[mirror_ubuntu-bionic-kernel.git] / fs / xfs / xfs_dfrag.c

/*
 * Copyright (c) 2000-2006 Silicon Graphics, Inc.
 * All Rights Reserved.
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public License as
 * published by the Free Software Foundation.
 *
 * This program is distributed in the hope that it would be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * 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"
#include "xfs_fs.h"
#include "xfs_types.h"
#include "xfs_bit.h"
#include "xfs_log.h"
#include "xfs_inum.h"
#include "xfs_trans.h"
#include "xfs_sb.h"
#include "xfs_ag.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_dir2_sf.h"
#include "xfs_attr_sf.h"
#include "xfs_dinode.h"
#include "xfs_inode.h"
#include "xfs_inode_item.h"
#include "xfs_bmap.h"
#include "xfs_btree.h"
#include "xfs_ialloc.h"
#include "xfs_itable.h"
#include "xfs_dfrag.h"
#include "xfs_error.h"
#include "xfs_mac.h"
#include "xfs_rw.h"

/*
 * Syssgi interface for swapext
 */
int
xfs_swapext(
	xfs_swapext_t	__user *sxu)
{
	xfs_swapext_t	*sxp;
	xfs_inode_t     *ip=NULL, *tip=NULL;
	xfs_mount_t     *mp;
	struct file	*fp = NULL, *tfp = NULL;
	bhv_vnode_t	*vp, *tvp;
	int		error = 0;

	sxp = kmem_alloc(sizeof(xfs_swapext_t), KM_MAYFAIL);
	if (!sxp) {
		error = XFS_ERROR(ENOMEM);
		goto error0;
	}

	if (copy_from_user(sxp, sxu, sizeof(xfs_swapext_t))) {
		error = XFS_ERROR(EFAULT);
		goto error0;
	}

	/* Pull information for the target fd */
	if (((fp = fget((int)sxp->sx_fdtarget)) == NULL) ||
	    ((vp = vn_from_inode(fp->f_dentry->d_inode)) == NULL))  {
		error = XFS_ERROR(EINVAL);
		goto error0;
	}

	ip = xfs_vtoi(vp);
	if (ip == NULL) {
		error = XFS_ERROR(EBADF);
		goto error0;
	}

	if (((tfp = fget((int)sxp->sx_fdtmp)) == NULL) ||
	    ((tvp = vn_from_inode(tfp->f_dentry->d_inode)) == NULL)) {
		error = XFS_ERROR(EINVAL);
		goto error0;
	}

	tip = xfs_vtoi(tvp);
	if (tip == NULL) {
		error = XFS_ERROR(EBADF);
		goto error0;
	}

	if (ip->i_mount != tip->i_mount) {
		error =  XFS_ERROR(EINVAL);
		goto error0;
	}

	if (ip->i_ino == tip->i_ino) {
		error =  XFS_ERROR(EINVAL);
		goto error0;
	}

	mp = ip->i_mount;

	if (XFS_FORCED_SHUTDOWN(mp)) {
		error =  XFS_ERROR(EIO);
		goto error0;
	}

	error = XFS_SWAP_EXTENTS(mp, &ip->i_iocore, &tip->i_iocore, sxp);

 error0:
	if (fp != NULL)
		fput(fp);
	if (tfp != NULL)
		fput(tfp);

	if (sxp != NULL)
		kmem_free(sxp, sizeof(xfs_swapext_t));

	return error;
}

int
xfs_swap_extents(
	xfs_inode_t	*ip,
	xfs_inode_t	*tip,
	xfs_swapext_t	*sxp)
{
	xfs_mount_t	*mp;
	xfs_inode_t	*ips[2];
	xfs_trans_t	*tp;
	xfs_bstat_t	*sbp = &sxp->sx_stat;
	bhv_vnode_t	*vp, *tvp;
	xfs_ifork_t	*tempifp, *ifp, *tifp;
	int		ilf_fields, tilf_fields;
	static uint	lock_flags = XFS_ILOCK_EXCL | XFS_IOLOCK_EXCL;
	int		error = 0;
	int		aforkblks = 0;
	int		taforkblks = 0;
	__uint64_t	tmp;
	char		locked = 0;

	mp = ip->i_mount;

	tempifp = kmem_alloc(sizeof(xfs_ifork_t), KM_MAYFAIL);
	if (!tempifp) {
		error = XFS_ERROR(ENOMEM);
		goto error0;
	}

	sbp = &sxp->sx_stat;
	vp = XFS_ITOV(ip);
	tvp = XFS_ITOV(tip);

	/* Lock in i_ino order */
	if (ip->i_ino < tip->i_ino) {
		ips[0] = ip;
		ips[1] = tip;
	} else {
		ips[0] = tip;
		ips[1] = ip;
	}

	xfs_lock_inodes(ips, 2, 0, lock_flags);
	locked = 1;

	/* Check permissions */
	error = xfs_iaccess(ip, S_IWUSR, NULL);
	if (error)
		goto error0;

	error = xfs_iaccess(tip, S_IWUSR, NULL);
	if (error)
		goto error0;

	/* Verify that both files have the same format */
	if ((ip->i_d.di_mode & S_IFMT) != (tip->i_d.di_mode & S_IFMT)) {
		error = XFS_ERROR(EINVAL);
		goto error0;
	}

	/* Verify both files are either real-time or non-realtime */
	if ((ip->i_d.di_flags & XFS_DIFLAG_REALTIME) !=
	    (tip->i_d.di_flags & XFS_DIFLAG_REALTIME)) {
		error = XFS_ERROR(EINVAL);
		goto error0;
	}

	/* Should never get a local format */
	if (ip->i_d.di_format == XFS_DINODE_FMT_LOCAL ||
	    tip->i_d.di_format == XFS_DINODE_FMT_LOCAL) {
		error = XFS_ERROR(EINVAL);
		goto error0;
	}

	if (VN_CACHED(tvp) != 0) {
		xfs_inval_cached_trace(&tip->i_iocore, 0, -1, 0, -1);
		bhv_vop_flushinval_pages(tvp, 0, -1, FI_REMAPF_LOCKED);
	}

	/* Verify O_DIRECT for ftmp */
	if (VN_CACHED(tvp) != 0) {
		error = XFS_ERROR(EINVAL);
		goto error0;
	}

	/* Verify all data are being swapped */
	if (sxp->sx_offset != 0 ||
	    sxp->sx_length != ip->i_d.di_size ||
	    sxp->sx_length != tip->i_d.di_size) {
		error = XFS_ERROR(EFAULT);
		goto error0;
	}

	/*
	 * If the target has extended attributes, the tmp file
	 * must also in order to ensure the correct data fork
	 * format.
	 */
	if ( XFS_IFORK_Q(ip) != XFS_IFORK_Q(tip) ) {
		error = XFS_ERROR(EINVAL);
		goto error0;
	}

	/*
	 * Compare the current change & modify times with that
	 * passed in.  If they differ, we abort this swap.
	 * This is the mechanism used to ensure the calling
	 * process that the file was not changed out from
	 * under it.
	 */
	if ((sbp->bs_ctime.tv_sec != ip->i_d.di_ctime.t_sec) ||
	    (sbp->bs_ctime.tv_nsec != ip->i_d.di_ctime.t_nsec) ||
	    (sbp->bs_mtime.tv_sec != ip->i_d.di_mtime.t_sec) ||
	    (sbp->bs_mtime.tv_nsec != ip->i_d.di_mtime.t_nsec)) {
		error = XFS_ERROR(EBUSY);
		goto error0;
	}

	/* We need to fail if the file is memory mapped.  Once we have tossed
	 * all existing pages, the page fault will have no option
	 * but to go to the filesystem for pages. By making the page fault call
	 * vop_read (or write in the case of autogrow) they block on the iolock
	 * until we have switched the extents.
	 */
	if (VN_MAPPED(vp)) {
		error = XFS_ERROR(EBUSY);
		goto error0;
	}

	xfs_iunlock(ip, XFS_ILOCK_EXCL);
	xfs_iunlock(tip, XFS_ILOCK_EXCL);

	/*
	 * There is a race condition here since we gave up the
	 * ilock.  However, the data fork will not change since
	 * we have the iolock (locked for truncation too) so we
	 * are safe.  We don't really care if non-io related
	 * fields change.
	 */

	bhv_vop_toss_pages(vp, 0, -1, FI_REMAPF);

	tp = xfs_trans_alloc(mp, XFS_TRANS_SWAPEXT);
	if ((error = xfs_trans_reserve(tp, 0,
				     XFS_ICHANGE_LOG_RES(mp), 0,
				     0, 0))) {
		xfs_iunlock(ip,  XFS_IOLOCK_EXCL);
		xfs_iunlock(tip, XFS_IOLOCK_EXCL);
		xfs_trans_cancel(tp, 0);
		locked = 0;
		goto error0;
	}
	xfs_lock_inodes(ips, 2, 0, XFS_ILOCK_EXCL);

	/*
	 * Count the number of extended attribute blocks
	 */
	if ( ((XFS_IFORK_Q(ip) != 0) && (ip->i_d.di_anextents > 0)) &&
	     (ip->i_d.di_aformat != XFS_DINODE_FMT_LOCAL)) {
		error = xfs_bmap_count_blocks(tp, ip, XFS_ATTR_FORK, &aforkblks);
		if (error) {
			xfs_trans_cancel(tp, 0);
			goto error0;
		}
	}
	if ( ((XFS_IFORK_Q(tip) != 0) && (tip->i_d.di_anextents > 0)) &&
	     (tip->i_d.di_aformat != XFS_DINODE_FMT_LOCAL)) {
		error = xfs_bmap_count_blocks(tp, tip, XFS_ATTR_FORK,
			&taforkblks);
		if (error) {
			xfs_trans_cancel(tp, 0);
			goto error0;
		}
	}

	/*
	 * Swap the data forks of the inodes
	 */
	ifp = &ip->i_df;
	tifp = &tip->i_df;
	*tempifp = *ifp;	/* struct copy */
	*ifp = *tifp;		/* struct copy */
	*tifp = *tempifp;	/* struct copy */

	/*
	 * Fix the on-disk inode values
	 */
	tmp = (__uint64_t)ip->i_d.di_nblocks;
	ip->i_d.di_nblocks = tip->i_d.di_nblocks - taforkblks + aforkblks;
	tip->i_d.di_nblocks = tmp + taforkblks - aforkblks;

	tmp = (__uint64_t) ip->i_d.di_nextents;
	ip->i_d.di_nextents = tip->i_d.di_nextents;
	tip->i_d.di_nextents = tmp;

	tmp = (__uint64_t) ip->i_d.di_format;
	ip->i_d.di_format = tip->i_d.di_format;
	tip->i_d.di_format = tmp;

	ilf_fields = XFS_ILOG_CORE;

	switch(ip->i_d.di_format) {
	case XFS_DINODE_FMT_EXTENTS:
		/* If the extents fit in the inode, fix the
		 * pointer.  Otherwise it's already NULL or
		 * pointing to the extent.
		 */
		if (ip->i_d.di_nextents <= XFS_INLINE_EXTS) {
			ifp->if_u1.if_extents =
				ifp->if_u2.if_inline_ext;
		}
		ilf_fields |= XFS_ILOG_DEXT;
		break;
	case XFS_DINODE_FMT_BTREE:
		ilf_fields |= XFS_ILOG_DBROOT;
		break;
	}

	tilf_fields = XFS_ILOG_CORE;

	switch(tip->i_d.di_format) {
	case XFS_DINODE_FMT_EXTENTS:
		/* If the extents fit in the inode, fix the
		 * pointer.  Otherwise it's already NULL or
		 * pointing to the extent.
		 */
		if (tip->i_d.di_nextents <= XFS_INLINE_EXTS) {
			tifp->if_u1.if_extents =
				tifp->if_u2.if_inline_ext;
		}
		tilf_fields |= XFS_ILOG_DEXT;
		break;
	case XFS_DINODE_FMT_BTREE:
		tilf_fields |= XFS_ILOG_DBROOT;
		break;
	}

	/*
	 * Increment vnode ref counts since xfs_trans_commit &
	 * xfs_trans_cancel will both unlock the inodes and
	 * decrement the associated ref counts.
	 */
	VN_HOLD(vp);
	VN_HOLD(tvp);

	xfs_trans_ijoin(tp, ip, lock_flags);
	xfs_trans_ijoin(tp, tip, lock_flags);

	xfs_trans_log_inode(tp, ip,  ilf_fields);
	xfs_trans_log_inode(tp, tip, tilf_fields);

	/*
	 * If this is a synchronous mount, make sure that the
	 * transaction goes to disk before returning to the user.
	 */
	if (mp->m_flags & XFS_MOUNT_WSYNC) {
		xfs_trans_set_sync(tp);
	}

	error = xfs_trans_commit(tp, XFS_TRANS_SWAPEXT, NULL);
	locked = 0;

 error0:
	if (locked) {
		xfs_iunlock(ip,  lock_flags);
		xfs_iunlock(tip, lock_flags);
	}
	if (tempifp != NULL)
		kmem_free(tempifp, sizeof(xfs_ifork_t));
	return error;
}
Commit	Line	Data
1da177e4	1	/*
3e57ecf6	2	* Copyright (c) 2000-2006 Silicon Graphics, Inc.
7b718769	3	* All Rights Reserved.
1da177e4	4	*
7b718769 NS	5	* This program is free software; you can redistribute it and/or
7b718769 NS	6	* modify it under the terms of the GNU General Public License as
1da177e4 LT	7	* published by the Free Software Foundation.
1da177e4 LT	8	*
7b718769 NS	9	* This program is distributed in the hope that it would be useful,
	10	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	11	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	12	* GNU General Public License for more details.
1da177e4	13	*
7b718769 NS	14	* You should have received a copy of the GNU General Public License
	15	* along with this program; if not, write the Free Software Foundation,
	16	* Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
1da177e4	17	*/
1da177e4	18	#include "xfs.h"
a844f451	19	#include "xfs_fs.h"
1da177e4	20	#include "xfs_types.h"
a844f451	21	#include "xfs_bit.h"
1da177e4	22	#include "xfs_log.h"
a844f451	23	#include "xfs_inum.h"
1da177e4 LT	24	#include "xfs_trans.h"
1da177e4 LT	25	#include "xfs_sb.h"
a844f451	26	#include "xfs_ag.h"
1da177e4 LT	27	#include "xfs_dir2.h"
	28	#include "xfs_dmapi.h"
	29	#include "xfs_mount.h"
1da177e4	30	#include "xfs_bmap_btree.h"
a844f451	31	#include "xfs_alloc_btree.h"
1da177e4	32	#include "xfs_ialloc_btree.h"
1da177e4	33	#include "xfs_dir2_sf.h"
a844f451	34	#include "xfs_attr_sf.h"
1da177e4	35	#include "xfs_dinode.h"
1da177e4	36	#include "xfs_inode.h"
a844f451	37	#include "xfs_inode_item.h"
1da177e4	38	#include "xfs_bmap.h"
a844f451	39	#include "xfs_btree.h"
1da177e4 LT	40	#include "xfs_ialloc.h"
	41	#include "xfs_itable.h"
	42	#include "xfs_dfrag.h"
	43	#include "xfs_error.h"
	44	#include "xfs_mac.h"
	45	#include "xfs_rw.h"
	46
	47	/*
	48	* Syssgi interface for swapext
	49	*/
	50	int
	51	xfs_swapext(
d0cfb373	52	xfs_swapext_t __user *sxu)
1da177e4	53	{
d0cfb373	54	xfs_swapext_t *sxp;
3e57ecf6	55	xfs_inode_t ip=NULL, tip=NULL;
1da177e4	56	xfs_mount_t *mp;
1da177e4	57	struct file fp = NULL, tfp = NULL;
67fcaa73	58	bhv_vnode_t vp, tvp;
1da177e4	59	int error = 0;
1da177e4	60
d0cfb373	61	sxp = kmem_alloc(sizeof(xfs_swapext_t), KM_MAYFAIL);
3e57ecf6	62	if (!sxp) {
d0cfb373 ES	63	error = XFS_ERROR(ENOMEM);
	64	goto error0;
	65	}
	66
	67	if (copy_from_user(sxp, sxu, sizeof(xfs_swapext_t))) {
	68	error = XFS_ERROR(EFAULT);
	69	goto error0;
	70	}
1da177e4 LT	71
1da177e4 LT	72	/* Pull information for the target fd */
d0cfb373	73	if (((fp = fget((int)sxp->sx_fdtarget)) == NULL) \|\|
ec86dc02	74	((vp = vn_from_inode(fp->f_dentry->d_inode)) == NULL)) {
1da177e4 LT	75	error = XFS_ERROR(EINVAL);
	76	goto error0;
	77	}
	78
75e17b3c CH	79	ip = xfs_vtoi(vp);
75e17b3c CH	80	if (ip == NULL) {
1da177e4 LT	81	error = XFS_ERROR(EBADF);
1da177e4 LT	82	goto error0;
1da177e4 LT	83	}
1da177e4 LT	84
d0cfb373	85	if (((tfp = fget((int)sxp->sx_fdtmp)) == NULL) \|\|
ec86dc02	86	((tvp = vn_from_inode(tfp->f_dentry->d_inode)) == NULL)) {
1da177e4 LT	87	error = XFS_ERROR(EINVAL);
	88	goto error0;
	89	}
	90
75e17b3c CH	91	tip = xfs_vtoi(tvp);
75e17b3c CH	92	if (tip == NULL) {
1da177e4 LT	93	error = XFS_ERROR(EBADF);
1da177e4 LT	94	goto error0;
1da177e4 LT	95	}
	96
	97	if (ip->i_mount != tip->i_mount) {
	98	error = XFS_ERROR(EINVAL);
	99	goto error0;
	100	}
	101
	102	if (ip->i_ino == tip->i_ino) {
	103	error = XFS_ERROR(EINVAL);
	104	goto error0;
	105	}
	106
	107	mp = ip->i_mount;
	108
1da177e4 LT	109	if (XFS_FORCED_SHUTDOWN(mp)) {
	110	error = XFS_ERROR(EIO);
	111	goto error0;
	112	}
	113
3e57ecf6 OW	114	error = XFS_SWAP_EXTENTS(mp, &ip->i_iocore, &tip->i_iocore, sxp);
	115
	116	error0:
	117	if (fp != NULL)
	118	fput(fp);
	119	if (tfp != NULL)
	120	fput(tfp);
	121
	122	if (sxp != NULL)
	123	kmem_free(sxp, sizeof(xfs_swapext_t));
	124
	125	return error;
	126	}
	127
	128	int
	129	xfs_swap_extents(
	130	xfs_inode_t *ip,
	131	xfs_inode_t *tip,
	132	xfs_swapext_t *sxp)
	133	{
	134	xfs_mount_t *mp;
	135	xfs_inode_t *ips[2];
	136	xfs_trans_t *tp;
	137	xfs_bstat_t *sbp = &sxp->sx_stat;
67fcaa73	138	bhv_vnode_t vp, tvp;
3e57ecf6 OW	139	xfs_ifork_t tempifp, ifp, *tifp;
	140	int ilf_fields, tilf_fields;
	141	static uint lock_flags = XFS_ILOCK_EXCL \| XFS_IOLOCK_EXCL;
	142	int error = 0;
	143	int aforkblks = 0;
	144	int taforkblks = 0;
	145	__uint64_t tmp;
	146	char locked = 0;
	147
	148	mp = ip->i_mount;
	149
	150	tempifp = kmem_alloc(sizeof(xfs_ifork_t), KM_MAYFAIL);
	151	if (!tempifp) {
	152	error = XFS_ERROR(ENOMEM);
	153	goto error0;
	154	}
	155
	156	sbp = &sxp->sx_stat;
	157	vp = XFS_ITOV(ip);
	158	tvp = XFS_ITOV(tip);
1da177e4 LT	159
	160	/* Lock in i_ino order */
	161	if (ip->i_ino < tip->i_ino) {
	162	ips[0] = ip;
	163	ips[1] = tip;
	164	} else {
	165	ips[0] = tip;
	166	ips[1] = ip;
	167	}
d0cfb373	168
1da177e4	169	xfs_lock_inodes(ips, 2, 0, lock_flags);
3e57ecf6	170	locked = 1;
1da177e4 LT	171
	172	/* Check permissions */
	173	error = xfs_iaccess(ip, S_IWUSR, NULL);
	174	if (error)
	175	goto error0;
	176
	177	error = xfs_iaccess(tip, S_IWUSR, NULL);
	178	if (error)
	179	goto error0;
	180
	181	/* Verify that both files have the same format */
	182	if ((ip->i_d.di_mode & S_IFMT) != (tip->i_d.di_mode & S_IFMT)) {
	183	error = XFS_ERROR(EINVAL);
	184	goto error0;
	185	}
	186
	187	/* Verify both files are either real-time or non-realtime */
	188	if ((ip->i_d.di_flags & XFS_DIFLAG_REALTIME) !=
	189	(tip->i_d.di_flags & XFS_DIFLAG_REALTIME)) {
	190	error = XFS_ERROR(EINVAL);
	191	goto error0;
	192	}
	193
	194	/* Should never get a local format */
	195	if (ip->i_d.di_format == XFS_DINODE_FMT_LOCAL \|\|
	196	tip->i_d.di_format == XFS_DINODE_FMT_LOCAL) {
	197	error = XFS_ERROR(EINVAL);
	198	goto error0;
	199	}
	200
bd5a876a CH	201	if (VN_CACHED(tvp) != 0) {
bd5a876a CH	202	xfs_inval_cached_trace(&tip->i_iocore, 0, -1, 0, -1);
67fcaa73	203	bhv_vop_flushinval_pages(tvp, 0, -1, FI_REMAPF_LOCKED);
bd5a876a	204	}
1da177e4 LT	205
	206	/* Verify O_DIRECT for ftmp */
	207	if (VN_CACHED(tvp) != 0) {
	208	error = XFS_ERROR(EINVAL);
	209	goto error0;
	210	}
	211
	212	/* Verify all data are being swapped */
d0cfb373 ES	213	if (sxp->sx_offset != 0 \|\|
	214	sxp->sx_length != ip->i_d.di_size \|\|
	215	sxp->sx_length != tip->i_d.di_size) {
1da177e4 LT	216	error = XFS_ERROR(EFAULT);
	217	goto error0;
	218	}
	219
	220	/*
	221	* If the target has extended attributes, the tmp file
	222	* must also in order to ensure the correct data fork
	223	* format.
	224	*/
	225	if ( XFS_IFORK_Q(ip) != XFS_IFORK_Q(tip) ) {
	226	error = XFS_ERROR(EINVAL);
	227	goto error0;
	228	}
	229
	230	/*
	231	* Compare the current change & modify times with that
	232	* passed in. If they differ, we abort this swap.
	233	* This is the mechanism used to ensure the calling
	234	* process that the file was not changed out from
	235	* under it.
	236	*/
	237	if ((sbp->bs_ctime.tv_sec != ip->i_d.di_ctime.t_sec) \|\|
	238	(sbp->bs_ctime.tv_nsec != ip->i_d.di_ctime.t_nsec) \|\|
	239	(sbp->bs_mtime.tv_sec != ip->i_d.di_mtime.t_sec) \|\|
	240	(sbp->bs_mtime.tv_nsec != ip->i_d.di_mtime.t_nsec)) {
	241	error = XFS_ERROR(EBUSY);
	242	goto error0;
	243	}
	244
	245	/* We need to fail if the file is memory mapped. Once we have tossed
	246	* all existing pages, the page fault will have no option
	247	* but to go to the filesystem for pages. By making the page fault call
67fcaa73	248	* vop_read (or write in the case of autogrow) they block on the iolock
1da177e4 LT	249	* until we have switched the extents.
	250	*/
	251	if (VN_MAPPED(vp)) {
	252	error = XFS_ERROR(EBUSY);
	253	goto error0;
	254	}
	255
	256	xfs_iunlock(ip, XFS_ILOCK_EXCL);
	257	xfs_iunlock(tip, XFS_ILOCK_EXCL);
	258
	259	/*
	260	* There is a race condition here since we gave up the
	261	* ilock. However, the data fork will not change since
	262	* we have the iolock (locked for truncation too) so we
	263	* are safe. We don't really care if non-io related
	264	* fields change.
	265	*/
	266
67fcaa73	267	bhv_vop_toss_pages(vp, 0, -1, FI_REMAPF);
1da177e4 LT	268
	269	tp = xfs_trans_alloc(mp, XFS_TRANS_SWAPEXT);
	270	if ((error = xfs_trans_reserve(tp, 0,
	271	XFS_ICHANGE_LOG_RES(mp), 0,
	272	0, 0))) {
	273	xfs_iunlock(ip, XFS_IOLOCK_EXCL);
	274	xfs_iunlock(tip, XFS_IOLOCK_EXCL);
	275	xfs_trans_cancel(tp, 0);
d0cfb373 ES	276	locked = 0;
d0cfb373 ES	277	goto error0;
1da177e4 LT	278	}
	279	xfs_lock_inodes(ips, 2, 0, XFS_ILOCK_EXCL);
	280
	281	/*
	282	* Count the number of extended attribute blocks
	283	*/
	284	if ( ((XFS_IFORK_Q(ip) != 0) && (ip->i_d.di_anextents > 0)) &&
	285	(ip->i_d.di_aformat != XFS_DINODE_FMT_LOCAL)) {
	286	error = xfs_bmap_count_blocks(tp, ip, XFS_ATTR_FORK, &aforkblks);
	287	if (error) {
1da177e4	288	xfs_trans_cancel(tp, 0);
d0cfb373	289	goto error0;
1da177e4 LT	290	}
	291	}
	292	if ( ((XFS_IFORK_Q(tip) != 0) && (tip->i_d.di_anextents > 0)) &&
	293	(tip->i_d.di_aformat != XFS_DINODE_FMT_LOCAL)) {
	294	error = xfs_bmap_count_blocks(tp, tip, XFS_ATTR_FORK,
	295	&taforkblks);
	296	if (error) {
1da177e4	297	xfs_trans_cancel(tp, 0);
d0cfb373	298	goto error0;
1da177e4 LT	299	}
	300	}
	301
	302	/*
	303	* Swap the data forks of the inodes
	304	*/
	305	ifp = &ip->i_df;
	306	tifp = &tip->i_df;
d0cfb373 ES	307	tempifp = ifp; /* struct copy */
	308	ifp = tifp; /* struct copy */
	309	tifp = tempifp; /* struct copy */
1da177e4 LT	310
	311	/*
	312	* Fix the on-disk inode values
	313	*/
	314	tmp = (__uint64_t)ip->i_d.di_nblocks;
	315	ip->i_d.di_nblocks = tip->i_d.di_nblocks - taforkblks + aforkblks;
	316	tip->i_d.di_nblocks = tmp + taforkblks - aforkblks;
	317
	318	tmp = (__uint64_t) ip->i_d.di_nextents;
	319	ip->i_d.di_nextents = tip->i_d.di_nextents;
	320	tip->i_d.di_nextents = tmp;
	321
	322	tmp = (__uint64_t) ip->i_d.di_format;
	323	ip->i_d.di_format = tip->i_d.di_format;
	324	tip->i_d.di_format = tmp;
	325
	326	ilf_fields = XFS_ILOG_CORE;
	327
	328	switch(ip->i_d.di_format) {
	329	case XFS_DINODE_FMT_EXTENTS:
	330	/* If the extents fit in the inode, fix the
	331	* pointer. Otherwise it's already NULL or
	332	* pointing to the extent.
	333	*/
	334	if (ip->i_d.di_nextents <= XFS_INLINE_EXTS) {
	335	ifp->if_u1.if_extents =
	336	ifp->if_u2.if_inline_ext;
	337	}
	338	ilf_fields \|= XFS_ILOG_DEXT;
	339	break;
	340	case XFS_DINODE_FMT_BTREE:
	341	ilf_fields \|= XFS_ILOG_DBROOT;
	342	break;
	343	}
	344
	345	tilf_fields = XFS_ILOG_CORE;
	346
	347	switch(tip->i_d.di_format) {
	348	case XFS_DINODE_FMT_EXTENTS:
	349	/* If the extents fit in the inode, fix the
	350	* pointer. Otherwise it's already NULL or
	351	* pointing to the extent.
	352	*/
	353	if (tip->i_d.di_nextents <= XFS_INLINE_EXTS) {
	354	tifp->if_u1.if_extents =
	355	tifp->if_u2.if_inline_ext;
	356	}
	357	tilf_fields \|= XFS_ILOG_DEXT;
	358	break;
	359	case XFS_DINODE_FMT_BTREE:
	360	tilf_fields \|= XFS_ILOG_DBROOT;
	361	break;
	362	}
	363
	364	/*
	365	* Increment vnode ref counts since xfs_trans_commit &
	366	* xfs_trans_cancel will both unlock the inodes and
	367	* decrement the associated ref counts.
	368	*/
	369	VN_HOLD(vp);
	370	VN_HOLD(tvp);
	371
	372	xfs_trans_ijoin(tp, ip, lock_flags);
	373	xfs_trans_ijoin(tp, tip, lock_flags);
374
375	xfs_trans_log_inode(tp, ip, ilf_fields);
376	xfs_trans_log_inode(tp, tip, tilf_fields);
377
378	/*
379	* If this is a synchronous mount, make sure that the
380	* transaction goes to disk before returning to the user.
381	*/
382	if (mp->m_flags & XFS_MOUNT_WSYNC) {
383	xfs_trans_set_sync(tp);
384	}
385
386	error = xfs_trans_commit(tp, XFS_TRANS_SWAPEXT, NULL);
d0cfb373	387	locked = 0;
1da177e4 LT	388
	389	error0:
	390	if (locked) {
	391	xfs_iunlock(ip, lock_flags);
	392	xfs_iunlock(tip, lock_flags);
	393	}
d0cfb373 ES	394	if (tempifp != NULL)
d0cfb373 ES	395	kmem_free(tempifp, sizeof(xfs_ifork_t));
1da177e4 LT	396	return error;
1da177e4 LT	397	}