Merge tag 'xfs-5.13-merge-3' of git://git.kernel.org/pub/scm/fs/xfs/xfs-linux

Pull xfs updates from Darrick Wong:
 "The notable user-visible addition this cycle is ability to remove
  space from the last AG in a filesystem. This is the first of many
  changes needed for full-fledged support for shrinking a filesystem.
  Still needed are (a) the ability to reorganize files and metadata away
  from the end of the fs; (b) the ability to remove entire allocation
  groups; (c) shrink support for realtime volumes; and (d) thorough
  testing of (a-c).

  There are a number of performance improvements in this code drop: Dave
  streamlined various parts of the buffer logging code and reduced the
  cost of various debugging checks, and added the ability to pre-create
  the xattr structures while creating files. Brian eliminated
  transaction reservations that were being held across writeback (thus
  reducing livelock potential.

  Other random pieces: Pavel fixed the repetitve warnings about
  deprecated mount options, I fixed online fsck to behave itself when a
  readonly remount comes in during scrub, and refactored various other
  parts of that code, Christoph contributed a lot of refactoring this
  cycle. The xfs_icdinode structure has been absorbed into the (incore)
  xfs_inode structure, and the format and flags handling around
  xfs_inode_fork structures has been simplified. Chandan provided a
  number of fixes for extent count overflow related problems that have
  been shaken out by debugging knobs added during 5.12.

  Summary:

   - Various minor fixes in online scrub.

   - Prevent metadata files from being automatically inactivated.

   - Validate btree heights by the computed per-btree limits.

   - Don't warn about remounting with deprecated mount options.

   - Initialize attr forks at create time if we suspect we're going to
     need to store them.

   - Reduce memory reallocation workouts in the logging code.

   - Fix some theoretical math calculation errors in logged buffers that
     span multiple discontig memory ranges but contiguous ondisk
     regions.

   - Speedups in dirty buffer bitmap handling.

   - Make type verifier functions more inline-happy to reduce overhead.

   - Reduce debug overhead in directory checking code.

   - Many many typo fixes.

   - Begin to handle the permanent loss of the very end of a filesystem.

   - Fold struct xfs_icdinode into xfs_inode.

   - Deprecate the long defunct BMV_IF_NO_DMAPI_READ from the bmapx
     ioctl.

   - Remove a broken directory block format check from online scrub.

   - Fix a bug where we could produce an unnecessarily tall data fork
     btree when creating an attr fork.

   - Fix scrub and readonly remounts racing.

   - Fix a writeback ioend log deadlock problem by dropping the behavior
     where we could preallocate a setfilesize transaction.

   - Fix some bugs in the new extent count checking code.

   - Fix some bugs in the attr fork preallocation code.

   - Refactor if_flags out of the incore inode fork data structure"

* tag 'xfs-5.13-merge-3' of git://git.kernel.org/pub/scm/fs/xfs/xfs-linux: (77 commits)
  xfs: remove xfs_quiesce_attr declaration
  xfs: remove XFS_IFEXTENTS
  xfs: remove XFS_IFINLINE
  xfs: remove XFS_IFBROOT
  xfs: only look at the fork format in xfs_idestroy_fork
  xfs: simplify xfs_attr_remove_args
  xfs: rename and simplify xfs_bmap_one_block
  xfs: move the XFS_IFEXTENTS check into xfs_iread_extents
  xfs: drop unnecessary setfilesize helper
  xfs: drop unused ioend private merge and setfilesize code
  xfs: open code ioend needs workqueue helper
  xfs: drop submit side trans alloc for append ioends
  xfs: fix return of uninitialized value in variable error
  xfs: get rid of the ip parameter to xchk_setup_*
  xfs: fix scrub and remount-ro protection when running scrub
  xfs: move the check for post-EOF mappings into xfs_can_free_eofblocks
  xfs: move the xfs_can_free_eofblocks call under the IOLOCK
  xfs: precalculate default inode attribute offset
  xfs: default attr fork size does not handle device inodes
  xfs: inode fork allocation depends on XFS_IFEXTENT flag
  ...

1 files changed, 135 insertions, 127 deletions

diff --git a/fs/xfs/xfs_inode.c b/fs/xfs/xfs_inode.c
index 2a8bdf33e6c4..0369eb22c1bb 100644
--- a/fs/xfs/xfs_inode.c
+++ b/fs/xfs/xfs_inode.c
@@ -60,8 +60,8 @@ xfs_get_extsz_hint(
 	 */
 	if (xfs_is_always_cow_inode(ip))
 		return 0;
-	if ((ip->i_d.di_flags & XFS_DIFLAG_EXTSIZE) && ip->i_d.di_extsize)
-		return ip->i_d.di_extsize;
+	if ((ip->i_diflags & XFS_DIFLAG_EXTSIZE) && ip->i_extsize)
+		return ip->i_extsize;
 	if (XFS_IS_REALTIME_INODE(ip))
 		return ip->i_mount->m_sb.sb_rextsize;
 	return 0;
@@ -80,8 +80,8 @@ xfs_get_cowextsz_hint(
 	xfs_extlen_t		a, b;
 
 	a = 0;
-	if (ip->i_d.di_flags2 & XFS_DIFLAG2_COWEXTSIZE)
-		a = ip->i_d.di_cowextsize;
+	if (ip->i_diflags2 & XFS_DIFLAG2_COWEXTSIZE)
+		a = ip->i_cowextsize;
 	b = xfs_get_extsz_hint(ip);
 
 	a = max(a, b);
@@ -111,8 +111,7 @@ xfs_ilock_data_map_shared(
 {
 	uint			lock_mode = XFS_ILOCK_SHARED;
 
-	if (ip->i_df.if_format == XFS_DINODE_FMT_BTREE &&
-	    (ip->i_df.if_flags & XFS_IFEXTENTS) == 0)
+	if (xfs_need_iread_extents(&ip->i_df))
 		lock_mode = XFS_ILOCK_EXCL;
 	xfs_ilock(ip, lock_mode);
 	return lock_mode;
@@ -124,9 +123,7 @@ xfs_ilock_attr_map_shared(
 {
 	uint			lock_mode = XFS_ILOCK_SHARED;
 
-	if (ip->i_afp &&
-	    ip->i_afp->if_format == XFS_DINODE_FMT_BTREE &&
-	    (ip->i_afp->if_flags & XFS_IFEXTENTS) == 0)
+	if (ip->i_afp && xfs_need_iread_extents(ip->i_afp))
 		lock_mode = XFS_ILOCK_EXCL;
 	xfs_ilock(ip, lock_mode);
 	return lock_mode;
@@ -598,67 +595,55 @@ xfs_lock_two_inodes(
 	}
 }
 
-STATIC uint
-_xfs_dic2xflags(
-	uint16_t		di_flags,
-	uint64_t		di_flags2,
-	bool			has_attr)
+uint
+xfs_ip2xflags(
+	struct xfs_inode	*ip)
 {
 	uint			flags = 0;
 
-	if (di_flags & XFS_DIFLAG_ANY) {
-		if (di_flags & XFS_DIFLAG_REALTIME)
+	if (ip->i_diflags & XFS_DIFLAG_ANY) {
+		if (ip->i_diflags & XFS_DIFLAG_REALTIME)
 			flags |= FS_XFLAG_REALTIME;
-		if (di_flags & XFS_DIFLAG_PREALLOC)
+		if (ip->i_diflags & XFS_DIFLAG_PREALLOC)
 			flags |= FS_XFLAG_PREALLOC;
-		if (di_flags & XFS_DIFLAG_IMMUTABLE)
+		if (ip->i_diflags & XFS_DIFLAG_IMMUTABLE)
 			flags |= FS_XFLAG_IMMUTABLE;
-		if (di_flags & XFS_DIFLAG_APPEND)
+		if (ip->i_diflags & XFS_DIFLAG_APPEND)
 			flags |= FS_XFLAG_APPEND;
-		if (di_flags & XFS_DIFLAG_SYNC)
+		if (ip->i_diflags & XFS_DIFLAG_SYNC)
 			flags |= FS_XFLAG_SYNC;
-		if (di_flags & XFS_DIFLAG_NOATIME)
+		if (ip->i_diflags & XFS_DIFLAG_NOATIME)
 			flags |= FS_XFLAG_NOATIME;
-		if (di_flags & XFS_DIFLAG_NODUMP)
+		if (ip->i_diflags & XFS_DIFLAG_NODUMP)
 			flags |= FS_XFLAG_NODUMP;
-		if (di_flags & XFS_DIFLAG_RTINHERIT)
+		if (ip->i_diflags & XFS_DIFLAG_RTINHERIT)
 			flags |= FS_XFLAG_RTINHERIT;
-		if (di_flags & XFS_DIFLAG_PROJINHERIT)
+		if (ip->i_diflags & XFS_DIFLAG_PROJINHERIT)
 			flags |= FS_XFLAG_PROJINHERIT;
-		if (di_flags & XFS_DIFLAG_NOSYMLINKS)
+		if (ip->i_diflags & XFS_DIFLAG_NOSYMLINKS)
 			flags |= FS_XFLAG_NOSYMLINKS;
-		if (di_flags & XFS_DIFLAG_EXTSIZE)
+		if (ip->i_diflags & XFS_DIFLAG_EXTSIZE)
 			flags |= FS_XFLAG_EXTSIZE;
-		if (di_flags & XFS_DIFLAG_EXTSZINHERIT)
+		if (ip->i_diflags & XFS_DIFLAG_EXTSZINHERIT)
 			flags |= FS_XFLAG_EXTSZINHERIT;
-		if (di_flags & XFS_DIFLAG_NODEFRAG)
+		if (ip->i_diflags & XFS_DIFLAG_NODEFRAG)
 			flags |= FS_XFLAG_NODEFRAG;
-		if (di_flags & XFS_DIFLAG_FILESTREAM)
+		if (ip->i_diflags & XFS_DIFLAG_FILESTREAM)
 			flags |= FS_XFLAG_FILESTREAM;
 	}
 
-	if (di_flags2 & XFS_DIFLAG2_ANY) {
-		if (di_flags2 & XFS_DIFLAG2_DAX)
+	if (ip->i_diflags2 & XFS_DIFLAG2_ANY) {
+		if (ip->i_diflags2 & XFS_DIFLAG2_DAX)
 			flags |= FS_XFLAG_DAX;
-		if (di_flags2 & XFS_DIFLAG2_COWEXTSIZE)
+		if (ip->i_diflags2 & XFS_DIFLAG2_COWEXTSIZE)
 			flags |= FS_XFLAG_COWEXTSIZE;
 	}
 
-	if (has_attr)
+	if (XFS_IFORK_Q(ip))
 		flags |= FS_XFLAG_HASATTR;
-
 	return flags;
 }
 
-uint
-xfs_ip2xflags(
-	struct xfs_inode	*ip)
-{
-	struct xfs_icdinode	*dic = &ip->i_d;
-
-	return _xfs_dic2xflags(dic->di_flags, dic->di_flags2, XFS_IFORK_Q(ip));
-}
-
 /*
  * Lookups up an inode from "name". If ci_name is not NULL, then a CI match
  * is allowed, otherwise it has to be an exact match. If a CI match is found,
@@ -708,42 +693,42 @@ xfs_inode_inherit_flags(
 	umode_t			mode = VFS_I(ip)->i_mode;
 
 	if (S_ISDIR(mode)) {
-		if (pip->i_d.di_flags & XFS_DIFLAG_RTINHERIT)
+		if (pip->i_diflags & XFS_DIFLAG_RTINHERIT)
 			di_flags |= XFS_DIFLAG_RTINHERIT;
-		if (pip->i_d.di_flags & XFS_DIFLAG_EXTSZINHERIT) {
+		if (pip->i_diflags & XFS_DIFLAG_EXTSZINHERIT) {
 			di_flags |= XFS_DIFLAG_EXTSZINHERIT;
-			ip->i_d.di_extsize = pip->i_d.di_extsize;
+			ip->i_extsize = pip->i_extsize;
 		}
-		if (pip->i_d.di_flags & XFS_DIFLAG_PROJINHERIT)
+		if (pip->i_diflags & XFS_DIFLAG_PROJINHERIT)
 			di_flags |= XFS_DIFLAG_PROJINHERIT;
 	} else if (S_ISREG(mode)) {
-		if ((pip->i_d.di_flags & XFS_DIFLAG_RTINHERIT) &&
+		if ((pip->i_diflags & XFS_DIFLAG_RTINHERIT) &&
 		    xfs_sb_version_hasrealtime(&ip->i_mount->m_sb))
 			di_flags |= XFS_DIFLAG_REALTIME;
-		if (pip->i_d.di_flags & XFS_DIFLAG_EXTSZINHERIT) {
+		if (pip->i_diflags & XFS_DIFLAG_EXTSZINHERIT) {
 			di_flags |= XFS_DIFLAG_EXTSIZE;
-			ip->i_d.di_extsize = pip->i_d.di_extsize;
+			ip->i_extsize = pip->i_extsize;
 		}
 	}
-	if ((pip->i_d.di_flags & XFS_DIFLAG_NOATIME) &&
+	if ((pip->i_diflags & XFS_DIFLAG_NOATIME) &&
 	    xfs_inherit_noatime)
 		di_flags |= XFS_DIFLAG_NOATIME;
-	if ((pip->i_d.di_flags & XFS_DIFLAG_NODUMP) &&
+	if ((pip->i_diflags & XFS_DIFLAG_NODUMP) &&
 	    xfs_inherit_nodump)
 		di_flags |= XFS_DIFLAG_NODUMP;
-	if ((pip->i_d.di_flags & XFS_DIFLAG_SYNC) &&
+	if ((pip->i_diflags & XFS_DIFLAG_SYNC) &&
 	    xfs_inherit_sync)
 		di_flags |= XFS_DIFLAG_SYNC;
-	if ((pip->i_d.di_flags & XFS_DIFLAG_NOSYMLINKS) &&
+	if ((pip->i_diflags & XFS_DIFLAG_NOSYMLINKS) &&
 	    xfs_inherit_nosymlinks)
 		di_flags |= XFS_DIFLAG_NOSYMLINKS;
-	if ((pip->i_d.di_flags & XFS_DIFLAG_NODEFRAG) &&
+	if ((pip->i_diflags & XFS_DIFLAG_NODEFRAG) &&
 	    xfs_inherit_nodefrag)
 		di_flags |= XFS_DIFLAG_NODEFRAG;
-	if (pip->i_d.di_flags & XFS_DIFLAG_FILESTREAM)
+	if (pip->i_diflags & XFS_DIFLAG_FILESTREAM)
 		di_flags |= XFS_DIFLAG_FILESTREAM;
 
-	ip->i_d.di_flags |= di_flags;
+	ip->i_diflags |= di_flags;
 }
 
 /* Propagate di_flags2 from a parent inode to a child inode. */
@@ -752,12 +737,12 @@ xfs_inode_inherit_flags2(
 	struct xfs_inode	*ip,
 	const struct xfs_inode	*pip)
 {
-	if (pip->i_d.di_flags2 & XFS_DIFLAG2_COWEXTSIZE) {
-		ip->i_d.di_flags2 |= XFS_DIFLAG2_COWEXTSIZE;
-		ip->i_d.di_cowextsize = pip->i_d.di_cowextsize;
+	if (pip->i_diflags2 & XFS_DIFLAG2_COWEXTSIZE) {
+		ip->i_diflags2 |= XFS_DIFLAG2_COWEXTSIZE;
+		ip->i_cowextsize = pip->i_cowextsize;
 	}
-	if (pip->i_d.di_flags2 & XFS_DIFLAG2_DAX)
-		ip->i_d.di_flags2 |= XFS_DIFLAG2_DAX;
+	if (pip->i_diflags2 & XFS_DIFLAG2_DAX)
+		ip->i_diflags2 |= XFS_DIFLAG2_DAX;
 }
 
 /*
@@ -774,6 +759,7 @@ xfs_init_new_inode(
 	xfs_nlink_t		nlink,
 	dev_t			rdev,
 	prid_t			prid,
+	bool			init_xattrs,
 	struct xfs_inode	**ipp)
 {
 	struct inode		*dir = pip ? VFS_I(pip) : NULL;
@@ -808,7 +794,7 @@ xfs_init_new_inode(
 	inode = VFS_I(ip);
 	set_nlink(inode, nlink);
 	inode->i_rdev = rdev;
-	ip->i_d.di_projid = prid;
+	ip->i_projid = prid;
 
 	if (dir && !(dir->i_mode & S_ISGID) &&
 	    (mp->m_flags & XFS_MOUNT_GRPID)) {
@@ -829,25 +815,22 @@ xfs_init_new_inode(
 	    !in_group_p(i_gid_into_mnt(mnt_userns, inode)))
 		inode->i_mode &= ~S_ISGID;
 
-	ip->i_d.di_size = 0;
+	ip->i_disk_size = 0;
 	ip->i_df.if_nextents = 0;
-	ASSERT(ip->i_d.di_nblocks == 0);
+	ASSERT(ip->i_nblocks == 0);
 
 	tv = current_time(inode);
 	inode->i_mtime = tv;
 	inode->i_atime = tv;
 	inode->i_ctime = tv;
 
-	ip->i_d.di_extsize = 0;
-	ip->i_d.di_dmevmask = 0;
-	ip->i_d.di_dmstate = 0;
-	ip->i_d.di_flags = 0;
+	ip->i_extsize = 0;
+	ip->i_diflags = 0;
 
 	if (xfs_sb_version_has_v3inode(&mp->m_sb)) {
 		inode_set_iversion(inode, 1);
-		ip->i_d.di_flags2 = mp->m_ino_geo.new_diflags2;
-		ip->i_d.di_cowextsize = 0;
-		ip->i_d.di_crtime = tv;
+		ip->i_cowextsize = 0;
+		ip->i_crtime = tv;
 	}
 
 	flags = XFS_ILOG_CORE;
@@ -857,19 +840,17 @@ xfs_init_new_inode(
 	case S_IFBLK:
 	case S_IFSOCK:
 		ip->i_df.if_format = XFS_DINODE_FMT_DEV;
-		ip->i_df.if_flags = 0;
 		flags |= XFS_ILOG_DEV;
 		break;
 	case S_IFREG:
 	case S_IFDIR:
-		if (pip && (pip->i_d.di_flags & XFS_DIFLAG_ANY))
+		if (pip && (pip->i_diflags & XFS_DIFLAG_ANY))
 			xfs_inode_inherit_flags(ip, pip);
-		if (pip && (pip->i_d.di_flags2 & XFS_DIFLAG2_ANY))
+		if (pip && (pip->i_diflags2 & XFS_DIFLAG2_ANY))
 			xfs_inode_inherit_flags2(ip, pip);
 		/* FALLTHROUGH */
 	case S_IFLNK:
 		ip->i_df.if_format = XFS_DINODE_FMT_EXTENTS;
-		ip->i_df.if_flags = XFS_IFEXTENTS;
 		ip->i_df.if_bytes = 0;
 		ip->i_df.if_u1.if_root = NULL;
 		break;
@@ -878,6 +859,20 @@ xfs_init_new_inode(
 	}
 
 	/*
+	 * If we need to create attributes immediately after allocating the
+	 * inode, initialise an empty attribute fork right now. We use the
+	 * default fork offset for attributes here as we don't know exactly what
+	 * size or how many attributes we might be adding. We can do this
+	 * safely here because we know the data fork is completely empty and
+	 * this saves us from needing to run a separate transaction to set the
+	 * fork offset in the immediate future.
+	 */
+	if (init_xattrs && xfs_sb_version_hasattr(&mp->m_sb)) {
+		ip->i_forkoff = xfs_default_attroffset(ip) >> 3;
+		ip->i_afp = xfs_ifork_alloc(XFS_DINODE_FMT_EXTENTS, 0);
+	}
+
+	/*
 	 * Log the new values stuffed into the inode.
 	 */
 	xfs_trans_ijoin(tp, ip, XFS_ILOCK_EXCL);
@@ -910,6 +905,7 @@ xfs_dir_ialloc(
 	xfs_nlink_t		nlink,
 	dev_t			rdev,
 	prid_t			prid,
+	bool			init_xattrs,
 	struct xfs_inode	**ipp)
 {
 	struct xfs_buf		*agibp;
@@ -937,7 +933,7 @@ xfs_dir_ialloc(
 	ASSERT(ino != NULLFSINO);
 
 	return xfs_init_new_inode(mnt_userns, *tpp, dp, ino, mode, nlink, rdev,
-				  prid, ipp);
+				  prid, init_xattrs, ipp);
 }
 
 /*
@@ -982,6 +978,7 @@ xfs_create(
 	struct xfs_name		*name,
 	umode_t			mode,
 	dev_t			rdev,
+	bool			init_xattrs,
 	xfs_inode_t		**ipp)
 {
 	int			is_dir = S_ISDIR(mode);
@@ -1053,7 +1050,7 @@ xfs_create(
 	 * pointing to itself.
 	 */
 	error = xfs_dir_ialloc(mnt_userns, &tp, dp, mode, is_dir ? 2 : 1, rdev,
-			       prid, &ip);
+			       prid, init_xattrs, &ip);
 	if (error)
 		goto out_trans_cancel;
 
@@ -1173,7 +1170,8 @@ xfs_create_tmpfile(
 	if (error)
 		goto out_release_dquots;
 
-	error = xfs_dir_ialloc(mnt_userns, &tp, dp, mode, 0, 0, prid, &ip);
+	error = xfs_dir_ialloc(mnt_userns, &tp, dp, mode, 0, 0, prid,
+				false, &ip);
 	if (error)
 		goto out_trans_cancel;
 
@@ -1272,8 +1270,8 @@ xfs_link(
 	 * creation in our tree when the project IDs are the same; else
 	 * the tree quota mechanism could be circumvented.
 	 */
-	if (unlikely((tdp->i_d.di_flags & XFS_DIFLAG_PROJINHERIT) &&
-		     tdp->i_d.di_projid != sip->i_d.di_projid)) {
+	if (unlikely((tdp->i_diflags & XFS_DIFLAG_PROJINHERIT) &&
+		     tdp->i_projid != sip->i_projid)) {
 		error = -EXDEV;
 		goto error_return;
 	}
@@ -1331,7 +1329,7 @@ xfs_itruncate_clear_reflink_flags(
 	dfork = XFS_IFORK_PTR(ip, XFS_DATA_FORK);
 	cfork = XFS_IFORK_PTR(ip, XFS_COW_FORK);
 	if (dfork->if_bytes == 0 && cfork->if_bytes == 0)
-		ip->i_d.di_flags2 &= ~XFS_DIFLAG2_REFLINK;
+		ip->i_diflags2 &= ~XFS_DIFLAG2_REFLINK;
 	if (cfork->if_bytes == 0)
 		xfs_inode_clear_cowblocks_tag(ip);
 }
@@ -1442,7 +1440,7 @@ xfs_release(
 	xfs_inode_t	*ip)
 {
 	xfs_mount_t	*mp = ip->i_mount;
-	int		error;
+	int		error = 0;
 
 	if (!S_ISREG(VFS_I(ip)->i_mode) || (VFS_I(ip)->i_mode == 0))
 		return 0;
@@ -1478,8 +1476,16 @@ xfs_release(
 	if (VFS_I(ip)->i_nlink == 0)
 		return 0;
 
-	if (xfs_can_free_eofblocks(ip, false)) {
+	/*
+	 * If we can't get the iolock just skip truncating the blocks past EOF
+	 * because we could deadlock with the mmap_lock otherwise. We'll get
+	 * another chance to drop them once the last reference to the inode is
+	 * dropped, so we'll never leak blocks permanently.
+	 */
+	if (!xfs_ilock_nowait(ip, XFS_IOLOCK_EXCL))
+		return 0;
 
+	if (xfs_can_free_eofblocks(ip, false)) {
 		/*
 		 * Check if the inode is being opened, written and closed
 		 * frequently and we have delayed allocation blocks outstanding
@@ -1495,26 +1501,20 @@ xfs_release(
 		 * place.
 		 */
 		if (xfs_iflags_test(ip, XFS_IDIRTY_RELEASE))
-			return 0;
-		/*
-		 * If we can't get the iolock just skip truncating the blocks
-		 * past EOF because we could deadlock with the mmap_lock
-		 * otherwise. We'll get another chance to drop them once the
-		 * last reference to the inode is dropped, so we'll never leak
-		 * blocks permanently.
-		 */
-		if (xfs_ilock_nowait(ip, XFS_IOLOCK_EXCL)) {
-			error = xfs_free_eofblocks(ip);
-			xfs_iunlock(ip, XFS_IOLOCK_EXCL);
-			if (error)
-				return error;
-		}
+			goto out_unlock;
+
+		error = xfs_free_eofblocks(ip);
+		if (error)
+			goto out_unlock;
 
 		/* delalloc blocks after truncation means it really is dirty */
 		if (ip->i_delayed_blks)
 			xfs_iflags_set(ip, XFS_IDIRTY_RELEASE);
 	}
-	return 0;
+
+out_unlock:
+	xfs_iunlock(ip, XFS_IOLOCK_EXCL);
+	return error;
 }
 
 /*
@@ -1543,7 +1543,7 @@ xfs_inactive_truncate(
 	 * of a system crash before the truncate completes. See the related
 	 * comment in xfs_vn_setattr_size() for details.
 	 */
-	ip->i_d.di_size = 0;
+	ip->i_disk_size = 0;
 	xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
 
 	error = xfs_itruncate_extents(&tp, ip, XFS_DATA_FORK, 0);
@@ -1697,6 +1697,10 @@ xfs_inactive(
 	if (mp->m_flags & XFS_MOUNT_RDONLY)
 		return;
 
+	/* Metadata inodes require explicit resource cleanup. */
+	if (xfs_is_metadata_inode(ip))
+		return;
+
 	/* Try to clean out the cow blocks if there are any. */
 	if (xfs_inode_has_cow_data(ip))
 		xfs_reflink_cancel_cow_range(ip, 0, NULLFILEOFF, true);
@@ -1718,7 +1722,7 @@ xfs_inactive(
 	}
 
 	if (S_ISREG(VFS_I(ip)->i_mode) &&
-	    (ip->i_d.di_size != 0 || XFS_ISIZE(ip) != 0 ||
+	    (ip->i_disk_size != 0 || XFS_ISIZE(ip) != 0 ||
 	     ip->i_df.if_nextents > 0 || ip->i_delayed_blks > 0))
 		truncate = 1;
 
@@ -1745,7 +1749,7 @@ xfs_inactive(
 	}
 
 	ASSERT(!ip->i_afp);
-	ASSERT(ip->i_d.di_forkoff == 0);
+	ASSERT(ip->i_forkoff == 0);
 
 	/*
 	 * Free the inode.
@@ -2053,9 +2057,10 @@ xfs_iunlink_update_inode(
 
 	ASSERT(xfs_verify_agino_or_null(mp, agno, next_agino));
 
-	error = xfs_imap_to_bp(mp, tp, &ip->i_imap, &dip, &ibp, 0);
+	error = xfs_imap_to_bp(mp, tp, &ip->i_imap, &ibp);
 	if (error)
 		return error;
+	dip = xfs_buf_offset(ibp, ip->i_imap.im_boffset);
 
 	/* Make sure the old pointer isn't garbage. */
 	old_value = be32_to_cpu(dip->di_next_unlinked);
@@ -2180,13 +2185,14 @@ xfs_iunlink_map_ino(
 		return error;
 	}
 
-	error = xfs_imap_to_bp(mp, tp, imap, dipp, bpp, 0);
+	error = xfs_imap_to_bp(mp, tp, imap, bpp);
 	if (error) {
 		xfs_warn(mp, "%s: xfs_imap_to_bp returned error %d.",
 				__func__, error);
 		return error;
 	}
 
+	*dipp = xfs_buf_offset(*bpp, imap->im_boffset);
 	return 0;
 }
 
@@ -2564,8 +2570,8 @@ xfs_ifree(
 	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));
 	ASSERT(VFS_I(ip)->i_nlink == 0);
 	ASSERT(ip->i_df.if_nextents == 0);
-	ASSERT(ip->i_d.di_size == 0 || !S_ISREG(VFS_I(ip)->i_mode));
-	ASSERT(ip->i_d.di_nblocks == 0);
+	ASSERT(ip->i_disk_size == 0 || !S_ISREG(VFS_I(ip)->i_mode));
+	ASSERT(ip->i_nblocks == 0);
 
 	/*
 	 * Pull the on-disk inode from the AGI unlinked list.
@@ -2590,11 +2596,12 @@ xfs_ifree(
 	}
 
 	VFS_I(ip)->i_mode = 0;		/* mark incore inode as free */
-	ip->i_d.di_flags = 0;
-	ip->i_d.di_flags2 = ip->i_mount->m_ino_geo.new_diflags2;
-	ip->i_d.di_dmevmask = 0;
-	ip->i_d.di_forkoff = 0;		/* mark the attr fork not in use */
+	ip->i_diflags = 0;
+	ip->i_diflags2 = ip->i_mount->m_ino_geo.new_diflags2;
+	ip->i_forkoff = 0;		/* mark the attr fork not in use */
 	ip->i_df.if_format = XFS_DINODE_FMT_EXTENTS;
+	if (xfs_iflags_test(ip, XFS_IPRESERVE_DM_FIELDS))
+		xfs_iflags_clear(ip, XFS_IPRESERVE_DM_FIELDS);
 
 	/* Don't attempt to replay owner changes for a deleted inode */
 	spin_lock(&iip->ili_lock);
@@ -2870,7 +2877,7 @@ xfs_finish_rename(
 /*
  * xfs_cross_rename()
  *
- * responsible for handling RENAME_EXCHANGE flag in renameat2() sytemcall
+ * responsible for handling RENAME_EXCHANGE flag in renameat2() syscall
  */
 STATIC int
 xfs_cross_rename(
@@ -3103,8 +3110,8 @@ xfs_rename(
 	 * into our tree when the project IDs are the same; else the
 	 * tree quota mechanism would be circumvented.
 	 */
-	if (unlikely((target_dp->i_d.di_flags & XFS_DIFLAG_PROJINHERIT) &&
-		     target_dp->i_d.di_projid != src_ip->i_d.di_projid)) {
+	if (unlikely((target_dp->i_diflags & XFS_DIFLAG_PROJINHERIT) &&
+		     target_dp->i_projid != src_ip->i_projid)) {
 		error = -EXDEV;
 		goto out_trans_cancel;
 	}
@@ -3414,34 +3421,33 @@ xfs_iflush(
 		}
 	}
 	if (XFS_TEST_ERROR(ip->i_df.if_nextents + xfs_ifork_nextents(ip->i_afp) >
-				ip->i_d.di_nblocks, mp, XFS_ERRTAG_IFLUSH_5)) {
+				ip->i_nblocks, mp, XFS_ERRTAG_IFLUSH_5)) {
 		xfs_alert_tag(mp, XFS_PTAG_IFLUSH,
 			"%s: detected corrupt incore inode %Lu, "
 			"total extents = %d, nblocks = %Ld, ptr "PTR_FMT,
 			__func__, ip->i_ino,
 			ip->i_df.if_nextents + xfs_ifork_nextents(ip->i_afp),
-			ip->i_d.di_nblocks, ip);
+			ip->i_nblocks, ip);
 		goto flush_out;
 	}
-	if (XFS_TEST_ERROR(ip->i_d.di_forkoff > mp->m_sb.sb_inodesize,
+	if (XFS_TEST_ERROR(ip->i_forkoff > mp->m_sb.sb_inodesize,
 				mp, XFS_ERRTAG_IFLUSH_6)) {
 		xfs_alert_tag(mp, XFS_PTAG_IFLUSH,
 			"%s: bad inode %Lu, forkoff 0x%x, ptr "PTR_FMT,
-			__func__, ip->i_ino, ip->i_d.di_forkoff, ip);
+			__func__, ip->i_ino, ip->i_forkoff, ip);
 		goto flush_out;
 	}
 
 	/*
-	 * Inode item log recovery for v2 inodes are dependent on the
-	 * di_flushiter count for correct sequencing. We bump the flush
-	 * iteration count so we can detect flushes which postdate a log record
-	 * during recovery. This is redundant as we now log every change and
-	 * hence this can't happen but we need to still do it to ensure
-	 * backwards compatibility with old kernels that predate logging all
-	 * inode changes.
+	 * Inode item log recovery for v2 inodes are dependent on the flushiter
+	 * count for correct sequencing.  We bump the flush iteration count so
+	 * we can detect flushes which postdate a log record during recovery.
+	 * This is redundant as we now log every change and hence this can't
+	 * happen but we need to still do it to ensure backwards compatibility
+	 * with old kernels that predate logging all inode changes.
 	 */
 	if (!xfs_sb_version_has_v3inode(&mp->m_sb))
-		ip->i_d.di_flushiter++;
+		ip->i_flushiter++;
 
 	/*
 	 * If there are inline format data / attr forks attached to this inode,
@@ -3462,8 +3468,10 @@ xfs_iflush(
 	xfs_inode_to_disk(ip, dip, iip->ili_item.li_lsn);
 
 	/* Wrap, we never let the log put out DI_MAX_FLUSH */
-	if (ip->i_d.di_flushiter == DI_MAX_FLUSH)
-		ip->i_d.di_flushiter = 0;
+	if (!xfs_sb_version_has_v3inode(&mp->m_sb)) {
+		if (ip->i_flushiter == DI_MAX_FLUSH)
+			ip->i_flushiter = 0;
+	}
 
 	xfs_iflush_fork(ip, dip, iip, XFS_DATA_FORK);
 	if (XFS_IFORK_Q(ip))