Merge git://oss.sgi.com:8090/xfs-2.6

* git://oss.sgi.com:8090/xfs-2.6: (43 commits)
  [XFS] Remove files from the build that are now unused.
  [XFS] Fix a Makefile issue related to exports.o handling.
  [XFS] Remove version 1 directory code.	Never functioned on Linux, just
  [XFS] Map EFSCORRUPTED to an actual error code, not just a made up one
  [XFS] Kill direct access to ->count in valusema(); all we ever use it for
  [XFS] Remove unneeded conditional code on NFS export interface related
  [XFS] Remove an incorrect use of unlikely() on a relatively likely code
  [XFS] Push some common code out of write path into core XFS code for
  [XFS] Remove unnecessary local from open_exec dmapi path.
  [XFS] Minor XFS documentation updates.
  [XFS] Fix broken const use inside local suffix_strtoul routine.
  [XFS] Fix nused counter.  It's currently getting set to -1 rather than
  [XFS] Fix mismerge of the fs_writable cleanup patch causing a freeze/thaw
  [XFS] Fix up debug code so that bulkstat wont generate thousands of
  [XFS] Remove unused parameter from di2xflags routine.
  [XFS] Cleanup a missed porting conversion, and freezing.
  [XFS] Resolve a namespace collision on remaining vtypes for FreeBSD
  [XFS] Resolve a namespace collision on vnode/vnodeops for FreeBSD porters.
  [XFS] Resolve a namespace collision on vfs/vfsops for FreeBSD porters.
  [XFS] statvfs component of directory/project quota support, code
  ...

MAINTAINERS

@@ -3191,7 +3191,7 @@ XFS FILESYSTEM
 P:	Silicon Graphics Inc
 M:	xfs-masters@oss.sgi.com
 M:	nathans@sgi.com
-L:	linux-xfs@oss.sgi.com
+L:	xfs@oss.sgi.com
 W:	http://oss.sgi.com/projects/xfs
 S:	Supported
 

fs/xfs/Kconfig

 config XFS_FS
 	tristate "XFS filesystem support"
-	select EXPORTFS if NFSD!=n
 	help
 	  XFS is a high performance journaling filesystem which originated
 	  on the SGI IRIX platform.  It is completely multi-threaded, can
@@ -18,11 +17,6 @@ config XFS_FS
 	  system of your root partition is compiled as a module, you'll need
 	  to use an initial ramdisk (initrd) to boot.
 
-config XFS_EXPORT
-	bool
-	depends on XFS_FS && EXPORTFS
-	default y
-
 config XFS_QUOTA
 	bool "XFS Quota support"
 	depends on XFS_FS
@@ -65,18 +59,19 @@ config XFS_POSIX_ACL
 	  If you don't know what Access Control Lists are, say N.
 
 config XFS_RT
-	bool "XFS Realtime support (EXPERIMENTAL)"
-	depends on XFS_FS && EXPERIMENTAL
+	bool "XFS Realtime subvolume support"
+	depends on XFS_FS
 	help
 	  If you say Y here you will be able to mount and use XFS filesystems
-	  which contain a realtime subvolume. The realtime subvolume is a
-	  separate area of disk space where only file data is stored. The
-	  realtime subvolume is designed to provide very deterministic
-	  data rates suitable for media streaming applications.
-
-	  See the xfs man page in section 5 for a bit more information.
+	  which contain a realtime subvolume.  The realtime subvolume is a
+	  separate area of disk space where only file data is stored.  It was
+	  originally designed to provide deterministic data rates suitable
+	  for media streaming applications, but is also useful as a generic
+	  mechanism for ensuring data and metadata/log I/Os are completely
+	  separated.  Regular file I/Os are isolated to a separate device
+	  from all other requests, and this can be done quite transparently
+	  to applications via the inherit-realtime directory inode flag.
 
-	  This feature is unsupported at this time, is not yet fully
-	  functional, and may cause serious problems.
+	  See the xfs man page in section 5 for additional information.
 
 	  If unsure, say N.

fs/xfs/Makefile-linux-2.6

@@ -59,7 +59,6 @@ xfs-$(CONFIG_XFS_POSIX_ACL)	+= xfs_acl.o
 xfs-$(CONFIG_PROC_FS)		+= $(XFS_LINUX)/xfs_stats.o
 xfs-$(CONFIG_SYSCTL)		+= $(XFS_LINUX)/xfs_sysctl.o
 xfs-$(CONFIG_COMPAT)		+= $(XFS_LINUX)/xfs_ioctl32.o
-xfs-$(CONFIG_XFS_EXPORT)	+= $(XFS_LINUX)/xfs_export.o
 
 
 xfs-y				+= xfs_alloc.o \
@@ -73,14 +72,12 @@ xfs-y				+= xfs_alloc.o \
 				   xfs_btree.o \
 				   xfs_buf_item.o \
 				   xfs_da_btree.o \
-				   xfs_dir.o \
 				   xfs_dir2.o \
 				   xfs_dir2_block.o \
 				   xfs_dir2_data.o \
 				   xfs_dir2_leaf.o \
 				   xfs_dir2_node.o \
 				   xfs_dir2_sf.o \
-				   xfs_dir_leaf.o \
 				   xfs_error.o \
 				   xfs_extfree_item.o \
 				   xfs_fsops.o \
@@ -117,6 +114,7 @@ xfs-y				+= $(addprefix $(XFS_LINUX)/, \
 				   kmem.o \
 				   xfs_aops.o \
 				   xfs_buf.o \
+				   xfs_export.o \
 				   xfs_file.o \
 				   xfs_fs_subr.o \
 				   xfs_globals.o \

fs/xfs/linux-2.6/kmem.h

@@ -22,42 +22,6 @@
 #include <linux/sched.h>
 #include <linux/mm.h>
 
-/*
- * Process flags handling
- */
-
-#define PFLAGS_TEST_NOIO()              (current->flags & PF_NOIO)
-#define PFLAGS_TEST_FSTRANS()           (current->flags & PF_FSTRANS)
-
-#define PFLAGS_SET_NOIO() do {		\
-	current->flags |= PF_NOIO;	\
-} while (0)
-
-#define PFLAGS_CLEAR_NOIO() do {	\
-	current->flags &= ~PF_NOIO;	\
-} while (0)
-
-/* these could be nested, so we save state */
-#define PFLAGS_SET_FSTRANS(STATEP) do {	\
-	*(STATEP) = current->flags;	\
-	current->flags |= PF_FSTRANS;	\
-} while (0)
-
-#define PFLAGS_CLEAR_FSTRANS(STATEP) do { \
-	*(STATEP) = current->flags;	\
-	current->flags &= ~PF_FSTRANS;	\
-} while (0)
-
-/* Restore the PF_FSTRANS state to what was saved in STATEP */
-#define PFLAGS_RESTORE_FSTRANS(STATEP) do {     		\
-	current->flags = ((current->flags & ~PF_FSTRANS) |	\
-			  (*(STATEP) & PF_FSTRANS));		\
-} while (0)
-
-#define PFLAGS_DUP(OSTATEP, NSTATEP) do { \
-	*(NSTATEP) = *(OSTATEP);	\
-} while (0)
-
 /*
  * General memory allocation interfaces
  */
@@ -83,7 +47,7 @@ kmem_flags_convert(unsigned int __nocast flags)
 		lflags = GFP_ATOMIC | __GFP_NOWARN;
 	} else {
 		lflags = GFP_KERNEL | __GFP_NOWARN;
-		if (PFLAGS_TEST_FSTRANS() || (flags & KM_NOFS))
+		if ((current->flags & PF_FSTRANS) || (flags & KM_NOFS))
 			lflags &= ~__GFP_FS;
 	}
 	return lflags;

fs/xfs/linux-2.6/mrlock.h

 /*
- * Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * Copyright (c) 2000-2006 Silicon Graphics, Inc.
  * All Rights Reserved.
  *
  * This program is free software; you can redistribute it and/or
@@ -28,7 +28,7 @@ typedef struct {
 } mrlock_t;
 
 #define mrinit(mrp, name)	\
-	( (mrp)->mr_writer = 0, init_rwsem(&(mrp)->mr_lock) )
+	do { (mrp)->mr_writer = 0; init_rwsem(&(mrp)->mr_lock); } while (0)
 #define mrlock_init(mrp, t,n,s)	mrinit(mrp, n)
 #define mrfree(mrp)		do { } while (0)
 #define mraccess(mrp)		mraccessf(mrp, 0)

fs/xfs/linux-2.6/sema.h

@@ -34,20 +34,21 @@ typedef struct semaphore sema_t;
 #define initnsema(sp, val, name)	sema_init(sp, val)
 #define psema(sp, b)			down(sp)
 #define vsema(sp)			up(sp)
-#define valusema(sp)			(atomic_read(&(sp)->count))
-#define freesema(sema)
+#define freesema(sema)			do { } while (0)
+
+static inline int issemalocked(sema_t *sp)
+{
+	return down_trylock(sp) || (up(sp), 0);
+}
 
 /*
  * Map cpsema (try to get the sema) to down_trylock. We need to switch
  * the return values since cpsema returns 1 (acquired) 0 (failed) and
  * down_trylock returns the reverse 0 (acquired) 1 (failed).
  */
-
-#define cpsema(sp)			(down_trylock(sp) ? 0 : 1)
-
-/*
- * Didn't do cvsema(sp). Not sure how to map this to up/down/...
- * It does a vsema if the values is < 0 other wise nothing.
- */
+static inline int cpsema(sema_t *sp)
+{
+	return down_trylock(sp) ? 0 : 1;
+}
 
 #endif /* __XFS_SUPPORT_SEMA_H__ */

fs/xfs/linux-2.6/xfs_aops.c

@@ -21,7 +21,6 @@
 #include "xfs_inum.h"
 #include "xfs_sb.h"
 #include "xfs_ag.h"
-#include "xfs_dir.h"
 #include "xfs_dir2.h"
 #include "xfs_trans.h"
 #include "xfs_dmapi.h"
@@ -29,7 +28,6 @@
 #include "xfs_bmap_btree.h"
 #include "xfs_alloc_btree.h"
 #include "xfs_ialloc_btree.h"
-#include "xfs_dir_sf.h"
 #include "xfs_dir2_sf.h"
 #include "xfs_attr_sf.h"
 #include "xfs_dinode.h"
@@ -76,7 +74,7 @@ xfs_page_trace(
 	int		mask)
 {
 	xfs_inode_t	*ip;
-	vnode_t		*vp = vn_from_inode(inode);
+	bhv_vnode_t	*vp = vn_from_inode(inode);
 	loff_t		isize = i_size_read(inode);
 	loff_t		offset = page_offset(page);
 	int		delalloc = -1, unmapped = -1, unwritten = -1;
@@ -136,9 +134,10 @@ xfs_destroy_ioend(
 
 	for (bh = ioend->io_buffer_head; bh; bh = next) {
 		next = bh->b_private;
-		bh->b_end_io(bh, ioend->io_uptodate);
+		bh->b_end_io(bh, !ioend->io_error);
 	}
-
+	if (unlikely(ioend->io_error))
+		vn_ioerror(ioend->io_vnode, ioend->io_error, __FILE__,__LINE__);
 	vn_iowake(ioend->io_vnode);
 	mempool_free(ioend, xfs_ioend_pool);
 }
@@ -180,13 +179,12 @@ xfs_end_bio_unwritten(
 	void			*data)
 {
 	xfs_ioend_t		*ioend = data;
-	vnode_t			*vp = ioend->io_vnode;
+	bhv_vnode_t		*vp = ioend->io_vnode;
 	xfs_off_t		offset = ioend->io_offset;
 	size_t			size = ioend->io_size;
-	int			error;
 
-	if (ioend->io_uptodate)
-		VOP_BMAP(vp, offset, size, BMAPI_UNWRITTEN, NULL, NULL, error);
+	if (likely(!ioend->io_error))
+		bhv_vop_bmap(vp, offset, size, BMAPI_UNWRITTEN, NULL, NULL);
 	xfs_destroy_ioend(ioend);
 }
 
@@ -211,7 +209,7 @@ xfs_alloc_ioend(
 	 * all the I/O from calling the completion routine too early.
 	 */
 	atomic_set(&ioend->io_remaining, 1);
-	ioend->io_uptodate = 1; /* cleared if any I/O fails */
+	ioend->io_error = 0;
 	ioend->io_list = NULL;
 	ioend->io_type = type;
 	ioend->io_vnode = vn_from_inode(inode);
@@ -239,10 +237,10 @@ xfs_map_blocks(
 	xfs_iomap_t		*mapp,
 	int			flags)
 {
-	vnode_t			*vp = vn_from_inode(inode);
+	bhv_vnode_t		*vp = vn_from_inode(inode);
 	int			error, nmaps = 1;
 
-	VOP_BMAP(vp, offset, count, flags, mapp, &nmaps, error);
+	error = bhv_vop_bmap(vp, offset, count, flags, mapp, &nmaps);
 	if (!error && (flags & (BMAPI_WRITE|BMAPI_ALLOCATE)))
 		VMODIFY(vp);
 	return -error;
@@ -271,16 +269,14 @@ xfs_end_bio(
 	if (bio->bi_size)
 		return 1;
 
-	ASSERT(ioend);
 	ASSERT(atomic_read(&bio->bi_cnt) >= 1);
+	ioend->io_error = test_bit(BIO_UPTODATE, &bio->bi_flags) ? 0 : error;
 
 	/* Toss bio and pass work off to an xfsdatad thread */
-	if (!test_bit(BIO_UPTODATE, &bio->bi_flags))
-		ioend->io_uptodate = 0;
 	bio->bi_private = NULL;
 	bio->bi_end_io = NULL;
-
 	bio_put(bio);
+
 	xfs_finish_ioend(ioend);
 	return 0;
 }
@@ -1127,7 +1123,7 @@ xfs_vm_writepage(
 	 * then mark the page dirty again and leave the page
 	 * as is.
 	 */
-	if (PFLAGS_TEST_FSTRANS() && need_trans)
+	if (current_test_flags(PF_FSTRANS) && need_trans)
 		goto out_fail;
 
 	/*
@@ -1158,6 +1154,18 @@ xfs_vm_writepage(
 	return error;
 }
 
+STATIC int
+xfs_vm_writepages(
+	struct address_space	*mapping,
+	struct writeback_control *wbc)
+{
+	struct bhv_vnode	*vp = vn_from_inode(mapping->host);
+
+	if (VN_TRUNC(vp))
+		VUNTRUNCATE(vp);
+	return generic_writepages(mapping, wbc);
+}
+
 /*
  * Called to move a page into cleanable state - and from there
  * to be released. Possibly the page is already clean. We always
@@ -1204,7 +1212,7 @@ xfs_vm_releasepage(
 	/* If we are already inside a transaction or the thread cannot
 	 * do I/O, we cannot release this page.
 	 */
-	if (PFLAGS_TEST_FSTRANS())
+	if (current_test_flags(PF_FSTRANS))
 		return 0;
 
 	/*
@@ -1231,7 +1239,7 @@ __xfs_get_blocks(
 	int			direct,
 	bmapi_flags_t		flags)
 {
-	vnode_t			*vp = vn_from_inode(inode);
+	bhv_vnode_t		*vp = vn_from_inode(inode);
 	xfs_iomap_t		iomap;
 	xfs_off_t		offset;
 	ssize_t			size;
@@ -1241,8 +1249,8 @@ __xfs_get_blocks(
 	offset = (xfs_off_t)iblock << inode->i_blkbits;
 	ASSERT(bh_result->b_size >= (1 << inode->i_blkbits));
 	size = bh_result->b_size;
-	VOP_BMAP(vp, offset, size,
-		create ? flags : BMAPI_READ, &iomap, &niomap, error);
+	error = bhv_vop_bmap(vp, offset, size,
+			     create ? flags : BMAPI_READ, &iomap, &niomap);
 	if (error)
 		return -error;
 	if (niomap == 0)
@@ -1370,13 +1378,13 @@ xfs_vm_direct_IO(
 {
 	struct file	*file = iocb->ki_filp;
 	struct inode	*inode = file->f_mapping->host;
-	vnode_t		*vp = vn_from_inode(inode);
+	bhv_vnode_t	*vp = vn_from_inode(inode);
 	xfs_iomap_t	iomap;
 	int		maps = 1;
 	int		error;
 	ssize_t		ret;
 
-	VOP_BMAP(vp, offset, 0, BMAPI_DEVICE, &iomap, &maps, error);
+	error = bhv_vop_bmap(vp, offset, 0, BMAPI_DEVICE, &iomap, &maps);
 	if (error)
 		return -error;
 
@@ -1409,14 +1417,12 @@ xfs_vm_bmap(
 	sector_t		block)
 {
 	struct inode		*inode = (struct inode *)mapping->host;
-	vnode_t			*vp = vn_from_inode(inode);
-	int			error;
+	bhv_vnode_t		*vp = vn_from_inode(inode);
 
 	vn_trace_entry(vp, __FUNCTION__, (inst_t *)__return_address);
-
-	VOP_RWLOCK(vp, VRWLOCK_READ);
-	VOP_FLUSH_PAGES(vp, (xfs_off_t)0, -1, 0, FI_REMAPF, error);
-	VOP_RWUNLOCK(vp, VRWLOCK_READ);
+	bhv_vop_rwlock(vp, VRWLOCK_READ);
+	bhv_vop_flush_pages(vp, (xfs_off_t)0, -1, 0, FI_REMAPF);
+	bhv_vop_rwunlock(vp, VRWLOCK_READ);
 	return generic_block_bmap(mapping, block, xfs_get_blocks);
 }
 
@@ -1452,6 +1458,7 @@ struct address_space_operations xfs_address_space_operations = {
 	.readpage		= xfs_vm_readpage,
 	.readpages		= xfs_vm_readpages,
 	.writepage		= xfs_vm_writepage,
+	.writepages		= xfs_vm_writepages,
 	.sync_page		= block_sync_page,
 	.releasepage		= xfs_vm_releasepage,
 	.invalidatepage		= xfs_vm_invalidatepage,

fs/xfs/linux-2.6/xfs_aops.h

 /*
- * Copyright (c) 2005 Silicon Graphics, Inc.
+ * Copyright (c) 2005-2006 Silicon Graphics, Inc.
  * All Rights Reserved.
  *
  * This program is free software; you can redistribute it and/or
@@ -30,9 +30,9 @@ typedef void (*xfs_ioend_func_t)(void *);
 typedef struct xfs_ioend {
 	struct xfs_ioend	*io_list;	/* next ioend in chain */
 	unsigned int		io_type;	/* delalloc / unwritten */
-	unsigned int		io_uptodate;	/* I/O status register */
+	int			io_error;	/* I/O error code */
 	atomic_t		io_remaining;	/* hold count */
-	struct vnode		*io_vnode;	/* file being written to */
+	struct bhv_vnode	*io_vnode;	/* file being written to */
 	struct buffer_head	*io_buffer_head;/* buffer linked list head */
 	struct buffer_head	*io_buffer_tail;/* buffer linked list tail */
 	size_t			io_size;	/* size of the extent */
@@ -43,4 +43,4 @@ typedef struct xfs_ioend {
 extern struct address_space_operations xfs_address_space_operations;
 extern int xfs_get_blocks(struct inode *, sector_t, struct buffer_head *, int);
 
-#endif /* __XFS_IOPS_H__ */
+#endif /* __XFS_AOPS_H__ */

fs/xfs/linux-2.6/xfs_export.c

@@ -21,7 +21,6 @@
 #include "xfs_log.h"
 #include "xfs_trans.h"
 #include "xfs_sb.h"
-#include "xfs_dir.h"
 #include "xfs_mount.h"
 #include "xfs_export.h"
 
@@ -97,7 +96,7 @@ xfs_fs_encode_fh(
 	int			len;
 	int			is64 = 0;
 #if XFS_BIG_INUMS
-	vfs_t			*vfs = vfs_from_sb(inode->i_sb);
+	bhv_vfs_t		*vfs = vfs_from_sb(inode->i_sb);
 
 	if (!(vfs->vfs_flag & VFS_32BITINODES)) {
 		/* filesystem may contain 64bit inode numbers */
@@ -136,13 +135,13 @@ xfs_fs_get_dentry(
 	struct super_block	*sb,
 	void			*data)
 {
-	vnode_t			*vp;
+	bhv_vnode_t		*vp;
 	struct inode		*inode;
 	struct dentry		*result;
-	vfs_t			*vfsp = vfs_from_sb(sb);
+	bhv_vfs_t		*vfsp = vfs_from_sb(sb);
 	int			error;
 
-	VFS_VGET(vfsp, &vp, (fid_t *)data, error);
+	error = bhv_vfs_vget(vfsp, &vp, (fid_t *)data);
 	if (error || vp == NULL)
 		return ERR_PTR(-ESTALE) ;
 
@@ -160,12 +159,12 @@ xfs_fs_get_parent(
 	struct dentry		*child)
 {
 	int			error;
-	vnode_t			*vp, *cvp;
+	bhv_vnode_t		*vp, *cvp;
 	struct dentry		*parent;
 
 	cvp = NULL;
 	vp = vn_from_inode(child->d_inode);
-	VOP_LOOKUP(vp, &dotdot, &cvp, 0, NULL, NULL, error);
+	error = bhv_vop_lookup(vp, &dotdot, &cvp, 0, NULL, NULL);
 	if (unlikely(error))
 		return ERR_PTR(-error);
 

fs/xfs/linux-2.6/xfs_fs_subr.c

 /*
- * Copyright (c) 2000-2002,2005 Silicon Graphics, Inc.
+ * Copyright (c) 2000-2002,2005-2006 Silicon Graphics, Inc.
  * All Rights Reserved.
  *
  * This program is free software; you can redistribute it and/or
@@ -15,40 +15,12 @@
  * along with this program; if not, write the Free Software Foundation,
  * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
  */
-
 #include "xfs.h"
 
-/*
- * Stub for no-op vnode operations that return error status.
- */
-int
-fs_noerr(void)
-{
-	return 0;
-}
+int  fs_noerr(void) { return 0; }
+int  fs_nosys(void) { return ENOSYS; }
+void fs_noval(void) { return; }
 
-/*
- * Operation unsupported under this file system.
- */
-int
-fs_nosys(void)
-{
-	return ENOSYS;
-}
-
-/*
- * Stub for inactive, strategy, and read/write lock/unlock.  Does nothing.
- */
-/* ARGSUSED */
-void
-fs_noval(void)
-{
-}
-
-/*
- * vnode pcache layer for vnode_tosspages.
- * 'last' parameter unused but left in for IRIX compatibility
- */
 void
 fs_tosspages(
 	bhv_desc_t	*bdp,
@@ -56,18 +28,13 @@ fs_tosspages(
 	xfs_off_t	last,
 	int		fiopt)
 {
-	vnode_t		*vp = BHV_TO_VNODE(bdp);
+	bhv_vnode_t	*vp = BHV_TO_VNODE(bdp);
 	struct inode	*ip = vn_to_inode(vp);
 
 	if (VN_CACHED(vp))
 		truncate_inode_pages(ip->i_mapping, first);
 }
 
-
-/*
- * vnode pcache layer for vnode_flushinval_pages.
- * 'last' parameter unused but left in for IRIX compatibility
- */
 void
 fs_flushinval_pages(
 	bhv_desc_t	*bdp,
@@ -75,20 +42,17 @@ fs_flushinval_pages(
 	xfs_off_t	last,
 	int		fiopt)
 {
-	vnode_t		*vp = BHV_TO_VNODE(bdp);
+	bhv_vnode_t	*vp = BHV_TO_VNODE(bdp);
 	struct inode	*ip = vn_to_inode(vp);
 
 	if (VN_CACHED(vp)) {
+		if (VN_TRUNC(vp))
+			VUNTRUNCATE(vp);
 		filemap_write_and_wait(ip->i_mapping);
-
 		truncate_inode_pages(ip->i_mapping, first);
 	}
 }
 
-/*
- * vnode pcache layer for vnode_flush_pages.
- * 'last' parameter unused but left in for IRIX compatibility
- */
 int
 fs_flush_pages(
 	bhv_desc_t	*bdp,
@@ -97,15 +61,16 @@ fs_flush_pages(
 	uint64_t	flags,
 	int		fiopt)
 {
-	vnode_t		*vp = BHV_TO_VNODE(bdp);
+	bhv_vnode_t	*vp = BHV_TO_VNODE(bdp);
 	struct inode	*ip = vn_to_inode(vp);
 
-	if (VN_CACHED(vp)) {
+	if (VN_DIRTY(vp)) {
+		if (VN_TRUNC(vp))
+			VUNTRUNCATE(vp);
 		filemap_fdatawrite(ip->i_mapping);
 		if (flags & XFS_B_ASYNC)
 			return 0;
 		filemap_fdatawait(ip->i_mapping);
 	}
-
 	return 0;
 }

fs/xfs/linux-2.6/xfs_globals.c

@@ -45,6 +45,7 @@ xfs_param_t xfs_params = {
 	.xfs_buf_age	= {	1*100,		15*100,		7200*100},
 	.inherit_nosym	= {	0,		0,		1	},
 	.rotorstep	= {	1,		1,		255	},
+	.inherit_nodfrg	= {	0,		1,		1	},
 };
 
 /*

fs/xfs/linux-2.6/xfs_ioctl.c

@@ -23,7 +23,6 @@
 #include "xfs_trans.h"
 #include "xfs_sb.h"
 #include "xfs_ag.h"
-#include "xfs_dir.h"
 #include "xfs_dir2.h"
 #include "xfs_alloc.h"
 #include "xfs_dmapi.h"
@@ -31,7 +30,6 @@
 #include "xfs_bmap_btree.h"
 #include "xfs_alloc_btree.h"
 #include "xfs_ialloc_btree.h"
-#include "xfs_dir_sf.h"
 #include "xfs_attr_sf.h"
 #include "xfs_dir2_sf.h"
 #include "xfs_dinode.h"
@@ -78,7 +76,7 @@ xfs_find_handle(
 	xfs_handle_t		handle;
 	xfs_fsop_handlereq_t	hreq;
 	struct inode		*inode;
-	struct vnode		*vp;
+	bhv_vnode_t		*vp;
 
 	if (copy_from_user(&hreq, arg, sizeof(hreq)))
 		return -XFS_ERROR(EFAULT);
@@ -192,7 +190,7 @@ xfs_vget_fsop_handlereq(
 	xfs_mount_t		*mp,
 	struct inode		*parinode,	/* parent inode pointer    */
 	xfs_fsop_handlereq_t	*hreq,
-	vnode_t			**vp,
+	bhv_vnode_t		**vp,
 	struct inode		**inode)
 {
 	void			__user *hanp;
@@ -202,7 +200,7 @@ xfs_vget_fsop_handlereq(
 	xfs_handle_t		handle;
 	xfs_inode_t		*ip;
 	struct inode		*inodep;
-	vnode_t			*vpp;
+	bhv_vnode_t		*vpp;
 	xfs_ino_t		ino;
 	__u32			igen;
 	int			error;
@@ -277,7 +275,7 @@ xfs_open_by_handle(
 	struct file		*filp;
 	struct inode		*inode;
 	struct dentry		*dentry;
-	vnode_t			*vp;
+	bhv_vnode_t		*vp;
 	xfs_fsop_handlereq_t	hreq;
 
 	if (!capable(CAP_SYS_ADMIN))
@@ -362,7 +360,7 @@ xfs_readlink_by_handle(
 	struct uio		auio;
 	struct inode		*inode;
 	xfs_fsop_handlereq_t	hreq;
-	vnode_t			*vp;
+	bhv_vnode_t		*vp;
 	__u32			olen;
 
 	if (!capable(CAP_SYS_ADMIN))
@@ -393,9 +391,11 @@ xfs_readlink_by_handle(
 	auio.uio_segflg	= UIO_USERSPACE;
 	auio.uio_resid	= olen;
 
-	VOP_READLINK(vp, &auio, IO_INVIS, NULL, error);
-
+	error = bhv_vop_readlink(vp, &auio, IO_INVIS, NULL);
 	VN_RELE(vp);
+	if (error)
+		return -error;
+
 	return (olen - auio.uio_resid);
 }
 
@@ -411,7 +411,7 @@ xfs_fssetdm_by_handle(
 	xfs_fsop_setdm_handlereq_t dmhreq;
 	struct inode		*inode;
 	bhv_desc_t		*bdp;
-	vnode_t			*vp;
+	bhv_vnode_t		*vp;
 
 	if (!capable(CAP_MKNOD))
 		return -XFS_ERROR(EPERM);
@@ -452,7 +452,7 @@ xfs_attrlist_by_handle(
 	attrlist_cursor_kern_t	*cursor;
 	xfs_fsop_attrlist_handlereq_t al_hreq;
 	struct inode		*inode;
-	vnode_t			*vp;
+	bhv_vnode_t		*vp;
 	char			*kbuf;
 
 	if (!capable(CAP_SYS_ADMIN))
@@ -472,8 +472,8 @@ xfs_attrlist_by_handle(
 		goto out_vn_rele;
 
 	cursor = (attrlist_cursor_kern_t *)&al_hreq.pos;
-	VOP_ATTR_LIST(vp, kbuf, al_hreq.buflen, al_hreq.flags,
-			cursor, NULL, error);
+	error = bhv_vop_attr_list(vp, kbuf, al_hreq.buflen, al_hreq.flags,
+					cursor, NULL);
 	if (error)
 		goto out_kfree;
 
@@ -490,7 +490,7 @@ xfs_attrlist_by_handle(
 
 STATIC int
 xfs_attrmulti_attr_get(
-	struct vnode		*vp,
+	bhv_vnode_t		*vp,
 	char			*name,
 	char			__user *ubuf,
 	__uint32_t		*len,
@@ -505,7 +505,7 @@ xfs_attrmulti_attr_get(
 	if (!kbuf)
 		return ENOMEM;
 
-	VOP_ATTR_GET(vp, name, kbuf, len, flags, NULL, error);
+	error = bhv_vop_attr_get(vp, name, kbuf, len, flags, NULL);
 	if (error)
 		goto out_kfree;
 
@@ -519,7 +519,7 @@ xfs_attrmulti_attr_get(
 
 STATIC int
 xfs_attrmulti_attr_set(
-	struct vnode		*vp,
+	bhv_vnode_t		*vp,
 	char			*name,
 	const char		__user *ubuf,
 	__uint32_t		len,
@@ -542,7 +542,7 @@ xfs_attrmulti_attr_set(
 	if (copy_from_user(kbuf, ubuf, len))
 		goto out_kfree;
 			
-	VOP_ATTR_SET(vp, name, kbuf, len, flags, NULL, error);
+	error = bhv_vop_attr_set(vp, name, kbuf, len, flags, NULL);
 
  out_kfree:
 	kfree(kbuf);
@@ -551,20 +551,15 @@ xfs_attrmulti_attr_set(
 
 STATIC int
 xfs_attrmulti_attr_remove(
-	struct vnode		*vp,
+	bhv_vnode_t		*vp,
 	char			*name,
 	__uint32_t		flags)
 {
-	int			error;
-
-
 	if (IS_RDONLY(&vp->v_inode))
 		return -EROFS;
 	if (IS_IMMUTABLE(&vp->v_inode) || IS_APPEND(&vp->v_inode))
 		return EPERM;
-
-	VOP_ATTR_REMOVE(vp, name, flags, NULL, error);
-	return error;
+	return bhv_vop_attr_remove(vp, name, flags, NULL);
 }
 
 STATIC int
@@ -578,7 +573,7 @@ xfs_attrmulti_by_handle(
 	xfs_attr_multiop_t	*ops;
 	xfs_fsop_attrmulti_handlereq_t am_hreq;
 	struct inode		*inode;
-	vnode_t			*vp;
+	bhv_vnode_t		*vp;
 	unsigned int		i, size;
 	char			*attr_name;
 
@@ -658,7 +653,7 @@ xfs_attrmulti_by_handle(
 STATIC int
 xfs_ioc_space(
 	bhv_desc_t		*bdp,
-	vnode_t			*vp,
+	bhv_vnode_t		*vp,
 	struct file		*filp,
 	int			flags,
 	unsigned int		cmd,
@@ -682,7 +677,7 @@ xfs_ioc_fsgeometry(
 
 STATIC int
 xfs_ioc_xattr(
-	vnode_t			*vp,
+	bhv_vnode_t		*vp,
 	xfs_inode_t		*ip,
 	struct file		*filp,
 	unsigned int		cmd,
@@ -711,7 +706,7 @@ xfs_ioctl(
 	void			__user *arg)
 {
 	int			error;
-	vnode_t			*vp;
+	bhv_vnode_t		*vp;
 	xfs_inode_t		*ip;
 	xfs_mount_t		*mp;
 
@@ -962,7 +957,7 @@ xfs_ioctl(
 STATIC int
 xfs_ioc_space(
 	bhv_desc_t		*bdp,
-	vnode_t			*vp,
+	bhv_vnode_t		*vp,
 	struct file		*filp,
 	int			ioflags,
 	unsigned int		cmd,
@@ -1153,14 +1148,14 @@ xfs_di2lxflags(
 
 STATIC int
 xfs_ioc_xattr(
-	vnode_t			*vp,
+	bhv_vnode_t		*vp,
 	xfs_inode_t		*ip,
 	struct file		*filp,
 	unsigned int		cmd,
 	void			__user *arg)
 {
 	struct fsxattr		fa;
-	struct vattr		*vattr;
+	struct bhv_vattr	*vattr;
 	int			error = 0;
 	int			attr_flags;
 	unsigned int		flags;
@@ -1173,7 +1168,7 @@ xfs_ioc_xattr(
 	case XFS_IOC_FSGETXATTR: {
 		vattr->va_mask = XFS_AT_XFLAGS | XFS_AT_EXTSIZE | \
 				 XFS_AT_NEXTENTS | XFS_AT_PROJID;
-		VOP_GETATTR(vp, vattr, 0, NULL, error);
+		error = bhv_vop_getattr(vp, vattr, 0, NULL);
 		if (unlikely(error)) {
 			error = -error;
 			break;
@@ -1206,7 +1201,7 @@ xfs_ioc_xattr(
 		vattr->va_extsize = fa.fsx_extsize;
 		vattr->va_projid  = fa.fsx_projid;
 
-		VOP_SETATTR(vp, vattr, attr_flags, NULL, error);
+		error = bhv_vop_setattr(vp, vattr, attr_flags, NULL);
 		if (likely(!error))
 			__vn_revalidate(vp, vattr);	/* update flags */
 		error = -error;
@@ -1216,7 +1211,7 @@ xfs_ioc_xattr(
 	case XFS_IOC_FSGETXATTRA: {
 		vattr->va_mask = XFS_AT_XFLAGS | XFS_AT_EXTSIZE | \
 				 XFS_AT_ANEXTENTS | XFS_AT_PROJID;
-		VOP_GETATTR(vp, vattr, 0, NULL, error);
+		error = bhv_vop_getattr(vp, vattr, 0, NULL);
 		if (unlikely(error)) {
 			error = -error;
 			break;
@@ -1262,7 +1257,7 @@ xfs_ioc_xattr(
 		vattr->va_xflags = xfs_merge_ioc_xflags(flags,
 							xfs_ip2xflags(ip));
 
-		VOP_SETATTR(vp, vattr, attr_flags, NULL, error);
+		error = bhv_vop_setattr(vp, vattr, attr_flags, NULL);
 		if (likely(!error))
 			__vn_revalidate(vp, vattr);	/* update flags */
 		error = -error;

fs/xfs/linux-2.6/xfs_ioctl32.c

@@ -114,7 +114,7 @@ xfs_compat_ioctl(
 	unsigned long	arg)
 {
 	struct inode	*inode = file->f_dentry->d_inode;
-	vnode_t		*vp = vn_from_inode(inode);
+	bhv_vnode_t	*vp = vn_from_inode(inode);
 	int		error;
 
 	switch (cmd) {
@@ -193,7 +193,7 @@ xfs_compat_ioctl(
 		return -ENOIOCTLCMD;
 	}
 
-	VOP_IOCTL(vp, inode, file, mode, cmd, (void __user *)arg, error);
+	error = bhv_vop_ioctl(vp, inode, file, mode, cmd, (void __user *)arg);
 	VMODIFY(vp);
 
 	return error;

fs/xfs/linux-2.6/xfs_iops.c

@@ -23,7 +23,6 @@
 #include "xfs_trans.h"
 #include "xfs_sb.h"
 #include "xfs_ag.h"
-#include "xfs_dir.h"
 #include "xfs_dir2.h"
 #include "xfs_alloc.h"
 #include "xfs_dmapi.h"
@@ -32,7 +31,6 @@
 #include "xfs_bmap_btree.h"
 #include "xfs_alloc_btree.h"
 #include "xfs_ialloc_btree.h"
-#include "xfs_dir_sf.h"
 #include "xfs_dir2_sf.h"
 #include "xfs_attr_sf.h"
 #include "xfs_dinode.h"
@@ -61,7 +59,7 @@
  */
 xfs_inode_t *
 xfs_vtoi(
-	struct vnode	*vp)
+	bhv_vnode_t	*vp)
 {
 	bhv_desc_t      *bdp;
 
@@ -80,7 +78,7 @@ void
 xfs_synchronize_atime(
 	xfs_inode_t	*ip)
 {
-	vnode_t		*vp;
+	bhv_vnode_t	*vp;
 
 	vp = XFS_ITOV_NULL(ip);
 	if (vp) {
@@ -200,14 +198,10 @@ xfs_ichgtime_fast(
 STATIC void
 xfs_validate_fields(
 	struct inode	*ip,
-	struct vattr	*vattr)
+	bhv_vattr_t	*vattr)
 {
-	vnode_t		*vp = vn_from_inode(ip);
-	int		error;
-
 	vattr->va_mask = XFS_AT_NLINK|XFS_AT_SIZE|XFS_AT_NBLOCKS;
-	VOP_GETATTR(vp, vattr, ATTR_LAZY, NULL, error);
-  	if (likely(!error)) {
+	if (!bhv_vop_getattr(vn_from_inode(ip), vattr, ATTR_LAZY, NULL)) {
 		ip->i_nlink = vattr->va_nlink;
 		ip->i_blocks = vattr->va_nblocks;
 
@@ -225,7 +219,7 @@ xfs_validate_fields(
  */
 STATIC int
 xfs_init_security(
-	struct vnode	*vp,
+	bhv_vnode_t	*vp,
 	struct inode	*dir)
 {
 	struct inode	*ip = vn_to_inode(vp);
@@ -241,7 +235,7 @@ xfs_init_security(
 		return -error;
 	}
 
-	VOP_ATTR_SET(vp, name, value, length, ATTR_SECURE, NULL, error);
+	error = bhv_vop_attr_set(vp, name, value, length, ATTR_SECURE, NULL);
 	if (!error)
 		VMODIFY(vp);
 
@@ -264,13 +258,12 @@ xfs_has_fs_struct(struct task_struct *task)
 
 STATIC inline void
 xfs_cleanup_inode(
-	vnode_t		*dvp,
-	vnode_t		*vp,
+	bhv_vnode_t	*dvp,
+	bhv_vnode_t	*vp,
 	struct dentry	*dentry,
 	int		mode)
 {
 	struct dentry   teardown = {};
-	int             error;
 
 	/* Oh, the horror.
 	 * If we can't add the ACL or we fail in
@@ -281,9 +274,9 @@ xfs_cleanup_inode(
 	teardown.d_name = dentry->d_name;
 
 	if (S_ISDIR(mode))
-	  	VOP_RMDIR(dvp, &teardown, NULL, error);
+	  	bhv_vop_rmdir(dvp, &teardown, NULL);
 	else
-		VOP_REMOVE(dvp, &teardown, NULL, error);
+		bhv_vop_remove(dvp, &teardown, NULL);
 	VN_RELE(vp);
 }
 
@@ -295,8 +288,8 @@ xfs_vn_mknod(
 	dev_t		rdev)
 {
 	struct inode	*ip;
-	vattr_t		vattr = { 0 };
-	vnode_t		*vp = NULL, *dvp = vn_from_inode(dir);
+	bhv_vattr_t	vattr = { 0 };
+	bhv_vnode_t	*vp = NULL, *dvp = vn_from_inode(dir);
 	xfs_acl_t	*default_acl = NULL;
 	attrexists_t	test_default_acl = _ACL_DEFAULT_EXISTS;
 	int		error;
@@ -330,10 +323,10 @@ xfs_vn_mknod(
 		vattr.va_mask |= XFS_AT_RDEV;
 		/*FALLTHROUGH*/
 	case S_IFREG:
-		VOP_CREATE(dvp, dentry, &vattr, &vp, NULL, error);
+		error = bhv_vop_create(dvp, dentry, &vattr, &vp, NULL);
 		break;
 	case S_IFDIR:
-		VOP_MKDIR(dvp, dentry, &vattr, &vp, NULL, error);
+		error = bhv_vop_mkdir(dvp, dentry, &vattr, &vp, NULL);
 		break;
 	default:
 		error = EINVAL;
@@ -396,14 +389,14 @@ xfs_vn_lookup(
 	struct dentry	*dentry,
 	struct nameidata *nd)
 {
-	struct vnode	*vp = vn_from_inode(dir), *cvp;
+	bhv_vnode_t	*vp = vn_from_inode(dir), *cvp;
 	int		error;
 
 	if (dentry->d_name.len >= MAXNAMELEN)
 		return ERR_PTR(-ENAMETOOLONG);
 
-	VOP_LOOKUP(vp, dentry, &cvp, 0, NULL, NULL, error);
-	if (error) {
+	error = bhv_vop_lookup(vp, dentry, &cvp, 0, NULL, NULL);
+	if (unlikely(error)) {
 		if (unlikely(error != ENOENT))
 			return ERR_PTR(-error);
 		d_add(dentry, NULL);
@@ -420,9 +413,9 @@ xfs_vn_link(
 	struct dentry	*dentry)
 {
 	struct inode	*ip;	/* inode of guy being linked to */
-	vnode_t		*tdvp;	/* target directory for new name/link */
-	vnode_t		*vp;	/* vp of name being linked */
-	vattr_t		vattr;
+	bhv_vnode_t	*tdvp;	/* target directory for new name/link */
+	bhv_vnode_t	*vp;	/* vp of name being linked */
+	bhv_vattr_t	vattr;
 	int		error;
 
 	ip = old_dentry->d_inode;	/* inode being linked to */
@@ -432,7 +425,7 @@ xfs_vn_link(
 	tdvp = vn_from_inode(dir);
 	vp = vn_from_inode(ip);
 
-	VOP_LINK(tdvp, vp, dentry, NULL, error);
+	error = bhv_vop_link(tdvp, vp, dentry, NULL);
 	if (likely(!error)) {
 		VMODIFY(tdvp);
 		VN_HOLD(vp);
@@ -448,14 +441,14 @@ xfs_vn_unlink(
 	struct dentry	*dentry)
 {
 	struct inode	*inode;
-	vnode_t		*dvp;	/* directory containing name to remove */
-	vattr_t		vattr;
+	bhv_vnode_t	*dvp;	/* directory containing name to remove */
+	bhv_vattr_t	vattr;
 	int		error;
 
 	inode = dentry->d_inode;
 	dvp = vn_from_inode(dir);
 
-	VOP_REMOVE(dvp, dentry, NULL, error);
+	error = bhv_vop_remove(dvp, dentry, NULL);
 	if (likely(!error)) {
 		xfs_validate_fields(dir, &vattr);	/* size needs update */
 		xfs_validate_fields(inode, &vattr);
@@ -470,27 +463,26 @@ xfs_vn_symlink(
 	const char	*symname)
 {
 	struct inode	*ip;
-	vattr_t		vattr = { 0 };
-	vnode_t		*dvp;	/* directory containing name of symlink */
-	vnode_t		*cvp;	/* used to lookup symlink to put in dentry */
+	bhv_vattr_t	va = { 0 };
+	bhv_vnode_t	*dvp;	/* directory containing name of symlink */
+	bhv_vnode_t	*cvp;	/* used to lookup symlink to put in dentry */
 	int		error;
 
 	dvp = vn_from_inode(dir);
 	cvp = NULL;
 
-	vattr.va_mode = S_IFLNK |
+	va.va_mode = S_IFLNK |
 		(irix_symlink_mode ? 0777 & ~current->fs->umask : S_IRWXUGO);
-	vattr.va_mask = XFS_AT_TYPE|XFS_AT_MODE;
+	va.va_mask = XFS_AT_TYPE|XFS_AT_MODE;
 
-	error = 0;
-	VOP_SYMLINK(dvp, dentry, &vattr, (char *)symname, &cvp, NULL, error);
+	error = bhv_vop_symlink(dvp, dentry, &va, (char *)symname, &cvp, NULL);
 	if (likely(!error && cvp)) {
 		error = xfs_init_security(cvp, dir);
 		if (likely(!error)) {
 			ip = vn_to_inode(cvp);
 			d_instantiate(dentry, ip);
-			xfs_validate_fields(dir, &vattr);
-			xfs_validate_fields(ip, &vattr);
+			xfs_validate_fields(dir, &va);
+			xfs_validate_fields(ip, &va);
 		} else {
 			xfs_cleanup_inode(dvp, cvp, dentry, 0);
 		}
@@ -504,11 +496,11 @@ xfs_vn_rmdir(
 	struct dentry	*dentry)
 {
 	struct inode	*inode = dentry->d_inode;
-	vnode_t		*dvp = vn_from_inode(dir);
-	vattr_t		vattr;
+	bhv_vnode_t	*dvp = vn_from_inode(dir);
+	bhv_vattr_t	vattr;
 	int		error;
 
-	VOP_RMDIR(dvp, dentry, NULL, error);
+	error = bhv_vop_rmdir(dvp, dentry, NULL);
 	if (likely(!error)) {
 		xfs_validate_fields(inode, &vattr);
 		xfs_validate_fields(dir, &vattr);
@@ -524,15 +516,15 @@ xfs_vn_rename(
 	struct dentry	*ndentry)
 {
 	struct inode	*new_inode = ndentry->d_inode;
-	vnode_t		*fvp;	/* from directory */
-	vnode_t		*tvp;	/* target directory */
-	vattr_t		vattr;
+	bhv_vnode_t	*fvp;	/* from directory */
+	bhv_vnode_t	*tvp;	/* target directory */
+	bhv_vattr_t	vattr;
 	int		error;
 
 	fvp = vn_from_inode(odir);
 	tvp = vn_from_inode(ndir);
 
-	VOP_RENAME(fvp, odentry, tvp, ndentry, NULL, error);
+	error = bhv_vop_rename(fvp, odentry, tvp, ndentry, NULL);
 	if (likely(!error)) {
 		if (new_inode)
 			xfs_validate_fields(new_inode, &vattr);
@@ -553,7 +545,7 @@ xfs_vn_follow_link(
 	struct dentry		*dentry,
 	struct nameidata	*nd)
 {
-	vnode_t			*vp;
+	bhv_vnode_t		*vp;
 	uio_t			*uio;
 	iovec_t			iov;
 	int			error;
@@ -586,8 +578,8 @@ xfs_vn_follow_link(
 	uio->uio_resid = MAXPATHLEN;
 	uio->uio_iovcnt = 1;
 
-	VOP_READLINK(vp, uio, 0, NULL, error);
-	if (error) {
+	error = bhv_vop_readlink(vp, uio, 0, NULL);
+	if (unlikely(error)) {
 		kfree(link);
 		link = ERR_PTR(-error);
 	} else {
@@ -618,12 +610,7 @@ xfs_vn_permission(
 	int		mode,
 	struct nameidata *nd)
 {
-	vnode_t		*vp = vn_from_inode(inode);
-	int		error;
-
-	mode <<= 6;		/* convert from linux to vnode access bits */
-	VOP_ACCESS(vp, mode, NULL, error);
-	return -error;
+	return -bhv_vop_access(vn_from_inode(inode), mode << 6, NULL);
 }
 #else
 #define xfs_vn_permission NULL
@@ -636,14 +623,14 @@ xfs_vn_getattr(
 	struct kstat	*stat)
 {
 	struct inode	*inode = dentry->d_inode;
-	vnode_t		*vp = vn_from_inode(inode);
+	bhv_vnode_t	*vp = vn_from_inode(inode);
 	int		error = 0;
 
 	if (unlikely(vp->v_flag & VMODIFIED))
 		error = vn_revalidate(vp);
 	if (!error)
 		generic_fillattr(inode, stat);
-	return 0;
+	return -error;
 }
 
 STATIC int
@@ -653,8 +640,8 @@ xfs_vn_setattr(
 {
 	struct inode	*inode = dentry->d_inode;
 	unsigned int	ia_valid = attr->ia_valid;
-	vnode_t		*vp = vn_from_inode(inode);
-	vattr_t		vattr = { 0 };
+	bhv_vnode_t	*vp = vn_from_inode(inode);
+	bhv_vattr_t	vattr = { 0 };
 	int		flags = 0;
 	int		error;
 
@@ -697,7 +684,7 @@ xfs_vn_setattr(
 		flags |= ATTR_NONBLOCK;
 #endif
 
-	VOP_SETATTR(vp, &vattr, flags, NULL, error);
+	error = bhv_vop_setattr(vp, &vattr, flags, NULL);
 	if (likely(!error))
 		__vn_revalidate(vp, &vattr);
 	return -error;
@@ -718,7 +705,7 @@ xfs_vn_setxattr(
 	size_t		size,
 	int		flags)
 {
-	vnode_t		*vp = vn_from_inode(dentry->d_inode);
+	bhv_vnode_t	*vp = vn_from_inode(dentry->d_inode);
 	char		*attr = (char *)name;
 	attrnames_t	*namesp;
 	int		xflags = 0;
@@ -748,7 +735,7 @@ xfs_vn_getxattr(
 	void		*data,
 	size_t		size)
 {
-	vnode_t		*vp = vn_from_inode(dentry->d_inode);
+	bhv_vnode_t	*vp = vn_from_inode(dentry->d_inode);
 	char		*attr = (char *)name;
 	attrnames_t	*namesp;
 	int		xflags = 0;
@@ -777,7 +764,7 @@ xfs_vn_listxattr(
 	char			*data,
 	size_t			size)
 {
-	vnode_t			*vp = vn_from_inode(dentry->d_inode);
+	bhv_vnode_t		*vp = vn_from_inode(dentry->d_inode);
 	int			error, xflags = ATTR_KERNAMELS;
 	ssize_t			result;
 
@@ -796,7 +783,7 @@ xfs_vn_removexattr(
 	struct dentry	*dentry,
 	const char	*name)
 {
-	vnode_t		*vp = vn_from_inode(dentry->d_inode);
+	bhv_vnode_t	*vp = vn_from_inode(dentry->d_inode);
 	char		*attr = (char *)name;
 	attrnames_t	*namesp;
 	int		xflags = 0;

fs/xfs/linux-2.6/xfs_linux.h

@@ -134,14 +134,21 @@ BUFFER_FNS(PrivateStart, unwritten);
 #define xfs_buf_age_centisecs	xfs_params.xfs_buf_age.val
 #define xfs_inherit_nosymlinks	xfs_params.inherit_nosym.val
 #define xfs_rotorstep		xfs_params.rotorstep.val
+#define xfs_inherit_nodefrag	xfs_params.inherit_nodfrg.val
 
-#ifndef raw_smp_processor_id
-#define raw_smp_processor_id()	smp_processor_id()
-#endif
-#define current_cpu()		raw_smp_processor_id()
+#define current_cpu()		(raw_smp_processor_id())
 #define current_pid()		(current->pid)
 #define current_fsuid(cred)	(current->fsuid)
 #define current_fsgid(cred)	(current->fsgid)
+#define current_set_flags(f)	(current->flags |= (f))
+#define current_test_flags(f)	(current->flags & (f))
+#define current_clear_flags(f)	(current->flags & ~(f))
+#define current_set_flags_nested(sp, f)		\
+		(*(sp) = current->flags, current->flags |= (f))
+#define current_clear_flags_nested(sp, f)	\
+		(*(sp) = current->flags, current->flags &= ~(f))
+#define current_restore_flags_nested(sp, f)	\
+		(current->flags = ((current->flags & ~(f)) | (*(sp) & (f))))
 
 #define NBPP		PAGE_SIZE
 #define DPPSHFT		(PAGE_SHIFT - 9)
@@ -187,25 +194,9 @@ BUFFER_FNS(PrivateStart, unwritten);
 /* bytes to clicks */
 #define btoc(x)         (((__psunsigned_t)(x)+(NBPC-1))>>BPCSHIFT)
 
-#ifndef ENOATTR
 #define ENOATTR		ENODATA		/* Attribute not found */
-#endif
-
-/* Note: EWRONGFS never visible outside the kernel */
-#define	EWRONGFS	EINVAL		/* Mount with wrong filesystem type */
-
-/*
- * XXX EFSCORRUPTED needs a real value in errno.h. asm-i386/errno.h won't
- *     return codes out of its known range in errno.
- * XXX Also note: needs to be < 1000 and fairly unique on Linux (mustn't
- *     conflict with any code we use already or any code a driver may use)
- * XXX Some options (currently we do #2):
- *	1/ New error code ["Filesystem is corrupted", _after_ glibc updated]
- *	2/ 990 ["Unknown error 990"]
- *	3/ EUCLEAN ["Structure needs cleaning"]
- *	4/ Convert EFSCORRUPTED to EIO [just prior to return into userspace]
- */
-#define EFSCORRUPTED    990		/* Filesystem is corrupted */
+#define EWRONGFS	EINVAL		/* Mount with wrong filesystem type */
+#define EFSCORRUPTED	EUCLEAN		/* Filesystem is corrupted */
 
 #define SYNCHRONIZE()	barrier()
 #define __return_address __builtin_return_address(0)

fs/xfs/linux-2.6/xfs_lrw.c

@@ -23,7 +23,6 @@
 #include "xfs_trans.h"
 #include "xfs_sb.h"
 #include "xfs_ag.h"
-#include "xfs_dir.h"
 #include "xfs_dir2.h"
 #include "xfs_alloc.h"
 #include "xfs_dmapi.h"
@@ -32,7 +31,6 @@
 #include "xfs_bmap_btree.h"
 #include "xfs_alloc_btree.h"
 #include "xfs_ialloc_btree.h"
-#include "xfs_dir_sf.h"
 #include "xfs_dir2_sf.h"
 #include "xfs_attr_sf.h"
 #include "xfs_dinode.h"
@@ -206,7 +204,7 @@ xfs_read(
 	xfs_fsize_t		n;
 	xfs_inode_t		*ip;
 	xfs_mount_t		*mp;
-	vnode_t			*vp;
+	bhv_vnode_t		*vp;
 	unsigned long		seg;
 
 	ip = XFS_BHVTOI(bdp);
@@ -258,7 +256,7 @@ xfs_read(
 
 	if (DM_EVENT_ENABLED(vp->v_vfsp, ip, DM_EVENT_READ) &&
 	    !(ioflags & IO_INVIS)) {
-		vrwlock_t locktype = VRWLOCK_READ;
+		bhv_vrwlock_t locktype = VRWLOCK_READ;
 		int dmflags = FILP_DELAY_FLAG(file) | DM_SEM_FLAG_RD(ioflags);
 
 		ret = -XFS_SEND_DATA(mp, DM_EVENT_READ,
@@ -271,7 +269,7 @@ xfs_read(
 	}
 
 	if (unlikely((ioflags & IO_ISDIRECT) && VN_CACHED(vp)))
-		VOP_FLUSHINVAL_PAGES(vp, ctooff(offtoct(*offset)),
+		bhv_vop_flushinval_pages(vp, ctooff(offtoct(*offset)),
 						-1, FI_REMAPF_LOCKED);
 
 	xfs_rw_enter_trace(XFS_READ_ENTER, &ip->i_iocore,
@@ -313,7 +311,7 @@ xfs_sendfile(
 
 	if (DM_EVENT_ENABLED(BHV_TO_VNODE(bdp)->v_vfsp, ip, DM_EVENT_READ) &&
 	    (!(ioflags & IO_INVIS))) {
-		vrwlock_t locktype = VRWLOCK_READ;
+		bhv_vrwlock_t locktype = VRWLOCK_READ;
 		int error;
 
 		error = XFS_SEND_DATA(mp, DM_EVENT_READ, BHV_TO_VNODE(bdp),
@@ -357,7 +355,7 @@ xfs_splice_read(
 
 	if (DM_EVENT_ENABLED(BHV_TO_VNODE(bdp)->v_vfsp, ip, DM_EVENT_READ) &&
 	    (!(ioflags & IO_INVIS))) {
-		vrwlock_t locktype = VRWLOCK_READ;
+		bhv_vrwlock_t locktype = VRWLOCK_READ;
 		int error;
 
 		error = XFS_SEND_DATA(mp, DM_EVENT_READ, BHV_TO_VNODE(bdp),
@@ -401,7 +399,7 @@ xfs_splice_write(
 
 	if (DM_EVENT_ENABLED(BHV_TO_VNODE(bdp)->v_vfsp, ip, DM_EVENT_WRITE) &&
 	    (!(ioflags & IO_INVIS))) {
-		vrwlock_t locktype = VRWLOCK_WRITE;
+		bhv_vrwlock_t locktype = VRWLOCK_WRITE;
 		int error;
 
 		error = XFS_SEND_DATA(mp, DM_EVENT_WRITE, BHV_TO_VNODE(bdp),
@@ -458,7 +456,7 @@ xfs_zero_last_block(
 	last_fsb = XFS_B_TO_FSBT(mp, isize);
 	nimaps = 1;
 	error = XFS_BMAPI(mp, NULL, io, last_fsb, 1, 0, NULL, 0, &imap,
-			  &nimaps, NULL);
+			  &nimaps, NULL, NULL);
 	if (error) {
 		return error;
 	}
@@ -499,7 +497,7 @@ xfs_zero_last_block(
 
 int					/* error (positive) */
 xfs_zero_eof(
-	vnode_t		*vp,
+	bhv_vnode_t	*vp,
 	xfs_iocore_t	*io,
 	xfs_off_t	offset,		/* starting I/O offset */
 	xfs_fsize_t	isize,		/* current inode size */
@@ -510,7 +508,6 @@ xfs_zero_eof(
 	xfs_fileoff_t	end_zero_fsb;
 	xfs_fileoff_t	zero_count_fsb;
 	xfs_fileoff_t	last_fsb;
-	xfs_extlen_t	buf_len_fsb;
 	xfs_mount_t	*mp = io->io_mount;
 	int		nimaps;
 	int		error = 0;
@@ -556,7 +553,7 @@ xfs_zero_eof(
 		nimaps = 1;
 		zero_count_fsb = end_zero_fsb - start_zero_fsb + 1;
 		error = XFS_BMAPI(mp, NULL, io, start_zero_fsb, zero_count_fsb,
-				  0, NULL, 0, &imap, &nimaps, NULL);
+				  0, NULL, 0, &imap, &nimaps, NULL, NULL);
 		if (error) {
 			ASSERT(ismrlocked(io->io_lock, MR_UPDATE));
 			ASSERT(ismrlocked(io->io_iolock, MR_UPDATE));
@@ -579,16 +576,7 @@ xfs_zero_eof(
 		}
 
 		/*
-		 * There are blocks in the range requested.
-		 * Zero them a single write at a time.  We actually
-		 * don't zero the entire range returned if it is
-		 * too big and simply loop around to get the rest.
-		 * That is not the most efficient thing to do, but it
-		 * is simple and this path should not be exercised often.
-		 */
-		buf_len_fsb = XFS_FILBLKS_MIN(imap.br_blockcount,
-					      mp->m_writeio_blocks << 8);
-		/*
+		 * There are blocks we need to zero.
 		 * Drop the inode lock while we're doing the I/O.
 		 * We'll still have the iolock to protect us.
 		 */
@@ -596,14 +584,13 @@ xfs_zero_eof(
 
 		error = xfs_iozero(ip,
 				   XFS_FSB_TO_B(mp, start_zero_fsb),
-				   XFS_FSB_TO_B(mp, buf_len_fsb),
+				   XFS_FSB_TO_B(mp, imap.br_blockcount),
 				   end_size);
-
 		if (error) {
 			goto out_lock;
 		}
 
-		start_zero_fsb = imap.br_startoff + buf_len_fsb;
+		start_zero_fsb = imap.br_startoff + imap.br_blockcount;
 		ASSERT(start_zero_fsb <= (end_zero_fsb + 1));
 
 		XFS_ILOCK(mp, io, XFS_ILOCK_EXCL|XFS_EXTSIZE_RD);
@@ -637,11 +624,11 @@ xfs_write(
 	ssize_t			ret = 0, error = 0;
 	xfs_fsize_t		isize, new_size;
 	xfs_iocore_t		*io;
-	vnode_t			*vp;
+	bhv_vnode_t		*vp;
 	unsigned long		seg;
 	int			iolock;
 	int			eventsent = 0;
-	vrwlock_t		locktype;
+	bhv_vrwlock_t		locktype;
 	size_t			ocount = 0, count;
 	loff_t			pos;
 	int			need_i_mutex = 1, need_flush = 0;
@@ -679,11 +666,11 @@ xfs_write(
 	io = &xip->i_iocore;
 	mp = io->io_mount;
 
+	vfs_wait_for_freeze(vp->v_vfsp, SB_FREEZE_WRITE);
+
 	if (XFS_FORCED_SHUTDOWN(mp))
 		return -EIO;
 
-	fs_check_frozen(vp->v_vfsp, SB_FREEZE_WRITE);
-
 	if (ioflags & IO_ISDIRECT) {
 		xfs_buftarg_t	*target =
 			(xip->i_d.di_flags & XFS_DIFLAG_REALTIME) ?
@@ -814,7 +801,7 @@ xfs_write(
 		if (need_flush) {
 			xfs_inval_cached_trace(io, pos, -1,
 					ctooff(offtoct(pos)), -1);
-			VOP_FLUSHINVAL_PAGES(vp, ctooff(offtoct(pos)),
+			bhv_vop_flushinval_pages(vp, ctooff(offtoct(pos)),
 					-1, FI_REMAPF_LOCKED);
 		}
 
@@ -903,79 +890,9 @@ xfs_write(
 
 	/* Handle various SYNC-type writes */
 	if ((file->f_flags & O_SYNC) || IS_SYNC(inode)) {
-		/*
-		 * If we're treating this as O_DSYNC and we have not updated the
-		 * size, force the log.
-		 */
-		if (!(mp->m_flags & XFS_MOUNT_OSYNCISOSYNC) &&
-		    !(xip->i_update_size)) {
-			xfs_inode_log_item_t	*iip = xip->i_itemp;
-
-			/*
-			 * If an allocation transaction occurred
-			 * without extending the size, then we have to force
-			 * the log up the proper point to ensure that the
-			 * allocation is permanent.  We can't count on
-			 * the fact that buffered writes lock out direct I/O
-			 * writes - the direct I/O write could have extended
-			 * the size nontransactionally, then finished before
-			 * we started.  xfs_write_file will think that the file
-			 * didn't grow but the update isn't safe unless the
-			 * size change is logged.
-			 *
-			 * Force the log if we've committed a transaction
-			 * against the inode or if someone else has and
-			 * the commit record hasn't gone to disk (e.g.
-			 * the inode is pinned).  This guarantees that
-			 * all changes affecting the inode are permanent
-			 * when we return.
-			 */
-			if (iip && iip->ili_last_lsn) {
-				xfs_log_force(mp, iip->ili_last_lsn,
-						XFS_LOG_FORCE | XFS_LOG_SYNC);
-			} else if (xfs_ipincount(xip) > 0) {
-				xfs_log_force(mp, (xfs_lsn_t)0,
-						XFS_LOG_FORCE | XFS_LOG_SYNC);
-			}
-
-		} else {
-			xfs_trans_t	*tp;
-
-			/*
-			 * O_SYNC or O_DSYNC _with_ a size update are handled
-			 * the same way.
-			 *
-			 * If the write was synchronous then we need to make
-			 * sure that the inode modification time is permanent.
-			 * We'll have updated the timestamp above, so here
-			 * we use a synchronous transaction to log the inode.
-			 * It's not fast, but it's necessary.
-			 *
-			 * If this a dsync write and the size got changed
-			 * non-transactionally, then we need to ensure that
-			 * the size change gets logged in a synchronous
-			 * transaction.
-			 */
-
-			tp = xfs_trans_alloc(mp, XFS_TRANS_WRITE_SYNC);
-			if ((error = xfs_trans_reserve(tp, 0,
-						      XFS_SWRITE_LOG_RES(mp),
-						      0, 0, 0))) {
-				/* Transaction reserve failed */
-				xfs_trans_cancel(tp, 0);
-			} else {
-				/* Transaction reserve successful */
-				xfs_ilock(xip, XFS_ILOCK_EXCL);
-				xfs_trans_ijoin(tp, xip, XFS_ILOCK_EXCL);
-				xfs_trans_ihold(tp, xip);
-				xfs_trans_log_inode(tp, xip, XFS_ILOG_CORE);
-				xfs_trans_set_sync(tp);
-				error = xfs_trans_commit(tp, 0, NULL);
-				xfs_iunlock(xip, XFS_ILOCK_EXCL);
-			}
-			if (error)
-				goto out_unlock_internal;
-		}
+		error = xfs_write_sync_logforce(mp, xip);
+		if (error)
+			goto out_unlock_internal;
 
 		xfs_rwunlock(bdp, locktype);
 		if (need_i_mutex)

fs/xfs/linux-2.6/xfs_lrw.h

@@ -18,8 +18,8 @@
 #ifndef __XFS_LRW_H__
 #define __XFS_LRW_H__
 
-struct vnode;
 struct bhv_desc;
+struct bhv_vnode;
 struct xfs_mount;
 struct xfs_iocore;
 struct xfs_inode;
@@ -49,7 +49,7 @@ struct xfs_iomap;
 #define	XFS_CTRUNC4		14
 #define	XFS_CTRUNC5		15
 #define	XFS_CTRUNC6		16
-#define	XFS_BUNMAPI		17
+#define	XFS_BUNMAP		17
 #define	XFS_INVAL_CACHED	18
 #define	XFS_DIORD_ENTER		19
 #define	XFS_DIOWR_ENTER		20
@@ -82,7 +82,7 @@ extern int xfsbdstrat(struct xfs_mount *, struct xfs_buf *);
 extern int xfs_bdstrat_cb(struct xfs_buf *);
 extern int xfs_dev_is_read_only(struct xfs_mount *, char *);
 
-extern int xfs_zero_eof(struct vnode *, struct xfs_iocore *, xfs_off_t,
+extern int xfs_zero_eof(struct bhv_vnode *, struct xfs_iocore *, xfs_off_t,
 				xfs_fsize_t, xfs_fsize_t);
 extern ssize_t xfs_read(struct bhv_desc *, struct kiocb *,
 				const struct iovec *, unsigned int,

fs/xfs/linux-2.6/xfs_super.h

@@ -105,7 +105,7 @@ struct block_device;
 
 extern __uint64_t xfs_max_file_offset(unsigned int);
 
-extern void xfs_initialize_vnode(bhv_desc_t *, vnode_t *, bhv_desc_t *, int);
+extern void xfs_initialize_vnode(bhv_desc_t *, bhv_vnode_t *, bhv_desc_t *, int);
 
 extern void xfs_flush_inode(struct xfs_inode *);
 extern void xfs_flush_device(struct xfs_inode *);

fs/xfs/linux-2.6/xfs_sysctl.c

@@ -120,6 +120,11 @@ STATIC ctl_table xfs_table[] = {
 	&sysctl_intvec, NULL,
 	&xfs_params.rotorstep.min, &xfs_params.rotorstep.max},
 
+	{XFS_INHERIT_NODFRG, "inherit_nodefrag", &xfs_params.inherit_nodfrg.val,
+	sizeof(int), 0644, NULL, &proc_dointvec_minmax,
+	&sysctl_intvec, NULL,
+	&xfs_params.inherit_nodfrg.min, &xfs_params.inherit_nodfrg.max},
+
 	/* please keep this the last entry */
 #ifdef CONFIG_PROC_FS
 	{XFS_STATS_CLEAR, "stats_clear", &xfs_params.stats_clear.val,

fs/xfs/linux-2.6/xfs_sysctl.h

@@ -46,6 +46,7 @@ typedef struct xfs_param {
 	xfs_sysctl_val_t xfs_buf_age;	/* Metadata buffer age before flush. */
 	xfs_sysctl_val_t inherit_nosym;	/* Inherit the "nosymlinks" flag. */
 	xfs_sysctl_val_t rotorstep;	/* inode32 AG rotoring control knob */
+	xfs_sysctl_val_t inherit_nodfrg;/* Inherit the "nodefrag" inode flag. */
 } xfs_param_t;
 
 /*
@@ -84,6 +85,7 @@ enum {
 	/* XFS_IO_BYPASS = 18 */
 	XFS_INHERIT_NOSYM = 19,
 	XFS_ROTORSTEP = 20,
+	XFS_INHERIT_NODFRG = 21,
 };
 
 extern xfs_param_t	xfs_params;

fs/xfs/linux-2.6/xfs_vfs.c

@@ -23,7 +23,6 @@
 #include "xfs_trans.h"
 #include "xfs_sb.h"
 #include "xfs_ag.h"
-#include "xfs_dir.h"
 #include "xfs_dir2.h"
 #include "xfs_imap.h"
 #include "xfs_alloc.h"
@@ -104,7 +103,7 @@ vfs_mntupdate(
 int
 vfs_root(
 	struct bhv_desc		*bdp,
-	struct vnode		**vpp)
+	struct bhv_vnode	**vpp)
 {
 	struct bhv_desc		*next = bdp;
 
@@ -117,15 +116,15 @@ vfs_root(
 int
 vfs_statvfs(
 	struct bhv_desc		*bdp,
-	xfs_statfs_t		*sp,
-	struct vnode		*vp)
+	bhv_statvfs_t		*statp,
+	struct bhv_vnode	*vp)
 {
 	struct bhv_desc		*next = bdp;
 
 	ASSERT(next);
 	while (! (bhvtovfsops(next))->vfs_statvfs)
 		next = BHV_NEXT(next);
-	return ((*bhvtovfsops(next)->vfs_statvfs)(next, sp, vp));
+	return ((*bhvtovfsops(next)->vfs_statvfs)(next, statp, vp));
 }
 
 int
@@ -145,7 +144,7 @@ vfs_sync(
 int
 vfs_vget(
 	struct bhv_desc		*bdp,
-	struct vnode		**vpp,
+	struct bhv_vnode	**vpp,
 	struct fid		*fidp)
 {
 	struct bhv_desc		*next = bdp;
@@ -187,7 +186,7 @@ vfs_quotactl(
 void
 vfs_init_vnode(
 	struct bhv_desc		*bdp,
-	struct vnode		*vp,
+	struct bhv_vnode	*vp,
 	struct bhv_desc		*bp,
 	int			unlock)
 {
@@ -226,13 +225,13 @@ vfs_freeze(
 	((*bhvtovfsops(next)->vfs_freeze)(next));
 }
 
-vfs_t *
+bhv_vfs_t *
 vfs_allocate(
 	struct super_block	*sb)
 {
-	struct vfs		*vfsp;
+	struct bhv_vfs		*vfsp;
 
-	vfsp = kmem_zalloc(sizeof(vfs_t), KM_SLEEP);
+	vfsp = kmem_zalloc(sizeof(bhv_vfs_t), KM_SLEEP);
 	bhv_head_init(VFS_BHVHEAD(vfsp), "vfs");
 	INIT_LIST_HEAD(&vfsp->vfs_sync_list);
 	spin_lock_init(&vfsp->vfs_sync_lock);
@@ -247,25 +246,25 @@ vfs_allocate(
 	return vfsp;
 }
 
-vfs_t *
+bhv_vfs_t *
 vfs_from_sb(
 	struct super_block	*sb)
 {
-	return (vfs_t *)sb->s_fs_info;
+	return (bhv_vfs_t *)sb->s_fs_info;
 }
 
 void
 vfs_deallocate(
-	struct vfs		*vfsp)
+	struct bhv_vfs		*vfsp)
 {
 	bhv_head_destroy(VFS_BHVHEAD(vfsp));
-	kmem_free(vfsp, sizeof(vfs_t));
+	kmem_free(vfsp, sizeof(bhv_vfs_t));
 }
 
 void
 vfs_insertops(
-	struct vfs		*vfsp,
-	struct bhv_vfsops	*vfsops)
+	struct bhv_vfs		*vfsp,
+	struct bhv_module_vfsops *vfsops)
 {
 	struct bhv_desc		*bdp;
 
@@ -276,9 +275,9 @@ vfs_insertops(
 
 void
 vfs_insertbhv(
-	struct vfs		*vfsp,
+	struct bhv_vfs		*vfsp,
 	struct bhv_desc		*bdp,
-	struct vfsops		*vfsops,
+	struct bhv_vfsops	*vfsops,
 	void			*mount)
 {
 	bhv_desc_init(bdp, mount, vfsp, vfsops);
@@ -287,7 +286,7 @@ vfs_insertbhv(
 
 void
 bhv_remove_vfsops(
-	struct vfs		*vfsp,
+	struct bhv_vfs		*vfsp,
 	int			pos)
 {
 	struct bhv_desc		*bhv;
@@ -301,7 +300,7 @@ bhv_remove_vfsops(
 
 void
 bhv_remove_all_vfsops(
-	struct vfs		*vfsp,
+	struct bhv_vfs		*vfsp,
 	int			freebase)
 {
 	struct xfs_mount	*mp;
@@ -317,7 +316,7 @@ bhv_remove_all_vfsops(
 
 void
 bhv_insert_all_vfsops(
-	struct vfs		*vfsp)
+	struct bhv_vfs		*vfsp)
 {
 	struct xfs_mount	*mp;
 

fs/xfs/linux-2.6/xfs_vnode.c

@@ -39,7 +39,7 @@ vn_init(void)
 
 void
 vn_iowait(
-	struct vnode	*vp)
+	bhv_vnode_t	*vp)
 {
 	wait_queue_head_t *wq = vptosync(vp);
 
@@ -48,17 +48,33 @@ vn_iowait(
 
 void
 vn_iowake(
-	struct vnode	*vp)
+	bhv_vnode_t	*vp)
 {
 	if (atomic_dec_and_test(&vp->v_iocount))
 		wake_up(vptosync(vp));
 }
 
-struct vnode *
+/*
+ * Volume managers supporting multiple paths can send back ENODEV when the
+ * final path disappears.  In this case continuing to fill the page cache
+ * with dirty data which cannot be written out is evil, so prevent that.
+ */
+void
+vn_ioerror(
+	bhv_vnode_t	*vp,
+	int		error,
+	char		*f,
+	int		l)
+{
+	if (unlikely(error == -ENODEV))
+		bhv_vfs_force_shutdown(vp->v_vfsp, SHUTDOWN_DEVICE_REQ, f, l);
+}
+
+bhv_vnode_t *
 vn_initialize(
 	struct inode	*inode)
 {
-	struct vnode	*vp = vn_from_inode(inode);
+	bhv_vnode_t	*vp = vn_from_inode(inode);
 
 	XFS_STATS_INC(vn_active);
 	XFS_STATS_INC(vn_alloc);
@@ -94,8 +110,8 @@ vn_initialize(
  */
 void
 vn_revalidate_core(
-	struct vnode	*vp,
-	vattr_t		*vap)
+	bhv_vnode_t	*vp,
+	bhv_vattr_t	*vap)
 {
 	struct inode	*inode = vn_to_inode(vp);
 
@@ -130,14 +146,14 @@ vn_revalidate_core(
  */
 int
 __vn_revalidate(
-	struct vnode	*vp,
-	struct vattr	*vattr)
+	bhv_vnode_t	*vp,
+	bhv_vattr_t	*vattr)
 {
 	int		error;
 
 	vn_trace_entry(vp, __FUNCTION__, (inst_t *)__return_address);
 	vattr->va_mask = XFS_AT_STAT | XFS_AT_XFLAGS;
-	VOP_GETATTR(vp, vattr, 0, NULL, error);
+	error = bhv_vop_getattr(vp, vattr, 0, NULL);
 	if (likely(!error)) {
 		vn_revalidate_core(vp, vattr);
 		VUNMODIFY(vp);
@@ -147,9 +163,9 @@ __vn_revalidate(
 
 int
 vn_revalidate(
-	struct vnode	*vp)
+	bhv_vnode_t	*vp)
 {
-	vattr_t		vattr;
+	bhv_vattr_t	vattr;
 
 	return __vn_revalidate(vp, &vattr);
 }
@@ -157,9 +173,9 @@ vn_revalidate(
 /*
  * Add a reference to a referenced vnode.
  */
-struct vnode *
+bhv_vnode_t *
 vn_hold(
-	struct vnode	*vp)
+	bhv_vnode_t	*vp)
 {
 	struct inode	*inode;
 
@@ -192,31 +208,31 @@ vn_hold(
  * Vnode tracing code.
  */
 void
-vn_trace_entry(vnode_t *vp, const char *func, inst_t *ra)
+vn_trace_entry(bhv_vnode_t *vp, const char *func, inst_t *ra)
 {
 	KTRACE_ENTER(vp, VNODE_KTRACE_ENTRY, func, 0, ra);
 }
 
 void
-vn_trace_exit(vnode_t *vp, const char *func, inst_t *ra)
+vn_trace_exit(bhv_vnode_t *vp, const char *func, inst_t *ra)
 {
 	KTRACE_ENTER(vp, VNODE_KTRACE_EXIT, func, 0, ra);
 }
 
 void
-vn_trace_hold(vnode_t *vp, char *file, int line, inst_t *ra)
+vn_trace_hold(bhv_vnode_t *vp, char *file, int line, inst_t *ra)
 {
 	KTRACE_ENTER(vp, VNODE_KTRACE_HOLD, file, line, ra);
 }
 
 void
-vn_trace_ref(vnode_t *vp, char *file, int line, inst_t *ra)
+vn_trace_ref(bhv_vnode_t *vp, char *file, int line, inst_t *ra)
 {
 	KTRACE_ENTER(vp, VNODE_KTRACE_REF, file, line, ra);
 }
 
 void
-vn_trace_rele(vnode_t *vp, char *file, int line, inst_t *ra)
+vn_trace_rele(bhv_vnode_t *vp, char *file, int line, inst_t *ra)
 {
 	KTRACE_ENTER(vp, VNODE_KTRACE_RELE, file, line, ra);
 }

fs/xfs/quota/xfs_dquot.c

@@ -23,7 +23,6 @@
 #include "xfs_trans.h"
 #include "xfs_sb.h"
 #include "xfs_ag.h"
-#include "xfs_dir.h"
 #include "xfs_dir2.h"
 #include "xfs_alloc.h"
 #include "xfs_dmapi.h"
@@ -32,7 +31,6 @@
 #include "xfs_bmap_btree.h"
 #include "xfs_alloc_btree.h"
 #include "xfs_ialloc_btree.h"
-#include "xfs_dir_sf.h"
 #include "xfs_dir2_sf.h"
 #include "xfs_attr_sf.h"
 #include "xfs_dinode.h"
@@ -444,7 +442,7 @@ xfs_qm_dqalloc(
 			      XFS_BMAPI_METADATA | XFS_BMAPI_WRITE,
 			      &firstblock,
 			      XFS_QM_DQALLOC_SPACE_RES(mp),
-			      &map, &nmaps, &flist))) {
+			      &map, &nmaps, &flist, NULL))) {
 		goto error0;
 	}
 	ASSERT(map.br_blockcount == XFS_DQUOT_CLUSTER_SIZE_FSB);
@@ -559,7 +557,7 @@ xfs_qm_dqtobp(
 		error = xfs_bmapi(NULL, quotip, dqp->q_fileoffset,
 				  XFS_DQUOT_CLUSTER_SIZE_FSB,
 				  XFS_BMAPI_METADATA,
-				  NULL, 0, &map, &nmaps, NULL);
+				  NULL, 0, &map, &nmaps, NULL, NULL);
 
 		xfs_iunlock(quotip, XFS_ILOCK_SHARED);
 		if (error)
@@ -1261,7 +1259,7 @@ xfs_qm_dqflush(
 
 	if (xfs_qm_dqcheck(&dqp->q_core, be32_to_cpu(ddqp->d_id),
 			   0, XFS_QMOPT_DOWARN, "dqflush (incore copy)")) {
-		xfs_force_shutdown(dqp->q_mount, XFS_CORRUPT_INCORE);
+		xfs_force_shutdown(dqp->q_mount, SHUTDOWN_CORRUPT_INCORE);
 		return XFS_ERROR(EIO);
 	}
 

fs/xfs/quota/xfs_dquot.h

@@ -119,7 +119,7 @@ XFS_DQ_IS_LOCKED(xfs_dquot_t *dqp)
  */
 #define xfs_dqflock(dqp)	 { psema(&((dqp)->q_flock), PINOD | PRECALC);\
 				   (dqp)->dq_flags |= XFS_DQ_FLOCKED; }
-#define xfs_dqfunlock(dqp)	 { ASSERT(valusema(&((dqp)->q_flock)) <= 0); \
+#define xfs_dqfunlock(dqp)	 { ASSERT(issemalocked(&((dqp)->q_flock))); \
 				   vsema(&((dqp)->q_flock)); \
 				   (dqp)->dq_flags &= ~(XFS_DQ_FLOCKED); }
 
@@ -128,7 +128,7 @@ XFS_DQ_IS_LOCKED(xfs_dquot_t *dqp)
 #define XFS_DQ_PINUNLOCK(dqp, s)   mutex_spinunlock( \
 				     &(XFS_DQ_TO_QINF(dqp)->qi_pinlock), s)
 
-#define XFS_DQ_IS_FLUSH_LOCKED(dqp) (valusema(&((dqp)->q_flock)) <= 0)
+#define XFS_DQ_IS_FLUSH_LOCKED(dqp) (issemalocked(&((dqp)->q_flock)))
 #define XFS_DQ_IS_ON_FREELIST(dqp)  ((dqp)->dq_flnext != (dqp))
 #define XFS_DQ_IS_DIRTY(dqp)	((dqp)->dq_flags & XFS_DQ_DIRTY)
 #define XFS_QM_ISUDQ(dqp)	((dqp)->dq_flags & XFS_DQ_USER)

fs/xfs/quota/xfs_dquot_item.c

@@ -23,7 +23,6 @@
 #include "xfs_trans.h"
 #include "xfs_sb.h"
 #include "xfs_ag.h"
-#include "xfs_dir.h"
 #include "xfs_dir2.h"
 #include "xfs_alloc.h"
 #include "xfs_dmapi.h"
@@ -32,7 +31,6 @@
 #include "xfs_bmap_btree.h"
 #include "xfs_alloc_btree.h"
 #include "xfs_ialloc_btree.h"
-#include "xfs_dir_sf.h"
 #include "xfs_dir2_sf.h"
 #include "xfs_attr_sf.h"
 #include "xfs_dinode.h"
@@ -248,7 +246,7 @@ xfs_qm_dquot_logitem_pushbuf(
 	 * inode flush completed and the inode was taken off the AIL.
 	 * So, just get out.
 	 */
-	if ((valusema(&(dqp->q_flock)) > 0)  ||
+	if (!issemalocked(&(dqp->q_flock))  ||
 	    ((qip->qli_item.li_flags & XFS_LI_IN_AIL) == 0)) {
 		qip->qli_pushbuf_flag = 0;
 		xfs_dqunlock(dqp);
@@ -261,7 +259,7 @@ xfs_qm_dquot_logitem_pushbuf(
 	if (bp != NULL) {
 		if (XFS_BUF_ISDELAYWRITE(bp)) {
 			dopush = ((qip->qli_item.li_flags & XFS_LI_IN_AIL) &&
-				  (valusema(&(dqp->q_flock)) <= 0));
+				  issemalocked(&(dqp->q_flock)));
 			qip->qli_pushbuf_flag = 0;
 			xfs_dqunlock(dqp);
 

fs/xfs/quota/xfs_qm.c

@@ -24,7 +24,6 @@
 #include "xfs_trans.h"
 #include "xfs_sb.h"
 #include "xfs_ag.h"
-#include "xfs_dir.h"
 #include "xfs_dir2.h"
 #include "xfs_alloc.h"
 #include "xfs_dmapi.h"
@@ -33,7 +32,6 @@
 #include "xfs_bmap_btree.h"
 #include "xfs_alloc_btree.h"
 #include "xfs_ialloc_btree.h"
-#include "xfs_dir_sf.h"
 #include "xfs_dir2_sf.h"
 #include "xfs_attr_sf.h"
 #include "xfs_dinode.h"
@@ -1603,7 +1601,7 @@ xfs_qm_dqiterate(
 				  maxlblkcnt - lblkno,
 				  XFS_BMAPI_METADATA,
 				  NULL,
-				  0, map, &nmaps, NULL);
+				  0, map, &nmaps, NULL, NULL);
 		xfs_iunlock(qip, XFS_ILOCK_SHARED);
 		if (error)
 			break;
@@ -1905,9 +1903,7 @@ xfs_qm_quotacheck(
 		 */
 		if ((error = xfs_bulkstat(mp, &lastino, &count,
 				     xfs_qm_dqusage_adjust, NULL,
-				     structsz, NULL,
-				     BULKSTAT_FG_IGET|BULKSTAT_FG_VFSLOCKED,
-				     &done)))
+				     structsz, NULL, BULKSTAT_FG_IGET, &done)))
 			break;
 
 	} while (! done);

fs/xfs/quota/xfs_qm_bhv.c

 /*
- * Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * Copyright (c) 2000-2006 Silicon Graphics, Inc.
  * All Rights Reserved.
  *
  * This program is free software; you can redistribute it and/or
@@ -24,7 +24,6 @@
 #include "xfs_trans.h"
 #include "xfs_sb.h"
 #include "xfs_ag.h"
-#include "xfs_dir.h"
 #include "xfs_dir2.h"
 #include "xfs_alloc.h"
 #include "xfs_dmapi.h"
@@ -33,7 +32,6 @@
 #include "xfs_bmap_btree.h"
 #include "xfs_alloc_btree.h"
 #include "xfs_ialloc_btree.h"
-#include "xfs_dir_sf.h"
 #include "xfs_dir2_sf.h"
 #include "xfs_attr_sf.h"
 #include "xfs_dinode.h"
@@ -129,7 +127,7 @@ xfs_qm_parseargs(
 		return XFS_ERROR(EINVAL);
 	}
 
-	PVFS_PARSEARGS(BHV_NEXT(bhv), options, args, update, error);
+	error = bhv_next_vfs_parseargs(BHV_NEXT(bhv), options, args, update);
 	if (!error && !referenced)
 		bhv_remove_vfsops(bhvtovfs(bhv), VFS_POSITION_QM);
 	return error;
@@ -140,9 +138,8 @@ xfs_qm_showargs(
 	struct bhv_desc		*bhv,
 	struct seq_file		*m)
 {
-	struct vfs		*vfsp = bhvtovfs(bhv);
+	struct bhv_vfs		*vfsp = bhvtovfs(bhv);
 	struct xfs_mount	*mp = XFS_VFSTOM(vfsp);
-	int			error;
 
 	if (mp->m_qflags & XFS_UQUOTA_ACCT) {
 		(mp->m_qflags & XFS_UQUOTA_ENFD) ?
@@ -165,8 +162,7 @@ xfs_qm_showargs(
 	if (!(mp->m_qflags & XFS_ALL_QUOTA_ACCT))
 		seq_puts(m, "," MNTOPT_NOQUOTA);
 
-	PVFS_SHOWARGS(BHV_NEXT(bhv), m, error);
-	return error;
+	return bhv_next_vfs_showargs(BHV_NEXT(bhv), m);
 }
 
 STATIC int
@@ -175,14 +171,67 @@ xfs_qm_mount(
 	struct xfs_mount_args	*args,
 	struct cred		*cr)
 {
-	struct vfs		*vfsp = bhvtovfs(bhv);
+	struct bhv_vfs		*vfsp = bhvtovfs(bhv);
 	struct xfs_mount	*mp = XFS_VFSTOM(vfsp);
-	int			error;
 
 	if (args->flags & (XFSMNT_UQUOTA | XFSMNT_GQUOTA | XFSMNT_PQUOTA))
 		xfs_qm_mount_quotainit(mp, args->flags);
-	PVFS_MOUNT(BHV_NEXT(bhv), args, cr, error);
-	return error;
+	return bhv_next_vfs_mount(BHV_NEXT(bhv), args, cr);
+}
+
+/*
+ * Directory tree accounting is implemented using project quotas, where
+ * the project identifier is inherited from parent directories.
+ * A statvfs (df, etc.) of a directory that is using project quota should
+ * return a statvfs of the project, not the entire filesystem.
+ * This makes such trees appear as if they are filesystems in themselves.
+ */
+STATIC int
+xfs_qm_statvfs(
+	struct bhv_desc		*bhv,
+	bhv_statvfs_t		*statp,
+	struct bhv_vnode	*vnode)
+{
+	xfs_mount_t		*mp;
+	xfs_inode_t		*ip;
+	xfs_dquot_t		*dqp;
+	xfs_disk_dquot_t	*dp;
+	__uint64_t		limit;
+	int			error;
+
+	error = bhv_next_vfs_statvfs(BHV_NEXT(bhv), statp, vnode);
+	if (error || !vnode)
+		return error;
+
+	mp = XFS_BHVTOM(bhv);
+	ip = xfs_vtoi(vnode);
+
+	if (!(ip->i_d.di_flags & XFS_DIFLAG_PROJINHERIT))
+		return 0;
+	if (!(mp->m_qflags & XFS_PQUOTA_ACCT))
+		return 0;
+	if (!(mp->m_qflags & XFS_OQUOTA_ENFD))
+		return 0;
+
+	if (xfs_qm_dqget(mp, NULL, ip->i_d.di_projid, XFS_DQ_PROJ, 0, &dqp))
+		return 0;
+	dp = &dqp->q_core;
+
+	limit = dp->d_blk_softlimit ? dp->d_blk_softlimit : dp->d_blk_hardlimit;
+	if (limit && statp->f_blocks > limit) {
+		statp->f_blocks = limit;
+		statp->f_bfree = (statp->f_blocks > dp->d_bcount) ?
+					(statp->f_blocks - dp->d_bcount) : 0;
+	}
+	limit = dp->d_ino_softlimit ? dp->d_ino_softlimit : dp->d_ino_hardlimit;
+	if (limit && statp->f_files > limit) {
+		statp->f_files = limit;
+		statp->f_ffree = (statp->f_files > dp->d_icount) ?
+					(statp->f_ffree - dp->d_icount) : 0;
+	}
+
+	xfs_qm_dqput(dqp);
+	return 0;
 }
 
 STATIC int
@@ -191,7 +240,7 @@ xfs_qm_syncall(
 	int			flags,
 	cred_t			*credp)
 {
-	struct vfs		*vfsp = bhvtovfs(bhv);
+	struct bhv_vfs		*vfsp = bhvtovfs(bhv);
 	struct xfs_mount	*mp = XFS_VFSTOM(vfsp);
 	int			error;
 
@@ -210,8 +259,7 @@ xfs_qm_syncall(
 			}
 		}
 	}
-	PVFS_SYNC(BHV_NEXT(bhv), flags, credp, error);
-	return error;
+	return bhv_next_vfs_sync(BHV_NEXT(bhv), flags, credp);
 }
 
 STATIC int
@@ -346,11 +394,12 @@ STATIC struct xfs_qmops xfs_qmcore_xfs = {
 	.xfs_dqtrxops		= &xfs_trans_dquot_ops,
 };
 
-struct bhv_vfsops xfs_qmops = { {
+struct bhv_module_vfsops xfs_qmops = { {
 	BHV_IDENTITY_INIT(VFS_BHV_QM, VFS_POSITION_QM),
 	.vfs_parseargs		= xfs_qm_parseargs,
 	.vfs_showargs		= xfs_qm_showargs,
 	.vfs_mount		= xfs_qm_mount,
+	.vfs_statvfs		= xfs_qm_statvfs,
 	.vfs_sync		= xfs_qm_syncall,
 	.vfs_quotactl		= xfs_qm_quotactl, },
 };

fs/xfs/quota/xfs_qm_stats.c

@@ -23,7 +23,6 @@
 #include "xfs_trans.h"
 #include "xfs_sb.h"
 #include "xfs_ag.h"
-#include "xfs_dir.h"
 #include "xfs_dir2.h"
 #include "xfs_alloc.h"
 #include "xfs_dmapi.h"
@@ -32,7 +31,6 @@
 #include "xfs_bmap_btree.h"
 #include "xfs_alloc_btree.h"
 #include "xfs_ialloc_btree.h"
-#include "xfs_dir_sf.h"
 #include "xfs_dir2_sf.h"
 #include "xfs_attr_sf.h"
 #include "xfs_dinode.h"

fs/xfs/quota/xfs_qm_syscalls.c

@@ -26,7 +26,6 @@
 #include "xfs_trans.h"
 #include "xfs_sb.h"
 #include "xfs_ag.h"
-#include "xfs_dir.h"
 #include "xfs_dir2.h"
 #include "xfs_alloc.h"
 #include "xfs_dmapi.h"
@@ -35,7 +34,6 @@
 #include "xfs_bmap_btree.h"
 #include "xfs_alloc_btree.h"
 #include "xfs_ialloc_btree.h"
-#include "xfs_dir_sf.h"
 #include "xfs_dir2_sf.h"
 #include "xfs_attr_sf.h"
 #include "xfs_dinode.h"
@@ -91,8 +89,8 @@ xfs_qm_quotactl(
 	xfs_caddr_t	addr)
 {
 	xfs_mount_t	*mp;
+	bhv_vfs_t	*vfsp;
 	int		error;
-	struct vfs	*vfsp;
 
 	vfsp = bhvtovfs(bdp);
 	mp = XFS_VFSTOM(vfsp);
@@ -1035,7 +1033,7 @@ xfs_qm_dqrele_all_inodes(
 {
 	xfs_inode_t	*ip, *topino;
 	uint		ireclaims;
-	vnode_t		*vp;
+	bhv_vnode_t	*vp;
 	boolean_t	vnode_refd;
 
 	ASSERT(mp->m_quotainfo);

fs/xfs/quota/xfs_trans_dquot.c

@@ -23,7 +23,6 @@
 #include "xfs_trans.h"
 #include "xfs_sb.h"
 #include "xfs_ag.h"
-#include "xfs_dir.h"
 #include "xfs_dir2.h"
 #include "xfs_alloc.h"
 #include "xfs_dmapi.h"
@@ -33,7 +32,6 @@
 #include "xfs_alloc_btree.h"
 #include "xfs_ialloc_btree.h"
 #include "xfs_attr_sf.h"
-#include "xfs_dir_sf.h"
 #include "xfs_dir2_sf.h"
 #include "xfs_dinode.h"
 #include "xfs_inode.h"

fs/xfs/support/debug.c

@@ -47,7 +47,7 @@ cmn_err(register int level, char *fmt, ...)
 	va_start(ap, fmt);
 	if (*fmt == '!') fp++;
 	len = vsprintf(message, fp, ap);
-	if (message[len-1] != '\n')
+	if (level != CE_DEBUG && message[len-1] != '\n')
 		strcat(message, "\n");
 	printk("%s%s", err_level[level], message);
 	va_end(ap);
@@ -68,7 +68,7 @@ icmn_err(register int level, char *fmt, va_list ap)
 		level = XFS_MAX_ERR_LEVEL;
 	spin_lock_irqsave(&xfs_err_lock,flags);
 	len = vsprintf(message, fmt, ap);
-	if (message[len-1] != '\n')
+	if (level != CE_DEBUG && message[len-1] != '\n')
 		strcat(message, "\n");
 	spin_unlock_irqrestore(&xfs_err_lock,flags);
 	printk("%s%s", err_level[level], message);

fs/xfs/support/debug.h

@@ -33,9 +33,6 @@ extern void cmn_err(int, char *, ...)
 	__attribute__ ((format (printf, 2, 3)));
 extern void assfail(char *expr, char *f, int l);
 
-#define prdev(fmt,targ,args...) \
-	printk("Device %s - " fmt "\n", XFS_BUFTARG_NAME(targ), ## args)
-
 #define ASSERT_ALWAYS(expr)	\
 	(unlikely((expr) != 0) ? (void)0 : assfail(#expr, __FILE__, __LINE__))
 

fs/xfs/xfs_acl.c

@@ -21,12 +21,10 @@
 #include "xfs_bit.h"
 #include "xfs_inum.h"
 #include "xfs_ag.h"
-#include "xfs_dir.h"
 #include "xfs_dir2.h"
 #include "xfs_bmap_btree.h"
 #include "xfs_alloc_btree.h"
 #include "xfs_ialloc_btree.h"
-#include "xfs_dir_sf.h"
 #include "xfs_dir2_sf.h"
 #include "xfs_attr_sf.h"
 #include "xfs_dinode.h"
@@ -39,15 +37,15 @@
 #include <linux/capability.h>
 #include <linux/posix_acl_xattr.h>
 
-STATIC int	xfs_acl_setmode(vnode_t *, xfs_acl_t *, int *);
+STATIC int	xfs_acl_setmode(bhv_vnode_t *, xfs_acl_t *, int *);
 STATIC void     xfs_acl_filter_mode(mode_t, xfs_acl_t *);
 STATIC void	xfs_acl_get_endian(xfs_acl_t *);
 STATIC int	xfs_acl_access(uid_t, gid_t, xfs_acl_t *, mode_t, cred_t *);
 STATIC int	xfs_acl_invalid(xfs_acl_t *);
 STATIC void	xfs_acl_sync_mode(mode_t, xfs_acl_t *);
-STATIC void	xfs_acl_get_attr(vnode_t *, xfs_acl_t *, int, int, int *);
-STATIC void	xfs_acl_set_attr(vnode_t *, xfs_acl_t *, int, int *);
-STATIC int	xfs_acl_allow_set(vnode_t *, int);
+STATIC void	xfs_acl_get_attr(bhv_vnode_t *, xfs_acl_t *, int, int, int *);
+STATIC void	xfs_acl_set_attr(bhv_vnode_t *, xfs_acl_t *, int, int *);
+STATIC int	xfs_acl_allow_set(bhv_vnode_t *, int);
 
 kmem_zone_t *xfs_acl_zone;
 
@@ -57,7 +55,7 @@ kmem_zone_t *xfs_acl_zone;
  */
 int
 xfs_acl_vhasacl_access(
-	vnode_t		*vp)
+	bhv_vnode_t	*vp)
 {
 	int		error;
 
@@ -70,7 +68,7 @@ xfs_acl_vhasacl_access(
  */
 int
 xfs_acl_vhasacl_default(
-	vnode_t		*vp)
+	bhv_vnode_t	*vp)
 {
 	int		error;
 
@@ -209,7 +207,7 @@ posix_acl_xfs_to_xattr(
 
 int
 xfs_acl_vget(
-	vnode_t		*vp,
+	bhv_vnode_t	*vp,
 	void		*acl,
 	size_t		size,
 	int		kind)
@@ -241,10 +239,10 @@ xfs_acl_vget(
 			goto out;
 		}
 		if (kind == _ACL_TYPE_ACCESS) {
-			vattr_t	va;
+			bhv_vattr_t	va;
 
 			va.va_mask = XFS_AT_MODE;
-			VOP_GETATTR(vp, &va, 0, sys_cred, error);
+			error = bhv_vop_getattr(vp, &va, 0, sys_cred);
 			if (error)
 				goto out;
 			xfs_acl_sync_mode(va.va_mode, xfs_acl);
@@ -260,7 +258,7 @@ xfs_acl_vget(
 
 int
 xfs_acl_vremove(
-	vnode_t		*vp,
+	bhv_vnode_t	*vp,
 	int		kind)
 {
 	int		error;
@@ -268,9 +266,9 @@ xfs_acl_vremove(
 	VN_HOLD(vp);
 	error = xfs_acl_allow_set(vp, kind);
 	if (!error) {
-		VOP_ATTR_REMOVE(vp, kind == _ACL_TYPE_DEFAULT?
-				SGI_ACL_DEFAULT: SGI_ACL_FILE,
-				ATTR_ROOT, sys_cred, error);
+		error = bhv_vop_attr_remove(vp, kind == _ACL_TYPE_DEFAULT?
+						SGI_ACL_DEFAULT: SGI_ACL_FILE,
+						ATTR_ROOT, sys_cred);
 		if (error == ENOATTR)
 			error = 0;	/* 'scool */
 	}
@@ -280,7 +278,7 @@ xfs_acl_vremove(
 
 int
 xfs_acl_vset(
-	vnode_t			*vp,
+	bhv_vnode_t		*vp,
 	void			*acl,
 	size_t			size,
 	int			kind)
@@ -370,10 +368,10 @@ xfs_acl_iaccess(
 
 STATIC int
 xfs_acl_allow_set(
-	vnode_t		*vp,
+	bhv_vnode_t	*vp,
 	int		kind)
 {
-	vattr_t		va;
+	bhv_vattr_t	va;
 	int		error;
 
 	if (vp->v_inode.i_flags & (S_IMMUTABLE|S_APPEND))
@@ -383,7 +381,7 @@ xfs_acl_allow_set(
 	if (vp->v_vfsp->vfs_flag & VFS_RDONLY)
 		return EROFS;
 	va.va_mask = XFS_AT_UID;
-	VOP_GETATTR(vp, &va, 0, NULL, error);
+	error = bhv_vop_getattr(vp, &va, 0, NULL);
 	if (error)
 		return error;
 	if (va.va_uid != current->fsuid && !capable(CAP_FOWNER))
@@ -606,7 +604,7 @@ xfs_acl_get_endian(
  */
 STATIC void
 xfs_acl_get_attr(
-	vnode_t		*vp,
+	bhv_vnode_t	*vp,
 	xfs_acl_t	*aclp,
 	int		kind,
 	int		flags,
@@ -616,9 +614,9 @@ xfs_acl_get_attr(
 
 	ASSERT((flags & ATTR_KERNOVAL) ? (aclp == NULL) : 1);
 	flags |= ATTR_ROOT;
-	VOP_ATTR_GET(vp,
-		kind == _ACL_TYPE_ACCESS ? SGI_ACL_FILE : SGI_ACL_DEFAULT,
-		(char *)aclp, &len, flags, sys_cred, *error);
+	*error = bhv_vop_attr_get(vp, kind == _ACL_TYPE_ACCESS ?
+					SGI_ACL_FILE : SGI_ACL_DEFAULT,
+					(char *)aclp, &len, flags, sys_cred);
 	if (*error || (flags & ATTR_KERNOVAL))
 		return;
 	xfs_acl_get_endian(aclp);
@@ -629,7 +627,7 @@ xfs_acl_get_attr(
  */
 STATIC void
 xfs_acl_set_attr(
-	vnode_t		*vp,
+	bhv_vnode_t	*vp,
 	xfs_acl_t	*aclp,
 	int		kind,
 	int		*error)
@@ -654,19 +652,19 @@ xfs_acl_set_attr(
 		INT_SET(newace->ae_perm, ARCH_CONVERT, ace->ae_perm);
 	}
 	INT_SET(newacl->acl_cnt, ARCH_CONVERT, aclp->acl_cnt);
-	VOP_ATTR_SET(vp,
-		kind == _ACL_TYPE_ACCESS ? SGI_ACL_FILE: SGI_ACL_DEFAULT,
-		(char *)newacl, len, ATTR_ROOT, sys_cred, *error);
+	*error = bhv_vop_attr_set(vp, kind == _ACL_TYPE_ACCESS ?
+				SGI_ACL_FILE: SGI_ACL_DEFAULT,
+				(char *)newacl, len, ATTR_ROOT, sys_cred);
 	_ACL_FREE(newacl);
 }
 
 int
 xfs_acl_vtoacl(
-	vnode_t		*vp,
+	bhv_vnode_t	*vp,
 	xfs_acl_t	*access_acl,
 	xfs_acl_t	*default_acl)
 {
-	vattr_t		va;
+	bhv_vattr_t	va;
 	int		error = 0;
 
 	if (access_acl) {
@@ -678,7 +676,7 @@ xfs_acl_vtoacl(
 		if (!error) {
 			/* Got the ACL, need the mode... */
 			va.va_mask = XFS_AT_MODE;
-			VOP_GETATTR(vp, &va, 0, sys_cred, error);
+			error = bhv_vop_getattr(vp, &va, 0, sys_cred);
 		}
 
 		if (error)
@@ -701,8 +699,8 @@ xfs_acl_vtoacl(
  */
 int
 xfs_acl_inherit(
-	vnode_t		*vp,
-	vattr_t		*vap,
+	bhv_vnode_t	*vp,
+	bhv_vattr_t	*vap,
 	xfs_acl_t	*pdaclp)
 {
 	xfs_acl_t	*cacl;
@@ -757,11 +755,11 @@ xfs_acl_inherit(
  */
 STATIC int
 xfs_acl_setmode(
-	vnode_t		*vp,
+	bhv_vnode_t	*vp,
 	xfs_acl_t	*acl,
 	int		*basicperms)
 {
-	vattr_t		va;
+	bhv_vattr_t	va;
 	xfs_acl_entry_t	*ap;
 	xfs_acl_entry_t	*gap = NULL;
 	int		i, error, nomask = 1;
@@ -776,7 +774,7 @@ xfs_acl_setmode(
 	 * mode.  The m:: bits take precedence over the g:: bits.
 	 */
 	va.va_mask = XFS_AT_MODE;
-	VOP_GETATTR(vp, &va, 0, sys_cred, error);
+	error = bhv_vop_getattr(vp, &va, 0, sys_cred);
 	if (error)
 		return error;
 
@@ -810,8 +808,7 @@ xfs_acl_setmode(
 	if (gap && nomask)
 		va.va_mode |= gap->ae_perm << 3;
 
-	VOP_SETATTR(vp, &va, 0, sys_cred, error);
-	return error;
+	return bhv_vop_setattr(vp, &va, 0, sys_cred);
 }
 
 /*

fs/xfs/xfs_acl.h

@@ -50,7 +50,7 @@ typedef struct xfs_acl {
 #ifdef CONFIG_XFS_POSIX_ACL
 
 struct vattr;
-struct vnode;
+struct bhv_vnode;
 struct xfs_inode;
 
 extern struct kmem_zone *xfs_acl_zone;
@@ -58,14 +58,14 @@ extern struct kmem_zone *xfs_acl_zone;
 		(zone) = kmem_zone_init(sizeof(xfs_acl_t), (name))
 #define xfs_acl_zone_destroy(zone)	kmem_zone_destroy(zone)
 
-extern int xfs_acl_inherit(struct vnode *, struct vattr *, xfs_acl_t *);
+extern int xfs_acl_inherit(struct bhv_vnode *, struct bhv_vattr *, xfs_acl_t *);
 extern int xfs_acl_iaccess(struct xfs_inode *, mode_t, cred_t *);
-extern int xfs_acl_vtoacl(struct vnode *, xfs_acl_t *, xfs_acl_t *);
-extern int xfs_acl_vhasacl_access(struct vnode *);
-extern int xfs_acl_vhasacl_default(struct vnode *);
-extern int xfs_acl_vset(struct vnode *, void *, size_t, int);
-extern int xfs_acl_vget(struct vnode *, void *, size_t, int);
-extern int xfs_acl_vremove(struct vnode *vp, int);
+extern int xfs_acl_vtoacl(struct bhv_vnode *, xfs_acl_t *, xfs_acl_t *);
+extern int xfs_acl_vhasacl_access(struct bhv_vnode *);
+extern int xfs_acl_vhasacl_default(struct bhv_vnode *);
+extern int xfs_acl_vset(struct bhv_vnode *, void *, size_t, int);
+extern int xfs_acl_vget(struct bhv_vnode *, void *, size_t, int);
+extern int xfs_acl_vremove(struct bhv_vnode *, int);
 
 #define _ACL_TYPE_ACCESS	1
 #define _ACL_TYPE_DEFAULT	2

fs/xfs/xfs_alloc.c

@@ -24,14 +24,12 @@
 #include "xfs_trans.h"
 #include "xfs_sb.h"
 #include "xfs_ag.h"
-#include "xfs_dir.h"
 #include "xfs_dir2.h"
 #include "xfs_dmapi.h"
 #include "xfs_mount.h"
 #include "xfs_bmap_btree.h"
 #include "xfs_alloc_btree.h"
 #include "xfs_ialloc_btree.h"
-#include "xfs_dir_sf.h"
 #include "xfs_dir2_sf.h"
 #include "xfs_attr_sf.h"
 #include "xfs_dinode.h"
@@ -1862,7 +1860,7 @@ xfs_alloc_fix_freelist(
 		(pag->pagf_longest - delta) :
 		(pag->pagf_flcount > 0 || pag->pagf_longest > 0);
 	if (args->minlen + args->alignment + args->minalignslop - 1 > longest ||
-	    (args->minleft &&
+	    (!(flags & XFS_ALLOC_FLAG_FREEING) &&
 	     (int)(pag->pagf_freeblks + pag->pagf_flcount -
 		   need - args->total) <
 	     (int)args->minleft)) {
@@ -1898,7 +1896,7 @@ xfs_alloc_fix_freelist(
 	longest = (longest > delta) ? (longest - delta) :
 		(be32_to_cpu(agf->agf_flcount) > 0 || longest > 0);
 	if (args->minlen + args->alignment + args->minalignslop - 1 > longest ||
-	     (args->minleft &&
+	     (!(flags & XFS_ALLOC_FLAG_FREEING) &&
 		(int)(be32_to_cpu(agf->agf_freeblks) +
 		   be32_to_cpu(agf->agf_flcount) - need - args->total) <
 	     (int)args->minleft)) {
@@ -1951,8 +1949,14 @@ xfs_alloc_fix_freelist(
 		 * the restrictions correctly.  Can happen for free calls
 		 * on a completely full ag.
 		 */
-		if (targs.agbno == NULLAGBLOCK)
+		if (targs.agbno == NULLAGBLOCK) {
+			if (!(flags & XFS_ALLOC_FLAG_FREEING)) {
+				xfs_trans_brelse(tp, agflbp);
+				args->agbp = NULL;
+				return 0;
+			}
 			break;
+		}
 		/*
 		 * Put each allocated block on the list.
 		 */
@@ -2360,8 +2364,19 @@ xfs_alloc_vextent(
 			if (args->agno == sagno &&
 			    type == XFS_ALLOCTYPE_START_BNO)
 				args->type = XFS_ALLOCTYPE_THIS_AG;
-			if (++(args->agno) == mp->m_sb.sb_agcount)
-				args->agno = 0;
+			/*
+			* For the first allocation, we can try any AG to get
+			* space.  However, if we already have allocated a
+			* block, we don't want to try AGs whose number is below
+			* sagno. Otherwise, we may end up with out-of-order
+			* locking of AGF, which might cause deadlock.
+			*/
+			if (++(args->agno) == mp->m_sb.sb_agcount) {
+				if (args->firstblock != NULLFSBLOCK)
+					args->agno = sagno;
+				else
+					args->agno = 0;
+			}
 			/*
 			 * Reached the starting a.g., must either be done
 			 * or switch to non-trylock mode.
@@ -2443,7 +2458,7 @@ xfs_free_extent(
 	args.minlen = args.minleft = args.minalignslop = 0;
 	down_read(&args.mp->m_peraglock);
 	args.pag = &args.mp->m_perag[args.agno];
-	if ((error = xfs_alloc_fix_freelist(&args, 0)))
+	if ((error = xfs_alloc_fix_freelist(&args, XFS_ALLOC_FLAG_FREEING)))
 		goto error0;
 #ifdef DEBUG
 	ASSERT(args.agbp != NULL);

fs/xfs/xfs_alloc.h

@@ -41,6 +41,7 @@ typedef enum xfs_alloctype
  * Flags for xfs_alloc_fix_freelist.
  */
 #define	XFS_ALLOC_FLAG_TRYLOCK	0x00000001  /* use trylock for buffer locking */
+#define	XFS_ALLOC_FLAG_FREEING	0x00000002  /* indicate caller is freeing extents*/
 
 /*
  * Argument structure for xfs_alloc routines.
@@ -70,6 +71,7 @@ typedef struct xfs_alloc_arg {
 	char		wasfromfl;	/* set if allocation is from freelist */
 	char		isfl;		/* set if is freelist blocks - !acctg */
 	char		userdata;	/* set if this is user data */
+	xfs_fsblock_t	firstblock;	/* io first block allocated */
 } xfs_alloc_arg_t;
 
 /*

fs/xfs/xfs_alloc_btree.c

@@ -24,14 +24,12 @@
 #include "xfs_trans.h"
 #include "xfs_sb.h"
 #include "xfs_ag.h"
-#include "xfs_dir.h"
 #include "xfs_dir2.h"
 #include "xfs_dmapi.h"
 #include "xfs_mount.h"
 #include "xfs_bmap_btree.h"
 #include "xfs_alloc_btree.h"
 #include "xfs_ialloc_btree.h"
-#include "xfs_dir_sf.h"
 #include "xfs_dir2_sf.h"
 #include "xfs_attr_sf.h"
 #include "xfs_dinode.h"

fs/xfs/xfs_attr.c

@@ -27,7 +27,6 @@
 #include "xfs_trans.h"
 #include "xfs_sb.h"
 #include "xfs_ag.h"
-#include "xfs_dir.h"
 #include "xfs_dir2.h"
 #include "xfs_dmapi.h"
 #include "xfs_mount.h"
@@ -35,7 +34,6 @@
 #include "xfs_bmap_btree.h"
 #include "xfs_alloc_btree.h"
 #include "xfs_ialloc_btree.h"
-#include "xfs_dir_sf.h"
 #include "xfs_dir2_sf.h"
 #include "xfs_attr_sf.h"
 #include "xfs_dinode.h"
@@ -1910,7 +1908,7 @@ xfs_attr_rmtval_get(xfs_da_args_t *args)
 		error = xfs_bmapi(args->trans, args->dp, (xfs_fileoff_t)lblkno,
 				  args->rmtblkcnt,
 				  XFS_BMAPI_ATTRFORK | XFS_BMAPI_METADATA,
-				  NULL, 0, map, &nmap, NULL);
+				  NULL, 0, map, &nmap, NULL, NULL);
 		if (error)
 			return(error);
 		ASSERT(nmap >= 1);
@@ -1988,7 +1986,7 @@ xfs_attr_rmtval_set(xfs_da_args_t *args)
 				  XFS_BMAPI_ATTRFORK | XFS_BMAPI_METADATA |
 							XFS_BMAPI_WRITE,
 				  args->firstblock, args->total, &map, &nmap,
-				  args->flist);
+				  args->flist, NULL);
 		if (!error) {
 			error = xfs_bmap_finish(&args->trans, args->flist,
 						*args->firstblock, &committed);
@@ -2039,7 +2037,8 @@ xfs_attr_rmtval_set(xfs_da_args_t *args)
 		error = xfs_bmapi(NULL, dp, (xfs_fileoff_t)lblkno,
 				  args->rmtblkcnt,
 				  XFS_BMAPI_ATTRFORK | XFS_BMAPI_METADATA,
-				  args->firstblock, 0, &map, &nmap, NULL);
+				  args->firstblock, 0, &map, &nmap,
+				  NULL, NULL);
 		if (error) {
 			return(error);
 		}
@@ -2104,7 +2103,7 @@ xfs_attr_rmtval_remove(xfs_da_args_t *args)
 					args->rmtblkcnt,
 					XFS_BMAPI_ATTRFORK | XFS_BMAPI_METADATA,
 					args->firstblock, 0, &map, &nmap,
-					args->flist);
+					args->flist, NULL);
 		if (error) {
 			return(error);
 		}
@@ -2142,7 +2141,8 @@ xfs_attr_rmtval_remove(xfs_da_args_t *args)
 		XFS_BMAP_INIT(args->flist, args->firstblock);
 		error = xfs_bunmapi(args->trans, args->dp, lblkno, blkcnt,
 				    XFS_BMAPI_ATTRFORK | XFS_BMAPI_METADATA,
-				    1, args->firstblock, args->flist, &done);
+				    1, args->firstblock, args->flist,
+				    NULL, &done);
 		if (!error) {
 			error = xfs_bmap_finish(&args->trans, args->flist,
 						*args->firstblock, &committed);
@@ -2322,56 +2322,56 @@ xfs_attr_trace_enter(int type, char *where,
 
 STATIC int
 posix_acl_access_set(
-	vnode_t	*vp, char *name, void *data, size_t size, int xflags)
+	bhv_vnode_t *vp, char *name, void *data, size_t size, int xflags)
 {
 	return xfs_acl_vset(vp, data, size, _ACL_TYPE_ACCESS);
 }
 
 STATIC int
 posix_acl_access_remove(
-	struct vnode *vp, char *name, int xflags)
+	bhv_vnode_t *vp, char *name, int xflags)
 {
 	return xfs_acl_vremove(vp, _ACL_TYPE_ACCESS);
 }
 
 STATIC int
 posix_acl_access_get(
-	vnode_t *vp, char *name, void *data, size_t size, int xflags)
+	bhv_vnode_t *vp, char *name, void *data, size_t size, int xflags)
 {
 	return xfs_acl_vget(vp, data, size, _ACL_TYPE_ACCESS);
 }
 
 STATIC int
 posix_acl_access_exists(
-	vnode_t *vp)
+	bhv_vnode_t *vp)
 {
 	return xfs_acl_vhasacl_access(vp);
 }
 
 STATIC int
 posix_acl_default_set(
-	vnode_t	*vp, char *name, void *data, size_t size, int xflags)
+	bhv_vnode_t *vp, char *name, void *data, size_t size, int xflags)
 {
 	return xfs_acl_vset(vp, data, size, _ACL_TYPE_DEFAULT);
 }
 
 STATIC int
 posix_acl_default_get(
-	vnode_t *vp, char *name, void *data, size_t size, int xflags)
+	bhv_vnode_t *vp, char *name, void *data, size_t size, int xflags)
 {
 	return xfs_acl_vget(vp, data, size, _ACL_TYPE_DEFAULT);
 }
 
 STATIC int
 posix_acl_default_remove(
-	struct vnode *vp, char *name, int xflags)
+	bhv_vnode_t *vp, char *name, int xflags)
 {
 	return xfs_acl_vremove(vp, _ACL_TYPE_DEFAULT);
 }
 
 STATIC int
 posix_acl_default_exists(
-	vnode_t *vp)
+	bhv_vnode_t *vp)
 {
 	return xfs_acl_vhasacl_default(vp);
 }
@@ -2404,21 +2404,18 @@ STATIC struct attrnames *attr_system_names[] =
 
 STATIC int
 attr_generic_set(
-	struct vnode *vp, char *name, void *data, size_t size, int xflags)
+	bhv_vnode_t *vp, char *name, void *data, size_t size, int xflags)
 {
-	int 	error;
-
-	VOP_ATTR_SET(vp, name, data, size, xflags, NULL, error);
-	return -error;
+	return -bhv_vop_attr_set(vp, name, data, size, xflags, NULL);
 }
 
 STATIC int
 attr_generic_get(
-	struct vnode *vp, char *name, void *data, size_t size, int xflags)
+	bhv_vnode_t *vp, char *name, void *data, size_t size, int xflags)
 {
 	int	error, asize = size;
 
-	VOP_ATTR_GET(vp, name, data, &asize, xflags, NULL, error);
+	error = bhv_vop_attr_get(vp, name, data, &asize, xflags, NULL);
 	if (!error)
 		return asize;
 	return -error;
@@ -2426,12 +2423,9 @@ attr_generic_get(
 
 STATIC int
 attr_generic_remove(
-	struct vnode *vp, char *name, int xflags)
+	bhv_vnode_t *vp, char *name, int xflags)
 {
-	int	error;
-
-	VOP_ATTR_REMOVE(vp, name, xflags, NULL, error);
-	return -error;
+	return -bhv_vop_attr_remove(vp, name, xflags, NULL);
 }
 
 STATIC int
@@ -2459,7 +2453,7 @@ attr_generic_listadd(
 
 STATIC int
 attr_system_list(
-	struct vnode		*vp,
+	bhv_vnode_t		*vp,
 	void			*data,
 	size_t			size,
 	ssize_t			*result)
@@ -2481,12 +2475,12 @@ attr_system_list(
 
 int
 attr_generic_list(
-	struct vnode *vp, void *data, size_t size, int xflags, ssize_t *result)
+	bhv_vnode_t *vp, void *data, size_t size, int xflags, ssize_t *result)
 {
 	attrlist_cursor_kern_t	cursor = { 0 };
 	int			error;
 
-	VOP_ATTR_LIST(vp, data, size, xflags, &cursor, NULL, error);
+	error = bhv_vop_attr_list(vp, data, size, xflags, &cursor, NULL);
 	if (error > 0)
 		return -error;
 	*result = -error;
@@ -2514,7 +2508,7 @@ attr_lookup_namespace(
  */
 STATIC int
 attr_user_capable(
-	struct vnode	*vp,
+	bhv_vnode_t	*vp,
 	cred_t		*cred)
 {
 	struct inode	*inode = vn_to_inode(vp);
@@ -2532,7 +2526,7 @@ attr_user_capable(
 
 STATIC int
 attr_trusted_capable(
-	struct vnode	*vp,
+	bhv_vnode_t	*vp,
 	cred_t		*cred)
 {
 	struct inode	*inode = vn_to_inode(vp);
@@ -2546,7 +2540,7 @@ attr_trusted_capable(
 
 STATIC int
 attr_secure_capable(
-	struct vnode	*vp,
+	bhv_vnode_t	*vp,
 	cred_t		*cred)
 {
 	return -ENOSECURITY;
@@ -2554,7 +2548,7 @@ attr_secure_capable(
 
 STATIC int
 attr_system_set(
-	struct vnode *vp, char *name, void *data, size_t size, int xflags)
+	bhv_vnode_t *vp, char *name, void *data, size_t size, int xflags)
 {
 	attrnames_t	*namesp;
 	int		error;
@@ -2573,7 +2567,7 @@ attr_system_set(
 
 STATIC int
 attr_system_get(
-	struct vnode *vp, char *name, void *data, size_t size, int xflags)
+	bhv_vnode_t *vp, char *name, void *data, size_t size, int xflags)
 {
 	attrnames_t	*namesp;
 
@@ -2585,7 +2579,7 @@ attr_system_get(
 
 STATIC int
 attr_system_remove(
-	struct vnode *vp, char *name, int xflags)
+	bhv_vnode_t *vp, char *name, int xflags)
 {
 	attrnames_t	*namesp;
 

fs/xfs/xfs_attr.h

@@ -36,13 +36,13 @@
  *========================================================================*/
 
 struct cred;
-struct vnode;
+struct bhv_vnode;
 
-typedef int (*attrset_t)(struct vnode *, char *, void *, size_t, int);
-typedef int (*attrget_t)(struct vnode *, char *, void *, size_t, int);
-typedef int (*attrremove_t)(struct vnode *, char *, int);
-typedef int (*attrexists_t)(struct vnode *);
-typedef int (*attrcapable_t)(struct vnode *, struct cred *);
+typedef int (*attrset_t)(struct bhv_vnode *, char *, void *, size_t, int);
+typedef int (*attrget_t)(struct bhv_vnode *, char *, void *, size_t, int);
+typedef int (*attrremove_t)(struct bhv_vnode *, char *, int);
+typedef int (*attrexists_t)(struct bhv_vnode *);
+typedef int (*attrcapable_t)(struct bhv_vnode *, struct cred *);
 
 typedef struct attrnames {
 	char *		attr_name;
@@ -63,7 +63,7 @@ extern struct attrnames attr_trusted;
 extern struct attrnames *attr_namespaces[ATTR_NAMECOUNT];
 
 extern attrnames_t *attr_lookup_namespace(char *, attrnames_t **, int);
-extern int attr_generic_list(struct vnode *, void *, size_t, int, ssize_t *);
+extern int attr_generic_list(struct bhv_vnode *, void *, size_t, int, ssize_t *);
 
 #define ATTR_DONTFOLLOW	0x0001	/* -- unused, from IRIX -- */
 #define ATTR_ROOT	0x0002	/* use attrs in root (trusted) namespace */

fs/xfs/xfs_attr_leaf.c

@@ -24,7 +24,6 @@
 #include "xfs_trans.h"
 #include "xfs_sb.h"
 #include "xfs_ag.h"
-#include "xfs_dir.h"
 #include "xfs_dir2.h"
 #include "xfs_dmapi.h"
 #include "xfs_mount.h"
@@ -34,7 +33,6 @@
 #include "xfs_ialloc_btree.h"
 #include "xfs_alloc.h"
 #include "xfs_btree.h"
-#include "xfs_dir_sf.h"
 #include "xfs_dir2_sf.h"
 #include "xfs_attr_sf.h"
 #include "xfs_dinode.h"
@@ -2990,7 +2988,7 @@ xfs_attr_leaf_freextent(xfs_trans_t **trans, xfs_inode_t *dp,
 		nmap = 1;
 		error = xfs_bmapi(*trans, dp, (xfs_fileoff_t)tblkno, tblkcnt,
 					XFS_BMAPI_ATTRFORK | XFS_BMAPI_METADATA,
-					NULL, 0, &map, &nmap, NULL);
+					NULL, 0, &map, &nmap, NULL, NULL);
 		if (error) {
 			return(error);
 		}

fs/xfs/xfs_bmap.h

 /*
- * Copyright (c) 2000-2003,2005 Silicon Graphics, Inc.
+ * Copyright (c) 2000-2006 Silicon Graphics, Inc.
  * All Rights Reserved.
  *
  * This program is free software; you can redistribute it and/or
@@ -25,6 +25,20 @@ struct xfs_inode;
 struct xfs_mount;
 struct xfs_trans;
 
+/*
+ * DELTA: describe a change to the in-core extent list.
+ *
+ * Internally the use of xed_blockount is somewhat funky.
+ * xed_blockcount contains an offset much of the time because this
+ * makes merging changes easier.  (xfs_fileoff_t and xfs_filblks_t are
+ * the same underlying type).
+ */
+typedef struct xfs_extdelta
+{
+	xfs_fileoff_t		xed_startoff;	/* offset of range */
+	xfs_filblks_t		xed_blockcount;	/* blocks in range */
+} xfs_extdelta_t;
+
 /*
  * List of extents to be free "later".
  * The list is kept sorted on xbf_startblock.
@@ -275,7 +289,9 @@ xfs_bmapi(
 	xfs_extlen_t		total,		/* total blocks needed */
 	struct xfs_bmbt_irec	*mval,		/* output: map values */
 	int			*nmap,		/* i/o: mval size/count */
-	xfs_bmap_free_t		*flist);	/* i/o: list extents to free */
+	xfs_bmap_free_t		*flist,		/* i/o: list extents to free */
+	xfs_extdelta_t		*delta);	/* o: change made to incore
+						   extents */
 
 /*
  * Map file blocks to filesystem blocks, simple version.
@@ -309,6 +325,8 @@ xfs_bunmapi(
 	xfs_fsblock_t		*firstblock,	/* first allocated block
 						   controls a.g. for allocs */
 	xfs_bmap_free_t		*flist,		/* i/o: list extents to free */
+	xfs_extdelta_t		*delta,		/* o: change made to incore
+						   extents */
 	int			*done);		/* set if not done yet */
 
 /*

fs/xfs/xfs_bmap_btree.c

@@ -24,14 +24,12 @@
 #include "xfs_trans.h"
 #include "xfs_sb.h"
 #include "xfs_ag.h"
-#include "xfs_dir.h"
 #include "xfs_dir2.h"
 #include "xfs_dmapi.h"
 #include "xfs_mount.h"
 #include "xfs_bmap_btree.h"
 #include "xfs_alloc_btree.h"
 #include "xfs_ialloc_btree.h"
-#include "xfs_dir_sf.h"
 #include "xfs_dir2_sf.h"
 #include "xfs_attr_sf.h"
 #include "xfs_dinode.h"
@@ -1569,12 +1567,11 @@ xfs_bmbt_split(
 	lbno = XFS_DADDR_TO_FSB(args.mp, XFS_BUF_ADDR(lbp));
 	left = XFS_BUF_TO_BMBT_BLOCK(lbp);
 	args.fsbno = cur->bc_private.b.firstblock;
+	args.firstblock = args.fsbno;
 	if (args.fsbno == NULLFSBLOCK) {
 		args.fsbno = lbno;
 		args.type = XFS_ALLOCTYPE_START_BNO;
-	} else if (cur->bc_private.b.flist->xbf_low)
-		args.type = XFS_ALLOCTYPE_FIRST_AG;
-	else
+	} else
 		args.type = XFS_ALLOCTYPE_NEAR_BNO;
 	args.mod = args.minleft = args.alignment = args.total = args.isfl =
 		args.userdata = args.minalignslop = 0;
@@ -2356,6 +2353,7 @@ xfs_bmbt_newroot(
 		args.userdata = args.minalignslop = 0;
 	args.minlen = args.maxlen = args.prod = 1;
 	args.wasdel = cur->bc_private.b.flags & XFS_BTCUR_BPRV_WASDEL;
+	args.firstblock = args.fsbno;
 	if (args.fsbno == NULLFSBLOCK) {
 #ifdef DEBUG
 		if ((error = xfs_btree_check_lptr(cur, INT_GET(*pp, ARCH_CONVERT), level))) {
@@ -2365,9 +2363,7 @@ xfs_bmbt_newroot(
 #endif
 		args.fsbno = INT_GET(*pp, ARCH_CONVERT);
 		args.type = XFS_ALLOCTYPE_START_BNO;
-	} else if (args.wasdel)
-		args.type = XFS_ALLOCTYPE_FIRST_AG;
-	else
+	} else
 		args.type = XFS_ALLOCTYPE_NEAR_BNO;
 	if ((error = xfs_alloc_vextent(&args))) {
 		XFS_BMBT_TRACE_CURSOR(cur, ERROR);

fs/xfs/xfs_btree.c

@@ -24,14 +24,12 @@
 #include "xfs_trans.h"
 #include "xfs_sb.h"
 #include "xfs_ag.h"
-#include "xfs_dir.h"
 #include "xfs_dir2.h"
 #include "xfs_dmapi.h"
 #include "xfs_mount.h"
 #include "xfs_bmap_btree.h"
 #include "xfs_alloc_btree.h"
 #include "xfs_ialloc_btree.h"
-#include "xfs_dir_sf.h"
 #include "xfs_dir2_sf.h"
 #include "xfs_attr_sf.h"
 #include "xfs_dinode.h"

fs/xfs/xfs_buf_item.c

@@ -23,7 +23,6 @@
 #include "xfs_inum.h"
 #include "xfs_trans.h"
 #include "xfs_sb.h"
-#include "xfs_dir.h"
 #include "xfs_dmapi.h"
 #include "xfs_mount.h"
 #include "xfs_buf_item.h"
@@ -1030,9 +1029,9 @@ xfs_buf_iodone_callbacks(
 		if ((XFS_BUF_TARGET(bp) != lasttarg) ||
 		    (time_after(jiffies, (lasttime + 5*HZ)))) {
 			lasttime = jiffies;
-			prdev("XFS write error in file system meta-data "
-			      "block 0x%llx in %s",
-			      XFS_BUF_TARGET(bp),
+			cmn_err(CE_ALERT, "Device %s, XFS metadata write error"
+					" block 0x%llx in %s",
+				XFS_BUFTARG_NAME(XFS_BUF_TARGET(bp)),
 			      (__uint64_t)XFS_BUF_ADDR(bp), mp->m_fsname);
 		}
 		lasttarg = XFS_BUF_TARGET(bp);
@@ -1108,7 +1107,7 @@ xfs_buf_error_relse(
 	XFS_BUF_ERROR(bp,0);
 	xfs_buftrace("BUF_ERROR_RELSE", bp);
 	if (! XFS_FORCED_SHUTDOWN(mp))
-		xfs_force_shutdown(mp, XFS_METADATA_IO_ERROR);
+		xfs_force_shutdown(mp, SHUTDOWN_META_IO_ERROR);
 	/*
 	 * We have to unpin the pinned buffers so do the
 	 * callbacks.

fs/xfs/xfs_cap.h

@@ -49,12 +49,12 @@ typedef struct xfs_cap_set {
 
 #include <linux/posix_cap_xattr.h>
 
-struct vnode;
+struct bhv_vnode;
 
-extern int xfs_cap_vhascap(struct vnode *);
-extern int xfs_cap_vset(struct vnode *, void *, size_t);
-extern int xfs_cap_vget(struct vnode *, void *, size_t);
-extern int xfs_cap_vremove(struct vnode *vp);
+extern int xfs_cap_vhascap(struct bhv_vnode *);
+extern int xfs_cap_vset(struct bhv_vnode *, void *, size_t);
+extern int xfs_cap_vget(struct bhv_vnode *, void *, size_t);
+extern int xfs_cap_vremove(struct bhv_vnode *);
 
 #define _CAP_EXISTS		xfs_cap_vhascap
 

fs/xfs/xfs_da_btree.h

@@ -36,14 +36,10 @@ struct zone;
  * level in the Btree, and to identify which type of block this is.
  */
 #define XFS_DA_NODE_MAGIC	0xfebe	/* magic number: non-leaf blocks */
-#define XFS_DIR_LEAF_MAGIC	0xfeeb	/* magic number: directory leaf blks */
 #define XFS_ATTR_LEAF_MAGIC	0xfbee	/* magic number: attribute leaf blks */
 #define	XFS_DIR2_LEAF1_MAGIC	0xd2f1	/* magic number: v2 dirlf single blks */
 #define	XFS_DIR2_LEAFN_MAGIC	0xd2ff	/* magic number: v2 dirlf multi blks */
 
-#define	XFS_DIRX_LEAF_MAGIC(mp)	\
-	(XFS_DIR_IS_V1(mp) ? XFS_DIR_LEAF_MAGIC : XFS_DIR2_LEAFN_MAGIC)
-
 typedef struct xfs_da_blkinfo {
 	__be32		forw;			/* previous block in list */
 	__be32		back;			/* following block in list */

fs/xfs/xfs_dfrag.h

@@ -48,6 +48,9 @@ typedef struct xfs_swapext
  */
 int	xfs_swapext(struct xfs_swapext __user *sx);
 
+int	xfs_swap_extents(struct xfs_inode *ip, struct xfs_inode *tip,
+		struct xfs_swapext *sxp);
+
 #endif	/* __KERNEL__ */
 
 #endif	/* __XFS_DFRAG_H__ */