Merge tag 'for-5.10-rc1-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux

Pull btrfs fixes from David Sterba:

 - lockdep fixes:
     - drop path locks before manipulating sysfs objects or qgroups
     - preliminary fixes before tree locks get switched to rwsem
     - use annotated seqlock

 - build warning fixes (printk format)

 - fix relocation vs fallocate race

 - tree checker properly validates number of stripes and parity

 - readahead vs device replace fixes

 - iomap dio fix for unnecessary buffered io fallback

* tag 'for-5.10-rc1-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux:
  btrfs: convert data_seqcount to seqcount_mutex_t
  btrfs: don't fallback to buffered read if we don't need to
  btrfs: add a helper to read the tree_root commit root for backref lookup
  btrfs: drop the path before adding qgroup items when enabling qgroups
  btrfs: fix readahead hang and use-after-free after removing a device
  btrfs: fix use-after-free on readahead extent after failure to create it
  btrfs: tree-checker: validate number of chunk stripes and parity
  btrfs: tree-checker: fix incorrect printk format
  btrfs: drop the path before adding block group sysfs files
  btrfs: fix relocation failure due to race with fallocate

fs/btrfs/backref.c

@@ -544,7 +544,18 @@ static int resolve_indirect_ref(struct btrfs_fs_info *fs_info,
 	int level = ref->level;
 	struct btrfs_key search_key = ref->key_for_search;
 
-	root = btrfs_get_fs_root(fs_info, ref->root_id, false);
+	/*
+	 * If we're search_commit_root we could possibly be holding locks on
+	 * other tree nodes.  This happens when qgroups does backref walks when
+	 * adding new delayed refs.  To deal with this we need to look in cache
+	 * for the root, and if we don't find it then we need to search the
+	 * tree_root's commit root, thus the btrfs_get_fs_root_commit_root usage
+	 * here.
+	 */
+	if (path->search_commit_root)
+		root = btrfs_get_fs_root_commit_root(fs_info, path, ref->root_id);
+	else
+		root = btrfs_get_fs_root(fs_info, ref->root_id, false);
 	if (IS_ERR(root)) {
 		ret = PTR_ERR(root);
 		goto out_free;

fs/btrfs/block-group.c

@@ -2024,6 +2024,7 @@ int btrfs_read_block_groups(struct btrfs_fs_info *info)
 		key.offset = 0;
 		btrfs_release_path(path);
 	}
+	btrfs_release_path(path);
 
 	list_for_each_entry(space_info, &info->space_info, list) {
 		int i;

fs/btrfs/ctree.h

@@ -3564,6 +3564,8 @@ struct reada_control *btrfs_reada_add(struct btrfs_root *root,
 int btrfs_reada_wait(void *handle);
 void btrfs_reada_detach(void *handle);
 int btree_readahead_hook(struct extent_buffer *eb, int err);
+void btrfs_reada_remove_dev(struct btrfs_device *dev);
+void btrfs_reada_undo_remove_dev(struct btrfs_device *dev);
 
 static inline int is_fstree(u64 rootid)
 {

fs/btrfs/dev-replace.c

@@ -688,6 +688,9 @@ static int btrfs_dev_replace_finishing(struct btrfs_fs_info *fs_info,
 	}
 	btrfs_wait_ordered_roots(fs_info, U64_MAX, 0, (u64)-1);
 
+	if (!scrub_ret)
+		btrfs_reada_remove_dev(src_device);
+
 	/*
 	 * We have to use this loop approach because at this point src_device
 	 * has to be available for transaction commit to complete, yet new
@@ -696,6 +699,7 @@ static int btrfs_dev_replace_finishing(struct btrfs_fs_info *fs_info,
 	while (1) {
 		trans = btrfs_start_transaction(root, 0);
 		if (IS_ERR(trans)) {
+			btrfs_reada_undo_remove_dev(src_device);
 			mutex_unlock(&dev_replace->lock_finishing_cancel_unmount);
 			return PTR_ERR(trans);
 		}
@@ -746,6 +750,7 @@ static int btrfs_dev_replace_finishing(struct btrfs_fs_info *fs_info,
 		up_write(&dev_replace->rwsem);
 		mutex_unlock(&fs_info->chunk_mutex);
 		mutex_unlock(&fs_info->fs_devices->device_list_mutex);
+		btrfs_reada_undo_remove_dev(src_device);
 		btrfs_rm_dev_replace_blocked(fs_info);
 		if (tgt_device)
 			btrfs_destroy_dev_replace_tgtdev(tgt_device);

fs/btrfs/disk-io.c

@@ -1281,32 +1281,26 @@ int btrfs_add_log_tree(struct btrfs_trans_handle *trans,
 	return 0;
 }
 
-struct btrfs_root *btrfs_read_tree_root(struct btrfs_root *tree_root,
-					struct btrfs_key *key)
+static struct btrfs_root *read_tree_root_path(struct btrfs_root *tree_root,
+					      struct btrfs_path *path,
+					      struct btrfs_key *key)
 {
 	struct btrfs_root *root;
 	struct btrfs_fs_info *fs_info = tree_root->fs_info;
-	struct btrfs_path *path;
 	u64 generation;
 	int ret;
 	int level;
 
-	path = btrfs_alloc_path();
-	if (!path)
-		return ERR_PTR(-ENOMEM);
-
 	root = btrfs_alloc_root(fs_info, key->objectid, GFP_NOFS);
-	if (!root) {
-		ret = -ENOMEM;
-		goto alloc_fail;
-	}
+	if (!root)
+		return ERR_PTR(-ENOMEM);
 
 	ret = btrfs_find_root(tree_root, key, path,
 			      &root->root_item, &root->root_key);
 	if (ret) {
 		if (ret > 0)
 			ret = -ENOENT;
-		goto find_fail;
+		goto fail;
 	}
 
 	generation = btrfs_root_generation(&root->root_item);
@@ -1317,21 +1311,31 @@ struct btrfs_root *btrfs_read_tree_root(struct btrfs_root *tree_root,
 	if (IS_ERR(root->node)) {
 		ret = PTR_ERR(root->node);
 		root->node = NULL;
-		goto find_fail;
+		goto fail;
 	} else if (!btrfs_buffer_uptodate(root->node, generation, 0)) {
 		ret = -EIO;
-		goto find_fail;
+		goto fail;
 	}
 	root->commit_root = btrfs_root_node(root);
-out:
-	btrfs_free_path(path);
 	return root;
-
-find_fail:
+fail:
 	btrfs_put_root(root);
-alloc_fail:
-	root = ERR_PTR(ret);
-	goto out;
+	return ERR_PTR(ret);
+}
+
+struct btrfs_root *btrfs_read_tree_root(struct btrfs_root *tree_root,
+					struct btrfs_key *key)
+{
+	struct btrfs_root *root;
+	struct btrfs_path *path;
+
+	path = btrfs_alloc_path();
+	if (!path)
+		return ERR_PTR(-ENOMEM);
+	root = read_tree_root_path(tree_root, path, key);
+	btrfs_free_path(path);
+
+	return root;
 }
 
 /*
@@ -1419,6 +1423,31 @@ static struct btrfs_root *btrfs_lookup_fs_root(struct btrfs_fs_info *fs_info,
 	return root;
 }
 
+static struct btrfs_root *btrfs_get_global_root(struct btrfs_fs_info *fs_info,
+						u64 objectid)
+{
+	if (objectid == BTRFS_ROOT_TREE_OBJECTID)
+		return btrfs_grab_root(fs_info->tree_root);
+	if (objectid == BTRFS_EXTENT_TREE_OBJECTID)
+		return btrfs_grab_root(fs_info->extent_root);
+	if (objectid == BTRFS_CHUNK_TREE_OBJECTID)
+		return btrfs_grab_root(fs_info->chunk_root);
+	if (objectid == BTRFS_DEV_TREE_OBJECTID)
+		return btrfs_grab_root(fs_info->dev_root);
+	if (objectid == BTRFS_CSUM_TREE_OBJECTID)
+		return btrfs_grab_root(fs_info->csum_root);
+	if (objectid == BTRFS_QUOTA_TREE_OBJECTID)
+		return btrfs_grab_root(fs_info->quota_root) ?
+			fs_info->quota_root : ERR_PTR(-ENOENT);
+	if (objectid == BTRFS_UUID_TREE_OBJECTID)
+		return btrfs_grab_root(fs_info->uuid_root) ?
+			fs_info->uuid_root : ERR_PTR(-ENOENT);
+	if (objectid == BTRFS_FREE_SPACE_TREE_OBJECTID)
+		return btrfs_grab_root(fs_info->free_space_root) ?
+			fs_info->free_space_root : ERR_PTR(-ENOENT);
+	return NULL;
+}
+
 int btrfs_insert_fs_root(struct btrfs_fs_info *fs_info,
 			 struct btrfs_root *root)
 {
@@ -1518,25 +1547,9 @@ static struct btrfs_root *btrfs_get_root_ref(struct btrfs_fs_info *fs_info,
 	struct btrfs_key key;
 	int ret;
 
-	if (objectid == BTRFS_ROOT_TREE_OBJECTID)
-		return btrfs_grab_root(fs_info->tree_root);
-	if (objectid == BTRFS_EXTENT_TREE_OBJECTID)
-		return btrfs_grab_root(fs_info->extent_root);
-	if (objectid == BTRFS_CHUNK_TREE_OBJECTID)
-		return btrfs_grab_root(fs_info->chunk_root);
-	if (objectid == BTRFS_DEV_TREE_OBJECTID)
-		return btrfs_grab_root(fs_info->dev_root);
-	if (objectid == BTRFS_CSUM_TREE_OBJECTID)
-		return btrfs_grab_root(fs_info->csum_root);
-	if (objectid == BTRFS_QUOTA_TREE_OBJECTID)
-		return btrfs_grab_root(fs_info->quota_root) ?
-			fs_info->quota_root : ERR_PTR(-ENOENT);
-	if (objectid == BTRFS_UUID_TREE_OBJECTID)
-		return btrfs_grab_root(fs_info->uuid_root) ?
-			fs_info->uuid_root : ERR_PTR(-ENOENT);
-	if (objectid == BTRFS_FREE_SPACE_TREE_OBJECTID)
-		return btrfs_grab_root(fs_info->free_space_root) ?
-			fs_info->free_space_root : ERR_PTR(-ENOENT);
+	root = btrfs_get_global_root(fs_info, objectid);
+	if (root)
+		return root;
 again:
 	root = btrfs_lookup_fs_root(fs_info, objectid);
 	if (root) {
@@ -1621,6 +1634,52 @@ struct btrfs_root *btrfs_get_new_fs_root(struct btrfs_fs_info *fs_info,
 	return btrfs_get_root_ref(fs_info, objectid, anon_dev, true);
 }
 
+/*
+ * btrfs_get_fs_root_commit_root - return a root for the given objectid
+ * @fs_info:	the fs_info
+ * @objectid:	the objectid we need to lookup
+ *
+ * This is exclusively used for backref walking, and exists specifically because
+ * of how qgroups does lookups.  Qgroups will do a backref lookup at delayed ref
+ * creation time, which means we may have to read the tree_root in order to look
+ * up a fs root that is not in memory.  If the root is not in memory we will
+ * read the tree root commit root and look up the fs root from there.  This is a
+ * temporary root, it will not be inserted into the radix tree as it doesn't
+ * have the most uptodate information, it'll simply be discarded once the
+ * backref code is finished using the root.
+ */
+struct btrfs_root *btrfs_get_fs_root_commit_root(struct btrfs_fs_info *fs_info,
+						 struct btrfs_path *path,
+						 u64 objectid)
+{
+	struct btrfs_root *root;
+	struct btrfs_key key;
+
+	ASSERT(path->search_commit_root && path->skip_locking);
+
+	/*
+	 * This can return -ENOENT if we ask for a root that doesn't exist, but
+	 * since this is called via the backref walking code we won't be looking
+	 * up a root that doesn't exist, unless there's corruption.  So if root
+	 * != NULL just return it.
+	 */
+	root = btrfs_get_global_root(fs_info, objectid);
+	if (root)
+		return root;
+
+	root = btrfs_lookup_fs_root(fs_info, objectid);
+	if (root)
+		return root;
+
+	key.objectid = objectid;
+	key.type = BTRFS_ROOT_ITEM_KEY;
+	key.offset = (u64)-1;
+	root = read_tree_root_path(fs_info->tree_root, path, &key);
+	btrfs_release_path(path);
+
+	return root;
+}
+
 /*
  * called by the kthread helper functions to finally call the bio end_io
  * functions.  This is where read checksum verification actually happens

fs/btrfs/disk-io.h

@@ -69,6 +69,9 @@ struct btrfs_root *btrfs_get_fs_root(struct btrfs_fs_info *fs_info,
 				     u64 objectid, bool check_ref);
 struct btrfs_root *btrfs_get_new_fs_root(struct btrfs_fs_info *fs_info,
 					 u64 objectid, dev_t anon_dev);
+struct btrfs_root *btrfs_get_fs_root_commit_root(struct btrfs_fs_info *fs_info,
+						 struct btrfs_path *path,
+						 u64 objectid);
 
 void btrfs_free_fs_info(struct btrfs_fs_info *fs_info);
 int btrfs_cleanup_fs_roots(struct btrfs_fs_info *fs_info);

fs/btrfs/extent-tree.c

@@ -3185,7 +3185,7 @@ static int __btrfs_free_extent(struct btrfs_trans_handle *trans,
 		struct btrfs_tree_block_info *bi;
 		if (item_size < sizeof(*ei) + sizeof(*bi)) {
 			btrfs_crit(info,
-"invalid extent item size for key (%llu, %u, %llu) owner %llu, has %u expect >= %lu",
+"invalid extent item size for key (%llu, %u, %llu) owner %llu, has %u expect >= %zu",
 				   key.objectid, key.type, key.offset,
 				   owner_objectid, item_size,
 				   sizeof(*ei) + sizeof(*bi));

fs/btrfs/file.c

@@ -3628,7 +3628,8 @@ static ssize_t btrfs_file_read_iter(struct kiocb *iocb, struct iov_iter *to)
 		inode_lock_shared(inode);
 		ret = btrfs_direct_IO(iocb, to);
 		inode_unlock_shared(inode);
-		if (ret < 0)
+		if (ret < 0 || !iov_iter_count(to) ||
+		    iocb->ki_pos >= i_size_read(file_inode(iocb->ki_filp)))
 			return ret;
 	}
 

fs/btrfs/inode.c

@@ -9672,10 +9672,16 @@ static int __btrfs_prealloc_file_range(struct inode *inode, int mode,
 		 * clear_offset by our extent size.
 		 */
 		clear_offset += ins.offset;
-		btrfs_dec_block_group_reservations(fs_info, ins.objectid);
 
 		last_alloc = ins.offset;
 		trans = insert_prealloc_file_extent(trans, inode, &ins, cur_offset);
+		/*
+		 * Now that we inserted the prealloc extent we can finally
+		 * decrement the number of reservations in the block group.
+		 * If we did it before, we could race with relocation and have
+		 * relocation miss the reserved extent, making it fail later.
+		 */
+		btrfs_dec_block_group_reservations(fs_info, ins.objectid);
 		if (IS_ERR(trans)) {
 			ret = PTR_ERR(trans);
 			btrfs_free_reserved_extent(fs_info, ins.objectid,

fs/btrfs/qgroup.c

@@ -1026,6 +1026,10 @@ int btrfs_quota_enable(struct btrfs_fs_info *fs_info)
 		btrfs_item_key_to_cpu(leaf, &found_key, slot);
 
 		if (found_key.type == BTRFS_ROOT_REF_KEY) {
+
+			/* Release locks on tree_root before we access quota_root */
+			btrfs_release_path(path);
+
 			ret = add_qgroup_item(trans, quota_root,
 					      found_key.offset);
 			if (ret) {
@@ -1044,6 +1048,20 @@ int btrfs_quota_enable(struct btrfs_fs_info *fs_info)
 				btrfs_abort_transaction(trans, ret);
 				goto out_free_path;
 			}
+			ret = btrfs_search_slot_for_read(tree_root, &found_key,
+							 path, 1, 0);
+			if (ret < 0) {
+				btrfs_abort_transaction(trans, ret);
+				goto out_free_path;
+			}
+			if (ret > 0) {
+				/*
+				 * Shouldn't happen, but in case it does we
+				 * don't need to do the btrfs_next_item, just
+				 * continue.
+				 */
+				continue;
+			}
 		}
 		ret = btrfs_next_item(tree_root, path);
 		if (ret < 0) {

fs/btrfs/reada.c

@@ -421,6 +421,9 @@ static struct reada_extent *reada_find_extent(struct btrfs_fs_info *fs_info,
 		if (!dev->bdev)
 			continue;
 
+		if (test_bit(BTRFS_DEV_STATE_NO_READA, &dev->dev_state))
+			continue;
+
 		if (dev_replace_is_ongoing &&
 		    dev == fs_info->dev_replace.tgtdev) {
 			/*
@@ -445,6 +448,8 @@ static struct reada_extent *reada_find_extent(struct btrfs_fs_info *fs_info,
 		}
 		have_zone = 1;
 	}
+	if (!have_zone)
+		radix_tree_delete(&fs_info->reada_tree, index);
 	spin_unlock(&fs_info->reada_lock);
 	up_read(&fs_info->dev_replace.rwsem);
 
@@ -1020,3 +1025,45 @@ void btrfs_reada_detach(void *handle)
 
 	kref_put(&rc->refcnt, reada_control_release);
 }
+
+/*
+ * Before removing a device (device replace or device remove ioctls), call this
+ * function to wait for all existing readahead requests on the device and to
+ * make sure no one queues more readahead requests for the device.
+ *
+ * Must be called without holding neither the device list mutex nor the device
+ * replace semaphore, otherwise it will deadlock.
+ */
+void btrfs_reada_remove_dev(struct btrfs_device *dev)
+{
+	struct btrfs_fs_info *fs_info = dev->fs_info;
+
+	/* Serialize with readahead extent creation at reada_find_extent(). */
+	spin_lock(&fs_info->reada_lock);
+	set_bit(BTRFS_DEV_STATE_NO_READA, &dev->dev_state);
+	spin_unlock(&fs_info->reada_lock);
+
+	/*
+	 * There might be readahead requests added to the radix trees which
+	 * were not yet added to the readahead work queue. We need to start
+	 * them and wait for their completion, otherwise we can end up with
+	 * use-after-free problems when dropping the last reference on the
+	 * readahead extents and their zones, as they need to access the
+	 * device structure.
+	 */
+	reada_start_machine(fs_info);
+	btrfs_flush_workqueue(fs_info->readahead_workers);
+}
+
+/*
+ * If when removing a device (device replace or device remove ioctls) an error
+ * happens after calling btrfs_reada_remove_dev(), call this to undo what that
+ * function did. This is safe to call even if btrfs_reada_remove_dev() was not
+ * called before.
+ */
+void btrfs_reada_undo_remove_dev(struct btrfs_device *dev)
+{
+	spin_lock(&dev->fs_info->reada_lock);
+	clear_bit(BTRFS_DEV_STATE_NO_READA, &dev->dev_state);
+	spin_unlock(&dev->fs_info->reada_lock);
+}

fs/btrfs/tree-checker.c

@@ -760,18 +760,36 @@ int btrfs_check_chunk_valid(struct extent_buffer *leaf,
 	u64 type;
 	u64 features;
 	bool mixed = false;
+	int raid_index;
+	int nparity;
+	int ncopies;
 
 	length = btrfs_chunk_length(leaf, chunk);
 	stripe_len = btrfs_chunk_stripe_len(leaf, chunk);
 	num_stripes = btrfs_chunk_num_stripes(leaf, chunk);
 	sub_stripes = btrfs_chunk_sub_stripes(leaf, chunk);
 	type = btrfs_chunk_type(leaf, chunk);
+	raid_index = btrfs_bg_flags_to_raid_index(type);
+	ncopies = btrfs_raid_array[raid_index].ncopies;
+	nparity = btrfs_raid_array[raid_index].nparity;
 
 	if (!num_stripes) {
 		chunk_err(leaf, chunk, logical,
 			  "invalid chunk num_stripes, have %u", num_stripes);
 		return -EUCLEAN;
 	}
+	if (num_stripes < ncopies) {
+		chunk_err(leaf, chunk, logical,
+			  "invalid chunk num_stripes < ncopies, have %u < %d",
+			  num_stripes, ncopies);
+		return -EUCLEAN;
+	}
+	if (nparity && num_stripes == nparity) {
+		chunk_err(leaf, chunk, logical,
+			  "invalid chunk num_stripes == nparity, have %u == %d",
+			  num_stripes, nparity);
+		return -EUCLEAN;
+	}
 	if (!IS_ALIGNED(logical, fs_info->sectorsize)) {
 		chunk_err(leaf, chunk, logical,
 		"invalid chunk logical, have %llu should aligned to %u",

fs/btrfs/volumes.c

@@ -431,7 +431,7 @@ static struct btrfs_device *__alloc_device(struct btrfs_fs_info *fs_info)
 
 	atomic_set(&dev->reada_in_flight, 0);
 	atomic_set(&dev->dev_stats_ccnt, 0);
-	btrfs_device_data_ordered_init(dev);
+	btrfs_device_data_ordered_init(dev, fs_info);
 	INIT_RADIX_TREE(&dev->reada_zones, GFP_NOFS & ~__GFP_DIRECT_RECLAIM);
 	INIT_RADIX_TREE(&dev->reada_extents, GFP_NOFS & ~__GFP_DIRECT_RECLAIM);
 	extent_io_tree_init(fs_info, &dev->alloc_state,
@@ -2099,6 +2099,8 @@ int btrfs_rm_device(struct btrfs_fs_info *fs_info, const char *device_path,
 
 	mutex_unlock(&uuid_mutex);
 	ret = btrfs_shrink_device(device, 0);
+	if (!ret)
+		btrfs_reada_remove_dev(device);
 	mutex_lock(&uuid_mutex);
 	if (ret)
 		goto error_undo;
@@ -2179,6 +2181,7 @@ int btrfs_rm_device(struct btrfs_fs_info *fs_info, const char *device_path,
 	return ret;
 
 error_undo:
+	btrfs_reada_undo_remove_dev(device);
 	if (test_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state)) {
 		mutex_lock(&fs_info->chunk_mutex);
 		list_add(&device->dev_alloc_list,

fs/btrfs/volumes.h

@@ -39,10 +39,10 @@ struct btrfs_io_geometry {
 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
 #include <linux/seqlock.h>
 #define __BTRFS_NEED_DEVICE_DATA_ORDERED
-#define btrfs_device_data_ordered_init(device)	\
-	seqcount_init(&device->data_seqcount)
+#define btrfs_device_data_ordered_init(device, info)				\
+	seqcount_mutex_init(&device->data_seqcount, &info->chunk_mutex)
 #else
-#define btrfs_device_data_ordered_init(device) do { } while (0)
+#define btrfs_device_data_ordered_init(device, info) do { } while (0)
 #endif
 
 #define BTRFS_DEV_STATE_WRITEABLE	(0)
@@ -50,6 +50,7 @@ struct btrfs_io_geometry {
 #define BTRFS_DEV_STATE_MISSING		(2)
 #define BTRFS_DEV_STATE_REPLACE_TGT	(3)
 #define BTRFS_DEV_STATE_FLUSH_SENT	(4)
+#define BTRFS_DEV_STATE_NO_READA	(5)
 
 struct btrfs_device {
 	struct list_head dev_list; /* device_list_mutex */
@@ -71,7 +72,8 @@ struct btrfs_device {
 	blk_status_t last_flush_error;
 
 #ifdef __BTRFS_NEED_DEVICE_DATA_ORDERED
-	seqcount_t data_seqcount;
+	/* A seqcount_t with associated chunk_mutex (for lockdep) */
+	seqcount_mutex_t data_seqcount;
 #endif
 
 	/* the internal btrfs device id */
@@ -162,11 +164,9 @@ btrfs_device_get_##name(const struct btrfs_device *dev)			\
 static inline void							\
 btrfs_device_set_##name(struct btrfs_device *dev, u64 size)		\
 {									\
-	preempt_disable();						\
 	write_seqcount_begin(&dev->data_seqcount);			\
 	dev->name = size;						\
 	write_seqcount_end(&dev->data_seqcount);			\
-	preempt_enable();						\
 }
 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION)
 #define BTRFS_DEVICE_GETSET_FUNCS(name)					\