Merge branch 'for-linus-4.2' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs

Pull btrfs fixes from Chris Mason:
 "This is an assortment of fixes.  Most of the commits are from Filipe
  (fsync, the inode allocation cache and a few others).  Mark kicked in
  a series fixing corners in the extent sharing ioctls, and everyone
  else fixed up on assorted other problems"

* 'for-linus-4.2' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs:
  Btrfs: fix wrong check for btrfs_force_chunk_alloc()
  Btrfs: fix warning of bytes_may_use
  Btrfs: fix hang when failing to submit bio of directIO
  Btrfs: fix a comment in inode.c:evict_inode_truncate_pages()
  Btrfs: fix memory corruption on failure to submit bio for direct IO
  btrfs: don't update mtime/ctime on deduped inodes
  btrfs: allow dedupe of same inode
  btrfs: fix deadlock with extent-same and readpage
  btrfs: pass unaligned length to btrfs_cmp_data()
  Btrfs: fix fsync after truncate when no_holes feature is enabled
  Btrfs: fix fsync xattr loss in the fast fsync path
  Btrfs: fix fsync data loss after append write
  Btrfs: fix crash on close_ctree() if cleaner starts new transaction
  Btrfs: fix race between caching kthread and returning inode to inode cache
  Btrfs: use kmem_cache_free when freeing entry in inode cache
  Btrfs: fix race between balance and unused block group deletion
  btrfs: add error handling for scrub_workers_get()
  btrfs: cleanup noused initialization of dev in btrfs_end_bio()
  btrfs: qgroup: allow user to clear the limitation on qgroup

fs/btrfs/btrfs_inode.h

@@ -44,6 +44,8 @@
 #define BTRFS_INODE_IN_DELALLOC_LIST		9
 #define BTRFS_INODE_READDIO_NEED_LOCK		10
 #define BTRFS_INODE_HAS_PROPS		        11
+/* DIO is ready to submit */
+#define BTRFS_INODE_DIO_READY		        12
 /*
  * The following 3 bits are meant only for the btree inode.
  * When any of them is set, it means an error happened while writing an

fs/btrfs/ctree.h

@@ -1778,6 +1778,7 @@ struct btrfs_fs_info {
 	spinlock_t unused_bgs_lock;
 	struct list_head unused_bgs;
 	struct mutex unused_bg_unpin_mutex;
+	struct mutex delete_unused_bgs_mutex;
 
 	/* For btrfs to record security options */
 	struct security_mnt_opts security_opts;

fs/btrfs/disk-io.c

@@ -1751,6 +1751,7 @@ static int cleaner_kthread(void *arg)
 {
 	struct btrfs_root *root = arg;
 	int again;
+	struct btrfs_trans_handle *trans;
 
 	do {
 		again = 0;
@@ -1772,7 +1773,6 @@ static int cleaner_kthread(void *arg)
 		}
 
 		btrfs_run_delayed_iputs(root);
-		btrfs_delete_unused_bgs(root->fs_info);
 		again = btrfs_clean_one_deleted_snapshot(root);
 		mutex_unlock(&root->fs_info->cleaner_mutex);
 
@@ -1781,6 +1781,16 @@ static int cleaner_kthread(void *arg)
 		 * needn't do anything special here.
 		 */
 		btrfs_run_defrag_inodes(root->fs_info);
+
+		/*
+		 * Acquires fs_info->delete_unused_bgs_mutex to avoid racing
+		 * with relocation (btrfs_relocate_chunk) and relocation
+		 * acquires fs_info->cleaner_mutex (btrfs_relocate_block_group)
+		 * after acquiring fs_info->delete_unused_bgs_mutex. So we
+		 * can't hold, nor need to, fs_info->cleaner_mutex when deleting
+		 * unused block groups.
+		 */
+		btrfs_delete_unused_bgs(root->fs_info);
 sleep:
 		if (!try_to_freeze() && !again) {
 			set_current_state(TASK_INTERRUPTIBLE);
@@ -1789,6 +1799,34 @@ static int cleaner_kthread(void *arg)
 			__set_current_state(TASK_RUNNING);
 		}
 	} while (!kthread_should_stop());
+
+	/*
+	 * Transaction kthread is stopped before us and wakes us up.
+	 * However we might have started a new transaction and COWed some
+	 * tree blocks when deleting unused block groups for example. So
+	 * make sure we commit the transaction we started to have a clean
+	 * shutdown when evicting the btree inode - if it has dirty pages
+	 * when we do the final iput() on it, eviction will trigger a
+	 * writeback for it which will fail with null pointer dereferences
+	 * since work queues and other resources were already released and
+	 * destroyed by the time the iput/eviction/writeback is made.
+	 */
+	trans = btrfs_attach_transaction(root);
+	if (IS_ERR(trans)) {
+		if (PTR_ERR(trans) != -ENOENT)
+			btrfs_err(root->fs_info,
+				  "cleaner transaction attach returned %ld",
+				  PTR_ERR(trans));
+	} else {
+		int ret;
+
+		ret = btrfs_commit_transaction(trans, root);
+		if (ret)
+			btrfs_err(root->fs_info,
+				  "cleaner open transaction commit returned %d",
+				  ret);
+	}
+
 	return 0;
 }
 
@@ -2492,6 +2530,7 @@ int open_ctree(struct super_block *sb,
 	spin_lock_init(&fs_info->unused_bgs_lock);
 	rwlock_init(&fs_info->tree_mod_log_lock);
 	mutex_init(&fs_info->unused_bg_unpin_mutex);
+	mutex_init(&fs_info->delete_unused_bgs_mutex);
 	mutex_init(&fs_info->reloc_mutex);
 	mutex_init(&fs_info->delalloc_root_mutex);
 	seqlock_init(&fs_info->profiles_lock);

fs/btrfs/extent-tree.c

@@ -9889,6 +9889,8 @@ void btrfs_delete_unused_bgs(struct btrfs_fs_info *fs_info)
 		}
 		spin_unlock(&fs_info->unused_bgs_lock);
 
+		mutex_lock(&root->fs_info->delete_unused_bgs_mutex);
+
 		/* Don't want to race with allocators so take the groups_sem */
 		down_write(&space_info->groups_sem);
 		spin_lock(&block_group->lock);
@@ -9983,6 +9985,7 @@ void btrfs_delete_unused_bgs(struct btrfs_fs_info *fs_info)
 end_trans:
 		btrfs_end_transaction(trans, root);
 next:
+		mutex_unlock(&root->fs_info->delete_unused_bgs_mutex);
 		btrfs_put_block_group(block_group);
 		spin_lock(&fs_info->unused_bgs_lock);
 	}

fs/btrfs/inode-map.c

@@ -246,6 +246,7 @@ void btrfs_unpin_free_ino(struct btrfs_root *root)
 {
 	struct btrfs_free_space_ctl *ctl = root->free_ino_ctl;
 	struct rb_root *rbroot = &root->free_ino_pinned->free_space_offset;
+	spinlock_t *rbroot_lock = &root->free_ino_pinned->tree_lock;
 	struct btrfs_free_space *info;
 	struct rb_node *n;
 	u64 count;
@@ -254,24 +255,30 @@ void btrfs_unpin_free_ino(struct btrfs_root *root)
 		return;
 
 	while (1) {
+		bool add_to_ctl = true;
+
+		spin_lock(rbroot_lock);
 		n = rb_first(rbroot);
-		if (!n)
+		if (!n) {
+			spin_unlock(rbroot_lock);
 			break;
+		}
 
 		info = rb_entry(n, struct btrfs_free_space, offset_index);
 		BUG_ON(info->bitmap); /* Logic error */
 
 		if (info->offset > root->ino_cache_progress)
-			goto free;
+			add_to_ctl = false;
 		else if (info->offset + info->bytes > root->ino_cache_progress)
 			count = root->ino_cache_progress - info->offset + 1;
 		else
 			count = info->bytes;
 
-		__btrfs_add_free_space(ctl, info->offset, count);
-free:
 		rb_erase(&info->offset_index, rbroot);
-		kfree(info);
+		spin_unlock(rbroot_lock);
+		if (add_to_ctl)
+			__btrfs_add_free_space(ctl, info->offset, count);
+		kmem_cache_free(btrfs_free_space_cachep, info);
 	}
 }
 

fs/btrfs/inode.c

@@ -4989,8 +4989,9 @@ static void evict_inode_truncate_pages(struct inode *inode)
 	/*
 	 * Keep looping until we have no more ranges in the io tree.
 	 * We can have ongoing bios started by readpages (called from readahead)
-	 * that didn't get their end io callbacks called yet or they are still
-	 * in progress ((extent_io.c:end_bio_extent_readpage()). This means some
+	 * that have their endio callback (extent_io.c:end_bio_extent_readpage)
+	 * still in progress (unlocked the pages in the bio but did not yet
+	 * unlocked the ranges in the io tree). Therefore this means some
 	 * ranges can still be locked and eviction started because before
 	 * submitting those bios, which are executed by a separate task (work
 	 * queue kthread), inode references (inode->i_count) were not taken
@@ -7546,6 +7547,7 @@ static int btrfs_get_blocks_direct(struct inode *inode, sector_t iblock,
 
 		current->journal_info = outstanding_extents;
 		btrfs_free_reserved_data_space(inode, len);
+		set_bit(BTRFS_INODE_DIO_READY, &BTRFS_I(inode)->runtime_flags);
 	}
 
 	/*
@@ -7871,8 +7873,6 @@ static void btrfs_endio_direct_write(struct bio *bio, int err)
 	struct bio *dio_bio;
 	int ret;
 
-	if (err)
-		goto out_done;
 again:
 	ret = btrfs_dec_test_first_ordered_pending(inode, &ordered,
 						   &ordered_offset,
@@ -7895,7 +7895,6 @@ static void btrfs_endio_direct_write(struct bio *bio, int err)
 		ordered = NULL;
 		goto again;
 	}
-out_done:
 	dio_bio = dip->dio_bio;
 
 	kfree(dip);
@@ -8163,9 +8162,8 @@ static int btrfs_submit_direct_hook(int rw, struct btrfs_dio_private *dip,
 static void btrfs_submit_direct(int rw, struct bio *dio_bio,
 				struct inode *inode, loff_t file_offset)
 {
-	struct btrfs_root *root = BTRFS_I(inode)->root;
-	struct btrfs_dio_private *dip;
-	struct bio *io_bio;
+	struct btrfs_dio_private *dip = NULL;
+	struct bio *io_bio = NULL;
 	struct btrfs_io_bio *btrfs_bio;
 	int skip_sum;
 	int write = rw & REQ_WRITE;
@@ -8182,7 +8180,7 @@ static void btrfs_submit_direct(int rw, struct bio *dio_bio,
 	dip = kzalloc(sizeof(*dip), GFP_NOFS);
 	if (!dip) {
 		ret = -ENOMEM;
-		goto free_io_bio;
+		goto free_ordered;
 	}
 
 	dip->private = dio_bio->bi_private;
@@ -8210,25 +8208,55 @@ static void btrfs_submit_direct(int rw, struct bio *dio_bio,
 
 	if (btrfs_bio->end_io)
 		btrfs_bio->end_io(btrfs_bio, ret);
-free_io_bio:
-	bio_put(io_bio);
 
 free_ordered:
 	/*
-	 * If this is a write, we need to clean up the reserved space and kill
-	 * the ordered extent.
+	 * If we arrived here it means either we failed to submit the dip
+	 * or we either failed to clone the dio_bio or failed to allocate the
+	 * dip. If we cloned the dio_bio and allocated the dip, we can just
+	 * call bio_endio against our io_bio so that we get proper resource
+	 * cleanup if we fail to submit the dip, otherwise, we must do the
+	 * same as btrfs_endio_direct_[write|read] because we can't call these
+	 * callbacks - they require an allocated dip and a clone of dio_bio.
+	 */
+	if (io_bio && dip) {
+		bio_endio(io_bio, ret);
+		/*
+		 * The end io callbacks free our dip, do the final put on io_bio
+		 * and all the cleanup and final put for dio_bio (through
+		 * dio_end_io()).
 		 */
+		dip = NULL;
+		io_bio = NULL;
+	} else {
 		if (write) {
 			struct btrfs_ordered_extent *ordered;
-		ordered = btrfs_lookup_ordered_extent(inode, file_offset);
-		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, 1);
-		btrfs_put_ordered_extent(ordered);
-		btrfs_put_ordered_extent(ordered);
+
+			ordered = btrfs_lookup_ordered_extent(inode,
+							      file_offset);
+			set_bit(BTRFS_ORDERED_IOERR, &ordered->flags);
+			/*
+			 * Decrements our ref on the ordered extent and removes
+			 * the ordered extent from the inode's ordered tree,
+			 * doing all the proper resource cleanup such as for the
+			 * reserved space and waking up any waiters for this
+			 * ordered extent (through btrfs_remove_ordered_extent).
+			 */
+			btrfs_finish_ordered_io(ordered);
+		} else {
+			unlock_extent(&BTRFS_I(inode)->io_tree, file_offset,
+			      file_offset + dio_bio->bi_iter.bi_size - 1);
+		}
+		clear_bit(BIO_UPTODATE, &dio_bio->bi_flags);
+		/*
+		 * Releases and cleans up our dio_bio, no need to bio_put()
+		 * nor bio_endio()/bio_io_error() against dio_bio.
+		 */
+		dio_end_io(dio_bio, ret);
 	}
-	bio_endio(dio_bio, ret);
+	if (io_bio)
+		bio_put(io_bio);
+	kfree(dip);
 }
 
 static ssize_t check_direct_IO(struct btrfs_root *root, struct kiocb *iocb,
@@ -8330,9 +8358,18 @@ static ssize_t btrfs_direct_IO(struct kiocb *iocb, struct iov_iter *iter,
 				   btrfs_submit_direct, flags);
 	if (iov_iter_rw(iter) == WRITE) {
 		current->journal_info = NULL;
-		if (ret < 0 && ret != -EIOCBQUEUED)
+		if (ret < 0 && ret != -EIOCBQUEUED) {
+			/*
+			 * If the error comes from submitting stage,
+			 * btrfs_get_blocsk_direct() has free'd data space,
+			 * and metadata space will be handled by
+			 * finish_ordered_fn, don't do that again to make
+			 * sure bytes_may_use is correct.
+			 */
+			if (!test_and_clear_bit(BTRFS_INODE_DIO_READY,
+				     &BTRFS_I(inode)->runtime_flags))
 				btrfs_delalloc_release_space(inode, count);
-		else if (ret >= 0 && (size_t)ret < count)
+		} else if (ret >= 0 && (size_t)ret < count)
 			btrfs_delalloc_release_space(inode,
 						     count - (size_t)ret);
 	}

fs/btrfs/ordered-data.c

@@ -552,6 +552,10 @@ void btrfs_put_ordered_extent(struct btrfs_ordered_extent *entry)
 	trace_btrfs_ordered_extent_put(entry->inode, entry);
 
 	if (atomic_dec_and_test(&entry->refs)) {
+		ASSERT(list_empty(&entry->log_list));
+		ASSERT(list_empty(&entry->trans_list));
+		ASSERT(list_empty(&entry->root_extent_list));
+		ASSERT(RB_EMPTY_NODE(&entry->rb_node));
 		if (entry->inode)
 			btrfs_add_delayed_iput(entry->inode);
 		while (!list_empty(&entry->list)) {
@@ -579,6 +583,7 @@ void btrfs_remove_ordered_extent(struct inode *inode,
 	spin_lock_irq(&tree->lock);
 	node = &entry->rb_node;
 	rb_erase(node, &tree->tree);
+	RB_CLEAR_NODE(node);
 	if (tree->last == node)
 		tree->last = NULL;
 	set_bit(BTRFS_ORDERED_COMPLETE, &entry->flags);

fs/btrfs/qgroup.c

@@ -1349,6 +1349,11 @@ int btrfs_limit_qgroup(struct btrfs_trans_handle *trans,
 	struct btrfs_root *quota_root;
 	struct btrfs_qgroup *qgroup;
 	int ret = 0;
+	/* Sometimes we would want to clear the limit on this qgroup.
+	 * To meet this requirement, we treat the -1 as a special value
+	 * which tell kernel to clear the limit on this qgroup.
+	 */
+	const u64 CLEAR_VALUE = -1;
 
 	mutex_lock(&fs_info->qgroup_ioctl_lock);
 	quota_root = fs_info->quota_root;
@@ -1364,14 +1369,42 @@ int btrfs_limit_qgroup(struct btrfs_trans_handle *trans,
 	}
 
 	spin_lock(&fs_info->qgroup_lock);
-	if (limit->flags & BTRFS_QGROUP_LIMIT_MAX_RFER)
+	if (limit->flags & BTRFS_QGROUP_LIMIT_MAX_RFER) {
+		if (limit->max_rfer == CLEAR_VALUE) {
+			qgroup->lim_flags &= ~BTRFS_QGROUP_LIMIT_MAX_RFER;
+			limit->flags &= ~BTRFS_QGROUP_LIMIT_MAX_RFER;
+			qgroup->max_rfer = 0;
+		} else {
 			qgroup->max_rfer = limit->max_rfer;
-	if (limit->flags & BTRFS_QGROUP_LIMIT_MAX_EXCL)
+		}
+	}
+	if (limit->flags & BTRFS_QGROUP_LIMIT_MAX_EXCL) {
+		if (limit->max_excl == CLEAR_VALUE) {
+			qgroup->lim_flags &= ~BTRFS_QGROUP_LIMIT_MAX_EXCL;
+			limit->flags &= ~BTRFS_QGROUP_LIMIT_MAX_EXCL;
+			qgroup->max_excl = 0;
+		} else {
 			qgroup->max_excl = limit->max_excl;
-	if (limit->flags & BTRFS_QGROUP_LIMIT_RSV_RFER)
+		}
+	}
+	if (limit->flags & BTRFS_QGROUP_LIMIT_RSV_RFER) {
+		if (limit->rsv_rfer == CLEAR_VALUE) {
+			qgroup->lim_flags &= ~BTRFS_QGROUP_LIMIT_RSV_RFER;
+			limit->flags &= ~BTRFS_QGROUP_LIMIT_RSV_RFER;
+			qgroup->rsv_rfer = 0;
+		} else {
 			qgroup->rsv_rfer = limit->rsv_rfer;
-	if (limit->flags & BTRFS_QGROUP_LIMIT_RSV_EXCL)
+		}
+	}
+	if (limit->flags & BTRFS_QGROUP_LIMIT_RSV_EXCL) {
+		if (limit->rsv_excl == CLEAR_VALUE) {
+			qgroup->lim_flags &= ~BTRFS_QGROUP_LIMIT_RSV_EXCL;
+			limit->flags &= ~BTRFS_QGROUP_LIMIT_RSV_EXCL;
+			qgroup->rsv_excl = 0;
+		} else {
 			qgroup->rsv_excl = limit->rsv_excl;
+		}
+	}
 	qgroup->lim_flags |= limit->flags;
 
 	spin_unlock(&fs_info->qgroup_lock);

fs/btrfs/relocation.c

@@ -4049,7 +4049,7 @@ static noinline_for_stack int relocate_block_group(struct reloc_control *rc)
 	if (trans && progress && err == -ENOSPC) {
 		ret = btrfs_force_chunk_alloc(trans, rc->extent_root,
 					      rc->block_group->flags);
-		if (ret == 0) {
+		if (ret == 1) {
 			err = 0;
 			progress = 0;
 			goto restart;

fs/btrfs/scrub.c

@@ -3571,7 +3571,6 @@ static noinline_for_stack int scrub_supers(struct scrub_ctx *sctx,
 static noinline_for_stack int scrub_workers_get(struct btrfs_fs_info *fs_info,
 						int is_dev_replace)
 {
-	int ret = 0;
 	unsigned int flags = WQ_FREEZABLE | WQ_UNBOUND;
 	int max_active = fs_info->thread_pool_size;
 
@@ -3584,34 +3583,36 @@ static noinline_for_stack int scrub_workers_get(struct btrfs_fs_info *fs_info,
 			fs_info->scrub_workers =
 				btrfs_alloc_workqueue("btrfs-scrub", flags,
 						      max_active, 4);
-		if (!fs_info->scrub_workers) {
-			ret = -ENOMEM;
-			goto out;
-		}
+		if (!fs_info->scrub_workers)
+			goto fail_scrub_workers;
+
 		fs_info->scrub_wr_completion_workers =
 			btrfs_alloc_workqueue("btrfs-scrubwrc", flags,
 					      max_active, 2);
-		if (!fs_info->scrub_wr_completion_workers) {
-			ret = -ENOMEM;
-			goto out;
-		}
+		if (!fs_info->scrub_wr_completion_workers)
+			goto fail_scrub_wr_completion_workers;
+
 		fs_info->scrub_nocow_workers =
 			btrfs_alloc_workqueue("btrfs-scrubnc", flags, 1, 0);
-		if (!fs_info->scrub_nocow_workers) {
-			ret = -ENOMEM;
-			goto out;
-		}
+		if (!fs_info->scrub_nocow_workers)
+			goto fail_scrub_nocow_workers;
 		fs_info->scrub_parity_workers =
 			btrfs_alloc_workqueue("btrfs-scrubparity", flags,
 					      max_active, 2);
-		if (!fs_info->scrub_parity_workers) {
-			ret = -ENOMEM;
-			goto out;
-		}
+		if (!fs_info->scrub_parity_workers)
+			goto fail_scrub_parity_workers;
 	}
 	++fs_info->scrub_workers_refcnt;
-out:
-	return ret;
+	return 0;
+
+fail_scrub_parity_workers:
+	btrfs_destroy_workqueue(fs_info->scrub_nocow_workers);
+fail_scrub_nocow_workers:
+	btrfs_destroy_workqueue(fs_info->scrub_wr_completion_workers);
+fail_scrub_wr_completion_workers:
+	btrfs_destroy_workqueue(fs_info->scrub_workers);
+fail_scrub_workers:
+	return -ENOMEM;
 }
 
 static noinline_for_stack void scrub_workers_put(struct btrfs_fs_info *fs_info)

fs/btrfs/tree-log.c

@@ -4117,6 +4117,187 @@ static int logged_inode_size(struct btrfs_root *log, struct inode *inode,
 	return 0;
 }
 
+/*
+ * At the moment we always log all xattrs. This is to figure out at log replay
+ * time which xattrs must have their deletion replayed. If a xattr is missing
+ * in the log tree and exists in the fs/subvol tree, we delete it. This is
+ * because if a xattr is deleted, the inode is fsynced and a power failure
+ * happens, causing the log to be replayed the next time the fs is mounted,
+ * we want the xattr to not exist anymore (same behaviour as other filesystems
+ * with a journal, ext3/4, xfs, f2fs, etc).
+ */
+static int btrfs_log_all_xattrs(struct btrfs_trans_handle *trans,
+				struct btrfs_root *root,
+				struct inode *inode,
+				struct btrfs_path *path,
+				struct btrfs_path *dst_path)
+{
+	int ret;
+	struct btrfs_key key;
+	const u64 ino = btrfs_ino(inode);
+	int ins_nr = 0;
+	int start_slot = 0;
+
+	key.objectid = ino;
+	key.type = BTRFS_XATTR_ITEM_KEY;
+	key.offset = 0;
+
+	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
+	if (ret < 0)
+		return ret;
+
+	while (true) {
+		int slot = path->slots[0];
+		struct extent_buffer *leaf = path->nodes[0];
+		int nritems = btrfs_header_nritems(leaf);
+
+		if (slot >= nritems) {
+			if (ins_nr > 0) {
+				u64 last_extent = 0;
+
+				ret = copy_items(trans, inode, dst_path, path,
+						 &last_extent, start_slot,
+						 ins_nr, 1, 0);
+				/* can't be 1, extent items aren't processed */
+				ASSERT(ret <= 0);
+				if (ret < 0)
+					return ret;
+				ins_nr = 0;
+			}
+			ret = btrfs_next_leaf(root, path);
+			if (ret < 0)
+				return ret;
+			else if (ret > 0)
+				break;
+			continue;
+		}
+
+		btrfs_item_key_to_cpu(leaf, &key, slot);
+		if (key.objectid != ino || key.type != BTRFS_XATTR_ITEM_KEY)
+			break;
+
+		if (ins_nr == 0)
+			start_slot = slot;
+		ins_nr++;
+		path->slots[0]++;
+		cond_resched();
+	}
+	if (ins_nr > 0) {
+		u64 last_extent = 0;
+
+		ret = copy_items(trans, inode, dst_path, path,
+				 &last_extent, start_slot,
+				 ins_nr, 1, 0);
+		/* can't be 1, extent items aren't processed */
+		ASSERT(ret <= 0);
+		if (ret < 0)
+			return ret;
+	}
+
+	return 0;
+}
+
+/*
+ * If the no holes feature is enabled we need to make sure any hole between the
+ * last extent and the i_size of our inode is explicitly marked in the log. This
+ * is to make sure that doing something like:
+ *
+ *      1) create file with 128Kb of data
+ *      2) truncate file to 64Kb
+ *      3) truncate file to 256Kb
+ *      4) fsync file
+ *      5) <crash/power failure>
+ *      6) mount fs and trigger log replay
+ *
+ * Will give us a file with a size of 256Kb, the first 64Kb of data match what
+ * the file had in its first 64Kb of data at step 1 and the last 192Kb of the
+ * file correspond to a hole. The presence of explicit holes in a log tree is
+ * what guarantees that log replay will remove/adjust file extent items in the
+ * fs/subvol tree.
+ *
+ * Here we do not need to care about holes between extents, that is already done
+ * by copy_items(). We also only need to do this in the full sync path, where we
+ * lookup for extents from the fs/subvol tree only. In the fast path case, we
+ * lookup the list of modified extent maps and if any represents a hole, we
+ * insert a corresponding extent representing a hole in the log tree.
+ */
+static int btrfs_log_trailing_hole(struct btrfs_trans_handle *trans,
+				   struct btrfs_root *root,
+				   struct inode *inode,
+				   struct btrfs_path *path)
+{
+	int ret;
+	struct btrfs_key key;
+	u64 hole_start;
+	u64 hole_size;
+	struct extent_buffer *leaf;
+	struct btrfs_root *log = root->log_root;
+	const u64 ino = btrfs_ino(inode);
+	const u64 i_size = i_size_read(inode);
+
+	if (!btrfs_fs_incompat(root->fs_info, NO_HOLES))
+		return 0;
+
+	key.objectid = ino;
+	key.type = BTRFS_EXTENT_DATA_KEY;
+	key.offset = (u64)-1;
+
+	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
+	ASSERT(ret != 0);
+	if (ret < 0)
+		return ret;
+
+	ASSERT(path->slots[0] > 0);
+	path->slots[0]--;
+	leaf = path->nodes[0];
+	btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
+
+	if (key.objectid != ino || key.type != BTRFS_EXTENT_DATA_KEY) {
+		/* inode does not have any extents */
+		hole_start = 0;
+		hole_size = i_size;
+	} else {
+		struct btrfs_file_extent_item *extent;
+		u64 len;
+
+		/*
+		 * If there's an extent beyond i_size, an explicit hole was
+		 * already inserted by copy_items().
+		 */
+		if (key.offset >= i_size)
+			return 0;
+
+		extent = btrfs_item_ptr(leaf, path->slots[0],
+					struct btrfs_file_extent_item);
+
+		if (btrfs_file_extent_type(leaf, extent) ==
+		    BTRFS_FILE_EXTENT_INLINE) {
+			len = btrfs_file_extent_inline_len(leaf,
+							   path->slots[0],
+							   extent);
+			ASSERT(len == i_size);
+			return 0;
+		}
+
+		len = btrfs_file_extent_num_bytes(leaf, extent);
+		/* Last extent goes beyond i_size, no need to log a hole. */
+		if (key.offset + len > i_size)
+			return 0;
+		hole_start = key.offset + len;
+		hole_size = i_size - hole_start;
+	}
+	btrfs_release_path(path);
+
+	/* Last extent ends at i_size. */
+	if (hole_size == 0)
+		return 0;
+
+	hole_size = ALIGN(hole_size, root->sectorsize);
+	ret = btrfs_insert_file_extent(trans, log, ino, hole_start, 0, 0,
+				       hole_size, 0, hole_size, 0, 0, 0);
+	return ret;
+}
+
 /* log a single inode in the tree log.
  * At least one parent directory for this inode must exist in the tree
  * or be logged already.
@@ -4155,6 +4336,7 @@ static int btrfs_log_inode(struct btrfs_trans_handle *trans,
 	u64 ino = btrfs_ino(inode);
 	struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
 	u64 logged_isize = 0;
+	bool need_log_inode_item = true;
 
 	path = btrfs_alloc_path();
 	if (!path)
@@ -4263,11 +4445,6 @@ static int btrfs_log_inode(struct btrfs_trans_handle *trans,
 		} else {
 			if (inode_only == LOG_INODE_ALL)
 				fast_search = true;
-			ret = log_inode_item(trans, log, dst_path, inode);
-			if (ret) {
-				err = ret;
-				goto out_unlock;
-			}
 			goto log_extents;
 		}
 
@@ -4290,6 +4467,28 @@ static int btrfs_log_inode(struct btrfs_trans_handle *trans,
 		if (min_key.type > max_key.type)
 			break;
 
+		if (min_key.type == BTRFS_INODE_ITEM_KEY)
+			need_log_inode_item = false;
+
+		/* Skip xattrs, we log them later with btrfs_log_all_xattrs() */
+		if (min_key.type == BTRFS_XATTR_ITEM_KEY) {
+			if (ins_nr == 0)
+				goto next_slot;
+			ret = copy_items(trans, inode, dst_path, path,
+					 &last_extent, ins_start_slot,
+					 ins_nr, inode_only, logged_isize);
+			if (ret < 0) {
+				err = ret;
+				goto out_unlock;
+			}
+			ins_nr = 0;
+			if (ret) {
+				btrfs_release_path(path);
+				continue;
+			}
+			goto next_slot;
+		}
+
 		src = path->nodes[0];
 		if (ins_nr && ins_start_slot + ins_nr == path->slots[0]) {
 			ins_nr++;
@@ -4357,9 +4556,26 @@ static int btrfs_log_inode(struct btrfs_trans_handle *trans,
 		ins_nr = 0;
 	}
 
+	btrfs_release_path(path);
+	btrfs_release_path(dst_path);
+	err = btrfs_log_all_xattrs(trans, root, inode, path, dst_path);
+	if (err)
+		goto out_unlock;
+	if (max_key.type >= BTRFS_EXTENT_DATA_KEY && !fast_search) {
+		btrfs_release_path(path);
+		btrfs_release_path(dst_path);
+		err = btrfs_log_trailing_hole(trans, root, inode, path);
+		if (err)
+			goto out_unlock;
+	}
 log_extents:
 	btrfs_release_path(path);
 	btrfs_release_path(dst_path);
+	if (need_log_inode_item) {
+		err = log_inode_item(trans, log, dst_path, inode);
+		if (err)
+			goto out_unlock;
+	}
 	if (fast_search) {
 		/*
 		 * Some ordered extents started by fsync might have completed

fs/btrfs/volumes.c

@@ -2766,6 +2766,20 @@ static int btrfs_relocate_chunk(struct btrfs_root *root,
 	root = root->fs_info->chunk_root;
 	extent_root = root->fs_info->extent_root;
 
+	/*
+	 * Prevent races with automatic removal of unused block groups.
+	 * After we relocate and before we remove the chunk with offset
+	 * chunk_offset, automatic removal of the block group can kick in,
+	 * resulting in a failure when calling btrfs_remove_chunk() below.
+	 *
+	 * Make sure to acquire this mutex before doing a tree search (dev
+	 * or chunk trees) to find chunks. Otherwise the cleaner kthread might
+	 * call btrfs_remove_chunk() (through btrfs_delete_unused_bgs()) after
+	 * we release the path used to search the chunk/dev tree and before
+	 * the current task acquires this mutex and calls us.
+	 */
+	ASSERT(mutex_is_locked(&root->fs_info->delete_unused_bgs_mutex));
+
 	ret = btrfs_can_relocate(extent_root, chunk_offset);
 	if (ret)
 		return -ENOSPC;
@@ -2814,13 +2828,18 @@ static int btrfs_relocate_sys_chunks(struct btrfs_root *root)
 	key.type = BTRFS_CHUNK_ITEM_KEY;
 
 	while (1) {
+		mutex_lock(&root->fs_info->delete_unused_bgs_mutex);
 		ret = btrfs_search_slot(NULL, chunk_root, &key, path, 0, 0);
-		if (ret < 0)
+		if (ret < 0) {
+			mutex_unlock(&root->fs_info->delete_unused_bgs_mutex);
 			goto error;
+		}
 		BUG_ON(ret == 0); /* Corruption */
 
 		ret = btrfs_previous_item(chunk_root, path, key.objectid,
 					  key.type);
+		if (ret)
+			mutex_unlock(&root->fs_info->delete_unused_bgs_mutex);
 		if (ret < 0)
 			goto error;
 		if (ret > 0)
@@ -2843,6 +2862,7 @@ static int btrfs_relocate_sys_chunks(struct btrfs_root *root)
 			else
 				BUG_ON(ret);
 		}
+		mutex_unlock(&root->fs_info->delete_unused_bgs_mutex);
 
 		if (found_key.offset == 0)
 			break;
@@ -3299,9 +3319,12 @@ static int __btrfs_balance(struct btrfs_fs_info *fs_info)
 			goto error;
 		}
 
+		mutex_lock(&fs_info->delete_unused_bgs_mutex);
 		ret = btrfs_search_slot(NULL, chunk_root, &key, path, 0, 0);
-		if (ret < 0)
+		if (ret < 0) {
+			mutex_unlock(&fs_info->delete_unused_bgs_mutex);
 			goto error;
+		}
 
 		/*
 		 * this shouldn't happen, it means the last relocate
@@ -3313,6 +3336,7 @@ static int __btrfs_balance(struct btrfs_fs_info *fs_info)
 		ret = btrfs_previous_item(chunk_root, path, 0,
 					  BTRFS_CHUNK_ITEM_KEY);
 		if (ret) {
+			mutex_unlock(&fs_info->delete_unused_bgs_mutex);
 			ret = 0;
 			break;
 		}
@@ -3321,8 +3345,10 @@ static int __btrfs_balance(struct btrfs_fs_info *fs_info)
 		slot = path->slots[0];
 		btrfs_item_key_to_cpu(leaf, &found_key, slot);
 
-		if (found_key.objectid != key.objectid)
+		if (found_key.objectid != key.objectid) {
+			mutex_unlock(&fs_info->delete_unused_bgs_mutex);
 			break;
+		}
 
 		chunk = btrfs_item_ptr(leaf, slot, struct btrfs_chunk);
 
@@ -3335,10 +3361,13 @@ static int __btrfs_balance(struct btrfs_fs_info *fs_info)
 		ret = should_balance_chunk(chunk_root, leaf, chunk,
 					   found_key.offset);
 		btrfs_release_path(path);
-		if (!ret)
+		if (!ret) {
+			mutex_unlock(&fs_info->delete_unused_bgs_mutex);
 			goto loop;
+		}
 
 		if (counting) {
+			mutex_unlock(&fs_info->delete_unused_bgs_mutex);
 			spin_lock(&fs_info->balance_lock);
 			bctl->stat.expected++;
 			spin_unlock(&fs_info->balance_lock);
@@ -3348,6 +3377,7 @@ static int __btrfs_balance(struct btrfs_fs_info *fs_info)
 		ret = btrfs_relocate_chunk(chunk_root,
 					   found_key.objectid,
 					   found_key.offset);
+		mutex_unlock(&fs_info->delete_unused_bgs_mutex);
 		if (ret && ret != -ENOSPC)
 			goto error;
 		if (ret == -ENOSPC) {
@@ -4087,11 +4117,16 @@ int btrfs_shrink_device(struct btrfs_device *device, u64 new_size)
 	key.type = BTRFS_DEV_EXTENT_KEY;
 
 	do {
+		mutex_lock(&root->fs_info->delete_unused_bgs_mutex);
 		ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
-		if (ret < 0)
+		if (ret < 0) {
+			mutex_unlock(&root->fs_info->delete_unused_bgs_mutex);
 			goto done;
+		}
 
 		ret = btrfs_previous_item(root, path, 0, key.type);
+		if (ret)
+			mutex_unlock(&root->fs_info->delete_unused_bgs_mutex);
 		if (ret < 0)
 			goto done;
 		if (ret) {
@@ -4105,6 +4140,7 @@ int btrfs_shrink_device(struct btrfs_device *device, u64 new_size)
 		btrfs_item_key_to_cpu(l, &key, path->slots[0]);
 
 		if (key.objectid != device->devid) {
+			mutex_unlock(&root->fs_info->delete_unused_bgs_mutex);
 			btrfs_release_path(path);
 			break;
 		}
@@ -4113,6 +4149,7 @@ int btrfs_shrink_device(struct btrfs_device *device, u64 new_size)
 		length = btrfs_dev_extent_length(l, dev_extent);
 
 		if (key.offset + length <= new_size) {
+			mutex_unlock(&root->fs_info->delete_unused_bgs_mutex);
 			btrfs_release_path(path);
 			break;
 		}
@@ -4122,6 +4159,7 @@ int btrfs_shrink_device(struct btrfs_device *device, u64 new_size)
 		btrfs_release_path(path);
 
 		ret = btrfs_relocate_chunk(root, chunk_objectid, chunk_offset);
+		mutex_unlock(&root->fs_info->delete_unused_bgs_mutex);
 		if (ret && ret != -ENOSPC)
 			goto done;
 		if (ret == -ENOSPC)
@@ -5715,7 +5753,6 @@ static inline void btrfs_end_bbio(struct btrfs_bio *bbio, struct bio *bio, int e
 static void btrfs_end_bio(struct bio *bio, int err)
 {
 	struct btrfs_bio *bbio = bio->bi_private;
-	struct btrfs_device *dev = bbio->stripes[0].dev;
 	int is_orig_bio = 0;
 
 	if (err) {
@@ -5723,6 +5760,7 @@ static void btrfs_end_bio(struct bio *bio, int err)
 		if (err == -EIO || err == -EREMOTEIO) {
 			unsigned int stripe_index =
 				btrfs_io_bio(bio)->stripe_index;
+			struct btrfs_device *dev;
 
 			BUG_ON(stripe_index >= bbio->num_stripes);
 			dev = bbio->stripes[stripe_index].dev;