Merge commit 'xfs/master'

Conflicts:

	fs/xfs/Makefile-linux-2.6

16 files changed, 402 insertions, 547 deletions

diff --git a/fs/xfs/linux-2.6/xfs_linux.h b/fs/xfs/linux-2.6/xfs_linux.h
index 3ca39c4e5d2a..e5143323e71f 100644
--- a/fs/xfs/linux-2.6/xfs_linux.h
+++ b/fs/xfs/linux-2.6/xfs_linux.h
@@ -99,7 +99,6 @@
 /*
  * Feature macros (disable/enable)
  */
-#undef  HAVE_REFCACHE	/* reference cache not needed for NFS in 2.6 */
 #define HAVE_SPLICE	/* a splice(2) exists in 2.6, but not in 2.4 */
 #ifdef CONFIG_SMP
 #define HAVE_PERCPU_SB	/* per cpu superblock counters are a 2.6 feature */
diff --git a/fs/xfs/linux-2.6/xfs_super.c b/fs/xfs/linux-2.6/xfs_super.c
index 21dfc9da235e..df24c6fe4ce4 100644
--- a/fs/xfs/linux-2.6/xfs_super.c
+++ b/fs/xfs/linux-2.6/xfs_super.c
@@ -896,7 +896,8 @@ xfs_fs_write_inode(
 	struct inode		*inode,
 	int			sync)
 {
-	int			error = 0, flags = FLUSH_INODE;
+	int			error = 0;
+	int			flags = 0;
 
 	xfs_itrace_entry(XFS_I(inode));
 	if (sync) {
diff --git a/fs/xfs/linux-2.6/xfs_vnode.h b/fs/xfs/linux-2.6/xfs_vnode.h
index b5ea418693b1..f200e0244082 100644
--- a/fs/xfs/linux-2.6/xfs_vnode.h
+++ b/fs/xfs/linux-2.6/xfs_vnode.h
@@ -73,12 +73,9 @@ typedef enum bhv_vrwlock {
 #define IO_INVIS	0x00020		/* don't update inode timestamps */
 
 /*
- * Flags for vop_iflush call
+ * Flags for xfs_inode_flush
  */
 #define FLUSH_SYNC		1	/* wait for flush to complete	*/
-#define FLUSH_INODE		2	/* flush the inode itself	*/
-#define FLUSH_LOG		4	/* force the last log entry for
-					 * this inode out to disk	*/
 
 /*
  * Flush/Invalidate options for vop_toss/flush/flushinval_pages.
diff --git a/fs/xfs/xfs_bmap.c b/fs/xfs/xfs_bmap.c
index 1c0a5a585a82..759b75b90b59 100644
--- a/fs/xfs/xfs_bmap.c
+++ b/fs/xfs/xfs_bmap.c
@@ -5869,6 +5869,10 @@ xfs_getbmap(
 		/* xfs_fsize_t last_byte = xfs_file_last_byte(ip); */
 		error = xfs_flush_pages(ip, (xfs_off_t)0,
 					       -1, 0, FI_REMAPF);
+		if (error) {
+			xfs_iunlock(ip, XFS_IOLOCK_SHARED);
+		return error;
+		}
 	}
 
 	ASSERT(whichfork == XFS_ATTR_FORK || ip->i_delayed_blks == 0);
diff --git a/fs/xfs/xfs_iget.c b/fs/xfs/xfs_iget.c
index f01b07687faf..a959e3336931 100644
--- a/fs/xfs/xfs_iget.c
+++ b/fs/xfs/xfs_iget.c
@@ -78,7 +78,6 @@ xfs_iget_core(
 	xfs_inode_t	*ip;
 	xfs_inode_t	*iq;
 	int		error;
-	xfs_icluster_t	*icl, *new_icl = NULL;
 	unsigned long	first_index, mask;
 	xfs_perag_t	*pag;
 	xfs_agino_t	agino;
@@ -229,11 +228,9 @@ finish_inode:
 	}
 
 	/*
-	 * This is a bit messy - we preallocate everything we _might_
-	 * need before we pick up the ici lock. That way we don't have to
-	 * juggle locks and go all the way back to the start.
+	 * Preload the radix tree so we can insert safely under the
+	 * write spinlock.
 	 */
-	new_icl = kmem_zone_alloc(xfs_icluster_zone, KM_SLEEP);
 	if (radix_tree_preload(GFP_KERNEL)) {
 		delay(1);
 		goto again;
@@ -241,17 +238,6 @@ finish_inode:
 	mask = ~(((XFS_INODE_CLUSTER_SIZE(mp) >> mp->m_sb.sb_inodelog)) - 1);
 	first_index = agino & mask;
 	write_lock(&pag->pag_ici_lock);
-
-	/*
-	 * Find the cluster if it exists
-	 */
-	icl = NULL;
-	if (radix_tree_gang_lookup(&pag->pag_ici_root, (void**)&iq,
-							first_index, 1)) {
-		if ((XFS_INO_TO_AGINO(mp, iq->i_ino) & mask) == first_index)
-			icl = iq->i_cluster;
-	}
-
 	/*
 	 * insert the new inode
 	 */
@@ -266,30 +252,13 @@ finish_inode:
 	}
 
 	/*
-	 * These values _must_ be set before releasing ihlock!
+	 * These values _must_ be set before releasing the radix tree lock!
 	 */
 	ip->i_udquot = ip->i_gdquot = NULL;
 	xfs_iflags_set(ip, XFS_INEW);
 
-	ASSERT(ip->i_cluster == NULL);
-
-	if (!icl) {
-		spin_lock_init(&new_icl->icl_lock);
-		INIT_HLIST_HEAD(&new_icl->icl_inodes);
-		icl = new_icl;
-		new_icl = NULL;
-	} else {
-		ASSERT(!hlist_empty(&icl->icl_inodes));
-	}
-	spin_lock(&icl->icl_lock);
-	hlist_add_head(&ip->i_cnode, &icl->icl_inodes);
-	ip->i_cluster = icl;
-	spin_unlock(&icl->icl_lock);
-
 	write_unlock(&pag->pag_ici_lock);
 	radix_tree_preload_end();
-	if (new_icl)
-		kmem_zone_free(xfs_icluster_zone, new_icl);
 
 	/*
 	 * Link ip to its mount and thread it on the mount's inode list.
@@ -528,18 +497,6 @@ xfs_iextract(
 	xfs_put_perag(mp, pag);
 
 	/*
-	 * Remove from cluster list
-	 */
-	mp = ip->i_mount;
-	spin_lock(&ip->i_cluster->icl_lock);
-	hlist_del(&ip->i_cnode);
-	spin_unlock(&ip->i_cluster->icl_lock);
-
-	/* was last inode in cluster? */
-	if (hlist_empty(&ip->i_cluster->icl_inodes))
-		kmem_zone_free(xfs_icluster_zone, ip->i_cluster);
-
-	/*
 	 * Remove from mount's inode list.
 	 */
 	XFS_MOUNT_ILOCK(mp);
diff --git a/fs/xfs/xfs_inode.c b/fs/xfs/xfs_inode.c
index a550546a7083..4e23a9bd5106 100644
--- a/fs/xfs/xfs_inode.c
+++ b/fs/xfs/xfs_inode.c
@@ -55,7 +55,6 @@
 
 kmem_zone_t *xfs_ifork_zone;
 kmem_zone_t *xfs_inode_zone;
-kmem_zone_t *xfs_icluster_zone;
 
 /*
  * Used in xfs_itruncate().  This is the maximum number of extents
@@ -126,6 +125,90 @@ xfs_inobp_check(
 #endif
 
 /*
+ * Find the buffer associated with the given inode map
+ * We do basic validation checks on the buffer once it has been
+ * retrieved from disk.
+ */
+STATIC int
+xfs_imap_to_bp(
+	xfs_mount_t	*mp,
+	xfs_trans_t	*tp,
+	xfs_imap_t	*imap,
+	xfs_buf_t	**bpp,
+	uint		buf_flags,
+	uint		imap_flags)
+{
+	int		error;
+	int		i;
+	int		ni;
+	xfs_buf_t	*bp;
+
+	error = xfs_trans_read_buf(mp, tp, mp->m_ddev_targp, imap->im_blkno,
+				   (int)imap->im_len, buf_flags, &bp);
+	if (error) {
+		if (error != EAGAIN) {
+			cmn_err(CE_WARN,
+				"xfs_imap_to_bp: xfs_trans_read_buf()returned "
+				"an error %d on %s.  Returning error.",
+				error, mp->m_fsname);
+		} else {
+			ASSERT(buf_flags & XFS_BUF_TRYLOCK);
+		}
+		return error;
+	}
+
+	/*
+	 * Validate the magic number and version of every inode in the buffer
+	 * (if DEBUG kernel) or the first inode in the buffer, otherwise.
+	 */
+#ifdef DEBUG
+	ni = BBTOB(imap->im_len) >> mp->m_sb.sb_inodelog;
+#else	/* usual case */
+	ni = 1;
+#endif
+
+	for (i = 0; i < ni; i++) {
+		int		di_ok;
+		xfs_dinode_t	*dip;
+
+		dip = (xfs_dinode_t *)xfs_buf_offset(bp,
+					(i << mp->m_sb.sb_inodelog));
+		di_ok = be16_to_cpu(dip->di_core.di_magic) == XFS_DINODE_MAGIC &&
+			    XFS_DINODE_GOOD_VERSION(dip->di_core.di_version);
+		if (unlikely(XFS_TEST_ERROR(!di_ok, mp,
+						XFS_ERRTAG_ITOBP_INOTOBP,
+						XFS_RANDOM_ITOBP_INOTOBP))) {
+			if (imap_flags & XFS_IMAP_BULKSTAT) {
+				xfs_trans_brelse(tp, bp);
+				return XFS_ERROR(EINVAL);
+			}
+			XFS_CORRUPTION_ERROR("xfs_imap_to_bp",
+						XFS_ERRLEVEL_HIGH, mp, dip);
+#ifdef DEBUG
+			cmn_err(CE_PANIC,
+					"Device %s - bad inode magic/vsn "
+					"daddr %lld #%d (magic=%x)",
+				XFS_BUFTARG_NAME(mp->m_ddev_targp),
+				(unsigned long long)imap->im_blkno, i,
+				be16_to_cpu(dip->di_core.di_magic));
+#endif
+			xfs_trans_brelse(tp, bp);
+			return XFS_ERROR(EFSCORRUPTED);
+		}
+	}
+
+	xfs_inobp_check(mp, bp);
+
+	/*
+	 * Mark the buffer as an inode buffer now that it looks good
+	 */
+	XFS_BUF_SET_VTYPE(bp, B_FS_INO);
+
+	*bpp = bp;
+	return 0;
+}
+
+/*
  * This routine is called to map an inode number within a file
  * system to the buffer containing the on-disk version of the
  * inode.  It returns a pointer to the buffer containing the
@@ -147,72 +230,19 @@ xfs_inotobp(
 	xfs_buf_t	**bpp,
 	int		*offset)
 {
-	int		di_ok;
 	xfs_imap_t	imap;
 	xfs_buf_t	*bp;
 	int		error;
-	xfs_dinode_t	*dip;
 
-	/*
-	 * Call the space management code to find the location of the
-	 * inode on disk.
-	 */
 	imap.im_blkno = 0;
 	error = xfs_imap(mp, tp, ino, &imap, XFS_IMAP_LOOKUP);
-	if (error != 0) {
-		cmn_err(CE_WARN,
-	"xfs_inotobp: xfs_imap()  returned an "
-	"error %d on %s.  Returning error.", error, mp->m_fsname);
+	if (error)
 		return error;
-	}
-
-	/*
-	 * If the inode number maps to a block outside the bounds of the
-	 * file system then return NULL rather than calling read_buf
-	 * and panicing when we get an error from the driver.
-	 */
-	if ((imap.im_blkno + imap.im_len) >
-	    XFS_FSB_TO_BB(mp, mp->m_sb.sb_dblocks)) {
-		cmn_err(CE_WARN,
-	"xfs_inotobp: inode number (%llu + %d) maps to a block outside the bounds "
-	"of the file system %s.  Returning EINVAL.",
-			(unsigned long long)imap.im_blkno,
-			imap.im_len, mp->m_fsname);
-		return XFS_ERROR(EINVAL);
-	}
 
-	/*
-	 * Read in the buffer.  If tp is NULL, xfs_trans_read_buf() will
-	 * default to just a read_buf() call.
-	 */
-	error = xfs_trans_read_buf(mp, tp, mp->m_ddev_targp, imap.im_blkno,
-				   (int)imap.im_len, XFS_BUF_LOCK, &bp);
-
-	if (error) {
-		cmn_err(CE_WARN,
-	"xfs_inotobp: xfs_trans_read_buf()  returned an "
-	"error %d on %s.  Returning error.", error, mp->m_fsname);
+	error = xfs_imap_to_bp(mp, tp, &imap, &bp, XFS_BUF_LOCK, 0);
+	if (error)
 		return error;
-	}
-	dip = (xfs_dinode_t *)xfs_buf_offset(bp, 0);
-	di_ok =
-		be16_to_cpu(dip->di_core.di_magic) == XFS_DINODE_MAGIC &&
-		XFS_DINODE_GOOD_VERSION(dip->di_core.di_version);
-	if (unlikely(XFS_TEST_ERROR(!di_ok, mp, XFS_ERRTAG_ITOBP_INOTOBP,
-			XFS_RANDOM_ITOBP_INOTOBP))) {
-		XFS_CORRUPTION_ERROR("xfs_inotobp", XFS_ERRLEVEL_LOW, mp, dip);
-		xfs_trans_brelse(tp, bp);
-		cmn_err(CE_WARN,
-	"xfs_inotobp: XFS_TEST_ERROR()  returned an "
-	"error on %s.  Returning EFSCORRUPTED.",  mp->m_fsname);
-		return XFS_ERROR(EFSCORRUPTED);
-	}
-
-	xfs_inobp_check(mp, bp);
 
-	/*
-	 * Set *dipp to point to the on-disk inode in the buffer.
-	 */
 	*dipp = (xfs_dinode_t *)xfs_buf_offset(bp, imap.im_boffset);
 	*bpp = bp;
 	*offset = imap.im_boffset;
@@ -248,46 +278,21 @@ xfs_itobp(
 	xfs_dinode_t	**dipp,
 	xfs_buf_t	**bpp,
 	xfs_daddr_t	bno,
-	uint		imap_flags)
+	uint		imap_flags,
+	uint		buf_flags)
 {
 	xfs_imap_t	imap;
 	xfs_buf_t	*bp;
 	int		error;
-	int		i;
-	int		ni;
 
 	if (ip->i_blkno == (xfs_daddr_t)0) {
-		/*
-		 * Call the space management code to find the location of the
-		 * inode on disk.
-		 */
 		imap.im_blkno = bno;
-		if ((error = xfs_imap(mp, tp, ip->i_ino, &imap,
-					XFS_IMAP_LOOKUP | imap_flags)))
+		error = xfs_imap(mp, tp, ip->i_ino, &imap,
+					XFS_IMAP_LOOKUP | imap_flags);
+		if (error)
 			return error;
 
 		/*
-		 * If the inode number maps to a block outside the bounds
-		 * of the file system then return NULL rather than calling
-		 * read_buf and panicing when we get an error from the
-		 * driver.
-		 */
-		if ((imap.im_blkno + imap.im_len) >
-		    XFS_FSB_TO_BB(mp, mp->m_sb.sb_dblocks)) {
-#ifdef DEBUG
-			xfs_fs_cmn_err(CE_ALERT, mp, "xfs_itobp: "
-					"(imap.im_blkno (0x%llx) "
-					"+ imap.im_len (0x%llx)) > "
-					" XFS_FSB_TO_BB(mp, "
-					"mp->m_sb.sb_dblocks) (0x%llx)",
-					(unsigned long long) imap.im_blkno,
-					(unsigned long long) imap.im_len,
-					XFS_FSB_TO_BB(mp, mp->m_sb.sb_dblocks));
-#endif /* DEBUG */
-			return XFS_ERROR(EINVAL);
-		}
-
-		/*
 		 * Fill in the fields in the inode that will be used to
 		 * map the inode to its buffer from now on.
 		 */
@@ -305,76 +310,17 @@ xfs_itobp(
 	}
 	ASSERT(bno == 0 || bno == imap.im_blkno);
 
-	/*
-	 * Read in the buffer.  If tp is NULL, xfs_trans_read_buf() will
-	 * default to just a read_buf() call.
-	 */
-	error = xfs_trans_read_buf(mp, tp, mp->m_ddev_targp, imap.im_blkno,
-				   (int)imap.im_len, XFS_BUF_LOCK, &bp);
-	if (error) {
-#ifdef DEBUG
-		xfs_fs_cmn_err(CE_ALERT, mp, "xfs_itobp: "
-				"xfs_trans_read_buf() returned error %d, "
-				"imap.im_blkno 0x%llx, imap.im_len 0x%llx",
-				error, (unsigned long long) imap.im_blkno,
-				(unsigned long long) imap.im_len);
-#endif /* DEBUG */
+	error = xfs_imap_to_bp(mp, tp, &imap, &bp, buf_flags, imap_flags);
+	if (error)
 		return error;
-	}
 
-	/*
-	 * Validate the magic number and version of every inode in the buffer
-	 * (if DEBUG kernel) or the first inode in the buffer, otherwise.
-	 * No validation is done here in userspace (xfs_repair).
-	 */
-#if !defined(__KERNEL__)
-	ni = 0;
-#elif defined(DEBUG)
-	ni = BBTOB(imap.im_len) >> mp->m_sb.sb_inodelog;
-#else	/* usual case */
-	ni = 1;
-#endif
-
-	for (i = 0; i < ni; i++) {
-		int		di_ok;
-		xfs_dinode_t	*dip;
-
-		dip = (xfs_dinode_t *)xfs_buf_offset(bp,
-					(i << mp->m_sb.sb_inodelog));
-		di_ok = be16_to_cpu(dip->di_core.di_magic) == XFS_DINODE_MAGIC &&
-			    XFS_DINODE_GOOD_VERSION(dip->di_core.di_version);
-		if (unlikely(XFS_TEST_ERROR(!di_ok, mp,
-						XFS_ERRTAG_ITOBP_INOTOBP,
-						XFS_RANDOM_ITOBP_INOTOBP))) {
-			if (imap_flags & XFS_IMAP_BULKSTAT) {
-				xfs_trans_brelse(tp, bp);
-				return XFS_ERROR(EINVAL);
-			}
-#ifdef DEBUG
-			cmn_err(CE_ALERT,
-					"Device %s - bad inode magic/vsn "
-					"daddr %lld #%d (magic=%x)",
-				XFS_BUFTARG_NAME(mp->m_ddev_targp),
-				(unsigned long long)imap.im_blkno, i,
-				be16_to_cpu(dip->di_core.di_magic));
-#endif
-			XFS_CORRUPTION_ERROR("xfs_itobp", XFS_ERRLEVEL_HIGH,
-					     mp, dip);
-			xfs_trans_brelse(tp, bp);
-			return XFS_ERROR(EFSCORRUPTED);
-		}
+	if (!bp) {
+		ASSERT(buf_flags & XFS_BUF_TRYLOCK);
+		ASSERT(tp == NULL);
+		*bpp = NULL;
+		return EAGAIN;
 	}
 
-	xfs_inobp_check(mp, bp);
-
-	/*
-	 * Mark the buffer as an inode buffer now that it looks good
-	 */
-	XFS_BUF_SET_VTYPE(bp, B_FS_INO);
-
-	/*
-	 * Set *dipp to point to the on-disk inode in the buffer.
-	 */
 	*dipp = (xfs_dinode_t *)xfs_buf_offset(bp, imap.im_boffset);
 	*bpp = bp;
 	return 0;
@@ -878,7 +824,7 @@ xfs_iread(
 	 * return NULL as well.  Set i_blkno to 0 so that xfs_itobp() will
 	 * know that this is a new incore inode.
 	 */
-	error = xfs_itobp(mp, tp, ip, &dip, &bp, bno, imap_flags);
+	error = xfs_itobp(mp, tp, ip, &dip, &bp, bno, imap_flags, XFS_BUF_LOCK);
 	if (error) {
 		kmem_zone_free(xfs_inode_zone, ip);
 		return error;
@@ -1967,7 +1913,7 @@ xfs_iunlink(
 		 * Here we put the head pointer into our next pointer,
 		 * and then we fall through to point the head at us.
 		 */
-		error = xfs_itobp(mp, tp, ip, &dip, &ibp, 0, 0);
+		error = xfs_itobp(mp, tp, ip, &dip, &ibp, 0, 0, XFS_BUF_LOCK);
 		if (error)
 			return error;
 
@@ -2075,7 +2021,7 @@ xfs_iunlink_remove(
 		 * of dealing with the buffer when there is no need to
 		 * change it.
 		 */
-		error = xfs_itobp(mp, tp, ip, &dip, &ibp, 0, 0);
+		error = xfs_itobp(mp, tp, ip, &dip, &ibp, 0, 0, XFS_BUF_LOCK);
 		if (error) {
 			cmn_err(CE_WARN,
 				"xfs_iunlink_remove: xfs_itobp()  returned an error %d on %s.  Returning error.",
@@ -2137,7 +2083,7 @@ xfs_iunlink_remove(
 		 * Now last_ibp points to the buffer previous to us on
 		 * the unlinked list.  Pull us from the list.
 		 */
-		error = xfs_itobp(mp, tp, ip, &dip, &ibp, 0, 0);
+		error = xfs_itobp(mp, tp, ip, &dip, &ibp, 0, 0, XFS_BUF_LOCK);
 		if (error) {
 			cmn_err(CE_WARN,
 				"xfs_iunlink_remove: xfs_itobp()  returned an error %d on %s.  Returning error.",
@@ -2172,13 +2118,6 @@ xfs_iunlink_remove(
 	return 0;
 }
 
-STATIC_INLINE int xfs_inode_clean(xfs_inode_t *ip)
-{
-	return (((ip->i_itemp == NULL) ||
-		!(ip->i_itemp->ili_format.ilf_fields & XFS_ILOG_ALL)) &&
-		(ip->i_update_core == 0));
-}
-
 STATIC void
 xfs_ifree_cluster(
 	xfs_inode_t	*free_ip,
@@ -2400,7 +2339,7 @@ xfs_ifree(
 
 	xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
 
-	error = xfs_itobp(ip->i_mount, tp, ip, &dip, &ibp, 0, 0);
+	error = xfs_itobp(ip->i_mount, tp, ip, &dip, &ibp, 0, 0, XFS_BUF_LOCK);
 	if (error)
 		return error;
 
@@ -2678,14 +2617,31 @@ xfs_imap(
 	fsbno = imap->im_blkno ?
 		XFS_DADDR_TO_FSB(mp, imap->im_blkno) : NULLFSBLOCK;
 	error = xfs_dilocate(mp, tp, ino, &fsbno, &len, &off, flags);
-	if (error != 0) {
+	if (error)
 		return error;
-	}
+
 	imap->im_blkno = XFS_FSB_TO_DADDR(mp, fsbno);
 	imap->im_len = XFS_FSB_TO_BB(mp, len);
 	imap->im_agblkno = XFS_FSB_TO_AGBNO(mp, fsbno);
 	imap->im_ioffset = (ushort)off;
 	imap->im_boffset = (ushort)(off << mp->m_sb.sb_inodelog);
+
+	/*
+	 * If the inode number maps to a block outside the bounds
+	 * of the file system then return NULL rather than calling
+	 * read_buf and panicing when we get an error from the
+	 * driver.
+	 */
+	if ((imap->im_blkno + imap->im_len) >
+	    XFS_FSB_TO_BB(mp, mp->m_sb.sb_dblocks)) {
+		xfs_fs_cmn_err(CE_ALERT, mp, "xfs_imap: "
+			"(imap->im_blkno (0x%llx) + imap->im_len (0x%llx)) > "
+			" XFS_FSB_TO_BB(mp, mp->m_sb.sb_dblocks) (0x%llx)",
+			(unsigned long long) imap->im_blkno,
+			(unsigned long long) imap->im_len,
+			XFS_FSB_TO_BB(mp, mp->m_sb.sb_dblocks));
+		return EINVAL;
+	}
 	return 0;
 }
 
@@ -2826,38 +2782,41 @@ xfs_iunpin(
 }
 
 /*
- * This is called to wait for the given inode to be unpinned.
- * It will sleep until this happens.  The caller must have the
- * inode locked in at least shared mode so that the buffer cannot
- * be subsequently pinned once someone is waiting for it to be
- * unpinned.
+ * This is called to unpin an inode. It can be directed to wait or to return
+ * immediately without waiting for the inode to be unpinned.  The caller must
+ * have the inode locked in at least shared mode so that the buffer cannot be
+ * subsequently pinned once someone is waiting for it to be unpinned.
  */
 STATIC void
-xfs_iunpin_wait(
-	xfs_inode_t	*ip)
+__xfs_iunpin_wait(
+	xfs_inode_t	*ip,
+	int		wait)
 {
-	xfs_inode_log_item_t	*iip;
-	xfs_lsn_t	lsn;
+	xfs_inode_log_item_t	*iip = ip->i_itemp;
 
 	ASSERT(ismrlocked(&ip->i_lock, MR_UPDATE | MR_ACCESS));
-
-	if (atomic_read(&ip->i_pincount) == 0) {
+	if (atomic_read(&ip->i_pincount) == 0)
 		return;
-	}
 
-	iip = ip->i_itemp;
-	if (iip && iip->ili_last_lsn) {
-		lsn = iip->ili_last_lsn;
-	} else {
-		lsn = (xfs_lsn_t)0;
-	}
+	/* Give the log a push to start the unpinning I/O */
+	xfs_log_force(ip->i_mount, (iip && iip->ili_last_lsn) ?
+				iip->ili_last_lsn : 0, XFS_LOG_FORCE);
+	if (wait)
+		wait_event(ip->i_ipin_wait, (atomic_read(&ip->i_pincount) == 0));
+}
 
-	/*
-	 * Give the log a push so we don't wait here too long.
-	 */
-	xfs_log_force(ip->i_mount, lsn, XFS_LOG_FORCE);
+static inline void
+xfs_iunpin_wait(
+	xfs_inode_t	*ip)
+{
+	__xfs_iunpin_wait(ip, 1);
+}
 
-	wait_event(ip->i_ipin_wait, (atomic_read(&ip->i_pincount) == 0));
+static inline void
+xfs_iunpin_nowait(
+	xfs_inode_t	*ip)
+{
+	__xfs_iunpin_wait(ip, 0);
 }
 
 
@@ -3027,6 +2986,145 @@ xfs_iflush_fork(
 	return 0;
 }
 
+STATIC int
+xfs_iflush_cluster(
+	xfs_inode_t	*ip,
+	xfs_buf_t	*bp)
+{
+	xfs_mount_t		*mp = ip->i_mount;
+	xfs_perag_t		*pag = xfs_get_perag(mp, ip->i_ino);
+	unsigned long		first_index, mask;
+	int			ilist_size;
+	xfs_inode_t		**ilist;
+	xfs_inode_t		*iq;
+	int			nr_found;
+	int			clcount = 0;
+	int			bufwasdelwri;
+	int			i;
+
+	ASSERT(pag->pagi_inodeok);
+	ASSERT(pag->pag_ici_init);
+
+	ilist_size = XFS_INODE_CLUSTER_SIZE(mp) * sizeof(xfs_inode_t *);
+	ilist = kmem_alloc(ilist_size, KM_MAYFAIL);
+	if (!ilist)
+		return 0;
+
+	mask = ~(((XFS_INODE_CLUSTER_SIZE(mp) >> mp->m_sb.sb_inodelog)) - 1);
+	first_index = XFS_INO_TO_AGINO(mp, ip->i_ino) & mask;
+	read_lock(&pag->pag_ici_lock);
+	/* really need a gang lookup range call here */
+	nr_found = radix_tree_gang_lookup(&pag->pag_ici_root, (void**)ilist,
+					first_index,
+					XFS_INODE_CLUSTER_SIZE(mp));
+	if (nr_found == 0)
+		goto out_free;
+
+	for (i = 0; i < nr_found; i++) {
+		iq = ilist[i];
+		if (iq == ip)
+			continue;
+		/* if the inode lies outside this cluster, we're done. */
+		if ((XFS_INO_TO_AGINO(mp, iq->i_ino) & mask) != first_index)
+			break;
+		/*
+		 * Do an un-protected check to see if the inode is dirty and
+		 * is a candidate for flushing.  These checks will be repeated
+		 * later after the appropriate locks are acquired.
+		 */
+		if (xfs_inode_clean(iq) && xfs_ipincount(iq) == 0)
+			continue;
+
+		/*
+		 * Try to get locks.  If any are unavailable or it is pinned,
+		 * then this inode cannot be flushed and is skipped.
+		 */
+
+		if (!xfs_ilock_nowait(iq, XFS_ILOCK_SHARED))
+			continue;
+		if (!xfs_iflock_nowait(iq)) {
+			xfs_iunlock(iq, XFS_ILOCK_SHARED);
+			continue;
+		}
+		if (xfs_ipincount(iq)) {
+			xfs_ifunlock(iq);
+			xfs_iunlock(iq, XFS_ILOCK_SHARED);
+			continue;
+		}
+
+		/*
+		 * arriving here means that this inode can be flushed.  First
+		 * re-check that it's dirty before flushing.
+		 */
+		if (!xfs_inode_clean(iq)) {
+			int	error;
+			error = xfs_iflush_int(iq, bp);
+			if (error) {
+				xfs_iunlock(iq, XFS_ILOCK_SHARED);
+				goto cluster_corrupt_out;
+			}
+			clcount++;
+		} else {
+			xfs_ifunlock(iq);
+		}
+		xfs_iunlock(iq, XFS_ILOCK_SHARED);
+	}
+
+	if (clcount) {
+		XFS_STATS_INC(xs_icluster_flushcnt);
+		XFS_STATS_ADD(xs_icluster_flushinode, clcount);
+	}
+
+out_free:
+	read_unlock(&pag->pag_ici_lock);
+	kmem_free(ilist, ilist_size);
+	return 0;
+
+
+cluster_corrupt_out:
+	/*
+	 * Corruption detected in the clustering loop.  Invalidate the
+	 * inode buffer and shut down the filesystem.
+	 */
+	read_unlock(&pag->pag_ici_lock);
+	/*
+	 * Clean up the buffer.  If it was B_DELWRI, just release it --
+	 * brelse can handle it with no problems.  If not, shut down the
+	 * filesystem before releasing the buffer.
+	 */
+	bufwasdelwri = XFS_BUF_ISDELAYWRITE(bp);
+	if (bufwasdelwri)
+		xfs_buf_relse(bp);
+
+	xfs_force_shutdown(mp, SHUTDOWN_CORRUPT_INCORE);
+
+	if (!bufwasdelwri) {
+		/*
+		 * Just like incore_relse: if we have b_iodone functions,
+		 * mark the buffer as an error and call them.  Otherwise
+		 * mark it as stale and brelse.
+		 */
+		if (XFS_BUF_IODONE_FUNC(bp)) {
+			XFS_BUF_CLR_BDSTRAT_FUNC(bp);
+			XFS_BUF_UNDONE(bp);
+			XFS_BUF_STALE(bp);
+			XFS_BUF_SHUT(bp);
+			XFS_BUF_ERROR(bp,EIO);
+			xfs_biodone(bp);
+		} else {
+			XFS_BUF_STALE(bp);
+			xfs_buf_relse(bp);
+		}
+	}
+
+	/*
+	 * Unlocks the flush lock
+	 */
+	xfs_iflush_abort(iq);
+	kmem_free(ilist, ilist_size);
+	return XFS_ERROR(EFSCORRUPTED);
+}
+
 /*
  * xfs_iflush() will write a modified inode's changes out to the
  * inode's on disk home.  The caller must have the inode lock held
@@ -3046,11 +3144,7 @@ xfs_iflush(
 	xfs_dinode_t		*dip;
 	xfs_mount_t		*mp;
 	int			error;
-	/* REFERENCED */
-	xfs_inode_t		*iq;
-	int			clcount;	/* count of inodes clustered */
-	int			bufwasdelwri;
-	struct hlist_node	*entry;
+	int			noblock = (flags == XFS_IFLUSH_ASYNC_NOBLOCK);
 	enum { INT_DELWRI = (1 << 0), INT_ASYNC = (1 << 1) };
 
 	XFS_STATS_INC(xs_iflush_count);
@@ -3067,8 +3161,7 @@ xfs_iflush(
 	 * If the inode isn't dirty, then just release the inode
 	 * flush lock and do nothing.
 	 */
-	if ((ip->i_update_core == 0) &&
-	    ((iip == NULL) || !(iip->ili_format.ilf_fields & XFS_ILOG_ALL))) {
+	if (xfs_inode_clean(ip)) {
 		ASSERT((iip != NULL) ?
 			 !(iip->ili_item.li_flags & XFS_LI_IN_AIL) : 1);
 		xfs_ifunlock(ip);
@@ -3076,11 +3169,21 @@ xfs_iflush(
 	}
 
 	/*
-	 * We can't flush the inode until it is unpinned, so
-	 * wait for it.  We know noone new can pin it, because
-	 * we are holding the inode lock shared and you need
-	 * to hold it exclusively to pin the inode.
+	 * We can't flush the inode until it is unpinned, so wait for it if we
+	 * are allowed to block.  We know noone new can pin it, because we are
+	 * holding the inode lock shared and you need to hold it exclusively to
+	 * pin the inode.
+	 *
+	 * If we are not allowed to block, force the log out asynchronously so
+	 * that when we come back the inode will be unpinned. If other inodes
+	 * in the same cluster are dirty, they will probably write the inode
+	 * out for us if they occur after the log force completes.
 	 */
+	if (noblock && xfs_ipincount(ip)) {
+		xfs_iunpin_nowait(ip);
+		xfs_ifunlock(ip);
+		return EAGAIN;
+	}
 	xfs_iunpin_wait(ip);
 
 	/*
@@ -3097,15 +3200,6 @@ xfs_iflush(
 	}
 
 	/*
-	 * Get the buffer containing the on-disk inode.
-	 */
-	error = xfs_itobp(mp, NULL, ip, &dip, &bp, 0, 0);
-	if (error) {
-		xfs_ifunlock(ip);
-		return error;
-	}
-
-	/*
 	 * Decide how buffer will be flushed out.  This is done before
 	 * the call to xfs_iflush_int because this field is zeroed by it.
 	 */
@@ -3121,6 +3215,7 @@ xfs_iflush(
 		case XFS_IFLUSH_DELWRI_ELSE_SYNC:
 			flags = 0;
 			break;
+		case XFS_IFLUSH_ASYNC_NOBLOCK:
 		case XFS_IFLUSH_ASYNC:
 		case XFS_IFLUSH_DELWRI_ELSE_ASYNC:
 			flags = INT_ASYNC;
@@ -3140,6 +3235,7 @@ xfs_iflush(
 		case XFS_IFLUSH_DELWRI:
 			flags = INT_DELWRI;
 			break;
+		case XFS_IFLUSH_ASYNC_NOBLOCK:
 		case XFS_IFLUSH_ASYNC:
 			flags = INT_ASYNC;
 			break;
@@ -3154,89 +3250,36 @@ xfs_iflush(
 	}
 
 	/*
-	 * First flush out the inode that xfs_iflush was called with.
+	 * Get the buffer containing the on-disk inode.
 	 */
-	error = xfs_iflush_int(ip, bp);
-	if (error) {
-		goto corrupt_out;
+	error = xfs_itobp(mp, NULL, ip, &dip, &bp, 0, 0,
+				noblock ? XFS_BUF_TRYLOCK : XFS_BUF_LOCK);
+	if (error || !bp) {
+		xfs_ifunlock(ip);
+		return error;
 	}
 
 	/*
-	 * inode clustering:
-	 * see if other inodes can be gathered into this write
+	 * First flush out the inode that xfs_iflush was called with.
 	 */
-	spin_lock(&ip->i_cluster->icl_lock);
-	ip->i_cluster->icl_buf = bp;
-
-	clcount = 0;
-	hlist_for_each_entry(iq, entry, &ip->i_cluster->icl_inodes, i_cnode) {
-		if (iq == ip)
-			continue;
-
-		/*
-		 * Do an un-protected check to see if the inode is dirty and
-		 * is a candidate for flushing.  These checks will be repeated
-		 * later after the appropriate locks are acquired.
-		 */
-		iip = iq->i_itemp;
-		if ((iq->i_update_core == 0) &&
-		    ((iip == NULL) ||
-		     !(iip->ili_format.ilf_fields & XFS_ILOG_ALL)) &&
-		      xfs_ipincount(iq) == 0) {
-			continue;
-		}
-
-		/*
-		 * Try to get locks.  If any are unavailable,
-		 * then this inode cannot be flushed and is skipped.
-		 */
-
-		/* get inode locks (just i_lock) */
-		if (xfs_ilock_nowait(iq, XFS_ILOCK_SHARED)) {
-			/* get inode flush lock */
-			if (xfs_iflock_nowait(iq)) {
-				/* check if pinned */
-				if (xfs_ipincount(iq) == 0) {
-					/* arriving here means that
-					 * this inode can be flushed.
-					 * first re-check that it's
-					 * dirty
-					 */
-					iip = iq->i_itemp;
-					if ((iq->i_update_core != 0)||
-					    ((iip != NULL) &&
-					     (iip->ili_format.ilf_fields & XFS_ILOG_ALL))) {
-						clcount++;
-						error = xfs_iflush_int(iq, bp);
-						if (error) {
-							xfs_iunlock(iq,
-								    XFS_ILOCK_SHARED);
-							goto cluster_corrupt_out;
-						}
-					} else {
-						xfs_ifunlock(iq);
-					}
-				} else {
-					xfs_ifunlock(iq);
-				}
-			}
-			xfs_iunlock(iq, XFS_ILOCK_SHARED);
-		}
-	}
-	spin_unlock(&ip->i_cluster->icl_lock);
-
-	if (clcount) {
-		XFS_STATS_INC(xs_icluster_flushcnt);
-		XFS_STATS_ADD(xs_icluster_flushinode, clcount);
-	}
+	error = xfs_iflush_int(ip, bp);
+	if (error)
+		goto corrupt_out;
 
 	/*
-	 * If the buffer is pinned then push on the log so we won't
+	 * If the buffer is pinned then push on the log now so we won't
 	 * get stuck waiting in the write for too long.
 	 */
-	if (XFS_BUF_ISPINNED(bp)){
+	if (XFS_BUF_ISPINNED(bp))
 		xfs_log_force(mp, (xfs_lsn_t)0, XFS_LOG_FORCE);
-	}
+
+	/*
+	 * inode clustering:
+	 * see if other inodes can be gathered into this write
+	 */
+	error = xfs_iflush_cluster(ip, bp);
+	if (error)
+		goto cluster_corrupt_out;
 
 	if (flags & INT_DELWRI) {
 		xfs_bdwrite(mp, bp);
@@ -3250,52 +3293,11 @@ xfs_iflush(
 corrupt_out:
 	xfs_buf_relse(bp);
 	xfs_force_shutdown(mp, SHUTDOWN_CORRUPT_INCORE);
-	xfs_iflush_abort(ip);
-	/*
-	 * Unlocks the flush lock
-	 */
-	return XFS_ERROR(EFSCORRUPTED);
-
 cluster_corrupt_out:
-	/* Corruption detected in the clustering loop.  Invalidate the
-	 * inode buffer and shut down the filesystem.
-	 */
-	spin_unlock(&ip->i_cluster->icl_lock);
-
-	/*
-	 * Clean up the buffer.  If it was B_DELWRI, just release it --
-	 * brelse can handle it with no problems.  If not, shut down the
-	 * filesystem before releasing the buffer.
-	 */
-	if ((bufwasdelwri= XFS_BUF_ISDELAYWRITE(bp))) {
-		xfs_buf_relse(bp);
-	}
-
-	xfs_force_shutdown(mp, SHUTDOWN_CORRUPT_INCORE);
-
-	if(!bufwasdelwri)  {
-		/*
-		 * Just like incore_relse: if we have b_iodone functions,
-		 * mark the buffer as an error and call them.  Otherwise
-		 * mark it as stale and brelse.
-		 */
-		if (XFS_BUF_IODONE_FUNC(bp)) {
-			XFS_BUF_CLR_BDSTRAT_FUNC(bp);
-			XFS_BUF_UNDONE(bp);
-			XFS_BUF_STALE(bp);
-			XFS_BUF_SHUT(bp);
-			XFS_BUF_ERROR(bp,EIO);
-			xfs_biodone(bp);
-		} else {
-			XFS_BUF_STALE(bp);
-			xfs_buf_relse(bp);
-		}
-	}
-
-	xfs_iflush_abort(iq);
 	/*
 	 * Unlocks the flush lock
 	 */
+	xfs_iflush_abort(ip);
 	return XFS_ERROR(EFSCORRUPTED);
 }
 
@@ -3325,8 +3327,7 @@ xfs_iflush_int(
 	 * If the inode isn't dirty, then just release the inode
 	 * flush lock and do nothing.
 	 */
-	if ((ip->i_update_core == 0) &&
-	    ((iip == NULL) || !(iip->ili_format.ilf_fields & XFS_ILOG_ALL))) {
+	if (xfs_inode_clean(ip)) {
 		xfs_ifunlock(ip);
 		return 0;
 	}
diff --git a/fs/xfs/xfs_inode.h b/fs/xfs/xfs_inode.h
index bfcd72cbaeea..93c37697a72c 100644
--- a/fs/xfs/xfs_inode.h
+++ b/fs/xfs/xfs_inode.h
@@ -133,19 +133,6 @@ typedef struct dm_attrs_s {
 } dm_attrs_t;
 
 /*
- * This is the xfs inode cluster structure.  This structure is used by
- * xfs_iflush to find inodes that share a cluster and can be flushed to disk at
- * the same time.
- */
-typedef struct xfs_icluster {
-	struct hlist_head	icl_inodes;	/* list of inodes on cluster */
-	xfs_daddr_t		icl_blkno;	/* starting block number of
-						 * the cluster */
-	struct xfs_buf		*icl_buf;	/* the inode buffer */
-	spinlock_t		icl_lock;	/* inode list lock */
-} xfs_icluster_t;
-
-/*
  * This is the xfs in-core inode structure.
  * Most of the on-disk inode is embedded in the i_d field.
  *
@@ -240,10 +227,6 @@ typedef struct xfs_inode {
 	atomic_t		i_pincount;	/* inode pin count */
 	wait_queue_head_t	i_ipin_wait;	/* inode pinning wait queue */
 	spinlock_t		i_flags_lock;	/* inode i_flags lock */
-#ifdef HAVE_REFCACHE
-	struct xfs_inode	**i_refcache;	/* ptr to entry in ref cache */
-	struct xfs_inode	*i_release;	/* inode to unref */
-#endif
 	/* Miscellaneous state. */
 	unsigned short		i_flags;	/* see defined flags below */
 	unsigned char		i_update_core;	/* timestamps/size is dirty */
@@ -252,8 +235,6 @@ typedef struct xfs_inode {
 	unsigned int		i_delayed_blks;	/* count of delay alloc blks */
 
 	xfs_icdinode_t		i_d;		/* most of ondisk inode */
-	xfs_icluster_t		*i_cluster;	/* cluster list header */
-	struct hlist_node	i_cnode;	/* cluster link node */
 
 	xfs_fsize_t		i_size;		/* in-memory size */
 	xfs_fsize_t		i_new_size;	/* size when write completes */
@@ -461,6 +442,7 @@ xfs_iflags_test_and_clear(xfs_inode_t *ip, unsigned short flags)
 #define	XFS_IFLUSH_SYNC			3
 #define	XFS_IFLUSH_ASYNC		4
 #define	XFS_IFLUSH_DELWRI		5
+#define	XFS_IFLUSH_ASYNC_NOBLOCK	6
 
 /*
  * Flags for xfs_itruncate_start().
@@ -515,7 +497,7 @@ int		xfs_finish_reclaim_all(struct xfs_mount *, int);
  */
 int		xfs_itobp(struct xfs_mount *, struct xfs_trans *,
 			  xfs_inode_t *, struct xfs_dinode **, struct xfs_buf **,
-			  xfs_daddr_t, uint);
+			  xfs_daddr_t, uint, uint);
 int		xfs_iread(struct xfs_mount *, struct xfs_trans *, xfs_ino_t,
 			  xfs_inode_t **, xfs_daddr_t, uint);
 int		xfs_iread_extents(struct xfs_trans *, xfs_inode_t *, int);
@@ -597,7 +579,6 @@ void		xfs_inobp_check(struct xfs_mount *, struct xfs_buf *);
 #define	xfs_inobp_check(mp, bp)
 #endif /* DEBUG */
 
-extern struct kmem_zone	*xfs_icluster_zone;
 extern struct kmem_zone	*xfs_ifork_zone;
 extern struct kmem_zone	*xfs_inode_zone;
 extern struct kmem_zone	*xfs_ili_zone;
diff --git a/fs/xfs/xfs_inode_item.h b/fs/xfs/xfs_inode_item.h
index bfe92ea17952..40513077ab36 100644
--- a/fs/xfs/xfs_inode_item.h
+++ b/fs/xfs/xfs_inode_item.h
@@ -168,6 +168,14 @@ static inline int xfs_ilog_fext(int w)
 	return (w == XFS_DATA_FORK ? XFS_ILOG_DEXT : XFS_ILOG_AEXT);
 }
 
+static inline int xfs_inode_clean(xfs_inode_t *ip)
+{
+	return (!ip->i_itemp ||
+		!(ip->i_itemp->ili_format.ilf_fields & XFS_ILOG_ALL)) &&
+	       !ip->i_update_core;
+}
+
+
 #ifdef __KERNEL__
 
 extern void xfs_inode_item_init(struct xfs_inode *, struct xfs_mount *);
diff --git a/fs/xfs/xfs_itable.c b/fs/xfs/xfs_itable.c
index 658aab6b1bbf..38390e7381de 100644
--- a/fs/xfs/xfs_itable.c
+++ b/fs/xfs/xfs_itable.c
@@ -614,7 +614,8 @@ xfs_bulkstat(
 							xfs_buf_relse(bp);
 						error = xfs_itobp(mp, NULL, ip,
 								&dip, &bp, bno,
-								XFS_IMAP_BULKSTAT);
+								XFS_IMAP_BULKSTAT,
+								XFS_BUF_LOCK);
 						if (!error)
 							clustidx = ip->i_boffset / mp->m_sb.sb_inodesize;
 						kmem_zone_free(xfs_inode_zone, ip);
diff --git a/fs/xfs/xfs_log.c b/fs/xfs/xfs_log.c
index a75edca1860f..6439c89826dc 100644
--- a/fs/xfs/xfs_log.c
+++ b/fs/xfs/xfs_log.c
@@ -675,7 +675,7 @@ xfs_log_unmount_write(xfs_mount_t *mp)
 
 		spin_lock(&log->l_icloglock);
 		iclog = log->l_iclog;
-		iclog->ic_refcnt++;
+		atomic_inc(&iclog->ic_refcnt);
 		spin_unlock(&log->l_icloglock);
 		xlog_state_want_sync(log, iclog);
 		(void) xlog_state_release_iclog(log, iclog);
@@ -713,7 +713,7 @@ xfs_log_unmount_write(xfs_mount_t *mp)
 		 */
 		spin_lock(&log->l_icloglock);
 		iclog = log->l_iclog;
-		iclog->ic_refcnt++;
+		atomic_inc(&iclog->ic_refcnt);
 		spin_unlock(&log->l_icloglock);
 
 		xlog_state_want_sync(log, iclog);
@@ -1405,7 +1405,7 @@ xlog_sync(xlog_t		*log,
 	int		v2 = XFS_SB_VERSION_HASLOGV2(&log->l_mp->m_sb);
 
 	XFS_STATS_INC(xs_log_writes);
-	ASSERT(iclog->ic_refcnt == 0);
+	ASSERT(atomic_read(&iclog->ic_refcnt) == 0);
 
 	/* Add for LR header */
 	count_init = log->l_iclog_hsize + iclog->ic_offset;
@@ -2309,7 +2309,7 @@ xlog_state_done_syncing(
 
 	ASSERT(iclog->ic_state == XLOG_STATE_SYNCING ||
 	       iclog->ic_state == XLOG_STATE_IOERROR);
-	ASSERT(iclog->ic_refcnt == 0);
+	ASSERT(atomic_read(&iclog->ic_refcnt) == 0);
 	ASSERT(iclog->ic_bwritecnt == 1 || iclog->ic_bwritecnt == 2);
 
 
@@ -2391,7 +2391,7 @@ restart:
 	ASSERT(iclog->ic_state == XLOG_STATE_ACTIVE);
 	head = &iclog->ic_header;
 
-	iclog->ic_refcnt++;			/* prevents sync */
+	atomic_inc(&iclog->ic_refcnt);	/* prevents sync */
 	log_offset = iclog->ic_offset;
 
 	/* On the 1st write to an iclog, figure out lsn.  This works
@@ -2423,12 +2423,12 @@ restart:
 		xlog_state_switch_iclogs(log, iclog, iclog->ic_size);
 
 		/* If I'm the only one writing to this iclog, sync it to disk */
-		if (iclog->ic_refcnt == 1) {
+		if (atomic_read(&iclog->ic_refcnt) == 1) {
 			spin_unlock(&log->l_icloglock);
 			if ((error = xlog_state_release_iclog(log, iclog)))
 				return error;
 		} else {
-			iclog->ic_refcnt--;
+			atomic_dec(&iclog->ic_refcnt);
 			spin_unlock(&log->l_icloglock);
 		}
 		goto restart;
@@ -2813,33 +2813,35 @@ xlog_state_put_ticket(xlog_t	    *log,
  *
  */
 STATIC int
-xlog_state_release_iclog(xlog_t		*log,
-			 xlog_in_core_t	*iclog)
+xlog_state_release_iclog(
+	xlog_t		*log,
+	xlog_in_core_t	*iclog)
 {
 	int		sync = 0;	/* do we sync? */
 
-	xlog_assign_tail_lsn(log->l_mp);
+	if (iclog->ic_state & XLOG_STATE_IOERROR)
+		return XFS_ERROR(EIO);
 
-	spin_lock(&log->l_icloglock);
+	ASSERT(atomic_read(&iclog->ic_refcnt) > 0);
+	if (!atomic_dec_and_lock(&iclog->ic_refcnt, &log->l_icloglock))
+		return 0;
 
 	if (iclog->ic_state & XLOG_STATE_IOERROR) {
 		spin_unlock(&log->l_icloglock);
 		return XFS_ERROR(EIO);
 	}
-
-	ASSERT(iclog->ic_refcnt > 0);
 	ASSERT(iclog->ic_state == XLOG_STATE_ACTIVE ||
 	       iclog->ic_state == XLOG_STATE_WANT_SYNC);
 
-	if (--iclog->ic_refcnt == 0 &&
-	    iclog->ic_state == XLOG_STATE_WANT_SYNC) {
+	if (iclog->ic_state == XLOG_STATE_WANT_SYNC) {
+		/* update tail before writing to iclog */
+		xlog_assign_tail_lsn(log->l_mp);
 		sync++;
 		iclog->ic_state = XLOG_STATE_SYNCING;
 		iclog->ic_header.h_tail_lsn = cpu_to_be64(log->l_tail_lsn);
 		xlog_verify_tail_lsn(log, iclog, log->l_tail_lsn);
 		/* cycle incremented when incrementing curr_block */
 	}
-
 	spin_unlock(&log->l_icloglock);
 
 	/*
@@ -2849,11 +2851,9 @@ xlog_state_release_iclog(xlog_t		*log,
 	 * this iclog has consistent data, so we ignore IOERROR
 	 * flags after this point.
 	 */
-	if (sync) {
+	if (sync)
 		return xlog_sync(log, iclog);
-	}
 	return 0;
-
 }	/* xlog_state_release_iclog */
 
 
@@ -2953,7 +2953,8 @@ xlog_state_sync_all(xlog_t *log, uint flags, int *log_flushed)
 		 * previous iclog and go to sleep.
 		 */
 		if (iclog->ic_state == XLOG_STATE_DIRTY ||
-		    (iclog->ic_refcnt == 0 && iclog->ic_offset == 0)) {
+		    (atomic_read(&iclog->ic_refcnt) == 0
+		     && iclog->ic_offset == 0)) {
 			iclog = iclog->ic_prev;
 			if (iclog->ic_state == XLOG_STATE_ACTIVE ||
 			    iclog->ic_state == XLOG_STATE_DIRTY)
@@ -2961,14 +2962,14 @@ xlog_state_sync_all(xlog_t *log, uint flags, int *log_flushed)
 			else
 				goto maybe_sleep;
 		} else {
-			if (iclog->ic_refcnt == 0) {
+			if (atomic_read(&iclog->ic_refcnt) == 0) {
 				/* We are the only one with access to this
 				 * iclog.  Flush it out now.  There should
 				 * be a roundoff of zero to show that someone
 				 * has already taken care of the roundoff from
 				 * the previous sync.
 				 */
-				iclog->ic_refcnt++;
+				atomic_inc(&iclog->ic_refcnt);
 				lsn = be64_to_cpu(iclog->ic_header.h_lsn);
 				xlog_state_switch_iclogs(log, iclog, 0);
 				spin_unlock(&log->l_icloglock);
@@ -3100,7 +3101,7 @@ try_again:
 			already_slept = 1;
 			goto try_again;
 		} else {
-			iclog->ic_refcnt++;
+			atomic_inc(&iclog->ic_refcnt);
 			xlog_state_switch_iclogs(log, iclog, 0);
 			spin_unlock(&log->l_icloglock);
 			if (xlog_state_release_iclog(log, iclog))
diff --git a/fs/xfs/xfs_log_priv.h b/fs/xfs/xfs_log_priv.h
index e008233ee249..8662ce245c1f 100644
--- a/fs/xfs/xfs_log_priv.h
+++ b/fs/xfs/xfs_log_priv.h
@@ -339,7 +339,7 @@ typedef struct xlog_iclog_fields {
 #endif
 	int			ic_size;
 	int			ic_offset;
-	int			ic_refcnt;
+	atomic_t		ic_refcnt;
 	int			ic_bwritecnt;
 	ushort_t		ic_state;
 	char			*ic_datap;	/* pointer to iclog data */
diff --git a/fs/xfs/xfs_log_recover.c b/fs/xfs/xfs_log_recover.c
index b82d5d4d2462..d8a6d3089b16 100644
--- a/fs/xfs/xfs_log_recover.c
+++ b/fs/xfs/xfs_log_recover.c
@@ -3214,7 +3214,8 @@ xlog_recover_process_iunlinks(
 					 * next inode in the bucket.
 					 */
 					error = xfs_itobp(mp, NULL, ip, &dip,
-							&ibp, 0, 0);
+							&ibp, 0, 0,
+							XFS_BUF_LOCK);
 					ASSERT(error || (dip != NULL));
 				}
 
diff --git a/fs/xfs/xfs_rename.c b/fs/xfs/xfs_rename.c
index 7eb157a59f9e..1c6d40ed6816 100644
--- a/fs/xfs/xfs_rename.c
+++ b/fs/xfs/xfs_rename.c
@@ -36,7 +36,6 @@
 #include "xfs_bmap.h"
 #include "xfs_error.h"
 #include "xfs_quota.h"
-#include "xfs_refcache.h"
 #include "xfs_utils.h"
 #include "xfs_trans_space.h"
 #include "xfs_vnodeops.h"
@@ -580,10 +579,8 @@ xfs_rename(
 	 * the vnode references.
 	 */
 	error = xfs_trans_commit(tp, XFS_TRANS_RELEASE_LOG_RES);
-	if (target_ip != NULL) {
-		xfs_refcache_purge_ip(target_ip);
+	if (target_ip != NULL)
 		IRELE(target_ip);
-	}
 	/*
 	 * Let interposed file systems know about removed links.
 	 */
diff --git a/fs/xfs/xfs_trans_buf.c b/fs/xfs/xfs_trans_buf.c
index 60b6b898022b..4e5c010f5040 100644
--- a/fs/xfs/xfs_trans_buf.c
+++ b/fs/xfs/xfs_trans_buf.c
@@ -304,7 +304,8 @@ xfs_trans_read_buf(
 	if (tp == NULL) {
 		bp = xfs_buf_read_flags(target, blkno, len, flags | BUF_BUSY);
 		if (!bp)
-			return XFS_ERROR(ENOMEM);
+			return (flags & XFS_BUF_TRYLOCK) ?
+					EAGAIN : XFS_ERROR(ENOMEM);
 
 		if ((bp != NULL) && (XFS_BUF_GETERROR(bp) != 0)) {
 			xfs_ioerror_alert("xfs_trans_read_buf", mp,
diff --git a/fs/xfs/xfs_vfsops.c b/fs/xfs/xfs_vfsops.c
index 413587f02155..2cdf40f64ccc 100644
--- a/fs/xfs/xfs_vfsops.c
+++ b/fs/xfs/xfs_vfsops.c
@@ -43,7 +43,6 @@
 #include "xfs_error.h"
 #include "xfs_bmap.h"
 #include "xfs_rw.h"
-#include "xfs_refcache.h"
 #include "xfs_buf_item.h"
 #include "xfs_log_priv.h"
 #include "xfs_dir2_trace.h"
@@ -113,9 +112,6 @@ xfs_init(void)
 	xfs_ili_zone =
 		kmem_zone_init_flags(sizeof(xfs_inode_log_item_t), "xfs_ili",
 					KM_ZONE_SPREAD, NULL);
-	xfs_icluster_zone =
-		kmem_zone_init_flags(sizeof(xfs_icluster_t), "xfs_icluster",
-					KM_ZONE_SPREAD, NULL);
 
 	/*
 	 * Allocate global trace buffers.
@@ -153,11 +149,9 @@ xfs_cleanup(void)
 	extern kmem_zone_t	*xfs_inode_zone;
 	extern kmem_zone_t	*xfs_efd_zone;
 	extern kmem_zone_t	*xfs_efi_zone;
-	extern kmem_zone_t	*xfs_icluster_zone;
 
 	xfs_cleanup_procfs();
 	xfs_sysctl_unregister();
-	xfs_refcache_destroy();
 	xfs_filestream_uninit();
 	xfs_mru_cache_uninit();
 	xfs_acl_zone_destroy(xfs_acl_zone);
@@ -189,7 +183,6 @@ xfs_cleanup(void)
 	kmem_zone_destroy(xfs_efi_zone);
 	kmem_zone_destroy(xfs_ifork_zone);
 	kmem_zone_destroy(xfs_ili_zone);
-	kmem_zone_destroy(xfs_icluster_zone);
 }
 
 /*
@@ -585,11 +578,6 @@ xfs_unmount(
 					0 : DM_FLAGS_UNWANTED;
 	}
 #endif
-	/*
-	 * First blow any referenced inode from this file system
-	 * out of the reference cache, and delete the timer.
-	 */
-	xfs_refcache_purge_mp(mp);
 
 	/*
 	 * Blow away any referenced inode in the filestreams cache.
@@ -653,7 +641,6 @@ xfs_quiesce_fs(
 {
 	int			count = 0, pincount;
 
-	xfs_refcache_purge_mp(mp);
 	xfs_flush_buftarg(mp->m_ddev_targp, 0);
 	xfs_finish_reclaim_all(mp, 0);
 
@@ -1324,18 +1311,6 @@ xfs_syncsub(
 	}
 
 	/*
-	 * If this is the periodic sync, then kick some entries out of
-	 * the reference cache.  This ensures that idle entries are
-	 * eventually kicked out of the cache.
-	 */
-	if (flags & SYNC_REFCACHE) {
-		if (flags & SYNC_WAIT)
-			xfs_refcache_purge_mp(mp);
-		else
-			xfs_refcache_purge_some(mp);
-	}
-
-	/*
 	 * If asked, update the disk superblock with incore counter values if we
 	 * are using non-persistent counters so that they don't get too far out
 	 * of sync if we crash or get a forced shutdown. We don't want to force
diff --git a/fs/xfs/xfs_vnodeops.c b/fs/xfs/xfs_vnodeops.c
index 51305242ff8c..e9d2feb842ed 100644
--- a/fs/xfs/xfs_vnodeops.c
+++ b/fs/xfs/xfs_vnodeops.c
@@ -48,7 +48,6 @@
 #include "xfs_quota.h"
 #include "xfs_utils.h"
 #include "xfs_rtalloc.h"
-#include "xfs_refcache.h"
 #include "xfs_trans_space.h"
 #include "xfs_log_priv.h"
 #include "xfs_filestream.h"
@@ -1520,12 +1519,6 @@ xfs_release(
 			xfs_flush_pages(ip, 0, -1, XFS_B_ASYNC, FI_NONE);
 	}
 
-#ifdef HAVE_REFCACHE
-	/* If we are in the NFS reference cache then don't do this now */
-	if (ip->i_refcache)
-		return 0;
-#endif
-
 	if (ip->i_d.di_nlink != 0) {
 		if ((((ip->i_d.di_mode & S_IFMT) == S_IFREG) &&
 		     ((ip->i_size > 0) || (VN_CACHED(vp) > 0 ||
@@ -2449,14 +2442,6 @@ xfs_remove(
 	}
 
 	/*
-	 * Before we drop our extra reference to the inode, purge it
-	 * from the refcache if it is there.  By waiting until afterwards
-	 * to do the IRELE, we ensure that we won't go inactive in the
-	 * xfs_refcache_purge_ip routine (although that would be OK).
-	 */
-	xfs_refcache_purge_ip(ip);
-
-	/*
 	 * If we are using filestreams, kill the stream association.
 	 * If the file is still open it may get a new one but that
 	 * will get killed on last close in xfs_close() so we don't
@@ -2495,14 +2480,6 @@ xfs_remove(
 	cancel_flags |= XFS_TRANS_ABORT;
 	xfs_trans_cancel(tp, cancel_flags);
 
-	/*
-	 * Before we drop our extra reference to the inode, purge it
-	 * from the refcache if it is there.  By waiting until afterwards
-	 * to do the IRELE, we ensure that we won't go inactive in the
-	 * xfs_refcache_purge_ip routine (although that would be OK).
-	 */
-	xfs_refcache_purge_ip(ip);
-
 	IRELE(ip);
 
 	goto std_return;
@@ -3461,14 +3438,7 @@ xfs_rwunlock(
  	if (S_ISDIR(ip->i_d.di_mode))
   		return;
 	if (locktype == VRWLOCK_WRITE) {
-		/*
-		 * In the write case, we may have added a new entry to
-		 * the reference cache.  This might store a pointer to
-		 * an inode to be released in this inode.  If it is there,
-		 * clear the pointer and release the inode after unlocking
-		 * this one.
-		 */
-		xfs_refcache_iunlock(ip, XFS_IOLOCK_EXCL);
+		xfs_iunlock(ip, XFS_IOLOCK_EXCL);
 	} else {
 		ASSERT((locktype == VRWLOCK_READ) ||
 		       (locktype == VRWLOCK_WRITE_DIRECT));
@@ -3484,7 +3454,6 @@ xfs_inode_flush(
 	int		flags)
 {
 	xfs_mount_t	*mp = ip->i_mount;
-	xfs_inode_log_item_t *iip = ip->i_itemp;
 	int		error = 0;
 
 	if (XFS_FORCED_SHUTDOWN(mp))
@@ -3494,33 +3463,9 @@ xfs_inode_flush(
 	 * Bypass inodes which have already been cleaned by
 	 * the inode flush clustering code inside xfs_iflush
 	 */
-	if ((ip->i_update_core == 0) &&
-	    ((iip == NULL) || !(iip->ili_format.ilf_fields & XFS_ILOG_ALL)))
+	if (xfs_inode_clean(ip))
 		return 0;
 
-	if (flags & FLUSH_LOG) {
-		if (iip && iip->ili_last_lsn) {
-			xlog_t		*log = mp->m_log;
-			xfs_lsn_t	sync_lsn;
-			int		log_flags = XFS_LOG_FORCE;
-
-			spin_lock(&log->l_grant_lock);
-			sync_lsn = log->l_last_sync_lsn;
-			spin_unlock(&log->l_grant_lock);
-
-			if ((XFS_LSN_CMP(iip->ili_last_lsn, sync_lsn) > 0)) {
-				if (flags & FLUSH_SYNC)
-					log_flags |= XFS_LOG_SYNC;
-				error = xfs_log_force(mp, iip->ili_last_lsn, log_flags);
-				if (error)
-					return error;
-			}
-
-			if (ip->i_update_core == 0)
-				return 0;
-		}
-	}
-
 	/*
 	 * We make this non-blocking if the inode is contended,
 	 * return EAGAIN to indicate to the caller that they
@@ -3528,30 +3473,22 @@ xfs_inode_flush(
 	 * blocking on inodes inside another operation right
 	 * now, they get caught later by xfs_sync.
 	 */
-	if (flags & FLUSH_INODE) {
-		int	flush_flags;
-
-		if (flags & FLUSH_SYNC) {
-			xfs_ilock(ip, XFS_ILOCK_SHARED);
-			xfs_iflock(ip);
-		} else if (xfs_ilock_nowait(ip, XFS_ILOCK_SHARED)) {
-			if (xfs_ipincount(ip) || !xfs_iflock_nowait(ip)) {
-				xfs_iunlock(ip, XFS_ILOCK_SHARED);
-				return EAGAIN;
-			}
-		} else {
+	if (flags & FLUSH_SYNC) {
+		xfs_ilock(ip, XFS_ILOCK_SHARED);
+		xfs_iflock(ip);
+	} else if (xfs_ilock_nowait(ip, XFS_ILOCK_SHARED)) {
+		if (xfs_ipincount(ip) || !xfs_iflock_nowait(ip)) {
+			xfs_iunlock(ip, XFS_ILOCK_SHARED);
 			return EAGAIN;
 		}
-
-		if (flags & FLUSH_SYNC)
-			flush_flags = XFS_IFLUSH_SYNC;
-		else
-			flush_flags = XFS_IFLUSH_ASYNC;
-
-		error = xfs_iflush(ip, flush_flags);
-		xfs_iunlock(ip, XFS_ILOCK_SHARED);
+	} else {
+		return EAGAIN;
 	}
 
+	error = xfs_iflush(ip, (flags & FLUSH_SYNC) ? XFS_IFLUSH_SYNC
+						    : XFS_IFLUSH_ASYNC_NOBLOCK);
+	xfs_iunlock(ip, XFS_ILOCK_SHARED);
+
 	return error;
 }
 
@@ -3694,12 +3631,12 @@ xfs_finish_reclaim(
 	 * We get the flush lock regardless, though, just to make sure
 	 * we don't free it while it is being flushed.
 	 */
-	if (!XFS_FORCED_SHUTDOWN(ip->i_mount)) {
-		if (!locked) {
-			xfs_ilock(ip, XFS_ILOCK_EXCL);
-			xfs_iflock(ip);
-		}
+	if (!locked) {
+		xfs_ilock(ip, XFS_ILOCK_EXCL);
+		xfs_iflock(ip);
+	}
 
+	if (!XFS_FORCED_SHUTDOWN(ip->i_mount)) {
 		if (ip->i_update_core ||
 		    ((ip->i_itemp != NULL) &&
 		     (ip->i_itemp->ili_format.ilf_fields != 0))) {
@@ -3719,17 +3656,11 @@ xfs_finish_reclaim(
 		ASSERT(ip->i_update_core == 0);
 		ASSERT(ip->i_itemp == NULL ||
 		       ip->i_itemp->ili_format.ilf_fields == 0);
-		xfs_iunlock(ip, XFS_ILOCK_EXCL);
-	} else if (locked) {
-		/*
-		 * We are not interested in doing an iflush if we're
-		 * in the process of shutting down the filesystem forcibly.
-		 * So, just reclaim the inode.
-		 */
-		xfs_ifunlock(ip);
-		xfs_iunlock(ip, XFS_ILOCK_EXCL);
 	}
 
+	xfs_ifunlock(ip);
+	xfs_iunlock(ip, XFS_ILOCK_EXCL);
+
  reclaim:
 	xfs_ireclaim(ip);
 	return 0;