Merge tag 'dm-4.11-changes' of git://git.kernel.org/pub/scm/linux/kernel/git/device-mapper/linux-dm

Pull device mapper updates from Mike Snitzer:

 - Fix dm-raid transient device failure processing and other smaller
   tweaks.

 - Add journal support to the DM raid target to close the 'write hole'
   on raid 4/5/6.

 - Fix dm-cache corruption, due to rounding bug, when cache exceeds 2TB.

 - Add 'metadata2' feature to dm-cache to separate the dirty bitset out
   from other cache metadata. This improves speed of shutting down a
   large cache device (which implies writing out dirty bits).

 - Fix a memory leak during dm-stats data structure destruction.

 - Fix a DM multipath round-robin path selector performance regression
   that was caused by less precise balancing across all paths.

 - Lastly, introduce a DM core fix for a long-standing DM snapshot
   deadlock that is rooted in the complexity of the device stack used in
   conjunction with block core maintaining bios on current->bio_list to
   manage recursion in generic_make_request(). A more comprehensive fix
   to block core (and its hook in the cpu scheduler) would be wonderful
   but this DM-specific fix is pragmatic considering how difficult it
   has been to make progress on a generic fix.

* tag 'dm-4.11-changes' of git://git.kernel.org/pub/scm/linux/kernel/git/device-mapper/linux-dm: (22 commits)
  dm: flush queued bios when process blocks to avoid deadlock
  dm round robin: revert "use percpu 'repeat_count' and 'current_path'"
  dm stats: fix a leaked s->histogram_boundaries array
  dm space map metadata: constify dm_space_map structures
  dm cache metadata: use cursor api in blocks_are_clean_separate_dirty()
  dm persistent data: add cursor skip functions to the cursor APIs
  dm cache metadata: use dm_bitset_new() to create the dirty bitset in format 2
  dm bitset: add dm_bitset_new()
  dm cache metadata: name the cache block that couldn't be loaded
  dm cache metadata: add "metadata2" feature
  dm cache metadata: use bitset cursor api to load discard bitset
  dm bitset: introduce cursor api
  dm btree: use GFP_NOFS in dm_btree_del()
  dm space map common: memcpy the disk root to ensure it's arch aligned
  dm block manager: add unlikely() annotations on dm_bufio error paths
  dm cache: fix corruption seen when using cache > 2TB
  dm raid: cleanup awkward branching in raid_message() option processing
  dm raid: use mddev rather than rdev->mddev
  dm raid: use read_disk_sb() throughout
  dm raid: add raid4/5/6 journaling support
  ...

9 files changed, 242 insertions, 12 deletions

diff --git a/drivers/md/persistent-data/dm-array.c b/drivers/md/persistent-data/dm-array.c
index 7938cd21fa4c..185dc60360b5 100644
--- a/drivers/md/persistent-data/dm-array.c
+++ b/drivers/md/persistent-data/dm-array.c
@@ -976,6 +976,27 @@ int dm_array_cursor_next(struct dm_array_cursor *c)
 }
 EXPORT_SYMBOL_GPL(dm_array_cursor_next);
 
+int dm_array_cursor_skip(struct dm_array_cursor *c, uint32_t count)
+{
+	int r;
+
+	do {
+		uint32_t remaining = le32_to_cpu(c->ab->nr_entries) - c->index;
+
+		if (count < remaining) {
+			c->index += count;
+			return 0;
+		}
+
+		count -= remaining;
+		r = dm_array_cursor_next(c);
+
+	} while (!r);
+
+	return r;
+}
+EXPORT_SYMBOL_GPL(dm_array_cursor_skip);
+
 void dm_array_cursor_get_value(struct dm_array_cursor *c, void **value_le)
 {
 	*value_le = element_at(c->info, c->ab, c->index);
diff --git a/drivers/md/persistent-data/dm-array.h b/drivers/md/persistent-data/dm-array.h
index 27ee49a55473..d7d2d579c662 100644
--- a/drivers/md/persistent-data/dm-array.h
+++ b/drivers/md/persistent-data/dm-array.h
@@ -207,6 +207,7 @@ void dm_array_cursor_end(struct dm_array_cursor *c);
 
 uint32_t dm_array_cursor_index(struct dm_array_cursor *c);
 int dm_array_cursor_next(struct dm_array_cursor *c);
+int dm_array_cursor_skip(struct dm_array_cursor *c, uint32_t count);
 
 /*
  * value_le is only valid while the cursor points at the current value.
diff --git a/drivers/md/persistent-data/dm-bitset.c b/drivers/md/persistent-data/dm-bitset.c
index 36f7cc2c7109..b7208d82e748 100644
--- a/drivers/md/persistent-data/dm-bitset.c
+++ b/drivers/md/persistent-data/dm-bitset.c
@@ -39,6 +39,48 @@ int dm_bitset_empty(struct dm_disk_bitset *info, dm_block_t *root)
 }
 EXPORT_SYMBOL_GPL(dm_bitset_empty);
 
+struct packer_context {
+	bit_value_fn fn;
+	unsigned nr_bits;
+	void *context;
+};
+
+static int pack_bits(uint32_t index, void *value, void *context)
+{
+	int r;
+	struct packer_context *p = context;
+	unsigned bit, nr = min(64u, p->nr_bits - (index * 64));
+	uint64_t word = 0;
+	bool bv;
+
+	for (bit = 0; bit < nr; bit++) {
+		r = p->fn(index * 64 + bit, &bv, p->context);
+		if (r)
+			return r;
+
+		if (bv)
+			set_bit(bit, (unsigned long *) &word);
+		else
+			clear_bit(bit, (unsigned long *) &word);
+	}
+
+	*((__le64 *) value) = cpu_to_le64(word);
+
+	return 0;
+}
+
+int dm_bitset_new(struct dm_disk_bitset *info, dm_block_t *root,
+		  uint32_t size, bit_value_fn fn, void *context)
+{
+	struct packer_context p;
+	p.fn = fn;
+	p.nr_bits = size;
+	p.context = context;
+
+	return dm_array_new(&info->array_info, root, dm_div_up(size, 64), pack_bits, &p);
+}
+EXPORT_SYMBOL_GPL(dm_bitset_new);
+
 int dm_bitset_resize(struct dm_disk_bitset *info, dm_block_t root,
 		     uint32_t old_nr_entries, uint32_t new_nr_entries,
 		     bool default_value, dm_block_t *new_root)
@@ -168,4 +210,108 @@ int dm_bitset_test_bit(struct dm_disk_bitset *info, dm_block_t root,
 }
 EXPORT_SYMBOL_GPL(dm_bitset_test_bit);
 
+static int cursor_next_array_entry(struct dm_bitset_cursor *c)
+{
+	int r;
+	__le64 *value;
+
+	r = dm_array_cursor_next(&c->cursor);
+	if (r)
+		return r;
+
+	dm_array_cursor_get_value(&c->cursor, (void **) &value);
+	c->array_index++;
+	c->bit_index = 0;
+	c->current_bits = le64_to_cpu(*value);
+	return 0;
+}
+
+int dm_bitset_cursor_begin(struct dm_disk_bitset *info,
+			   dm_block_t root, uint32_t nr_entries,
+			   struct dm_bitset_cursor *c)
+{
+	int r;
+	__le64 *value;
+
+	if (!nr_entries)
+		return -ENODATA;
+
+	c->info = info;
+	c->entries_remaining = nr_entries;
+
+	r = dm_array_cursor_begin(&info->array_info, root, &c->cursor);
+	if (r)
+		return r;
+
+	dm_array_cursor_get_value(&c->cursor, (void **) &value);
+	c->array_index = 0;
+	c->bit_index = 0;
+	c->current_bits = le64_to_cpu(*value);
+
+	return r;
+}
+EXPORT_SYMBOL_GPL(dm_bitset_cursor_begin);
+
+void dm_bitset_cursor_end(struct dm_bitset_cursor *c)
+{
+	return dm_array_cursor_end(&c->cursor);
+}
+EXPORT_SYMBOL_GPL(dm_bitset_cursor_end);
+
+int dm_bitset_cursor_next(struct dm_bitset_cursor *c)
+{
+	int r = 0;
+
+	if (!c->entries_remaining)
+		return -ENODATA;
+
+	c->entries_remaining--;
+	if (++c->bit_index > 63)
+		r = cursor_next_array_entry(c);
+
+	return r;
+}
+EXPORT_SYMBOL_GPL(dm_bitset_cursor_next);
+
+int dm_bitset_cursor_skip(struct dm_bitset_cursor *c, uint32_t count)
+{
+	int r;
+	__le64 *value;
+	uint32_t nr_array_skip;
+	uint32_t remaining_in_word = 64 - c->bit_index;
+
+	if (c->entries_remaining < count)
+		return -ENODATA;
+
+	if (count < remaining_in_word) {
+		c->bit_index += count;
+		c->entries_remaining -= count;
+		return 0;
+
+	} else {
+		c->entries_remaining -= remaining_in_word;
+		count -= remaining_in_word;
+	}
+
+	nr_array_skip = (count / 64) + 1;
+	r = dm_array_cursor_skip(&c->cursor, nr_array_skip);
+	if (r)
+		return r;
+
+	dm_array_cursor_get_value(&c->cursor, (void **) &value);
+	c->entries_remaining -= count;
+	c->array_index += nr_array_skip;
+	c->bit_index = count & 63;
+	c->current_bits = le64_to_cpu(*value);
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(dm_bitset_cursor_skip);
+
+bool dm_bitset_cursor_get_value(struct dm_bitset_cursor *c)
+{
+	return test_bit(c->bit_index, (unsigned long *) &c->current_bits);
+}
+EXPORT_SYMBOL_GPL(dm_bitset_cursor_get_value);
+
 /*----------------------------------------------------------------*/
diff --git a/drivers/md/persistent-data/dm-bitset.h b/drivers/md/persistent-data/dm-bitset.h
index c2287d672ef5..df888da04ee1 100644
--- a/drivers/md/persistent-data/dm-bitset.h
+++ b/drivers/md/persistent-data/dm-bitset.h
@@ -93,6 +93,22 @@ void dm_disk_bitset_init(struct dm_transaction_manager *tm,
 int dm_bitset_empty(struct dm_disk_bitset *info, dm_block_t *new_root);
 
 /*
+ * Creates a new bitset populated with values provided by a callback
+ * function.  This is more efficient than creating an empty bitset,
+ * resizing, and then setting values since that process incurs a lot of
+ * copying.
+ *
+ * info - describes the array
+ * root - the root block of the array on disk
+ * size - the number of entries in the array
+ * fn - the callback
+ * context - passed to the callback
+ */
+typedef int (*bit_value_fn)(uint32_t index, bool *value, void *context);
+int dm_bitset_new(struct dm_disk_bitset *info, dm_block_t *root,
+		  uint32_t size, bit_value_fn fn, void *context);
+
+/*
  * Resize the bitset.
  *
  * info - describes the bitset
@@ -161,6 +177,29 @@ int dm_bitset_test_bit(struct dm_disk_bitset *info, dm_block_t root,
 int dm_bitset_flush(struct dm_disk_bitset *info, dm_block_t root,
 		    dm_block_t *new_root);
 
+struct dm_bitset_cursor {
+	struct dm_disk_bitset *info;
+	struct dm_array_cursor cursor;
+
+	uint32_t entries_remaining;
+	uint32_t array_index;
+	uint32_t bit_index;
+	uint64_t current_bits;
+};
+
+/*
+ * Make sure you've flush any dm_disk_bitset and updated the root before
+ * using this.
+ */
+int dm_bitset_cursor_begin(struct dm_disk_bitset *info,
+			   dm_block_t root, uint32_t nr_entries,
+			   struct dm_bitset_cursor *c);
+void dm_bitset_cursor_end(struct dm_bitset_cursor *c);
+
+int dm_bitset_cursor_next(struct dm_bitset_cursor *c);
+int dm_bitset_cursor_skip(struct dm_bitset_cursor *c, uint32_t count);
+bool dm_bitset_cursor_get_value(struct dm_bitset_cursor *c);
+
 /*----------------------------------------------------------------*/
 
 #endif /* _LINUX_DM_BITSET_H */
diff --git a/drivers/md/persistent-data/dm-block-manager.c b/drivers/md/persistent-data/dm-block-manager.c
index 758d90cc2733..0863905dee02 100644
--- a/drivers/md/persistent-data/dm-block-manager.c
+++ b/drivers/md/persistent-data/dm-block-manager.c
@@ -462,7 +462,7 @@ int dm_bm_read_lock(struct dm_block_manager *bm, dm_block_t b,
 	int r;
 
 	p = dm_bufio_read(bm->bufio, b, (struct dm_buffer **) result);
-	if (IS_ERR(p))
+	if (unlikely(IS_ERR(p)))
 		return PTR_ERR(p);
 
 	aux = dm_bufio_get_aux_data(to_buffer(*result));
@@ -498,7 +498,7 @@ int dm_bm_write_lock(struct dm_block_manager *bm,
 		return -EPERM;
 
 	p = dm_bufio_read(bm->bufio, b, (struct dm_buffer **) result);
-	if (IS_ERR(p))
+	if (unlikely(IS_ERR(p)))
 		return PTR_ERR(p);
 
 	aux = dm_bufio_get_aux_data(to_buffer(*result));
@@ -531,7 +531,7 @@ int dm_bm_read_try_lock(struct dm_block_manager *bm,
 	int r;
 
 	p = dm_bufio_get(bm->bufio, b, (struct dm_buffer **) result);
-	if (IS_ERR(p))
+	if (unlikely(IS_ERR(p)))
 		return PTR_ERR(p);
 	if (unlikely(!p))
 		return -EWOULDBLOCK;
@@ -567,7 +567,7 @@ int dm_bm_write_lock_zero(struct dm_block_manager *bm,
 		return -EPERM;
 
 	p = dm_bufio_new(bm->bufio, b, (struct dm_buffer **) result);
-	if (IS_ERR(p))
+	if (unlikely(IS_ERR(p)))
 		return PTR_ERR(p);
 
 	memset(p, 0, dm_bm_block_size(bm));
diff --git a/drivers/md/persistent-data/dm-btree.c b/drivers/md/persistent-data/dm-btree.c
index 20a40329d84a..02e2ee0d8a00 100644
--- a/drivers/md/persistent-data/dm-btree.c
+++ b/drivers/md/persistent-data/dm-btree.c
@@ -272,7 +272,12 @@ int dm_btree_del(struct dm_btree_info *info, dm_block_t root)
 	int r;
 	struct del_stack *s;
 
-	s = kmalloc(sizeof(*s), GFP_NOIO);
+	/*
+	 * dm_btree_del() is called via an ioctl, as such should be
+	 * considered an FS op.  We can't recurse back into the FS, so we
+	 * allocate GFP_NOFS.
+	 */
+	s = kmalloc(sizeof(*s), GFP_NOFS);
 	if (!s)
 		return -ENOMEM;
 	s->info = info;
@@ -1139,6 +1144,17 @@ int dm_btree_cursor_next(struct dm_btree_cursor *c)
 }
 EXPORT_SYMBOL_GPL(dm_btree_cursor_next);
 
+int dm_btree_cursor_skip(struct dm_btree_cursor *c, uint32_t count)
+{
+	int r = 0;
+
+	while (count-- && !r)
+		r = dm_btree_cursor_next(c);
+
+	return r;
+}
+EXPORT_SYMBOL_GPL(dm_btree_cursor_skip);
+
 int dm_btree_cursor_get_value(struct dm_btree_cursor *c, uint64_t *key, void *value_le)
 {
 	if (c->depth) {
diff --git a/drivers/md/persistent-data/dm-btree.h b/drivers/md/persistent-data/dm-btree.h
index db9bd26adf31..3dc5bb1a4748 100644
--- a/drivers/md/persistent-data/dm-btree.h
+++ b/drivers/md/persistent-data/dm-btree.h
@@ -209,6 +209,7 @@ int dm_btree_cursor_begin(struct dm_btree_info *info, dm_block_t root,
 			  bool prefetch_leaves, struct dm_btree_cursor *c);
 void dm_btree_cursor_end(struct dm_btree_cursor *c);
 int dm_btree_cursor_next(struct dm_btree_cursor *c);
+int dm_btree_cursor_skip(struct dm_btree_cursor *c, uint32_t count);
 int dm_btree_cursor_get_value(struct dm_btree_cursor *c, uint64_t *key, void *value_le);
 
 #endif	/* _LINUX_DM_BTREE_H */
diff --git a/drivers/md/persistent-data/dm-space-map-common.c b/drivers/md/persistent-data/dm-space-map-common.c
index 4c28608a0c94..829b4ce057d8 100644
--- a/drivers/md/persistent-data/dm-space-map-common.c
+++ b/drivers/md/persistent-data/dm-space-map-common.c
@@ -626,13 +626,19 @@ int sm_ll_open_metadata(struct ll_disk *ll, struct dm_transaction_manager *tm,
 			void *root_le, size_t len)
 {
 	int r;
-	struct disk_sm_root *smr = root_le;
+	struct disk_sm_root smr;
 
 	if (len < sizeof(struct disk_sm_root)) {
 		DMERR("sm_metadata root too small");
 		return -ENOMEM;
 	}
 
+	/*
+	 * We don't know the alignment of the root_le buffer, so need to
+	 * copy into a new structure.
+	 */
+	memcpy(&smr, root_le, sizeof(smr));
+
 	r = sm_ll_init(ll, tm);
 	if (r < 0)
 		return r;
@@ -644,10 +650,10 @@ int sm_ll_open_metadata(struct ll_disk *ll, struct dm_transaction_manager *tm,
 	ll->max_entries = metadata_ll_max_entries;
 	ll->commit = metadata_ll_commit;
 
-	ll->nr_blocks = le64_to_cpu(smr->nr_blocks);
-	ll->nr_allocated = le64_to_cpu(smr->nr_allocated);
-	ll->bitmap_root = le64_to_cpu(smr->bitmap_root);
-	ll->ref_count_root = le64_to_cpu(smr->ref_count_root);
+	ll->nr_blocks = le64_to_cpu(smr.nr_blocks);
+	ll->nr_allocated = le64_to_cpu(smr.nr_allocated);
+	ll->bitmap_root = le64_to_cpu(smr.bitmap_root);
+	ll->ref_count_root = le64_to_cpu(smr.ref_count_root);
 
 	return ll->open_index(ll);
 }
diff --git a/drivers/md/persistent-data/dm-space-map-metadata.c b/drivers/md/persistent-data/dm-space-map-metadata.c
index 20557e2c60c6..4aed69d9dd17 100644
--- a/drivers/md/persistent-data/dm-space-map-metadata.c
+++ b/drivers/md/persistent-data/dm-space-map-metadata.c
@@ -544,7 +544,7 @@ static int sm_metadata_copy_root(struct dm_space_map *sm, void *where_le, size_t
 
 static int sm_metadata_extend(struct dm_space_map *sm, dm_block_t extra_blocks);
 
-static struct dm_space_map ops = {
+static const struct dm_space_map ops = {
 	.destroy = sm_metadata_destroy,
 	.extend = sm_metadata_extend,
 	.get_nr_blocks = sm_metadata_get_nr_blocks,
@@ -671,7 +671,7 @@ static int sm_bootstrap_copy_root(struct dm_space_map *sm, void *where,
 	return -EINVAL;
 }
 
-static struct dm_space_map bootstrap_ops = {
+static const struct dm_space_map bootstrap_ops = {
 	.destroy = sm_bootstrap_destroy,
 	.extend = sm_bootstrap_extend,
 	.get_nr_blocks = sm_bootstrap_get_nr_blocks,