Merge tag 'dm-4.3-fixes-2' of git://git.kernel.org/pub/scm/linux/kern…

…el/git/device-mapper/linux-dm

Pull dm fixes from Mike Snitzer:
 "Three stable fixes:

   - DM core AB-BA deadlock fix in the device destruction path (vs
     device creation's DM table swap).

   - DM raid fix to properly round up the region_size to the next
     power-of-2.

   - DM cache fix for a NULL pointer seen while switching from the
     "cleaner" cache policy.

  Two fixes for regressions introduced during the 4.3 merge:

   - request-based DM error propagation regressed due to incorrect
     changes introduced when adding the bi_error field to bio.

   - DM snapshot fix to only support snapshots that overflow if the
     client (e.g. lvm2) is prepared to deal with the associated
     snapshot status interface change"

* tag 'dm-4.3-fixes-2' of git://git.kernel.org/pub/scm/linux/kernel/git/device-mapper/linux-dm:
  dm snapshot: add new persistent store option to support overflow
  dm cache: fix NULL pointer when switching from cleaner policy
  dm: fix request-based dm error reporting
  dm raid: fix round up of default region size
  dm: fix AB-BA deadlock in __dm_destroy()

Documentation/device-mapper/snapshot.txt

@@ -41,9 +41,13 @@ useless and be disabled, returning errors.  So it is important to monitor
 the amount of free space and expand the <COW device> before it fills up.
 
 <persistent?> is P (Persistent) or N (Not persistent - will not survive
-after reboot).
-The difference is that for transient snapshots less metadata must be
-saved on disk - they can be kept in memory by the kernel.
+after reboot).  O (Overflow) can be added as a persistent store option
+to allow userspace to advertise its support for seeing "Overflow" in the
+snapshot status.  So supported store types are "P", "PO" and "N".
+
+The difference between persistent and transient is with transient
+snapshots less metadata must be saved on disk - they can be kept in
+memory by the kernel.
 
 
 * snapshot-merge <origin> <COW device> <persistent> <chunksize>

drivers/md/dm-cache-policy-cleaner.c

@@ -436,7 +436,7 @@ static struct dm_cache_policy *wb_create(dm_cblock_t cache_size,
 static struct dm_cache_policy_type wb_policy_type = {
 	.name = "cleaner",
 	.version = {1, 0, 0},
-	.hint_size = 0,
+	.hint_size = 4,
 	.owner = THIS_MODULE,
 	.create = wb_create
 };

drivers/md/dm-exception-store.c

@@ -203,7 +203,7 @@ int dm_exception_store_create(struct dm_target *ti, int argc, char **argv,
 		return -EINVAL;
 	}
 
-	tmp_store = kmalloc(sizeof(*tmp_store), GFP_KERNEL);
+	tmp_store = kzalloc(sizeof(*tmp_store), GFP_KERNEL);
 	if (!tmp_store) {
 		ti->error = "Exception store allocation failed";
 		return -ENOMEM;
@@ -215,7 +215,7 @@ int dm_exception_store_create(struct dm_target *ti, int argc, char **argv,
 	else if (persistent == 'N')
 		type = get_type("N");
 	else {
-		ti->error = "Persistent flag is not P or N";
+		ti->error = "Exception store type is not P or N";
 		r = -EINVAL;
 		goto bad_type;
 	}
@@ -233,7 +233,7 @@ int dm_exception_store_create(struct dm_target *ti, int argc, char **argv,
 	if (r)
 		goto bad;
 
-	r = type->ctr(tmp_store, 0, NULL);
+	r = type->ctr(tmp_store, (strlen(argv[0]) > 1 ? &argv[0][1] : NULL));
 	if (r) {
 		ti->error = "Exception store type constructor failed";
 		goto bad;

drivers/md/dm-exception-store.h

@@ -42,8 +42,7 @@ struct dm_exception_store_type {
 	const char *name;
 	struct module *module;
 
-	int (*ctr) (struct dm_exception_store *store,
-		    unsigned argc, char **argv);
+	int (*ctr) (struct dm_exception_store *store, char *options);
 
 	/*
 	 * Destroys this object when you've finished with it.
@@ -123,6 +122,8 @@ struct dm_exception_store {
 	unsigned chunk_shift;
 
 	void *context;
+
+	bool userspace_supports_overflow;
 };
 
 /*

drivers/md/dm-raid.c

@@ -329,8 +329,7 @@ static int validate_region_size(struct raid_set *rs, unsigned long region_size)
 		 */
 		if (min_region_size > (1 << 13)) {
 			/* If not a power of 2, make it the next power of 2 */
-			if (min_region_size & (min_region_size - 1))
-				region_size = 1 << fls(region_size);
+			region_size = roundup_pow_of_two(min_region_size);
 			DMINFO("Choosing default region size of %lu sectors",
 			       region_size);
 		} else {

drivers/md/dm-snap-persistent.c

@@ -7,6 +7,7 @@
 
 #include "dm-exception-store.h"
 
+#include <linux/ctype.h>
 #include <linux/mm.h>
 #include <linux/pagemap.h>
 #include <linux/vmalloc.h>
@@ -843,8 +844,7 @@ static void persistent_drop_snapshot(struct dm_exception_store *store)
 		DMWARN("write header failed");
 }
 
-static int persistent_ctr(struct dm_exception_store *store,
-			  unsigned argc, char **argv)
+static int persistent_ctr(struct dm_exception_store *store, char *options)
 {
 	struct pstore *ps;
 
@@ -873,6 +873,16 @@ static int persistent_ctr(struct dm_exception_store *store,
 		return -ENOMEM;
 	}
 
+	if (options) {
+		char overflow = toupper(options[0]);
+		if (overflow == 'O')
+			store->userspace_supports_overflow = true;
+		else {
+			DMERR("Unsupported persistent store option: %s", options);
+			return -EINVAL;
+		}
+	}
+
 	store->context = ps;
 
 	return 0;
@@ -888,7 +898,8 @@ static unsigned persistent_status(struct dm_exception_store *store,
 	case STATUSTYPE_INFO:
 		break;
 	case STATUSTYPE_TABLE:
-		DMEMIT(" P %llu", (unsigned long long)store->chunk_size);
+		DMEMIT(" %s %llu", store->userspace_supports_overflow ? "PO" : "P",
+		       (unsigned long long)store->chunk_size);
 	}
 
 	return sz;

drivers/md/dm-snap-transient.c

@@ -70,8 +70,7 @@ static void transient_usage(struct dm_exception_store *store,
 	*metadata_sectors = 0;
 }
 
-static int transient_ctr(struct dm_exception_store *store,
-			 unsigned argc, char **argv)
+static int transient_ctr(struct dm_exception_store *store, char *options)
 {
 	struct transient_c *tc;
 

drivers/md/dm-snap.c

@@ -1098,7 +1098,7 @@ static void stop_merge(struct dm_snapshot *s)
 }
 
 /*
- * Construct a snapshot mapping: <origin_dev> <COW-dev> <p/n> <chunk-size>
+ * Construct a snapshot mapping: <origin_dev> <COW-dev> <p|po|n> <chunk-size>
  */
 static int snapshot_ctr(struct dm_target *ti, unsigned int argc, char **argv)
 {
@@ -1302,6 +1302,7 @@ static void __handover_exceptions(struct dm_snapshot *snap_src,
 
 	u.store_swap = snap_dest->store;
 	snap_dest->store = snap_src->store;
+	snap_dest->store->userspace_supports_overflow = u.store_swap->userspace_supports_overflow;
 	snap_src->store = u.store_swap;
 
 	snap_dest->store->snap = snap_dest;
@@ -1739,8 +1740,11 @@ static int snapshot_map(struct dm_target *ti, struct bio *bio)
 
 			pe = __find_pending_exception(s, pe, chunk);
 			if (!pe) {
-				s->snapshot_overflowed = 1;
-				DMERR("Snapshot overflowed: Unable to allocate exception.");
+				if (s->store->userspace_supports_overflow) {
+					s->snapshot_overflowed = 1;
+					DMERR("Snapshot overflowed: Unable to allocate exception.");
+				} else
+					__invalidate_snapshot(s, -ENOMEM);
 				r = -EIO;
 				goto out_unlock;
 			}
@@ -2365,7 +2369,7 @@ static struct target_type origin_target = {
 
 static struct target_type snapshot_target = {
 	.name    = "snapshot",
-	.version = {1, 14, 0},
+	.version = {1, 15, 0},
 	.module  = THIS_MODULE,
 	.ctr     = snapshot_ctr,
 	.dtr     = snapshot_dtr,
@@ -2379,7 +2383,7 @@ static struct target_type snapshot_target = {
 
 static struct target_type merge_target = {
 	.name    = dm_snapshot_merge_target_name,
-	.version = {1, 3, 0},
+	.version = {1, 4, 0},
 	.module  = THIS_MODULE,
 	.ctr     = snapshot_ctr,
 	.dtr     = snapshot_dtr,

drivers/md/dm.c

@@ -1001,6 +1001,7 @@ static void end_clone_bio(struct bio *clone)
 	struct dm_rq_target_io *tio = info->tio;
 	struct bio *bio = info->orig;
 	unsigned int nr_bytes = info->orig->bi_iter.bi_size;
+	int error = clone->bi_error;
 
 	bio_put(clone);
 
@@ -1011,13 +1012,13 @@ static void end_clone_bio(struct bio *clone)
 		 * the remainder.
 		 */
 		return;
-	else if (bio->bi_error) {
+	else if (error) {
 		/*
 		 * Don't notice the error to the upper layer yet.
 		 * The error handling decision is made by the target driver,
 		 * when the request is completed.
 		 */
-		tio->error = bio->bi_error;
+		tio->error = error;
 		return;
 	}
 
@@ -2837,8 +2838,6 @@ static void __dm_destroy(struct mapped_device *md, bool wait)
 
 	might_sleep();
 
-	map = dm_get_live_table(md, &srcu_idx);
-
 	spin_lock(&_minor_lock);
 	idr_replace(&_minor_idr, MINOR_ALLOCED, MINOR(disk_devt(dm_disk(md))));
 	set_bit(DMF_FREEING, &md->flags);
@@ -2852,14 +2851,14 @@ static void __dm_destroy(struct mapped_device *md, bool wait)
 	 * do not race with internal suspend.
 	 */
 	mutex_lock(&md->suspend_lock);
+	map = dm_get_live_table(md, &srcu_idx);
 	if (!dm_suspended_md(md)) {
 		dm_table_presuspend_targets(map);
 		dm_table_postsuspend_targets(map);
 	}
-	mutex_unlock(&md->suspend_lock);
-
 	/* dm_put_live_table must be before msleep, otherwise deadlock is possible */
 	dm_put_live_table(md, srcu_idx);
+	mutex_unlock(&md->suspend_lock);
 
 	/*
 	 * Rare, but there may be I/O requests still going to complete,