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, 114 insertions, 1 deletions

diff --git a/fs/btrfs/tree-checker.c b/fs/btrfs/tree-checker.c
index 8d40e7dd8c30..db835635372f 100644
--- a/fs/btrfs/tree-checker.c
+++ b/fs/btrfs/tree-checker.c
@@ -19,6 +19,7 @@
 #include "tree-checker.h"
 #include "disk-io.h"
 #include "compression.h"
+#include "volumes.h"
 
 /*
  * Error message should follow the following format:
@@ -353,6 +354,102 @@ static int check_dir_item(struct btrfs_fs_info *fs_info,
 	return 0;
 }
 
+__printf(4, 5)
+__cold
+static void block_group_err(const struct btrfs_fs_info *fs_info,
+			    const struct extent_buffer *eb, int slot,
+			    const char *fmt, ...)
+{
+	struct btrfs_key key;
+	struct va_format vaf;
+	va_list args;
+
+	btrfs_item_key_to_cpu(eb, &key, slot);
+	va_start(args, fmt);
+
+	vaf.fmt = fmt;
+	vaf.va = &args;
+
+	btrfs_crit(fs_info,
+	"corrupt %s: root=%llu block=%llu slot=%d bg_start=%llu bg_len=%llu, %pV",
+		btrfs_header_level(eb) == 0 ? "leaf" : "node",
+		btrfs_header_owner(eb), btrfs_header_bytenr(eb), slot,
+		key.objectid, key.offset, &vaf);
+	va_end(args);
+}
+
+static int check_block_group_item(struct btrfs_fs_info *fs_info,
+				  struct extent_buffer *leaf,
+				  struct btrfs_key *key, int slot)
+{
+	struct btrfs_block_group_item bgi;
+	u32 item_size = btrfs_item_size_nr(leaf, slot);
+	u64 flags;
+	u64 type;
+
+	/*
+	 * Here we don't really care about alignment since extent allocator can
+	 * handle it.  We care more about the size, as if one block group is
+	 * larger than maximum size, it's must be some obvious corruption.
+	 */
+	if (key->offset > BTRFS_MAX_DATA_CHUNK_SIZE || key->offset == 0) {
+		block_group_err(fs_info, leaf, slot,
+			"invalid block group size, have %llu expect (0, %llu]",
+				key->offset, BTRFS_MAX_DATA_CHUNK_SIZE);
+		return -EUCLEAN;
+	}
+
+	if (item_size != sizeof(bgi)) {
+		block_group_err(fs_info, leaf, slot,
+			"invalid item size, have %u expect %zu",
+				item_size, sizeof(bgi));
+		return -EUCLEAN;
+	}
+
+	read_extent_buffer(leaf, &bgi, btrfs_item_ptr_offset(leaf, slot),
+			   sizeof(bgi));
+	if (btrfs_block_group_chunk_objectid(&bgi) !=
+	    BTRFS_FIRST_CHUNK_TREE_OBJECTID) {
+		block_group_err(fs_info, leaf, slot,
+		"invalid block group chunk objectid, have %llu expect %llu",
+				btrfs_block_group_chunk_objectid(&bgi),
+				BTRFS_FIRST_CHUNK_TREE_OBJECTID);
+		return -EUCLEAN;
+	}
+
+	if (btrfs_block_group_used(&bgi) > key->offset) {
+		block_group_err(fs_info, leaf, slot,
+			"invalid block group used, have %llu expect [0, %llu)",
+				btrfs_block_group_used(&bgi), key->offset);
+		return -EUCLEAN;
+	}
+
+	flags = btrfs_block_group_flags(&bgi);
+	if (hweight64(flags & BTRFS_BLOCK_GROUP_PROFILE_MASK) > 1) {
+		block_group_err(fs_info, leaf, slot,
+"invalid profile flags, have 0x%llx (%lu bits set) expect no more than 1 bit set",
+			flags & BTRFS_BLOCK_GROUP_PROFILE_MASK,
+			hweight64(flags & BTRFS_BLOCK_GROUP_PROFILE_MASK));
+		return -EUCLEAN;
+	}
+
+	type = flags & BTRFS_BLOCK_GROUP_TYPE_MASK;
+	if (type != BTRFS_BLOCK_GROUP_DATA &&
+	    type != BTRFS_BLOCK_GROUP_METADATA &&
+	    type != BTRFS_BLOCK_GROUP_SYSTEM &&
+	    type != (BTRFS_BLOCK_GROUP_METADATA |
+			   BTRFS_BLOCK_GROUP_DATA)) {
+		block_group_err(fs_info, leaf, slot,
+"invalid type, have 0x%llx (%lu bits set) expect either 0x%llx, 0x%llx, 0x%llu or 0x%llx",
+			type, hweight64(type),
+			BTRFS_BLOCK_GROUP_DATA, BTRFS_BLOCK_GROUP_METADATA,
+			BTRFS_BLOCK_GROUP_SYSTEM,
+			BTRFS_BLOCK_GROUP_METADATA | BTRFS_BLOCK_GROUP_DATA);
+		return -EUCLEAN;
+	}
+	return 0;
+}
+
 /*
  * Common point to switch the item-specific validation.
  */
@@ -374,6 +471,9 @@ static int check_leaf_item(struct btrfs_fs_info *fs_info,
 	case BTRFS_XATTR_ITEM_KEY:
 		ret = check_dir_item(fs_info, leaf, key, slot);
 		break;
+	case BTRFS_BLOCK_GROUP_ITEM_KEY:
+		ret = check_block_group_item(fs_info, leaf, key, slot);
+		break;
 	}
 	return ret;
 }
@@ -396,9 +496,22 @@ static int check_leaf(struct btrfs_fs_info *fs_info, struct extent_buffer *leaf,
 	 * skip this check for relocation trees.
 	 */
 	if (nritems == 0 && !btrfs_header_flag(leaf, BTRFS_HEADER_FLAG_RELOC)) {
+		u64 owner = btrfs_header_owner(leaf);
 		struct btrfs_root *check_root;
 
-		key.objectid = btrfs_header_owner(leaf);
+		/* These trees must never be empty */
+		if (owner == BTRFS_ROOT_TREE_OBJECTID ||
+		    owner == BTRFS_CHUNK_TREE_OBJECTID ||
+		    owner == BTRFS_EXTENT_TREE_OBJECTID ||
+		    owner == BTRFS_DEV_TREE_OBJECTID ||
+		    owner == BTRFS_FS_TREE_OBJECTID ||
+		    owner == BTRFS_DATA_RELOC_TREE_OBJECTID) {
+			generic_err(fs_info, leaf, 0,
+			"invalid root, root %llu must never be empty",
+				    owner);
+			return -EUCLEAN;
+		}
+		key.objectid = owner;
 		key.type = BTRFS_ROOT_ITEM_KEY;
 		key.offset = (u64)-1;