Merge tag 'for-4.19-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux

Pull btrfs updates from David Sterba:
 "Mostly fixes and cleanups, nothing big, though the notable thing is
  the inserted/deleted lines delta -1124.

  User visible changes:
   - allow defrag on opened read-only files that have rw permissions;
     similar to what dedupe will allow on such files

  Core changes:
   - tree checker improvements, reported by fuzzing:
      * more checks for: block group items, essential trees
      * chunk type validation
      * mount time cross-checks that physical and logical chunks match
      * switch more error codes to EUCLEAN aka EFSCORRUPTED

  Fixes:
   - fsync corner case fixes

   - fix send failure when root has deleted files still open

   - send, fix incorrect file layout after hole punching beyond eof

   - fix races between mount and deice scan ioctl, found by fuzzing

   - fix deadlock when delayed iput is called from writeback on the same
     inode; rare but has been observed in practice, also removes code

   - fix pinned byte accounting, using the right percpu helpers; this
     should avoid some write IO inefficiency during low space conditions

   - don't remove block group that still has pinned bytes

   - reset on-disk device stats value after replace, otherwise this
     would report stale values for the new device

  Cleanups:
   - time64_t/timespec64 cleanups

   - remove remaining dead code in scrub handling NOCOW extents after
     disabling it in previous cycle

   - simplify fsync regarding ordered extents logic and remove all the
     related code

   - remove redundant arguments in order to reduce stack space
     consumption

   - remove support for V0 type of extents, not in use since 2.6.30

   - remove several unused structure members

   - fewer indirect function calls by inlining some callbacks

   - qgroup rescan timing fixes

   - vfs: iget cleanups"

* tag 'for-4.19-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux: (182 commits)
  btrfs: revert fs_devices state on error of btrfs_init_new_device
  btrfs: Exit gracefully when chunk map cannot be inserted to the tree
  btrfs: Introduce mount time chunk <-> dev extent mapping check
  btrfs: Verify that every chunk has corresponding block group at mount time
  btrfs: Check that each block group has corresponding chunk at mount time
  Btrfs: send, fix incorrect file layout after hole punching beyond eof
  btrfs: Use wrapper macro for rcu string to remove duplicate code
  btrfs: simplify btrfs_iget
  btrfs: lift make_bad_inode into btrfs_iget
  btrfs: simplify IS_ERR/PTR_ERR checks
  btrfs: btrfs_iget never returns an is_bad_inode inode
  btrfs: replace: Reset on-disk dev stats value after replace
  btrfs: extent-tree: Remove unused __btrfs_free_block_rsv
  btrfs: backref: Use ERR_CAST to return error code
  btrfs: Remove redundant btrfs_release_path from btrfs_unlink_subvol
  btrfs: Remove root parameter from btrfs_unlink_subvol
  btrfs: Remove fs_info from btrfs_add_root_ref
  btrfs: Remove fs_info from btrfs_del_root_ref
  btrfs: Remove fs_info from btrfs_del_root
  btrfs: Remove fs_info from btrfs_delete_delayed_dir_index
  ...

1 files changed, 147 insertions, 25 deletions

diff --git a/fs/btrfs/send.c b/fs/btrfs/send.c
index c47f62b19226..ba8950bfd9c7 100644
--- a/fs/btrfs/send.c
+++ b/fs/btrfs/send.c
@@ -100,6 +100,7 @@ struct send_ctx {
 	u64 cur_inode_rdev;
 	u64 cur_inode_last_extent;
 	u64 cur_inode_next_write_offset;
+	bool ignore_cur_inode;
 
 	u64 send_progress;
 
@@ -1500,7 +1501,7 @@ static int read_symlink(struct btrfs_root *root,
 	BUG_ON(compression);
 
 	off = btrfs_file_extent_inline_start(ei);
-	len = btrfs_file_extent_inline_len(path->nodes[0], path->slots[0], ei);
+	len = btrfs_file_extent_ram_bytes(path->nodes[0], ei);
 
 	ret = fs_path_add_from_extent_buffer(dest, path->nodes[0], off, len);
 
@@ -5006,6 +5007,15 @@ static int send_hole(struct send_ctx *sctx, u64 end)
 	u64 len;
 	int ret = 0;
 
+	/*
+	 * A hole that starts at EOF or beyond it. Since we do not yet support
+	 * fallocate (for extent preallocation and hole punching), sending a
+	 * write of zeroes starting at EOF or beyond would later require issuing
+	 * a truncate operation which would undo the write and achieve nothing.
+	 */
+	if (offset >= sctx->cur_inode_size)
+		return 0;
+
 	if (sctx->flags & BTRFS_SEND_FLAG_NO_FILE_DATA)
 		return send_update_extent(sctx, offset, end - offset);
 
@@ -5160,7 +5170,7 @@ static int clone_range(struct send_ctx *sctx,
 		ei = btrfs_item_ptr(leaf, slot, struct btrfs_file_extent_item);
 		type = btrfs_file_extent_type(leaf, ei);
 		if (type == BTRFS_FILE_EXTENT_INLINE) {
-			ext_len = btrfs_file_extent_inline_len(leaf, slot, ei);
+			ext_len = btrfs_file_extent_ram_bytes(leaf, ei);
 			ext_len = PAGE_ALIGN(ext_len);
 		} else {
 			ext_len = btrfs_file_extent_num_bytes(leaf, ei);
@@ -5236,8 +5246,7 @@ static int send_write_or_clone(struct send_ctx *sctx,
 			struct btrfs_file_extent_item);
 	type = btrfs_file_extent_type(path->nodes[0], ei);
 	if (type == BTRFS_FILE_EXTENT_INLINE) {
-		len = btrfs_file_extent_inline_len(path->nodes[0],
-						   path->slots[0], ei);
+		len = btrfs_file_extent_ram_bytes(path->nodes[0], ei);
 		/*
 		 * it is possible the inline item won't cover the whole page,
 		 * but there may be items after this page.  Make
@@ -5375,7 +5384,7 @@ static int is_extent_unchanged(struct send_ctx *sctx,
 		}
 
 		if (right_type == BTRFS_FILE_EXTENT_INLINE) {
-			right_len = btrfs_file_extent_inline_len(eb, slot, ei);
+			right_len = btrfs_file_extent_ram_bytes(eb, ei);
 			right_len = PAGE_ALIGN(right_len);
 		} else {
 			right_len = btrfs_file_extent_num_bytes(eb, ei);
@@ -5496,8 +5505,7 @@ static int get_last_extent(struct send_ctx *sctx, u64 offset)
 			    struct btrfs_file_extent_item);
 	type = btrfs_file_extent_type(path->nodes[0], fi);
 	if (type == BTRFS_FILE_EXTENT_INLINE) {
-		u64 size = btrfs_file_extent_inline_len(path->nodes[0],
-							path->slots[0], fi);
+		u64 size = btrfs_file_extent_ram_bytes(path->nodes[0], fi);
 		extent_end = ALIGN(key.offset + size,
 				   sctx->send_root->fs_info->sectorsize);
 	} else {
@@ -5560,7 +5568,7 @@ static int range_is_hole_in_parent(struct send_ctx *sctx,
 		fi = btrfs_item_ptr(leaf, slot, struct btrfs_file_extent_item);
 		if (btrfs_file_extent_type(leaf, fi) ==
 		    BTRFS_FILE_EXTENT_INLINE) {
-			u64 size = btrfs_file_extent_inline_len(leaf, slot, fi);
+			u64 size = btrfs_file_extent_ram_bytes(leaf, fi);
 
 			extent_end = ALIGN(key.offset + size,
 					   root->fs_info->sectorsize);
@@ -5606,8 +5614,7 @@ static int maybe_send_hole(struct send_ctx *sctx, struct btrfs_path *path,
 			    struct btrfs_file_extent_item);
 	type = btrfs_file_extent_type(path->nodes[0], fi);
 	if (type == BTRFS_FILE_EXTENT_INLINE) {
-		u64 size = btrfs_file_extent_inline_len(path->nodes[0],
-							path->slots[0], fi);
+		u64 size = btrfs_file_extent_ram_bytes(path->nodes[0], fi);
 		extent_end = ALIGN(key->offset + size,
 				   sctx->send_root->fs_info->sectorsize);
 	} else {
@@ -5799,6 +5806,9 @@ static int finish_inode_if_needed(struct send_ctx *sctx, int at_end)
 	int pending_move = 0;
 	int refs_processed = 0;
 
+	if (sctx->ignore_cur_inode)
+		return 0;
+
 	ret = process_recorded_refs_if_needed(sctx, at_end, &pending_move,
 					      &refs_processed);
 	if (ret < 0)
@@ -5917,6 +5927,93 @@ out:
 	return ret;
 }
 
+struct parent_paths_ctx {
+	struct list_head *refs;
+	struct send_ctx *sctx;
+};
+
+static int record_parent_ref(int num, u64 dir, int index, struct fs_path *name,
+			     void *ctx)
+{
+	struct parent_paths_ctx *ppctx = ctx;
+
+	return record_ref(ppctx->sctx->parent_root, dir, name, ppctx->sctx,
+			  ppctx->refs);
+}
+
+/*
+ * Issue unlink operations for all paths of the current inode found in the
+ * parent snapshot.
+ */
+static int btrfs_unlink_all_paths(struct send_ctx *sctx)
+{
+	LIST_HEAD(deleted_refs);
+	struct btrfs_path *path;
+	struct btrfs_key key;
+	struct parent_paths_ctx ctx;
+	int ret;
+
+	path = alloc_path_for_send();
+	if (!path)
+		return -ENOMEM;
+
+	key.objectid = sctx->cur_ino;
+	key.type = BTRFS_INODE_REF_KEY;
+	key.offset = 0;
+	ret = btrfs_search_slot(NULL, sctx->parent_root, &key, path, 0, 0);
+	if (ret < 0)
+		goto out;
+
+	ctx.refs = &deleted_refs;
+	ctx.sctx = sctx;
+
+	while (true) {
+		struct extent_buffer *eb = path->nodes[0];
+		int slot = path->slots[0];
+
+		if (slot >= btrfs_header_nritems(eb)) {
+			ret = btrfs_next_leaf(sctx->parent_root, path);
+			if (ret < 0)
+				goto out;
+			else if (ret > 0)
+				break;
+			continue;
+		}
+
+		btrfs_item_key_to_cpu(eb, &key, slot);
+		if (key.objectid != sctx->cur_ino)
+			break;
+		if (key.type != BTRFS_INODE_REF_KEY &&
+		    key.type != BTRFS_INODE_EXTREF_KEY)
+			break;
+
+		ret = iterate_inode_ref(sctx->parent_root, path, &key, 1,
+					record_parent_ref, &ctx);
+		if (ret < 0)
+			goto out;
+
+		path->slots[0]++;
+	}
+
+	while (!list_empty(&deleted_refs)) {
+		struct recorded_ref *ref;
+
+		ref = list_first_entry(&deleted_refs, struct recorded_ref, list);
+		ret = send_unlink(sctx, ref->full_path);
+		if (ret < 0)
+			goto out;
+		fs_path_free(ref->full_path);
+		list_del(&ref->list);
+		kfree(ref);
+	}
+	ret = 0;
+out:
+	btrfs_free_path(path);
+	if (ret)
+		__free_recorded_refs(&deleted_refs);
+	return ret;
+}
+
 static int changed_inode(struct send_ctx *sctx,
 			 enum btrfs_compare_tree_result result)
 {
@@ -5931,6 +6028,7 @@ static int changed_inode(struct send_ctx *sctx,
 	sctx->cur_inode_new_gen = 0;
 	sctx->cur_inode_last_extent = (u64)-1;
 	sctx->cur_inode_next_write_offset = 0;
+	sctx->ignore_cur_inode = false;
 
 	/*
 	 * Set send_progress to current inode. This will tell all get_cur_xxx
@@ -5971,6 +6069,33 @@ static int changed_inode(struct send_ctx *sctx,
 			sctx->cur_inode_new_gen = 1;
 	}
 
+	/*
+	 * Normally we do not find inodes with a link count of zero (orphans)
+	 * because the most common case is to create a snapshot and use it
+	 * for a send operation. However other less common use cases involve
+	 * using a subvolume and send it after turning it to RO mode just
+	 * after deleting all hard links of a file while holding an open
+	 * file descriptor against it or turning a RO snapshot into RW mode,
+	 * keep an open file descriptor against a file, delete it and then
+	 * turn the snapshot back to RO mode before using it for a send
+	 * operation. So if we find such cases, ignore the inode and all its
+	 * items completely if it's a new inode, or if it's a changed inode
+	 * make sure all its previous paths (from the parent snapshot) are all
+	 * unlinked and all other the inode items are ignored.
+	 */
+	if (result == BTRFS_COMPARE_TREE_NEW ||
+	    result == BTRFS_COMPARE_TREE_CHANGED) {
+		u32 nlinks;
+
+		nlinks = btrfs_inode_nlink(sctx->left_path->nodes[0], left_ii);
+		if (nlinks == 0) {
+			sctx->ignore_cur_inode = true;
+			if (result == BTRFS_COMPARE_TREE_CHANGED)
+				ret = btrfs_unlink_all_paths(sctx);
+			goto out;
+		}
+	}
+
 	if (result == BTRFS_COMPARE_TREE_NEW) {
 		sctx->cur_inode_gen = left_gen;
 		sctx->cur_inode_new = 1;
@@ -6309,15 +6434,17 @@ static int changed_cb(struct btrfs_path *left_path,
 	    key->objectid == BTRFS_FREE_SPACE_OBJECTID)
 		goto out;
 
-	if (key->type == BTRFS_INODE_ITEM_KEY)
+	if (key->type == BTRFS_INODE_ITEM_KEY) {
 		ret = changed_inode(sctx, result);
-	else if (key->type == BTRFS_INODE_REF_KEY ||
-		 key->type == BTRFS_INODE_EXTREF_KEY)
-		ret = changed_ref(sctx, result);
-	else if (key->type == BTRFS_XATTR_ITEM_KEY)
-		ret = changed_xattr(sctx, result);
-	else if (key->type == BTRFS_EXTENT_DATA_KEY)
-		ret = changed_extent(sctx, result);
+	} else if (!sctx->ignore_cur_inode) {
+		if (key->type == BTRFS_INODE_REF_KEY ||
+		    key->type == BTRFS_INODE_EXTREF_KEY)
+			ret = changed_ref(sctx, result);
+		else if (key->type == BTRFS_XATTR_ITEM_KEY)
+			ret = changed_xattr(sctx, result);
+		else if (key->type == BTRFS_EXTENT_DATA_KEY)
+			ret = changed_extent(sctx, result);
+	}
 
 out:
 	return ret;
@@ -6328,7 +6455,6 @@ static int full_send_tree(struct send_ctx *sctx)
 	int ret;
 	struct btrfs_root *send_root = sctx->send_root;
 	struct btrfs_key key;
-	struct btrfs_key found_key;
 	struct btrfs_path *path;
 	struct extent_buffer *eb;
 	int slot;
@@ -6350,17 +6476,13 @@ static int full_send_tree(struct send_ctx *sctx)
 	while (1) {
 		eb = path->nodes[0];
 		slot = path->slots[0];
-		btrfs_item_key_to_cpu(eb, &found_key, slot);
+		btrfs_item_key_to_cpu(eb, &key, slot);
 
-		ret = changed_cb(path, NULL, &found_key,
+		ret = changed_cb(path, NULL, &key,
 				 BTRFS_COMPARE_TREE_NEW, sctx);
 		if (ret < 0)
 			goto out;
 
-		key.objectid = found_key.objectid;
-		key.type = found_key.type;
-		key.offset = found_key.offset + 1;
-
 		ret = btrfs_next_item(send_root, path);
 		if (ret < 0)
 			goto out;