Merge branch 'for-linus-4.3' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs

Pull btrfs updates from Chris Mason:
 "This has Jeff Mahoney's long standing trim patch that fixes corners
  where trims were missing.  Omar has some raid5/6 fixes, especially for
  using scrub and device replace when devices are missing.

  Zhao Lie continues cleaning and fixing things, this series fixes some
  really hard to hit corners in xfstests.  I had to pull it last merge
  window due to some deadlocks, but those are now resolved.

  I added support for Tejun's new blkio controllers.  It seems to work
  well for single devices, we'll expand to multi-device as well"

* 'for-linus-4.3' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs: (47 commits)
  btrfs: fix compile when block cgroups are not enabled
  Btrfs: fix file read corruption after extent cloning and fsync
  Btrfs: check if previous transaction aborted to avoid fs corruption
  btrfs: use __GFP_NOFAIL in alloc_btrfs_bio
  btrfs: Prevent from early transaction abort
  btrfs: Remove unused arguments in tree-log.c
  btrfs: Remove useless condition in start_log_trans()
  Btrfs: add support for blkio controllers
  Btrfs: remove unused mutex from struct 'btrfs_fs_info'
  Btrfs: fix parity scrub of RAID 5/6 with missing device
  Btrfs: fix device replace of a missing RAID 5/6 device
  Btrfs: add RAID 5/6 BTRFS_RBIO_REBUILD_MISSING operation
  Btrfs: count devices correctly in readahead during RAID 5/6 replace
  Btrfs: remove misleading handling of missing device scrub
  btrfs: fix clone / extent-same deadlocks
  Btrfs: fix defrag to merge tail file extent
  Btrfs: fix warning in backref walking
  btrfs: Add WARN_ON() for double lock in btrfs_tree_lock()
  btrfs: Remove root argument in extent_data_ref_count()
  btrfs: Fix wrong comment of btrfs_alloc_tree_block()
  ...

fs/btrfs/backref.c

@@ -206,10 +206,33 @@ static int __add_prelim_ref(struct list_head *head, u64 root_id,
 		return -ENOMEM;
 
 	ref->root_id = root_id;
-	if (key)
+	if (key) {
 		ref->key_for_search = *key;
-	else
+		/*
+		 * We can often find data backrefs with an offset that is too
+		 * large (>= LLONG_MAX, maximum allowed file offset) due to
+		 * underflows when subtracting a file's offset with the data
+		 * offset of its corresponding extent data item. This can
+		 * happen for example in the clone ioctl.
+		 * So if we detect such case we set the search key's offset to
+		 * zero to make sure we will find the matching file extent item
+		 * at add_all_parents(), otherwise we will miss it because the
+		 * offset taken form the backref is much larger then the offset
+		 * of the file extent item. This can make us scan a very large
+		 * number of file extent items, but at least it will not make
+		 * us miss any.
+		 * This is an ugly workaround for a behaviour that should have
+		 * never existed, but it does and a fix for the clone ioctl
+		 * would touch a lot of places, cause backwards incompatibility
+		 * and would not fix the problem for extents cloned with older
+		 * kernels.
+		 */
+		if (ref->key_for_search.type == BTRFS_EXTENT_DATA_KEY &&
+		    ref->key_for_search.offset >= LLONG_MAX)
+			ref->key_for_search.offset = 0;
+	} else {
 		memset(&ref->key_for_search, 0, sizeof(ref->key_for_search));
+	}
 
 	ref->inode_list = NULL;
 	ref->level = level;
@@ -632,7 +655,7 @@ static int __add_delayed_refs(struct btrfs_delayed_ref_head *head, u64 seq,
 			struct btrfs_delayed_tree_ref *ref;
 
 			ref = btrfs_delayed_node_to_tree_ref(node);
-			ret = __add_prelim_ref(prefs, ref->root, NULL,
+			ret = __add_prelim_ref(prefs, 0, NULL,
 					       ref->level + 1, ref->parent,
 					       node->bytenr,
 					       node->ref_mod * sgn, GFP_ATOMIC);
@@ -664,11 +687,7 @@ static int __add_delayed_refs(struct btrfs_delayed_ref_head *head, u64 seq,
 			struct btrfs_delayed_data_ref *ref;
 
 			ref = btrfs_delayed_node_to_data_ref(node);
-
-			key.objectid = ref->objectid;
-			key.type = BTRFS_EXTENT_DATA_KEY;
-			key.offset = ref->offset;
-			ret = __add_prelim_ref(prefs, ref->root, &key, 0,
+			ret = __add_prelim_ref(prefs, 0, NULL, 0,
 					       ref->parent, node->bytenr,
 					       node->ref_mod * sgn, GFP_ATOMIC);
 			break;

fs/btrfs/ctree.c

@@ -1159,8 +1159,10 @@ static noinline int __btrfs_cow_block(struct btrfs_trans_handle *trans,
 
 	if (test_bit(BTRFS_ROOT_REF_COWS, &root->state)) {
 		ret = btrfs_reloc_cow_block(trans, root, buf, cow);
-		if (ret)
+		if (ret) {
+			btrfs_abort_transaction(trans, root, ret);
 			return ret;
+		}
 	}
 
 	if (buf == root->node) {

fs/btrfs/ctree.h

@@ -1300,7 +1300,7 @@ struct btrfs_block_group_cache {
 	/* for raid56, this is a full stripe, without parity */
 	unsigned long full_stripe_len;
 
-	unsigned int ro:1;
+	unsigned int ro;
 	unsigned int iref:1;
 	unsigned int has_caching_ctl:1;
 	unsigned int removed:1;
@@ -1518,12 +1518,6 @@ struct btrfs_fs_info {
 	 */
 	struct mutex ordered_operations_mutex;
 
-	/*
-	 * Same as ordered_operations_mutex except this is for ordered extents
-	 * and not the operations.
-	 */
-	struct mutex ordered_extent_flush_mutex;
-
 	struct rw_semaphore commit_root_sem;
 
 	struct rw_semaphore cleanup_work_sem;
@@ -3437,6 +3431,8 @@ int btrfs_remove_block_group(struct btrfs_trans_handle *trans,
 			     struct btrfs_root *root, u64 group_start,
 			     struct extent_map *em);
 void btrfs_delete_unused_bgs(struct btrfs_fs_info *fs_info);
+void btrfs_get_block_group_trimming(struct btrfs_block_group_cache *cache);
+void btrfs_put_block_group_trimming(struct btrfs_block_group_cache *cache);
 void btrfs_create_pending_block_groups(struct btrfs_trans_handle *trans,
 				       struct btrfs_root *root);
 u64 btrfs_get_alloc_profile(struct btrfs_root *root, int data);
@@ -3495,9 +3491,9 @@ int btrfs_cond_migrate_bytes(struct btrfs_fs_info *fs_info,
 void btrfs_block_rsv_release(struct btrfs_root *root,
 			     struct btrfs_block_rsv *block_rsv,
 			     u64 num_bytes);
-int btrfs_set_block_group_ro(struct btrfs_root *root,
+int btrfs_inc_block_group_ro(struct btrfs_root *root,
 			     struct btrfs_block_group_cache *cache);
-void btrfs_set_block_group_rw(struct btrfs_root *root,
+void btrfs_dec_block_group_ro(struct btrfs_root *root,
 			      struct btrfs_block_group_cache *cache);
 void btrfs_put_block_group_cache(struct btrfs_fs_info *info);
 u64 btrfs_account_ro_block_groups_free_space(struct btrfs_space_info *sinfo);
@@ -4073,6 +4069,7 @@ __cold
 void __btrfs_std_error(struct btrfs_fs_info *fs_info, const char *function,
 		     unsigned int line, int errno, const char *fmt, ...);
 
+const char *btrfs_decode_error(int errno);
 
 __cold
 void __btrfs_abort_transaction(struct btrfs_trans_handle *trans,
@@ -4185,8 +4182,7 @@ int btrfs_reloc_clone_csums(struct inode *inode, u64 file_pos, u64 len);
 int btrfs_reloc_cow_block(struct btrfs_trans_handle *trans,
 			  struct btrfs_root *root, struct extent_buffer *buf,
 			  struct extent_buffer *cow);
-void btrfs_reloc_pre_snapshot(struct btrfs_trans_handle *trans,
-			      struct btrfs_pending_snapshot *pending,
+void btrfs_reloc_pre_snapshot(struct btrfs_pending_snapshot *pending,
 			      u64 *bytes_to_reserve);
 int btrfs_reloc_post_snapshot(struct btrfs_trans_handle *trans,
 			      struct btrfs_pending_snapshot *pending);

fs/btrfs/disk-io.c

@@ -1730,6 +1730,7 @@ static int setup_bdi(struct btrfs_fs_info *info, struct backing_dev_info *bdi)
 	bdi->ra_pages = VM_MAX_READAHEAD * 1024 / PAGE_CACHE_SIZE;
 	bdi->congested_fn	= btrfs_congested_fn;
 	bdi->congested_data	= info;
+	bdi->capabilities |= BDI_CAP_CGROUP_WRITEBACK;
 	return 0;
 }
 
@@ -2613,7 +2614,6 @@ int open_ctree(struct super_block *sb,
 
 
 	mutex_init(&fs_info->ordered_operations_mutex);
-	mutex_init(&fs_info->ordered_extent_flush_mutex);
 	mutex_init(&fs_info->tree_log_mutex);
 	mutex_init(&fs_info->chunk_mutex);
 	mutex_init(&fs_info->transaction_kthread_mutex);
@@ -2955,8 +2955,9 @@ retry_root_backup:
 	if (fs_info->fs_devices->missing_devices >
 	     fs_info->num_tolerated_disk_barrier_failures &&
 	    !(sb->s_flags & MS_RDONLY)) {
-		printk(KERN_WARNING "BTRFS: "
-			"too many missing devices, writeable mount is not allowed\n");
+		pr_warn("BTRFS: missing devices(%llu) exceeds the limit(%d), writeable mount is not allowed\n",
+			fs_info->fs_devices->missing_devices,
+			fs_info->num_tolerated_disk_barrier_failures);
 		goto fail_sysfs;
 	}
 
@@ -3763,6 +3764,15 @@ void close_ctree(struct btrfs_root *root)
 	cancel_work_sync(&fs_info->async_reclaim_work);
 
 	if (!(fs_info->sb->s_flags & MS_RDONLY)) {
+		/*
+		 * If the cleaner thread is stopped and there are
+		 * block groups queued for removal, the deletion will be
+		 * skipped when we quit the cleaner thread.
+		 */
+		mutex_lock(&root->fs_info->cleaner_mutex);
+		btrfs_delete_unused_bgs(root->fs_info);
+		mutex_unlock(&root->fs_info->cleaner_mutex);
+
 		ret = btrfs_commit_super(root);
 		if (ret)
 			btrfs_err(fs_info, "commit super ret %d", ret);

fs/btrfs/extent-tree.c

@@ -1316,8 +1316,7 @@ static noinline int remove_extent_data_ref(struct btrfs_trans_handle *trans,
 	return ret;
 }
 
-static noinline u32 extent_data_ref_count(struct btrfs_root *root,
-					  struct btrfs_path *path,
+static noinline u32 extent_data_ref_count(struct btrfs_path *path,
 					  struct btrfs_extent_inline_ref *iref)
 {
 	struct btrfs_key key;
@@ -1883,10 +1882,77 @@ static int remove_extent_backref(struct btrfs_trans_handle *trans,
 	return ret;
 }
 
-static int btrfs_issue_discard(struct block_device *bdev,
-				u64 start, u64 len)
+#define in_range(b, first, len)        ((b) >= (first) && (b) < (first) + (len))
+static int btrfs_issue_discard(struct block_device *bdev, u64 start, u64 len,
+			       u64 *discarded_bytes)
 {
-	return blkdev_issue_discard(bdev, start >> 9, len >> 9, GFP_NOFS, 0);
+	int j, ret = 0;
+	u64 bytes_left, end;
+	u64 aligned_start = ALIGN(start, 1 << 9);
+
+	if (WARN_ON(start != aligned_start)) {
+		len -= aligned_start - start;
+		len = round_down(len, 1 << 9);
+		start = aligned_start;
+	}
+
+	*discarded_bytes = 0;
+
+	if (!len)
+		return 0;
+
+	end = start + len;
+	bytes_left = len;
+
+	/* Skip any superblocks on this device. */
+	for (j = 0; j < BTRFS_SUPER_MIRROR_MAX; j++) {
+		u64 sb_start = btrfs_sb_offset(j);
+		u64 sb_end = sb_start + BTRFS_SUPER_INFO_SIZE;
+		u64 size = sb_start - start;
+
+		if (!in_range(sb_start, start, bytes_left) &&
+		    !in_range(sb_end, start, bytes_left) &&
+		    !in_range(start, sb_start, BTRFS_SUPER_INFO_SIZE))
+			continue;
+
+		/*
+		 * Superblock spans beginning of range.  Adjust start and
+		 * try again.
+		 */
+		if (sb_start <= start) {
+			start += sb_end - start;
+			if (start > end) {
+				bytes_left = 0;
+				break;
+			}
+			bytes_left = end - start;
+			continue;
+		}
+
+		if (size) {
+			ret = blkdev_issue_discard(bdev, start >> 9, size >> 9,
+						   GFP_NOFS, 0);
+			if (!ret)
+				*discarded_bytes += size;
+			else if (ret != -EOPNOTSUPP)
+				return ret;
+		}
+
+		start = sb_end;
+		if (start > end) {
+			bytes_left = 0;
+			break;
+		}
+		bytes_left = end - start;
+	}
+
+	if (bytes_left) {
+		ret = blkdev_issue_discard(bdev, start >> 9, bytes_left >> 9,
+					   GFP_NOFS, 0);
+		if (!ret)
+			*discarded_bytes += bytes_left;
+	}
+	return ret;
 }
 
 int btrfs_discard_extent(struct btrfs_root *root, u64 bytenr,
@@ -1907,14 +1973,16 @@ int btrfs_discard_extent(struct btrfs_root *root, u64 bytenr,
 
 
 		for (i = 0; i < bbio->num_stripes; i++, stripe++) {
+			u64 bytes;
 			if (!stripe->dev->can_discard)
 				continue;
 
 			ret = btrfs_issue_discard(stripe->dev->bdev,
 						  stripe->physical,
-						  stripe->length);
+						  stripe->length,
+						  &bytes);
 			if (!ret)
-				discarded_bytes += stripe->length;
+				discarded_bytes += bytes;
 			else if (ret != -EOPNOTSUPP)
 				break; /* Logic errors or -ENOMEM, or -EIO but I don't know how that could happen JDM */
 
@@ -6062,20 +6130,19 @@ int btrfs_finish_extent_commit(struct btrfs_trans_handle *trans,
 			       struct btrfs_root *root)
 {
 	struct btrfs_fs_info *fs_info = root->fs_info;
+	struct btrfs_block_group_cache *block_group, *tmp;
+	struct list_head *deleted_bgs;
 	struct extent_io_tree *unpin;
 	u64 start;
 	u64 end;
 	int ret;
 
-	if (trans->aborted)
-		return 0;
-
 	if (fs_info->pinned_extents == &fs_info->freed_extents[0])
 		unpin = &fs_info->freed_extents[1];
 	else
 		unpin = &fs_info->freed_extents[0];
 
-	while (1) {
+	while (!trans->aborted) {
 		mutex_lock(&fs_info->unused_bg_unpin_mutex);
 		ret = find_first_extent_bit(unpin, 0, &start, &end,
 					    EXTENT_DIRTY, NULL);
@@ -6094,6 +6161,34 @@ int btrfs_finish_extent_commit(struct btrfs_trans_handle *trans,
 		cond_resched();
 	}
 
+	/*
+	 * Transaction is finished.  We don't need the lock anymore.  We
+	 * do need to clean up the block groups in case of a transaction
+	 * abort.
+	 */
+	deleted_bgs = &trans->transaction->deleted_bgs;
+	list_for_each_entry_safe(block_group, tmp, deleted_bgs, bg_list) {
+		u64 trimmed = 0;
+
+		ret = -EROFS;
+		if (!trans->aborted)
+			ret = btrfs_discard_extent(root,
+						   block_group->key.objectid,
+						   block_group->key.offset,
+						   &trimmed);
+
+		list_del_init(&block_group->bg_list);
+		btrfs_put_block_group_trimming(block_group);
+		btrfs_put_block_group(block_group);
+
+		if (ret) {
+			const char *errstr = btrfs_decode_error(ret);
+			btrfs_warn(fs_info,
+				   "Discard failed while removing blockgroup: errno=%d %s\n",
+				   ret, errstr);
+		}
+	}
+
 	return 0;
 }
 
@@ -6349,7 +6444,7 @@ static int __btrfs_free_extent(struct btrfs_trans_handle *trans,
 	} else {
 		if (found_extent) {
 			BUG_ON(is_data && refs_to_drop !=
-			       extent_data_ref_count(root, path, iref));
+			       extent_data_ref_count(path, iref));
 			if (iref) {
 				BUG_ON(path->slots[0] != extent_slot);
 			} else {
@@ -7567,9 +7662,6 @@ static void unuse_block_rsv(struct btrfs_fs_info *fs_info,
 
 /*
  * finds a free extent and does all the dirty work required for allocation
- * returns the key for the extent through ins, and a tree buffer for
- * the first block of the extent through buf.
- *
  * returns the tree buffer or an ERR_PTR on error.
  */
 struct extent_buffer *btrfs_alloc_tree_block(struct btrfs_trans_handle *trans,
@@ -8723,14 +8815,13 @@ static u64 update_block_group_flags(struct btrfs_root *root, u64 flags)
 	return flags;
 }
 
-static int set_block_group_ro(struct btrfs_block_group_cache *cache, int force)
+static int inc_block_group_ro(struct btrfs_block_group_cache *cache, int force)
 {
 	struct btrfs_space_info *sinfo = cache->space_info;
 	u64 num_bytes;
 	u64 min_allocable_bytes;
 	int ret = -ENOSPC;
 
-
 	/*
 	 * We need some metadata space and system metadata space for
 	 * allocating chunks in some corner cases until we force to set
@@ -8747,6 +8838,7 @@ static int set_block_group_ro(struct btrfs_block_group_cache *cache, int force)
 	spin_lock(&cache->lock);
 
 	if (cache->ro) {
+		cache->ro++;
 		ret = 0;
 		goto out;
 	}
@@ -8758,7 +8850,7 @@ static int set_block_group_ro(struct btrfs_block_group_cache *cache, int force)
 	    sinfo->bytes_may_use + sinfo->bytes_readonly + num_bytes +
 	    min_allocable_bytes <= sinfo->total_bytes) {
 		sinfo->bytes_readonly += num_bytes;
-		cache->ro = 1;
+		cache->ro++;
 		list_add_tail(&cache->ro_list, &sinfo->ro_bgs);
 		ret = 0;
 	}
@@ -8768,7 +8860,7 @@ out:
 	return ret;
 }
 
-int btrfs_set_block_group_ro(struct btrfs_root *root,
+int btrfs_inc_block_group_ro(struct btrfs_root *root,
 			     struct btrfs_block_group_cache *cache)
 
 {
@@ -8776,8 +8868,6 @@ int btrfs_set_block_group_ro(struct btrfs_root *root,
 	u64 alloc_flags;
 	int ret;
 
-	BUG_ON(cache->ro);
-
 again:
 	trans = btrfs_join_transaction(root);
 	if (IS_ERR(trans))
@@ -8820,7 +8910,7 @@ again:
 			goto out;
 	}
 
-	ret = set_block_group_ro(cache, 0);
+	ret = inc_block_group_ro(cache, 0);
 	if (!ret)
 		goto out;
 	alloc_flags = get_alloc_profile(root, cache->space_info->flags);
@@ -8828,7 +8918,7 @@ again:
 			     CHUNK_ALLOC_FORCE);
 	if (ret < 0)
 		goto out;
-	ret = set_block_group_ro(cache, 0);
+	ret = inc_block_group_ro(cache, 0);
 out:
 	if (cache->flags & BTRFS_BLOCK_GROUP_SYSTEM) {
 		alloc_flags = update_block_group_flags(root, cache->flags);
@@ -8891,7 +8981,7 @@ u64 btrfs_account_ro_block_groups_free_space(struct btrfs_space_info *sinfo)
 	return free_bytes;
 }
 
-void btrfs_set_block_group_rw(struct btrfs_root *root,
+void btrfs_dec_block_group_ro(struct btrfs_root *root,
 			      struct btrfs_block_group_cache *cache)
 {
 	struct btrfs_space_info *sinfo = cache->space_info;
@@ -8901,11 +8991,13 @@ void btrfs_set_block_group_rw(struct btrfs_root *root,
 
 	spin_lock(&sinfo->lock);
 	spin_lock(&cache->lock);
-	num_bytes = cache->key.offset - cache->reserved - cache->pinned -
-		    cache->bytes_super - btrfs_block_group_used(&cache->item);
-	sinfo->bytes_readonly -= num_bytes;
-	cache->ro = 0;
-	list_del_init(&cache->ro_list);
+	if (!--cache->ro) {
+		num_bytes = cache->key.offset - cache->reserved -
+			    cache->pinned - cache->bytes_super -
+			    btrfs_block_group_used(&cache->item);
+		sinfo->bytes_readonly -= num_bytes;
+		list_del_init(&cache->ro_list);
+	}
 	spin_unlock(&cache->lock);
 	spin_unlock(&sinfo->lock);
 }
@@ -9421,7 +9513,7 @@ int btrfs_read_block_groups(struct btrfs_root *root)
 
 		set_avail_alloc_bits(root->fs_info, cache->flags);
 		if (btrfs_chunk_readonly(root, cache->key.objectid)) {
-			set_block_group_ro(cache, 1);
+			inc_block_group_ro(cache, 1);
 		} else if (btrfs_block_group_used(&cache->item) == 0) {
 			spin_lock(&info->unused_bgs_lock);
 			/* Should always be true but just in case. */
@@ -9449,11 +9541,11 @@ int btrfs_read_block_groups(struct btrfs_root *root)
 		list_for_each_entry(cache,
 				&space_info->block_groups[BTRFS_RAID_RAID0],
 				list)
-			set_block_group_ro(cache, 1);
+			inc_block_group_ro(cache, 1);
 		list_for_each_entry(cache,
 				&space_info->block_groups[BTRFS_RAID_SINGLE],
 				list)
-			set_block_group_ro(cache, 1);
+			inc_block_group_ro(cache, 1);
 	}
 
 	init_global_block_rsv(info);
@@ -9834,6 +9926,11 @@ int btrfs_remove_block_group(struct btrfs_trans_handle *trans,
 	 * currently running transaction might finish and a new one start,
 	 * allowing for new block groups to be created that can reuse the same
 	 * physical device locations unless we take this special care.
+	 *
+	 * There may also be an implicit trim operation if the file system
+	 * is mounted with -odiscard. The same protections must remain
+	 * in place until the extents have been discarded completely when
+	 * the transaction commit has completed.
 	 */
 	remove_em = (atomic_read(&block_group->trimming) == 0);
 	/*
@@ -9908,6 +10005,7 @@ void btrfs_delete_unused_bgs(struct btrfs_fs_info *fs_info)
 	spin_lock(&fs_info->unused_bgs_lock);
 	while (!list_empty(&fs_info->unused_bgs)) {
 		u64 start, end;
+		int trimming;
 
 		block_group = list_first_entry(&fs_info->unused_bgs,
 					       struct btrfs_block_group_cache,
@@ -9941,7 +10039,7 @@ void btrfs_delete_unused_bgs(struct btrfs_fs_info *fs_info)
 		spin_unlock(&block_group->lock);
 
 		/* We don't want to force the issue, only flip if it's ok. */
-		ret = set_block_group_ro(block_group, 0);
+		ret = inc_block_group_ro(block_group, 0);
 		up_write(&space_info->groups_sem);
 		if (ret < 0) {
 			ret = 0;
@@ -9955,7 +10053,7 @@ void btrfs_delete_unused_bgs(struct btrfs_fs_info *fs_info)
 		/* 1 for btrfs_orphan_reserve_metadata() */
 		trans = btrfs_start_transaction(root, 1);
 		if (IS_ERR(trans)) {
-			btrfs_set_block_group_rw(root, block_group);
+			btrfs_dec_block_group_ro(root, block_group);
 			ret = PTR_ERR(trans);
 			goto next;
 		}
@@ -9982,14 +10080,14 @@ void btrfs_delete_unused_bgs(struct btrfs_fs_info *fs_info)
 				  EXTENT_DIRTY, GFP_NOFS);
 		if (ret) {
 			mutex_unlock(&fs_info->unused_bg_unpin_mutex);
-			btrfs_set_block_group_rw(root, block_group);
+			btrfs_dec_block_group_ro(root, block_group);
 			goto end_trans;
 		}
 		ret = clear_extent_bits(&fs_info->freed_extents[1], start, end,
 				  EXTENT_DIRTY, GFP_NOFS);
 		if (ret) {
 			mutex_unlock(&fs_info->unused_bg_unpin_mutex);
-			btrfs_set_block_group_rw(root, block_group);
+			btrfs_dec_block_group_ro(root, block_group);
 			goto end_trans;
 		}
 		mutex_unlock(&fs_info->unused_bg_unpin_mutex);
@@ -10007,12 +10105,39 @@ void btrfs_delete_unused_bgs(struct btrfs_fs_info *fs_info)
 		spin_unlock(&block_group->lock);
 		spin_unlock(&space_info->lock);
 
+		/* DISCARD can flip during remount */
+		trimming = btrfs_test_opt(root, DISCARD);
+
+		/* Implicit trim during transaction commit. */
+		if (trimming)
+			btrfs_get_block_group_trimming(block_group);
+
 		/*
 		 * Btrfs_remove_chunk will abort the transaction if things go
 		 * horribly wrong.
 		 */
 		ret = btrfs_remove_chunk(trans, root,
 					 block_group->key.objectid);
+
+		if (ret) {
+			if (trimming)
+				btrfs_put_block_group_trimming(block_group);
+			goto end_trans;
+		}
+
+		/*
+		 * If we're not mounted with -odiscard, we can just forget
+		 * about this block group. Otherwise we'll need to wait
+		 * until transaction commit to do the actual discard.
+		 */
+		if (trimming) {
+			WARN_ON(!list_empty(&block_group->bg_list));
+			spin_lock(&trans->transaction->deleted_bgs_lock);
+			list_move(&block_group->bg_list,
+				  &trans->transaction->deleted_bgs);
+			spin_unlock(&trans->transaction->deleted_bgs_lock);
+			btrfs_get_block_group(block_group);
+		}
 end_trans:
 		btrfs_end_transaction(trans, root);
 next:
@@ -10066,10 +10191,99 @@ int btrfs_error_unpin_extent_range(struct btrfs_root *root, u64 start, u64 end)
 	return unpin_extent_range(root, start, end, false);
 }
 
+/*
+ * It used to be that old block groups would be left around forever.
+ * Iterating over them would be enough to trim unused space.  Since we
+ * now automatically remove them, we also need to iterate over unallocated
+ * space.
+ *
+ * We don't want a transaction for this since the discard may take a
+ * substantial amount of time.  We don't require that a transaction be
+ * running, but we do need to take a running transaction into account
+ * to ensure that we're not discarding chunks that were released in
+ * the current transaction.
+ *
+ * Holding the chunks lock will prevent other threads from allocating
+ * or releasing chunks, but it won't prevent a running transaction
+ * from committing and releasing the memory that the pending chunks
+ * list head uses.  For that, we need to take a reference to the
+ * transaction.
+ */
+static int btrfs_trim_free_extents(struct btrfs_device *device,
+				   u64 minlen, u64 *trimmed)
+{
+	u64 start = 0, len = 0;
+	int ret;
+
+	*trimmed = 0;
+
+	/* Not writeable = nothing to do. */
+	if (!device->writeable)
+		return 0;
+
+	/* No free space = nothing to do. */
+	if (device->total_bytes <= device->bytes_used)
+		return 0;
+
+	ret = 0;
+
+	while (1) {
+		struct btrfs_fs_info *fs_info = device->dev_root->fs_info;
+		struct btrfs_transaction *trans;
+		u64 bytes;
+
+		ret = mutex_lock_interruptible(&fs_info->chunk_mutex);
+		if (ret)
+			return ret;
+
+		down_read(&fs_info->commit_root_sem);
+
+		spin_lock(&fs_info->trans_lock);
+		trans = fs_info->running_transaction;
+		if (trans)
+			atomic_inc(&trans->use_count);
+		spin_unlock(&fs_info->trans_lock);
+
+		ret = find_free_dev_extent_start(trans, device, minlen, start,
+						 &start, &len);
+		if (trans)
+			btrfs_put_transaction(trans);
+
+		if (ret) {
+			up_read(&fs_info->commit_root_sem);
+			mutex_unlock(&fs_info->chunk_mutex);
+			if (ret == -ENOSPC)
+				ret = 0;
+			break;
+		}
+
+		ret = btrfs_issue_discard(device->bdev, start, len, &bytes);
+		up_read(&fs_info->commit_root_sem);
+		mutex_unlock(&fs_info->chunk_mutex);
+
+		if (ret)
+			break;
+
+		start += len;
+		*trimmed += bytes;
+
+		if (fatal_signal_pending(current)) {
+			ret = -ERESTARTSYS;
+			break;
+		}
+
+		cond_resched();
+	}
+
+	return ret;
+}
+
 int btrfs_trim_fs(struct btrfs_root *root, struct fstrim_range *range)
 {
 	struct btrfs_fs_info *fs_info = root->fs_info;
 	struct btrfs_block_group_cache *cache = NULL;
+	struct btrfs_device *device;
+	struct list_head *devices;
 	u64 group_trimmed;
 	u64 start;
 	u64 end;
@@ -10124,6 +10338,18 @@ int btrfs_trim_fs(struct btrfs_root *root, struct fstrim_range *range)
 		cache = next_block_group(fs_info->tree_root, cache);
 	}
 
+	mutex_lock(&root->fs_info->fs_devices->device_list_mutex);
+	devices = &root->fs_info->fs_devices->alloc_list;
+	list_for_each_entry(device, devices, dev_alloc_list) {
+		ret = btrfs_trim_free_extents(device, range->minlen,
+					      &group_trimmed);
+		if (ret)
+			break;
+
+		trimmed += group_trimmed;
+	}
+	mutex_unlock(&root->fs_info->fs_devices->device_list_mutex);
+
 	range->len = trimmed;
 	return ret;
 }

fs/btrfs/extent_io.c

@@ -2723,6 +2723,12 @@ struct bio *btrfs_bio_clone(struct bio *bio, gfp_t gfp_mask)
 		btrfs_bio->csum = NULL;
 		btrfs_bio->csum_allocated = NULL;
 		btrfs_bio->end_io = NULL;
+
+#ifdef CONFIG_BLK_CGROUP
+		/* FIXME, put this into bio_clone_bioset */
+		if (bio->bi_css)
+			bio_associate_blkcg(new, bio->bi_css);
+#endif
 	}
 	return new;
 }
@@ -2783,6 +2789,7 @@ static int merge_bio(int rw, struct extent_io_tree *tree, struct page *page,
 }
 
 static int submit_extent_page(int rw, struct extent_io_tree *tree,
+			      struct writeback_control *wbc,
 			      struct page *page, sector_t sector,
 			      size_t size, unsigned long offset,
 			      struct block_device *bdev,
@@ -2817,6 +2824,8 @@ static int submit_extent_page(int rw, struct extent_io_tree *tree,
 			}
 			bio = NULL;
 		} else {
+			if (wbc)
+				wbc_account_io(wbc, page, page_size);
 			return 0;
 		}
 	}
@@ -2829,6 +2838,10 @@ static int submit_extent_page(int rw, struct extent_io_tree *tree,
 	bio_add_page(bio, page, page_size, offset);
 	bio->bi_end_io = end_io_func;
 	bio->bi_private = tree;
+	if (wbc) {
+		wbc_init_bio(wbc, bio);
+		wbc_account_io(wbc, page, page_size);
+	}
 
 	if (bio_ret)
 		*bio_ret = bio;
@@ -3039,7 +3052,7 @@ static int __do_readpage(struct extent_io_tree *tree,
 		}
 
 		pnr -= page->index;
-		ret = submit_extent_page(rw, tree, page,
+		ret = submit_extent_page(rw, tree, NULL, page,
 					 sector, disk_io_size, pg_offset,
 					 bdev, bio, pnr,
 					 end_bio_extent_readpage, mirror_num,
@@ -3434,7 +3447,7 @@ static noinline_for_stack int __extent_writepage_io(struct inode *inode,
 				       page->index, cur, end);
 			}
 
-			ret = submit_extent_page(write_flags, tree, page,
+			ret = submit_extent_page(write_flags, tree, wbc, page,
 						 sector, iosize, pg_offset,
 						 bdev, &epd->bio, max_nr,
 						 end_bio_extent_writepage,
@@ -3738,7 +3751,7 @@ static noinline_for_stack int write_one_eb(struct extent_buffer *eb,
 
 		clear_page_dirty_for_io(p);
 		set_page_writeback(p);
-		ret = submit_extent_page(rw, tree, p, offset >> 9,
+		ret = submit_extent_page(rw, tree, wbc, p, offset >> 9,
 					 PAGE_CACHE_SIZE, 0, bdev, &epd->bio,
 					 -1, end_bio_extent_buffer_writepage,
 					 0, epd->bio_flags, bio_flags);
@@ -4603,9 +4616,7 @@ __alloc_extent_buffer(struct btrfs_fs_info *fs_info, u64 start,
 {
 	struct extent_buffer *eb = NULL;
 
-	eb = kmem_cache_zalloc(extent_buffer_cache, GFP_NOFS);
-	if (eb == NULL)
-		return NULL;
+	eb = kmem_cache_zalloc(extent_buffer_cache, GFP_NOFS|__GFP_NOFAIL);
 	eb->start = start;
 	eb->len = len;
 	eb->fs_info = fs_info;
@@ -4863,7 +4874,7 @@ struct extent_buffer *alloc_extent_buffer(struct btrfs_fs_info *fs_info,
 		return NULL;
 
 	for (i = 0; i < num_pages; i++, index++) {
-		p = find_or_create_page(mapping, index, GFP_NOFS);
+		p = find_or_create_page(mapping, index, GFP_NOFS|__GFP_NOFAIL);
 		if (!p)
 			goto free_eb;
 

fs/btrfs/free-space-cache.c

@@ -3272,35 +3272,23 @@ next:
 	return ret;
 }
 
-int btrfs_trim_block_group(struct btrfs_block_group_cache *block_group,
-			   u64 *trimmed, u64 start, u64 end, u64 minlen)
+void btrfs_get_block_group_trimming(struct btrfs_block_group_cache *cache)
 {
-	int ret;
+	atomic_inc(&cache->trimming);
+}
 
-	*trimmed = 0;
+void btrfs_put_block_group_trimming(struct btrfs_block_group_cache *block_group)
+{
+	struct extent_map_tree *em_tree;
+	struct extent_map *em;
+	bool cleanup;
 
 	spin_lock(&block_group->lock);
-	if (block_group->removed) {
-		spin_unlock(&block_group->lock);
-		return 0;
-	}
-	atomic_inc(&block_group->trimming);
+	cleanup = (atomic_dec_and_test(&block_group->trimming) &&
+		   block_group->removed);
 	spin_unlock(&block_group->lock);
 
-	ret = trim_no_bitmap(block_group, trimmed, start, end, minlen);
-	if (ret)
-		goto out;
-
-	ret = trim_bitmaps(block_group, trimmed, start, end, minlen);
-out:
-	spin_lock(&block_group->lock);
-	if (atomic_dec_and_test(&block_group->trimming) &&
-	    block_group->removed) {
-		struct extent_map_tree *em_tree;
-		struct extent_map *em;
-
-		spin_unlock(&block_group->lock);
-
+	if (cleanup) {
 		lock_chunks(block_group->fs_info->chunk_root);
 		em_tree = &block_group->fs_info->mapping_tree.map_tree;
 		write_lock(&em_tree->lock);
@@ -3324,10 +3312,31 @@ out:
 		 * this block group have left 1 entry each one. Free them.
 		 */
 		__btrfs_remove_free_space_cache(block_group->free_space_ctl);
-	} else {
+	}
+}
+
+int btrfs_trim_block_group(struct btrfs_block_group_cache *block_group,
+			   u64 *trimmed, u64 start, u64 end, u64 minlen)
+{
+	int ret;
+
+	*trimmed = 0;
+
+	spin_lock(&block_group->lock);
+	if (block_group->removed) {
 		spin_unlock(&block_group->lock);
+		return 0;
 	}
+	btrfs_get_block_group_trimming(block_group);
+	spin_unlock(&block_group->lock);
+
+	ret = trim_no_bitmap(block_group, trimmed, start, end, minlen);
+	if (ret)
+		goto out;
 
+	ret = trim_bitmaps(block_group, trimmed, start, end, minlen);
+out:
+	btrfs_put_block_group_trimming(block_group);
 	return ret;
 }
 

fs/btrfs/inode.c

@@ -3658,6 +3658,35 @@ cache_index:
 		set_bit(BTRFS_INODE_NEEDS_FULL_SYNC,
 			&BTRFS_I(inode)->runtime_flags);
 
+	/*
+	 * We don't persist the id of the transaction where an unlink operation
+	 * against the inode was last made. So here we assume the inode might
+	 * have been evicted, and therefore the exact value of last_unlink_trans
+	 * lost, and set it to last_trans to avoid metadata inconsistencies
+	 * between the inode and its parent if the inode is fsync'ed and the log
+	 * replayed. For example, in the scenario:
+	 *
+	 * touch mydir/foo
+	 * ln mydir/foo mydir/bar
+	 * sync
+	 * unlink mydir/bar
+	 * echo 2 > /proc/sys/vm/drop_caches   # evicts inode
+	 * xfs_io -c fsync mydir/foo
+	 * <power failure>
+	 * mount fs, triggers fsync log replay
+	 *
+	 * We must make sure that when we fsync our inode foo we also log its
+	 * parent inode, otherwise after log replay the parent still has the
+	 * dentry with the "bar" name but our inode foo has a link count of 1
+	 * and doesn't have an inode ref with the name "bar" anymore.
+	 *
+	 * Setting last_unlink_trans to last_trans is a pessimistic approach,
+	 * but it guarantees correctness at the expense of ocassional full
+	 * transaction commits on fsync if our inode is a directory, or if our
+	 * inode is not a directory, logging its parent unnecessarily.
+	 */
+	BTRFS_I(inode)->last_unlink_trans = BTRFS_I(inode)->last_trans;
+
 	path->slots[0]++;
 	if (inode->i_nlink != 1 ||
 	    path->slots[0] >= btrfs_header_nritems(leaf))
@@ -7958,7 +7987,11 @@ out:
 static struct bio *btrfs_dio_bio_alloc(struct block_device *bdev,
 				       u64 first_sector, gfp_t gfp_flags)
 {
-	return btrfs_bio_alloc(bdev, first_sector, BIO_MAX_PAGES, gfp_flags);
+	struct bio *bio;
+	bio = btrfs_bio_alloc(bdev, first_sector, BIO_MAX_PAGES, gfp_flags);
+	if (bio)
+		bio_associate_current(bio);
+	return bio;
 }
 
 static inline int btrfs_lookup_and_bind_dio_csum(struct btrfs_root *root,

fs/btrfs/ioctl.c

@@ -1030,6 +1030,7 @@ static int should_defrag_range(struct inode *inode, u64 start, u32 thresh,
 	struct extent_map *em;
 	int ret = 1;
 	bool next_mergeable = true;
+	bool prev_mergeable = true;
 
 	/*
 	 * make sure that once we start defragging an extent, we keep on
@@ -1050,13 +1051,16 @@ static int should_defrag_range(struct inode *inode, u64 start, u32 thresh,
 		goto out;
 	}
 
+	if (!*defrag_end)
+		prev_mergeable = false;
+
 	next_mergeable = defrag_check_next_extent(inode, em);
 	/*
 	 * we hit a real extent, if it is big or the next extent is not a
 	 * real extent, don't bother defragging it
 	 */
 	if (!compress && (*last_len == 0 || *last_len >= thresh) &&
-	    (em->len >= thresh || !next_mergeable))
+	    (em->len >= thresh || (!next_mergeable && !prev_mergeable)))
 		ret = 0;
 out:
 	/*
@@ -1933,6 +1937,7 @@ static noinline int copy_to_sk(struct btrfs_root *root,
 	u64 found_transid;
 	struct extent_buffer *leaf;
 	struct btrfs_ioctl_search_header sh;
+	struct btrfs_key test;
 	unsigned long item_off;
 	unsigned long item_len;
 	int nritems;
@@ -2016,12 +2021,17 @@ static noinline int copy_to_sk(struct btrfs_root *root,
 	}
 advance_key:
 	ret = 0;
-	if (key->offset < (u64)-1 && key->offset < sk->max_offset)
+	test.objectid = sk->max_objectid;
+	test.type = sk->max_type;
+	test.offset = sk->max_offset;
+	if (btrfs_comp_cpu_keys(key, &test) >= 0)
+		ret = 1;
+	else if (key->offset < (u64)-1)
 		key->offset++;
-	else if (key->type < (u8)-1 && key->type < sk->max_type) {
+	else if (key->type < (u8)-1) {
 		key->offset = 0;
 		key->type++;
-	} else if (key->objectid < (u64)-1 && key->objectid < sk->max_objectid) {
+	} else if (key->objectid < (u64)-1) {
 		key->offset = 0;
 		key->type = 0;
 		key->objectid++;
@@ -2842,8 +2852,7 @@ static void btrfs_double_inode_lock(struct inode *inode1, struct inode *inode2)
 		swap(inode1, inode2);
 
 	mutex_lock_nested(&inode1->i_mutex, I_MUTEX_PARENT);
-	if (inode1 != inode2)
-		mutex_lock_nested(&inode2->i_mutex, I_MUTEX_CHILD);
+	mutex_lock_nested(&inode2->i_mutex, I_MUTEX_CHILD);
 }
 
 static void btrfs_double_extent_unlock(struct inode *inode1, u64 loff1,
@@ -2861,8 +2870,7 @@ static void btrfs_double_extent_lock(struct inode *inode1, u64 loff1,
 		swap(loff1, loff2);
 	}
 	lock_extent_range(inode1, loff1, len);
-	if (inode1 != inode2)
-		lock_extent_range(inode2, loff2, len);
+	lock_extent_range(inode2, loff2, len);
 }
 
 struct cmp_pages {
@@ -3787,13 +3795,7 @@ static noinline long btrfs_ioctl_clone(struct file *file, unsigned long srcfd,
 		goto out_fput;
 
 	if (!same_inode) {
-		if (inode < src) {
-			mutex_lock_nested(&inode->i_mutex, I_MUTEX_PARENT);
-			mutex_lock_nested(&src->i_mutex, I_MUTEX_CHILD);
-		} else {
-			mutex_lock_nested(&src->i_mutex, I_MUTEX_PARENT);
-			mutex_lock_nested(&inode->i_mutex, I_MUTEX_CHILD);
-		}
+		btrfs_double_inode_lock(src, inode);
 	} else {
 		mutex_lock(&src->i_mutex);
 	}
@@ -3843,8 +3845,7 @@ static noinline long btrfs_ioctl_clone(struct file *file, unsigned long srcfd,
 
 		lock_extent_range(src, lock_start, lock_len);
 	} else {
-		lock_extent_range(src, off, len);
-		lock_extent_range(inode, destoff, len);
+		btrfs_double_extent_lock(src, off, inode, destoff, len);
 	}
 
 	ret = btrfs_clone(src, inode, off, olen, len, destoff, 0);
@@ -3855,9 +3856,7 @@ static noinline long btrfs_ioctl_clone(struct file *file, unsigned long srcfd,
 
 		unlock_extent(&BTRFS_I(src)->io_tree, lock_start, lock_end);
 	} else {
-		unlock_extent(&BTRFS_I(src)->io_tree, off, off + len - 1);
-		unlock_extent(&BTRFS_I(inode)->io_tree, destoff,
-			      destoff + len - 1);
+		btrfs_double_extent_unlock(src, off, inode, destoff, len);
 	}
 	/*
 	 * Truncate page cache pages so that future reads will see the cloned
@@ -3866,17 +3865,10 @@ static noinline long btrfs_ioctl_clone(struct file *file, unsigned long srcfd,
 	truncate_inode_pages_range(&inode->i_data, destoff,
 				   PAGE_CACHE_ALIGN(destoff + len) - 1);
 out_unlock:
-	if (!same_inode) {
-		if (inode < src) {
-			mutex_unlock(&src->i_mutex);
-			mutex_unlock(&inode->i_mutex);
-		} else {
-			mutex_unlock(&inode->i_mutex);
-			mutex_unlock(&src->i_mutex);
-		}
-	} else {
+	if (!same_inode)
+		btrfs_double_inode_unlock(src, inode);
+	else
 		mutex_unlock(&src->i_mutex);
-	}
 out_fput:
 	fdput(src_file);
 out_drop_write:

fs/btrfs/locking.c

@@ -241,6 +241,7 @@ void btrfs_tree_read_unlock_blocking(struct extent_buffer *eb)
  */
 void btrfs_tree_lock(struct extent_buffer *eb)
 {
+	WARN_ON(eb->lock_owner == current->pid);
 again:
 	wait_event(eb->read_lock_wq, atomic_read(&eb->blocking_readers) == 0);
 	wait_event(eb->write_lock_wq, atomic_read(&eb->blocking_writers) == 0);

fs/btrfs/raid56.c

@@ -61,9 +61,10 @@
 #define RBIO_CACHE_SIZE 1024
 
 enum btrfs_rbio_ops {
-	BTRFS_RBIO_WRITE	= 0,
-	BTRFS_RBIO_READ_REBUILD	= 1,
-	BTRFS_RBIO_PARITY_SCRUB	= 2,
+	BTRFS_RBIO_WRITE,
+	BTRFS_RBIO_READ_REBUILD,
+	BTRFS_RBIO_PARITY_SCRUB,
+	BTRFS_RBIO_REBUILD_MISSING,
 };
 
 struct btrfs_raid_bio {
@@ -602,6 +603,10 @@ static int rbio_can_merge(struct btrfs_raid_bio *last,
 	    cur->operation == BTRFS_RBIO_PARITY_SCRUB)
 		return 0;
 
+	if (last->operation == BTRFS_RBIO_REBUILD_MISSING ||
+	    cur->operation == BTRFS_RBIO_REBUILD_MISSING)
+		return 0;
+
 	return 1;
 }
 
@@ -793,7 +798,10 @@ static noinline void unlock_stripe(struct btrfs_raid_bio *rbio)
 
 			if (next->operation == BTRFS_RBIO_READ_REBUILD)
 				async_read_rebuild(next);
-			else if (next->operation == BTRFS_RBIO_WRITE) {
+			else if (next->operation == BTRFS_RBIO_REBUILD_MISSING) {
+				steal_rbio(rbio, next);
+				async_read_rebuild(next);
+			} else if (next->operation == BTRFS_RBIO_WRITE) {
 				steal_rbio(rbio, next);
 				async_rmw_stripe(next);
 			} else if (next->operation == BTRFS_RBIO_PARITY_SCRUB) {
@@ -1805,7 +1813,8 @@ static void __raid_recover_end_io(struct btrfs_raid_bio *rbio)
 	faila = rbio->faila;
 	failb = rbio->failb;
 
-	if (rbio->operation == BTRFS_RBIO_READ_REBUILD) {
+	if (rbio->operation == BTRFS_RBIO_READ_REBUILD ||
+	    rbio->operation == BTRFS_RBIO_REBUILD_MISSING) {
 		spin_lock_irq(&rbio->bio_list_lock);
 		set_bit(RBIO_RMW_LOCKED_BIT, &rbio->flags);
 		spin_unlock_irq(&rbio->bio_list_lock);
@@ -1830,7 +1839,8 @@ static void __raid_recover_end_io(struct btrfs_raid_bio *rbio)
 			 * if we're rebuilding a read, we have to use
 			 * pages from the bio list
 			 */
-			if (rbio->operation == BTRFS_RBIO_READ_REBUILD &&
+			if ((rbio->operation == BTRFS_RBIO_READ_REBUILD ||
+			     rbio->operation == BTRFS_RBIO_REBUILD_MISSING) &&
 			    (stripe == faila || stripe == failb)) {
 				page = page_in_rbio(rbio, stripe, pagenr, 0);
 			} else {
@@ -1939,7 +1949,8 @@ pstripe:
 			 * if we're rebuilding a read, we have to use
 			 * pages from the bio list
 			 */
-			if (rbio->operation == BTRFS_RBIO_READ_REBUILD &&
+			if ((rbio->operation == BTRFS_RBIO_READ_REBUILD ||
+			     rbio->operation == BTRFS_RBIO_REBUILD_MISSING) &&
 			    (stripe == faila || stripe == failb)) {
 				page = page_in_rbio(rbio, stripe, pagenr, 0);
 			} else {
@@ -1960,6 +1971,8 @@ cleanup_io:
 		else
 			clear_bit(RBIO_CACHE_READY_BIT, &rbio->flags);
 
+		rbio_orig_end_io(rbio, err);
+	} else if (rbio->operation == BTRFS_RBIO_REBUILD_MISSING) {
 		rbio_orig_end_io(rbio, err);
 	} else if (err == 0) {
 		rbio->faila = -1;
@@ -2096,7 +2109,8 @@ out:
 	return 0;
 
 cleanup:
-	if (rbio->operation == BTRFS_RBIO_READ_REBUILD)
+	if (rbio->operation == BTRFS_RBIO_READ_REBUILD ||
+	    rbio->operation == BTRFS_RBIO_REBUILD_MISSING)
 		rbio_orig_end_io(rbio, -EIO);
 	return -EIO;
 }
@@ -2227,8 +2241,9 @@ raid56_parity_alloc_scrub_rbio(struct btrfs_root *root, struct bio *bio,
 	return rbio;
 }
 
-void raid56_parity_add_scrub_pages(struct btrfs_raid_bio *rbio,
-				   struct page *page, u64 logical)
+/* Used for both parity scrub and missing. */
+void raid56_add_scrub_pages(struct btrfs_raid_bio *rbio, struct page *page,
+			    u64 logical)
 {
 	int stripe_offset;
 	int index;
@@ -2662,3 +2677,55 @@ void raid56_parity_submit_scrub_rbio(struct btrfs_raid_bio *rbio)
 	if (!lock_stripe_add(rbio))
 		async_scrub_parity(rbio);
 }
+
+/* The following code is used for dev replace of a missing RAID 5/6 device. */
+
+struct btrfs_raid_bio *
+raid56_alloc_missing_rbio(struct btrfs_root *root, struct bio *bio,
+			  struct btrfs_bio *bbio, u64 length)
+{
+	struct btrfs_raid_bio *rbio;
+
+	rbio = alloc_rbio(root, bbio, length);
+	if (IS_ERR(rbio))
+		return NULL;
+
+	rbio->operation = BTRFS_RBIO_REBUILD_MISSING;
+	bio_list_add(&rbio->bio_list, bio);
+	/*
+	 * This is a special bio which is used to hold the completion handler
+	 * and make the scrub rbio is similar to the other types
+	 */
+	ASSERT(!bio->bi_iter.bi_size);
+
+	rbio->faila = find_logical_bio_stripe(rbio, bio);
+	if (rbio->faila == -1) {
+		BUG();
+		kfree(rbio);
+		return NULL;
+	}
+
+	return rbio;
+}
+
+static void missing_raid56_work(struct btrfs_work *work)
+{
+	struct btrfs_raid_bio *rbio;
+
+	rbio = container_of(work, struct btrfs_raid_bio, work);
+	__raid56_parity_recover(rbio);
+}
+
+static void async_missing_raid56(struct btrfs_raid_bio *rbio)
+{
+	btrfs_init_work(&rbio->work, btrfs_rmw_helper,
+			missing_raid56_work, NULL, NULL);
+
+	btrfs_queue_work(rbio->fs_info->rmw_workers, &rbio->work);
+}
+
+void raid56_submit_missing_rbio(struct btrfs_raid_bio *rbio)
+{
+	if (!lock_stripe_add(rbio))
+		async_missing_raid56(rbio);
+}

fs/btrfs/raid56.h

@@ -48,15 +48,21 @@ int raid56_parity_recover(struct btrfs_root *root, struct bio *bio,
 int raid56_parity_write(struct btrfs_root *root, struct bio *bio,
 			       struct btrfs_bio *bbio, u64 stripe_len);
 
+void raid56_add_scrub_pages(struct btrfs_raid_bio *rbio, struct page *page,
+			    u64 logical);
+
 struct btrfs_raid_bio *
 raid56_parity_alloc_scrub_rbio(struct btrfs_root *root, struct bio *bio,
 			       struct btrfs_bio *bbio, u64 stripe_len,
 			       struct btrfs_device *scrub_dev,
 			       unsigned long *dbitmap, int stripe_nsectors);
-void raid56_parity_add_scrub_pages(struct btrfs_raid_bio *rbio,
-				   struct page *page, u64 logical);
 void raid56_parity_submit_scrub_rbio(struct btrfs_raid_bio *rbio);
 
+struct btrfs_raid_bio *
+raid56_alloc_missing_rbio(struct btrfs_root *root, struct bio *bio,
+			  struct btrfs_bio *bbio, u64 length);
+void raid56_submit_missing_rbio(struct btrfs_raid_bio *rbio);
+
 int btrfs_alloc_stripe_hash_table(struct btrfs_fs_info *info);
 void btrfs_free_stripe_hash_table(struct btrfs_fs_info *info);
 #endif

fs/btrfs/reada.c

@@ -328,6 +328,7 @@ static struct reada_extent *reada_find_extent(struct btrfs_root *root,
 	struct btrfs_device *prev_dev;
 	u32 blocksize;
 	u64 length;
+	int real_stripes;
 	int nzones = 0;
 	int i;
 	unsigned long index = logical >> PAGE_CACHE_SHIFT;
@@ -369,7 +370,8 @@ static struct reada_extent *reada_find_extent(struct btrfs_root *root,
 		goto error;
 	}
 
-	for (nzones = 0; nzones < bbio->num_stripes; ++nzones) {
+	real_stripes = bbio->num_stripes - bbio->num_tgtdevs;
+	for (nzones = 0; nzones < real_stripes; ++nzones) {
 		struct reada_zone *zone;
 
 		dev = bbio->stripes[nzones].dev;

fs/btrfs/relocation.c

@@ -2523,8 +2523,7 @@ struct btrfs_root *select_reloc_root(struct btrfs_trans_handle *trans,
  * counted. return -ENOENT if the block is root of reloc tree.
  */
 static noinline_for_stack
-struct btrfs_root *select_one_root(struct btrfs_trans_handle *trans,
-				   struct backref_node *node)
+struct btrfs_root *select_one_root(struct backref_node *node)
 {
 	struct backref_node *next;
 	struct btrfs_root *root;
@@ -2912,7 +2911,7 @@ static int relocate_tree_block(struct btrfs_trans_handle *trans,
 		return 0;
 
 	BUG_ON(node->processed);
-	root = select_one_root(trans, node);
+	root = select_one_root(node);
 	if (root == ERR_PTR(-ENOENT)) {
 		update_processed_blocks(rc, node);
 		goto out;
@@ -3755,8 +3754,7 @@ out:
  * helper to find next unprocessed extent
  */
 static noinline_for_stack
-int find_next_extent(struct btrfs_trans_handle *trans,
-		     struct reloc_control *rc, struct btrfs_path *path,
+int find_next_extent(struct reloc_control *rc, struct btrfs_path *path,
 		     struct btrfs_key *extent_key)
 {
 	struct btrfs_key key;
@@ -3951,7 +3949,7 @@ restart:
 			continue;
 		}
 
-		ret = find_next_extent(trans, rc, path, &key);
+		ret = find_next_extent(rc, path, &key);
 		if (ret < 0)
 			err = ret;
 		if (ret != 0)
@@ -3976,6 +3974,10 @@ restart:
 			       sizeof(struct btrfs_extent_item_v0));
 			ret = get_ref_objectid_v0(rc, path, &key, &ref_owner,
 						  &path_change);
+			if (ret < 0) {
+				err = ret;
+				break;
+			}
 			if (ref_owner < BTRFS_FIRST_FREE_OBJECTID)
 				flags = BTRFS_EXTENT_FLAG_TREE_BLOCK;
 			else
@@ -4140,7 +4142,7 @@ struct inode *create_reloc_inode(struct btrfs_fs_info *fs_info,
 	struct btrfs_trans_handle *trans;
 	struct btrfs_root *root;
 	struct btrfs_key key;
-	u64 objectid = BTRFS_FIRST_FREE_OBJECTID;
+	u64 objectid;
 	int err = 0;
 
 	root = read_fs_root(fs_info, BTRFS_DATA_RELOC_TREE_OBJECTID);
@@ -4215,14 +4217,12 @@ int btrfs_relocate_block_group(struct btrfs_root *extent_root, u64 group_start)
 	rc->block_group = btrfs_lookup_block_group(fs_info, group_start);
 	BUG_ON(!rc->block_group);
 
-	if (!rc->block_group->ro) {
-		ret = btrfs_set_block_group_ro(extent_root, rc->block_group);
-		if (ret) {
-			err = ret;
-			goto out;
-		}
-		rw = 1;
+	ret = btrfs_inc_block_group_ro(extent_root, rc->block_group);
+	if (ret) {
+		err = ret;
+		goto out;
 	}
+	rw = 1;
 
 	path = btrfs_alloc_path();
 	if (!path) {
@@ -4294,7 +4294,7 @@ int btrfs_relocate_block_group(struct btrfs_root *extent_root, u64 group_start)
 	WARN_ON(btrfs_block_group_used(&rc->block_group->item) > 0);
 out:
 	if (err && rw)
-		btrfs_set_block_group_rw(extent_root, rc->block_group);
+		btrfs_dec_block_group_ro(extent_root, rc->block_group);
 	iput(rc->data_inode);
 	btrfs_put_block_group(rc->block_group);
 	kfree(rc);
@@ -4594,8 +4594,7 @@ int btrfs_reloc_cow_block(struct btrfs_trans_handle *trans,
  * called before creating snapshot. it calculates metadata reservation
  * requried for relocating tree blocks in the snapshot
  */
-void btrfs_reloc_pre_snapshot(struct btrfs_trans_handle *trans,
-			      struct btrfs_pending_snapshot *pending,
+void btrfs_reloc_pre_snapshot(struct btrfs_pending_snapshot *pending,
 			      u64 *bytes_to_reserve)
 {
 	struct btrfs_root *root;

fs/btrfs/scrub.c

@@ -125,6 +125,7 @@ struct scrub_block {
 		/* It is for the data with checksum */
 		unsigned int	data_corrected:1;
 	};
+	struct btrfs_work	work;
 };
 
 /* Used for the chunks with parity stripe such RAID5/6 */
@@ -332,11 +333,14 @@ static void __scrub_blocked_if_needed(struct btrfs_fs_info *fs_info)
 	}
 }
 
-static void scrub_blocked_if_needed(struct btrfs_fs_info *fs_info)
+static void scrub_pause_on(struct btrfs_fs_info *fs_info)
 {
 	atomic_inc(&fs_info->scrubs_paused);
 	wake_up(&fs_info->scrub_pause_wait);
+}
 
+static void scrub_pause_off(struct btrfs_fs_info *fs_info)
+{
 	mutex_lock(&fs_info->scrub_lock);
 	__scrub_blocked_if_needed(fs_info);
 	atomic_dec(&fs_info->scrubs_paused);
@@ -345,6 +349,12 @@ static void scrub_blocked_if_needed(struct btrfs_fs_info *fs_info)
 	wake_up(&fs_info->scrub_pause_wait);
 }
 
+static void scrub_blocked_if_needed(struct btrfs_fs_info *fs_info)
+{
+	scrub_pause_on(fs_info);
+	scrub_pause_off(fs_info);
+}
+
 /*
  * used for workers that require transaction commits (i.e., for the
  * NOCOW case)
@@ -2074,21 +2084,7 @@ static void scrub_submit(struct scrub_ctx *sctx)
 	sbio = sctx->bios[sctx->curr];
 	sctx->curr = -1;
 	scrub_pending_bio_inc(sctx);
-
-	if (!sbio->bio->bi_bdev) {
-		/*
-		 * this case should not happen. If btrfs_map_block() is
-		 * wrong, it could happen for dev-replace operations on
-		 * missing devices when no mirrors are available, but in
-		 * this case it should already fail the mount.
-		 * This case is handled correctly (but _very_ slowly).
-		 */
-		printk_ratelimited(KERN_WARNING
-			"BTRFS: scrub_submit(bio bdev == NULL) is unexpected!\n");
-		bio_io_error(sbio->bio);
-	} else {
-		btrfsic_submit_bio(READ, sbio->bio);
-	}
+	btrfsic_submit_bio(READ, sbio->bio);
 }
 
 static int scrub_add_page_to_rd_bio(struct scrub_ctx *sctx,
@@ -2165,6 +2161,134 @@ again:
 	return 0;
 }
 
+static void scrub_missing_raid56_end_io(struct bio *bio)
+{
+	struct scrub_block *sblock = bio->bi_private;
+	struct btrfs_fs_info *fs_info = sblock->sctx->dev_root->fs_info;
+
+	if (bio->bi_error)
+		sblock->no_io_error_seen = 0;
+
+	btrfs_queue_work(fs_info->scrub_workers, &sblock->work);
+}
+
+static void scrub_missing_raid56_worker(struct btrfs_work *work)
+{
+	struct scrub_block *sblock = container_of(work, struct scrub_block, work);
+	struct scrub_ctx *sctx = sblock->sctx;
+	struct btrfs_fs_info *fs_info = sctx->dev_root->fs_info;
+	unsigned int is_metadata;
+	unsigned int have_csum;
+	u8 *csum;
+	u64 generation;
+	u64 logical;
+	struct btrfs_device *dev;
+
+	is_metadata = !(sblock->pagev[0]->flags & BTRFS_EXTENT_FLAG_DATA);
+	have_csum = sblock->pagev[0]->have_csum;
+	csum = sblock->pagev[0]->csum;
+	generation = sblock->pagev[0]->generation;
+	logical = sblock->pagev[0]->logical;
+	dev = sblock->pagev[0]->dev;
+
+	if (sblock->no_io_error_seen) {
+		scrub_recheck_block_checksum(fs_info, sblock, is_metadata,
+					     have_csum, csum, generation,
+					     sctx->csum_size);
+	}
+
+	if (!sblock->no_io_error_seen) {
+		spin_lock(&sctx->stat_lock);
+		sctx->stat.read_errors++;
+		spin_unlock(&sctx->stat_lock);
+		printk_ratelimited_in_rcu(KERN_ERR
+			"BTRFS: I/O error rebulding logical %llu for dev %s\n",
+			logical, rcu_str_deref(dev->name));
+	} else if (sblock->header_error || sblock->checksum_error) {
+		spin_lock(&sctx->stat_lock);
+		sctx->stat.uncorrectable_errors++;
+		spin_unlock(&sctx->stat_lock);
+		printk_ratelimited_in_rcu(KERN_ERR
+			"BTRFS: failed to rebuild valid logical %llu for dev %s\n",
+			logical, rcu_str_deref(dev->name));
+	} else {
+		scrub_write_block_to_dev_replace(sblock);
+	}
+
+	scrub_block_put(sblock);
+
+	if (sctx->is_dev_replace &&
+	    atomic_read(&sctx->wr_ctx.flush_all_writes)) {
+		mutex_lock(&sctx->wr_ctx.wr_lock);
+		scrub_wr_submit(sctx);
+		mutex_unlock(&sctx->wr_ctx.wr_lock);
+	}
+
+	scrub_pending_bio_dec(sctx);
+}
+
+static void scrub_missing_raid56_pages(struct scrub_block *sblock)
+{
+	struct scrub_ctx *sctx = sblock->sctx;
+	struct btrfs_fs_info *fs_info = sctx->dev_root->fs_info;
+	u64 length = sblock->page_count * PAGE_SIZE;
+	u64 logical = sblock->pagev[0]->logical;
+	struct btrfs_bio *bbio;
+	struct bio *bio;
+	struct btrfs_raid_bio *rbio;
+	int ret;
+	int i;
+
+	ret = btrfs_map_sblock(fs_info, REQ_GET_READ_MIRRORS, logical, &length,
+			       &bbio, 0, 1);
+	if (ret || !bbio || !bbio->raid_map)
+		goto bbio_out;
+
+	if (WARN_ON(!sctx->is_dev_replace ||
+		    !(bbio->map_type & BTRFS_BLOCK_GROUP_RAID56_MASK))) {
+		/*
+		 * We shouldn't be scrubbing a missing device. Even for dev
+		 * replace, we should only get here for RAID 5/6. We either
+		 * managed to mount something with no mirrors remaining or
+		 * there's a bug in scrub_remap_extent()/btrfs_map_block().
+		 */
+		goto bbio_out;
+	}
+
+	bio = btrfs_io_bio_alloc(GFP_NOFS, 0);
+	if (!bio)
+		goto bbio_out;
+
+	bio->bi_iter.bi_sector = logical >> 9;
+	bio->bi_private = sblock;
+	bio->bi_end_io = scrub_missing_raid56_end_io;
+
+	rbio = raid56_alloc_missing_rbio(sctx->dev_root, bio, bbio, length);
+	if (!rbio)
+		goto rbio_out;
+
+	for (i = 0; i < sblock->page_count; i++) {
+		struct scrub_page *spage = sblock->pagev[i];
+
+		raid56_add_scrub_pages(rbio, spage->page, spage->logical);
+	}
+
+	btrfs_init_work(&sblock->work, btrfs_scrub_helper,
+			scrub_missing_raid56_worker, NULL, NULL);
+	scrub_block_get(sblock);
+	scrub_pending_bio_inc(sctx);
+	raid56_submit_missing_rbio(rbio);
+	return;
+
+rbio_out:
+	bio_put(bio);
+bbio_out:
+	btrfs_put_bbio(bbio);
+	spin_lock(&sctx->stat_lock);
+	sctx->stat.malloc_errors++;
+	spin_unlock(&sctx->stat_lock);
+}
+
 static int scrub_pages(struct scrub_ctx *sctx, u64 logical, u64 len,
 		       u64 physical, struct btrfs_device *dev, u64 flags,
 		       u64 gen, int mirror_num, u8 *csum, int force,
@@ -2228,19 +2352,27 @@ leave_nomem:
 	}
 
 	WARN_ON(sblock->page_count == 0);
-	for (index = 0; index < sblock->page_count; index++) {
-		struct scrub_page *spage = sblock->pagev[index];
-		int ret;
+	if (dev->missing) {
+		/*
+		 * This case should only be hit for RAID 5/6 device replace. See
+		 * the comment in scrub_missing_raid56_pages() for details.
+		 */
+		scrub_missing_raid56_pages(sblock);
+	} else {
+		for (index = 0; index < sblock->page_count; index++) {
+			struct scrub_page *spage = sblock->pagev[index];
+			int ret;
 
-		ret = scrub_add_page_to_rd_bio(sctx, spage);
-		if (ret) {
-			scrub_block_put(sblock);
-			return ret;
+			ret = scrub_add_page_to_rd_bio(sctx, spage);
+			if (ret) {
+				scrub_block_put(sblock);
+				return ret;
+			}
 		}
-	}
 
-	if (force)
-		scrub_submit(sctx);
+		if (force)
+			scrub_submit(sctx);
+	}
 
 	/* last one frees, either here or in bio completion for last page */
 	scrub_block_put(sblock);
@@ -2551,6 +2683,11 @@ static int scrub_extent_for_parity(struct scrub_parity *sparity,
 	u8 csum[BTRFS_CSUM_SIZE];
 	u32 blocksize;
 
+	if (dev->missing) {
+		scrub_parity_mark_sectors_error(sparity, logical, len);
+		return 0;
+	}
+
 	if (flags & BTRFS_EXTENT_FLAG_DATA) {
 		blocksize = sctx->sectorsize;
 	} else if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) {
@@ -2689,7 +2826,7 @@ static void scrub_parity_check_and_repair(struct scrub_parity *sparity)
 			   sparity->nsectors))
 		goto out;
 
-	length = sparity->logic_end - sparity->logic_start + 1;
+	length = sparity->logic_end - sparity->logic_start;
 	ret = btrfs_map_sblock(sctx->dev_root->fs_info, WRITE,
 			       sparity->logic_start,
 			       &length, &bbio, 0, 1);
@@ -2712,8 +2849,7 @@ static void scrub_parity_check_and_repair(struct scrub_parity *sparity)
 		goto rbio_out;
 
 	list_for_each_entry(spage, &sparity->spages, list)
-		raid56_parity_add_scrub_pages(rbio, spage->page,
-					      spage->logical);
+		raid56_add_scrub_pages(rbio, spage->page, spage->logical);
 
 	scrub_pending_bio_inc(sctx);
 	raid56_parity_submit_scrub_rbio(rbio);
@@ -2761,6 +2897,7 @@ static noinline_for_stack int scrub_raid56_parity(struct scrub_ctx *sctx,
 	struct btrfs_root *root = fs_info->extent_root;
 	struct btrfs_root *csum_root = fs_info->csum_root;
 	struct btrfs_extent_item *extent;
+	struct btrfs_bio *bbio = NULL;
 	u64 flags;
 	int ret;
 	int slot;
@@ -2770,6 +2907,7 @@ static noinline_for_stack int scrub_raid56_parity(struct scrub_ctx *sctx,
 	u64 extent_logical;
 	u64 extent_physical;
 	u64 extent_len;
+	u64 mapped_length;
 	struct btrfs_device *extent_dev;
 	struct scrub_parity *sparity;
 	int nsectors;
@@ -2843,6 +2981,10 @@ static noinline_for_stack int scrub_raid56_parity(struct scrub_ctx *sctx,
 			}
 			btrfs_item_key_to_cpu(l, &key, slot);
 
+			if (key.type != BTRFS_EXTENT_ITEM_KEY &&
+			    key.type != BTRFS_METADATA_ITEM_KEY)
+				goto next;
+
 			if (key.type == BTRFS_METADATA_ITEM_KEY)
 				bytes = root->nodesize;
 			else
@@ -2851,11 +2993,7 @@ static noinline_for_stack int scrub_raid56_parity(struct scrub_ctx *sctx,
 			if (key.objectid + bytes <= logic_start)
 				goto next;
 
-			if (key.type != BTRFS_EXTENT_ITEM_KEY &&
-			    key.type != BTRFS_METADATA_ITEM_KEY)
-				goto next;
-
-			if (key.objectid > logic_end) {
+			if (key.objectid >= logic_end) {
 				stop_loop = 1;
 				break;
 			}
@@ -2868,11 +3006,12 @@ static noinline_for_stack int scrub_raid56_parity(struct scrub_ctx *sctx,
 			flags = btrfs_extent_flags(l, extent);
 			generation = btrfs_extent_generation(l, extent);
 
-			if (key.objectid < logic_start &&
-			    (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK)) {
-				btrfs_err(fs_info,
-					  "scrub: tree block %llu spanning stripes, ignored. logical=%llu",
-					   key.objectid, logic_start);
+			if ((flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) &&
+			    (key.objectid < logic_start ||
+			     key.objectid + bytes >
+			     logic_start + map->stripe_len)) {
+				btrfs_err(fs_info, "scrub: tree block %llu spanning stripes, ignored. logical=%llu",
+					  key.objectid, logic_start);
 				goto next;
 			}
 again:
@@ -2892,10 +3031,21 @@ again:
 			scrub_parity_mark_sectors_data(sparity, extent_logical,
 						       extent_len);
 
-			scrub_remap_extent(fs_info, extent_logical,
-					   extent_len, &extent_physical,
-					   &extent_dev,
-					   &extent_mirror_num);
+			mapped_length = extent_len;
+			ret = btrfs_map_block(fs_info, READ, extent_logical,
+					      &mapped_length, &bbio, 0);
+			if (!ret) {
+				if (!bbio || mapped_length < extent_len)
+					ret = -EIO;
+			}
+			if (ret) {
+				btrfs_put_bbio(bbio);
+				goto out;
+			}
+			extent_physical = bbio->stripes[0].physical;
+			extent_mirror_num = bbio->mirror_num;
+			extent_dev = bbio->stripes[0].dev;
+			btrfs_put_bbio(bbio);
 
 			ret = btrfs_lookup_csums_range(csum_root,
 						extent_logical,
@@ -2910,10 +3060,12 @@ again:
 						      extent_dev, flags,
 						      generation,
 						      extent_mirror_num);
+
+			scrub_free_csums(sctx);
+
 			if (ret)
 				goto out;
 
-			scrub_free_csums(sctx);
 			if (extent_logical + extent_len <
 			    key.objectid + bytes) {
 				logic_start += map->stripe_len;
@@ -2942,7 +3094,7 @@ next:
 out:
 	if (ret < 0)
 		scrub_parity_mark_sectors_error(sparity, logic_start,
-						logic_end - logic_start + 1);
+						logic_end - logic_start);
 	scrub_parity_put(sparity);
 	scrub_submit(sctx);
 	mutex_lock(&sctx->wr_ctx.wr_lock);
@@ -3091,22 +3243,6 @@ static noinline_for_stack int scrub_stripe(struct scrub_ctx *sctx,
 	 */
 	ret = 0;
 	while (physical < physical_end) {
-		/* for raid56, we skip parity stripe */
-		if (map->type & BTRFS_BLOCK_GROUP_RAID56_MASK) {
-			ret = get_raid56_logic_offset(physical, num,
-					map, &logical, &stripe_logical);
-			logical += base;
-			if (ret) {
-				stripe_logical += base;
-				stripe_end = stripe_logical + increment - 1;
-				ret = scrub_raid56_parity(sctx, map, scrub_dev,
-						ppath, stripe_logical,
-						stripe_end);
-				if (ret)
-					goto out;
-				goto skip;
-			}
-		}
 		/*
 		 * canceled?
 		 */
@@ -3131,6 +3267,24 @@ static noinline_for_stack int scrub_stripe(struct scrub_ctx *sctx,
 			scrub_blocked_if_needed(fs_info);
 		}
 
+		/* for raid56, we skip parity stripe */
+		if (map->type & BTRFS_BLOCK_GROUP_RAID56_MASK) {
+			ret = get_raid56_logic_offset(physical, num, map,
+						      &logical,
+						      &stripe_logical);
+			logical += base;
+			if (ret) {
+				stripe_logical += base;
+				stripe_end = stripe_logical + increment;
+				ret = scrub_raid56_parity(sctx, map, scrub_dev,
+							  ppath, stripe_logical,
+							  stripe_end);
+				if (ret)
+					goto out;
+				goto skip;
+			}
+		}
+
 		if (btrfs_fs_incompat(fs_info, SKINNY_METADATA))
 			key.type = BTRFS_METADATA_ITEM_KEY;
 		else
@@ -3175,6 +3329,10 @@ static noinline_for_stack int scrub_stripe(struct scrub_ctx *sctx,
 			}
 			btrfs_item_key_to_cpu(l, &key, slot);
 
+			if (key.type != BTRFS_EXTENT_ITEM_KEY &&
+			    key.type != BTRFS_METADATA_ITEM_KEY)
+				goto next;
+
 			if (key.type == BTRFS_METADATA_ITEM_KEY)
 				bytes = root->nodesize;
 			else
@@ -3183,10 +3341,6 @@ static noinline_for_stack int scrub_stripe(struct scrub_ctx *sctx,
 			if (key.objectid + bytes <= logical)
 				goto next;
 
-			if (key.type != BTRFS_EXTENT_ITEM_KEY &&
-			    key.type != BTRFS_METADATA_ITEM_KEY)
-				goto next;
-
 			if (key.objectid >= logical + map->stripe_len) {
 				/* out of this device extent */
 				if (key.objectid >= logic_end)
@@ -3199,8 +3353,10 @@ static noinline_for_stack int scrub_stripe(struct scrub_ctx *sctx,
 			flags = btrfs_extent_flags(l, extent);
 			generation = btrfs_extent_generation(l, extent);
 
-			if (key.objectid < logical &&
-			    (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK)) {
+			if ((flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) &&
+			    (key.objectid < logical ||
+			     key.objectid + bytes >
+			     logical + map->stripe_len)) {
 				btrfs_err(fs_info,
 					   "scrub: tree block %llu spanning "
 					   "stripes, ignored. logical=%llu",
@@ -3234,9 +3390,11 @@ again:
 						   &extent_dev,
 						   &extent_mirror_num);
 
-			ret = btrfs_lookup_csums_range(csum_root, logical,
-						logical + map->stripe_len - 1,
-						&sctx->csum_list, 1);
+			ret = btrfs_lookup_csums_range(csum_root,
+						       extent_logical,
+						       extent_logical +
+						       extent_len - 1,
+						       &sctx->csum_list, 1);
 			if (ret)
 				goto out;
 
@@ -3244,10 +3402,12 @@ again:
 					   extent_physical, extent_dev, flags,
 					   generation, extent_mirror_num,
 					   extent_logical - logical + physical);
+
+			scrub_free_csums(sctx);
+
 			if (ret)
 				goto out;
 
-			scrub_free_csums(sctx);
 			if (extent_logical + extent_len <
 			    key.objectid + bytes) {
 				if (map->type & BTRFS_BLOCK_GROUP_RAID56_MASK) {
@@ -3265,7 +3425,7 @@ loop:
 					if (ret && physical < physical_end) {
 						stripe_logical += base;
 						stripe_end = stripe_logical +
-								increment - 1;
+								increment;
 						ret = scrub_raid56_parity(sctx,
 							map, scrub_dev, ppath,
 							stripe_logical,
@@ -3374,7 +3534,7 @@ int scrub_enumerate_chunks(struct scrub_ctx *sctx,
 	u64 chunk_tree;
 	u64 chunk_objectid;
 	u64 chunk_offset;
-	int ret;
+	int ret = 0;
 	int slot;
 	struct extent_buffer *l;
 	struct btrfs_key key;
@@ -3402,8 +3562,14 @@ int scrub_enumerate_chunks(struct scrub_ctx *sctx,
 			if (path->slots[0] >=
 			    btrfs_header_nritems(path->nodes[0])) {
 				ret = btrfs_next_leaf(root, path);
-				if (ret)
+				if (ret < 0)
+					break;
+				if (ret > 0) {
+					ret = 0;
 					break;
+				}
+			} else {
+				ret = 0;
 			}
 		}
 
@@ -3445,6 +3611,22 @@ int scrub_enumerate_chunks(struct scrub_ctx *sctx,
 		if (!cache)
 			goto skip;
 
+		/*
+		 * we need call btrfs_inc_block_group_ro() with scrubs_paused,
+		 * to avoid deadlock caused by:
+		 * btrfs_inc_block_group_ro()
+		 * -> btrfs_wait_for_commit()
+		 * -> btrfs_commit_transaction()
+		 * -> btrfs_scrub_pause()
+		 */
+		scrub_pause_on(fs_info);
+		ret = btrfs_inc_block_group_ro(root, cache);
+		scrub_pause_off(fs_info);
+		if (ret) {
+			btrfs_put_block_group(cache);
+			break;
+		}
+
 		dev_replace->cursor_right = found_key.offset + length;
 		dev_replace->cursor_left = found_key.offset;
 		dev_replace->item_needs_writeback = 1;
@@ -3470,8 +3652,8 @@ int scrub_enumerate_chunks(struct scrub_ctx *sctx,
 
 		wait_event(sctx->list_wait,
 			   atomic_read(&sctx->bios_in_flight) == 0);
-		atomic_inc(&fs_info->scrubs_paused);
-		wake_up(&fs_info->scrub_pause_wait);
+
+		scrub_pause_on(fs_info);
 
 		/*
 		 * must be called before we decrease @scrub_paused.
@@ -3482,11 +3664,9 @@ int scrub_enumerate_chunks(struct scrub_ctx *sctx,
 			   atomic_read(&sctx->workers_pending) == 0);
 		atomic_set(&sctx->wr_ctx.flush_all_writes, 0);
 
-		mutex_lock(&fs_info->scrub_lock);
-		__scrub_blocked_if_needed(fs_info);
-		atomic_dec(&fs_info->scrubs_paused);
-		mutex_unlock(&fs_info->scrub_lock);
-		wake_up(&fs_info->scrub_pause_wait);
+		scrub_pause_off(fs_info);
+
+		btrfs_dec_block_group_ro(root, cache);
 
 		btrfs_put_block_group(cache);
 		if (ret)
@@ -3510,11 +3690,7 @@ skip:
 
 	btrfs_free_path(path);
 
-	/*
-	 * ret can still be 1 from search_slot or next_leaf,
-	 * that's not an error
-	 */
-	return ret < 0 ? ret : 0;
+	return ret;
 }
 
 static noinline_for_stack int scrub_supers(struct scrub_ctx *sctx,

fs/btrfs/super.c

@@ -69,7 +69,7 @@ static struct file_system_type btrfs_fs_type;
 
 static int btrfs_remount(struct super_block *sb, int *flags, char *data);
 
-static const char *btrfs_decode_error(int errno)
+const char *btrfs_decode_error(int errno)
 {
 	char *errstr = "unknown";
 
@@ -1033,6 +1033,7 @@ static int btrfs_fill_super(struct super_block *sb,
 	sb->s_flags |= MS_POSIXACL;
 #endif
 	sb->s_flags |= MS_I_VERSION;
+	sb->s_iflags |= SB_I_CGROUPWB;
 	err = open_ctree(sb, fs_devices, (char *)data);
 	if (err) {
 		printk(KERN_ERR "BTRFS: open_ctree failed\n");
@@ -1650,6 +1651,17 @@ static int btrfs_remount(struct super_block *sb, int *flags, char *data)
 
 		sb->s_flags |= MS_RDONLY;
 
+		/*
+		 * Setting MS_RDONLY will put the cleaner thread to
+		 * sleep at the next loop if it's already active.
+		 * If it's already asleep, we'll leave unused block
+		 * groups on disk until we're mounted read-write again
+		 * unless we clean them up here.
+		 */
+		mutex_lock(&root->fs_info->cleaner_mutex);
+		btrfs_delete_unused_bgs(fs_info);
+		mutex_unlock(&root->fs_info->cleaner_mutex);
+
 		btrfs_dev_replace_suspend_for_unmount(fs_info);
 		btrfs_scrub_cancel(fs_info);
 		btrfs_pause_balance(fs_info);

fs/btrfs/transaction.c

@@ -258,6 +258,8 @@ loop:
 	mutex_init(&cur_trans->cache_write_mutex);
 	cur_trans->num_dirty_bgs = 0;
 	spin_lock_init(&cur_trans->dirty_bgs_lock);
+	INIT_LIST_HEAD(&cur_trans->deleted_bgs);
+	spin_lock_init(&cur_trans->deleted_bgs_lock);
 	list_add_tail(&cur_trans->list, &fs_info->trans_list);
 	extent_io_tree_init(&cur_trans->dirty_pages,
 			     fs_info->btree_inode->i_mapping);
@@ -1301,7 +1303,7 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 	 */
 	btrfs_set_skip_qgroup(trans, objectid);
 
-	btrfs_reloc_pre_snapshot(trans, pending, &to_reserve);
+	btrfs_reloc_pre_snapshot(pending, &to_reserve);
 
 	if (to_reserve > 0) {
 		pending->error = btrfs_block_rsv_add(root,
@@ -1893,8 +1895,11 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
 			spin_unlock(&root->fs_info->trans_lock);
 
 			wait_for_commit(root, prev_trans);
+			ret = prev_trans->aborted;
 
 			btrfs_put_transaction(prev_trans);
+			if (ret)
+				goto cleanup_transaction;
 		} else {
 			spin_unlock(&root->fs_info->trans_lock);
 		}

fs/btrfs/transaction.h

@@ -74,6 +74,8 @@ struct btrfs_transaction {
 	 */
 	struct mutex cache_write_mutex;
 	spinlock_t dirty_bgs_lock;
+	struct list_head deleted_bgs;
+	spinlock_t deleted_bgs_lock;
 	struct btrfs_delayed_ref_root delayed_refs;
 	int aborted;
 	int dirty_bg_run;

fs/btrfs/tree-log.c

@@ -140,55 +140,46 @@ static int start_log_trans(struct btrfs_trans_handle *trans,
 			   struct btrfs_root *root,
 			   struct btrfs_log_ctx *ctx)
 {
-	int index;
-	int ret;
+	int ret = 0;
 
 	mutex_lock(&root->log_mutex);
+
 	if (root->log_root) {
 		if (btrfs_need_log_full_commit(root->fs_info, trans)) {
 			ret = -EAGAIN;
 			goto out;
 		}
+
 		if (!root->log_start_pid) {
-			root->log_start_pid = current->pid;
 			clear_bit(BTRFS_ROOT_MULTI_LOG_TASKS, &root->state);
+			root->log_start_pid = current->pid;
 		} else if (root->log_start_pid != current->pid) {
 			set_bit(BTRFS_ROOT_MULTI_LOG_TASKS, &root->state);
 		}
+	} else {
+		mutex_lock(&root->fs_info->tree_log_mutex);
+		if (!root->fs_info->log_root_tree)
+			ret = btrfs_init_log_root_tree(trans, root->fs_info);
+		mutex_unlock(&root->fs_info->tree_log_mutex);
+		if (ret)
+			goto out;
 
-		atomic_inc(&root->log_batch);
-		atomic_inc(&root->log_writers);
-		if (ctx) {
-			index = root->log_transid % 2;
-			list_add_tail(&ctx->list, &root->log_ctxs[index]);
-			ctx->log_transid = root->log_transid;
-		}
-		mutex_unlock(&root->log_mutex);
-		return 0;
-	}
-
-	ret = 0;
-	mutex_lock(&root->fs_info->tree_log_mutex);
-	if (!root->fs_info->log_root_tree)
-		ret = btrfs_init_log_root_tree(trans, root->fs_info);
-	mutex_unlock(&root->fs_info->tree_log_mutex);
-	if (ret)
-		goto out;
-
-	if (!root->log_root) {
 		ret = btrfs_add_log_tree(trans, root);
 		if (ret)
 			goto out;
+
+		clear_bit(BTRFS_ROOT_MULTI_LOG_TASKS, &root->state);
+		root->log_start_pid = current->pid;
 	}
-	clear_bit(BTRFS_ROOT_MULTI_LOG_TASKS, &root->state);
-	root->log_start_pid = current->pid;
+
 	atomic_inc(&root->log_batch);
 	atomic_inc(&root->log_writers);
 	if (ctx) {
-		index = root->log_transid % 2;
+		int index = root->log_transid % 2;
 		list_add_tail(&ctx->list, &root->log_ctxs[index]);
 		ctx->log_transid = root->log_transid;
 	}
+
 out:
 	mutex_unlock(&root->log_mutex);
 	return ret;
@@ -731,11 +722,65 @@ static noinline int replay_one_extent(struct btrfs_trans_handle *trans,
 						&ordered_sums, 0);
 			if (ret)
 				goto out;
+			/*
+			 * Now delete all existing cums in the csum root that
+			 * cover our range. We do this because we can have an
+			 * extent that is completely referenced by one file
+			 * extent item and partially referenced by another
+			 * file extent item (like after using the clone or
+			 * extent_same ioctls). In this case if we end up doing
+			 * the replay of the one that partially references the
+			 * extent first, and we do not do the csum deletion
+			 * below, we can get 2 csum items in the csum tree that
+			 * overlap each other. For example, imagine our log has
+			 * the two following file extent items:
+			 *
+			 * key (257 EXTENT_DATA 409600)
+			 *     extent data disk byte 12845056 nr 102400
+			 *     extent data offset 20480 nr 20480 ram 102400
+			 *
+			 * key (257 EXTENT_DATA 819200)
+			 *     extent data disk byte 12845056 nr 102400
+			 *     extent data offset 0 nr 102400 ram 102400
+			 *
+			 * Where the second one fully references the 100K extent
+			 * that starts at disk byte 12845056, and the log tree
+			 * has a single csum item that covers the entire range
+			 * of the extent:
+			 *
+			 * key (EXTENT_CSUM EXTENT_CSUM 12845056) itemsize 100
+			 *
+			 * After the first file extent item is replayed, the
+			 * csum tree gets the following csum item:
+			 *
+			 * key (EXTENT_CSUM EXTENT_CSUM 12865536) itemsize 20
+			 *
+			 * Which covers the 20K sub-range starting at offset 20K
+			 * of our extent. Now when we replay the second file
+			 * extent item, if we do not delete existing csum items
+			 * that cover any of its blocks, we end up getting two
+			 * csum items in our csum tree that overlap each other:
+			 *
+			 * key (EXTENT_CSUM EXTENT_CSUM 12845056) itemsize 100
+			 * key (EXTENT_CSUM EXTENT_CSUM 12865536) itemsize 20
+			 *
+			 * Which is a problem, because after this anyone trying
+			 * to lookup up for the checksum of any block of our
+			 * extent starting at an offset of 40K or higher, will
+			 * end up looking at the second csum item only, which
+			 * does not contain the checksum for any block starting
+			 * at offset 40K or higher of our extent.
+			 */
 			while (!list_empty(&ordered_sums)) {
 				struct btrfs_ordered_sum *sums;
 				sums = list_entry(ordered_sums.next,
 						struct btrfs_ordered_sum,
 						list);
+				if (!ret)
+					ret = btrfs_del_csums(trans,
+						      root->fs_info->csum_root,
+						      sums->bytenr,
+						      sums->len);
 				if (!ret)
 					ret = btrfs_csum_file_blocks(trans,
 						root->fs_info->csum_root,
@@ -1549,9 +1594,8 @@ static noinline int link_to_fixup_dir(struct btrfs_trans_handle *trans,
  */
 static noinline int insert_one_name(struct btrfs_trans_handle *trans,
 				    struct btrfs_root *root,
-				    struct btrfs_path *path,
 				    u64 dirid, u64 index,
-				    char *name, int name_len, u8 type,
+				    char *name, int name_len,
 				    struct btrfs_key *location)
 {
 	struct inode *inode;
@@ -1613,6 +1657,9 @@ static bool name_in_log_ref(struct btrfs_root *log_root,
  * not exist in the FS, it is skipped.  fsyncs on directories
  * do not force down inodes inside that directory, just changes to the
  * names or unlinks in a directory.
+ *
+ * Returns < 0 on error, 0 if the name wasn't replayed (dentry points to a
+ * non-existing inode) and 1 if the name was replayed.
  */
 static noinline int replay_one_name(struct btrfs_trans_handle *trans,
 				    struct btrfs_root *root,
@@ -1631,6 +1678,7 @@ static noinline int replay_one_name(struct btrfs_trans_handle *trans,
 	int exists;
 	int ret = 0;
 	bool update_size = (key->type == BTRFS_DIR_INDEX_KEY);
+	bool name_added = false;
 
 	dir = read_one_inode(root, key->objectid);
 	if (!dir)
@@ -1708,6 +1756,8 @@ out:
 	}
 	kfree(name);
 	iput(dir);
+	if (!ret && name_added)
+		ret = 1;
 	return ret;
 
 insert:
@@ -1719,10 +1769,12 @@ insert:
 		goto out;
 	}
 	btrfs_release_path(path);
-	ret = insert_one_name(trans, root, path, key->objectid, key->offset,
-			      name, name_len, log_type, &log_key);
+	ret = insert_one_name(trans, root, key->objectid, key->offset,
+			      name, name_len, &log_key);
 	if (ret && ret != -ENOENT && ret != -EEXIST)
 		goto out;
+	if (!ret)
+		name_added = true;
 	update_size = false;
 	ret = 0;
 	goto out;
@@ -1740,12 +1792,13 @@ static noinline int replay_one_dir_item(struct btrfs_trans_handle *trans,
 					struct extent_buffer *eb, int slot,
 					struct btrfs_key *key)
 {
-	int ret;
+	int ret = 0;
 	u32 item_size = btrfs_item_size_nr(eb, slot);
 	struct btrfs_dir_item *di;
 	int name_len;
 	unsigned long ptr;
 	unsigned long ptr_end;
+	struct btrfs_path *fixup_path = NULL;
 
 	ptr = btrfs_item_ptr_offset(eb, slot);
 	ptr_end = ptr + item_size;
@@ -1755,12 +1808,59 @@ static noinline int replay_one_dir_item(struct btrfs_trans_handle *trans,
 			return -EIO;
 		name_len = btrfs_dir_name_len(eb, di);
 		ret = replay_one_name(trans, root, path, eb, di, key);
-		if (ret)
-			return ret;
+		if (ret < 0)
+			break;
 		ptr = (unsigned long)(di + 1);
 		ptr += name_len;
+
+		/*
+		 * If this entry refers to a non-directory (directories can not
+		 * have a link count > 1) and it was added in the transaction
+		 * that was not committed, make sure we fixup the link count of
+		 * the inode it the entry points to. Otherwise something like
+		 * the following would result in a directory pointing to an
+		 * inode with a wrong link that does not account for this dir
+		 * entry:
+		 *
+		 * mkdir testdir
+		 * touch testdir/foo
+		 * touch testdir/bar
+		 * sync
+		 *
+		 * ln testdir/bar testdir/bar_link
+		 * ln testdir/foo testdir/foo_link
+		 * xfs_io -c "fsync" testdir/bar
+		 *
+		 * <power failure>
+		 *
+		 * mount fs, log replay happens
+		 *
+		 * File foo would remain with a link count of 1 when it has two
+		 * entries pointing to it in the directory testdir. This would
+		 * make it impossible to ever delete the parent directory has
+		 * it would result in stale dentries that can never be deleted.
+		 */
+		if (ret == 1 && btrfs_dir_type(eb, di) != BTRFS_FT_DIR) {
+			struct btrfs_key di_key;
+
+			if (!fixup_path) {
+				fixup_path = btrfs_alloc_path();
+				if (!fixup_path) {
+					ret = -ENOMEM;
+					break;
+				}
+			}
+
+			btrfs_dir_item_key_to_cpu(eb, di, &di_key);
+			ret = link_to_fixup_dir(trans, root, fixup_path,
+						di_key.objectid);
+			if (ret)
+				break;
+		}
+		ret = 0;
 	}
-	return 0;
+	btrfs_free_path(fixup_path);
+	return ret;
 }
 
 /*
@@ -2535,8 +2635,7 @@ static int update_log_root(struct btrfs_trans_handle *trans,
 	return ret;
 }
 
-static void wait_log_commit(struct btrfs_trans_handle *trans,
-			    struct btrfs_root *root, int transid)
+static void wait_log_commit(struct btrfs_root *root, int transid)
 {
 	DEFINE_WAIT(wait);
 	int index = transid % 2;
@@ -2561,8 +2660,7 @@ static void wait_log_commit(struct btrfs_trans_handle *trans,
 		 atomic_read(&root->log_commit[index]));
 }
 
-static void wait_for_writer(struct btrfs_trans_handle *trans,
-			    struct btrfs_root *root)
+static void wait_for_writer(struct btrfs_root *root)
 {
 	DEFINE_WAIT(wait);
 
@@ -2642,7 +2740,7 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans,
 
 	index1 = log_transid % 2;
 	if (atomic_read(&root->log_commit[index1])) {
-		wait_log_commit(trans, root, log_transid);
+		wait_log_commit(root, log_transid);
 		mutex_unlock(&root->log_mutex);
 		return ctx->log_ret;
 	}
@@ -2651,7 +2749,7 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans,
 
 	/* wait for previous tree log sync to complete */
 	if (atomic_read(&root->log_commit[(index1 + 1) % 2]))
-		wait_log_commit(trans, root, log_transid - 1);
+		wait_log_commit(root, log_transid - 1);
 
 	while (1) {
 		int batch = atomic_read(&root->log_batch);
@@ -2662,7 +2760,7 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans,
 			schedule_timeout_uninterruptible(1);
 			mutex_lock(&root->log_mutex);
 		}
-		wait_for_writer(trans, root);
+		wait_for_writer(root);
 		if (batch == atomic_read(&root->log_batch))
 			break;
 	}
@@ -2759,7 +2857,7 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans,
 		ret = btrfs_wait_marked_extents(log, &log->dirty_log_pages,
 						mark);
 		btrfs_wait_logged_extents(trans, log, log_transid);
-		wait_log_commit(trans, log_root_tree,
+		wait_log_commit(log_root_tree,
 				root_log_ctx.log_transid);
 		mutex_unlock(&log_root_tree->log_mutex);
 		if (!ret)
@@ -2770,11 +2868,11 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans,
 	atomic_set(&log_root_tree->log_commit[index2], 1);
 
 	if (atomic_read(&log_root_tree->log_commit[(index2 + 1) % 2])) {
-		wait_log_commit(trans, log_root_tree,
+		wait_log_commit(log_root_tree,
 				root_log_ctx.log_transid - 1);
 	}
 
-	wait_for_writer(trans, log_root_tree);
+	wait_for_writer(log_root_tree);
 
 	/*
 	 * now that we've moved on to the tree of log tree roots,
@@ -4904,6 +5002,94 @@ next_dir_inode:
 	return ret;
 }
 
+static int btrfs_log_all_parents(struct btrfs_trans_handle *trans,
+				 struct inode *inode,
+				 struct btrfs_log_ctx *ctx)
+{
+	int ret;
+	struct btrfs_path *path;
+	struct btrfs_key key;
+	struct btrfs_root *root = BTRFS_I(inode)->root;
+	const u64 ino = btrfs_ino(inode);
+
+	path = btrfs_alloc_path();
+	if (!path)
+		return -ENOMEM;
+	path->skip_locking = 1;
+	path->search_commit_root = 1;
+
+	key.objectid = ino;
+	key.type = BTRFS_INODE_REF_KEY;
+	key.offset = 0;
+	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
+	if (ret < 0)
+		goto out;
+
+	while (true) {
+		struct extent_buffer *leaf = path->nodes[0];
+		int slot = path->slots[0];
+		u32 cur_offset = 0;
+		u32 item_size;
+		unsigned long ptr;
+
+		if (slot >= btrfs_header_nritems(leaf)) {
+			ret = btrfs_next_leaf(root, path);
+			if (ret < 0)
+				goto out;
+			else if (ret > 0)
+				break;
+			continue;
+		}
+
+		btrfs_item_key_to_cpu(leaf, &key, slot);
+		/* BTRFS_INODE_EXTREF_KEY is BTRFS_INODE_REF_KEY + 1 */
+		if (key.objectid != ino || key.type > BTRFS_INODE_EXTREF_KEY)
+			break;
+
+		item_size = btrfs_item_size_nr(leaf, slot);
+		ptr = btrfs_item_ptr_offset(leaf, slot);
+		while (cur_offset < item_size) {
+			struct btrfs_key inode_key;
+			struct inode *dir_inode;
+
+			inode_key.type = BTRFS_INODE_ITEM_KEY;
+			inode_key.offset = 0;
+
+			if (key.type == BTRFS_INODE_EXTREF_KEY) {
+				struct btrfs_inode_extref *extref;
+
+				extref = (struct btrfs_inode_extref *)
+					(ptr + cur_offset);
+				inode_key.objectid = btrfs_inode_extref_parent(
+					leaf, extref);
+				cur_offset += sizeof(*extref);
+				cur_offset += btrfs_inode_extref_name_len(leaf,
+					extref);
+			} else {
+				inode_key.objectid = key.offset;
+				cur_offset = item_size;
+			}
+
+			dir_inode = btrfs_iget(root->fs_info->sb, &inode_key,
+					       root, NULL);
+			/* If parent inode was deleted, skip it. */
+			if (IS_ERR(dir_inode))
+				continue;
+
+			ret = btrfs_log_inode(trans, root, dir_inode,
+					      LOG_INODE_ALL, 0, LLONG_MAX, ctx);
+			iput(dir_inode);
+			if (ret)
+				goto out;
+		}
+		path->slots[0]++;
+	}
+	ret = 0;
+out:
+	btrfs_free_path(path);
+	return ret;
+}
+
 /*
  * helper function around btrfs_log_inode to make sure newly created
  * parent directories also end up in the log.  A minimal inode and backref
@@ -4923,9 +5109,6 @@ static int btrfs_log_inode_parent(struct btrfs_trans_handle *trans,
 	struct dentry *old_parent = NULL;
 	int ret = 0;
 	u64 last_committed = root->fs_info->last_trans_committed;
-	const struct dentry * const first_parent = parent;
-	const bool did_unlink = (BTRFS_I(inode)->last_unlink_trans >
-				 last_committed);
 	bool log_dentries = false;
 	struct inode *orig_inode = inode;
 
@@ -4986,6 +5169,53 @@ static int btrfs_log_inode_parent(struct btrfs_trans_handle *trans,
 	if (S_ISDIR(inode->i_mode) && ctx && ctx->log_new_dentries)
 		log_dentries = true;
 
+	/*
+	 * On unlink we must make sure all our current and old parent directores
+	 * inodes are fully logged. This is to prevent leaving dangling
+	 * directory index entries in directories that were our parents but are
+	 * not anymore. Not doing this results in old parent directory being
+	 * impossible to delete after log replay (rmdir will always fail with
+	 * error -ENOTEMPTY).
+	 *
+	 * Example 1:
+	 *
+	 * mkdir testdir
+	 * touch testdir/foo
+	 * ln testdir/foo testdir/bar
+	 * sync
+	 * unlink testdir/bar
+	 * xfs_io -c fsync testdir/foo
+	 * <power failure>
+	 * mount fs, triggers log replay
+	 *
+	 * If we don't log the parent directory (testdir), after log replay the
+	 * directory still has an entry pointing to the file inode using the bar
+	 * name, but a matching BTRFS_INODE_[REF|EXTREF]_KEY does not exist and
+	 * the file inode has a link count of 1.
+	 *
+	 * Example 2:
+	 *
+	 * mkdir testdir
+	 * touch foo
+	 * ln foo testdir/foo2
+	 * ln foo testdir/foo3
+	 * sync
+	 * unlink testdir/foo3
+	 * xfs_io -c fsync foo
+	 * <power failure>
+	 * mount fs, triggers log replay
+	 *
+	 * Similar as the first example, after log replay the parent directory
+	 * testdir still has an entry pointing to the inode file with name foo3
+	 * but the file inode does not have a matching BTRFS_INODE_REF_KEY item
+	 * and has a link count of 2.
+	 */
+	if (BTRFS_I(inode)->last_unlink_trans > last_committed) {
+		ret = btrfs_log_all_parents(trans, orig_inode, ctx);
+		if (ret)
+			goto end_trans;
+	}
+
 	while (1) {
 		if (!parent || d_really_is_negative(parent) || sb != d_inode(parent)->i_sb)
 			break;
@@ -4994,23 +5224,9 @@ static int btrfs_log_inode_parent(struct btrfs_trans_handle *trans,
 		if (root != BTRFS_I(inode)->root)
 			break;
 
-		/*
-		 * On unlink we must make sure our immediate parent directory
-		 * inode is fully logged. This is to prevent leaving dangling
-		 * directory index entries and a wrong directory inode's i_size.
-		 * Not doing so can result in a directory being impossible to
-		 * delete after log replay (rmdir will always fail with error
-		 * -ENOTEMPTY).
-		 */
-		if (did_unlink && parent == first_parent)
-			inode_only = LOG_INODE_ALL;
-		else
-			inode_only = LOG_INODE_EXISTS;
-
-		if (BTRFS_I(inode)->generation >
-		    root->fs_info->last_trans_committed ||
-		    inode_only == LOG_INODE_ALL) {
-			ret = btrfs_log_inode(trans, root, inode, inode_only,
+		if (BTRFS_I(inode)->generation > last_committed) {
+			ret = btrfs_log_inode(trans, root, inode,
+					      LOG_INODE_EXISTS,
 					      0, LLONG_MAX, ctx);
 			if (ret)
 				goto end_trans;

fs/btrfs/volumes.c

@@ -1116,15 +1116,18 @@ out:
 	return ret;
 }
 
-static int contains_pending_extent(struct btrfs_trans_handle *trans,
+static int contains_pending_extent(struct btrfs_transaction *transaction,
 				   struct btrfs_device *device,
 				   u64 *start, u64 len)
 {
+	struct btrfs_fs_info *fs_info = device->dev_root->fs_info;
 	struct extent_map *em;
-	struct list_head *search_list = &trans->transaction->pending_chunks;
+	struct list_head *search_list = &fs_info->pinned_chunks;
 	int ret = 0;
 	u64 physical_start = *start;
 
+	if (transaction)
+		search_list = &transaction->pending_chunks;
 again:
 	list_for_each_entry(em, search_list, list) {
 		struct map_lookup *map;
@@ -1159,8 +1162,8 @@ again:
 			}
 		}
 	}
-	if (search_list == &trans->transaction->pending_chunks) {
-		search_list = &trans->root->fs_info->pinned_chunks;
+	if (search_list != &fs_info->pinned_chunks) {
+		search_list = &fs_info->pinned_chunks;
 		goto again;
 	}
 
@@ -1169,12 +1172,13 @@ again:
 
 
 /*
- * find_free_dev_extent - find free space in the specified device
- * @device:	the device which we search the free space in
- * @num_bytes:	the size of the free space that we need
- * @start:	store the start of the free space.
- * @len:	the size of the free space. that we find, or the size of the max
- * 		free space if we don't find suitable free space
+ * find_free_dev_extent_start - find free space in the specified device
+ * @device:	  the device which we search the free space in
+ * @num_bytes:	  the size of the free space that we need
+ * @search_start: the position from which to begin the search
+ * @start:	  store the start of the free space.
+ * @len:	  the size of the free space. that we find, or the size
+ *		  of the max free space if we don't find suitable free space
  *
  * this uses a pretty simple search, the expectation is that it is
  * called very infrequently and that a given device has a small number
@@ -1188,9 +1192,9 @@ again:
  * But if we don't find suitable free space, it is used to store the size of
  * the max free space.
  */
-int find_free_dev_extent(struct btrfs_trans_handle *trans,
-			 struct btrfs_device *device, u64 num_bytes,
-			 u64 *start, u64 *len)
+int find_free_dev_extent_start(struct btrfs_transaction *transaction,
+			       struct btrfs_device *device, u64 num_bytes,
+			       u64 search_start, u64 *start, u64 *len)
 {
 	struct btrfs_key key;
 	struct btrfs_root *root = device->dev_root;
@@ -1200,19 +1204,11 @@ int find_free_dev_extent(struct btrfs_trans_handle *trans,
 	u64 max_hole_start;
 	u64 max_hole_size;
 	u64 extent_end;
-	u64 search_start;
 	u64 search_end = device->total_bytes;
 	int ret;
 	int slot;
 	struct extent_buffer *l;
 
-	/* FIXME use last free of some kind */
-
-	/* we don't want to overwrite the superblock on the drive,
-	 * so we make sure to start at an offset of at least 1MB
-	 */
-	search_start = max(root->fs_info->alloc_start, 1024ull * 1024);
-
 	path = btrfs_alloc_path();
 	if (!path)
 		return -ENOMEM;
@@ -1273,7 +1269,7 @@ again:
 			 * Have to check before we set max_hole_start, otherwise
 			 * we could end up sending back this offset anyway.
 			 */
-			if (contains_pending_extent(trans, device,
+			if (contains_pending_extent(transaction, device,
 						    &search_start,
 						    hole_size)) {
 				if (key.offset >= search_start) {
@@ -1322,7 +1318,7 @@ next:
 	if (search_end > search_start) {
 		hole_size = search_end - search_start;
 
-		if (contains_pending_extent(trans, device, &search_start,
+		if (contains_pending_extent(transaction, device, &search_start,
 					    hole_size)) {
 			btrfs_release_path(path);
 			goto again;
@@ -1348,6 +1344,24 @@ out:
 	return ret;
 }
 
+int find_free_dev_extent(struct btrfs_trans_handle *trans,
+			 struct btrfs_device *device, u64 num_bytes,
+			 u64 *start, u64 *len)
+{
+	struct btrfs_root *root = device->dev_root;
+	u64 search_start;
+
+	/* FIXME use last free of some kind */
+
+	/*
+	 * we don't want to overwrite the superblock on the drive,
+	 * so we make sure to start at an offset of at least 1MB
+	 */
+	search_start = max(root->fs_info->alloc_start, 1024ull * 1024);
+	return find_free_dev_extent_start(trans->transaction, device,
+					  num_bytes, search_start, start, len);
+}
+
 static int btrfs_free_dev_extent(struct btrfs_trans_handle *trans,
 			  struct btrfs_device *device,
 			  u64 start, u64 *dev_extent_len)
@@ -2755,9 +2769,7 @@ out:
 	return ret;
 }
 
-static int btrfs_relocate_chunk(struct btrfs_root *root,
-				u64 chunk_objectid,
-				u64 chunk_offset)
+static int btrfs_relocate_chunk(struct btrfs_root *root, u64 chunk_offset)
 {
 	struct btrfs_root *extent_root;
 	struct btrfs_trans_handle *trans;
@@ -2785,7 +2797,9 @@ static int btrfs_relocate_chunk(struct btrfs_root *root,
 		return -ENOSPC;
 
 	/* step one, relocate all the extents inside this chunk */
+	btrfs_scrub_pause(root);
 	ret = btrfs_relocate_block_group(extent_root, chunk_offset);
+	btrfs_scrub_continue(root);
 	if (ret)
 		return ret;
 
@@ -2855,7 +2869,6 @@ again:
 
 		if (chunk_type & BTRFS_BLOCK_GROUP_SYSTEM) {
 			ret = btrfs_relocate_chunk(chunk_root,
-						   found_key.objectid,
 						   found_key.offset);
 			if (ret == -ENOSPC)
 				failed++;
@@ -3375,7 +3388,6 @@ again:
 		}
 
 		ret = btrfs_relocate_chunk(chunk_root,
-					   found_key.objectid,
 					   found_key.offset);
 		mutex_unlock(&fs_info->delete_unused_bgs_mutex);
 		if (ret && ret != -ENOSPC)
@@ -4077,7 +4089,6 @@ int btrfs_shrink_device(struct btrfs_device *device, u64 new_size)
 	struct btrfs_dev_extent *dev_extent = NULL;
 	struct btrfs_path *path;
 	u64 length;
-	u64 chunk_objectid;
 	u64 chunk_offset;
 	int ret;
 	int slot;
@@ -4154,11 +4165,10 @@ again:
 			break;
 		}
 
-		chunk_objectid = btrfs_dev_extent_chunk_objectid(l, dev_extent);
 		chunk_offset = btrfs_dev_extent_chunk_offset(l, dev_extent);
 		btrfs_release_path(path);
 
-		ret = btrfs_relocate_chunk(root, chunk_objectid, chunk_offset);
+		ret = btrfs_relocate_chunk(root, chunk_offset);
 		mutex_unlock(&root->fs_info->delete_unused_bgs_mutex);
 		if (ret && ret != -ENOSPC)
 			goto done;
@@ -4200,7 +4210,8 @@ again:
 		u64 start = new_size;
 		u64 len = old_size - new_size;
 
-		if (contains_pending_extent(trans, device, &start, len)) {
+		if (contains_pending_extent(trans->transaction, device,
+					    &start, len)) {
 			unlock_chunks(root);
 			checked_pending_chunks = true;
 			failed = 0;
@@ -5071,9 +5082,7 @@ static struct btrfs_bio *alloc_btrfs_bio(int total_stripes, int real_stripes)
 		 * and the stripes
 		 */
 		sizeof(u64) * (total_stripes),
-		GFP_NOFS);
-	if (!bbio)
-		return NULL;
+		GFP_NOFS|__GFP_NOFAIL);
 
 	atomic_set(&bbio->error, 0);
 	atomic_set(&bbio->refs, 1);

fs/btrfs/volumes.h

@@ -453,6 +453,9 @@ int btrfs_cancel_balance(struct btrfs_fs_info *fs_info);
 int btrfs_create_uuid_tree(struct btrfs_fs_info *fs_info);
 int btrfs_check_uuid_tree(struct btrfs_fs_info *fs_info);
 int btrfs_chunk_readonly(struct btrfs_root *root, u64 chunk_offset);
+int find_free_dev_extent_start(struct btrfs_transaction *transaction,
+			 struct btrfs_device *device, u64 num_bytes,
+			 u64 search_start, u64 *start, u64 *max_avail);
 int find_free_dev_extent(struct btrfs_trans_handle *trans,
 			 struct btrfs_device *device, u64 num_bytes,
 			 u64 *start, u64 *max_avail);