Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs

Pull btrfs fixes from Chris Mason:
 "This fixes some lockups in btrfs reported with rc1.  It probably has
  some performance impact because it is backing off our spinning locks
  more often and switching to a blocking lock.  I'll be able to nail
  that down next week, but for now I want to get the lockups taken care
  of.

  Otherwise some more stack reduction and assorted fixes"

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs:
  Btrfs: fix wrong error handle when the device is missing or is not writeable
  Btrfs: fix deadlock when mounting a degraded fs
  Btrfs: use bio_endio_nodec instead of open code
  Btrfs: fix NULL pointer crash when running balance and scrub concurrently
  btrfs: Skip scrubbing removed chunks to avoid -ENOENT.
  Btrfs: fix broken free space cache after the system crashed
  Btrfs: make free space cache write out functions more readable
  Btrfs: remove unused wait queue in struct extent_buffer
  Btrfs: fix deadlocks with trylock on tree nodes

fs/btrfs/ctree.h

@@ -1259,11 +1259,19 @@ struct btrfs_block_group_cache {
 	spinlock_t lock;
 	u64 pinned;
 	u64 reserved;
+	u64 delalloc_bytes;
 	u64 bytes_super;
 	u64 flags;
 	u64 sectorsize;
 	u64 cache_generation;
 
+	/*
+	 * It is just used for the delayed data space allocation because
+	 * only the data space allocation and the relative metadata update
+	 * can be done cross the transaction.
+	 */
+	struct rw_semaphore data_rwsem;
+
 	/* for raid56, this is a full stripe, without parity */
 	unsigned long full_stripe_len;
 
@@ -3316,7 +3324,7 @@ int btrfs_alloc_logged_file_extent(struct btrfs_trans_handle *trans,
 				   struct btrfs_key *ins);
 int btrfs_reserve_extent(struct btrfs_root *root, u64 num_bytes,
 			 u64 min_alloc_size, u64 empty_size, u64 hint_byte,
-			 struct btrfs_key *ins, int is_data);
+			 struct btrfs_key *ins, int is_data, int delalloc);
 int btrfs_inc_ref(struct btrfs_trans_handle *trans, struct btrfs_root *root,
 		  struct extent_buffer *buf, int full_backref, int no_quota);
 int btrfs_dec_ref(struct btrfs_trans_handle *trans, struct btrfs_root *root,
@@ -3330,7 +3338,8 @@ int btrfs_free_extent(struct btrfs_trans_handle *trans,
 		      u64 bytenr, u64 num_bytes, u64 parent, u64 root_objectid,
 		      u64 owner, u64 offset, int no_quota);
 
-int btrfs_free_reserved_extent(struct btrfs_root *root, u64 start, u64 len);
+int btrfs_free_reserved_extent(struct btrfs_root *root, u64 start, u64 len,
+			       int delalloc);
 int btrfs_free_and_pin_reserved_extent(struct btrfs_root *root,
 				       u64 start, u64 len);
 void btrfs_prepare_extent_commit(struct btrfs_trans_handle *trans,

fs/btrfs/extent-tree.c

@@ -105,7 +105,8 @@ static int find_next_key(struct btrfs_path *path, int level,
 static void dump_space_info(struct btrfs_space_info *info, u64 bytes,
 			    int dump_block_groups);
 static int btrfs_update_reserved_bytes(struct btrfs_block_group_cache *cache,
-				       u64 num_bytes, int reserve);
+				       u64 num_bytes, int reserve,
+				       int delalloc);
 static int block_rsv_use_bytes(struct btrfs_block_rsv *block_rsv,
 			       u64 num_bytes);
 int btrfs_pin_extent(struct btrfs_root *root,
@@ -3260,7 +3261,8 @@ static int cache_save_setup(struct btrfs_block_group_cache *block_group,
 
 	spin_lock(&block_group->lock);
 	if (block_group->cached != BTRFS_CACHE_FINISHED ||
-	    !btrfs_test_opt(root, SPACE_CACHE)) {
+	    !btrfs_test_opt(root, SPACE_CACHE) ||
+	    block_group->delalloc_bytes) {
 		/*
 		 * don't bother trying to write stuff out _if_
 		 * a) we're not cached,
@@ -5613,6 +5615,7 @@ int btrfs_exclude_logged_extents(struct btrfs_root *log,
  * @cache:	The cache we are manipulating
  * @num_bytes:	The number of bytes in question
  * @reserve:	One of the reservation enums
+ * @delalloc:   The blocks are allocated for the delalloc write
  *
  * This is called by the allocator when it reserves space, or by somebody who is
  * freeing space that was never actually used on disk.  For example if you
@@ -5631,7 +5634,7 @@ int btrfs_exclude_logged_extents(struct btrfs_root *log,
  * succeeds.
  */
 static int btrfs_update_reserved_bytes(struct btrfs_block_group_cache *cache,
-				       u64 num_bytes, int reserve)
+				       u64 num_bytes, int reserve, int delalloc)
 {
 	struct btrfs_space_info *space_info = cache->space_info;
 	int ret = 0;
@@ -5650,12 +5653,18 @@ static int btrfs_update_reserved_bytes(struct btrfs_block_group_cache *cache,
 						num_bytes, 0);
 				space_info->bytes_may_use -= num_bytes;
 			}
+
+			if (delalloc)
+				cache->delalloc_bytes += num_bytes;
 		}
 	} else {
 		if (cache->ro)
 			space_info->bytes_readonly += num_bytes;
 		cache->reserved -= num_bytes;
 		space_info->bytes_reserved -= num_bytes;
+
+		if (delalloc)
+			cache->delalloc_bytes -= num_bytes;
 	}
 	spin_unlock(&cache->lock);
 	spin_unlock(&space_info->lock);
@@ -6206,7 +6215,7 @@ void btrfs_free_tree_block(struct btrfs_trans_handle *trans,
 		WARN_ON(test_bit(EXTENT_BUFFER_DIRTY, &buf->bflags));
 
 		btrfs_add_free_space(cache, buf->start, buf->len);
-		btrfs_update_reserved_bytes(cache, buf->len, RESERVE_FREE);
+		btrfs_update_reserved_bytes(cache, buf->len, RESERVE_FREE, 0);
 		trace_btrfs_reserved_extent_free(root, buf->start, buf->len);
 		pin = 0;
 	}
@@ -6365,6 +6374,70 @@ enum btrfs_loop_type {
 	LOOP_NO_EMPTY_SIZE = 3,
 };
 
+static inline void
+btrfs_lock_block_group(struct btrfs_block_group_cache *cache,
+		       int delalloc)
+{
+	if (delalloc)
+		down_read(&cache->data_rwsem);
+}
+
+static inline void
+btrfs_grab_block_group(struct btrfs_block_group_cache *cache,
+		       int delalloc)
+{
+	btrfs_get_block_group(cache);
+	if (delalloc)
+		down_read(&cache->data_rwsem);
+}
+
+static struct btrfs_block_group_cache *
+btrfs_lock_cluster(struct btrfs_block_group_cache *block_group,
+		   struct btrfs_free_cluster *cluster,
+		   int delalloc)
+{
+	struct btrfs_block_group_cache *used_bg;
+	bool locked = false;
+again:
+	spin_lock(&cluster->refill_lock);
+	if (locked) {
+		if (used_bg == cluster->block_group)
+			return used_bg;
+
+		up_read(&used_bg->data_rwsem);
+		btrfs_put_block_group(used_bg);
+	}
+
+	used_bg = cluster->block_group;
+	if (!used_bg)
+		return NULL;
+
+	if (used_bg == block_group)
+		return used_bg;
+
+	btrfs_get_block_group(used_bg);
+
+	if (!delalloc)
+		return used_bg;
+
+	if (down_read_trylock(&used_bg->data_rwsem))
+		return used_bg;
+
+	spin_unlock(&cluster->refill_lock);
+	down_read(&used_bg->data_rwsem);
+	locked = true;
+	goto again;
+}
+
+static inline void
+btrfs_release_block_group(struct btrfs_block_group_cache *cache,
+			 int delalloc)
+{
+	if (delalloc)
+		up_read(&cache->data_rwsem);
+	btrfs_put_block_group(cache);
+}
+
 /*
  * walks the btree of allocated extents and find a hole of a given size.
  * The key ins is changed to record the hole:
@@ -6379,7 +6452,7 @@ enum btrfs_loop_type {
 static noinline int find_free_extent(struct btrfs_root *orig_root,
 				     u64 num_bytes, u64 empty_size,
 				     u64 hint_byte, struct btrfs_key *ins,
-				     u64 flags)
+				     u64 flags, int delalloc)
 {
 	int ret = 0;
 	struct btrfs_root *root = orig_root->fs_info->extent_root;
@@ -6467,6 +6540,7 @@ static noinline int find_free_extent(struct btrfs_root *orig_root,
 				up_read(&space_info->groups_sem);
 			} else {
 				index = get_block_group_index(block_group);
+				btrfs_lock_block_group(block_group, delalloc);
 				goto have_block_group;
 			}
 		} else if (block_group) {
@@ -6481,7 +6555,7 @@ static noinline int find_free_extent(struct btrfs_root *orig_root,
 		u64 offset;
 		int cached;
 
-		btrfs_get_block_group(block_group);
+		btrfs_grab_block_group(block_group, delalloc);
 		search_start = block_group->key.objectid;
 
 		/*
@@ -6529,16 +6603,16 @@ static noinline int find_free_extent(struct btrfs_root *orig_root,
 			 * the refill lock keeps out other
 			 * people trying to start a new cluster
 			 */
-			spin_lock(&last_ptr->refill_lock);
-			used_block_group = last_ptr->block_group;
-			if (used_block_group != block_group &&
-			    (!used_block_group ||
-			     used_block_group->ro ||
-			     !block_group_bits(used_block_group, flags)))
+			used_block_group = btrfs_lock_cluster(block_group,
+							      last_ptr,
+							      delalloc);
+			if (!used_block_group)
 				goto refill_cluster;
 
-			if (used_block_group != block_group)
-				btrfs_get_block_group(used_block_group);
+			if (used_block_group != block_group &&
+			    (used_block_group->ro ||
+			     !block_group_bits(used_block_group, flags)))
+				goto release_cluster;
 
 			offset = btrfs_alloc_from_cluster(used_block_group,
 						last_ptr,
@@ -6552,16 +6626,15 @@ static noinline int find_free_extent(struct btrfs_root *orig_root,
 						used_block_group,
 						search_start, num_bytes);
 				if (used_block_group != block_group) {
-					btrfs_put_block_group(block_group);
+					btrfs_release_block_group(block_group,
+								  delalloc);
 					block_group = used_block_group;
 				}
 				goto checks;
 			}
 
 			WARN_ON(last_ptr->block_group != used_block_group);
-			if (used_block_group != block_group)
-				btrfs_put_block_group(used_block_group);
-refill_cluster:
+release_cluster:
 			/* If we are on LOOP_NO_EMPTY_SIZE, we can't
 			 * set up a new clusters, so lets just skip it
 			 * and let the allocator find whatever block
@@ -6578,8 +6651,10 @@ static noinline int find_free_extent(struct btrfs_root *orig_root,
 			 * succeeding in the unclustered
 			 * allocation.  */
 			if (loop >= LOOP_NO_EMPTY_SIZE &&
-			    last_ptr->block_group != block_group) {
+			    used_block_group != block_group) {
 				spin_unlock(&last_ptr->refill_lock);
+				btrfs_release_block_group(used_block_group,
+							  delalloc);
 				goto unclustered_alloc;
 			}
 
@@ -6589,6 +6664,10 @@ static noinline int find_free_extent(struct btrfs_root *orig_root,
 			 */
 			btrfs_return_cluster_to_free_space(NULL, last_ptr);
 
+			if (used_block_group != block_group)
+				btrfs_release_block_group(used_block_group,
+							  delalloc);
+refill_cluster:
 			if (loop >= LOOP_NO_EMPTY_SIZE) {
 				spin_unlock(&last_ptr->refill_lock);
 				goto unclustered_alloc;
@@ -6696,7 +6775,7 @@ static noinline int find_free_extent(struct btrfs_root *orig_root,
 		BUG_ON(offset > search_start);
 
 		ret = btrfs_update_reserved_bytes(block_group, num_bytes,
-						  alloc_type);
+						  alloc_type, delalloc);
 		if (ret == -EAGAIN) {
 			btrfs_add_free_space(block_group, offset, num_bytes);
 			goto loop;
@@ -6708,13 +6787,13 @@ static noinline int find_free_extent(struct btrfs_root *orig_root,
 
 		trace_btrfs_reserve_extent(orig_root, block_group,
 					   search_start, num_bytes);
-		btrfs_put_block_group(block_group);
+		btrfs_release_block_group(block_group, delalloc);
 		break;
 loop:
 		failed_cluster_refill = false;
 		failed_alloc = false;
 		BUG_ON(index != get_block_group_index(block_group));
-		btrfs_put_block_group(block_group);
+		btrfs_release_block_group(block_group, delalloc);
 	}
 	up_read(&space_info->groups_sem);
 
@@ -6827,7 +6906,7 @@ static void dump_space_info(struct btrfs_space_info *info, u64 bytes,
 int btrfs_reserve_extent(struct btrfs_root *root,
 			 u64 num_bytes, u64 min_alloc_size,
 			 u64 empty_size, u64 hint_byte,
-			 struct btrfs_key *ins, int is_data)
+			 struct btrfs_key *ins, int is_data, int delalloc)
 {
 	bool final_tried = false;
 	u64 flags;
@@ -6837,7 +6916,7 @@ int btrfs_reserve_extent(struct btrfs_root *root,
 again:
 	WARN_ON(num_bytes < root->sectorsize);
 	ret = find_free_extent(root, num_bytes, empty_size, hint_byte, ins,
-			       flags);
+			       flags, delalloc);
 
 	if (ret == -ENOSPC) {
 		if (!final_tried && ins->offset) {
@@ -6862,7 +6941,8 @@ int btrfs_reserve_extent(struct btrfs_root *root,
 }
 
 static int __btrfs_free_reserved_extent(struct btrfs_root *root,
-					u64 start, u64 len, int pin)
+					u64 start, u64 len,
+					int pin, int delalloc)
 {
 	struct btrfs_block_group_cache *cache;
 	int ret = 0;
@@ -6881,7 +6961,7 @@ static int __btrfs_free_reserved_extent(struct btrfs_root *root,
 		pin_down_extent(root, cache, start, len, 1);
 	else {
 		btrfs_add_free_space(cache, start, len);
-		btrfs_update_reserved_bytes(cache, len, RESERVE_FREE);
+		btrfs_update_reserved_bytes(cache, len, RESERVE_FREE, delalloc);
 	}
 	btrfs_put_block_group(cache);
 
@@ -6891,15 +6971,15 @@ static int __btrfs_free_reserved_extent(struct btrfs_root *root,
 }
 
 int btrfs_free_reserved_extent(struct btrfs_root *root,
-					u64 start, u64 len)
+			       u64 start, u64 len, int delalloc)
 {
-	return __btrfs_free_reserved_extent(root, start, len, 0);
+	return __btrfs_free_reserved_extent(root, start, len, 0, delalloc);
 }
 
 int btrfs_free_and_pin_reserved_extent(struct btrfs_root *root,
 				       u64 start, u64 len)
 {
-	return __btrfs_free_reserved_extent(root, start, len, 1);
+	return __btrfs_free_reserved_extent(root, start, len, 1, 0);
 }
 
 static int alloc_reserved_file_extent(struct btrfs_trans_handle *trans,
@@ -7114,7 +7194,7 @@ int btrfs_alloc_logged_file_extent(struct btrfs_trans_handle *trans,
 		return -EINVAL;
 
 	ret = btrfs_update_reserved_bytes(block_group, ins->offset,
-					  RESERVE_ALLOC_NO_ACCOUNT);
+					  RESERVE_ALLOC_NO_ACCOUNT, 0);
 	BUG_ON(ret); /* logic error */
 	ret = alloc_reserved_file_extent(trans, root, 0, root_objectid,
 					 0, owner, offset, ins, 1);
@@ -7256,7 +7336,7 @@ struct extent_buffer *btrfs_alloc_free_block(struct btrfs_trans_handle *trans,
 		return ERR_CAST(block_rsv);
 
 	ret = btrfs_reserve_extent(root, blocksize, blocksize,
-				   empty_size, hint, &ins, 0);
+				   empty_size, hint, &ins, 0, 0);
 	if (ret) {
 		unuse_block_rsv(root->fs_info, block_rsv, blocksize);
 		return ERR_PTR(ret);
@@ -8659,6 +8739,7 @@ btrfs_create_block_group_cache(struct btrfs_root *root, u64 start, u64 size)
 					       start);
 	atomic_set(&cache->count, 1);
 	spin_lock_init(&cache->lock);
+	init_rwsem(&cache->data_rwsem);
 	INIT_LIST_HEAD(&cache->list);
 	INIT_LIST_HEAD(&cache->cluster_list);
 	INIT_LIST_HEAD(&cache->new_bg_list);

fs/btrfs/extent_io.h

@@ -158,7 +158,6 @@ struct extent_buffer {
 	 * to unlock
 	 */
 	wait_queue_head_t read_lock_wq;
-	wait_queue_head_t lock_wq;
 	struct page *pages[INLINE_EXTENT_BUFFER_PAGES];
 #ifdef CONFIG_BTRFS_DEBUG
 	struct list_head leak_list;

fs/btrfs/extent_map.c

@@ -75,6 +75,8 @@ void free_extent_map(struct extent_map *em)
 	if (atomic_dec_and_test(&em->refs)) {
 		WARN_ON(extent_map_in_tree(em));
 		WARN_ON(!list_empty(&em->list));
+		if (test_bit(EXTENT_FLAG_FS_MAPPING, &em->flags))
+			kfree(em->bdev);
 		kmem_cache_free(extent_map_cache, em);
 	}
 }

fs/btrfs/extent_map.h

@@ -15,6 +15,7 @@
 #define EXTENT_FLAG_PREALLOC 3 /* pre-allocated extent */
 #define EXTENT_FLAG_LOGGING 4 /* Logging this extent */
 #define EXTENT_FLAG_FILLING 5 /* Filling in a preallocated extent */
+#define EXTENT_FLAG_FS_MAPPING 6 /* filesystem extent mapping type */
 
 struct extent_map {
 	struct rb_node rb_node;

fs/btrfs/free-space-cache.c

@@ -274,18 +274,32 @@ struct io_ctl {
 };
 
 static int io_ctl_init(struct io_ctl *io_ctl, struct inode *inode,
-		       struct btrfs_root *root)
+		       struct btrfs_root *root, int write)
 {
+	int num_pages;
+	int check_crcs = 0;
+
+	num_pages = (i_size_read(inode) + PAGE_CACHE_SIZE - 1) >>
+		    PAGE_CACHE_SHIFT;
+
+	if (btrfs_ino(inode) != BTRFS_FREE_INO_OBJECTID)
+		check_crcs = 1;
+
+	/* Make sure we can fit our crcs into the first page */
+	if (write && check_crcs &&
+	    (num_pages * sizeof(u32)) >= PAGE_CACHE_SIZE)
+		return -ENOSPC;
+
 	memset(io_ctl, 0, sizeof(struct io_ctl));
-	io_ctl->num_pages = (i_size_read(inode) + PAGE_CACHE_SIZE - 1) >>
-		PAGE_CACHE_SHIFT;
-	io_ctl->pages = kzalloc(sizeof(struct page *) * io_ctl->num_pages,
-				GFP_NOFS);
+
+	io_ctl->pages = kzalloc(sizeof(struct page *) * num_pages, GFP_NOFS);
 	if (!io_ctl->pages)
 		return -ENOMEM;
+
+	io_ctl->num_pages = num_pages;
 	io_ctl->root = root;
-	if (btrfs_ino(inode) != BTRFS_FREE_INO_OBJECTID)
-		io_ctl->check_crcs = 1;
+	io_ctl->check_crcs = check_crcs;
+
 	return 0;
 }
 
@@ -666,6 +680,13 @@ static int __load_free_space_cache(struct btrfs_root *root, struct inode *inode,
 	generation = btrfs_free_space_generation(leaf, header);
 	btrfs_release_path(path);
 
+	if (!BTRFS_I(inode)->generation) {
+		btrfs_info(root->fs_info,
+			   "The free space cache file (%llu) is invalid. skip it\n",
+			   offset);
+		return 0;
+	}
+
 	if (BTRFS_I(inode)->generation != generation) {
 		btrfs_err(root->fs_info,
 			"free space inode generation (%llu) "
@@ -677,7 +698,7 @@ static int __load_free_space_cache(struct btrfs_root *root, struct inode *inode,
 	if (!num_entries)
 		return 0;
 
-	ret = io_ctl_init(&io_ctl, inode, root);
+	ret = io_ctl_init(&io_ctl, inode, root, 0);
 	if (ret)
 		return ret;
 
@@ -957,19 +978,18 @@ update_cache_item(struct btrfs_trans_handle *trans,
 }
 
 static noinline_for_stack int
-add_ioctl_entries(struct btrfs_root *root,
-		  struct inode *inode,
-		  struct btrfs_block_group_cache *block_group,
-		  struct io_ctl *io_ctl,
-		  struct extent_state **cached_state,
-		  struct list_head *bitmap_list,
-		  int *entries)
+write_pinned_extent_entries(struct btrfs_root *root,
+			    struct btrfs_block_group_cache *block_group,
+			    struct io_ctl *io_ctl,
+			    int *entries)
 {
 	u64 start, extent_start, extent_end, len;
-	struct list_head *pos, *n;
 	struct extent_io_tree *unpin = NULL;
 	int ret;
 
+	if (!block_group)
+		return 0;
+
 	/*
 	 * We want to add any pinned extents to our free space cache
 	 * so we don't leak the space
@@ -979,23 +999,19 @@ add_ioctl_entries(struct btrfs_root *root,
 	 */
 	unpin = root->fs_info->pinned_extents;
 
-	if (block_group)
-		start = block_group->key.objectid;
+	start = block_group->key.objectid;
 
-	while (block_group && (start < block_group->key.objectid +
-			       block_group->key.offset)) {
+	while (start < block_group->key.objectid + block_group->key.offset) {
 		ret = find_first_extent_bit(unpin, start,
 					    &extent_start, &extent_end,
 					    EXTENT_DIRTY, NULL);
-		if (ret) {
-			ret = 0;
-			break;
-		}
+		if (ret)
+			return 0;
 
 		/* This pinned extent is out of our range */
 		if (extent_start >= block_group->key.objectid +
 		    block_group->key.offset)
-			break;
+			return 0;
 
 		extent_start = max(extent_start, start);
 		extent_end = min(block_group->key.objectid +
@@ -1005,11 +1021,20 @@ add_ioctl_entries(struct btrfs_root *root,
 		*entries += 1;
 		ret = io_ctl_add_entry(io_ctl, extent_start, len, NULL);
 		if (ret)
-			goto out_nospc;
+			return -ENOSPC;
 
 		start = extent_end;
 	}
 
+	return 0;
+}
+
+static noinline_for_stack int
+write_bitmap_entries(struct io_ctl *io_ctl, struct list_head *bitmap_list)
+{
+	struct list_head *pos, *n;
+	int ret;
+
 	/* Write out the bitmaps */
 	list_for_each_safe(pos, n, bitmap_list) {
 		struct btrfs_free_space *entry =
@@ -1017,36 +1042,24 @@ add_ioctl_entries(struct btrfs_root *root,
 
 		ret = io_ctl_add_bitmap(io_ctl, entry->bitmap);
 		if (ret)
-			goto out_nospc;
+			return -ENOSPC;
 		list_del_init(&entry->list);
 	}
 
-	/* Zero out the rest of the pages just to make sure */
-	io_ctl_zero_remaining_pages(io_ctl);
-
-	ret = btrfs_dirty_pages(root, inode, io_ctl->pages, io_ctl->num_pages,
-				0, i_size_read(inode), cached_state);
-	io_ctl_drop_pages(io_ctl);
-	unlock_extent_cached(&BTRFS_I(inode)->io_tree, 0,
-			     i_size_read(inode) - 1, cached_state, GFP_NOFS);
+	return 0;
+}
 
-	if (ret)
-		goto fail;
+static int flush_dirty_cache(struct inode *inode)
+{
+	int ret;
 
 	ret = btrfs_wait_ordered_range(inode, 0, (u64)-1);
-	if (ret) {
+	if (ret)
 		clear_extent_bit(&BTRFS_I(inode)->io_tree, 0, inode->i_size - 1,
 				 EXTENT_DIRTY | EXTENT_DELALLOC, 0, 0, NULL,
 				 GFP_NOFS);
-		goto fail;
-	}
-	return 0;
 
-fail:
-	return -1;
-
-out_nospc:
-	return -ENOSPC;
+	return ret;
 }
 
 static void noinline_for_stack
@@ -1056,6 +1069,7 @@ cleanup_write_cache_enospc(struct inode *inode,
 			   struct list_head *bitmap_list)
 {
 	struct list_head *pos, *n;
+
 	list_for_each_safe(pos, n, bitmap_list) {
 		struct btrfs_free_space *entry =
 			list_entry(pos, struct btrfs_free_space, list);
@@ -1088,64 +1102,104 @@ static int __btrfs_write_out_cache(struct btrfs_root *root, struct inode *inode,
 {
 	struct extent_state *cached_state = NULL;
 	struct io_ctl io_ctl;
-	struct list_head bitmap_list;
+	LIST_HEAD(bitmap_list);
 	int entries = 0;
 	int bitmaps = 0;
 	int ret;
-	int err = -1;
-
-	INIT_LIST_HEAD(&bitmap_list);
 
 	if (!i_size_read(inode))
 		return -1;
 
-	ret = io_ctl_init(&io_ctl, inode, root);
+	ret = io_ctl_init(&io_ctl, inode, root, 1);
 	if (ret)
 		return -1;
 
+	if (block_group && (block_group->flags & BTRFS_BLOCK_GROUP_DATA)) {
+		down_write(&block_group->data_rwsem);
+		spin_lock(&block_group->lock);
+		if (block_group->delalloc_bytes) {
+			block_group->disk_cache_state = BTRFS_DC_WRITTEN;
+			spin_unlock(&block_group->lock);
+			up_write(&block_group->data_rwsem);
+			BTRFS_I(inode)->generation = 0;
+			ret = 0;
+			goto out;
+		}
+		spin_unlock(&block_group->lock);
+	}
+
 	/* Lock all pages first so we can lock the extent safely. */
 	io_ctl_prepare_pages(&io_ctl, inode, 0);
 
 	lock_extent_bits(&BTRFS_I(inode)->io_tree, 0, i_size_read(inode) - 1,
 			 0, &cached_state);
 
-
-	/* Make sure we can fit our crcs into the first page */
-	if (io_ctl.check_crcs &&
-	    (io_ctl.num_pages * sizeof(u32)) >= PAGE_CACHE_SIZE)
-		goto out_nospc;
-
 	io_ctl_set_generation(&io_ctl, trans->transid);
 
+	/* Write out the extent entries in the free space cache */
 	ret = write_cache_extent_entries(&io_ctl, ctl,
 					 block_group, &entries, &bitmaps,
 					 &bitmap_list);
 	if (ret)
 		goto out_nospc;
 
-	ret = add_ioctl_entries(root, inode, block_group, &io_ctl,
-				&cached_state, &bitmap_list, &entries);
+	/*
+	 * Some spaces that are freed in the current transaction are pinned,
+	 * they will be added into free space cache after the transaction is
+	 * committed, we shouldn't lose them.
+	 */
+	ret = write_pinned_extent_entries(root, block_group, &io_ctl, &entries);
+	if (ret)
+		goto out_nospc;
 
-	if (ret == -ENOSPC)
+	/* At last, we write out all the bitmaps. */
+	ret = write_bitmap_entries(&io_ctl, &bitmap_list);
+	if (ret)
 		goto out_nospc;
-	else if (ret)
+
+	/* Zero out the rest of the pages just to make sure */
+	io_ctl_zero_remaining_pages(&io_ctl);
+
+	/* Everything is written out, now we dirty the pages in the file. */
+	ret = btrfs_dirty_pages(root, inode, io_ctl.pages, io_ctl.num_pages,
+				0, i_size_read(inode), &cached_state);
+	if (ret)
+		goto out_nospc;
+
+	if (block_group && (block_group->flags & BTRFS_BLOCK_GROUP_DATA))
+		up_write(&block_group->data_rwsem);
+	/*
+	 * Release the pages and unlock the extent, we will flush
+	 * them out later
+	 */
+	io_ctl_drop_pages(&io_ctl);
+
+	unlock_extent_cached(&BTRFS_I(inode)->io_tree, 0,
+			     i_size_read(inode) - 1, &cached_state, GFP_NOFS);
+
+	/* Flush the dirty pages in the cache file. */
+	ret = flush_dirty_cache(inode);
+	if (ret)
 		goto out;
 
-	err = update_cache_item(trans, root, inode, path, offset,
+	/* Update the cache item to tell everyone this cache file is valid. */
+	ret = update_cache_item(trans, root, inode, path, offset,
 				entries, bitmaps);
-
 out:
 	io_ctl_free(&io_ctl);
-	if (err) {
+	if (ret) {
 		invalidate_inode_pages2(inode->i_mapping);
 		BTRFS_I(inode)->generation = 0;
 	}
 	btrfs_update_inode(trans, root, inode);
-	return err;
+	return ret;
 
 out_nospc:
-
 	cleanup_write_cache_enospc(inode, &io_ctl, &cached_state, &bitmap_list);
+
+	if (block_group && (block_group->flags & BTRFS_BLOCK_GROUP_DATA))
+		up_write(&block_group->data_rwsem);
+
 	goto out;
 }
 
@@ -1165,6 +1219,12 @@ int btrfs_write_out_cache(struct btrfs_root *root,
 		spin_unlock(&block_group->lock);
 		return 0;
 	}
+
+	if (block_group->delalloc_bytes) {
+		block_group->disk_cache_state = BTRFS_DC_WRITTEN;
+		spin_unlock(&block_group->lock);
+		return 0;
+	}
 	spin_unlock(&block_group->lock);
 
 	inode = lookup_free_space_inode(root, block_group, path);

fs/btrfs/inode.c

@@ -693,7 +693,7 @@ static noinline int submit_compressed_extents(struct inode *inode,
 		ret = btrfs_reserve_extent(root,
 					   async_extent->compressed_size,
 					   async_extent->compressed_size,
-					   0, alloc_hint, &ins, 1);
+					   0, alloc_hint, &ins, 1, 1);
 		if (ret) {
 			int i;
 
@@ -794,7 +794,7 @@ static noinline int submit_compressed_extents(struct inode *inode,
 out:
 	return ret;
 out_free_reserve:
-	btrfs_free_reserved_extent(root, ins.objectid, ins.offset);
+	btrfs_free_reserved_extent(root, ins.objectid, ins.offset, 1);
 out_free:
 	extent_clear_unlock_delalloc(inode, async_extent->start,
 				     async_extent->start +
@@ -917,7 +917,7 @@ static noinline int cow_file_range(struct inode *inode,
 		cur_alloc_size = disk_num_bytes;
 		ret = btrfs_reserve_extent(root, cur_alloc_size,
 					   root->sectorsize, 0, alloc_hint,
-					   &ins, 1);
+					   &ins, 1, 1);
 		if (ret < 0)
 			goto out_unlock;
 
@@ -995,7 +995,7 @@ static noinline int cow_file_range(struct inode *inode,
 	return ret;
 
 out_reserve:
-	btrfs_free_reserved_extent(root, ins.objectid, ins.offset);
+	btrfs_free_reserved_extent(root, ins.objectid, ins.offset, 1);
 out_unlock:
 	extent_clear_unlock_delalloc(inode, start, end, locked_page,
 				     EXTENT_LOCKED | EXTENT_DO_ACCOUNTING |
@@ -2599,6 +2599,21 @@ record_old_file_extents(struct inode *inode,
 	return NULL;
 }
 
+static void btrfs_release_delalloc_bytes(struct btrfs_root *root,
+					 u64 start, u64 len)
+{
+	struct btrfs_block_group_cache *cache;
+
+	cache = btrfs_lookup_block_group(root->fs_info, start);
+	ASSERT(cache);
+
+	spin_lock(&cache->lock);
+	cache->delalloc_bytes -= len;
+	spin_unlock(&cache->lock);
+
+	btrfs_put_block_group(cache);
+}
+
 /* as ordered data IO finishes, this gets called so we can finish
  * an ordered extent if the range of bytes in the file it covers are
  * fully written.
@@ -2698,6 +2713,10 @@ static int btrfs_finish_ordered_io(struct btrfs_ordered_extent *ordered_extent)
 						logical_len, logical_len,
 						compress_type, 0, 0,
 						BTRFS_FILE_EXTENT_REG);
+		if (!ret)
+			btrfs_release_delalloc_bytes(root,
+						     ordered_extent->start,
+						     ordered_extent->disk_len);
 	}
 	unpin_extent_cache(&BTRFS_I(inode)->extent_tree,
 			   ordered_extent->file_offset, ordered_extent->len,
@@ -2750,7 +2769,7 @@ static int btrfs_finish_ordered_io(struct btrfs_ordered_extent *ordered_extent)
 		    !test_bit(BTRFS_ORDERED_NOCOW, &ordered_extent->flags) &&
 		    !test_bit(BTRFS_ORDERED_PREALLOC, &ordered_extent->flags))
 			btrfs_free_reserved_extent(root, ordered_extent->start,
-						   ordered_extent->disk_len);
+						   ordered_extent->disk_len, 1);
 	}
 
 
@@ -6535,21 +6554,21 @@ static struct extent_map *btrfs_new_extent_direct(struct inode *inode,
 
 	alloc_hint = get_extent_allocation_hint(inode, start, len);
 	ret = btrfs_reserve_extent(root, len, root->sectorsize, 0,
-				   alloc_hint, &ins, 1);
+				   alloc_hint, &ins, 1, 1);
 	if (ret)
 		return ERR_PTR(ret);
 
 	em = create_pinned_em(inode, start, ins.offset, start, ins.objectid,
 			      ins.offset, ins.offset, ins.offset, 0);
 	if (IS_ERR(em)) {
-		btrfs_free_reserved_extent(root, ins.objectid, ins.offset);
+		btrfs_free_reserved_extent(root, ins.objectid, ins.offset, 1);
 		return em;
 	}
 
 	ret = btrfs_add_ordered_extent_dio(inode, start, ins.objectid,
 					   ins.offset, ins.offset, 0);
 	if (ret) {
-		btrfs_free_reserved_extent(root, ins.objectid, ins.offset);
+		btrfs_free_reserved_extent(root, ins.objectid, ins.offset, 1);
 		free_extent_map(em);
 		return ERR_PTR(ret);
 	}
@@ -7437,7 +7456,7 @@ static void btrfs_submit_direct(int rw, struct bio *dio_bio,
 		if (!test_bit(BTRFS_ORDERED_PREALLOC, &ordered->flags) &&
 		    !test_bit(BTRFS_ORDERED_NOCOW, &ordered->flags))
 			btrfs_free_reserved_extent(root, ordered->start,
-						   ordered->disk_len);
+						   ordered->disk_len, 1);
 		btrfs_put_ordered_extent(ordered);
 		btrfs_put_ordered_extent(ordered);
 	}
@@ -8808,7 +8827,7 @@ static int __btrfs_prealloc_file_range(struct inode *inode, int mode,
 		cur_bytes = min(num_bytes, 256ULL * 1024 * 1024);
 		cur_bytes = max(cur_bytes, min_size);
 		ret = btrfs_reserve_extent(root, cur_bytes, min_size, 0,
-					   *alloc_hint, &ins, 1);
+					   *alloc_hint, &ins, 1, 0);
 		if (ret) {
 			if (own_trans)
 				btrfs_end_transaction(trans, root);
@@ -8822,7 +8841,7 @@ static int __btrfs_prealloc_file_range(struct inode *inode, int mode,
 						  BTRFS_FILE_EXTENT_PREALLOC);
 		if (ret) {
 			btrfs_free_reserved_extent(root, ins.objectid,
-						   ins.offset);
+						   ins.offset, 0);
 			btrfs_abort_transaction(trans, root, ret);
 			if (own_trans)
 				btrfs_end_transaction(trans, root);

fs/btrfs/locking.c

@@ -33,14 +33,14 @@ static void btrfs_assert_tree_read_locked(struct extent_buffer *eb);
  */
 void btrfs_set_lock_blocking_rw(struct extent_buffer *eb, int rw)
 {
-	if (eb->lock_nested) {
-		read_lock(&eb->lock);
-		if (eb->lock_nested && current->pid == eb->lock_owner) {
-			read_unlock(&eb->lock);
-			return;
-		}
-		read_unlock(&eb->lock);
-	}
+	/*
+	 * no lock is required.  The lock owner may change if
+	 * we have a read lock, but it won't change to or away
+	 * from us.  If we have the write lock, we are the owner
+	 * and it'll never change.
+	 */
+	if (eb->lock_nested && current->pid == eb->lock_owner)
+		return;
 	if (rw == BTRFS_WRITE_LOCK) {
 		if (atomic_read(&eb->blocking_writers) == 0) {
 			WARN_ON(atomic_read(&eb->spinning_writers) != 1);
@@ -65,14 +65,15 @@ void btrfs_set_lock_blocking_rw(struct extent_buffer *eb, int rw)
  */
 void btrfs_clear_lock_blocking_rw(struct extent_buffer *eb, int rw)
 {
-	if (eb->lock_nested) {
-		read_lock(&eb->lock);
-		if (eb->lock_nested && current->pid == eb->lock_owner) {
-			read_unlock(&eb->lock);
-			return;
-		}
-		read_unlock(&eb->lock);
-	}
+	/*
+	 * no lock is required.  The lock owner may change if
+	 * we have a read lock, but it won't change to or away
+	 * from us.  If we have the write lock, we are the owner
+	 * and it'll never change.
+	 */
+	if (eb->lock_nested && current->pid == eb->lock_owner)
+		return;
+
 	if (rw == BTRFS_WRITE_LOCK_BLOCKING) {
 		BUG_ON(atomic_read(&eb->blocking_writers) != 1);
 		write_lock(&eb->lock);
@@ -99,6 +100,9 @@ void btrfs_clear_lock_blocking_rw(struct extent_buffer *eb, int rw)
 void btrfs_tree_read_lock(struct extent_buffer *eb)
 {
 again:
+	BUG_ON(!atomic_read(&eb->blocking_writers) &&
+	       current->pid == eb->lock_owner);
+
 	read_lock(&eb->lock);
 	if (atomic_read(&eb->blocking_writers) &&
 	    current->pid == eb->lock_owner) {
@@ -132,7 +136,9 @@ int btrfs_try_tree_read_lock(struct extent_buffer *eb)
 	if (atomic_read(&eb->blocking_writers))
 		return 0;
 
-	read_lock(&eb->lock);
+	if (!read_trylock(&eb->lock))
+		return 0;
+
 	if (atomic_read(&eb->blocking_writers)) {
 		read_unlock(&eb->lock);
 		return 0;
@@ -151,7 +157,10 @@ int btrfs_try_tree_write_lock(struct extent_buffer *eb)
 	if (atomic_read(&eb->blocking_writers) ||
 	    atomic_read(&eb->blocking_readers))
 		return 0;
-	write_lock(&eb->lock);
+
+	if (!write_trylock(&eb->lock))
+		return 0;
+
 	if (atomic_read(&eb->blocking_writers) ||
 	    atomic_read(&eb->blocking_readers)) {
 		write_unlock(&eb->lock);
@@ -168,14 +177,15 @@ int btrfs_try_tree_write_lock(struct extent_buffer *eb)
  */
 void btrfs_tree_read_unlock(struct extent_buffer *eb)
 {
-	if (eb->lock_nested) {
-		read_lock(&eb->lock);
-		if (eb->lock_nested && current->pid == eb->lock_owner) {
-			eb->lock_nested = 0;
-			read_unlock(&eb->lock);
-			return;
-		}
-		read_unlock(&eb->lock);
+	/*
+	 * if we're nested, we have the write lock.  No new locking
+	 * is needed as long as we are the lock owner.
+	 * The write unlock will do a barrier for us, and the lock_nested
+	 * field only matters to the lock owner.
+	 */
+	if (eb->lock_nested && current->pid == eb->lock_owner) {
+		eb->lock_nested = 0;
+		return;
 	}
 	btrfs_assert_tree_read_locked(eb);
 	WARN_ON(atomic_read(&eb->spinning_readers) == 0);
@@ -189,14 +199,15 @@ void btrfs_tree_read_unlock(struct extent_buffer *eb)
  */
 void btrfs_tree_read_unlock_blocking(struct extent_buffer *eb)
 {
-	if (eb->lock_nested) {
-		read_lock(&eb->lock);
-		if (eb->lock_nested && current->pid == eb->lock_owner) {
-			eb->lock_nested = 0;
-			read_unlock(&eb->lock);
-			return;
-		}
-		read_unlock(&eb->lock);
+	/*
+	 * if we're nested, we have the write lock.  No new locking
+	 * is needed as long as we are the lock owner.
+	 * The write unlock will do a barrier for us, and the lock_nested
+	 * field only matters to the lock owner.
+	 */
+	if (eb->lock_nested && current->pid == eb->lock_owner) {
+		eb->lock_nested = 0;
+		return;
 	}
 	btrfs_assert_tree_read_locked(eb);
 	WARN_ON(atomic_read(&eb->blocking_readers) == 0);
@@ -244,6 +255,7 @@ void btrfs_tree_unlock(struct extent_buffer *eb)
 	BUG_ON(blockers > 1);
 
 	btrfs_assert_tree_locked(eb);
+	eb->lock_owner = 0;
 	atomic_dec(&eb->write_locks);
 
 	if (blockers) {

fs/btrfs/scrub.c

@@ -2725,11 +2725,8 @@ int scrub_enumerate_chunks(struct scrub_ctx *sctx,
 		dev_extent = btrfs_item_ptr(l, slot, struct btrfs_dev_extent);
 		length = btrfs_dev_extent_length(l, dev_extent);
 
-		if (found_key.offset + length <= start) {
-			key.offset = found_key.offset + length;
-			btrfs_release_path(path);
-			continue;
-		}
+		if (found_key.offset + length <= start)
+			goto skip;
 
 		chunk_tree = btrfs_dev_extent_chunk_tree(l, dev_extent);
 		chunk_objectid = btrfs_dev_extent_chunk_objectid(l, dev_extent);
@@ -2740,10 +2737,12 @@ int scrub_enumerate_chunks(struct scrub_ctx *sctx,
 		 * the chunk from going away while we scrub it
 		 */
 		cache = btrfs_lookup_block_group(fs_info, chunk_offset);
-		if (!cache) {
-			ret = -ENOENT;
-			break;
-		}
+
+		/* some chunks are removed but not committed to disk yet,
+		 * continue scrubbing */
+		if (!cache)
+			goto skip;
+
 		dev_replace->cursor_right = found_key.offset + length;
 		dev_replace->cursor_left = found_key.offset;
 		dev_replace->item_needs_writeback = 1;
@@ -2802,7 +2801,7 @@ int scrub_enumerate_chunks(struct scrub_ctx *sctx,
 
 		dev_replace->cursor_left = dev_replace->cursor_right;
 		dev_replace->item_needs_writeback = 1;
-
+skip:
 		key.offset = found_key.offset + length;
 		btrfs_release_path(path);
 	}

fs/btrfs/volumes.c

@@ -2543,9 +2543,6 @@ static int btrfs_relocate_chunk(struct btrfs_root *root,
 	remove_extent_mapping(em_tree, em);
 	write_unlock(&em_tree->lock);
 
-	kfree(map);
-	em->bdev = NULL;
-
 	/* once for the tree */
 	free_extent_map(em);
 	/* once for us */
@@ -4301,9 +4298,11 @@ static int __btrfs_alloc_chunk(struct btrfs_trans_handle *trans,
 
 	em = alloc_extent_map();
 	if (!em) {
+		kfree(map);
 		ret = -ENOMEM;
 		goto error;
 	}
+	set_bit(EXTENT_FLAG_FS_MAPPING, &em->flags);
 	em->bdev = (struct block_device *)map;
 	em->start = start;
 	em->len = num_bytes;
@@ -4346,7 +4345,6 @@ static int __btrfs_alloc_chunk(struct btrfs_trans_handle *trans,
 	/* One for the tree reference */
 	free_extent_map(em);
 error:
-	kfree(map);
 	kfree(devices_info);
 	return ret;
 }
@@ -4558,7 +4556,6 @@ void btrfs_mapping_tree_free(struct btrfs_mapping_tree *tree)
 		write_unlock(&tree->map_tree.lock);
 		if (!em)
 			break;
-		kfree(em->bdev);
 		/* once for us */
 		free_extent_map(em);
 		/* once for the tree */
@@ -5362,6 +5359,15 @@ int btrfs_rmap_block(struct btrfs_mapping_tree *map_tree,
 	return 0;
 }
 
+static inline void btrfs_end_bbio(struct btrfs_bio *bbio, struct bio *bio, int err)
+{
+	if (likely(bbio->flags & BTRFS_BIO_ORIG_BIO_SUBMITTED))
+		bio_endio_nodec(bio, err);
+	else
+		bio_endio(bio, err);
+	kfree(bbio);
+}
+
 static void btrfs_end_bio(struct bio *bio, int err)
 {
 	struct btrfs_bio *bbio = bio->bi_private;
@@ -5402,12 +5408,6 @@ static void btrfs_end_bio(struct bio *bio, int err)
 			bio = bbio->orig_bio;
 		}
 
- 		/*
-		 * We have original bio now. So increment bi_remaining to
-		 * account for it in endio
-		 */
-		atomic_inc(&bio->bi_remaining);
-
 		bio->bi_private = bbio->private;
 		bio->bi_end_io = bbio->end_io;
 		btrfs_io_bio(bio)->mirror_num = bbio->mirror_num;
@@ -5424,9 +5424,8 @@ static void btrfs_end_bio(struct bio *bio, int err)
 			set_bit(BIO_UPTODATE, &bio->bi_flags);
 			err = 0;
 		}
-		kfree(bbio);
 
-		bio_endio(bio, err);
+		btrfs_end_bbio(bbio, bio, err);
 	} else if (!is_orig_bio) {
 		bio_put(bio);
 	}
@@ -5589,12 +5588,15 @@ static void bbio_error(struct btrfs_bio *bbio, struct bio *bio, u64 logical)
 {
 	atomic_inc(&bbio->error);
 	if (atomic_dec_and_test(&bbio->stripes_pending)) {
+		/* Shoud be the original bio. */
+		WARN_ON(bio != bbio->orig_bio);
+
 		bio->bi_private = bbio->private;
 		bio->bi_end_io = bbio->end_io;
 		btrfs_io_bio(bio)->mirror_num = bbio->mirror_num;
 		bio->bi_iter.bi_sector = logical >> 9;
-		kfree(bbio);
-		bio_endio(bio, -EIO);
+
+		btrfs_end_bbio(bbio, bio, -EIO);
 	}
 }
 
@@ -5681,6 +5683,7 @@ int btrfs_map_bio(struct btrfs_root *root, int rw, struct bio *bio,
 			BUG_ON(!bio); /* -ENOMEM */
 		} else {
 			bio = first_bio;
+			bbio->flags |= BTRFS_BIO_ORIG_BIO_SUBMITTED;
 		}
 
 		submit_stripe_bio(root, bbio, bio,
@@ -5822,6 +5825,7 @@ static int read_one_chunk(struct btrfs_root *root, struct btrfs_key *key,
 		return -ENOMEM;
 	}
 
+	set_bit(EXTENT_FLAG_FS_MAPPING, &em->flags);
 	em->bdev = (struct block_device *)map;
 	em->start = logical;
 	em->len = length;
@@ -5846,7 +5850,6 @@ static int read_one_chunk(struct btrfs_root *root, struct btrfs_key *key,
 		map->stripes[i].dev = btrfs_find_device(root->fs_info, devid,
 							uuid, NULL);
 		if (!map->stripes[i].dev && !btrfs_test_opt(root, DEGRADED)) {
-			kfree(map);
 			free_extent_map(em);
 			return -EIO;
 		}
@@ -5854,7 +5857,6 @@ static int read_one_chunk(struct btrfs_root *root, struct btrfs_key *key,
 			map->stripes[i].dev =
 				add_missing_dev(root, devid, uuid);
 			if (!map->stripes[i].dev) {
-				kfree(map);
 				free_extent_map(em);
 				return -EIO;
 			}

fs/btrfs/volumes.h

@@ -190,11 +190,14 @@ struct btrfs_bio_stripe {
 struct btrfs_bio;
 typedef void (btrfs_bio_end_io_t) (struct btrfs_bio *bio, int err);
 
+#define BTRFS_BIO_ORIG_BIO_SUBMITTED	0x1
+
 struct btrfs_bio {
 	atomic_t stripes_pending;
 	struct btrfs_fs_info *fs_info;
 	bio_end_io_t *end_io;
 	struct bio *orig_bio;
+	unsigned long flags;
 	void *private;
 	atomic_t error;
 	int max_errors;