio *bi, int dd_idx, in
 		if (sector >= sh->dev[dd_idx].sector + STRIPE_SECTORS)
 			set_bit(R5_OVERWRITE, &sh->dev[dd_idx].flags);
 	}
-	spin_unlock_irq(&sh->stripe_lock);
 
 	pr_debug("added bi b#%llu to stripe s#%llu, disk %d.\n",
 		(unsigned long long)(*bip)->bi_sector,
 		(unsigned long long)sh->sector, dd_idx);
+	spin_unlock_irq(&sh->stripe_lock);
 
 	if (conf->mddev->bitmap && firstwrite) {
 		bitmap_startwrite(conf->mddev->bitmap, sh->sector,
@@ -2479,10 +2512,8 @@ handle_failed_stripe(struct r5conf *conf, struct stripe_head *sh,
 		bi = sh->dev[i].towrite;
 		sh->dev[i].towrite = NULL;
 		spin_unlock_irq(&sh->stripe_lock);
-		if (bi) {
-			s->to_write--;
+		if (bi)
 			bitmap_end = 1;
-		}
 
 		if (test_and_clear_bit(R5_Overlap, &sh->dev[i].flags))
 			wake_up(&conf->wait_for_overlap);
@@ -2524,11 +2555,12 @@ handle_failed_stripe(struct r5conf *conf, struct stripe_head *sh,
 		if (!test_bit(R5_Wantfill, &sh->dev[i].flags) &&
 		    (!test_bit(R5_Insync, &sh->dev[i].flags) ||
 		      test_bit(R5_ReadError, &sh->dev[i].flags))) {
+			spin_lock_irq(&sh->stripe_lock);
 			bi = sh->dev[i].toread;
 			sh->dev[i].toread = NULL;
+			spin_unlock_irq(&sh->stripe_lock);
 			if (test_and_clear_bit(R5_Overlap, &sh->dev[i].flags))
 				wake_up(&conf->wait_for_overlap);
-			if (bi) s->to_read--;
 			while (bi && bi->bi_sector <
 			       sh->dev[i].sector + STRIPE_SECTORS) {
 				struct bio *nextbi =
@@ -2741,7 +2773,8 @@ static void handle_stripe_clean_event(struct r5conf *conf,
 		if (sh->dev[i].written) {
 			dev = &sh->dev[i];
 			if (!test_bit(R5_LOCKED, &dev->flags) &&
-				test_bit(R5_UPTODATE, &dev->flags)) {
+			    (test_bit(R5_UPTODATE, &dev->flags) ||
+			     test_and_clear_bit(R5_Discard, &dev->flags))) {
 				/* We can return any write requests */
 				struct bio *wbi, *wbi2;
 				pr_debug("Return write for disc %d\n", i);
@@ -2775,12 +2808,25 @@ static void handle_stripe_dirtying(struct r5conf *conf,
 				   int disks)
 {
 	int rmw = 0, rcw = 0, i;
-	if (conf->max_degraded == 2) {
-		/* RAID6 requires 'rcw' in current implementation
-		 * Calculate the real rcw later - for now fake it
+	sector_t recovery_cp = conf->mddev->recovery_cp;
+
+	/* RAID6 requires 'rcw' in current implementation.
+	 * Otherwise, check whether resync is now happening or should start.
+	 * If yes, then the array is dirty (after unclean shutdown or
+	 * initial creation), so parity in some stripes might be inconsistent.
+	 * In this case, we need to always do reconstruct-write, to ensure
+	 * that in case of drive failure or read-error correction, we
+	 * generate correct data from the parity.
+	 */
+	if (conf->max_degraded == 2 ||
+	    (recovery_cp < MaxSector && sh->sector >= recovery_cp)) {
+		/* Calculate the real rcw later - for now make it
 		 * look like rcw is cheaper
 		 */
 		rcw = 1; rmw = 2;
+		pr_debug("force RCW max_degraded=%u, recovery_cp=%llu sh->sector=%llu\n",
+			 conf->max_degraded, (unsigned long long)recovery_cp,
+			 (unsigned long long)sh->sector);
 	} else for (i = disks; i--; ) {
 		/* would I have to read this buffer for read_modify_write */
 		struct r5dev *dev = &sh->dev[i];
@@ -2932,7 +2978,7 @@ static void handle_parity_checks5(struct r5conf *conf, struct stripe_head *sh,
 			 */
 			set_bit(STRIPE_INSYNC, &sh->state);
 		else {
-			conf->mddev->resync_mismatches += STRIPE_SECTORS;
+			atomic64_add(STRIPE_SECTORS, &conf->mddev->resync_mismatches);
 			if (test_bit(MD_RECOVERY_CHECK, &conf->mddev->recovery))
 				/* don't try to repair!! */
 				set_bit(STRIPE_INSYNC, &sh->state);
@@ -3084,7 +3130,7 @@ static void handle_parity_checks6(struct r5conf *conf, struct stripe_head *sh,
 				 */
 			}
 		} else {
-			conf->mddev->resync_mismatches += STRIPE_SECTORS;
+			atomic64_add(STRIPE_SECTORS, &conf->mddev->resync_mismatches);
 			if (test_bit(MD_RECOVERY_CHECK, &conf->mddev->recovery))
 				/* don't try to repair!! */
 				set_bit(STRIPE_INSYNC, &sh->state);
@@ -3459,10 +3505,12 @@ static void handle_stripe(struct stripe_head *sh)
 	if (s.written &&
 	    (s.p_failed || ((test_bit(R5_Insync, &pdev->flags)
 			     && !test_bit(R5_LOCKED, &pdev->flags)
-			     && test_bit(R5_UPTODATE, &pdev->flags)))) &&
+			     && (test_bit(R5_UPTODATE, &pdev->flags) ||
+				 test_bit(R5_Discard, &pdev->flags))))) &&
 	    (s.q_failed || ((test_bit(R5_Insync, &qdev->flags)
 			     && !test_bit(R5_LOCKED, &qdev->flags)
-			     && test_bit(R5_UPTODATE, &qdev->flags)))))
+			     && (test_bit(R5_UPTODATE, &qdev->flags) ||
+				 test_bit(R5_Discard, &qdev->flags))))))
 		handle_stripe_clean_event(conf, sh, disks, &s.return_bi);
 
 	/* Now we might consider reading some blocks, either to check/generate
@@ -3489,9 +3537,11 @@ static void handle_stripe(struct stripe_head *sh)
 		/* All the 'written' buffers and the parity block are ready to
 		 * be written back to disk
 		 */
-		BUG_ON(!test_bit(R5_UPTODATE, &sh->dev[sh->pd_idx].flags));
+		BUG_ON(!test_bit(R5_UPTODATE, &sh->dev[sh->pd_idx].flags) &&
+		       !test_bit(R5_Discard, &sh->dev[sh->pd_idx].flags));
 		BUG_ON(sh->qd_idx >= 0 &&
-		       !test_bit(R5_UPTODATE, &sh->dev[sh->qd_idx].flags));
+		       !test_bit(R5_UPTODATE, &sh->dev[sh->qd_idx].flags) &&
+		       !test_bit(R5_Discard, &sh->dev[sh->qd_idx].flags));
 		for (i = disks; i--; ) {
 			struct r5dev *dev = &sh->dev[i];
 			if (test_bit(R5_LOCKED, &dev->flags) &&
@@ -4072,6 +4122,88 @@ static void release_stripe_plug(struct mddev *mddev,
 		release_stripe(sh);
 }
 
+static void make_discard_request(struct mddev *mddev, struct bio *bi)
+{
+	struct r5conf *conf = mddev->private;
+	sector_t logical_sector, last_sector;
+	struct stripe_head *sh;
+	int remaining;
+	int stripe_sectors;
+
+	if (mddev->reshape_position != MaxSector)
+		/* Skip discard while reshape is happening */
+		return;
+
+	logical_sector = bi->bi_sector & ~((sector_t)STRIPE_SECTORS-1);
+	last_sector = bi->bi_sector + (bi->bi_size>>9);
+
+	bi->bi_next = NULL;
+	bi->bi_phys_segments = 1; /* over-loaded to count active stripes */
+
+	stripe_sectors = conf->chunk_sectors *
+		(conf->raid_disks - conf->max_degraded);
+	logical_sector = DIV_ROUND_UP_SECTOR_T(logical_sector,
+					       stripe_sectors);
+	sector_div(last_sector, stripe_sectors);
+
+	logical_sector *= conf->chunk_sectors;
+	last_sector *= conf->chunk_sectors;
+
+	for (; logical_sector < last_sector;
+	     logical_sector += STRIPE_SECTORS) {
+		DEFINE_WAIT(w);
+		int d;
+	again:
+		sh = get_active_stripe(conf, logical_sector, 0, 0, 0);
+		prepare_to_wait(&conf->wait_for_overlap, &w,
+				TASK_UNINTERRUPTIBLE);
+		spin_lock_irq(&sh->stripe_lock);
+		for (d = 0; d < conf->raid_disks; d++) {
+			if (d == sh->pd_idx || d == sh->qd_idx)
+				continue;
+			if (sh->dev[d].towrite || sh->dev[d].toread) {
+				set_bit(R5_Overlap, &sh->dev[d].flags);
+				spin_unlock_irq(&sh->stripe_lock);
+				release_stripe(sh);
+				schedule();
+				goto again;
+			}
+		}
+		finish_wait(&conf->wait_for_overlap, &w);
+		for (d = 0; d < conf->raid_disks; d++) {
+			if (d == sh->pd_idx || d == sh->qd_idx)
+				continue;
+			sh->dev[d].towrite = bi;
+			set_bit(R5_OVERWRITE, &sh->dev[d].flags);
+			raid5_inc_bi_active_stripes(bi);
+		}
+		spin_unlock_irq(&sh->stripe_lock);
+		if (conf->mddev->bitmap) {
+			for (d = 0;
+			     d < conf->raid_disks - conf->max_degraded;
+			     d++)
+				bitmap_startwrite(mddev->bitmap,
+						  sh->sector,
+						  STRIPE_SECTORS,
+						  0);
+			sh->bm_seq = conf->seq_flush + 1;
+			set_bit(STRIPE_BIT_DELAY, &sh->state);
+		}
+
+		set_bit(STRIPE_HANDLE, &sh->state);
+		clear_bit(STRIPE_DELAYED, &sh->state);
+		if (!test_and_set_bit(STRIPE_PREREAD_ACTIVE, &sh->state))
+			atomic_inc(&conf->preread_active_stripes);
+		release_stripe_plug(mddev, sh);
+	}
+
+	remaining = raid5_dec_bi_active_stripes(bi);
+	if (remaining == 0) {
+		md_write_end(mddev);
+		bio_endio(bi, 0);
+	}
+}
+
 static void make_request(struct mddev *mddev, struct bio * bi)
 {
 	struct r5conf *conf = mddev->private;
@@ -4094,6 +4226,11 @@ static void make_request(struct mddev *mddev, struct bio * bi)
 	     chunk_aligned_read(mddev,bi))
 		return;
 
+	if (unlikely(bi->bi_rw & REQ_DISCARD)) {
+		make_discard_request(mddev, bi);
+		return;
+	}
+
 	logical_sector = bi->bi_sector & ~((sector_t)STRIPE_SECTORS-1);
 	last_sector = bi->bi_sector + (bi->bi_size>>9);
 	bi->bi_next = NULL;
@@ -4630,8 +4767,9 @@ static int handle_active_stripes(struct r5conf *conf)
  * During the scan, completed stripes are saved for us by the interrupt
  * handler, so that they will not have to wait for our next wakeup.
  */
-static void raid5d(struct mddev *mddev)
+static void raid5d(struct md_thread *thread)
 {
+	struct mddev *mddev = thread->mddev;
 	struct r5conf *conf = mddev->private;
 	int handled;
 	struct blk_plug plug;
@@ -5366,6 +5504,7 @@ static int run(struct mddev *mddev)
 
 	if (mddev->queue) {
 		int chunk_size;
+		bool discard_supported = true;
 		/* read-ahead size must cover two whole stripes, which
 		 * is 2 * (datadisks) * chunksize where 'n' is the
 		 * number of raid devices
@@ -5385,13 +5524,48 @@ static int run(struct mddev *mddev)
 		blk_queue_io_min(mddev->queue, chunk_size);
 		blk_queue_io_opt(mddev->queue, chunk_size *
 				 (conf->raid_disks - conf->max_degraded));
+		/*
+		 * We can only discard a whole stripe. It doesn't make sense to
+		 * discard data disk but write parity disk
+		 */
+		stripe = stripe * PAGE_SIZE;
+		mddev->queue->limits.discard_alignment = stripe;
+		mddev->queue->limits.discard_granularity = stripe;
+		/*
+		 * unaligned part of discard request will be ignored, so can't
+		 * guarantee discard_zerors_data
+		 */
+		mddev->queue->limits.discard_zeroes_data = 0;
 
 		rdev_for_each(rdev, mddev) {
 			disk_stack_limits(mddev->gendisk, rdev->bdev,
 					  rdev->data_offset << 9);
 			disk_stack_limits(mddev->gendisk, rdev->bdev,
 					  rdev->new_data_offset << 9);
+			/*
+			 * discard_zeroes_data is required, otherwise data
+			 * could be lost. Consider a scenario: discard a stripe
+			 * (the stripe could be inconsistent if
+			 * discard_zeroes_data is 0); write one disk of the
+			 * stripe (the stripe could be inconsistent again
+			 * depending on which disks are used to calculate
+			 * parity); the disk is broken; The stripe data of this
+			 * disk is lost.
+			 */
+			if (!blk_queue_discard(bdev_get_queue(rdev->bdev)) ||
+			    !bdev_get_queue(rdev->bdev)->
+						limits.discard_zeroes_data)
+				discard_supported = false;
 		}
+
+		if (discard_supported &&
+		   mddev->queue->limits.max_discard_sectors >= stripe &&
+		   mddev->queue->limits.discard_granularity >= stripe)
+			queue_flag_set_unlocked(QUEUE_FLAG_DISCARD,
+						mddev->queue);
+		else
+			queue_flag_clear_unlocked(QUEUE_FLAG_DISCARD,
+						mddev->queue);
 	}
 
 	return 0;
@@ -5702,7 +5876,8 @@ static int check_reshape(struct mddev *mddev)
 	if (!check_stripe_cache(mddev))
 		return -ENOSPC;
 
-	return resize_stripes(conf, conf->raid_disks + mddev->delta_disks);
+	return resize_stripes(conf, (conf->previous_raid_disks
+				     + mddev->delta_disks));
 }
 
 static int raid5_start_reshape(struct mddev *mddev)

drivers/md/raid5.h

@@ -298,6 +298,7 @@ enum r5dev_flags {
 	R5_WantReplace, /* We need to update the replacement, we have read
 			 * data in, and now is a good time to write it out.
 			 */
+	R5_Discard,	/* Discard the stripe */
 };
 
 /*