Merge tag 'dm-4.0-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/device-mapper/linux-dm

Pull devicemapper fixes from Mike Snitzer:
 "A handful of stable fixes for DM:
   - fix thin target to always zero-fill reads to unprovisioned blocks
   - fix to interlock device destruction's suspend from internal
     suspends
   - fix 2 snapshot exception store handover bugs
   - fix dm-io to cope with DISCARD and WRITE_SAME capabilities changing"

* tag 'dm-4.0-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/device-mapper/linux-dm:
  dm io: deal with wandering queue limits when handling REQ_DISCARD and REQ_WRITE_SAME
  dm snapshot: suspend merging snapshot when doing exception handover
  dm snapshot: suspend origin when doing exception handover
  dm: hold suspend_lock while suspending device during device deletion
  dm thin: fix to consistently zero-fill reads to unprovisioned blocks

drivers/md/dm-io.c

@@ -289,9 +289,16 @@ static void do_region(int rw, unsigned region, struct dm_io_region *where,
 	struct request_queue *q = bdev_get_queue(where->bdev);
 	unsigned short logical_block_size = queue_logical_block_size(q);
 	sector_t num_sectors;
+	unsigned int uninitialized_var(special_cmd_max_sectors);
 
-	/* Reject unsupported discard requests */
-	if ((rw & REQ_DISCARD) && !blk_queue_discard(q)) {
+	/*
+	 * Reject unsupported discard and write same requests.
+	 */
+	if (rw & REQ_DISCARD)
+		special_cmd_max_sectors = q->limits.max_discard_sectors;
+	else if (rw & REQ_WRITE_SAME)
+		special_cmd_max_sectors = q->limits.max_write_same_sectors;
+	if ((rw & (REQ_DISCARD | REQ_WRITE_SAME)) && special_cmd_max_sectors == 0) {
 		dec_count(io, region, -EOPNOTSUPP);
 		return;
 	}
@@ -317,7 +324,7 @@ static void do_region(int rw, unsigned region, struct dm_io_region *where,
 		store_io_and_region_in_bio(bio, io, region);
 
 		if (rw & REQ_DISCARD) {
-			num_sectors = min_t(sector_t, q->limits.max_discard_sectors, remaining);
+			num_sectors = min_t(sector_t, special_cmd_max_sectors, remaining);
 			bio->bi_iter.bi_size = num_sectors << SECTOR_SHIFT;
 			remaining -= num_sectors;
 		} else if (rw & REQ_WRITE_SAME) {
@@ -326,7 +333,7 @@ static void do_region(int rw, unsigned region, struct dm_io_region *where,
 			 */
 			dp->get_page(dp, &page, &len, &offset);
 			bio_add_page(bio, page, logical_block_size, offset);
-			num_sectors = min_t(sector_t, q->limits.max_write_same_sectors, remaining);
+			num_sectors = min_t(sector_t, special_cmd_max_sectors, remaining);
 			bio->bi_iter.bi_size = num_sectors << SECTOR_SHIFT;
 
 			offset = 0;

drivers/md/dm-snap.c

@@ -20,6 +20,8 @@
 #include <linux/log2.h>
 #include <linux/dm-kcopyd.h>
 
+#include "dm.h"
+
 #include "dm-exception-store.h"
 
 #define DM_MSG_PREFIX "snapshots"
@@ -290,6 +292,16 @@ struct origin {
 	struct list_head snapshots;
 };
 
+/*
+ * This structure is allocated for each origin target
+ */
+struct dm_origin {
+	struct dm_dev *dev;
+	struct dm_target *ti;
+	unsigned split_boundary;
+	struct list_head hash_list;
+};
+
 /*
  * Size of the hash table for origin volumes. If we make this
  * the size of the minors list then it should be nearly perfect
@@ -297,6 +309,7 @@ struct origin {
 #define ORIGIN_HASH_SIZE 256
 #define ORIGIN_MASK      0xFF
 static struct list_head *_origins;
+static struct list_head *_dm_origins;
 static struct rw_semaphore _origins_lock;
 
 static DECLARE_WAIT_QUEUE_HEAD(_pending_exceptions_done);
@@ -310,12 +323,22 @@ static int init_origin_hash(void)
 	_origins = kmalloc(ORIGIN_HASH_SIZE * sizeof(struct list_head),
 			   GFP_KERNEL);
 	if (!_origins) {
-		DMERR("unable to allocate memory");
+		DMERR("unable to allocate memory for _origins");
 		return -ENOMEM;
 	}
-
 	for (i = 0; i < ORIGIN_HASH_SIZE; i++)
 		INIT_LIST_HEAD(_origins + i);
+
+	_dm_origins = kmalloc(ORIGIN_HASH_SIZE * sizeof(struct list_head),
+			      GFP_KERNEL);
+	if (!_dm_origins) {
+		DMERR("unable to allocate memory for _dm_origins");
+		kfree(_origins);
+		return -ENOMEM;
+	}
+	for (i = 0; i < ORIGIN_HASH_SIZE; i++)
+		INIT_LIST_HEAD(_dm_origins + i);
+
 	init_rwsem(&_origins_lock);
 
 	return 0;
@@ -324,6 +347,7 @@ static int init_origin_hash(void)
 static void exit_origin_hash(void)
 {
 	kfree(_origins);
+	kfree(_dm_origins);
 }
 
 static unsigned origin_hash(struct block_device *bdev)
@@ -350,6 +374,30 @@ static void __insert_origin(struct origin *o)
 	list_add_tail(&o->hash_list, sl);
 }
 
+static struct dm_origin *__lookup_dm_origin(struct block_device *origin)
+{
+	struct list_head *ol;
+	struct dm_origin *o;
+
+	ol = &_dm_origins[origin_hash(origin)];
+	list_for_each_entry (o, ol, hash_list)
+		if (bdev_equal(o->dev->bdev, origin))
+			return o;
+
+	return NULL;
+}
+
+static void __insert_dm_origin(struct dm_origin *o)
+{
+	struct list_head *sl = &_dm_origins[origin_hash(o->dev->bdev)];
+	list_add_tail(&o->hash_list, sl);
+}
+
+static void __remove_dm_origin(struct dm_origin *o)
+{
+	list_del(&o->hash_list);
+}
+
 /*
  * _origins_lock must be held when calling this function.
  * Returns number of snapshots registered using the supplied cow device, plus:
@@ -1840,9 +1888,40 @@ static int snapshot_preresume(struct dm_target *ti)
 static void snapshot_resume(struct dm_target *ti)
 {
 	struct dm_snapshot *s = ti->private;
-	struct dm_snapshot *snap_src = NULL, *snap_dest = NULL;
+	struct dm_snapshot *snap_src = NULL, *snap_dest = NULL, *snap_merging = NULL;
+	struct dm_origin *o;
+	struct mapped_device *origin_md = NULL;
+	bool must_restart_merging = false;
 
 	down_read(&_origins_lock);
+
+	o = __lookup_dm_origin(s->origin->bdev);
+	if (o)
+		origin_md = dm_table_get_md(o->ti->table);
+	if (!origin_md) {
+		(void) __find_snapshots_sharing_cow(s, NULL, NULL, &snap_merging);
+		if (snap_merging)
+			origin_md = dm_table_get_md(snap_merging->ti->table);
+	}
+	if (origin_md == dm_table_get_md(ti->table))
+		origin_md = NULL;
+	if (origin_md) {
+		if (dm_hold(origin_md))
+			origin_md = NULL;
+	}
+
+	up_read(&_origins_lock);
+
+	if (origin_md) {
+		dm_internal_suspend_fast(origin_md);
+		if (snap_merging && test_bit(RUNNING_MERGE, &snap_merging->state_bits)) {
+			must_restart_merging = true;
+			stop_merge(snap_merging);
+		}
+	}
+
+	down_read(&_origins_lock);
+
 	(void) __find_snapshots_sharing_cow(s, &snap_src, &snap_dest, NULL);
 	if (snap_src && snap_dest) {
 		down_write(&snap_src->lock);
@@ -1851,8 +1930,16 @@ static void snapshot_resume(struct dm_target *ti)
 		up_write(&snap_dest->lock);
 		up_write(&snap_src->lock);
 	}
+
 	up_read(&_origins_lock);
 
+	if (origin_md) {
+		if (must_restart_merging)
+			start_merge(snap_merging);
+		dm_internal_resume_fast(origin_md);
+		dm_put(origin_md);
+	}
+
 	/* Now we have correct chunk size, reregister */
 	reregister_snapshot(s);
 
@@ -2133,11 +2220,6 @@ static int origin_write_extent(struct dm_snapshot *merging_snap,
  * Origin: maps a linear range of a device, with hooks for snapshotting.
  */
 
-struct dm_origin {
-	struct dm_dev *dev;
-	unsigned split_boundary;
-};
-
 /*
  * Construct an origin mapping: <dev_path>
  * The context for an origin is merely a 'struct dm_dev *'
@@ -2166,6 +2248,7 @@ static int origin_ctr(struct dm_target *ti, unsigned int argc, char **argv)
 		goto bad_open;
 	}
 
+	o->ti = ti;
 	ti->private = o;
 	ti->num_flush_bios = 1;
 
@@ -2180,6 +2263,7 @@ static int origin_ctr(struct dm_target *ti, unsigned int argc, char **argv)
 static void origin_dtr(struct dm_target *ti)
 {
 	struct dm_origin *o = ti->private;
+
 	dm_put_device(ti, o->dev);
 	kfree(o);
 }
@@ -2216,6 +2300,19 @@ static void origin_resume(struct dm_target *ti)
 	struct dm_origin *o = ti->private;
 
 	o->split_boundary = get_origin_minimum_chunksize(o->dev->bdev);
+
+	down_write(&_origins_lock);
+	__insert_dm_origin(o);
+	up_write(&_origins_lock);
+}
+
+static void origin_postsuspend(struct dm_target *ti)
+{
+	struct dm_origin *o = ti->private;
+
+	down_write(&_origins_lock);
+	__remove_dm_origin(o);
+	up_write(&_origins_lock);
 }
 
 static void origin_status(struct dm_target *ti, status_type_t type,
@@ -2258,12 +2355,13 @@ static int origin_iterate_devices(struct dm_target *ti,
 
 static struct target_type origin_target = {
 	.name    = "snapshot-origin",
-	.version = {1, 8, 1},
+	.version = {1, 9, 0},
 	.module  = THIS_MODULE,
 	.ctr     = origin_ctr,
 	.dtr     = origin_dtr,
 	.map     = origin_map,
 	.resume  = origin_resume,
+	.postsuspend = origin_postsuspend,
 	.status  = origin_status,
 	.merge	 = origin_merge,
 	.iterate_devices = origin_iterate_devices,
@@ -2271,7 +2369,7 @@ static struct target_type origin_target = {
 
 static struct target_type snapshot_target = {
 	.name    = "snapshot",
-	.version = {1, 12, 0},
+	.version = {1, 13, 0},
 	.module  = THIS_MODULE,
 	.ctr     = snapshot_ctr,
 	.dtr     = snapshot_dtr,
@@ -2285,7 +2383,7 @@ static struct target_type snapshot_target = {
 
 static struct target_type merge_target = {
 	.name    = dm_snapshot_merge_target_name,
-	.version = {1, 2, 0},
+	.version = {1, 3, 0},
 	.module  = THIS_MODULE,
 	.ctr     = snapshot_ctr,
 	.dtr     = snapshot_dtr,

drivers/md/dm-thin.c

@@ -2358,17 +2358,6 @@ static int thin_bio_map(struct dm_target *ti, struct bio *bio)
 		return DM_MAPIO_REMAPPED;
 
 	case -ENODATA:
-		if (get_pool_mode(tc->pool) == PM_READ_ONLY) {
-			/*
-			 * This block isn't provisioned, and we have no way
-			 * of doing so.
-			 */
-			handle_unserviceable_bio(tc->pool, bio);
-			cell_defer_no_holder(tc, virt_cell);
-			return DM_MAPIO_SUBMITTED;
-		}
-		/* fall through */
-
 	case -EWOULDBLOCK:
 		thin_defer_cell(tc, virt_cell);
 		return DM_MAPIO_SUBMITTED;

drivers/md/dm.c

@@ -2616,6 +2616,19 @@ void dm_get(struct mapped_device *md)
 	BUG_ON(test_bit(DMF_FREEING, &md->flags));
 }
 
+int dm_hold(struct mapped_device *md)
+{
+	spin_lock(&_minor_lock);
+	if (test_bit(DMF_FREEING, &md->flags)) {
+		spin_unlock(&_minor_lock);
+		return -EBUSY;
+	}
+	dm_get(md);
+	spin_unlock(&_minor_lock);
+	return 0;
+}
+EXPORT_SYMBOL_GPL(dm_hold);
+
 const char *dm_device_name(struct mapped_device *md)
 {
 	return md->name;
@@ -2638,10 +2651,16 @@ static void __dm_destroy(struct mapped_device *md, bool wait)
 	if (dm_request_based(md))
 		flush_kthread_worker(&md->kworker);
 
+	/*
+	 * Take suspend_lock so that presuspend and postsuspend methods
+	 * do not race with internal suspend.
+	 */
+	mutex_lock(&md->suspend_lock);
 	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);
@@ -3115,6 +3134,7 @@ void dm_internal_suspend_fast(struct mapped_device *md)
 	flush_workqueue(md->wq);
 	dm_wait_for_completion(md, TASK_UNINTERRUPTIBLE);
 }
+EXPORT_SYMBOL_GPL(dm_internal_suspend_fast);
 
 void dm_internal_resume_fast(struct mapped_device *md)
 {
@@ -3126,6 +3146,7 @@ void dm_internal_resume_fast(struct mapped_device *md)
 done:
 	mutex_unlock(&md->suspend_lock);
 }
+EXPORT_SYMBOL_GPL(dm_internal_resume_fast);
 
 /*-----------------------------------------------------------------
  * Event notification.

include/linux/device-mapper.h

@@ -375,6 +375,7 @@ int dm_create(int minor, struct mapped_device **md);
  */
 struct mapped_device *dm_get_md(dev_t dev);
 void dm_get(struct mapped_device *md);
+int dm_hold(struct mapped_device *md);
 void dm_put(struct mapped_device *md);
 
 /*