Merge tag 'dm-3.6-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/agk/linux-dm

Pull dm fixes from Alasdair G Kergon:
 "A few fixes for problems discovered during the 3.6 cycle.

  Of particular note, are fixes to the thin target's discard support,
  which I hope is finally working correctly; and fixes for multipath
  ioctls and device limits when there are no paths."

* tag 'dm-3.6-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/agk/linux-dm:
  dm verity: fix overflow check
  dm thin: fix discard support for data devices
  dm thin: tidy discard support
  dm: retain table limits when swapping to new table with no devices
  dm table: clear add_random unless all devices have it set
  dm: handle requests beyond end of device instead of using BUG_ON
  dm mpath: only retry ioctl when no paths if queue_if_no_path set
  dm thin: do not set discard_zeroes_data

drivers/md/dm-mpath.c

@@ -1555,6 +1555,7 @@ static int multipath_ioctl(struct dm_target *ti, unsigned int cmd,
 			   unsigned long arg)
 {
 	struct multipath *m = ti->private;
+	struct pgpath *pgpath;
 	struct block_device *bdev;
 	fmode_t mode;
 	unsigned long flags;
@@ -1570,12 +1571,14 @@ static int multipath_ioctl(struct dm_target *ti, unsigned int cmd,
 	if (!m->current_pgpath)
 		__choose_pgpath(m, 0);
 
-	if (m->current_pgpath) {
-		bdev = m->current_pgpath->path.dev->bdev;
-		mode = m->current_pgpath->path.dev->mode;
+	pgpath = m->current_pgpath;
+
+	if (pgpath) {
+		bdev = pgpath->path.dev->bdev;
+		mode = pgpath->path.dev->mode;
 	}
 
-	if (m->queue_io)
+	if ((pgpath && m->queue_io) || (!pgpath && m->queue_if_no_path))
 		r = -EAGAIN;
 	else if (!bdev)
 		r = -EIO;

drivers/md/dm-table.c

@@ -1212,6 +1212,41 @@ struct dm_target *dm_table_find_target(struct dm_table *t, sector_t sector)
 	return &t->targets[(KEYS_PER_NODE * n) + k];
 }
 
+static int count_device(struct dm_target *ti, struct dm_dev *dev,
+			sector_t start, sector_t len, void *data)
+{
+	unsigned *num_devices = data;
+
+	(*num_devices)++;
+
+	return 0;
+}
+
+/*
+ * Check whether a table has no data devices attached using each
+ * target's iterate_devices method.
+ * Returns false if the result is unknown because a target doesn't
+ * support iterate_devices.
+ */
+bool dm_table_has_no_data_devices(struct dm_table *table)
+{
+	struct dm_target *uninitialized_var(ti);
+	unsigned i = 0, num_devices = 0;
+
+	while (i < dm_table_get_num_targets(table)) {
+		ti = dm_table_get_target(table, i++);
+
+		if (!ti->type->iterate_devices)
+			return false;
+
+		ti->type->iterate_devices(ti, count_device, &num_devices);
+		if (num_devices)
+			return false;
+	}
+
+	return true;
+}
+
 /*
  * Establish the new table's queue_limits and validate them.
  */
@@ -1354,17 +1389,25 @@ static int device_is_nonrot(struct dm_target *ti, struct dm_dev *dev,
 	return q && blk_queue_nonrot(q);
 }
 
-static bool dm_table_is_nonrot(struct dm_table *t)
+static int device_is_not_random(struct dm_target *ti, struct dm_dev *dev,
+			     sector_t start, sector_t len, void *data)
+{
+	struct request_queue *q = bdev_get_queue(dev->bdev);
+
+	return q && !blk_queue_add_random(q);
+}
+
+static bool dm_table_all_devices_attribute(struct dm_table *t,
+					   iterate_devices_callout_fn func)
 {
 	struct dm_target *ti;
 	unsigned i = 0;
 
-	/* Ensure that all underlying device are non-rotational. */
 	while (i < dm_table_get_num_targets(t)) {
 		ti = dm_table_get_target(t, i++);
 
 		if (!ti->type->iterate_devices ||
-		    !ti->type->iterate_devices(ti, device_is_nonrot, NULL))
+		    !ti->type->iterate_devices(ti, func, NULL))
 			return 0;
 	}
 
@@ -1396,13 +1439,23 @@ void dm_table_set_restrictions(struct dm_table *t, struct request_queue *q,
 	if (!dm_table_discard_zeroes_data(t))
 		q->limits.discard_zeroes_data = 0;
 
-	if (dm_table_is_nonrot(t))
+	/* Ensure that all underlying devices are non-rotational. */
+	if (dm_table_all_devices_attribute(t, device_is_nonrot))
 		queue_flag_set_unlocked(QUEUE_FLAG_NONROT, q);
 	else
 		queue_flag_clear_unlocked(QUEUE_FLAG_NONROT, q);
 
 	dm_table_set_integrity(t);
 
+	/*
+	 * Determine whether or not this queue's I/O timings contribute
+	 * to the entropy pool, Only request-based targets use this.
+	 * Clear QUEUE_FLAG_ADD_RANDOM if any underlying device does not
+	 * have it set.
+	 */
+	if (blk_queue_add_random(q) && dm_table_all_devices_attribute(t, device_is_not_random))
+		queue_flag_clear_unlocked(QUEUE_FLAG_ADD_RANDOM, q);
+
 	/*
 	 * QUEUE_FLAG_STACKABLE must be set after all queue settings are
 	 * visible to other CPUs because, once the flag is set, incoming bios

drivers/md/dm-thin.c

@@ -509,9 +509,9 @@ enum pool_mode {
 struct pool_features {
 	enum pool_mode mode;
 
-	unsigned zero_new_blocks:1;
-	unsigned discard_enabled:1;
-	unsigned discard_passdown:1;
+	bool zero_new_blocks:1;
+	bool discard_enabled:1;
+	bool discard_passdown:1;
 };
 
 struct thin_c;
@@ -580,7 +580,8 @@ struct pool_c {
 	struct dm_target_callbacks callbacks;
 
 	dm_block_t low_water_blocks;
-	struct pool_features pf;
+	struct pool_features requested_pf; /* Features requested during table load */
+	struct pool_features adjusted_pf;  /* Features used after adjusting for constituent devices */
 };
 
 /*
@@ -1839,6 +1840,47 @@ static void __requeue_bios(struct pool *pool)
 /*----------------------------------------------------------------
  * Binding of control targets to a pool object
  *--------------------------------------------------------------*/
+static bool data_dev_supports_discard(struct pool_c *pt)
+{
+	struct request_queue *q = bdev_get_queue(pt->data_dev->bdev);
+
+	return q && blk_queue_discard(q);
+}
+
+/*
+ * If discard_passdown was enabled verify that the data device
+ * supports discards.  Disable discard_passdown if not.
+ */
+static void disable_passdown_if_not_supported(struct pool_c *pt)
+{
+	struct pool *pool = pt->pool;
+	struct block_device *data_bdev = pt->data_dev->bdev;
+	struct queue_limits *data_limits = &bdev_get_queue(data_bdev)->limits;
+	sector_t block_size = pool->sectors_per_block << SECTOR_SHIFT;
+	const char *reason = NULL;
+	char buf[BDEVNAME_SIZE];
+
+	if (!pt->adjusted_pf.discard_passdown)
+		return;
+
+	if (!data_dev_supports_discard(pt))
+		reason = "discard unsupported";
+
+	else if (data_limits->max_discard_sectors < pool->sectors_per_block)
+		reason = "max discard sectors smaller than a block";
+
+	else if (data_limits->discard_granularity > block_size)
+		reason = "discard granularity larger than a block";
+
+	else if (block_size & (data_limits->discard_granularity - 1))
+		reason = "discard granularity not a factor of block size";
+
+	if (reason) {
+		DMWARN("Data device (%s) %s: Disabling discard passdown.", bdevname(data_bdev, buf), reason);
+		pt->adjusted_pf.discard_passdown = false;
+	}
+}
+
 static int bind_control_target(struct pool *pool, struct dm_target *ti)
 {
 	struct pool_c *pt = ti->private;
@@ -1847,31 +1889,16 @@ static int bind_control_target(struct pool *pool, struct dm_target *ti)
 	 * We want to make sure that degraded pools are never upgraded.
 	 */
 	enum pool_mode old_mode = pool->pf.mode;
-	enum pool_mode new_mode = pt->pf.mode;
+	enum pool_mode new_mode = pt->adjusted_pf.mode;
 
 	if (old_mode > new_mode)
 		new_mode = old_mode;
 
 	pool->ti = ti;
 	pool->low_water_blocks = pt->low_water_blocks;
-	pool->pf = pt->pf;
-	set_pool_mode(pool, new_mode);
+	pool->pf = pt->adjusted_pf;
 
-	/*
-	 * If discard_passdown was enabled verify that the data device
-	 * supports discards.  Disable discard_passdown if not; otherwise
-	 * -EOPNOTSUPP will be returned.
-	 */
-	/* FIXME: pull this out into a sep fn. */
-	if (pt->pf.discard_passdown) {
-		struct request_queue *q = bdev_get_queue(pt->data_dev->bdev);
-		if (!q || !blk_queue_discard(q)) {
-			char buf[BDEVNAME_SIZE];
-			DMWARN("Discard unsupported by data device (%s): Disabling discard passdown.",
-			       bdevname(pt->data_dev->bdev, buf));
-			pool->pf.discard_passdown = 0;
-		}
-	}
+	set_pool_mode(pool, new_mode);
 
 	return 0;
 }
@@ -1889,9 +1916,9 @@ static void unbind_control_target(struct pool *pool, struct dm_target *ti)
 static void pool_features_init(struct pool_features *pf)
 {
 	pf->mode = PM_WRITE;
-	pf->zero_new_blocks = 1;
-	pf->discard_enabled = 1;
-	pf->discard_passdown = 1;
+	pf->zero_new_blocks = true;
+	pf->discard_enabled = true;
+	pf->discard_passdown = true;
 }
 
 static void __pool_destroy(struct pool *pool)
@@ -2119,13 +2146,13 @@ static int parse_pool_features(struct dm_arg_set *as, struct pool_features *pf,
 		argc--;
 
 		if (!strcasecmp(arg_name, "skip_block_zeroing"))
-			pf->zero_new_blocks = 0;
+			pf->zero_new_blocks = false;
 
 		else if (!strcasecmp(arg_name, "ignore_discard"))
-			pf->discard_enabled = 0;
+			pf->discard_enabled = false;
 
 		else if (!strcasecmp(arg_name, "no_discard_passdown"))
-			pf->discard_passdown = 0;
+			pf->discard_passdown = false;
 
 		else if (!strcasecmp(arg_name, "read_only"))
 			pf->mode = PM_READ_ONLY;
@@ -2259,8 +2286,9 @@ static int pool_ctr(struct dm_target *ti, unsigned argc, char **argv)
 	pt->metadata_dev = metadata_dev;
 	pt->data_dev = data_dev;
 	pt->low_water_blocks = low_water_blocks;
-	pt->pf = pf;
+	pt->adjusted_pf = pt->requested_pf = pf;
 	ti->num_flush_requests = 1;
+
 	/*
 	 * Only need to enable discards if the pool should pass
 	 * them down to the data device.  The thin device's discard
@@ -2268,12 +2296,14 @@ static int pool_ctr(struct dm_target *ti, unsigned argc, char **argv)
 	 */
 	if (pf.discard_enabled && pf.discard_passdown) {
 		ti->num_discard_requests = 1;
+
 		/*
 		 * Setting 'discards_supported' circumvents the normal
 		 * stacking of discard limits (this keeps the pool and
 		 * thin devices' discard limits consistent).
 		 */
 		ti->discards_supported = true;
+		ti->discard_zeroes_data_unsupported = true;
 	}
 	ti->private = pt;
 
@@ -2703,7 +2733,7 @@ static int pool_status(struct dm_target *ti, status_type_t type,
 		       format_dev_t(buf2, pt->data_dev->bdev->bd_dev),
 		       (unsigned long)pool->sectors_per_block,
 		       (unsigned long long)pt->low_water_blocks);
-		emit_flags(&pt->pf, result, sz, maxlen);
+		emit_flags(&pt->requested_pf, result, sz, maxlen);
 		break;
 	}
 
@@ -2732,20 +2762,21 @@ static int pool_merge(struct dm_target *ti, struct bvec_merge_data *bvm,
 	return min(max_size, q->merge_bvec_fn(q, bvm, biovec));
 }
 
-static void set_discard_limits(struct pool *pool, struct queue_limits *limits)
+static void set_discard_limits(struct pool_c *pt, struct queue_limits *limits)
 {
-	/*
-	 * FIXME: these limits may be incompatible with the pool's data device
-	 */
+	struct pool *pool = pt->pool;
+	struct queue_limits *data_limits;
+
 	limits->max_discard_sectors = pool->sectors_per_block;
 
 	/*
-	 * This is just a hint, and not enforced.  We have to cope with
-	 * bios that cover a block partially.  A discard that spans a block
-	 * boundary is not sent to this target.
+	 * discard_granularity is just a hint, and not enforced.
 	 */
-	limits->discard_granularity = pool->sectors_per_block << SECTOR_SHIFT;
-	limits->discard_zeroes_data = pool->pf.zero_new_blocks;
+	if (pt->adjusted_pf.discard_passdown) {
+		data_limits = &bdev_get_queue(pt->data_dev->bdev)->limits;
+		limits->discard_granularity = data_limits->discard_granularity;
+	} else
+		limits->discard_granularity = pool->sectors_per_block << SECTOR_SHIFT;
 }
 
 static void pool_io_hints(struct dm_target *ti, struct queue_limits *limits)
@@ -2755,15 +2786,25 @@ static void pool_io_hints(struct dm_target *ti, struct queue_limits *limits)
 
 	blk_limits_io_min(limits, 0);
 	blk_limits_io_opt(limits, pool->sectors_per_block << SECTOR_SHIFT);
-	if (pool->pf.discard_enabled)
-		set_discard_limits(pool, limits);
+
+	/*
+	 * pt->adjusted_pf is a staging area for the actual features to use.
+	 * They get transferred to the live pool in bind_control_target()
+	 * called from pool_preresume().
+	 */
+	if (!pt->adjusted_pf.discard_enabled)
+		return;
+
+	disable_passdown_if_not_supported(pt);
+
+	set_discard_limits(pt, limits);
 }
 
 static struct target_type pool_target = {
 	.name = "thin-pool",
 	.features = DM_TARGET_SINGLETON | DM_TARGET_ALWAYS_WRITEABLE |
 		    DM_TARGET_IMMUTABLE,
-	.version = {1, 3, 0},
+	.version = {1, 4, 0},
 	.module = THIS_MODULE,
 	.ctr = pool_ctr,
 	.dtr = pool_dtr,
@@ -3042,19 +3083,19 @@ static int thin_iterate_devices(struct dm_target *ti,
 	return 0;
 }
 
+/*
+ * A thin device always inherits its queue limits from its pool.
+ */
 static void thin_io_hints(struct dm_target *ti, struct queue_limits *limits)
 {
 	struct thin_c *tc = ti->private;
-	struct pool *pool = tc->pool;
 
-	blk_limits_io_min(limits, 0);
-	blk_limits_io_opt(limits, pool->sectors_per_block << SECTOR_SHIFT);
-	set_discard_limits(pool, limits);
+	*limits = bdev_get_queue(tc->pool_dev->bdev)->limits;
 }
 
 static struct target_type thin_target = {
 	.name = "thin",
-	.version = {1, 3, 0},
+	.version = {1, 4, 0},
 	.module	= THIS_MODULE,
 	.ctr = thin_ctr,
 	.dtr = thin_dtr,

drivers/md/dm-verity.c

@@ -718,8 +718,8 @@ static int verity_ctr(struct dm_target *ti, unsigned argc, char **argv)
 	v->hash_dev_block_bits = ffs(num) - 1;
 
 	if (sscanf(argv[5], "%llu%c", &num_ll, &dummy) != 1 ||
-	    num_ll << (v->data_dev_block_bits - SECTOR_SHIFT) !=
-	    (sector_t)num_ll << (v->data_dev_block_bits - SECTOR_SHIFT)) {
+	    (sector_t)(num_ll << (v->data_dev_block_bits - SECTOR_SHIFT))
+	    >> (v->data_dev_block_bits - SECTOR_SHIFT) != num_ll) {
 		ti->error = "Invalid data blocks";
 		r = -EINVAL;
 		goto bad;
@@ -733,8 +733,8 @@ static int verity_ctr(struct dm_target *ti, unsigned argc, char **argv)
 	}
 
 	if (sscanf(argv[6], "%llu%c", &num_ll, &dummy) != 1 ||
-	    num_ll << (v->hash_dev_block_bits - SECTOR_SHIFT) !=
-	    (sector_t)num_ll << (v->hash_dev_block_bits - SECTOR_SHIFT)) {
+	    (sector_t)(num_ll << (v->hash_dev_block_bits - SECTOR_SHIFT))
+	    >> (v->hash_dev_block_bits - SECTOR_SHIFT) != num_ll) {
 		ti->error = "Invalid hash start";
 		r = -EINVAL;
 		goto bad;

drivers/md/dm.c

@@ -865,10 +865,14 @@ static void dm_done(struct request *clone, int error, bool mapped)
 {
 	int r = error;
 	struct dm_rq_target_io *tio = clone->end_io_data;
-	dm_request_endio_fn rq_end_io = tio->ti->type->rq_end_io;
+	dm_request_endio_fn rq_end_io = NULL;
 
-	if (mapped && rq_end_io)
-		r = rq_end_io(tio->ti, clone, error, &tio->info);
+	if (tio->ti) {
+		rq_end_io = tio->ti->type->rq_end_io;
+
+		if (mapped && rq_end_io)
+			r = rq_end_io(tio->ti, clone, error, &tio->info);
+	}
 
 	if (r <= 0)
 		/* The target wants to complete the I/O */
@@ -1588,15 +1592,6 @@ static int map_request(struct dm_target *ti, struct request *clone,
 	int r, requeued = 0;
 	struct dm_rq_target_io *tio = clone->end_io_data;
 
-	/*
-	 * Hold the md reference here for the in-flight I/O.
-	 * We can't rely on the reference count by device opener,
-	 * because the device may be closed during the request completion
-	 * when all bios are completed.
-	 * See the comment in rq_completed() too.
-	 */
-	dm_get(md);
-
 	tio->ti = ti;
 	r = ti->type->map_rq(ti, clone, &tio->info);
 	switch (r) {
@@ -1628,6 +1623,26 @@ static int map_request(struct dm_target *ti, struct request *clone,
 	return requeued;
 }
 
+static struct request *dm_start_request(struct mapped_device *md, struct request *orig)
+{
+	struct request *clone;
+
+	blk_start_request(orig);
+	clone = orig->special;
+	atomic_inc(&md->pending[rq_data_dir(clone)]);
+
+	/*
+	 * Hold the md reference here for the in-flight I/O.
+	 * We can't rely on the reference count by device opener,
+	 * because the device may be closed during the request completion
+	 * when all bios are completed.
+	 * See the comment in rq_completed() too.
+	 */
+	dm_get(md);
+
+	return clone;
+}
+
 /*
  * q->request_fn for request-based dm.
  * Called with the queue lock held.
@@ -1657,14 +1672,21 @@ static void dm_request_fn(struct request_queue *q)
 			pos = blk_rq_pos(rq);
 
 		ti = dm_table_find_target(map, pos);
-		BUG_ON(!dm_target_is_valid(ti));
+		if (!dm_target_is_valid(ti)) {
+			/*
+			 * Must perform setup, that dm_done() requires,
+			 * before calling dm_kill_unmapped_request
+			 */
+			DMERR_LIMIT("request attempted access beyond the end of device");
+			clone = dm_start_request(md, rq);
+			dm_kill_unmapped_request(clone, -EIO);
+			continue;
+		}
 
 		if (ti->type->busy && ti->type->busy(ti))
 			goto delay_and_out;
 
-		blk_start_request(rq);
-		clone = rq->special;
-		atomic_inc(&md->pending[rq_data_dir(clone)]);
+		clone = dm_start_request(md, rq);
 
 		spin_unlock(q->queue_lock);
 		if (map_request(ti, clone, md))
@@ -1684,8 +1706,6 @@ static void dm_request_fn(struct request_queue *q)
 	blk_delay_queue(q, HZ / 10);
 out:
 	dm_table_put(map);
-
-	return;
 }
 
 int dm_underlying_device_busy(struct request_queue *q)
@@ -2409,7 +2429,7 @@ static void dm_queue_flush(struct mapped_device *md)
  */
 struct dm_table *dm_swap_table(struct mapped_device *md, struct dm_table *table)
 {
-	struct dm_table *map = ERR_PTR(-EINVAL);
+	struct dm_table *live_map, *map = ERR_PTR(-EINVAL);
 	struct queue_limits limits;
 	int r;
 
@@ -2419,6 +2439,19 @@ struct dm_table *dm_swap_table(struct mapped_device *md, struct dm_table *table)
 	if (!dm_suspended_md(md))
 		goto out;
 
+	/*
+	 * If the new table has no data devices, retain the existing limits.
+	 * This helps multipath with queue_if_no_path if all paths disappear,
+	 * then new I/O is queued based on these limits, and then some paths
+	 * reappear.
+	 */
+	if (dm_table_has_no_data_devices(table)) {
+		live_map = dm_get_live_table(md);
+		if (live_map)
+			limits = md->queue->limits;
+		dm_table_put(live_map);
+	}
+
 	r = dm_calculate_queue_limits(table, &limits);
 	if (r) {
 		map = ERR_PTR(r);

drivers/md/dm.h

@@ -54,6 +54,7 @@ void dm_table_event_callback(struct dm_table *t,
 			     void (*fn)(void *), void *context);
 struct dm_target *dm_table_get_target(struct dm_table *t, unsigned int index);
 struct dm_target *dm_table_find_target(struct dm_table *t, sector_t sector);
+bool dm_table_has_no_data_devices(struct dm_table *table);
 int dm_calculate_queue_limits(struct dm_table *table,
 			      struct queue_limits *limits);
 void dm_table_set_restrictions(struct dm_table *t, struct request_queue *q,