1 files changed, 85 insertions, 45 deletions

diff --git a/drivers/md/md.c b/drivers/md/md.c
index 6efe49f7b..09c7f5215 100644
--- a/drivers/md/md.c
+++ b/drivers/md/md.c
@@ -1179,6 +1179,7 @@ struct super_type  {
 					  struct md_rdev *refdev,
 					  int minor_version);
 	int		    (*validate_super)(struct mddev *mddev,
+					      struct md_rdev *freshest,
 					      struct md_rdev *rdev);
 	void		    (*sync_super)(struct mddev *mddev,
 					  struct md_rdev *rdev);
@@ -1317,8 +1318,9 @@ static int super_90_load(struct md_rdev *rdev, struct md_rdev *refdev, int minor
 
 /*
  * validate_super for 0.90.0
+ * note: we are not using "freshest" for 0.9 superblock
  */
-static int super_90_validate(struct mddev *mddev, struct md_rdev *rdev)
+static int super_90_validate(struct mddev *mddev, struct md_rdev *freshest, struct md_rdev *rdev)
 {
 	mdp_disk_t *desc;
 	mdp_super_t *sb = page_address(rdev->sb_page);
@@ -1833,7 +1835,7 @@ static int super_1_load(struct md_rdev *rdev, struct md_rdev *refdev, int minor_
 	return ret;
 }
 
-static int super_1_validate(struct mddev *mddev, struct md_rdev *rdev)
+static int super_1_validate(struct mddev *mddev, struct md_rdev *freshest, struct md_rdev *rdev)
 {
 	struct mdp_superblock_1 *sb = page_address(rdev->sb_page);
 	__u64 ev1 = le64_to_cpu(sb->events);
@@ -1929,13 +1931,15 @@ static int super_1_validate(struct mddev *mddev, struct md_rdev *rdev)
 		}
 	} else if (mddev->pers == NULL) {
 		/* Insist of good event counter while assembling, except for
-		 * spares (which don't need an event count) */
-		++ev1;
+		 * spares (which don't need an event count).
+		 * Similar to mdadm, we allow event counter difference of 1
+		 * from the freshest device.
+		 */
 		if (rdev->desc_nr >= 0 &&
 		    rdev->desc_nr < le32_to_cpu(sb->max_dev) &&
 		    (le16_to_cpu(sb->dev_roles[rdev->desc_nr]) < MD_DISK_ROLE_MAX ||
 		     le16_to_cpu(sb->dev_roles[rdev->desc_nr]) == MD_DISK_ROLE_JOURNAL))
-			if (ev1 < mddev->events)
+			if (ev1 + 1 < mddev->events)
 				return -EINVAL;
 	} else if (mddev->bitmap) {
 		/* If adding to array with a bitmap, then we can accept an
@@ -1956,8 +1960,38 @@ static int super_1_validate(struct mddev *mddev, struct md_rdev *rdev)
 		    rdev->desc_nr >= le32_to_cpu(sb->max_dev)) {
 			role = MD_DISK_ROLE_SPARE;
 			rdev->desc_nr = -1;
-		} else
+		} else if (mddev->pers == NULL && freshest && ev1 < mddev->events) {
+			/*
+			 * If we are assembling, and our event counter is smaller than the
+			 * highest event counter, we cannot trust our superblock about the role.
+			 * It could happen that our rdev was marked as Faulty, and all other
+			 * superblocks were updated with +1 event counter.
+			 * Then, before the next superblock update, which typically happens when
+			 * remove_and_add_spares() removes the device from the array, there was
+			 * a crash or reboot.
+			 * If we allow current rdev without consulting the freshest superblock,
+			 * we could cause data corruption.
+			 * Note that in this case our event counter is smaller by 1 than the
+			 * highest, otherwise, this rdev would not be allowed into array;
+			 * both kernel and mdadm allow event counter difference of 1.
+			 */
+			struct mdp_superblock_1 *freshest_sb = page_address(freshest->sb_page);
+			u32 freshest_max_dev = le32_to_cpu(freshest_sb->max_dev);
+
+			if (rdev->desc_nr >= freshest_max_dev) {
+				/* this is unexpected, better not proceed */
+				pr_warn("md: %s: rdev[%pg]: desc_nr(%d) >= freshest(%pg)->sb->max_dev(%u)\n",
+						mdname(mddev), rdev->bdev, rdev->desc_nr,
+						freshest->bdev, freshest_max_dev);
+				return -EUCLEAN;
+			}
+
+			role = le16_to_cpu(freshest_sb->dev_roles[rdev->desc_nr]);
+			pr_debug("md: %s: rdev[%pg]: role=%d(0x%x) according to freshest %pg\n",
+				     mdname(mddev), rdev->bdev, role, role, freshest->bdev);
+		} else {
 			role = le16_to_cpu(sb->dev_roles[rdev->desc_nr]);
+		}
 		switch(role) {
 		case MD_DISK_ROLE_SPARE: /* spare */
 			break;
@@ -2896,7 +2930,7 @@ static int add_bound_rdev(struct md_rdev *rdev)
 		 * and should be added immediately.
 		 */
 		super_types[mddev->major_version].
-			validate_super(mddev, rdev);
+			validate_super(mddev, NULL/*freshest*/, rdev);
 		if (add_journal)
 			mddev_suspend(mddev);
 		err = mddev->pers->hot_add_disk(mddev, rdev);
@@ -3814,7 +3848,7 @@ static int analyze_sbs(struct mddev *mddev)
 	}
 
 	super_types[mddev->major_version].
-		validate_super(mddev, freshest);
+		validate_super(mddev, NULL/*freshest*/, freshest);
 
 	i = 0;
 	rdev_for_each_safe(rdev, tmp, mddev) {
@@ -3829,7 +3863,7 @@ static int analyze_sbs(struct mddev *mddev)
 		}
 		if (rdev != freshest) {
 			if (super_types[mddev->major_version].
-			    validate_super(mddev, rdev)) {
+			    validate_super(mddev, freshest, rdev)) {
 				pr_warn("md: kicking non-fresh %s from array!\n",
 					bdevname(rdev->bdev,b));
 				md_kick_rdev_from_array(rdev);
@@ -6209,7 +6243,15 @@ static void md_clean(struct mddev *mddev)
 	mddev->persistent = 0;
 	mddev->level = LEVEL_NONE;
 	mddev->clevel[0] = 0;
-	mddev->flags = 0;
+	/*
+	 * Don't clear MD_CLOSING, or mddev can be opened again.
+	 * 'hold_active != 0' means mddev is still in the creation
+	 * process and will be used later.
+	 */
+	if (mddev->hold_active)
+		mddev->flags = 0;
+	else
+		mddev->flags &= BIT_ULL_MASK(MD_CLOSING);
 	mddev->sb_flags = 0;
 	mddev->ro = 0;
 	mddev->metadata_type[0] = 0;
@@ -6817,7 +6859,7 @@ int md_add_new_disk(struct mddev *mddev, struct mdu_disk_info_s *info)
 			rdev->saved_raid_disk = rdev->raid_disk;
 		} else
 			super_types[mddev->major_version].
-				validate_super(mddev, rdev);
+				validate_super(mddev, NULL/*freshest*/, rdev);
 		if ((info->state & (1<<MD_DISK_SYNC)) &&
 		     rdev->raid_disk != info->raid_disk) {
 			/* This was a hot-add request, but events doesn't
@@ -7502,7 +7544,6 @@ static inline bool md_ioctl_valid(unsigned int cmd)
 {
 	switch (cmd) {
 	case ADD_NEW_DISK:
-	case BLKROSET:
 	case GET_ARRAY_INFO:
 	case GET_BITMAP_FILE:
 	case GET_DISK_INFO:
@@ -7529,8 +7570,6 @@ static int md_ioctl(struct block_device *bdev, fmode_t mode,
 	int err = 0;
 	void __user *argp = (void __user *)arg;
 	struct mddev *mddev = NULL;
-	int ro;
-	bool did_set_md_closing = false;
 
 	if (!md_ioctl_valid(cmd))
 		return -ENOTTY;
@@ -7617,7 +7656,6 @@ static int md_ioctl(struct block_device *bdev, fmode_t mode,
 			err = -EBUSY;
 			goto out;
 		}
-		did_set_md_closing = true;
 		mutex_unlock(&mddev->open_mutex);
 		sync_blockdev(bdev);
 	}
@@ -7712,35 +7750,6 @@ static int md_ioctl(struct block_device *bdev, fmode_t mode,
 			goto unlock;
 		}
 		break;
-
-	case BLKROSET:
-		if (get_user(ro, (int __user *)(arg))) {
-			err = -EFAULT;
-			goto unlock;
-		}
-		err = -EINVAL;
-
-		/* if the bdev is going readonly the value of mddev->ro
-		 * does not matter, no writes are coming
-		 */
-		if (ro)
-			goto unlock;
-
-		/* are we are already prepared for writes? */
-		if (mddev->ro != 1)
-			goto unlock;
-
-		/* transitioning to readauto need only happen for
-		 * arrays that call md_write_start
-		 */
-		if (mddev->pers) {
-			err = restart_array(mddev);
-			if (err == 0) {
-				mddev->ro = 2;
-				set_disk_ro(mddev->gendisk, 0);
-			}
-		}
-		goto unlock;
 	}
 
 	/*
@@ -7810,7 +7819,7 @@ unlock:
 		mddev->hold_active = 0;
 	mddev_unlock(mddev);
 out:
-	if(did_set_md_closing)
+	if (cmd == STOP_ARRAY_RO || (err && cmd == STOP_ARRAY))
 		clear_bit(MD_CLOSING, &mddev->flags);
 	return err;
 }
@@ -7834,6 +7843,36 @@ static int md_compat_ioctl(struct block_device *bdev, fmode_t mode,
 }
 #endif /* CONFIG_COMPAT */
 
+static int md_set_read_only(struct block_device *bdev, bool ro)
+{
+	struct mddev *mddev = bdev->bd_disk->private_data;
+	int err;
+
+	err = mddev_lock(mddev);
+	if (err)
+		return err;
+
+	if (!mddev->raid_disks && !mddev->external) {
+		err = -ENODEV;
+		goto out_unlock;
+	}
+
+	/*
+	 * Transitioning to read-auto need only happen for arrays that call
+	 * md_write_start and which are not ready for writes yet.
+	 */
+	if (!ro && mddev->ro == 1 && mddev->pers) {
+		err = restart_array(mddev);
+		if (err)
+			goto out_unlock;
+		mddev->ro = 2;
+	}
+
+out_unlock:
+	mddev_unlock(mddev);
+	return err;
+}
+
 static int md_open(struct block_device *bdev, fmode_t mode)
 {
 	/*
@@ -7910,6 +7949,7 @@ const struct block_device_operations md_fops =
 #endif
 	.getgeo		= md_getgeo,
 	.check_events	= md_check_events,
+	.set_read_only	= md_set_read_only,
 };
 
 static int md_thread(void *arg)