1 files changed, 81 insertions, 86 deletions

diff --git a/fs/xfs/xfs_inode.c b/fs/xfs/xfs_inode.c
index 4e560e6a12c1..7605d8396596 100644
--- a/fs/xfs/xfs_inode.c
+++ b/fs/xfs/xfs_inode.c
@@ -50,6 +50,7 @@
 #include "xfs_log.h"
 #include "xfs_bmap_btree.h"
 #include "xfs_reflink.h"
+#include "xfs_dir2_priv.h"
 
 kmem_zone_t *xfs_inode_zone;
 
@@ -142,31 +143,31 @@ xfs_ilock_attr_map_shared(
 }
 
 /*
- * The xfs inode contains 3 multi-reader locks: the i_iolock the i_mmap_lock and
- * the i_lock.  This routine allows various combinations of the locks to be
- * obtained.
+ * In addition to i_rwsem in the VFS inode, the xfs inode contains 2
+ * multi-reader locks: i_mmap_lock and the i_lock.  This routine allows
+ * various combinations of the locks to be obtained.
  *
  * The 3 locks should always be ordered so that the IO lock is obtained first,
  * the mmap lock second and the ilock last in order to prevent deadlock.
  *
  * Basic locking order:
  *
- * i_iolock -> i_mmap_lock -> page_lock -> i_ilock
+ * i_rwsem -> i_mmap_lock -> page_lock -> i_ilock
  *
  * mmap_sem locking order:
  *
- * i_iolock -> page lock -> mmap_sem
+ * i_rwsem -> page lock -> mmap_sem
  * mmap_sem -> i_mmap_lock -> page_lock
  *
  * The difference in mmap_sem locking order mean that we cannot hold the
  * i_mmap_lock over syscall based read(2)/write(2) based IO. These IO paths can
  * fault in pages during copy in/out (for buffered IO) or require the mmap_sem
  * in get_user_pages() to map the user pages into the kernel address space for
- * direct IO. Similarly the i_iolock cannot be taken inside a page fault because
+ * direct IO. Similarly the i_rwsem cannot be taken inside a page fault because
  * page faults already hold the mmap_sem.
  *
  * Hence to serialise fully against both syscall and mmap based IO, we need to
- * take both the i_iolock and the i_mmap_lock. These locks should *only* be both
+ * take both the i_rwsem and the i_mmap_lock. These locks should *only* be both
  * taken in places where we need to invalidate the page cache in a race
  * free manner (e.g. truncate, hole punch and other extent manipulation
  * functions).
@@ -191,10 +192,13 @@ xfs_ilock(
 	       (XFS_ILOCK_SHARED | XFS_ILOCK_EXCL));
 	ASSERT((lock_flags & ~(XFS_LOCK_MASK | XFS_LOCK_SUBCLASS_MASK)) == 0);
 
-	if (lock_flags & XFS_IOLOCK_EXCL)
-		mrupdate_nested(&ip->i_iolock, XFS_IOLOCK_DEP(lock_flags));
-	else if (lock_flags & XFS_IOLOCK_SHARED)
-		mraccess_nested(&ip->i_iolock, XFS_IOLOCK_DEP(lock_flags));
+	if (lock_flags & XFS_IOLOCK_EXCL) {
+		down_write_nested(&VFS_I(ip)->i_rwsem,
+				  XFS_IOLOCK_DEP(lock_flags));
+	} else if (lock_flags & XFS_IOLOCK_SHARED) {
+		down_read_nested(&VFS_I(ip)->i_rwsem,
+				 XFS_IOLOCK_DEP(lock_flags));
+	}
 
 	if (lock_flags & XFS_MMAPLOCK_EXCL)
 		mrupdate_nested(&ip->i_mmaplock, XFS_MMAPLOCK_DEP(lock_flags));
@@ -240,10 +244,10 @@ xfs_ilock_nowait(
 	ASSERT((lock_flags & ~(XFS_LOCK_MASK | XFS_LOCK_SUBCLASS_MASK)) == 0);
 
 	if (lock_flags & XFS_IOLOCK_EXCL) {
-		if (!mrtryupdate(&ip->i_iolock))
+		if (!down_write_trylock(&VFS_I(ip)->i_rwsem))
 			goto out;
 	} else if (lock_flags & XFS_IOLOCK_SHARED) {
-		if (!mrtryaccess(&ip->i_iolock))
+		if (!down_read_trylock(&VFS_I(ip)->i_rwsem))
 			goto out;
 	}
 
@@ -271,9 +275,9 @@ out_undo_mmaplock:
 		mrunlock_shared(&ip->i_mmaplock);
 out_undo_iolock:
 	if (lock_flags & XFS_IOLOCK_EXCL)
-		mrunlock_excl(&ip->i_iolock);
+		up_write(&VFS_I(ip)->i_rwsem);
 	else if (lock_flags & XFS_IOLOCK_SHARED)
-		mrunlock_shared(&ip->i_iolock);
+		up_read(&VFS_I(ip)->i_rwsem);
 out:
 	return 0;
 }
@@ -310,9 +314,9 @@ xfs_iunlock(
 	ASSERT(lock_flags != 0);
 
 	if (lock_flags & XFS_IOLOCK_EXCL)
-		mrunlock_excl(&ip->i_iolock);
+		up_write(&VFS_I(ip)->i_rwsem);
 	else if (lock_flags & XFS_IOLOCK_SHARED)
-		mrunlock_shared(&ip->i_iolock);
+		up_read(&VFS_I(ip)->i_rwsem);
 
 	if (lock_flags & XFS_MMAPLOCK_EXCL)
 		mrunlock_excl(&ip->i_mmaplock);
@@ -345,7 +349,7 @@ xfs_ilock_demote(
 	if (lock_flags & XFS_MMAPLOCK_EXCL)
 		mrdemote(&ip->i_mmaplock);
 	if (lock_flags & XFS_IOLOCK_EXCL)
-		mrdemote(&ip->i_iolock);
+		downgrade_write(&VFS_I(ip)->i_rwsem);
 
 	trace_xfs_ilock_demote(ip, lock_flags, _RET_IP_);
 }
@@ -370,8 +374,9 @@ xfs_isilocked(
 
 	if (lock_flags & (XFS_IOLOCK_EXCL|XFS_IOLOCK_SHARED)) {
 		if (!(lock_flags & XFS_IOLOCK_SHARED))
-			return !!ip->i_iolock.mr_writer;
-		return rwsem_is_locked(&ip->i_iolock.mr_lock);
+			return !debug_locks ||
+				lockdep_is_held_type(&VFS_I(ip)->i_rwsem, 0);
+		return rwsem_is_locked(&VFS_I(ip)->i_rwsem);
 	}
 
 	ASSERT(0);
@@ -421,11 +426,7 @@ xfs_lock_inumorder(int lock_mode, int subclass)
 
 	if (lock_mode & (XFS_IOLOCK_SHARED|XFS_IOLOCK_EXCL)) {
 		ASSERT(subclass <= XFS_IOLOCK_MAX_SUBCLASS);
-		ASSERT(xfs_lockdep_subclass_ok(subclass +
-						XFS_IOLOCK_PARENT_VAL));
 		class += subclass << XFS_IOLOCK_SHIFT;
-		if (lock_mode & XFS_IOLOCK_PARENT)
-			class += XFS_IOLOCK_PARENT_VAL << XFS_IOLOCK_SHIFT;
 	}
 
 	if (lock_mode & (XFS_MMAPLOCK_SHARED|XFS_MMAPLOCK_EXCL)) {
@@ -477,8 +478,6 @@ xfs_lock_inodes(
 			    XFS_ILOCK_EXCL));
 	ASSERT(!(lock_mode & (XFS_IOLOCK_SHARED | XFS_MMAPLOCK_SHARED |
 			      XFS_ILOCK_SHARED)));
-	ASSERT(!(lock_mode & XFS_IOLOCK_EXCL) ||
-		inodes <= XFS_IOLOCK_MAX_SUBCLASS + 1);
 	ASSERT(!(lock_mode & XFS_MMAPLOCK_EXCL) ||
 		inodes <= XFS_MMAPLOCK_MAX_SUBCLASS + 1);
 	ASSERT(!(lock_mode & XFS_ILOCK_EXCL) ||
@@ -581,10 +580,8 @@ xfs_lock_two_inodes(
 	int			attempts = 0;
 	xfs_log_item_t		*lp;
 
-	if (lock_mode & (XFS_IOLOCK_SHARED|XFS_IOLOCK_EXCL)) {
-		ASSERT(!(lock_mode & (XFS_MMAPLOCK_SHARED|XFS_MMAPLOCK_EXCL)));
-		ASSERT(!(lock_mode & (XFS_ILOCK_SHARED|XFS_ILOCK_EXCL)));
-	} else if (lock_mode & (XFS_MMAPLOCK_SHARED|XFS_MMAPLOCK_EXCL))
+	ASSERT(!(lock_mode & (XFS_IOLOCK_SHARED|XFS_IOLOCK_EXCL)));
+	if (lock_mode & (XFS_MMAPLOCK_SHARED|XFS_MMAPLOCK_EXCL))
 		ASSERT(!(lock_mode & (XFS_ILOCK_SHARED|XFS_ILOCK_EXCL)));
 
 	ASSERT(ip0->i_ino != ip1->i_ino);
@@ -715,7 +712,6 @@ xfs_lookup(
 	if (XFS_FORCED_SHUTDOWN(dp->i_mount))
 		return -EIO;
 
-	xfs_ilock(dp, XFS_IOLOCK_SHARED);
 	error = xfs_dir_lookup(NULL, dp, name, &inum, ci_name);
 	if (error)
 		goto out_unlock;
@@ -724,14 +720,12 @@ xfs_lookup(
 	if (error)
 		goto out_free_name;
 
-	xfs_iunlock(dp, XFS_IOLOCK_SHARED);
 	return 0;
 
 out_free_name:
 	if (ci_name)
 		kmem_free(ci_name->name);
 out_unlock:
-	xfs_iunlock(dp, XFS_IOLOCK_SHARED);
 	*ipp = NULL;
 	return error;
 }
@@ -1215,8 +1209,7 @@ xfs_create(
 	if (error)
 		goto out_release_inode;
 
-	xfs_ilock(dp, XFS_IOLOCK_EXCL | XFS_ILOCK_EXCL |
-		      XFS_IOLOCK_PARENT | XFS_ILOCK_PARENT);
+	xfs_ilock(dp, XFS_ILOCK_EXCL | XFS_ILOCK_PARENT);
 	unlock_dp_on_error = true;
 
 	xfs_defer_init(&dfops, &first_block);
@@ -1252,7 +1245,7 @@ xfs_create(
 	 * the transaction cancel unlocking dp so don't do it explicitly in the
 	 * error path.
 	 */
-	xfs_trans_ijoin(tp, dp, XFS_IOLOCK_EXCL | XFS_ILOCK_EXCL);
+	xfs_trans_ijoin(tp, dp, XFS_ILOCK_EXCL);
 	unlock_dp_on_error = false;
 
 	error = xfs_dir_createname(tp, dp, name, ip->i_ino,
@@ -1325,7 +1318,7 @@ xfs_create(
 	xfs_qm_dqrele(pdqp);
 
 	if (unlock_dp_on_error)
-		xfs_iunlock(dp, XFS_IOLOCK_EXCL | XFS_ILOCK_EXCL);
+		xfs_iunlock(dp, XFS_ILOCK_EXCL);
 	return error;
 }
 
@@ -1466,11 +1459,10 @@ xfs_link(
 	if (error)
 		goto std_return;
 
-	xfs_ilock(tdp, XFS_IOLOCK_EXCL | XFS_IOLOCK_PARENT);
 	xfs_lock_two_inodes(sip, tdp, XFS_ILOCK_EXCL);
 
 	xfs_trans_ijoin(tp, sip, XFS_ILOCK_EXCL);
-	xfs_trans_ijoin(tp, tdp, XFS_IOLOCK_EXCL | XFS_ILOCK_EXCL);
+	xfs_trans_ijoin(tp, tdp, XFS_ILOCK_EXCL);
 
 	/*
 	 * If we are using project inheritance, we only allow hard link
@@ -1624,7 +1616,7 @@ xfs_itruncate_extents(
 
 	/* Remove all pending CoW reservations. */
 	error = xfs_reflink_cancel_cow_blocks(ip, &tp, first_unmap_block,
-			last_block);
+			last_block, true);
 	if (error)
 		goto out;
 
@@ -1701,32 +1693,34 @@ xfs_release(
 	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
+		 * (e.g. streaming writes from the NFS server), truncating the
+		 * blocks past EOF will cause fragmentation to occur.
+		 *
+		 * In this case don't do the truncation, but we have to be
+		 * careful how we detect this case. Blocks beyond EOF show up as
+		 * i_delayed_blks even when the inode is clean, so we need to
+		 * truncate them away first before checking for a dirty release.
+		 * Hence on the first dirty close we will still remove the
+		 * speculative allocation, but after that we will leave it in
+		 * 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_sem
-		 * otherwise.  We'll get another chance to drop them once the
+		 * 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.
-		 *
-		 * Further, check if the inode is being opened, written and
-		 * closed frequently and we have delayed allocation blocks
-		 * outstanding (e.g. streaming writes from the NFS server),
-		 * truncating the blocks past EOF will cause fragmentation to
-		 * occur.
-		 *
-		 * In this case don't do the truncation, either, but we have to
-		 * be careful how we detect this case. Blocks beyond EOF show
-		 * up as i_delayed_blks even when the inode is clean, so we
-		 * need to truncate them away first before checking for a dirty
-		 * release. Hence on the first dirty close we will still remove
-		 * the speculative allocation, but after that we will leave it
-		 * in place.
 		 */
-		if (xfs_iflags_test(ip, XFS_IDIRTY_RELEASE))
-			return 0;
-
-		error = xfs_free_eofblocks(mp, ip, true);
-		if (error && error != -EAGAIN)
-			return error;
+		if (xfs_ilock_nowait(ip, XFS_IOLOCK_EXCL)) {
+			error = xfs_free_eofblocks(ip);
+			xfs_iunlock(ip, XFS_IOLOCK_EXCL);
+			if (error)
+				return error;
+		}
 
 		/* delalloc blocks after truncation means it really is dirty */
 		if (ip->i_delayed_blks)
@@ -1801,22 +1795,23 @@ xfs_inactive_ifree(
 	int			error;
 
 	/*
-	 * The ifree transaction might need to allocate blocks for record
-	 * insertion to the finobt. We don't want to fail here at ENOSPC, so
-	 * allow ifree to dip into the reserved block pool if necessary.
-	 *
-	 * Freeing large sets of inodes generally means freeing inode chunks,
-	 * directory and file data blocks, so this should be relatively safe.
-	 * Only under severe circumstances should it be possible to free enough
-	 * inodes to exhaust the reserve block pool via finobt expansion while
-	 * at the same time not creating free space in the filesystem.
+	 * We try to use a per-AG reservation for any block needed by the finobt
+	 * tree, but as the finobt feature predates the per-AG reservation
+	 * support a degraded file system might not have enough space for the
+	 * reservation at mount time.  In that case try to dip into the reserved
+	 * pool and pray.
 	 *
 	 * Send a warning if the reservation does happen to fail, as the inode
 	 * now remains allocated and sits on the unlinked list until the fs is
 	 * repaired.
 	 */
-	error = xfs_trans_alloc(mp, &M_RES(mp)->tr_ifree,
-			XFS_IFREE_SPACE_RES(mp), 0, XFS_TRANS_RESERVE, &tp);
+	if (unlikely(mp->m_inotbt_nores)) {
+		error = xfs_trans_alloc(mp, &M_RES(mp)->tr_ifree,
+				XFS_IFREE_SPACE_RES(mp), 0, XFS_TRANS_RESERVE,
+				&tp);
+	} else {
+		error = xfs_trans_alloc(mp, &M_RES(mp)->tr_ifree, 0, 0, 0, &tp);
+	}
 	if (error) {
 		if (error == -ENOSPC) {
 			xfs_warn_ratelimited(mp,
@@ -1912,8 +1907,11 @@ xfs_inactive(
 		 * cache. Post-eof blocks must be freed, lest we end up with
 		 * broken free space accounting.
 		 */
-		if (xfs_can_free_eofblocks(ip, true))
-			xfs_free_eofblocks(mp, ip, false);
+		if (xfs_can_free_eofblocks(ip, true)) {
+			xfs_ilock(ip, XFS_IOLOCK_EXCL);
+			xfs_free_eofblocks(ip);
+			xfs_iunlock(ip, XFS_IOLOCK_EXCL);
+		}
 
 		return;
 	}
@@ -2041,7 +2039,6 @@ xfs_iunlink(
 	agi->agi_unlinked[bucket_index] = cpu_to_be32(agino);
 	offset = offsetof(xfs_agi_t, agi_unlinked) +
 		(sizeof(xfs_agino_t) * bucket_index);
-	xfs_trans_buf_set_type(tp, agibp, XFS_BLFT_AGI_BUF);
 	xfs_trans_log_buf(tp, agibp, offset,
 			  (offset + sizeof(xfs_agino_t) - 1));
 	return 0;
@@ -2133,7 +2130,6 @@ xfs_iunlink_remove(
 		agi->agi_unlinked[bucket_index] = cpu_to_be32(next_agino);
 		offset = offsetof(xfs_agi_t, agi_unlinked) +
 			(sizeof(xfs_agino_t) * bucket_index);
-		xfs_trans_buf_set_type(tp, agibp, XFS_BLFT_AGI_BUF);
 		xfs_trans_log_buf(tp, agibp, offset,
 				  (offset + sizeof(xfs_agino_t) - 1));
 	} else {
@@ -2579,10 +2575,9 @@ xfs_remove(
 		goto std_return;
 	}
 
-	xfs_ilock(dp, XFS_IOLOCK_EXCL | XFS_IOLOCK_PARENT);
 	xfs_lock_two_inodes(dp, ip, XFS_ILOCK_EXCL);
 
-	xfs_trans_ijoin(tp, dp, XFS_IOLOCK_EXCL | XFS_ILOCK_EXCL);
+	xfs_trans_ijoin(tp, dp, XFS_ILOCK_EXCL);
 	xfs_trans_ijoin(tp, ip, XFS_ILOCK_EXCL);
 
 	/*
@@ -2963,12 +2958,6 @@ xfs_rename(
 	 * whether the target directory is the same as the source
 	 * directory, we can lock from 2 to 4 inodes.
 	 */
-	if (!new_parent)
-		xfs_ilock(src_dp, XFS_IOLOCK_EXCL | XFS_IOLOCK_PARENT);
-	else
-		xfs_lock_two_inodes(src_dp, target_dp,
-				    XFS_IOLOCK_EXCL | XFS_IOLOCK_PARENT);
-
 	xfs_lock_inodes(inodes, num_inodes, XFS_ILOCK_EXCL);
 
 	/*
@@ -2976,9 +2965,9 @@ xfs_rename(
 	 * we can rely on either trans_commit or trans_cancel to unlock
 	 * them.
 	 */
-	xfs_trans_ijoin(tp, src_dp, XFS_IOLOCK_EXCL | XFS_ILOCK_EXCL);
+	xfs_trans_ijoin(tp, src_dp, XFS_ILOCK_EXCL);
 	if (new_parent)
-		xfs_trans_ijoin(tp, target_dp, XFS_IOLOCK_EXCL | XFS_ILOCK_EXCL);
+		xfs_trans_ijoin(tp, target_dp, XFS_ILOCK_EXCL);
 	xfs_trans_ijoin(tp, src_ip, XFS_ILOCK_EXCL);
 	if (target_ip)
 		xfs_trans_ijoin(tp, target_ip, XFS_ILOCK_EXCL);
@@ -3558,6 +3547,12 @@ xfs_iflush_int(
 	if (ip->i_d.di_version < 3)
 		ip->i_d.di_flushiter++;
 
+	/* Check the inline directory data. */
+	if (S_ISDIR(VFS_I(ip)->i_mode) &&
+	    ip->i_d.di_format == XFS_DINODE_FMT_LOCAL &&
+	    xfs_dir2_sf_verify(ip))
+		goto corrupt_out;
+
 	/*
 	 * Copy the dirty parts of the inode into the on-disk inode.  We always
 	 * copy out the core of the inode, because if the inode is dirty at all