Doug / smarc-fsl-linux-kernel | Embedian Git Server

Commit 745b1f47fc0c68dbb1ff440eec8889f61e57194b

Authored by Nathan Scott 2006-09-28 09:02:23 +0800

Committed by Tim Shimmin 2006-09-28 09:02:23 +0800

Exists in master and in 7 other branches

[XFS] Remove last bulkstat false-positives with debug kernels.

SGI-PV: 953819
SGI-Modid: xfs-linux-melb:xfs-kern:26628a

Signed-off-by: Nathan Scott <nathans@sgi.com>
Signed-off-by: Tim Shimmin <tes@sgi.com>

Showing 5 changed files with 22 additions and 17 deletions Inline Diff

fs/xfs/xfs_ialloc.c
fs/xfs/xfs_iget.c
fs/xfs/xfs_inode.c
fs/xfs/xfs_inode.h
fs/xfs/xfs_itable.c

fs/xfs/xfs_ialloc.c

Diff comments View file @ 745b1f4

 /*
  * Copyright (c) 2000-2002,2005 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_btree.h"
 #include "xfs_ialloc.h"
 #include "xfs_alloc.h"
 #include "xfs_rtalloc.h"
 #include "xfs_error.h"
 #include "xfs_bmap.h"
 /*
  * Log specified fields for the inode given by bp and off.
  */
 STATIC void
 xfs_ialloc_log_di(
 	xfs_trans_t	*tp,		/* transaction pointer */
 	xfs_buf_t	*bp,		/* inode buffer */
 	int		off,		/* index of inode in buffer */
 	int		fields)		/* bitmask of fields to log */
 {
 	int			first;		/* first byte number */
 	int			ioffset;	/* off in bytes */
 	int			last;		/* last byte number */
 	xfs_mount_t		*mp;		/* mount point structure */
 	static const short	offsets[] = {	/* field offsets */
 						/* keep in sync with bits */
 		offsetof(xfs_dinode_core_t, di_magic),
 		offsetof(xfs_dinode_core_t, di_mode),
 		offsetof(xfs_dinode_core_t, di_version),
 		offsetof(xfs_dinode_core_t, di_format),
 		offsetof(xfs_dinode_core_t, di_onlink),
 		offsetof(xfs_dinode_core_t, di_uid),
 		offsetof(xfs_dinode_core_t, di_gid),
 		offsetof(xfs_dinode_core_t, di_nlink),
 		offsetof(xfs_dinode_core_t, di_projid),
 		offsetof(xfs_dinode_core_t, di_pad),
 		offsetof(xfs_dinode_core_t, di_atime),
 		offsetof(xfs_dinode_core_t, di_mtime),
 		offsetof(xfs_dinode_core_t, di_ctime),
 		offsetof(xfs_dinode_core_t, di_size),
 		offsetof(xfs_dinode_core_t, di_nblocks),
 		offsetof(xfs_dinode_core_t, di_extsize),
 		offsetof(xfs_dinode_core_t, di_nextents),
 		offsetof(xfs_dinode_core_t, di_anextents),
 		offsetof(xfs_dinode_core_t, di_forkoff),
 		offsetof(xfs_dinode_core_t, di_aformat),
 		offsetof(xfs_dinode_core_t, di_dmevmask),
 		offsetof(xfs_dinode_core_t, di_dmstate),
 		offsetof(xfs_dinode_core_t, di_flags),
 		offsetof(xfs_dinode_core_t, di_gen),
 		offsetof(xfs_dinode_t, di_next_unlinked),
 		offsetof(xfs_dinode_t, di_u),
 		offsetof(xfs_dinode_t, di_a),
 		sizeof(xfs_dinode_t)
 	};
 	ASSERT(offsetof(xfs_dinode_t, di_core) == 0);
 	ASSERT((fields & (XFS_DI_U|XFS_DI_A)) == 0);
 	mp = tp->t_mountp;
 	/*
 	 * Get the inode-relative first and last bytes for these fields
 	 */
 	xfs_btree_offsets(fields, offsets, XFS_DI_NUM_BITS, &first, &last);
 	/*
 	 * Convert to buffer offsets and log it.
 	 */
 	ioffset = off << mp->m_sb.sb_inodelog;
 	first += ioffset;
 	last += ioffset;
 	xfs_trans_log_buf(tp, bp, first, last);
 }
 /*
  * Allocation group level functions.
  */
 /*
  * Allocate new inodes in the allocation group specified by agbp.
  * Return 0 for success, else error code.
  */
 STATIC int				/* error code or 0 */
 xfs_ialloc_ag_alloc(
 	xfs_trans_t	*tp,		/* transaction pointer */
 	xfs_buf_t	*agbp,		/* alloc group buffer */
 	int		*alloc)
 {
 	xfs_agi_t	*agi;		/* allocation group header */
 	xfs_alloc_arg_t	args;		/* allocation argument structure */
 	int		blks_per_cluster;  /* fs blocks per inode cluster */
 	xfs_btree_cur_t	*cur;		/* inode btree cursor */
 	xfs_daddr_t	d;		/* disk addr of buffer */
 	int		error;
 	xfs_buf_t	*fbuf;		/* new free inodes' buffer */
 	xfs_dinode_t	*free;		/* new free inode structure */
 	int		i;		/* inode counter */
 	int		j;		/* block counter */
 	int		nbufs;		/* num bufs of new inodes */
 	xfs_agino_t	newino;		/* new first inode's number */
 	xfs_agino_t	newlen;		/* new number of inodes */
 	int		ninodes;	/* num inodes per buf */
 	xfs_agino_t	thisino;	/* current inode number, for loop */
 	int		version;	/* inode version number to use */
 	int		isaligned = 0;	/* inode allocation at stripe unit */
 					/* boundary */
 	args.tp = tp;
 	args.mp = tp->t_mountp;
 	/*
 	 * Locking will ensure that we don't have two callers in here
 	 * at one time.
 	 */
 	newlen = XFS_IALLOC_INODES(args.mp);
 	if (args.mp->m_maxicount &&
 	    args.mp->m_sb.sb_icount + newlen > args.mp->m_maxicount)
 		return XFS_ERROR(ENOSPC);
 	args.minlen = args.maxlen = XFS_IALLOC_BLOCKS(args.mp);
 	/*
 	 * First try to allocate inodes contiguous with the last-allocated
 	 * chunk of inodes.  If the filesystem is striped, this will fill
 	 * an entire stripe unit with inodes.
  	 */
 	agi = XFS_BUF_TO_AGI(agbp);
 	newino = be32_to_cpu(agi->agi_newino);
 	args.agbno = XFS_AGINO_TO_AGBNO(args.mp, newino) +
 			XFS_IALLOC_BLOCKS(args.mp);
 	if (likely(newino != NULLAGINO &&
 		  (args.agbno < be32_to_cpu(agi->agi_length)))) {
 		args.fsbno = XFS_AGB_TO_FSB(args.mp,
 				be32_to_cpu(agi->agi_seqno), args.agbno);
 		args.type = XFS_ALLOCTYPE_THIS_BNO;
 		args.mod = args.total = args.wasdel = args.isfl =
 			args.userdata = args.minalignslop = 0;
 		args.prod = 1;
 		args.alignment = 1;
 		/*
 		 * Allow space for the inode btree to split.
 		 */
 		args.minleft = XFS_IN_MAXLEVELS(args.mp) - 1;
 		if ((error = xfs_alloc_vextent(&args)))
 			return error;
 	} else
 		args.fsbno = NULLFSBLOCK;
 	if (unlikely(args.fsbno == NULLFSBLOCK)) {
 		/*
 		 * Set the alignment for the allocation.
 		 * If stripe alignment is turned on then align at stripe unit
 		 * boundary.
 		 * If the cluster size is smaller than a filesystem block
 		 * then we're doing I/O for inodes in filesystem block size
 		 * pieces, so don't need alignment anyway.
 		 */
 		isaligned = 0;
 		if (args.mp->m_sinoalign) {
 			ASSERT(!(args.mp->m_flags & XFS_MOUNT_NOALIGN));
 			args.alignment = args.mp->m_dalign;
 			isaligned = 1;
 		} else if (XFS_SB_VERSION_HASALIGN(&args.mp->m_sb) &&
 			   args.mp->m_sb.sb_inoalignmt >=
 			   XFS_B_TO_FSBT(args.mp,
 			  	XFS_INODE_CLUSTER_SIZE(args.mp)))
 				args.alignment = args.mp->m_sb.sb_inoalignmt;
 		else
 			args.alignment = 1;
 		/*
 		 * Need to figure out where to allocate the inode blocks.
 		 * Ideally they should be spaced out through the a.g.
 		 * For now, just allocate blocks up front.
 		 */
 		args.agbno = be32_to_cpu(agi->agi_root);
 		args.fsbno = XFS_AGB_TO_FSB(args.mp,
 				be32_to_cpu(agi->agi_seqno), args.agbno);
 		/*
 		 * Allocate a fixed-size extent of inodes.
 		 */
 		args.type = XFS_ALLOCTYPE_NEAR_BNO;
 		args.mod = args.total = args.wasdel = args.isfl =
 			args.userdata = args.minalignslop = 0;
 		args.prod = 1;
 		/*
 		 * Allow space for the inode btree to split.
 		 */
 		args.minleft = XFS_IN_MAXLEVELS(args.mp) - 1;
 		if ((error = xfs_alloc_vextent(&args)))
 			return error;
 	}
 	/*
 	 * If stripe alignment is turned on, then try again with cluster
 	 * alignment.
 	 */
 	if (isaligned && args.fsbno == NULLFSBLOCK) {
 		args.type = XFS_ALLOCTYPE_NEAR_BNO;
 		args.agbno = be32_to_cpu(agi->agi_root);
 		args.fsbno = XFS_AGB_TO_FSB(args.mp,
 				be32_to_cpu(agi->agi_seqno), args.agbno);
 		if (XFS_SB_VERSION_HASALIGN(&args.mp->m_sb) &&
 			args.mp->m_sb.sb_inoalignmt >=
 			XFS_B_TO_FSBT(args.mp, XFS_INODE_CLUSTER_SIZE(args.mp)))
 				args.alignment = args.mp->m_sb.sb_inoalignmt;
 		else
 			args.alignment = 1;
 		if ((error = xfs_alloc_vextent(&args)))
 			return error;
 	}
 	if (args.fsbno == NULLFSBLOCK) {
 		*alloc = 0;
 		return 0;
 	}
 	ASSERT(args.len == args.minlen);
 	/*
 	 * Convert the results.
 	 */
 	newino = XFS_OFFBNO_TO_AGINO(args.mp, args.agbno, 0);
 	/*
 	 * Loop over the new block(s), filling in the inodes.
 	 * For small block sizes, manipulate the inodes in buffers
 	 * which are multiples of the blocks size.
 	 */
 	if (args.mp->m_sb.sb_blocksize >= XFS_INODE_CLUSTER_SIZE(args.mp)) {
 		blks_per_cluster = 1;
 		nbufs = (int)args.len;
 		ninodes = args.mp->m_sb.sb_inopblock;
 	} else {
 		blks_per_cluster = XFS_INODE_CLUSTER_SIZE(args.mp) /
 				   args.mp->m_sb.sb_blocksize;
 		nbufs = (int)args.len / blks_per_cluster;
 		ninodes = blks_per_cluster * args.mp->m_sb.sb_inopblock;
 	}
 	/*
 	 * Figure out what version number to use in the inodes we create.
 	 * If the superblock version has caught up to the one that supports
 	 * the new inode format, then use the new inode version.  Otherwise
 	 * use the old version so that old kernels will continue to be
 	 * able to use the file system.
 	 */
 	if (XFS_SB_VERSION_HASNLINK(&args.mp->m_sb))
 		version = XFS_DINODE_VERSION_2;
 	else
 		version = XFS_DINODE_VERSION_1;
 	for (j = 0; j < nbufs; j++) {
 		/*
 		 * Get the block.
 		 */
 		d = XFS_AGB_TO_DADDR(args.mp, be32_to_cpu(agi->agi_seqno),
 				     args.agbno + (j * blks_per_cluster));
 		fbuf = xfs_trans_get_buf(tp, args.mp->m_ddev_targp, d,
 					 args.mp->m_bsize * blks_per_cluster,
 					 XFS_BUF_LOCK);
 		ASSERT(fbuf);
 		ASSERT(!XFS_BUF_GETERROR(fbuf));
 		/*
 		 * Set initial values for the inodes in this buffer.
 		 */
 		xfs_biozero(fbuf, 0, ninodes << args.mp->m_sb.sb_inodelog);
 		for (i = 0; i < ninodes; i++) {
 			free = XFS_MAKE_IPTR(args.mp, fbuf, i);
 			INT_SET(free->di_core.di_magic, ARCH_CONVERT, XFS_DINODE_MAGIC);
 			INT_SET(free->di_core.di_version, ARCH_CONVERT, version);
 			INT_SET(free->di_next_unlinked, ARCH_CONVERT, NULLAGINO);
 			xfs_ialloc_log_di(tp, fbuf, i,
 				XFS_DI_CORE_BITS | XFS_DI_NEXT_UNLINKED);
 		}
 		xfs_trans_inode_alloc_buf(tp, fbuf);
 	}
 	be32_add(&agi->agi_count, newlen);
 	be32_add(&agi->agi_freecount, newlen);
 	down_read(&args.mp->m_peraglock);
 	args.mp->m_perag[be32_to_cpu(agi->agi_seqno)].pagi_freecount += newlen;
 	up_read(&args.mp->m_peraglock);
 	agi->agi_newino = cpu_to_be32(newino);
 	/*
 	 * Insert records describing the new inode chunk into the btree.
 	 */
 	cur = xfs_btree_init_cursor(args.mp, tp, agbp,
 			be32_to_cpu(agi->agi_seqno),
 			XFS_BTNUM_INO, (xfs_inode_t *)0, 0);
 	for (thisino = newino;
 	     thisino < newino + newlen;
 	     thisino += XFS_INODES_PER_CHUNK) {
 		if ((error = xfs_inobt_lookup_eq(cur, thisino,
 				XFS_INODES_PER_CHUNK, XFS_INOBT_ALL_FREE, &i))) {
 			xfs_btree_del_cursor(cur, XFS_BTREE_ERROR);
 			return error;
 		}
 		ASSERT(i == 0);
 		if ((error = xfs_inobt_insert(cur, &i))) {
 			xfs_btree_del_cursor(cur, XFS_BTREE_ERROR);
 			return error;
 		}
 		ASSERT(i == 1);
 	}
 	xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
 	/*
 	 * Log allocation group header fields
 	 */
 	xfs_ialloc_log_agi(tp, agbp,
 		XFS_AGI_COUNT | XFS_AGI_FREECOUNT | XFS_AGI_NEWINO);
 	/*
 	 * Modify/log superblock values for inode count and inode free count.
 	 */
 	xfs_trans_mod_sb(tp, XFS_TRANS_SB_ICOUNT, (long)newlen);
 	xfs_trans_mod_sb(tp, XFS_TRANS_SB_IFREE, (long)newlen);
 	*alloc = 1;
 	return 0;
 }
 STATIC __inline xfs_agnumber_t
 xfs_ialloc_next_ag(
 	xfs_mount_t	*mp)
 {
 	xfs_agnumber_t	agno;
 	spin_lock(&mp->m_agirotor_lock);
 	agno = mp->m_agirotor;
 	if (++mp->m_agirotor == mp->m_maxagi)
 		mp->m_agirotor = 0;
 	spin_unlock(&mp->m_agirotor_lock);
 	return agno;
 }
 /*
  * Select an allocation group to look for a free inode in, based on the parent
  * inode and then mode.  Return the allocation group buffer.
  */
 STATIC xfs_buf_t *			/* allocation group buffer */
 xfs_ialloc_ag_select(
 	xfs_trans_t	*tp,		/* transaction pointer */
 	xfs_ino_t	parent,		/* parent directory inode number */
 	mode_t		mode,		/* bits set to indicate file type */
 	int		okalloc)	/* ok to allocate more space */
 {
 	xfs_buf_t	*agbp;		/* allocation group header buffer */
 	xfs_agnumber_t	agcount;	/* number of ag's in the filesystem */
 	xfs_agnumber_t	agno;		/* current ag number */
 	int		flags;		/* alloc buffer locking flags */
 	xfs_extlen_t	ineed;		/* blocks needed for inode allocation */
 	xfs_extlen_t	longest = 0;	/* longest extent available */
 	xfs_mount_t	*mp;		/* mount point structure */
 	int		needspace;	/* file mode implies space allocated */
 	xfs_perag_t	*pag;		/* per allocation group data */
 	xfs_agnumber_t	pagno;		/* parent (starting) ag number */
 	/*
 	 * Files of these types need at least one block if length > 0
 	 * (and they won't fit in the inode, but that's hard to figure out).
 	 */
 	needspace = S_ISDIR(mode) || S_ISREG(mode) || S_ISLNK(mode);
 	mp = tp->t_mountp;
 	agcount = mp->m_maxagi;
 	if (S_ISDIR(mode))
 		pagno = xfs_ialloc_next_ag(mp);
 	else {
 		pagno = XFS_INO_TO_AGNO(mp, parent);
 		if (pagno >= agcount)
 			pagno = 0;
 	}
 	ASSERT(pagno < agcount);
 	/*
 	 * Loop through allocation groups, looking for one with a little
 	 * free space in it.  Note we don't look for free inodes, exactly.
 	 * Instead, we include whether there is a need to allocate inodes
 	 * to mean that blocks must be allocated for them,
 	 * if none are currently free.
 	 */
 	agno = pagno;
 	flags = XFS_ALLOC_FLAG_TRYLOCK;
 	down_read(&mp->m_peraglock);
 	for (;;) {
 		pag = &mp->m_perag[agno];
 		if (!pag->pagi_init) {
 			if (xfs_ialloc_read_agi(mp, tp, agno, &agbp)) {
 				agbp = NULL;
 				goto nextag;
 			}
 		} else
 			agbp = NULL;
 		if (!pag->pagi_inodeok) {
 			xfs_ialloc_next_ag(mp);
 			goto unlock_nextag;
 		}
 		/*
 		 * Is there enough free space for the file plus a block
 		 * of inodes (if we need to allocate some)?
 		 */
 		ineed = pag->pagi_freecount ? 0 : XFS_IALLOC_BLOCKS(mp);
 		if (ineed && !pag->pagf_init) {
 			if (agbp == NULL &&
 			    xfs_ialloc_read_agi(mp, tp, agno, &agbp)) {
 				agbp = NULL;
 				goto nextag;
 			}
 			(void)xfs_alloc_pagf_init(mp, tp, agno, flags);
 		}
 		if (!ineed || pag->pagf_init) {
 			if (ineed && !(longest = pag->pagf_longest))
 				longest = pag->pagf_flcount > 0;
 			if (!ineed ||
 			    (pag->pagf_freeblks >= needspace + ineed &&
 			     longest >= ineed &&
 			     okalloc)) {
 				if (agbp == NULL &&
 				    xfs_ialloc_read_agi(mp, tp, agno, &agbp)) {
 					agbp = NULL;
 					goto nextag;
 				}
 				up_read(&mp->m_peraglock);
 				return agbp;
 			}
 		}
 unlock_nextag:
 		if (agbp)
 			xfs_trans_brelse(tp, agbp);
 nextag:
 		/*
 		 * No point in iterating over the rest, if we're shutting
 		 * down.
 		 */
 		if (XFS_FORCED_SHUTDOWN(mp)) {
 			up_read(&mp->m_peraglock);
 			return NULL;
 		}
 		agno++;
 		if (agno >= agcount)
 			agno = 0;
 		if (agno == pagno) {
 			if (flags == 0) {
 				up_read(&mp->m_peraglock);
 				return NULL;
 			}
 			flags = 0;
 		}
 	}
 }
 /*
  * Visible inode allocation functions.
  */
 /*
  * Allocate an inode on disk.
  * Mode is used to tell whether the new inode will need space, and whether
  * it is a directory.
  *
  * The arguments IO_agbp and alloc_done are defined to work within
  * the constraint of one allocation per transaction.
  * xfs_dialloc() is designed to be called twice if it has to do an
  * allocation to make more free inodes.  On the first call,
  * IO_agbp should be set to NULL. If an inode is available,
  * i.e., xfs_dialloc() did not need to do an allocation, an inode
  * number is returned.  In this case, IO_agbp would be set to the
  * current ag_buf and alloc_done set to false.
  * If an allocation needed to be done, xfs_dialloc would return
  * the current ag_buf in IO_agbp and set alloc_done to true.
  * The caller should then commit the current transaction, allocate a new
  * transaction, and call xfs_dialloc() again, passing in the previous
  * value of IO_agbp.  IO_agbp should be held across the transactions.
  * Since the agbp is locked across the two calls, the second call is
  * guaranteed to have a free inode available.
  *
  * Once we successfully pick an inode its number is returned and the
  * on-disk data structures are updated.  The inode itself is not read
  * in, since doing so would break ordering constraints with xfs_reclaim.
  */
 int
 xfs_dialloc(
 	xfs_trans_t	*tp,		/* transaction pointer */
 	xfs_ino_t	parent,		/* parent inode (directory) */
 	mode_t		mode,		/* mode bits for new inode */
 	int		okalloc,	/* ok to allocate more space */
 	xfs_buf_t	**IO_agbp,	/* in/out ag header's buffer */
 	boolean_t	*alloc_done,	/* true if we needed to replenish
 					   inode freelist */
 	xfs_ino_t	*inop)		/* inode number allocated */
 {
 	xfs_agnumber_t	agcount;	/* number of allocation groups */
 	xfs_buf_t	*agbp;		/* allocation group header's buffer */
 	xfs_agnumber_t	agno;		/* allocation group number */
 	xfs_agi_t	*agi;		/* allocation group header structure */
 	xfs_btree_cur_t	*cur;		/* inode allocation btree cursor */
 	int		error;		/* error return value */
 	int		i;		/* result code */
 	int		ialloced;	/* inode allocation status */
 	int		noroom = 0;	/* no space for inode blk allocation */
 	xfs_ino_t	ino;		/* fs-relative inode to be returned */
 	/* REFERENCED */
 	int		j;		/* result code */
 	xfs_mount_t	*mp;		/* file system mount structure */
 	int		offset;		/* index of inode in chunk */
 	xfs_agino_t	pagino;		/* parent's a.g. relative inode # */
 	xfs_agnumber_t	pagno;		/* parent's allocation group number */
 	xfs_inobt_rec_incore_t rec;	/* inode allocation record */
 	xfs_agnumber_t	tagno;		/* testing allocation group number */
 	xfs_btree_cur_t	*tcur;		/* temp cursor */
 	xfs_inobt_rec_incore_t trec;	/* temp inode allocation record */
 	if (*IO_agbp == NULL) {
 		/*
 		 * We do not have an agbp, so select an initial allocation
 		 * group for inode allocation.
 		 */
 		agbp = xfs_ialloc_ag_select(tp, parent, mode, okalloc);
 		/*
 		 * Couldn't find an allocation group satisfying the
 		 * criteria, give up.
 		 */
 		if (!agbp) {
 			*inop = NULLFSINO;
 			return 0;
 		}
 		agi = XFS_BUF_TO_AGI(agbp);
 		ASSERT(be32_to_cpu(agi->agi_magicnum) == XFS_AGI_MAGIC);
 	} else {
 		/*
 		 * Continue where we left off before.  In this case, we
 		 * know that the allocation group has free inodes.
 		 */
 		agbp = *IO_agbp;
 		agi = XFS_BUF_TO_AGI(agbp);
 		ASSERT(be32_to_cpu(agi->agi_magicnum) == XFS_AGI_MAGIC);
 		ASSERT(be32_to_cpu(agi->agi_freecount) > 0);
 	}
 	mp = tp->t_mountp;
 	agcount = mp->m_sb.sb_agcount;
 	agno = be32_to_cpu(agi->agi_seqno);
 	tagno = agno;
 	pagno = XFS_INO_TO_AGNO(mp, parent);
 	pagino = XFS_INO_TO_AGINO(mp, parent);
 	/*
 	 * If we have already hit the ceiling of inode blocks then clear
 	 * okalloc so we scan all available agi structures for a free
 	 * inode.
 	 */
 	if (mp->m_maxicount &&
 	    mp->m_sb.sb_icount + XFS_IALLOC_INODES(mp) > mp->m_maxicount) {
 		noroom = 1;
 		okalloc = 0;
 	}
 	/*
 	 * Loop until we find an allocation group that either has free inodes
 	 * or in which we can allocate some inodes.  Iterate through the
 	 * allocation groups upward, wrapping at the end.
 	 */
 	*alloc_done = B_FALSE;
 	while (!agi->agi_freecount) {
 		/*
 		 * Don't do anything if we're not supposed to allocate
 		 * any blocks, just go on to the next ag.
 		 */
 		if (okalloc) {
 			/*
 			 * Try to allocate some new inodes in the allocation
 			 * group.
 			 */
 			if ((error = xfs_ialloc_ag_alloc(tp, agbp, &ialloced))) {
 				xfs_trans_brelse(tp, agbp);
 				if (error == ENOSPC) {
 					*inop = NULLFSINO;
 					return 0;
 				} else
 					return error;
 			}
 			if (ialloced) {
 				/*
 				 * We successfully allocated some inodes, return
 				 * the current context to the caller so that it
 				 * can commit the current transaction and call
 				 * us again where we left off.
 				 */
 				ASSERT(be32_to_cpu(agi->agi_freecount) > 0);
 				*alloc_done = B_TRUE;
 				*IO_agbp = agbp;
 				*inop = NULLFSINO;
 				return 0;
 			}
 		}
 		/*
 		 * If it failed, give up on this ag.
 		 */
 		xfs_trans_brelse(tp, agbp);
 		/*
 		 * Go on to the next ag: get its ag header.
 		 */
 nextag:
 		if (++tagno == agcount)
 			tagno = 0;
 		if (tagno == agno) {
 			*inop = NULLFSINO;
 			return noroom ? ENOSPC : 0;
 		}
 		down_read(&mp->m_peraglock);
 		if (mp->m_perag[tagno].pagi_inodeok == 0) {
 			up_read(&mp->m_peraglock);
 			goto nextag;
 		}
 		error = xfs_ialloc_read_agi(mp, tp, tagno, &agbp);
 		up_read(&mp->m_peraglock);
 		if (error)
 			goto nextag;
 		agi = XFS_BUF_TO_AGI(agbp);
 		ASSERT(be32_to_cpu(agi->agi_magicnum) == XFS_AGI_MAGIC);
 	}
 	/*
 	 * Here with an allocation group that has a free inode.
 	 * Reset agno since we may have chosen a new ag in the
 	 * loop above.
 	 */
 	agno = tagno;
 	*IO_agbp = NULL;
 	cur = xfs_btree_init_cursor(mp, tp, agbp, be32_to_cpu(agi->agi_seqno),
 				    XFS_BTNUM_INO, (xfs_inode_t *)0, 0);
 	/*
 	 * If pagino is 0 (this is the root inode allocation) use newino.
 	 * This must work because we've just allocated some.
 	 */
 	if (!pagino)
 		pagino = be32_to_cpu(agi->agi_newino);
 #ifdef DEBUG
 	if (cur->bc_nlevels == 1) {
 		int	freecount = 0;
 		if ((error = xfs_inobt_lookup_ge(cur, 0, 0, 0, &i)))
 			goto error0;
 		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
 		do {
 			if ((error = xfs_inobt_get_rec(cur, &rec.ir_startino,
 					&rec.ir_freecount, &rec.ir_free, &i)))
 				goto error0;
 			XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
 			freecount += rec.ir_freecount;
 			if ((error = xfs_inobt_increment(cur, 0, &i)))
 				goto error0;
 		} while (i == 1);
 		ASSERT(freecount == be32_to_cpu(agi->agi_freecount) ||
 		       XFS_FORCED_SHUTDOWN(mp));
 	}
 #endif
 	/*
 	 * If in the same a.g. as the parent, try to get near the parent.
 	 */
 	if (pagno == agno) {
 		if ((error = xfs_inobt_lookup_le(cur, pagino, 0, 0, &i)))
 			goto error0;
 		if (i != 0 &&
 		    (error = xfs_inobt_get_rec(cur, &rec.ir_startino,
 			    &rec.ir_freecount, &rec.ir_free, &j)) == 0 &&
 		    j == 1 &&
 		    rec.ir_freecount > 0) {
 			/*
 			 * Found a free inode in the same chunk
 			 * as parent, done.
 			 */
 		}
 		/*
 		 * In the same a.g. as parent, but parent's chunk is full.
 		 */
 		else {
 			int	doneleft;	/* done, to the left */
 			int	doneright;	/* done, to the right */
 			if (error)
 				goto error0;
 			ASSERT(i == 1);
 			ASSERT(j == 1);
 			/*
 			 * Duplicate the cursor, search left & right
 			 * simultaneously.
 			 */
 			if ((error = xfs_btree_dup_cursor(cur, &tcur)))
 				goto error0;
 			/*
 			 * Search left with tcur, back up 1 record.
 			 */
 			if ((error = xfs_inobt_decrement(tcur, 0, &i)))
 				goto error1;
 			doneleft = !i;
 			if (!doneleft) {
 				if ((error = xfs_inobt_get_rec(tcur,
 						&trec.ir_startino,
 						&trec.ir_freecount,
 						&trec.ir_free, &i)))
 					goto error1;
 				XFS_WANT_CORRUPTED_GOTO(i == 1, error1);
 			}
 			/*
 			 * Search right with cur, go forward 1 record.
 			 */
 			if ((error = xfs_inobt_increment(cur, 0, &i)))
 				goto error1;
 			doneright = !i;
 			if (!doneright) {
 				if ((error = xfs_inobt_get_rec(cur,
 						&rec.ir_startino,
 						&rec.ir_freecount,
 						&rec.ir_free, &i)))
 					goto error1;
 				XFS_WANT_CORRUPTED_GOTO(i == 1, error1);
 			}
 			/*
 			 * Loop until we find the closest inode chunk
 			 * with a free one.
 			 */
 			while (!doneleft || !doneright) {
 				int	useleft;  /* using left inode
 						     chunk this time */
 				/*
 				 * Figure out which block is closer,
 				 * if both are valid.
 				 */
 				if (!doneleft && !doneright)
 					useleft =
 						pagino -
 						(trec.ir_startino +
 						 XFS_INODES_PER_CHUNK - 1) <
 						 rec.ir_startino - pagino;
 				else
 					useleft = !doneleft;
 				/*
 				 * If checking the left, does it have
 				 * free inodes?
 				 */
 				if (useleft && trec.ir_freecount) {
 					/*
 					 * Yes, set it up as the chunk to use.
 					 */
 					rec = trec;
 					xfs_btree_del_cursor(cur,
 						XFS_BTREE_NOERROR);
 					cur = tcur;
 					break;
 				}
 				/*
 				 * If checking the right, does it have
 				 * free inodes?
 				 */
 				if (!useleft && rec.ir_freecount) {
 					/*
 					 * Yes, it's already set up.
 					 */
 					xfs_btree_del_cursor(tcur,
 						XFS_BTREE_NOERROR);
 					break;
 				}
 				/*
 				 * If used the left, get another one
 				 * further left.
 				 */
 				if (useleft) {
 					if ((error = xfs_inobt_decrement(tcur, 0,
 							&i)))
 						goto error1;
 					doneleft = !i;
 					if (!doneleft) {
 						if ((error = xfs_inobt_get_rec(
 							    tcur,
 							    &trec.ir_startino,
 							    &trec.ir_freecount,
 							    &trec.ir_free, &i)))
 							goto error1;
 						XFS_WANT_CORRUPTED_GOTO(i == 1,
 							error1);
 					}
 				}
 				/*
 				 * If used the right, get another one
 				 * further right.
 				 */
 				else {
 					if ((error = xfs_inobt_increment(cur, 0,
 							&i)))
 						goto error1;
 					doneright = !i;
 					if (!doneright) {
 						if ((error = xfs_inobt_get_rec(
 							    cur,
 							    &rec.ir_startino,
 							    &rec.ir_freecount,
 							    &rec.ir_free, &i)))
 							goto error1;
 						XFS_WANT_CORRUPTED_GOTO(i == 1,
 							error1);
 					}
 				}
 			}
 			ASSERT(!doneleft || !doneright);
 		}
 	}
 	/*
 	 * In a different a.g. from the parent.
 	 * See if the most recently allocated block has any free.
 	 */
 	else if (be32_to_cpu(agi->agi_newino) != NULLAGINO) {
 		if ((error = xfs_inobt_lookup_eq(cur,
 				be32_to_cpu(agi->agi_newino), 0, 0, &i)))
 			goto error0;
 		if (i == 1 &&
 		    (error = xfs_inobt_get_rec(cur, &rec.ir_startino,
 			    &rec.ir_freecount, &rec.ir_free, &j)) == 0 &&
 		    j == 1 &&
 		    rec.ir_freecount > 0) {
 			/*
 			 * The last chunk allocated in the group still has
 			 * a free inode.
 			 */
 		}
 		/*
 		 * None left in the last group, search the whole a.g.
 		 */
 		else {
 			if (error)
 				goto error0;
 			if ((error = xfs_inobt_lookup_ge(cur, 0, 0, 0, &i)))
 				goto error0;
 			ASSERT(i == 1);
 			for (;;) {
 				if ((error = xfs_inobt_get_rec(cur,
 						&rec.ir_startino,
 						&rec.ir_freecount, &rec.ir_free,
 						&i)))
 					goto error0;
 				XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
 				if (rec.ir_freecount > 0)
 					break;
 				if ((error = xfs_inobt_increment(cur, 0, &i)))
 					goto error0;
 				XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
 			}
 		}
 	}
 	offset = XFS_IALLOC_FIND_FREE(&rec.ir_free);
 	ASSERT(offset >= 0);
 	ASSERT(offset < XFS_INODES_PER_CHUNK);
 	ASSERT((XFS_AGINO_TO_OFFSET(mp, rec.ir_startino) %
 				   XFS_INODES_PER_CHUNK) == 0);
 	ino = XFS_AGINO_TO_INO(mp, agno, rec.ir_startino + offset);
 	XFS_INOBT_CLR_FREE(&rec, offset);
 	rec.ir_freecount--;
 	if ((error = xfs_inobt_update(cur, rec.ir_startino, rec.ir_freecount,
 			rec.ir_free)))
 		goto error0;
 	be32_add(&agi->agi_freecount, -1);
 	xfs_ialloc_log_agi(tp, agbp, XFS_AGI_FREECOUNT);
 	down_read(&mp->m_peraglock);
 	mp->m_perag[tagno].pagi_freecount--;
 	up_read(&mp->m_peraglock);
 #ifdef DEBUG
 	if (cur->bc_nlevels == 1) {
 		int	freecount = 0;
 		if ((error = xfs_inobt_lookup_ge(cur, 0, 0, 0, &i)))
 			goto error0;
 		do {
 			if ((error = xfs_inobt_get_rec(cur, &rec.ir_startino,
 					&rec.ir_freecount, &rec.ir_free, &i)))
 				goto error0;
 			XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
 			freecount += rec.ir_freecount;
 			if ((error = xfs_inobt_increment(cur, 0, &i)))
 				goto error0;
 		} while (i == 1);
 		ASSERT(freecount == be32_to_cpu(agi->agi_freecount) ||
 		       XFS_FORCED_SHUTDOWN(mp));
 	}
 #endif
 	xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
 	xfs_trans_mod_sb(tp, XFS_TRANS_SB_IFREE, -1);
 	*inop = ino;
 	return 0;
 error1:
 	xfs_btree_del_cursor(tcur, XFS_BTREE_ERROR);
 error0:
 	xfs_btree_del_cursor(cur, XFS_BTREE_ERROR);
 	return error;
 }
 /*
  * Free disk inode.  Carefully avoids touching the incore inode, all
  * manipulations incore are the caller's responsibility.
  * The on-disk inode is not changed by this operation, only the
  * btree (free inode mask) is changed.
  */
 int
 xfs_difree(
 	xfs_trans_t	*tp,		/* transaction pointer */
 	xfs_ino_t	inode,		/* inode to be freed */
 	xfs_bmap_free_t	*flist,		/* extents to free */
 	int		*delete,	/* set if inode cluster was deleted */
 	xfs_ino_t	*first_ino)	/* first inode in deleted cluster */
 {
 	/* REFERENCED */
 	xfs_agblock_t	agbno;	/* block number containing inode */
 	xfs_buf_t	*agbp;	/* buffer containing allocation group header */
 	xfs_agino_t	agino;	/* inode number relative to allocation group */
 	xfs_agnumber_t	agno;	/* allocation group number */
 	xfs_agi_t	*agi;	/* allocation group header */
 	xfs_btree_cur_t	*cur;	/* inode btree cursor */
 	int		error;	/* error return value */
 	int		i;	/* result code */
 	int		ilen;	/* inodes in an inode cluster */
 	xfs_mount_t	*mp;	/* mount structure for filesystem */
 	int		off;	/* offset of inode in inode chunk */
 	xfs_inobt_rec_incore_t rec;	/* btree record */
 	mp = tp->t_mountp;
 	/*
 	 * Break up inode number into its components.
 	 */
 	agno = XFS_INO_TO_AGNO(mp, inode);
 	if (agno >= mp->m_sb.sb_agcount)  {
 		cmn_err(CE_WARN,
 			"xfs_difree: agno >= mp->m_sb.sb_agcount (%d >= %d) on %s.  Returning EINVAL.",
 			agno, mp->m_sb.sb_agcount, mp->m_fsname);
 		ASSERT(0);
 		return XFS_ERROR(EINVAL);
 	}
 	agino = XFS_INO_TO_AGINO(mp, inode);
 	if (inode != XFS_AGINO_TO_INO(mp, agno, agino))  {
 		cmn_err(CE_WARN,
 			"xfs_difree: inode != XFS_AGINO_TO_INO() "
 			"(%llu != %llu) on %s.  Returning EINVAL.",
 			(unsigned long long)inode,
 			(unsigned long long)XFS_AGINO_TO_INO(mp, agno, agino),
 			mp->m_fsname);
 		ASSERT(0);
 		return XFS_ERROR(EINVAL);
 	}
 	agbno = XFS_AGINO_TO_AGBNO(mp, agino);
 	if (agbno >= mp->m_sb.sb_agblocks)  {
 		cmn_err(CE_WARN,
 			"xfs_difree: agbno >= mp->m_sb.sb_agblocks (%d >= %d) on %s.  Returning EINVAL.",
 			agbno, mp->m_sb.sb_agblocks, mp->m_fsname);
 		ASSERT(0);
 		return XFS_ERROR(EINVAL);
 	}
 	/*
 	 * Get the allocation group header.
 	 */
 	down_read(&mp->m_peraglock);
 	error = xfs_ialloc_read_agi(mp, tp, agno, &agbp);
 	up_read(&mp->m_peraglock);
 	if (error) {
 		cmn_err(CE_WARN,
 			"xfs_difree: xfs_ialloc_read_agi() returned an error %d on %s.  Returning error.",
 			error, mp->m_fsname);
 		return error;
 	}
 	agi = XFS_BUF_TO_AGI(agbp);
 	ASSERT(be32_to_cpu(agi->agi_magicnum) == XFS_AGI_MAGIC);
 	ASSERT(agbno < be32_to_cpu(agi->agi_length));
 	/*
 	 * Initialize the cursor.
 	 */
 	cur = xfs_btree_init_cursor(mp, tp, agbp, agno, XFS_BTNUM_INO,
 		(xfs_inode_t *)0, 0);
 #ifdef DEBUG
 	if (cur->bc_nlevels == 1) {
 		int freecount = 0;
 		if ((error = xfs_inobt_lookup_ge(cur, 0, 0, 0, &i)))
 			goto error0;
 		do {
 			if ((error = xfs_inobt_get_rec(cur, &rec.ir_startino,
 					&rec.ir_freecount, &rec.ir_free, &i)))
 				goto error0;
 			if (i) {
 				freecount += rec.ir_freecount;
 				if ((error = xfs_inobt_increment(cur, 0, &i)))
 					goto error0;
 			}
 		} while (i == 1);
 		ASSERT(freecount == be32_to_cpu(agi->agi_freecount) ||
 		       XFS_FORCED_SHUTDOWN(mp));
 	}
 #endif
 	/*
 	 * Look for the entry describing this inode.
 	 */
 	if ((error = xfs_inobt_lookup_le(cur, agino, 0, 0, &i))) {
 		cmn_err(CE_WARN,
 			"xfs_difree: xfs_inobt_lookup_le returned()  an error %d on %s.  Returning error.",
 			error, mp->m_fsname);
 		goto error0;
 	}
 	XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
 	if ((error = xfs_inobt_get_rec(cur, &rec.ir_startino, &rec.ir_freecount,
 			&rec.ir_free, &i))) {
 		cmn_err(CE_WARN,
 			"xfs_difree: xfs_inobt_get_rec()  returned an error %d on %s.  Returning error.",
 			error, mp->m_fsname);
 		goto error0;
 	}
 	XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
 	/*
 	 * Get the offset in the inode chunk.
 	 */
 	off = agino - rec.ir_startino;
 	ASSERT(off >= 0 && off < XFS_INODES_PER_CHUNK);
 	ASSERT(!XFS_INOBT_IS_FREE(&rec, off));
 	/*
 	 * Mark the inode free & increment the count.
 	 */
 	XFS_INOBT_SET_FREE(&rec, off);
 	rec.ir_freecount++;
 	/*
 	 * When an inode cluster is free, it becomes eligible for removal
 	 */
 	if ((mp->m_flags & XFS_MOUNT_IDELETE) &&
 	    (rec.ir_freecount == XFS_IALLOC_INODES(mp))) {
 		*delete = 1;
 		*first_ino = XFS_AGINO_TO_INO(mp, agno, rec.ir_startino);
 		/*
 		 * Remove the inode cluster from the AGI B+Tree, adjust the
 		 * AGI and Superblock inode counts, and mark the disk space
 		 * to be freed when the transaction is committed.
 		 */
 		ilen = XFS_IALLOC_INODES(mp);
 		be32_add(&agi->agi_count, -ilen);
 		be32_add(&agi->agi_freecount, -(ilen - 1));
 		xfs_ialloc_log_agi(tp, agbp, XFS_AGI_COUNT | XFS_AGI_FREECOUNT);
 		down_read(&mp->m_peraglock);
 		mp->m_perag[agno].pagi_freecount -= ilen - 1;
 		up_read(&mp->m_peraglock);
 		xfs_trans_mod_sb(tp, XFS_TRANS_SB_ICOUNT, -ilen);
 		xfs_trans_mod_sb(tp, XFS_TRANS_SB_IFREE, -(ilen - 1));
 		if ((error = xfs_inobt_delete(cur, &i))) {
 			cmn_err(CE_WARN, "xfs_difree: xfs_inobt_delete returned an error %d on %s.\n",
 				error, mp->m_fsname);
 			goto error0;
 		}
 		xfs_bmap_add_free(XFS_AGB_TO_FSB(mp,
 				agno, XFS_INO_TO_AGBNO(mp,rec.ir_startino)),
 				XFS_IALLOC_BLOCKS(mp), flist, mp);
 	} else {
 		*delete = 0;
 		if ((error = xfs_inobt_update(cur, rec.ir_startino, rec.ir_freecount, rec.ir_free))) {
 			cmn_err(CE_WARN,
 				"xfs_difree: xfs_inobt_update()  returned an error %d on %s.  Returning error.",
 				error, mp->m_fsname);
 			goto error0;
 		}
 		/*
 		 * Change the inode free counts and log the ag/sb changes.
 		 */
 		be32_add(&agi->agi_freecount, 1);
 		xfs_ialloc_log_agi(tp, agbp, XFS_AGI_FREECOUNT);
 		down_read(&mp->m_peraglock);
 		mp->m_perag[agno].pagi_freecount++;
 		up_read(&mp->m_peraglock);
 		xfs_trans_mod_sb(tp, XFS_TRANS_SB_IFREE, 1);
 	}
 #ifdef DEBUG
 	if (cur->bc_nlevels == 1) {
 		int freecount = 0;
 		if ((error = xfs_inobt_lookup_ge(cur, 0, 0, 0, &i)))
 			goto error0;
 		do {
 			if ((error = xfs_inobt_get_rec(cur,
 					&rec.ir_startino,
 					&rec.ir_freecount,
 					&rec.ir_free, &i)))
 				goto error0;
 			if (i) {
 				freecount += rec.ir_freecount;
 				if ((error = xfs_inobt_increment(cur, 0, &i)))
 					goto error0;
 			}
 		} while (i == 1);
 		ASSERT(freecount == be32_to_cpu(agi->agi_freecount) ||
 		       XFS_FORCED_SHUTDOWN(mp));
 	}
 #endif
 	xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
 	return 0;
 error0:
 	xfs_btree_del_cursor(cur, XFS_BTREE_ERROR);
 	return error;
 }
 /*
  * Return the location of the inode in bno/off, for mapping it into a buffer.
  */
 /*ARGSUSED*/
 int
 xfs_dilocate(
 	xfs_mount_t	*mp,	/* file system mount structure */
 	xfs_trans_t	*tp,	/* transaction pointer */
 	xfs_ino_t	ino,	/* inode to locate */
 	xfs_fsblock_t	*bno,	/* output: block containing inode */
 	int		*len,	/* output: num blocks in inode cluster */
 	int		*off,	/* output: index in block of inode */
 	uint		flags)	/* flags concerning inode lookup */
 {
 	xfs_agblock_t	agbno;	/* block number of inode in the alloc group */
 	xfs_buf_t	*agbp;	/* agi buffer */
 	xfs_agino_t	agino;	/* inode number within alloc group */
 	xfs_agnumber_t	agno;	/* allocation group number */
 	int		blks_per_cluster; /* num blocks per inode cluster */
 	xfs_agblock_t	chunk_agbno;	/* first block in inode chunk */
 	xfs_agino_t	chunk_agino;	/* first agino in inode chunk */
 	__int32_t	chunk_cnt;	/* count of free inodes in chunk */
 	xfs_inofree_t	chunk_free;	/* mask of free inodes in chunk */
 	xfs_agblock_t	cluster_agbno;	/* first block in inode cluster */
 	xfs_btree_cur_t	*cur;	/* inode btree cursor */
 	int		error;	/* error code */
 	int		i;	/* temp state */
 	int		offset;	/* index of inode in its buffer */
 	int		offset_agbno;	/* blks from chunk start to inode */
 	ASSERT(ino != NULLFSINO);
 	/*
 	 * Split up the inode number into its parts.
 	 */
 	agno = XFS_INO_TO_AGNO(mp, ino);
 	agino = XFS_INO_TO_AGINO(mp, ino);
 	agbno = XFS_AGINO_TO_AGBNO(mp, agino);
 	if (agno >= mp->m_sb.sb_agcount || agbno >= mp->m_sb.sb_agblocks ||
 	    ino != XFS_AGINO_TO_INO(mp, agno, agino)) {
 #ifdef DEBUG
 		/* no diagnostics for bulkstat, ino comes from userspace */
 		if (flags & XFS_IMAP_BULKSTAT)
 			return XFS_ERROR(EINVAL);
 		if (agno >= mp->m_sb.sb_agcount) {
 			xfs_fs_cmn_err(CE_ALERT, mp,
 					"xfs_dilocate: agno (%d) >= "
 					"mp->m_sb.sb_agcount (%d)",
 					agno,  mp->m_sb.sb_agcount);
 		}
 		if (agbno >= mp->m_sb.sb_agblocks) {
 			xfs_fs_cmn_err(CE_ALERT, mp,
 					"xfs_dilocate: agbno (0x%llx) >= "
 					"mp->m_sb.sb_agblocks (0x%lx)",
 					(unsigned long long) agbno,
 					(unsigned long) mp->m_sb.sb_agblocks);
 		}
 		if (ino != XFS_AGINO_TO_INO(mp, agno, agino)) {
 			xfs_fs_cmn_err(CE_ALERT, mp,
 					"xfs_dilocate: ino (0x%llx) != "
 					"XFS_AGINO_TO_INO(mp, agno, agino) "
 					"(0x%llx)",
 					ino, XFS_AGINO_TO_INO(mp, agno, agino));
 		}
+		xfs_stack_trace();
 #endif /* DEBUG */
 		return XFS_ERROR(EINVAL);
 	}
 	if ((mp->m_sb.sb_blocksize >= XFS_INODE_CLUSTER_SIZE(mp)) ||
 	    !(flags & XFS_IMAP_LOOKUP)) {
 		offset = XFS_INO_TO_OFFSET(mp, ino);
 		ASSERT(offset < mp->m_sb.sb_inopblock);
 		*bno = XFS_AGB_TO_FSB(mp, agno, agbno);
 		*off = offset;
 		*len = 1;
 		return 0;
 	}
 	blks_per_cluster = XFS_INODE_CLUSTER_SIZE(mp) >> mp->m_sb.sb_blocklog;
 	if (*bno != NULLFSBLOCK) {
 		offset = XFS_INO_TO_OFFSET(mp, ino);
 		ASSERT(offset < mp->m_sb.sb_inopblock);
 		cluster_agbno = XFS_FSB_TO_AGBNO(mp, *bno);
 		*off = ((agbno - cluster_agbno) * mp->m_sb.sb_inopblock) +
 			offset;
 		*len = blks_per_cluster;
 		return 0;
 	}
 	if (mp->m_inoalign_mask) {
 		offset_agbno = agbno & mp->m_inoalign_mask;
 		chunk_agbno = agbno - offset_agbno;
 	} else {
 		down_read(&mp->m_peraglock);
 		error = xfs_ialloc_read_agi(mp, tp, agno, &agbp);
 		up_read(&mp->m_peraglock);
 		if (error) {
 #ifdef DEBUG
 			xfs_fs_cmn_err(CE_ALERT, mp, "xfs_dilocate: "
 					"xfs_ialloc_read_agi() returned "
 					"error %d, agno %d",
 					error, agno);
 #endif /* DEBUG */
 			return error;
 		}
 		cur = xfs_btree_init_cursor(mp, tp, agbp, agno, XFS_BTNUM_INO,
 			(xfs_inode_t *)0, 0);
 		if ((error = xfs_inobt_lookup_le(cur, agino, 0, 0, &i))) {
 #ifdef DEBUG
 			xfs_fs_cmn_err(CE_ALERT, mp, "xfs_dilocate: "
 					"xfs_inobt_lookup_le() failed");
 #endif /* DEBUG */
 			goto error0;
 		}
 		if ((error = xfs_inobt_get_rec(cur, &chunk_agino, &chunk_cnt,
 				&chunk_free, &i))) {
 #ifdef DEBUG
 			xfs_fs_cmn_err(CE_ALERT, mp, "xfs_dilocate: "
 					"xfs_inobt_get_rec() failed");
 #endif /* DEBUG */
 			goto error0;
 		}
 		if (i == 0) {
 #ifdef DEBUG
 			xfs_fs_cmn_err(CE_ALERT, mp, "xfs_dilocate: "
 					"xfs_inobt_get_rec() failed");
 #endif /* DEBUG */
 			error = XFS_ERROR(EINVAL);
 		}
 		xfs_trans_brelse(tp, agbp);
 		xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
 		if (error)
 			return error;
 		chunk_agbno = XFS_AGINO_TO_AGBNO(mp, chunk_agino);
 		offset_agbno = agbno - chunk_agbno;
 	}
 	ASSERT(agbno >= chunk_agbno);
 	cluster_agbno = chunk_agbno +
 		((offset_agbno / blks_per_cluster) * blks_per_cluster);
 	offset = ((agbno - cluster_agbno) * mp->m_sb.sb_inopblock) +
 		XFS_INO_TO_OFFSET(mp, ino);
 	*bno = XFS_AGB_TO_FSB(mp, agno, cluster_agbno);
 	*off = offset;
 	*len = blks_per_cluster;
 	return 0;
 error0:
 	xfs_trans_brelse(tp, agbp);
 	xfs_btree_del_cursor(cur, XFS_BTREE_ERROR);
 	return error;
 }
 /*
  * Compute and fill in value of m_in_maxlevels.
  */
 void
 xfs_ialloc_compute_maxlevels(
 	xfs_mount_t	*mp)		/* file system mount structure */
 {
 	int		level;
 	uint		maxblocks;
 	uint		maxleafents;
 	int		minleafrecs;
 	int		minnoderecs;
 	maxleafents = (1LL << XFS_INO_AGINO_BITS(mp)) >>
 		XFS_INODES_PER_CHUNK_LOG;
 	minleafrecs = mp->m_alloc_mnr[0];
 	minnoderecs = mp->m_alloc_mnr[1];
 	maxblocks = (maxleafents + minleafrecs - 1) / minleafrecs;
 	for (level = 1; maxblocks > 1; level++)
 		maxblocks = (maxblocks + minnoderecs - 1) / minnoderecs;
 	mp->m_in_maxlevels = level;
 }
 /*
  * Log specified fields for the ag hdr (inode section)
  */
 void
 xfs_ialloc_log_agi(
 	xfs_trans_t	*tp,		/* transaction pointer */
 	xfs_buf_t	*bp,		/* allocation group header buffer */
 	int		fields)		/* bitmask of fields to log */
 {
 	int			first;		/* first byte number */
 	int			last;		/* last byte number */
 	static const short	offsets[] = {	/* field starting offsets */
 					/* keep in sync with bit definitions */
 		offsetof(xfs_agi_t, agi_magicnum),
 		offsetof(xfs_agi_t, agi_versionnum),
 		offsetof(xfs_agi_t, agi_seqno),
 		offsetof(xfs_agi_t, agi_length),
 		offsetof(xfs_agi_t, agi_count),
 		offsetof(xfs_agi_t, agi_root),
 		offsetof(xfs_agi_t, agi_level),
 		offsetof(xfs_agi_t, agi_freecount),
 		offsetof(xfs_agi_t, agi_newino),
 		offsetof(xfs_agi_t, agi_dirino),
 		offsetof(xfs_agi_t, agi_unlinked),
 		sizeof(xfs_agi_t)
 	};
 #ifdef DEBUG
 	xfs_agi_t		*agi;	/* allocation group header */
 	agi = XFS_BUF_TO_AGI(bp);
 	ASSERT(be32_to_cpu(agi->agi_magicnum) == XFS_AGI_MAGIC);
 #endif
 	/*
 	 * Compute byte offsets for the first and last fields.
 	 */
 	xfs_btree_offsets(fields, offsets, XFS_AGI_NUM_BITS, &first, &last);
 	/*
 	 * Log the allocation group inode header buffer.
 	 */
 	xfs_trans_log_buf(tp, bp, first, last);
 }
 /*
  * Read in the allocation group header (inode allocation section)
  */
 int
 xfs_ialloc_read_agi(
 	xfs_mount_t	*mp,		/* file system mount structure */
 	xfs_trans_t	*tp,		/* transaction pointer */
 	xfs_agnumber_t	agno,		/* allocation group number */
 	xfs_buf_t	**bpp)		/* allocation group hdr buf */
 {
 	xfs_agi_t	*agi;		/* allocation group header */
 	int		agi_ok;		/* agi is consistent */
 	xfs_buf_t	*bp;		/* allocation group hdr buf */
 	xfs_perag_t	*pag;		/* per allocation group data */
 	int		error;
 	ASSERT(agno != NULLAGNUMBER);
 	error = xfs_trans_read_buf(
 			mp, tp, mp->m_ddev_targp,
 			XFS_AG_DADDR(mp, agno, XFS_AGI_DADDR(mp)),
 			XFS_FSS_TO_BB(mp, 1), 0, &bp);
 	if (error)
 		return error;
 	ASSERT(bp && !XFS_BUF_GETERROR(bp));
 	/*
 	 * Validate the magic number of the agi block.
 	 */
 	agi = XFS_BUF_TO_AGI(bp);
 	agi_ok =
 		be32_to_cpu(agi->agi_magicnum) == XFS_AGI_MAGIC &&
 		XFS_AGI_GOOD_VERSION(be32_to_cpu(agi->agi_versionnum));
 	if (unlikely(XFS_TEST_ERROR(!agi_ok, mp, XFS_ERRTAG_IALLOC_READ_AGI,
 			XFS_RANDOM_IALLOC_READ_AGI))) {
 		XFS_CORRUPTION_ERROR("xfs_ialloc_read_agi", XFS_ERRLEVEL_LOW,
 				     mp, agi);
 		xfs_trans_brelse(tp, bp);
 		return XFS_ERROR(EFSCORRUPTED);
 	}
 	pag = &mp->m_perag[agno];
 	if (!pag->pagi_init) {
 		pag->pagi_freecount = be32_to_cpu(agi->agi_freecount);
 		pag->pagi_init = 1;
 	} else {
 		/*
 		 * It's possible for these to be out of sync if
 		 * we are in the middle of a forced shutdown.
 		 */
 		ASSERT(pag->pagi_freecount == be32_to_cpu(agi->agi_freecount) ||
 			XFS_FORCED_SHUTDOWN(mp));
 	}
 #ifdef DEBUG
 	{
 		int	i;
 		for (i = 0; i < XFS_AGI_UNLINKED_BUCKETS; i++)
 			ASSERT(agi->agi_unlinked[i]);
 	}
 #endif
 	XFS_BUF_SET_VTYPE_REF(bp, B_FS_AGI, XFS_AGI_REF);
 	*bpp = bp;
 	return 0;
 }

fs/xfs/xfs_iget.c

Diff comments View file @ 745b1f4

1	/*	1	/*
2	* Copyright (c) 2000-2005 Silicon Graphics, Inc.	2	* Copyright (c) 2000-2005 Silicon Graphics, Inc.
3	* All Rights Reserved.	3	* All Rights Reserved.
4	*	4	*
5	* This program is free software; you can redistribute it and/or	5	* This program is free software; you can redistribute it and/or
6	* modify it under the terms of the GNU General Public License as	6	* modify it under the terms of the GNU General Public License as
7	* published by the Free Software Foundation.	7	* published by the Free Software Foundation.
8	*	8	*
9	* This program is distributed in the hope that it would be useful,	9	* This program is distributed in the hope that it would be useful,
10	* but WITHOUT ANY WARRANTY; without even the implied warranty of	10	* but WITHOUT ANY WARRANTY; without even the implied warranty of
11	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the	11	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
12	* GNU General Public License for more details.	12	* GNU General Public License for more details.
13	*	13	*
14	* You should have received a copy of the GNU General Public License	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,	15	* along with this program; if not, write the Free Software Foundation,
16	* Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA	16	* Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
17	*/	17	*/
18	#include "xfs.h"	18	#include "xfs.h"
19	#include "xfs_fs.h"	19	#include "xfs_fs.h"
20	#include "xfs_types.h"	20	#include "xfs_types.h"
21	#include "xfs_bit.h"	21	#include "xfs_bit.h"
22	#include "xfs_log.h"	22	#include "xfs_log.h"
23	#include "xfs_inum.h"	23	#include "xfs_inum.h"
24	#include "xfs_trans.h"	24	#include "xfs_trans.h"
25	#include "xfs_sb.h"	25	#include "xfs_sb.h"
26	#include "xfs_ag.h"	26	#include "xfs_ag.h"
27	#include "xfs_dir2.h"	27	#include "xfs_dir2.h"
28	#include "xfs_dmapi.h"	28	#include "xfs_dmapi.h"
29	#include "xfs_mount.h"	29	#include "xfs_mount.h"
30	#include "xfs_bmap_btree.h"	30	#include "xfs_bmap_btree.h"
31	#include "xfs_alloc_btree.h"	31	#include "xfs_alloc_btree.h"
32	#include "xfs_ialloc_btree.h"	32	#include "xfs_ialloc_btree.h"
33	#include "xfs_dir2_sf.h"	33	#include "xfs_dir2_sf.h"
34	#include "xfs_attr_sf.h"	34	#include "xfs_attr_sf.h"
35	#include "xfs_dinode.h"	35	#include "xfs_dinode.h"
36	#include "xfs_inode.h"	36	#include "xfs_inode.h"
37	#include "xfs_btree.h"	37	#include "xfs_btree.h"
38	#include "xfs_ialloc.h"	38	#include "xfs_ialloc.h"
39	#include "xfs_quota.h"	39	#include "xfs_quota.h"
40	#include "xfs_utils.h"	40	#include "xfs_utils.h"
41		41
42	/*	42	/*
43	* Initialize the inode hash table for the newly mounted file system.	43	* Initialize the inode hash table for the newly mounted file system.
44	* Choose an initial table size based on user specified value, else	44	* Choose an initial table size based on user specified value, else
45	* use a simple algorithm using the maximum number of inodes as an	45	* use a simple algorithm using the maximum number of inodes as an
46	* indicator for table size, and clamp it between one and some large	46	* indicator for table size, and clamp it between one and some large
47	* number of pages.	47	* number of pages.
48	*/	48	*/
49	void	49	void
50	xfs_ihash_init(xfs_mount_t *mp)	50	xfs_ihash_init(xfs_mount_t *mp)
51	{	51	{
52	__uint64_t icount;	52	__uint64_t icount;
53	uint i, flags = KM_SLEEP \| KM_MAYFAIL;	53	uint i, flags = KM_SLEEP \| KM_MAYFAIL;
54		54
55	if (!mp->m_ihsize) {	55	if (!mp->m_ihsize) {
56	icount = mp->m_maxicount ? mp->m_maxicount :	56	icount = mp->m_maxicount ? mp->m_maxicount :
57	(mp->m_sb.sb_dblocks << mp->m_sb.sb_inopblog);	57	(mp->m_sb.sb_dblocks << mp->m_sb.sb_inopblog);
58	mp->m_ihsize = 1 << max_t(uint, 8,	58	mp->m_ihsize = 1 << max_t(uint, 8,
59	(xfs_highbit64(icount) + 1) / 2);	59	(xfs_highbit64(icount) + 1) / 2);
60	mp->m_ihsize = min_t(uint, mp->m_ihsize,	60	mp->m_ihsize = min_t(uint, mp->m_ihsize,
61	(64 * NBPP) / sizeof(xfs_ihash_t));	61	(64 * NBPP) / sizeof(xfs_ihash_t));
62	}	62	}
63		63
64	while (!(mp->m_ihash = (xfs_ihash_t )kmem_zalloc(mp->m_ihsize	64	while (!(mp->m_ihash = (xfs_ihash_t )kmem_zalloc(mp->m_ihsize
65	sizeof(xfs_ihash_t), flags))) {	65	sizeof(xfs_ihash_t), flags))) {
66	if ((mp->m_ihsize >>= 1) <= NBPP)	66	if ((mp->m_ihsize >>= 1) <= NBPP)
67	flags = KM_SLEEP;	67	flags = KM_SLEEP;
68	}	68	}
69	for (i = 0; i < mp->m_ihsize; i++) {	69	for (i = 0; i < mp->m_ihsize; i++) {
70	rwlock_init(&(mp->m_ihash[i].ih_lock));	70	rwlock_init(&(mp->m_ihash[i].ih_lock));
71	}	71	}
72	}	72	}
73		73
74	/*	74	/*
75	* Free up structures allocated by xfs_ihash_init, at unmount time.	75	* Free up structures allocated by xfs_ihash_init, at unmount time.
76	*/	76	*/
77	void	77	void
78	xfs_ihash_free(xfs_mount_t *mp)	78	xfs_ihash_free(xfs_mount_t *mp)
79	{	79	{
80	kmem_free(mp->m_ihash, mp->m_ihsize*sizeof(xfs_ihash_t));	80	kmem_free(mp->m_ihash, mp->m_ihsize*sizeof(xfs_ihash_t));
81	mp->m_ihash = NULL;	81	mp->m_ihash = NULL;
82	}	82	}
83		83
84	/*	84	/*
85	* Initialize the inode cluster hash table for the newly mounted file system.	85	* Initialize the inode cluster hash table for the newly mounted file system.
86	* Its size is derived from the ihash table size.	86	* Its size is derived from the ihash table size.
87	*/	87	*/
88	void	88	void
89	xfs_chash_init(xfs_mount_t *mp)	89	xfs_chash_init(xfs_mount_t *mp)
90	{	90	{
91	uint i;	91	uint i;
92		92
93	mp->m_chsize = max_t(uint, 1, mp->m_ihsize /	93	mp->m_chsize = max_t(uint, 1, mp->m_ihsize /
94	(XFS_INODE_CLUSTER_SIZE(mp) >> mp->m_sb.sb_inodelog));	94	(XFS_INODE_CLUSTER_SIZE(mp) >> mp->m_sb.sb_inodelog));
95	mp->m_chsize = min_t(uint, mp->m_chsize, mp->m_ihsize);	95	mp->m_chsize = min_t(uint, mp->m_chsize, mp->m_ihsize);
96	mp->m_chash = (xfs_chash_t *)kmem_zalloc(mp->m_chsize	96	mp->m_chash = (xfs_chash_t *)kmem_zalloc(mp->m_chsize
97	* sizeof(xfs_chash_t),	97	* sizeof(xfs_chash_t),
98	KM_SLEEP);	98	KM_SLEEP);
99	for (i = 0; i < mp->m_chsize; i++) {	99	for (i = 0; i < mp->m_chsize; i++) {
100	spinlock_init(&mp->m_chash[i].ch_lock,"xfshash");	100	spinlock_init(&mp->m_chash[i].ch_lock,"xfshash");
101	}	101	}
102	}	102	}
103		103
104	/*	104	/*
105	* Free up structures allocated by xfs_chash_init, at unmount time.	105	* Free up structures allocated by xfs_chash_init, at unmount time.
106	*/	106	*/
107	void	107	void
108	xfs_chash_free(xfs_mount_t *mp)	108	xfs_chash_free(xfs_mount_t *mp)
109	{	109	{
110	int i;	110	int i;
111		111
112	for (i = 0; i < mp->m_chsize; i++) {	112	for (i = 0; i < mp->m_chsize; i++) {
113	spinlock_destroy(&mp->m_chash[i].ch_lock);	113	spinlock_destroy(&mp->m_chash[i].ch_lock);
114	}	114	}
115		115
116	kmem_free(mp->m_chash, mp->m_chsize*sizeof(xfs_chash_t));	116	kmem_free(mp->m_chash, mp->m_chsize*sizeof(xfs_chash_t));
117	mp->m_chash = NULL;	117	mp->m_chash = NULL;
118	}	118	}
119		119
120	/*	120	/*
121	* Try to move an inode to the front of its hash list if possible	121	* Try to move an inode to the front of its hash list if possible
122	* (and if its not there already). Called right after obtaining	122	* (and if its not there already). Called right after obtaining
123	* the list version number and then dropping the read_lock on the	123	* the list version number and then dropping the read_lock on the
124	* hash list in question (which is done right after looking up the	124	* hash list in question (which is done right after looking up the
125	* inode in question...).	125	* inode in question...).
126	*/	126	*/
127	STATIC void	127	STATIC void
128	xfs_ihash_promote(	128	xfs_ihash_promote(
129	xfs_ihash_t *ih,	129	xfs_ihash_t *ih,
130	xfs_inode_t *ip,	130	xfs_inode_t *ip,
131	ulong version)	131	ulong version)
132	{	132	{
133	xfs_inode_t *iq;	133	xfs_inode_t *iq;
134		134
135	if ((ip->i_prevp != &ih->ih_next) && write_trylock(&ih->ih_lock)) {	135	if ((ip->i_prevp != &ih->ih_next) && write_trylock(&ih->ih_lock)) {
136	if (likely(version == ih->ih_version)) {	136	if (likely(version == ih->ih_version)) {
137	/* remove from list */	137	/* remove from list */
138	if ((iq = ip->i_next)) {	138	if ((iq = ip->i_next)) {
139	iq->i_prevp = ip->i_prevp;	139	iq->i_prevp = ip->i_prevp;
140	}	140	}
141	*ip->i_prevp = iq;	141	*ip->i_prevp = iq;
142		142
143	/* insert at list head */	143	/* insert at list head */
144	iq = ih->ih_next;	144	iq = ih->ih_next;
145	iq->i_prevp = &ip->i_next;	145	iq->i_prevp = &ip->i_next;
146	ip->i_next = iq;	146	ip->i_next = iq;
147	ip->i_prevp = &ih->ih_next;	147	ip->i_prevp = &ih->ih_next;
148	ih->ih_next = ip;	148	ih->ih_next = ip;
149	}	149	}
150	write_unlock(&ih->ih_lock);	150	write_unlock(&ih->ih_lock);
151	}	151	}
152	}	152	}
153		153
154	/*	154	/*
155	* Look up an inode by number in the given file system.	155	* Look up an inode by number in the given file system.
156	* The inode is looked up in the hash table for the file system	156	* The inode is looked up in the hash table for the file system
157	* represented by the mount point parameter mp. Each bucket of	157	* represented by the mount point parameter mp. Each bucket of
158	* the hash table is guarded by an individual semaphore.	158	* the hash table is guarded by an individual semaphore.
159	*	159	*
160	* If the inode is found in the hash table, its corresponding vnode	160	* If the inode is found in the hash table, its corresponding vnode
161	* is obtained with a call to vn_get(). This call takes care of	161	* is obtained with a call to vn_get(). This call takes care of
162	* coordination with the reclamation of the inode and vnode. Note	162	* coordination with the reclamation of the inode and vnode. Note
163	* that the vmap structure is filled in while holding the hash lock.	163	* that the vmap structure is filled in while holding the hash lock.
164	* This gives us the state of the inode/vnode when we found it and	164	* This gives us the state of the inode/vnode when we found it and
165	* is used for coordination in vn_get().	165	* is used for coordination in vn_get().
166	*	166	*
167	* If it is not in core, read it in from the file system's device and	167	* If it is not in core, read it in from the file system's device and
168	* add the inode into the hash table.	168	* add the inode into the hash table.
169	*	169	*
170	* The inode is locked according to the value of the lock_flags parameter.	170	* The inode is locked according to the value of the lock_flags parameter.
171	* This flag parameter indicates how and if the inode's IO lock and inode lock	171	* This flag parameter indicates how and if the inode's IO lock and inode lock
172	* should be taken.	172	* should be taken.
173	*	173	*
174	* mp -- the mount point structure for the current file system. It points	174	* mp -- the mount point structure for the current file system. It points
175	* to the inode hash table.	175	* to the inode hash table.
176	* tp -- a pointer to the current transaction if there is one. This is	176	* tp -- a pointer to the current transaction if there is one. This is
177	* simply passed through to the xfs_iread() call.	177	* simply passed through to the xfs_iread() call.
178	* ino -- the number of the inode desired. This is the unique identifier	178	* ino -- the number of the inode desired. This is the unique identifier
179	* within the file system for the inode being requested.	179	* within the file system for the inode being requested.
180	* lock_flags -- flags indicating how to lock the inode. See the comment	180	* lock_flags -- flags indicating how to lock the inode. See the comment
181	* for xfs_ilock() for a list of valid values.	181	* for xfs_ilock() for a list of valid values.
182	* bno -- the block number starting the buffer containing the inode,	182	* bno -- the block number starting the buffer containing the inode,
183	* if known (as by bulkstat), else 0.	183	* if known (as by bulkstat), else 0.
184	*/	184	*/
185	STATIC int	185	STATIC int
186	xfs_iget_core(	186	xfs_iget_core(
187	bhv_vnode_t *vp,	187	bhv_vnode_t *vp,
188	xfs_mount_t *mp,	188	xfs_mount_t *mp,
189	xfs_trans_t *tp,	189	xfs_trans_t *tp,
190	xfs_ino_t ino,	190	xfs_ino_t ino,
191	uint flags,	191	uint flags,
192	uint lock_flags,	192	uint lock_flags,
193	xfs_inode_t **ipp,	193	xfs_inode_t **ipp,
194	xfs_daddr_t bno)	194	xfs_daddr_t bno)
195	{	195	{
196	xfs_ihash_t *ih;	196	xfs_ihash_t *ih;
197	xfs_inode_t *ip;	197	xfs_inode_t *ip;
198	xfs_inode_t *iq;	198	xfs_inode_t *iq;
199	bhv_vnode_t *inode_vp;	199	bhv_vnode_t *inode_vp;
200	ulong version;	200	ulong version;
201	int error;	201	int error;
202	/* REFERENCED */	202	/* REFERENCED */
203	xfs_chash_t *ch;	203	xfs_chash_t *ch;
204	xfs_chashlist_t chl, chlnew;	204	xfs_chashlist_t chl, chlnew;
205	SPLDECL(s);	205	SPLDECL(s);
206		206
207		207
208	ih = XFS_IHASH(mp, ino);	208	ih = XFS_IHASH(mp, ino);
209		209
210	again:	210	again:
211	read_lock(&ih->ih_lock);	211	read_lock(&ih->ih_lock);
212		212
213	for (ip = ih->ih_next; ip != NULL; ip = ip->i_next) {	213	for (ip = ih->ih_next; ip != NULL; ip = ip->i_next) {
214	if (ip->i_ino == ino) {	214	if (ip->i_ino == ino) {
215	/*	215	/*
216	* If INEW is set this inode is being set up	216	* If INEW is set this inode is being set up
217	* we need to pause and try again.	217	* we need to pause and try again.
218	*/	218	*/
219	if (ip->i_flags & XFS_INEW) {	219	if (ip->i_flags & XFS_INEW) {
220	read_unlock(&ih->ih_lock);	220	read_unlock(&ih->ih_lock);
221	delay(1);	221	delay(1);
222	XFS_STATS_INC(xs_ig_frecycle);	222	XFS_STATS_INC(xs_ig_frecycle);
223		223
224	goto again;	224	goto again;
225	}	225	}
226		226
227	inode_vp = XFS_ITOV_NULL(ip);	227	inode_vp = XFS_ITOV_NULL(ip);
228	if (inode_vp == NULL) {	228	if (inode_vp == NULL) {
229	/*	229	/*
230	* If IRECLAIM is set this inode is	230	* If IRECLAIM is set this inode is
231	* on its way out of the system,	231	* on its way out of the system,
232	* we need to pause and try again.	232	* we need to pause and try again.
233	*/	233	*/
234	if (ip->i_flags & XFS_IRECLAIM) {	234	if (ip->i_flags & XFS_IRECLAIM) {
235	read_unlock(&ih->ih_lock);	235	read_unlock(&ih->ih_lock);
236	delay(1);	236	delay(1);
237	XFS_STATS_INC(xs_ig_frecycle);	237	XFS_STATS_INC(xs_ig_frecycle);
238		238
239	goto again;	239	goto again;
240	}	240	}
241		241
242	vn_trace_exit(vp, "xfs_iget.alloc",	242	vn_trace_exit(vp, "xfs_iget.alloc",
243	(inst_t *)__return_address);	243	(inst_t *)__return_address);
244		244
245	XFS_STATS_INC(xs_ig_found);	245	XFS_STATS_INC(xs_ig_found);
246		246
247	ip->i_flags &= ~XFS_IRECLAIMABLE;	247	ip->i_flags &= ~XFS_IRECLAIMABLE;
248	version = ih->ih_version;	248	version = ih->ih_version;
249	read_unlock(&ih->ih_lock);	249	read_unlock(&ih->ih_lock);
250	xfs_ihash_promote(ih, ip, version);	250	xfs_ihash_promote(ih, ip, version);
251		251
252	XFS_MOUNT_ILOCK(mp);	252	XFS_MOUNT_ILOCK(mp);
253	list_del_init(&ip->i_reclaim);	253	list_del_init(&ip->i_reclaim);
254	XFS_MOUNT_IUNLOCK(mp);	254	XFS_MOUNT_IUNLOCK(mp);
255		255
256	goto finish_inode;	256	goto finish_inode;
257		257
258	} else if (vp != inode_vp) {	258	} else if (vp != inode_vp) {
259	struct inode *inode = vn_to_inode(inode_vp);	259	struct inode *inode = vn_to_inode(inode_vp);
260		260
261	/* The inode is being torn down, pause and	261	/* The inode is being torn down, pause and
262	* try again.	262	* try again.
263	*/	263	*/
264	if (inode->i_state & (I_FREEING \| I_CLEAR)) {	264	if (inode->i_state & (I_FREEING \| I_CLEAR)) {
265	read_unlock(&ih->ih_lock);	265	read_unlock(&ih->ih_lock);
266	delay(1);	266	delay(1);
267	XFS_STATS_INC(xs_ig_frecycle);	267	XFS_STATS_INC(xs_ig_frecycle);
268		268
269	goto again;	269	goto again;
270	}	270	}
271	/* Chances are the other vnode (the one in the inode) is being torn	271	/* Chances are the other vnode (the one in the inode) is being torn
272	* down right now, and we landed on top of it. Question is, what do	272	* down right now, and we landed on top of it. Question is, what do
273	* we do? Unhook the old inode and hook up the new one?	273	* we do? Unhook the old inode and hook up the new one?
274	*/	274	*/
275	cmn_err(CE_PANIC,	275	cmn_err(CE_PANIC,
276	"xfs_iget_core: ambiguous vns: vp/0x%p, invp/0x%p",	276	"xfs_iget_core: ambiguous vns: vp/0x%p, invp/0x%p",
277	inode_vp, vp);	277	inode_vp, vp);
278	}	278	}
279		279
280	/*	280	/*
281	* Inode cache hit: if ip is not at the front of	281	* Inode cache hit: if ip is not at the front of
282	* its hash chain, move it there now.	282	* its hash chain, move it there now.
283	* Do this with the lock held for update, but	283	* Do this with the lock held for update, but
284	* do statistics after releasing the lock.	284	* do statistics after releasing the lock.
285	*/	285	*/
286	version = ih->ih_version;	286	version = ih->ih_version;
287	read_unlock(&ih->ih_lock);	287	read_unlock(&ih->ih_lock);
288	xfs_ihash_promote(ih, ip, version);	288	xfs_ihash_promote(ih, ip, version);
289	XFS_STATS_INC(xs_ig_found);	289	XFS_STATS_INC(xs_ig_found);
290		290
291	finish_inode:	291	finish_inode:
292	if (ip->i_d.di_mode == 0) {	292	if (ip->i_d.di_mode == 0) {
293	if (!(flags & IGET_CREATE))	293	if (!(flags & XFS_IGET_CREATE))
294	return ENOENT;	294	return ENOENT;
295	xfs_iocore_inode_reinit(ip);	295	xfs_iocore_inode_reinit(ip);
296	}	296	}
297		297
298	if (lock_flags != 0)	298	if (lock_flags != 0)
299	xfs_ilock(ip, lock_flags);	299	xfs_ilock(ip, lock_flags);
300		300
301	ip->i_flags &= ~XFS_ISTALE;	301	ip->i_flags &= ~XFS_ISTALE;
302		302
303	vn_trace_exit(vp, "xfs_iget.found",	303	vn_trace_exit(vp, "xfs_iget.found",
304	(inst_t *)__return_address);	304	(inst_t *)__return_address);
305	goto return_ip;	305	goto return_ip;
306	}	306	}
307	}	307	}
308		308
309	/*	309	/*
310	* Inode cache miss: save the hash chain version stamp and unlock	310	* Inode cache miss: save the hash chain version stamp and unlock
311	* the chain, so we don't deadlock in vn_alloc.	311	* the chain, so we don't deadlock in vn_alloc.
312	*/	312	*/
313	XFS_STATS_INC(xs_ig_missed);	313	XFS_STATS_INC(xs_ig_missed);
314		314
315	version = ih->ih_version;	315	version = ih->ih_version;
316		316
317	read_unlock(&ih->ih_lock);	317	read_unlock(&ih->ih_lock);
318		318
319	/*	319	/*
320	* Read the disk inode attributes into a new inode structure and get	320	* Read the disk inode attributes into a new inode structure and get
321	* a new vnode for it. This should also initialize i_ino and i_mount.	321	* a new vnode for it. This should also initialize i_ino and i_mount.
322	*/	322	*/
323	error = xfs_iread(mp, tp, ino, &ip, bno);	323	error = xfs_iread(mp, tp, ino, &ip, bno,
324	if (error) {	324	(flags & XFS_IGET_BULKSTAT) ? XFS_IMAP_BULKSTAT : 0);
		325	if (error)
325	return error;	326	return error;
326	}
327		327
328	vn_trace_exit(vp, "xfs_iget.alloc", (inst_t *)__return_address);	328	vn_trace_exit(vp, "xfs_iget.alloc", (inst_t *)__return_address);
329		329
330	xfs_inode_lock_init(ip, vp);	330	xfs_inode_lock_init(ip, vp);
331	xfs_iocore_inode_init(ip);	331	xfs_iocore_inode_init(ip);
332		332
333	if (lock_flags != 0) {	333	if (lock_flags)
334	xfs_ilock(ip, lock_flags);	334	xfs_ilock(ip, lock_flags);
335	}	335
336		336	if ((ip->i_d.di_mode == 0) && !(flags & XFS_IGET_CREATE)) {
337	if ((ip->i_d.di_mode == 0) && !(flags & IGET_CREATE)) {
338	xfs_idestroy(ip);	337	xfs_idestroy(ip);
339	return ENOENT;	338	return ENOENT;
340	}	339	}
341		340
342	/*	341	/*
343	* Put ip on its hash chain, unless someone else hashed a duplicate	342	* Put ip on its hash chain, unless someone else hashed a duplicate
344	* after we released the hash lock.	343	* after we released the hash lock.
345	*/	344	*/
346	write_lock(&ih->ih_lock);	345	write_lock(&ih->ih_lock);
347		346
348	if (ih->ih_version != version) {	347	if (ih->ih_version != version) {
349	for (iq = ih->ih_next; iq != NULL; iq = iq->i_next) {	348	for (iq = ih->ih_next; iq != NULL; iq = iq->i_next) {
350	if (iq->i_ino == ino) {	349	if (iq->i_ino == ino) {
351	write_unlock(&ih->ih_lock);	350	write_unlock(&ih->ih_lock);
352	xfs_idestroy(ip);	351	xfs_idestroy(ip);
353		352
354	XFS_STATS_INC(xs_ig_dup);	353	XFS_STATS_INC(xs_ig_dup);
355	goto again;	354	goto again;
356	}	355	}
357	}	356	}
358	}	357	}
359		358
360	/*	359	/*
361	* These values _must_ be set before releasing ihlock!	360	* These values _must_ be set before releasing ihlock!
362	*/	361	*/
363	ip->i_hash = ih;	362	ip->i_hash = ih;
364	if ((iq = ih->ih_next)) {	363	if ((iq = ih->ih_next)) {
365	iq->i_prevp = &ip->i_next;	364	iq->i_prevp = &ip->i_next;
366	}	365	}
367	ip->i_next = iq;	366	ip->i_next = iq;
368	ip->i_prevp = &ih->ih_next;	367	ip->i_prevp = &ih->ih_next;
369	ih->ih_next = ip;	368	ih->ih_next = ip;
370	ip->i_udquot = ip->i_gdquot = NULL;	369	ip->i_udquot = ip->i_gdquot = NULL;
371	ih->ih_version++;	370	ih->ih_version++;
372	ip->i_flags \|= XFS_INEW;	371	ip->i_flags \|= XFS_INEW;
373		372
374	write_unlock(&ih->ih_lock);	373	write_unlock(&ih->ih_lock);
375		374
376	/*	375	/*
377	* put ip on its cluster's hash chain	376	* put ip on its cluster's hash chain
378	*/	377	*/
379	ASSERT(ip->i_chash == NULL && ip->i_cprev == NULL &&	378	ASSERT(ip->i_chash == NULL && ip->i_cprev == NULL &&
380	ip->i_cnext == NULL);	379	ip->i_cnext == NULL);
381		380
382	chlnew = NULL;	381	chlnew = NULL;
383	ch = XFS_CHASH(mp, ip->i_blkno);	382	ch = XFS_CHASH(mp, ip->i_blkno);
384	chlredo:	383	chlredo:
385	s = mutex_spinlock(&ch->ch_lock);	384	s = mutex_spinlock(&ch->ch_lock);
386	for (chl = ch->ch_list; chl != NULL; chl = chl->chl_next) {	385	for (chl = ch->ch_list; chl != NULL; chl = chl->chl_next) {
387	if (chl->chl_blkno == ip->i_blkno) {	386	if (chl->chl_blkno == ip->i_blkno) {
388		387
389	/* insert this inode into the doubly-linked list	388	/* insert this inode into the doubly-linked list
390	* where chl points */	389	* where chl points */
391	if ((iq = chl->chl_ip)) {	390	if ((iq = chl->chl_ip)) {
392	ip->i_cprev = iq->i_cprev;	391	ip->i_cprev = iq->i_cprev;
393	iq->i_cprev->i_cnext = ip;	392	iq->i_cprev->i_cnext = ip;
394	iq->i_cprev = ip;	393	iq->i_cprev = ip;
395	ip->i_cnext = iq;	394	ip->i_cnext = iq;
396	} else {	395	} else {
397	ip->i_cnext = ip;	396	ip->i_cnext = ip;
398	ip->i_cprev = ip;	397	ip->i_cprev = ip;
399	}	398	}
400	chl->chl_ip = ip;	399	chl->chl_ip = ip;
401	ip->i_chash = chl;	400	ip->i_chash = chl;
402	break;	401	break;
403	}	402	}
404	}	403	}
405		404
406	/* no hash list found for this block; add a new hash list */	405	/* no hash list found for this block; add a new hash list */
407	if (chl == NULL) {	406	if (chl == NULL) {
408	if (chlnew == NULL) {	407	if (chlnew == NULL) {
409	mutex_spinunlock(&ch->ch_lock, s);	408	mutex_spinunlock(&ch->ch_lock, s);
410	ASSERT(xfs_chashlist_zone != NULL);	409	ASSERT(xfs_chashlist_zone != NULL);
411	chlnew = (xfs_chashlist_t *)	410	chlnew = (xfs_chashlist_t *)
412	kmem_zone_alloc(xfs_chashlist_zone,	411	kmem_zone_alloc(xfs_chashlist_zone,
413	KM_SLEEP);	412	KM_SLEEP);
414	ASSERT(chlnew != NULL);	413	ASSERT(chlnew != NULL);
415	goto chlredo;	414	goto chlredo;
416	} else {	415	} else {
417	ip->i_cnext = ip;	416	ip->i_cnext = ip;
418	ip->i_cprev = ip;	417	ip->i_cprev = ip;
419	ip->i_chash = chlnew;	418	ip->i_chash = chlnew;
420	chlnew->chl_ip = ip;	419	chlnew->chl_ip = ip;
421	chlnew->chl_blkno = ip->i_blkno;	420	chlnew->chl_blkno = ip->i_blkno;
422	if (ch->ch_list)	421	if (ch->ch_list)
423	ch->ch_list->chl_prev = chlnew;	422	ch->ch_list->chl_prev = chlnew;
424	chlnew->chl_next = ch->ch_list;	423	chlnew->chl_next = ch->ch_list;
425	chlnew->chl_prev = NULL;	424	chlnew->chl_prev = NULL;
426	ch->ch_list = chlnew;	425	ch->ch_list = chlnew;
427	chlnew = NULL;	426	chlnew = NULL;
428	}	427	}
429	} else {	428	} else {
430	if (chlnew != NULL) {	429	if (chlnew != NULL) {
431	kmem_zone_free(xfs_chashlist_zone, chlnew);	430	kmem_zone_free(xfs_chashlist_zone, chlnew);
432	}	431	}
433	}	432	}
434		433
435	mutex_spinunlock(&ch->ch_lock, s);	434	mutex_spinunlock(&ch->ch_lock, s);
436		435
437		436
438	/*	437	/*
439	* Link ip to its mount and thread it on the mount's inode list.	438	* Link ip to its mount and thread it on the mount's inode list.
440	*/	439	*/
441	XFS_MOUNT_ILOCK(mp);	440	XFS_MOUNT_ILOCK(mp);
442	if ((iq = mp->m_inodes)) {	441	if ((iq = mp->m_inodes)) {
443	ASSERT(iq->i_mprev->i_mnext == iq);	442	ASSERT(iq->i_mprev->i_mnext == iq);
444	ip->i_mprev = iq->i_mprev;	443	ip->i_mprev = iq->i_mprev;
445	iq->i_mprev->i_mnext = ip;	444	iq->i_mprev->i_mnext = ip;
446	iq->i_mprev = ip;	445	iq->i_mprev = ip;
447	ip->i_mnext = iq;	446	ip->i_mnext = iq;
448	} else {	447	} else {
449	ip->i_mnext = ip;	448	ip->i_mnext = ip;
450	ip->i_mprev = ip;	449	ip->i_mprev = ip;
451	}	450	}
452	mp->m_inodes = ip;	451	mp->m_inodes = ip;
453		452
454	XFS_MOUNT_IUNLOCK(mp);	453	XFS_MOUNT_IUNLOCK(mp);
455		454
456	return_ip:	455	return_ip:
457	ASSERT(ip->i_df.if_ext_max ==	456	ASSERT(ip->i_df.if_ext_max ==
458	XFS_IFORK_DSIZE(ip) / sizeof(xfs_bmbt_rec_t));	457	XFS_IFORK_DSIZE(ip) / sizeof(xfs_bmbt_rec_t));
459		458
460	ASSERT(((ip->i_d.di_flags & XFS_DIFLAG_REALTIME) != 0) ==	459	ASSERT(((ip->i_d.di_flags & XFS_DIFLAG_REALTIME) != 0) ==
461	((ip->i_iocore.io_flags & XFS_IOCORE_RT) != 0));	460	((ip->i_iocore.io_flags & XFS_IOCORE_RT) != 0));
462		461
463	*ipp = ip;	462	*ipp = ip;
464		463
465	/*	464	/*
466	* If we have a real type for an on-disk inode, we can set ops(&unlock)	465	* If we have a real type for an on-disk inode, we can set ops(&unlock)
467	* now. If it's a new inode being created, xfs_ialloc will handle it.	466	* now. If it's a new inode being created, xfs_ialloc will handle it.
468	*/	467	*/
469	bhv_vfs_init_vnode(XFS_MTOVFS(mp), vp, XFS_ITOBHV(ip), 1);	468	bhv_vfs_init_vnode(XFS_MTOVFS(mp), vp, XFS_ITOBHV(ip), 1);
470		469
471	return 0;	470	return 0;
472	}	471	}
473		472
474		473
475	/*	474	/*
476	* The 'normal' internal xfs_iget, if needed it will	475	* The 'normal' internal xfs_iget, if needed it will
477	* 'allocate', or 'get', the vnode.	476	* 'allocate', or 'get', the vnode.
478	*/	477	*/
479	int	478	int
480	xfs_iget(	479	xfs_iget(
481	xfs_mount_t *mp,	480	xfs_mount_t *mp,
482	xfs_trans_t *tp,	481	xfs_trans_t *tp,
483	xfs_ino_t ino,	482	xfs_ino_t ino,
484	uint flags,	483	uint flags,
485	uint lock_flags,	484	uint lock_flags,
486	xfs_inode_t **ipp,	485	xfs_inode_t **ipp,
487	xfs_daddr_t bno)	486	xfs_daddr_t bno)
488	{	487	{
489	struct inode *inode;	488	struct inode *inode;
490	bhv_vnode_t *vp = NULL;	489	bhv_vnode_t *vp = NULL;
491	int error;	490	int error;
492		491
493	XFS_STATS_INC(xs_ig_attempts);	492	XFS_STATS_INC(xs_ig_attempts);
494		493
495	retry:	494	retry:
496	if ((inode = iget_locked(XFS_MTOVFS(mp)->vfs_super, ino))) {	495	if ((inode = iget_locked(XFS_MTOVFS(mp)->vfs_super, ino))) {
497	xfs_inode_t *ip;	496	xfs_inode_t *ip;
498		497
499	vp = vn_from_inode(inode);	498	vp = vn_from_inode(inode);
500	if (inode->i_state & I_NEW) {	499	if (inode->i_state & I_NEW) {
501	vn_initialize(inode);	500	vn_initialize(inode);
502	error = xfs_iget_core(vp, mp, tp, ino, flags,	501	error = xfs_iget_core(vp, mp, tp, ino, flags,
503	lock_flags, ipp, bno);	502	lock_flags, ipp, bno);
504	if (error) {	503	if (error) {
505	vn_mark_bad(vp);	504	vn_mark_bad(vp);
506	if (inode->i_state & I_NEW)	505	if (inode->i_state & I_NEW)
507	unlock_new_inode(inode);	506	unlock_new_inode(inode);
508	iput(inode);	507	iput(inode);
509	}	508	}
510	} else {	509	} else {
511	/*	510	/*
512	* If the inode is not fully constructed due to	511	* If the inode is not fully constructed due to
513	* filehandle mismatches wait for the inode to go	512	* filehandle mismatches wait for the inode to go
514	* away and try again.	513	* away and try again.
515	*	514	*
516	* iget_locked will call __wait_on_freeing_inode	515	* iget_locked will call __wait_on_freeing_inode
517	* to wait for the inode to go away.	516	* to wait for the inode to go away.
518	*/	517	*/
519	if (is_bad_inode(inode) \|\|	518	if (is_bad_inode(inode) \|\|
520	((ip = xfs_vtoi(vp)) == NULL)) {	519	((ip = xfs_vtoi(vp)) == NULL)) {
521	iput(inode);	520	iput(inode);
522	delay(1);	521	delay(1);
523	goto retry;	522	goto retry;
524	}	523	}
525		524
526	if (lock_flags != 0)	525	if (lock_flags != 0)
527	xfs_ilock(ip, lock_flags);	526	xfs_ilock(ip, lock_flags);
528	XFS_STATS_INC(xs_ig_found);	527	XFS_STATS_INC(xs_ig_found);
529	*ipp = ip;	528	*ipp = ip;
530	error = 0;	529	error = 0;
531	}	530	}
532	} else	531	} else
533	error = ENOMEM; /* If we got no inode we are out of memory */	532	error = ENOMEM; /* If we got no inode we are out of memory */
534		533
535	return error;	534	return error;
536	}	535	}
537		536
538	/*	537	/*
539	* Do the setup for the various locks within the incore inode.	538	* Do the setup for the various locks within the incore inode.
540	*/	539	*/
541	void	540	void
542	xfs_inode_lock_init(	541	xfs_inode_lock_init(
543	xfs_inode_t *ip,	542	xfs_inode_t *ip,
544	bhv_vnode_t *vp)	543	bhv_vnode_t *vp)
545	{	544	{
546	mrlock_init(&ip->i_lock, MRLOCK_ALLOW_EQUAL_PRI\|MRLOCK_BARRIER,	545	mrlock_init(&ip->i_lock, MRLOCK_ALLOW_EQUAL_PRI\|MRLOCK_BARRIER,
547	"xfsino", (long)vp->v_number);	546	"xfsino", (long)vp->v_number);
548	mrlock_init(&ip->i_iolock, MRLOCK_BARRIER, "xfsio", vp->v_number);	547	mrlock_init(&ip->i_iolock, MRLOCK_BARRIER, "xfsio", vp->v_number);
549	init_waitqueue_head(&ip->i_ipin_wait);	548	init_waitqueue_head(&ip->i_ipin_wait);
550	atomic_set(&ip->i_pincount, 0);	549	atomic_set(&ip->i_pincount, 0);
551	init_sema(&ip->i_flock, 1, "xfsfino", vp->v_number);	550	init_sema(&ip->i_flock, 1, "xfsfino", vp->v_number);
552	}	551	}
553		552
554	/*	553	/*
555	* Look for the inode corresponding to the given ino in the hash table.	554	* Look for the inode corresponding to the given ino in the hash table.
556	* If it is there and its i_transp pointer matches tp, return it.	555	* If it is there and its i_transp pointer matches tp, return it.
557	* Otherwise, return NULL.	556	* Otherwise, return NULL.
558	*/	557	*/
559	xfs_inode_t *	558	xfs_inode_t *
560	xfs_inode_incore(xfs_mount_t *mp,	559	xfs_inode_incore(xfs_mount_t *mp,
561	xfs_ino_t ino,	560	xfs_ino_t ino,
562	xfs_trans_t *tp)	561	xfs_trans_t *tp)
563	{	562	{
564	xfs_ihash_t *ih;	563	xfs_ihash_t *ih;
565	xfs_inode_t *ip;	564	xfs_inode_t *ip;
566	ulong version;	565	ulong version;
567		566
568	ih = XFS_IHASH(mp, ino);	567	ih = XFS_IHASH(mp, ino);
569	read_lock(&ih->ih_lock);	568	read_lock(&ih->ih_lock);
570	for (ip = ih->ih_next; ip != NULL; ip = ip->i_next) {	569	for (ip = ih->ih_next; ip != NULL; ip = ip->i_next) {
571	if (ip->i_ino == ino) {	570	if (ip->i_ino == ino) {
572	/*	571	/*
573	* If we find it and tp matches, return it.	572	* If we find it and tp matches, return it.
574	* Also move it to the front of the hash list	573	* Also move it to the front of the hash list
575	* if we find it and it is not already there.	574	* if we find it and it is not already there.
576	* Otherwise break from the loop and return	575	* Otherwise break from the loop and return
577	* NULL.	576	* NULL.
578	*/	577	*/
579	if (ip->i_transp == tp) {	578	if (ip->i_transp == tp) {
580	version = ih->ih_version;	579	version = ih->ih_version;
581	read_unlock(&ih->ih_lock);	580	read_unlock(&ih->ih_lock);
582	xfs_ihash_promote(ih, ip, version);	581	xfs_ihash_promote(ih, ip, version);
583	return (ip);	582	return (ip);
584	}	583	}
585	break;	584	break;
586	}	585	}
587	}	586	}
588	read_unlock(&ih->ih_lock);	587	read_unlock(&ih->ih_lock);
589	return (NULL);	588	return (NULL);
590	}	589	}
591		590
592	/*	591	/*
593	* Decrement reference count of an inode structure and unlock it.	592	* Decrement reference count of an inode structure and unlock it.
594	*	593	*
595	* ip -- the inode being released	594	* ip -- the inode being released
596	* lock_flags -- this parameter indicates the inode's locks to be	595	* lock_flags -- this parameter indicates the inode's locks to be
597	* to be released. See the comment on xfs_iunlock() for a list	596	* to be released. See the comment on xfs_iunlock() for a list
598	* of valid values.	597	* of valid values.
599	*/	598	*/
600	void	599	void
601	xfs_iput(xfs_inode_t *ip,	600	xfs_iput(xfs_inode_t *ip,
602	uint lock_flags)	601	uint lock_flags)
603	{	602	{
604	bhv_vnode_t *vp = XFS_ITOV(ip);	603	bhv_vnode_t *vp = XFS_ITOV(ip);
605		604
606	vn_trace_entry(vp, "xfs_iput", (inst_t *)__return_address);	605	vn_trace_entry(vp, "xfs_iput", (inst_t *)__return_address);
607	xfs_iunlock(ip, lock_flags);	606	xfs_iunlock(ip, lock_flags);
608	VN_RELE(vp);	607	VN_RELE(vp);
609	}	608	}
610		609
611	/*	610	/*
612	* Special iput for brand-new inodes that are still locked	611	* Special iput for brand-new inodes that are still locked
613	*/	612	*/
614	void	613	void
615	xfs_iput_new(xfs_inode_t *ip,	614	xfs_iput_new(xfs_inode_t *ip,
616	uint lock_flags)	615	uint lock_flags)
617	{	616	{
618	bhv_vnode_t *vp = XFS_ITOV(ip);	617	bhv_vnode_t *vp = XFS_ITOV(ip);
619	struct inode *inode = vn_to_inode(vp);	618	struct inode *inode = vn_to_inode(vp);
620		619
621	vn_trace_entry(vp, "xfs_iput_new", (inst_t *)__return_address);	620	vn_trace_entry(vp, "xfs_iput_new", (inst_t *)__return_address);
622		621
623	if ((ip->i_d.di_mode == 0)) {	622	if ((ip->i_d.di_mode == 0)) {
624	ASSERT(!(ip->i_flags & XFS_IRECLAIMABLE));	623	ASSERT(!(ip->i_flags & XFS_IRECLAIMABLE));
625	vn_mark_bad(vp);	624	vn_mark_bad(vp);
626	}	625	}
627	if (inode->i_state & I_NEW)	626	if (inode->i_state & I_NEW)
628	unlock_new_inode(inode);	627	unlock_new_inode(inode);
629	if (lock_flags)	628	if (lock_flags)
630	xfs_iunlock(ip, lock_flags);	629	xfs_iunlock(ip, lock_flags);
631	VN_RELE(vp);	630	VN_RELE(vp);
632	}	631	}
633		632
634		633
635	/*	634	/*
636	* This routine embodies the part of the reclaim code that pulls	635	* This routine embodies the part of the reclaim code that pulls
637	* the inode from the inode hash table and the mount structure's	636	* the inode from the inode hash table and the mount structure's
638	* inode list.	637	* inode list.
639	* This should only be called from xfs_reclaim().	638	* This should only be called from xfs_reclaim().
640	*/	639	*/
641	void	640	void
642	xfs_ireclaim(xfs_inode_t *ip)	641	xfs_ireclaim(xfs_inode_t *ip)
643	{	642	{
644	bhv_vnode_t *vp;	643	bhv_vnode_t *vp;
645		644
646	/*	645	/*
647	* Remove from old hash list and mount list.	646	* Remove from old hash list and mount list.
648	*/	647	*/
649	XFS_STATS_INC(xs_ig_reclaims);	648	XFS_STATS_INC(xs_ig_reclaims);
650		649
651	xfs_iextract(ip);	650	xfs_iextract(ip);
652		651
653	/*	652	/*
654	* Here we do a spurious inode lock in order to coordinate with	653	* Here we do a spurious inode lock in order to coordinate with
655	* xfs_sync(). This is because xfs_sync() references the inodes	654	* xfs_sync(). This is because xfs_sync() references the inodes
656	* in the mount list without taking references on the corresponding	655	* in the mount list without taking references on the corresponding
657	* vnodes. We make that OK here by ensuring that we wait until	656	* vnodes. We make that OK here by ensuring that we wait until
658	* the inode is unlocked in xfs_sync() before we go ahead and	657	* the inode is unlocked in xfs_sync() before we go ahead and
659	* free it. We get both the regular lock and the io lock because	658	* free it. We get both the regular lock and the io lock because
660	* the xfs_sync() code may need to drop the regular one but will	659	* the xfs_sync() code may need to drop the regular one but will
661	* still hold the io lock.	660	* still hold the io lock.
662	*/	661	*/
663	xfs_ilock(ip, XFS_ILOCK_EXCL \| XFS_IOLOCK_EXCL);	662	xfs_ilock(ip, XFS_ILOCK_EXCL \| XFS_IOLOCK_EXCL);
664		663
665	/*	664	/*
666	* Release dquots (and their references) if any. An inode may escape	665	* Release dquots (and their references) if any. An inode may escape
667	* xfs_inactive and get here via vn_alloc->vn_reclaim path.	666	* xfs_inactive and get here via vn_alloc->vn_reclaim path.
668	*/	667	*/
669	XFS_QM_DQDETACH(ip->i_mount, ip);	668	XFS_QM_DQDETACH(ip->i_mount, ip);
670		669
671	/*	670	/*
672	* Pull our behavior descriptor from the vnode chain.	671	* Pull our behavior descriptor from the vnode chain.
673	*/	672	*/
674	vp = XFS_ITOV_NULL(ip);	673	vp = XFS_ITOV_NULL(ip);
675	if (vp) {	674	if (vp) {
676	vn_bhv_remove(VN_BHV_HEAD(vp), XFS_ITOBHV(ip));	675	vn_bhv_remove(VN_BHV_HEAD(vp), XFS_ITOBHV(ip));
677	}	676	}
678		677
679	/*	678	/*
680	* Free all memory associated with the inode.	679	* Free all memory associated with the inode.
681	*/	680	*/
682	xfs_idestroy(ip);	681	xfs_idestroy(ip);
683	}	682	}
684		683
685	/*	684	/*
686	* This routine removes an about-to-be-destroyed inode from	685	* This routine removes an about-to-be-destroyed inode from
687	* all of the lists in which it is located with the exception	686	* all of the lists in which it is located with the exception
688	* of the behavior chain.	687	* of the behavior chain.
689	*/	688	*/
690	void	689	void
691	xfs_iextract(	690	xfs_iextract(
692	xfs_inode_t *ip)	691	xfs_inode_t *ip)
693	{	692	{
694	xfs_ihash_t *ih;	693	xfs_ihash_t *ih;
695	xfs_inode_t *iq;	694	xfs_inode_t *iq;
696	xfs_mount_t *mp;	695	xfs_mount_t *mp;
697	xfs_chash_t *ch;	696	xfs_chash_t *ch;
698	xfs_chashlist_t chl, chm;	697	xfs_chashlist_t chl, chm;
699	SPLDECL(s);	698	SPLDECL(s);
700		699
701	ih = ip->i_hash;	700	ih = ip->i_hash;
702	write_lock(&ih->ih_lock);	701	write_lock(&ih->ih_lock);
703	if ((iq = ip->i_next)) {	702	if ((iq = ip->i_next)) {
704	iq->i_prevp = ip->i_prevp;	703	iq->i_prevp = ip->i_prevp;
705	}	704	}
706	*ip->i_prevp = iq;	705	*ip->i_prevp = iq;
707	ih->ih_version++;	706	ih->ih_version++;
708	write_unlock(&ih->ih_lock);	707	write_unlock(&ih->ih_lock);
709		708
710	/*	709	/*
711	* Remove from cluster hash list	710	* Remove from cluster hash list
712	* 1) delete the chashlist if this is the last inode on the chashlist	711	* 1) delete the chashlist if this is the last inode on the chashlist
713	* 2) unchain from list of inodes	712	* 2) unchain from list of inodes
714	* 3) point chashlist->chl_ip to 'chl_next' if to this inode.	713	* 3) point chashlist->chl_ip to 'chl_next' if to this inode.
715	*/	714	*/
716	mp = ip->i_mount;	715	mp = ip->i_mount;
717	ch = XFS_CHASH(mp, ip->i_blkno);	716	ch = XFS_CHASH(mp, ip->i_blkno);
718	s = mutex_spinlock(&ch->ch_lock);	717	s = mutex_spinlock(&ch->ch_lock);
719		718
720	if (ip->i_cnext == ip) {	719	if (ip->i_cnext == ip) {
721	/* Last inode on chashlist */	720	/* Last inode on chashlist */
722	ASSERT(ip->i_cnext == ip && ip->i_cprev == ip);	721	ASSERT(ip->i_cnext == ip && ip->i_cprev == ip);
723	ASSERT(ip->i_chash != NULL);	722	ASSERT(ip->i_chash != NULL);
724	chm=NULL;	723	chm=NULL;
725	chl = ip->i_chash;	724	chl = ip->i_chash;
726	if (chl->chl_prev)	725	if (chl->chl_prev)
727	chl->chl_prev->chl_next = chl->chl_next;	726	chl->chl_prev->chl_next = chl->chl_next;
728	else	727	else
729	ch->ch_list = chl->chl_next;	728	ch->ch_list = chl->chl_next;
730	if (chl->chl_next)	729	if (chl->chl_next)
731	chl->chl_next->chl_prev = chl->chl_prev;	730	chl->chl_next->chl_prev = chl->chl_prev;
732	kmem_zone_free(xfs_chashlist_zone, chl);	731	kmem_zone_free(xfs_chashlist_zone, chl);
733	} else {	732	} else {
734	/* delete one inode from a non-empty list */	733	/* delete one inode from a non-empty list */
735	iq = ip->i_cnext;	734	iq = ip->i_cnext;
736	iq->i_cprev = ip->i_cprev;	735	iq->i_cprev = ip->i_cprev;
737	ip->i_cprev->i_cnext = iq;	736	ip->i_cprev->i_cnext = iq;
738	if (ip->i_chash->chl_ip == ip) {	737	if (ip->i_chash->chl_ip == ip) {
739	ip->i_chash->chl_ip = iq;	738	ip->i_chash->chl_ip = iq;
740	}	739	}
741	ip->i_chash = __return_address;	740	ip->i_chash = __return_address;
742	ip->i_cprev = __return_address;	741	ip->i_cprev = __return_address;
743	ip->i_cnext = __return_address;	742	ip->i_cnext = __return_address;
744	}	743	}
745	mutex_spinunlock(&ch->ch_lock, s);	744	mutex_spinunlock(&ch->ch_lock, s);
746		745
747	/*	746	/*
748	* Remove from mount's inode list.	747	* Remove from mount's inode list.
749	*/	748	*/
750	XFS_MOUNT_ILOCK(mp);	749	XFS_MOUNT_ILOCK(mp);
751	ASSERT((ip->i_mnext != NULL) && (ip->i_mprev != NULL));	750	ASSERT((ip->i_mnext != NULL) && (ip->i_mprev != NULL));
752	iq = ip->i_mnext;	751	iq = ip->i_mnext;
753	iq->i_mprev = ip->i_mprev;	752	iq->i_mprev = ip->i_mprev;
754	ip->i_mprev->i_mnext = iq;	753	ip->i_mprev->i_mnext = iq;
755		754
756	/*	755	/*
757	* Fix up the head pointer if it points to the inode being deleted.	756	* Fix up the head pointer if it points to the inode being deleted.
758	*/	757	*/
759	if (mp->m_inodes == ip) {	758	if (mp->m_inodes == ip) {
760	if (ip == iq) {	759	if (ip == iq) {
761	mp->m_inodes = NULL;	760	mp->m_inodes = NULL;
762	} else {	761	} else {
763	mp->m_inodes = iq;	762	mp->m_inodes = iq;
764	}	763	}
765	}	764	}
766		765
767	/* Deal with the deleted inodes list */	766	/* Deal with the deleted inodes list */
768	list_del_init(&ip->i_reclaim);	767	list_del_init(&ip->i_reclaim);
769		768
770	mp->m_ireclaims++;	769	mp->m_ireclaims++;
771	XFS_MOUNT_IUNLOCK(mp);	770	XFS_MOUNT_IUNLOCK(mp);
772	}	771	}
773		772
774	/*	773	/*
775	* This is a wrapper routine around the xfs_ilock() routine	774	* This is a wrapper routine around the xfs_ilock() routine
776	* used to centralize some grungy code. It is used in places	775	* used to centralize some grungy code. It is used in places
777	* that wish to lock the inode solely for reading the extents.	776	* that wish to lock the inode solely for reading the extents.
778	* The reason these places can't just call xfs_ilock(SHARED)	777	* The reason these places can't just call xfs_ilock(SHARED)
779	* is that the inode lock also guards to bringing in of the	778	* is that the inode lock also guards to bringing in of the
780	* extents from disk for a file in b-tree format. If the inode	779	* extents from disk for a file in b-tree format. If the inode
781	* is in b-tree format, then we need to lock the inode exclusively	780	* is in b-tree format, then we need to lock the inode exclusively
782	* until the extents are read in. Locking it exclusively all	781	* until the extents are read in. Locking it exclusively all
783	* the time would limit our parallelism unnecessarily, though.	782	* the time would limit our parallelism unnecessarily, though.
784	* What we do instead is check to see if the extents have been	783	* What we do instead is check to see if the extents have been
785	* read in yet, and only lock the inode exclusively if they	784	* read in yet, and only lock the inode exclusively if they
786	* have not.	785	* have not.
787	*	786	*
788	* The function returns a value which should be given to the	787	* The function returns a value which should be given to the
789	* corresponding xfs_iunlock_map_shared(). This value is	788	* corresponding xfs_iunlock_map_shared(). This value is
790	* the mode in which the lock was actually taken.	789	* the mode in which the lock was actually taken.
791	*/	790	*/
792	uint	791	uint
793	xfs_ilock_map_shared(	792	xfs_ilock_map_shared(
794	xfs_inode_t *ip)	793	xfs_inode_t *ip)
795	{	794	{
796	uint lock_mode;	795	uint lock_mode;
797		796
798	if ((ip->i_d.di_format == XFS_DINODE_FMT_BTREE) &&	797	if ((ip->i_d.di_format == XFS_DINODE_FMT_BTREE) &&
799	((ip->i_df.if_flags & XFS_IFEXTENTS) == 0)) {	798	((ip->i_df.if_flags & XFS_IFEXTENTS) == 0)) {
800	lock_mode = XFS_ILOCK_EXCL;	799	lock_mode = XFS_ILOCK_EXCL;
801	} else {	800	} else {
802	lock_mode = XFS_ILOCK_SHARED;	801	lock_mode = XFS_ILOCK_SHARED;
803	}	802	}
804		803
805	xfs_ilock(ip, lock_mode);	804	xfs_ilock(ip, lock_mode);
806		805
807	return lock_mode;	806	return lock_mode;
808	}	807	}
809		808
810	/*	809	/*
811	* This is simply the unlock routine to go with xfs_ilock_map_shared().	810	* This is simply the unlock routine to go with xfs_ilock_map_shared().
812	* All it does is call xfs_iunlock() with the given lock_mode.	811	* All it does is call xfs_iunlock() with the given lock_mode.
813	*/	812	*/
814	void	813	void
815	xfs_iunlock_map_shared(	814	xfs_iunlock_map_shared(
816	xfs_inode_t *ip,	815	xfs_inode_t *ip,
817	unsigned int lock_mode)	816	unsigned int lock_mode)
818	{	817	{
819	xfs_iunlock(ip, lock_mode);	818	xfs_iunlock(ip, lock_mode);
820	}	819	}
821		820
822	/*	821	/*
823	* The xfs inode contains 2 locks: a multi-reader lock called the	822	* The xfs inode contains 2 locks: a multi-reader lock called the
824	* i_iolock and a multi-reader lock called the i_lock. This routine	823	* i_iolock and a multi-reader lock called the i_lock. This routine
825	* allows either or both of the locks to be obtained.	824	* allows either or both of the locks to be obtained.
826	*	825	*
827	* The 2 locks should always be ordered so that the IO lock is	826	* The 2 locks should always be ordered so that the IO lock is
828	* obtained first in order to prevent deadlock.	827	* obtained first in order to prevent deadlock.
829	*	828	*
830	* ip -- the inode being locked	829	* ip -- the inode being locked
831	* lock_flags -- this parameter indicates the inode's locks	830	* lock_flags -- this parameter indicates the inode's locks
832	* to be locked. It can be:	831	* to be locked. It can be:
833	* XFS_IOLOCK_SHARED,	832	* XFS_IOLOCK_SHARED,
834	* XFS_IOLOCK_EXCL,	833	* XFS_IOLOCK_EXCL,
835	* XFS_ILOCK_SHARED,	834	* XFS_ILOCK_SHARED,
836	* XFS_ILOCK_EXCL,	835	* XFS_ILOCK_EXCL,
837	* XFS_IOLOCK_SHARED \| XFS_ILOCK_SHARED,	836	* XFS_IOLOCK_SHARED \| XFS_ILOCK_SHARED,
838	* XFS_IOLOCK_SHARED \| XFS_ILOCK_EXCL,	837	* XFS_IOLOCK_SHARED \| XFS_ILOCK_EXCL,
839	* XFS_IOLOCK_EXCL \| XFS_ILOCK_SHARED,	838	* XFS_IOLOCK_EXCL \| XFS_ILOCK_SHARED,
840	* XFS_IOLOCK_EXCL \| XFS_ILOCK_EXCL	839	* XFS_IOLOCK_EXCL \| XFS_ILOCK_EXCL
841	*/	840	*/
842	void	841	void
843	xfs_ilock(xfs_inode_t *ip,	842	xfs_ilock(xfs_inode_t *ip,
844	uint lock_flags)	843	uint lock_flags)
845	{	844	{
846	/*	845	/*
847	* You can't set both SHARED and EXCL for the same lock,	846	* You can't set both SHARED and EXCL for the same lock,
848	* and only XFS_IOLOCK_SHARED, XFS_IOLOCK_EXCL, XFS_ILOCK_SHARED,	847	* and only XFS_IOLOCK_SHARED, XFS_IOLOCK_EXCL, XFS_ILOCK_SHARED,
849	* and XFS_ILOCK_EXCL are valid values to set in lock_flags.	848	* and XFS_ILOCK_EXCL are valid values to set in lock_flags.
850	*/	849	*/
851	ASSERT((lock_flags & (XFS_IOLOCK_SHARED \| XFS_IOLOCK_EXCL)) !=	850	ASSERT((lock_flags & (XFS_IOLOCK_SHARED \| XFS_IOLOCK_EXCL)) !=
852	(XFS_IOLOCK_SHARED \| XFS_IOLOCK_EXCL));	851	(XFS_IOLOCK_SHARED \| XFS_IOLOCK_EXCL));
853	ASSERT((lock_flags & (XFS_ILOCK_SHARED \| XFS_ILOCK_EXCL)) !=	852	ASSERT((lock_flags & (XFS_ILOCK_SHARED \| XFS_ILOCK_EXCL)) !=
854	(XFS_ILOCK_SHARED \| XFS_ILOCK_EXCL));	853	(XFS_ILOCK_SHARED \| XFS_ILOCK_EXCL));
855	ASSERT((lock_flags & ~XFS_LOCK_MASK) == 0);	854	ASSERT((lock_flags & ~XFS_LOCK_MASK) == 0);
856		855
857	if (lock_flags & XFS_IOLOCK_EXCL) {	856	if (lock_flags & XFS_IOLOCK_EXCL) {
858	mrupdate(&ip->i_iolock);	857	mrupdate(&ip->i_iolock);
859	} else if (lock_flags & XFS_IOLOCK_SHARED) {	858	} else if (lock_flags & XFS_IOLOCK_SHARED) {
860	mraccess(&ip->i_iolock);	859	mraccess(&ip->i_iolock);
861	}	860	}
862	if (lock_flags & XFS_ILOCK_EXCL) {	861	if (lock_flags & XFS_ILOCK_EXCL) {
863	mrupdate(&ip->i_lock);	862	mrupdate(&ip->i_lock);
864	} else if (lock_flags & XFS_ILOCK_SHARED) {	863	} else if (lock_flags & XFS_ILOCK_SHARED) {
865	mraccess(&ip->i_lock);	864	mraccess(&ip->i_lock);
866	}	865	}
867	xfs_ilock_trace(ip, 1, lock_flags, (inst_t *)__return_address);	866	xfs_ilock_trace(ip, 1, lock_flags, (inst_t *)__return_address);
868	}	867	}
869		868
870	/*	869	/*
871	* This is just like xfs_ilock(), except that the caller	870	* This is just like xfs_ilock(), except that the caller
872	* is guaranteed not to sleep. It returns 1 if it gets	871	* is guaranteed not to sleep. It returns 1 if it gets
873	* the requested locks and 0 otherwise. If the IO lock is	872	* the requested locks and 0 otherwise. If the IO lock is
874	* obtained but the inode lock cannot be, then the IO lock	873	* obtained but the inode lock cannot be, then the IO lock
875	* is dropped before returning.	874	* is dropped before returning.
876	*	875	*
877	* ip -- the inode being locked	876	* ip -- the inode being locked
878	* lock_flags -- this parameter indicates the inode's locks to be	877	* lock_flags -- this parameter indicates the inode's locks to be
879	* to be locked. See the comment for xfs_ilock() for a list	878	* to be locked. See the comment for xfs_ilock() for a list
880	* of valid values.	879	* of valid values.
881	*	880	*
882	*/	881	*/
883	int	882	int
884	xfs_ilock_nowait(xfs_inode_t *ip,	883	xfs_ilock_nowait(xfs_inode_t *ip,
885	uint lock_flags)	884	uint lock_flags)
886	{	885	{
887	int iolocked;	886	int iolocked;
888	int ilocked;	887	int ilocked;
889		888
890	/*	889	/*
891	* You can't set both SHARED and EXCL for the same lock,	890	* You can't set both SHARED and EXCL for the same lock,
892	* and only XFS_IOLOCK_SHARED, XFS_IOLOCK_EXCL, XFS_ILOCK_SHARED,	891	* and only XFS_IOLOCK_SHARED, XFS_IOLOCK_EXCL, XFS_ILOCK_SHARED,
893	* and XFS_ILOCK_EXCL are valid values to set in lock_flags.	892	* and XFS_ILOCK_EXCL are valid values to set in lock_flags.
894	*/	893	*/
895	ASSERT((lock_flags & (XFS_IOLOCK_SHARED \| XFS_IOLOCK_EXCL)) !=	894	ASSERT((lock_flags & (XFS_IOLOCK_SHARED \| XFS_IOLOCK_EXCL)) !=
896	(XFS_IOLOCK_SHARED \| XFS_IOLOCK_EXCL));	895	(XFS_IOLOCK_SHARED \| XFS_IOLOCK_EXCL));
897	ASSERT((lock_flags & (XFS_ILOCK_SHARED \| XFS_ILOCK_EXCL)) !=	896	ASSERT((lock_flags & (XFS_ILOCK_SHARED \| XFS_ILOCK_EXCL)) !=
898	(XFS_ILOCK_SHARED \| XFS_ILOCK_EXCL));	897	(XFS_ILOCK_SHARED \| XFS_ILOCK_EXCL));
899	ASSERT((lock_flags & ~XFS_LOCK_MASK) == 0);	898	ASSERT((lock_flags & ~XFS_LOCK_MASK) == 0);
900		899
901	iolocked = 0;	900	iolocked = 0;
902	if (lock_flags & XFS_IOLOCK_EXCL) {	901	if (lock_flags & XFS_IOLOCK_EXCL) {
903	iolocked = mrtryupdate(&ip->i_iolock);	902	iolocked = mrtryupdate(&ip->i_iolock);
904	if (!iolocked) {	903	if (!iolocked) {
905	return 0;	904	return 0;
906	}	905	}
907	} else if (lock_flags & XFS_IOLOCK_SHARED) {	906	} else if (lock_flags & XFS_IOLOCK_SHARED) {
908	iolocked = mrtryaccess(&ip->i_iolock);	907	iolocked = mrtryaccess(&ip->i_iolock);
909	if (!iolocked) {	908	if (!iolocked) {
910	return 0;	909	return 0;
911	}	910	}
912	}	911	}
913	if (lock_flags & XFS_ILOCK_EXCL) {	912	if (lock_flags & XFS_ILOCK_EXCL) {
914	ilocked = mrtryupdate(&ip->i_lock);	913	ilocked = mrtryupdate(&ip->i_lock);
915	if (!ilocked) {	914	if (!ilocked) {
916	if (iolocked) {	915	if (iolocked) {
917	mrunlock(&ip->i_iolock);	916	mrunlock(&ip->i_iolock);
918	}	917	}
919	return 0;	918	return 0;
920	}	919	}
921	} else if (lock_flags & XFS_ILOCK_SHARED) {	920	} else if (lock_flags & XFS_ILOCK_SHARED) {
922	ilocked = mrtryaccess(&ip->i_lock);	921	ilocked = mrtryaccess(&ip->i_lock);
923	if (!ilocked) {	922	if (!ilocked) {
924	if (iolocked) {	923	if (iolocked) {
925	mrunlock(&ip->i_iolock);	924	mrunlock(&ip->i_iolock);
926	}	925	}
927	return 0;	926	return 0;
928	}	927	}
929	}	928	}
930	xfs_ilock_trace(ip, 2, lock_flags, (inst_t *)__return_address);	929	xfs_ilock_trace(ip, 2, lock_flags, (inst_t *)__return_address);
931	return 1;	930	return 1;
932	}	931	}
933		932
934	/*	933	/*
935	* xfs_iunlock() is used to drop the inode locks acquired with	934	* xfs_iunlock() is used to drop the inode locks acquired with
936	* xfs_ilock() and xfs_ilock_nowait(). The caller must pass	935	* xfs_ilock() and xfs_ilock_nowait(). The caller must pass
937	* in the flags given to xfs_ilock() or xfs_ilock_nowait() so	936	* in the flags given to xfs_ilock() or xfs_ilock_nowait() so
938	* that we know which locks to drop.	937	* that we know which locks to drop.
939	*	938	*
940	* ip -- the inode being unlocked	939	* ip -- the inode being unlocked
941	* lock_flags -- this parameter indicates the inode's locks to be	940	* lock_flags -- this parameter indicates the inode's locks to be
942	* to be unlocked. See the comment for xfs_ilock() for a list	941	* to be unlocked. See the comment for xfs_ilock() for a list
943	* of valid values for this parameter.	942	* of valid values for this parameter.
944	*	943	*
945	*/	944	*/
946	void	945	void
947	xfs_iunlock(xfs_inode_t *ip,	946	xfs_iunlock(xfs_inode_t *ip,
948	uint lock_flags)	947	uint lock_flags)
949	{	948	{
950	/*	949	/*
951	* You can't set both SHARED and EXCL for the same lock,	950	* You can't set both SHARED and EXCL for the same lock,
952	* and only XFS_IOLOCK_SHARED, XFS_IOLOCK_EXCL, XFS_ILOCK_SHARED,	951	* and only XFS_IOLOCK_SHARED, XFS_IOLOCK_EXCL, XFS_ILOCK_SHARED,
953	* and XFS_ILOCK_EXCL are valid values to set in lock_flags.	952	* and XFS_ILOCK_EXCL are valid values to set in lock_flags.
954	*/	953	*/
955	ASSERT((lock_flags & (XFS_IOLOCK_SHARED \| XFS_IOLOCK_EXCL)) !=	954	ASSERT((lock_flags & (XFS_IOLOCK_SHARED \| XFS_IOLOCK_EXCL)) !=
956	(XFS_IOLOCK_SHARED \| XFS_IOLOCK_EXCL));	955	(XFS_IOLOCK_SHARED \| XFS_IOLOCK_EXCL));
957	ASSERT((lock_flags & (XFS_ILOCK_SHARED \| XFS_ILOCK_EXCL)) !=	956	ASSERT((lock_flags & (XFS_ILOCK_SHARED \| XFS_ILOCK_EXCL)) !=
958	(XFS_ILOCK_SHARED \| XFS_ILOCK_EXCL));	957	(XFS_ILOCK_SHARED \| XFS_ILOCK_EXCL));
959	ASSERT((lock_flags & ~(XFS_LOCK_MASK \| XFS_IUNLOCK_NONOTIFY)) == 0);	958	ASSERT((lock_flags & ~(XFS_LOCK_MASK \| XFS_IUNLOCK_NONOTIFY)) == 0);
960	ASSERT(lock_flags != 0);	959	ASSERT(lock_flags != 0);
961		960
962	if (lock_flags & (XFS_IOLOCK_SHARED \| XFS_IOLOCK_EXCL)) {	961	if (lock_flags & (XFS_IOLOCK_SHARED \| XFS_IOLOCK_EXCL)) {
963	ASSERT(!(lock_flags & XFS_IOLOCK_SHARED) \|\|	962	ASSERT(!(lock_flags & XFS_IOLOCK_SHARED) \|\|
964	(ismrlocked(&ip->i_iolock, MR_ACCESS)));	963	(ismrlocked(&ip->i_iolock, MR_ACCESS)));
965	ASSERT(!(lock_flags & XFS_IOLOCK_EXCL) \|\|	964	ASSERT(!(lock_flags & XFS_IOLOCK_EXCL) \|\|
966	(ismrlocked(&ip->i_iolock, MR_UPDATE)));	965	(ismrlocked(&ip->i_iolock, MR_UPDATE)));
967	mrunlock(&ip->i_iolock);	966	mrunlock(&ip->i_iolock);
968	}	967	}
969		968
970	if (lock_flags & (XFS_ILOCK_SHARED \| XFS_ILOCK_EXCL)) {	969	if (lock_flags & (XFS_ILOCK_SHARED \| XFS_ILOCK_EXCL)) {
971	ASSERT(!(lock_flags & XFS_ILOCK_SHARED) \|\|	970	ASSERT(!(lock_flags & XFS_ILOCK_SHARED) \|\|
972	(ismrlocked(&ip->i_lock, MR_ACCESS)));	971	(ismrlocked(&ip->i_lock, MR_ACCESS)));
973	ASSERT(!(lock_flags & XFS_ILOCK_EXCL) \|\|	972	ASSERT(!(lock_flags & XFS_ILOCK_EXCL) \|\|
974	(ismrlocked(&ip->i_lock, MR_UPDATE)));	973	(ismrlocked(&ip->i_lock, MR_UPDATE)));
975	mrunlock(&ip->i_lock);	974	mrunlock(&ip->i_lock);
976		975
977	/*	976	/*
978	* Let the AIL know that this item has been unlocked in case	977	* Let the AIL know that this item has been unlocked in case
979	* it is in the AIL and anyone is waiting on it. Don't do	978	* it is in the AIL and anyone is waiting on it. Don't do
980	* this if the caller has asked us not to.	979	* this if the caller has asked us not to.
981	*/	980	*/
982	if (!(lock_flags & XFS_IUNLOCK_NONOTIFY) &&	981	if (!(lock_flags & XFS_IUNLOCK_NONOTIFY) &&
983	ip->i_itemp != NULL) {	982	ip->i_itemp != NULL) {
984	xfs_trans_unlocked_item(ip->i_mount,	983	xfs_trans_unlocked_item(ip->i_mount,
985	(xfs_log_item_t*)(ip->i_itemp));	984	(xfs_log_item_t*)(ip->i_itemp));
986	}	985	}
987	}	986	}
988	xfs_ilock_trace(ip, 3, lock_flags, (inst_t *)__return_address);	987	xfs_ilock_trace(ip, 3, lock_flags, (inst_t *)__return_address);
989	}	988	}
990		989
991	/*	990	/*
992	* give up write locks. the i/o lock cannot be held nested	991	* give up write locks. the i/o lock cannot be held nested
993	* if it is being demoted.	992	* if it is being demoted.
994	*/	993	*/
995	void	994	void
996	xfs_ilock_demote(xfs_inode_t *ip,	995	xfs_ilock_demote(xfs_inode_t *ip,
997	uint lock_flags)	996	uint lock_flags)
998	{	997	{
999	ASSERT(lock_flags & (XFS_IOLOCK_EXCL\|XFS_ILOCK_EXCL));	998	ASSERT(lock_flags & (XFS_IOLOCK_EXCL\|XFS_ILOCK_EXCL));
1000	ASSERT((lock_flags & ~(XFS_IOLOCK_EXCL\|XFS_ILOCK_EXCL)) == 0);	999	ASSERT((lock_flags & ~(XFS_IOLOCK_EXCL\|XFS_ILOCK_EXCL)) == 0);
1001		1000
1002	if (lock_flags & XFS_ILOCK_EXCL) {	1001	if (lock_flags & XFS_ILOCK_EXCL) {
1003	ASSERT(ismrlocked(&ip->i_lock, MR_UPDATE));	1002	ASSERT(ismrlocked(&ip->i_lock, MR_UPDATE));
1004	mrdemote(&ip->i_lock);	1003	mrdemote(&ip->i_lock);
1005	}	1004	}
1006	if (lock_flags & XFS_IOLOCK_EXCL) {	1005	if (lock_flags & XFS_IOLOCK_EXCL) {
1007	ASSERT(ismrlocked(&ip->i_iolock, MR_UPDATE));	1006	ASSERT(ismrlocked(&ip->i_iolock, MR_UPDATE));
1008	mrdemote(&ip->i_iolock);	1007	mrdemote(&ip->i_iolock);
1009	}	1008	}
1010	}	1009	}
1011		1010
1012	/*	1011	/*
1013	* The following three routines simply manage the i_flock	1012	* The following three routines simply manage the i_flock
1014	* semaphore embedded in the inode. This semaphore synchronizes	1013	* semaphore embedded in the inode. This semaphore synchronizes
1015	* processes attempting to flush the in-core inode back to disk.	1014	* processes attempting to flush the in-core inode back to disk.
1016	*/	1015	*/
1017	void	1016	void
1018	xfs_iflock(xfs_inode_t *ip)	1017	xfs_iflock(xfs_inode_t *ip)
1019	{	1018	{
1020	psema(&(ip->i_flock), PINOD\|PLTWAIT);	1019	psema(&(ip->i_flock), PINOD\|PLTWAIT);
1021	}	1020	}
1022		1021
1023	int	1022	int
1024	xfs_iflock_nowait(xfs_inode_t *ip)	1023	xfs_iflock_nowait(xfs_inode_t *ip)
1025	{	1024	{
1026	return (cpsema(&(ip->i_flock)));	1025	return (cpsema(&(ip->i_flock)));
1027	}	1026	}
1028		1027
1029	void	1028	void
1030	xfs_ifunlock(xfs_inode_t *ip)	1029	xfs_ifunlock(xfs_inode_t *ip)
1031	{	1030	{
1032	ASSERT(issemalocked(&(ip->i_flock)));	1031	ASSERT(issemalocked(&(ip->i_flock)));
1033	vsema(&(ip->i_flock));	1032	vsema(&(ip->i_flock));
1034	}	1033	}

fs/xfs/xfs_inode.c

Diff comments View file @ 745b1f4

 /*
  * 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_imap.h"
 #include "xfs_trans.h"
 #include "xfs_trans_priv.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_buf_item.h"
 #include "xfs_inode_item.h"
 #include "xfs_btree.h"
 #include "xfs_alloc.h"
 #include "xfs_ialloc.h"
 #include "xfs_bmap.h"
 #include "xfs_rw.h"
 #include "xfs_error.h"
 #include "xfs_utils.h"
 #include "xfs_dir2_trace.h"
 #include "xfs_quota.h"
 #include "xfs_mac.h"
 #include "xfs_acl.h"
 kmem_zone_t *xfs_ifork_zone;
 kmem_zone_t *xfs_inode_zone;
 kmem_zone_t *xfs_chashlist_zone;
 /*
  * Used in xfs_itruncate().  This is the maximum number of extents
  * freed from a file in a single transaction.
  */
 #define	XFS_ITRUNC_MAX_EXTENTS	2
 STATIC int xfs_iflush_int(xfs_inode_t *, xfs_buf_t *);
 STATIC int xfs_iformat_local(xfs_inode_t *, xfs_dinode_t *, int, int);
 STATIC int xfs_iformat_extents(xfs_inode_t *, xfs_dinode_t *, int);
 STATIC int xfs_iformat_btree(xfs_inode_t *, xfs_dinode_t *, int);
 #ifdef DEBUG
 /*
  * Make sure that the extents in the given memory buffer
  * are valid.
  */
 STATIC void
 xfs_validate_extents(
 	xfs_ifork_t		*ifp,
 	int			nrecs,
 	int			disk,
 	xfs_exntfmt_t		fmt)
 {
 	xfs_bmbt_rec_t		*ep;
 	xfs_bmbt_irec_t		irec;
 	xfs_bmbt_rec_t		rec;
 	int			i;
 	for (i = 0; i < nrecs; i++) {
 		ep = xfs_iext_get_ext(ifp, i);
 		rec.l0 = get_unaligned((__uint64_t*)&ep->l0);
 		rec.l1 = get_unaligned((__uint64_t*)&ep->l1);
 		if (disk)
 			xfs_bmbt_disk_get_all(&rec, &irec);
 		else
 			xfs_bmbt_get_all(&rec, &irec);
 		if (fmt == XFS_EXTFMT_NOSTATE)
 			ASSERT(irec.br_state == XFS_EXT_NORM);
 	}
 }
 #else /* DEBUG */
 #define xfs_validate_extents(ifp, nrecs, disk, fmt)
 #endif /* DEBUG */
 /*
  * Check that none of the inode's in the buffer have a next
  * unlinked field of 0.
  */
 #if defined(DEBUG)
 void
 xfs_inobp_check(
 	xfs_mount_t	*mp,
 	xfs_buf_t	*bp)
 {
 	int		i;
 	int		j;
 	xfs_dinode_t	*dip;
 	j = mp->m_inode_cluster_size >> mp->m_sb.sb_inodelog;
 	for (i = 0; i < j; i++) {
 		dip = (xfs_dinode_t *)xfs_buf_offset(bp,
 					i * mp->m_sb.sb_inodesize);
 		if (!dip->di_next_unlinked)  {
 			xfs_fs_cmn_err(CE_ALERT, mp,
 				"Detected a bogus zero next_unlinked field in incore inode buffer 0x%p.  About to pop an ASSERT.",
 				bp);
 			ASSERT(dip->di_next_unlinked);
 		}
 	}
 }
 #endif
 /*
  * This routine is called to map an inode number within a file
  * system to the buffer containing the on-disk version of the
  * inode.  It returns a pointer to the buffer containing the
  * on-disk inode in the bpp parameter, and in the dip parameter
  * it returns a pointer to the on-disk inode within that buffer.
  *
  * If a non-zero error is returned, then the contents of bpp and
  * dipp are undefined.
  *
  * Use xfs_imap() to determine the size and location of the
  * buffer to read from disk.
  */
 STATIC int
 xfs_inotobp(
 	xfs_mount_t	*mp,
 	xfs_trans_t	*tp,
 	xfs_ino_t	ino,
 	xfs_dinode_t	**dipp,
 	xfs_buf_t	**bpp,
 	int		*offset)
 {
 	int		di_ok;
 	xfs_imap_t	imap;
 	xfs_buf_t	*bp;
 	int		error;
 	xfs_dinode_t	*dip;
 	/*
 	 * Call the space management code to find the location of the
 	 * inode on disk.
 	 */
 	imap.im_blkno = 0;
 	error = xfs_imap(mp, tp, ino, &imap, XFS_IMAP_LOOKUP);
 	if (error != 0) {
 		cmn_err(CE_WARN,
 	"xfs_inotobp: xfs_imap()  returned an "
 	"error %d on %s.  Returning error.", error, mp->m_fsname);
 		return error;
 	}
 	/*
 	 * If the inode number maps to a block outside the bounds of the
 	 * file system then return NULL rather than calling read_buf
 	 * and panicing when we get an error from the driver.
 	 */
 	if ((imap.im_blkno + imap.im_len) >
 	    XFS_FSB_TO_BB(mp, mp->m_sb.sb_dblocks)) {
 		cmn_err(CE_WARN,
 	"xfs_inotobp: inode number (%llu + %d) maps to a block outside the bounds "
 	"of the file system %s.  Returning EINVAL.",
 			(unsigned long long)imap.im_blkno,
 			imap.im_len, mp->m_fsname);
 		return XFS_ERROR(EINVAL);
 	}
 	/*
 	 * Read in the buffer.  If tp is NULL, xfs_trans_read_buf() will
 	 * default to just a read_buf() call.
 	 */
 	error = xfs_trans_read_buf(mp, tp, mp->m_ddev_targp, imap.im_blkno,
 				   (int)imap.im_len, XFS_BUF_LOCK, &bp);
 	if (error) {
 		cmn_err(CE_WARN,
 	"xfs_inotobp: xfs_trans_read_buf()  returned an "
 	"error %d on %s.  Returning error.", error, mp->m_fsname);
 		return error;
 	}
 	dip = (xfs_dinode_t *)xfs_buf_offset(bp, 0);
 	di_ok =
 		INT_GET(dip->di_core.di_magic, ARCH_CONVERT) == XFS_DINODE_MAGIC &&
 		XFS_DINODE_GOOD_VERSION(INT_GET(dip->di_core.di_version, ARCH_CONVERT));
 	if (unlikely(XFS_TEST_ERROR(!di_ok, mp, XFS_ERRTAG_ITOBP_INOTOBP,
 			XFS_RANDOM_ITOBP_INOTOBP))) {
 		XFS_CORRUPTION_ERROR("xfs_inotobp", XFS_ERRLEVEL_LOW, mp, dip);
 		xfs_trans_brelse(tp, bp);
 		cmn_err(CE_WARN,
 	"xfs_inotobp: XFS_TEST_ERROR()  returned an "
 	"error on %s.  Returning EFSCORRUPTED.",  mp->m_fsname);
 		return XFS_ERROR(EFSCORRUPTED);
 	}
 	xfs_inobp_check(mp, bp);
 	/*
 	 * Set *dipp to point to the on-disk inode in the buffer.
 	 */
 	*dipp = (xfs_dinode_t *)xfs_buf_offset(bp, imap.im_boffset);
 	*bpp = bp;
 	*offset = imap.im_boffset;
 	return 0;
 }
 /*
  * This routine is called to map an inode to the buffer containing
  * the on-disk version of the inode.  It returns a pointer to the
  * buffer containing the on-disk inode in the bpp parameter, and in
  * the dip parameter it returns a pointer to the on-disk inode within
  * that buffer.
  *
  * If a non-zero error is returned, then the contents of bpp and
  * dipp are undefined.
  *
  * If the inode is new and has not yet been initialized, use xfs_imap()
  * to determine the size and location of the buffer to read from disk.
  * If the inode has already been mapped to its buffer and read in once,
  * then use the mapping information stored in the inode rather than
  * calling xfs_imap().  This allows us to avoid the overhead of looking
  * at the inode btree for small block file systems (see xfs_dilocate()).
  * We can tell whether the inode has been mapped in before by comparing
  * its disk block address to 0.  Only uninitialized inodes will have
  * 0 for the disk block address.
  */
 int
 xfs_itobp(
 	xfs_mount_t	*mp,
 	xfs_trans_t	*tp,
 	xfs_inode_t	*ip,
 	xfs_dinode_t	**dipp,
 	xfs_buf_t	**bpp,
 	xfs_daddr_t	bno,
 	uint		imap_flags)
 {
 	xfs_imap_t	imap;
 	xfs_buf_t	*bp;
 	int		error;
 	int		i;
 	int		ni;
 	if (ip->i_blkno == (xfs_daddr_t)0) {
 		/*
 		 * Call the space management code to find the location of the
 		 * inode on disk.
 		 */
 		imap.im_blkno = bno;
 		if ((error = xfs_imap(mp, tp, ip->i_ino, &imap,
 					XFS_IMAP_LOOKUP | imap_flags)))
 			return error;
 		/*
 		 * If the inode number maps to a block outside the bounds
 		 * of the file system then return NULL rather than calling
 		 * read_buf and panicing when we get an error from the
 		 * driver.
 		 */
 		if ((imap.im_blkno + imap.im_len) >
 		    XFS_FSB_TO_BB(mp, mp->m_sb.sb_dblocks)) {
 #ifdef DEBUG
 			xfs_fs_cmn_err(CE_ALERT, mp, "xfs_itobp: "
 					"(imap.im_blkno (0x%llx) "
 					"+ imap.im_len (0x%llx)) > "
 					" XFS_FSB_TO_BB(mp, "
 					"mp->m_sb.sb_dblocks) (0x%llx)",
 					(unsigned long long) imap.im_blkno,
 					(unsigned long long) imap.im_len,
 					XFS_FSB_TO_BB(mp, mp->m_sb.sb_dblocks));
 #endif /* DEBUG */
 			return XFS_ERROR(EINVAL);
 		}
 		/*
 		 * Fill in the fields in the inode that will be used to
 		 * map the inode to its buffer from now on.
 		 */
 		ip->i_blkno = imap.im_blkno;
 		ip->i_len = imap.im_len;
 		ip->i_boffset = imap.im_boffset;
 	} else {
 		/*
 		 * We've already mapped the inode once, so just use the
 		 * mapping that we saved the first time.
 		 */
 		imap.im_blkno = ip->i_blkno;
 		imap.im_len = ip->i_len;
 		imap.im_boffset = ip->i_boffset;
 	}
 	ASSERT(bno == 0 || bno == imap.im_blkno);
 	/*
 	 * Read in the buffer.  If tp is NULL, xfs_trans_read_buf() will
 	 * default to just a read_buf() call.
 	 */
 	error = xfs_trans_read_buf(mp, tp, mp->m_ddev_targp, imap.im_blkno,
 				   (int)imap.im_len, XFS_BUF_LOCK, &bp);
 	if (error) {
 #ifdef DEBUG
 		xfs_fs_cmn_err(CE_ALERT, mp, "xfs_itobp: "
 				"xfs_trans_read_buf() returned error %d, "
 				"imap.im_blkno 0x%llx, imap.im_len 0x%llx",
 				error, (unsigned long long) imap.im_blkno,
 				(unsigned long long) imap.im_len);
 #endif /* DEBUG */
 		return error;
 	}
 	/*
 	 * Validate the magic number and version of every inode in the buffer
 	 * (if DEBUG kernel) or the first inode in the buffer, otherwise.
 	 * No validation is done here in userspace (xfs_repair).
 	 */
 #if !defined(__KERNEL__)
 	ni = 0;
 #elif defined(DEBUG)
 	ni = BBTOB(imap.im_len) >> mp->m_sb.sb_inodelog;
 #else	/* usual case */
 	ni = 1;
 #endif
 	for (i = 0; i < ni; i++) {
 		int		di_ok;
 		xfs_dinode_t	*dip;
 		dip = (xfs_dinode_t *)xfs_buf_offset(bp,
 					(i << mp->m_sb.sb_inodelog));
 		di_ok = INT_GET(dip->di_core.di_magic, ARCH_CONVERT) == XFS_DINODE_MAGIC &&
 			    XFS_DINODE_GOOD_VERSION(INT_GET(dip->di_core.di_version, ARCH_CONVERT));
 		if (unlikely(XFS_TEST_ERROR(!di_ok, mp,
 						XFS_ERRTAG_ITOBP_INOTOBP,
 						XFS_RANDOM_ITOBP_INOTOBP))) {
 			if (imap_flags & XFS_IMAP_BULKSTAT) {
 				xfs_trans_brelse(tp, bp);
 				return XFS_ERROR(EINVAL);
 			}
 #ifdef DEBUG
 			cmn_err(CE_ALERT,
 					"Device %s - bad inode magic/vsn "
 					"daddr %lld #%d (magic=%x)",
 				XFS_BUFTARG_NAME(mp->m_ddev_targp),
 				(unsigned long long)imap.im_blkno, i,
 				INT_GET(dip->di_core.di_magic, ARCH_CONVERT));
 #endif
 			XFS_CORRUPTION_ERROR("xfs_itobp", XFS_ERRLEVEL_HIGH,
 					     mp, dip);
 			xfs_trans_brelse(tp, bp);
 			return XFS_ERROR(EFSCORRUPTED);
 		}
 	}
 	xfs_inobp_check(mp, bp);
 	/*
 	 * Mark the buffer as an inode buffer now that it looks good
 	 */
 	XFS_BUF_SET_VTYPE(bp, B_FS_INO);
 	/*
 	 * Set *dipp to point to the on-disk inode in the buffer.
 	 */
 	*dipp = (xfs_dinode_t *)xfs_buf_offset(bp, imap.im_boffset);
 	*bpp = bp;
 	return 0;
 }
 /*
  * Move inode type and inode format specific information from the
  * on-disk inode to the in-core inode.  For fifos, devs, and sockets
  * this means set if_rdev to the proper value.  For files, directories,
  * and symlinks this means to bring in the in-line data or extent
  * pointers.  For a file in B-tree format, only the root is immediately
  * brought in-core.  The rest will be in-lined in if_extents when it
  * is first referenced (see xfs_iread_extents()).
  */
 STATIC int
 xfs_iformat(
 	xfs_inode_t		*ip,
 	xfs_dinode_t		*dip)
 {
 	xfs_attr_shortform_t	*atp;
 	int			size;
 	int			error;
 	xfs_fsize_t             di_size;
 	ip->i_df.if_ext_max =
 		XFS_IFORK_DSIZE(ip) / (uint)sizeof(xfs_bmbt_rec_t);
 	error = 0;
 	if (unlikely(
 	    INT_GET(dip->di_core.di_nextents, ARCH_CONVERT) +
 		INT_GET(dip->di_core.di_anextents, ARCH_CONVERT) >
 	    INT_GET(dip->di_core.di_nblocks, ARCH_CONVERT))) {
 		xfs_fs_repair_cmn_err(CE_WARN, ip->i_mount,
 			"corrupt dinode %Lu, extent total = %d, nblocks = %Lu.",
 			(unsigned long long)ip->i_ino,
 			(int)(INT_GET(dip->di_core.di_nextents, ARCH_CONVERT)
 			    + INT_GET(dip->di_core.di_anextents, ARCH_CONVERT)),
 			(unsigned long long)
 			INT_GET(dip->di_core.di_nblocks, ARCH_CONVERT));
 		XFS_CORRUPTION_ERROR("xfs_iformat(1)", XFS_ERRLEVEL_LOW,
 				     ip->i_mount, dip);
 		return XFS_ERROR(EFSCORRUPTED);
 	}
 	if (unlikely(INT_GET(dip->di_core.di_forkoff, ARCH_CONVERT) > ip->i_mount->m_sb.sb_inodesize)) {
 		xfs_fs_repair_cmn_err(CE_WARN, ip->i_mount,
 			"corrupt dinode %Lu, forkoff = 0x%x.",
 			(unsigned long long)ip->i_ino,
 			(int)(INT_GET(dip->di_core.di_forkoff, ARCH_CONVERT)));
 		XFS_CORRUPTION_ERROR("xfs_iformat(2)", XFS_ERRLEVEL_LOW,
 				     ip->i_mount, dip);
 		return XFS_ERROR(EFSCORRUPTED);
 	}
 	switch (ip->i_d.di_mode & S_IFMT) {
 	case S_IFIFO:
 	case S_IFCHR:
 	case S_IFBLK:
 	case S_IFSOCK:
 		if (unlikely(INT_GET(dip->di_core.di_format, ARCH_CONVERT) != XFS_DINODE_FMT_DEV)) {
 			XFS_CORRUPTION_ERROR("xfs_iformat(3)", XFS_ERRLEVEL_LOW,
 					      ip->i_mount, dip);
 			return XFS_ERROR(EFSCORRUPTED);
 		}
 		ip->i_d.di_size = 0;
 		ip->i_df.if_u2.if_rdev = INT_GET(dip->di_u.di_dev, ARCH_CONVERT);
 		break;
 	case S_IFREG:
 	case S_IFLNK:
 	case S_IFDIR:
 		switch (INT_GET(dip->di_core.di_format, ARCH_CONVERT)) {
 		case XFS_DINODE_FMT_LOCAL:
 			/*
 			 * no local regular files yet
 			 */
 			if (unlikely((INT_GET(dip->di_core.di_mode, ARCH_CONVERT) & S_IFMT) == S_IFREG)) {
 				xfs_fs_repair_cmn_err(CE_WARN, ip->i_mount,
 					"corrupt inode %Lu "
 					"(local format for regular file).",
 					(unsigned long long) ip->i_ino);
 				XFS_CORRUPTION_ERROR("xfs_iformat(4)",
 						     XFS_ERRLEVEL_LOW,
 						     ip->i_mount, dip);
 				return XFS_ERROR(EFSCORRUPTED);
 			}
 			di_size = INT_GET(dip->di_core.di_size, ARCH_CONVERT);
 			if (unlikely(di_size > XFS_DFORK_DSIZE(dip, ip->i_mount))) {
 				xfs_fs_repair_cmn_err(CE_WARN, ip->i_mount,
 					"corrupt inode %Lu "
 					"(bad size %Ld for local inode).",
 					(unsigned long long) ip->i_ino,
 					(long long) di_size);
 				XFS_CORRUPTION_ERROR("xfs_iformat(5)",
 						     XFS_ERRLEVEL_LOW,
 						     ip->i_mount, dip);
 				return XFS_ERROR(EFSCORRUPTED);
 			}
 			size = (int)di_size;
 			error = xfs_iformat_local(ip, dip, XFS_DATA_FORK, size);
 			break;
 		case XFS_DINODE_FMT_EXTENTS:
 			error = xfs_iformat_extents(ip, dip, XFS_DATA_FORK);
 			break;
 		case XFS_DINODE_FMT_BTREE:
 			error = xfs_iformat_btree(ip, dip, XFS_DATA_FORK);
 			break;
 		default:
 			XFS_ERROR_REPORT("xfs_iformat(6)", XFS_ERRLEVEL_LOW,
 					 ip->i_mount);
 			return XFS_ERROR(EFSCORRUPTED);
 		}
 		break;
 	default:
 		XFS_ERROR_REPORT("xfs_iformat(7)", XFS_ERRLEVEL_LOW, ip->i_mount);
 		return XFS_ERROR(EFSCORRUPTED);
 	}
 	if (error) {
 		return error;
 	}
 	if (!XFS_DFORK_Q(dip))
 		return 0;
 	ASSERT(ip->i_afp == NULL);
 	ip->i_afp = kmem_zone_zalloc(xfs_ifork_zone, KM_SLEEP);
 	ip->i_afp->if_ext_max =
 		XFS_IFORK_ASIZE(ip) / (uint)sizeof(xfs_bmbt_rec_t);
 	switch (INT_GET(dip->di_core.di_aformat, ARCH_CONVERT)) {
 	case XFS_DINODE_FMT_LOCAL:
 		atp = (xfs_attr_shortform_t *)XFS_DFORK_APTR(dip);
 		size = be16_to_cpu(atp->hdr.totsize);
 		error = xfs_iformat_local(ip, dip, XFS_ATTR_FORK, size);
 		break;
 	case XFS_DINODE_FMT_EXTENTS:
 		error = xfs_iformat_extents(ip, dip, XFS_ATTR_FORK);
 		break;
 	case XFS_DINODE_FMT_BTREE:
 		error = xfs_iformat_btree(ip, dip, XFS_ATTR_FORK);
 		break;
 	default:
 		error = XFS_ERROR(EFSCORRUPTED);
 		break;
 	}
 	if (error) {
 		kmem_zone_free(xfs_ifork_zone, ip->i_afp);
 		ip->i_afp = NULL;
 		xfs_idestroy_fork(ip, XFS_DATA_FORK);
 	}
 	return error;
 }
 /*
  * The file is in-lined in the on-disk inode.
  * If it fits into if_inline_data, then copy
  * it there, otherwise allocate a buffer for it
  * and copy the data there.  Either way, set
  * if_data to point at the data.
  * If we allocate a buffer for the data, make
  * sure that its size is a multiple of 4 and
  * record the real size in i_real_bytes.
  */
 STATIC int
 xfs_iformat_local(
 	xfs_inode_t	*ip,
 	xfs_dinode_t	*dip,
 	int		whichfork,
 	int		size)
 {
 	xfs_ifork_t	*ifp;
 	int		real_size;
 	/*
 	 * If the size is unreasonable, then something
 	 * is wrong and we just bail out rather than crash in
 	 * kmem_alloc() or memcpy() below.
 	 */
 	if (unlikely(size > XFS_DFORK_SIZE(dip, ip->i_mount, whichfork))) {
 		xfs_fs_repair_cmn_err(CE_WARN, ip->i_mount,
 			"corrupt inode %Lu "
 			"(bad size %d for local fork, size = %d).",
 			(unsigned long long) ip->i_ino, size,
 			XFS_DFORK_SIZE(dip, ip->i_mount, whichfork));
 		XFS_CORRUPTION_ERROR("xfs_iformat_local", XFS_ERRLEVEL_LOW,
 				     ip->i_mount, dip);
 		return XFS_ERROR(EFSCORRUPTED);
 	}
 	ifp = XFS_IFORK_PTR(ip, whichfork);
 	real_size = 0;
 	if (size == 0)
 		ifp->if_u1.if_data = NULL;
 	else if (size <= sizeof(ifp->if_u2.if_inline_data))
 		ifp->if_u1.if_data = ifp->if_u2.if_inline_data;
 	else {
 		real_size = roundup(size, 4);
 		ifp->if_u1.if_data = kmem_alloc(real_size, KM_SLEEP);
 	}
 	ifp->if_bytes = size;
 	ifp->if_real_bytes = real_size;
 	if (size)
 		memcpy(ifp->if_u1.if_data, XFS_DFORK_PTR(dip, whichfork), size);
 	ifp->if_flags &= ~XFS_IFEXTENTS;
 	ifp->if_flags |= XFS_IFINLINE;
 	return 0;
 }
 /*
  * The file consists of a set of extents all
  * of which fit into the on-disk inode.
  * If there are few enough extents to fit into
  * the if_inline_ext, then copy them there.
  * Otherwise allocate a buffer for them and copy
  * them into it.  Either way, set if_extents
  * to point at the extents.
  */
 STATIC int
 xfs_iformat_extents(
 	xfs_inode_t	*ip,
 	xfs_dinode_t	*dip,
 	int		whichfork)
 {
 	xfs_bmbt_rec_t	*ep, *dp;
 	xfs_ifork_t	*ifp;
 	int		nex;
 	int		size;
 	int		i;
 	ifp = XFS_IFORK_PTR(ip, whichfork);
 	nex = XFS_DFORK_NEXTENTS(dip, whichfork);
 	size = nex * (uint)sizeof(xfs_bmbt_rec_t);
 	/*
 	 * If the number of extents is unreasonable, then something
 	 * is wrong and we just bail out rather than crash in
 	 * kmem_alloc() or memcpy() below.
 	 */
 	if (unlikely(size < 0 || size > XFS_DFORK_SIZE(dip, ip->i_mount, whichfork))) {
 		xfs_fs_repair_cmn_err(CE_WARN, ip->i_mount,
 			"corrupt inode %Lu ((a)extents = %d).",
 			(unsigned long long) ip->i_ino, nex);
 		XFS_CORRUPTION_ERROR("xfs_iformat_extents(1)", XFS_ERRLEVEL_LOW,
 				     ip->i_mount, dip);
 		return XFS_ERROR(EFSCORRUPTED);
 	}
 	ifp->if_real_bytes = 0;
 	if (nex == 0)
 		ifp->if_u1.if_extents = NULL;
 	else if (nex <= XFS_INLINE_EXTS)
 		ifp->if_u1.if_extents = ifp->if_u2.if_inline_ext;
 	else
 		xfs_iext_add(ifp, 0, nex);
 	ifp->if_bytes = size;
 	if (size) {
 		dp = (xfs_bmbt_rec_t *) XFS_DFORK_PTR(dip, whichfork);
 		xfs_validate_extents(ifp, nex, 1, XFS_EXTFMT_INODE(ip));
 		for (i = 0; i < nex; i++, dp++) {
 			ep = xfs_iext_get_ext(ifp, i);
 			ep->l0 = INT_GET(get_unaligned((__uint64_t*)&dp->l0),
 								ARCH_CONVERT);
 			ep->l1 = INT_GET(get_unaligned((__uint64_t*)&dp->l1),
 								ARCH_CONVERT);
 		}
 		xfs_bmap_trace_exlist("xfs_iformat_extents", ip, nex,
 			whichfork);
 		if (whichfork != XFS_DATA_FORK ||
 			XFS_EXTFMT_INODE(ip) == XFS_EXTFMT_NOSTATE)
 				if (unlikely(xfs_check_nostate_extents(
 				    ifp, 0, nex))) {
 					XFS_ERROR_REPORT("xfs_iformat_extents(2)",
 							 XFS_ERRLEVEL_LOW,
 							 ip->i_mount);
 					return XFS_ERROR(EFSCORRUPTED);
 				}
 	}
 	ifp->if_flags |= XFS_IFEXTENTS;
 	return 0;
 }
 /*
  * The file has too many extents to fit into
  * the inode, so they are in B-tree format.
  * Allocate a buffer for the root of the B-tree
  * and copy the root into it.  The i_extents
  * field will remain NULL until all of the
  * extents are read in (when they are needed).
  */
 STATIC int
 xfs_iformat_btree(
 	xfs_inode_t		*ip,
 	xfs_dinode_t		*dip,
 	int			whichfork)
 {
 	xfs_bmdr_block_t	*dfp;
 	xfs_ifork_t		*ifp;
 	/* REFERENCED */
 	int			nrecs;
 	int			size;
 	ifp = XFS_IFORK_PTR(ip, whichfork);
 	dfp = (xfs_bmdr_block_t *)XFS_DFORK_PTR(dip, whichfork);
 	size = XFS_BMAP_BROOT_SPACE(dfp);
 	nrecs = XFS_BMAP_BROOT_NUMRECS(dfp);
 	/*
 	 * blow out if -- fork has less extents than can fit in
 	 * fork (fork shouldn't be a btree format), root btree
 	 * block has more records than can fit into the fork,
 	 * or the number of extents is greater than the number of
 	 * blocks.
 	 */
 	if (unlikely(XFS_IFORK_NEXTENTS(ip, whichfork) <= ifp->if_ext_max
 	    || XFS_BMDR_SPACE_CALC(nrecs) >
 			XFS_DFORK_SIZE(dip, ip->i_mount, whichfork)
 	    || XFS_IFORK_NEXTENTS(ip, whichfork) > ip->i_d.di_nblocks)) {
 		xfs_fs_repair_cmn_err(CE_WARN, ip->i_mount,
 			"corrupt inode %Lu (btree).",
 			(unsigned long long) ip->i_ino);
 		XFS_ERROR_REPORT("xfs_iformat_btree", XFS_ERRLEVEL_LOW,
 				 ip->i_mount);
 		return XFS_ERROR(EFSCORRUPTED);
 	}
 	ifp->if_broot_bytes = size;
 	ifp->if_broot = kmem_alloc(size, KM_SLEEP);
 	ASSERT(ifp->if_broot != NULL);
 	/*
 	 * Copy and convert from the on-disk structure
 	 * to the in-memory structure.
 	 */
 	xfs_bmdr_to_bmbt(dfp, XFS_DFORK_SIZE(dip, ip->i_mount, whichfork),
 		ifp->if_broot, size);
 	ifp->if_flags &= ~XFS_IFEXTENTS;
 	ifp->if_flags |= XFS_IFBROOT;
 	return 0;
 }
 /*
  * xfs_xlate_dinode_core - translate an xfs_inode_core_t between ondisk
  * and native format
  *
  * buf  = on-disk representation
  * dip  = native representation
  * dir  = direction - +ve -> disk to native
  *                    -ve -> native to disk
  */
 void
 xfs_xlate_dinode_core(
 	xfs_caddr_t		buf,
 	xfs_dinode_core_t	*dip,
 	int			dir)
 {
 	xfs_dinode_core_t	*buf_core = (xfs_dinode_core_t *)buf;
 	xfs_dinode_core_t	*mem_core = (xfs_dinode_core_t *)dip;
 	xfs_arch_t		arch = ARCH_CONVERT;
 	ASSERT(dir);
 	INT_XLATE(buf_core->di_magic, mem_core->di_magic, dir, arch);
 	INT_XLATE(buf_core->di_mode, mem_core->di_mode, dir, arch);
 	INT_XLATE(buf_core->di_version,	mem_core->di_version, dir, arch);
 	INT_XLATE(buf_core->di_format, mem_core->di_format, dir, arch);
 	INT_XLATE(buf_core->di_onlink, mem_core->di_onlink, dir, arch);
 	INT_XLATE(buf_core->di_uid, mem_core->di_uid, dir, arch);
 	INT_XLATE(buf_core->di_gid, mem_core->di_gid, dir, arch);
 	INT_XLATE(buf_core->di_nlink, mem_core->di_nlink, dir, arch);
 	INT_XLATE(buf_core->di_projid, mem_core->di_projid, dir, arch);
 	if (dir > 0) {
 		memcpy(mem_core->di_pad, buf_core->di_pad,
 			sizeof(buf_core->di_pad));
 	} else {
 		memcpy(buf_core->di_pad, mem_core->di_pad,
 			sizeof(buf_core->di_pad));
 	}
 	INT_XLATE(buf_core->di_flushiter, mem_core->di_flushiter, dir, arch);
 	INT_XLATE(buf_core->di_atime.t_sec, mem_core->di_atime.t_sec,
 			dir, arch);
 	INT_XLATE(buf_core->di_atime.t_nsec, mem_core->di_atime.t_nsec,
 			dir, arch);
 	INT_XLATE(buf_core->di_mtime.t_sec, mem_core->di_mtime.t_sec,
 			dir, arch);
 	INT_XLATE(buf_core->di_mtime.t_nsec, mem_core->di_mtime.t_nsec,
 			dir, arch);
 	INT_XLATE(buf_core->di_ctime.t_sec, mem_core->di_ctime.t_sec,
 			dir, arch);
 	INT_XLATE(buf_core->di_ctime.t_nsec, mem_core->di_ctime.t_nsec,
 			dir, arch);
 	INT_XLATE(buf_core->di_size, mem_core->di_size, dir, arch);
 	INT_XLATE(buf_core->di_nblocks, mem_core->di_nblocks, dir, arch);
 	INT_XLATE(buf_core->di_extsize, mem_core->di_extsize, dir, arch);
 	INT_XLATE(buf_core->di_nextents, mem_core->di_nextents, dir, arch);
 	INT_XLATE(buf_core->di_anextents, mem_core->di_anextents, dir, arch);
 	INT_XLATE(buf_core->di_forkoff, mem_core->di_forkoff, dir, arch);
 	INT_XLATE(buf_core->di_aformat, mem_core->di_aformat, dir, arch);
 	INT_XLATE(buf_core->di_dmevmask, mem_core->di_dmevmask, dir, arch);
 	INT_XLATE(buf_core->di_dmstate, mem_core->di_dmstate, dir, arch);
 	INT_XLATE(buf_core->di_flags, mem_core->di_flags, dir, arch);
 	INT_XLATE(buf_core->di_gen, mem_core->di_gen, dir, arch);
 }
 STATIC uint
 _xfs_dic2xflags(
 	__uint16_t		di_flags)
 {
 	uint			flags = 0;
 	if (di_flags & XFS_DIFLAG_ANY) {
 		if (di_flags & XFS_DIFLAG_REALTIME)
 			flags |= XFS_XFLAG_REALTIME;
 		if (di_flags & XFS_DIFLAG_PREALLOC)
 			flags |= XFS_XFLAG_PREALLOC;
 		if (di_flags & XFS_DIFLAG_IMMUTABLE)
 			flags |= XFS_XFLAG_IMMUTABLE;
 		if (di_flags & XFS_DIFLAG_APPEND)
 			flags |= XFS_XFLAG_APPEND;
 		if (di_flags & XFS_DIFLAG_SYNC)
 			flags |= XFS_XFLAG_SYNC;
 		if (di_flags & XFS_DIFLAG_NOATIME)
 			flags |= XFS_XFLAG_NOATIME;
 		if (di_flags & XFS_DIFLAG_NODUMP)
 			flags |= XFS_XFLAG_NODUMP;
 		if (di_flags & XFS_DIFLAG_RTINHERIT)
 			flags |= XFS_XFLAG_RTINHERIT;
 		if (di_flags & XFS_DIFLAG_PROJINHERIT)
 			flags |= XFS_XFLAG_PROJINHERIT;
 		if (di_flags & XFS_DIFLAG_NOSYMLINKS)
 			flags |= XFS_XFLAG_NOSYMLINKS;
 		if (di_flags & XFS_DIFLAG_EXTSIZE)
 			flags |= XFS_XFLAG_EXTSIZE;
 		if (di_flags & XFS_DIFLAG_EXTSZINHERIT)
 			flags |= XFS_XFLAG_EXTSZINHERIT;
 		if (di_flags & XFS_DIFLAG_NODEFRAG)
 			flags |= XFS_XFLAG_NODEFRAG;
 	}
 	return flags;
 }
 uint
 xfs_ip2xflags(
 	xfs_inode_t		*ip)
 {
 	xfs_dinode_core_t	*dic = &ip->i_d;
 	return _xfs_dic2xflags(dic->di_flags) |
 				(XFS_CFORK_Q(dic) ? XFS_XFLAG_HASATTR : 0);
 }
 uint
 xfs_dic2xflags(
 	xfs_dinode_core_t	*dic)
 {
 	return _xfs_dic2xflags(INT_GET(dic->di_flags, ARCH_CONVERT)) |
 				(XFS_CFORK_Q_DISK(dic) ? XFS_XFLAG_HASATTR : 0);
 }
 /*
  * Given a mount structure and an inode number, return a pointer
  * to a newly allocated in-core inode corresponding to the given
  * inode number.
  *
  * Initialize the inode's attributes and extent pointers if it
  * already has them (it will not if the inode has no links).
  */
 int
 xfs_iread(
 	xfs_mount_t	*mp,
 	xfs_trans_t	*tp,
 	xfs_ino_t	ino,
 	xfs_inode_t	**ipp,
-	xfs_daddr_t	bno)
+	xfs_daddr_t	bno,
+	uint		imap_flags)
 {
 	xfs_buf_t	*bp;
 	xfs_dinode_t	*dip;
 	xfs_inode_t	*ip;
 	int		error;
 	ASSERT(xfs_inode_zone != NULL);
 	ip = kmem_zone_zalloc(xfs_inode_zone, KM_SLEEP);
 	ip->i_ino = ino;
 	ip->i_mount = mp;
 	/*
 	 * Get pointer's to the on-disk inode and the buffer containing it.
 	 * If the inode number refers to a block outside the file system
 	 * then xfs_itobp() will return NULL.  In this case we should
 	 * return NULL as well.  Set i_blkno to 0 so that xfs_itobp() will
 	 * know that this is a new incore inode.
 	 */
-	error = xfs_itobp(mp, tp, ip, &dip, &bp, bno, 0);
+	error = xfs_itobp(mp, tp, ip, &dip, &bp, bno, imap_flags);
 	if (error) {
 		kmem_zone_free(xfs_inode_zone, ip);
 		return error;
 	}
 	/*
 	 * Initialize inode's trace buffers.
 	 * Do this before xfs_iformat in case it adds entries.
 	 */
 #ifdef XFS_BMAP_TRACE
 	ip->i_xtrace = ktrace_alloc(XFS_BMAP_KTRACE_SIZE, KM_SLEEP);
 #endif
 #ifdef XFS_BMBT_TRACE
 	ip->i_btrace = ktrace_alloc(XFS_BMBT_KTRACE_SIZE, KM_SLEEP);
 #endif
 #ifdef XFS_RW_TRACE
 	ip->i_rwtrace = ktrace_alloc(XFS_RW_KTRACE_SIZE, KM_SLEEP);
 #endif
 #ifdef XFS_ILOCK_TRACE
 	ip->i_lock_trace = ktrace_alloc(XFS_ILOCK_KTRACE_SIZE, KM_SLEEP);
 #endif
 #ifdef XFS_DIR2_TRACE
 	ip->i_dir_trace = ktrace_alloc(XFS_DIR2_KTRACE_SIZE, KM_SLEEP);
 #endif
 	/*
 	 * If we got something that isn't an inode it means someone
 	 * (nfs or dmi) has a stale handle.
 	 */
 	if (INT_GET(dip->di_core.di_magic, ARCH_CONVERT) != XFS_DINODE_MAGIC) {
 		kmem_zone_free(xfs_inode_zone, ip);
 		xfs_trans_brelse(tp, bp);
 #ifdef DEBUG
 		xfs_fs_cmn_err(CE_ALERT, mp, "xfs_iread: "
 				"dip->di_core.di_magic (0x%x) != "
 				"XFS_DINODE_MAGIC (0x%x)",
 				INT_GET(dip->di_core.di_magic, ARCH_CONVERT),
 				XFS_DINODE_MAGIC);
 #endif /* DEBUG */
 		return XFS_ERROR(EINVAL);
 	}
 	/*
 	 * If the on-disk inode is already linked to a directory
 	 * entry, copy all of the inode into the in-core inode.
 	 * xfs_iformat() handles copying in the inode format
 	 * specific information.
 	 * Otherwise, just get the truly permanent information.
 	 */
 	if (dip->di_core.di_mode) {
 		xfs_xlate_dinode_core((xfs_caddr_t)&dip->di_core,
 		     &(ip->i_d), 1);
 		error = xfs_iformat(ip, dip);
 		if (error)  {
 			kmem_zone_free(xfs_inode_zone, ip);
 			xfs_trans_brelse(tp, bp);
 #ifdef DEBUG
 			xfs_fs_cmn_err(CE_ALERT, mp, "xfs_iread: "
 					"xfs_iformat() returned error %d",
 					error);
 #endif /* DEBUG */
 			return error;
 		}
 	} else {
 		ip->i_d.di_magic = INT_GET(dip->di_core.di_magic, ARCH_CONVERT);
 		ip->i_d.di_version = INT_GET(dip->di_core.di_version, ARCH_CONVERT);
 		ip->i_d.di_gen = INT_GET(dip->di_core.di_gen, ARCH_CONVERT);
 		ip->i_d.di_flushiter = INT_GET(dip->di_core.di_flushiter, ARCH_CONVERT);
 		/*
 		 * Make sure to pull in the mode here as well in
 		 * case the inode is released without being used.
 		 * This ensures that xfs_inactive() will see that
 		 * the inode is already free and not try to mess
 		 * with the uninitialized part of it.
 		 */
 		ip->i_d.di_mode = 0;
 		/*
 		 * Initialize the per-fork minima and maxima for a new
 		 * inode here.  xfs_iformat will do it for old inodes.
 		 */
 		ip->i_df.if_ext_max =
 			XFS_IFORK_DSIZE(ip) / (uint)sizeof(xfs_bmbt_rec_t);
 	}
 	INIT_LIST_HEAD(&ip->i_reclaim);
 	/*
 	 * The inode format changed when we moved the link count and
 	 * made it 32 bits long.  If this is an old format inode,
 	 * convert it in memory to look like a new one.  If it gets
 	 * flushed to disk we will convert back before flushing or
 	 * logging it.  We zero out the new projid field and the old link
 	 * count field.  We'll handle clearing the pad field (the remains
 	 * of the old uuid field) when we actually convert the inode to
 	 * the new format. We don't change the version number so that we
 	 * can distinguish this from a real new format inode.
 	 */
 	if (ip->i_d.di_version == XFS_DINODE_VERSION_1) {
 		ip->i_d.di_nlink = ip->i_d.di_onlink;
 		ip->i_d.di_onlink = 0;
 		ip->i_d.di_projid = 0;
 	}
 	ip->i_delayed_blks = 0;
 	/*
 	 * Mark the buffer containing the inode as something to keep
 	 * around for a while.  This helps to keep recently accessed
 	 * meta-data in-core longer.
 	 */
 	 XFS_BUF_SET_REF(bp, XFS_INO_REF);
 	/*
 	 * Use xfs_trans_brelse() to release the buffer containing the
 	 * on-disk inode, because it was acquired with xfs_trans_read_buf()
 	 * in xfs_itobp() above.  If tp is NULL, this is just a normal
 	 * brelse().  If we're within a transaction, then xfs_trans_brelse()
 	 * will only release the buffer if it is not dirty within the
 	 * transaction.  It will be OK to release the buffer in this case,
 	 * because inodes on disk are never destroyed and we will be
 	 * locking the new in-core inode before putting it in the hash
 	 * table where other processes can find it.  Thus we don't have
 	 * to worry about the inode being changed just because we released
 	 * the buffer.
 	 */
 	xfs_trans_brelse(tp, bp);
 	*ipp = ip;
 	return 0;
 }
 /*
  * Read in extents from a btree-format inode.
  * Allocate and fill in if_extents.  Real work is done in xfs_bmap.c.
  */
 int
 xfs_iread_extents(
 	xfs_trans_t	*tp,
 	xfs_inode_t	*ip,
 	int		whichfork)
 {
 	int		error;
 	xfs_ifork_t	*ifp;
 	xfs_extnum_t	nextents;
 	size_t		size;
 	if (unlikely(XFS_IFORK_FORMAT(ip, whichfork) != XFS_DINODE_FMT_BTREE)) {
 		XFS_ERROR_REPORT("xfs_iread_extents", XFS_ERRLEVEL_LOW,
 				 ip->i_mount);
 		return XFS_ERROR(EFSCORRUPTED);
 	}
 	nextents = XFS_IFORK_NEXTENTS(ip, whichfork);
 	size = nextents * sizeof(xfs_bmbt_rec_t);
 	ifp = XFS_IFORK_PTR(ip, whichfork);
 	/*
 	 * We know that the size is valid (it's checked in iformat_btree)
 	 */
 	ifp->if_lastex = NULLEXTNUM;
 	ifp->if_bytes = ifp->if_real_bytes = 0;
 	ifp->if_flags |= XFS_IFEXTENTS;
 	xfs_iext_add(ifp, 0, nextents);
 	error = xfs_bmap_read_extents(tp, ip, whichfork);
 	if (error) {
 		xfs_iext_destroy(ifp);
 		ifp->if_flags &= ~XFS_IFEXTENTS;
 		return error;
 	}
 	xfs_validate_extents(ifp, nextents, 0, XFS_EXTFMT_INODE(ip));
 	return 0;
 }
 /*
  * Allocate an inode on disk and return a copy of its in-core version.
  * The in-core inode is locked exclusively.  Set mode, nlink, and rdev
  * appropriately within the inode.  The uid and gid for the inode are
  * set according to the contents of the given cred structure.
  *
  * Use xfs_dialloc() to allocate the on-disk inode. If xfs_dialloc()
  * has a free inode available, call xfs_iget()
  * to obtain the in-core version of the allocated inode.  Finally,
  * fill in the inode and log its initial contents.  In this case,
  * ialloc_context would be set to NULL and call_again set to false.
  *
  * If xfs_dialloc() does not have an available inode,
  * it will replenish its supply by doing an allocation. Since we can
  * only do one allocation within a transaction without deadlocks, we
  * must commit the current transaction before returning the inode itself.
  * In this case, therefore, we will set call_again to true and return.
  * The caller should then commit the current transaction, start a new
  * transaction, and call xfs_ialloc() again to actually get the inode.
  *
  * To ensure that some other process does not grab the inode that
  * was allocated during the first call to xfs_ialloc(), this routine
  * also returns the [locked] bp pointing to the head of the freelist
  * as ialloc_context.  The caller should hold this buffer across
  * the commit and pass it back into this routine on the second call.
  */
 int
 xfs_ialloc(
 	xfs_trans_t	*tp,
 	xfs_inode_t	*pip,
 	mode_t		mode,
 	xfs_nlink_t	nlink,
 	xfs_dev_t	rdev,
 	cred_t		*cr,
 	xfs_prid_t	prid,
 	int		okalloc,
 	xfs_buf_t	**ialloc_context,
 	boolean_t	*call_again,
 	xfs_inode_t	**ipp)
 {
 	xfs_ino_t	ino;
 	xfs_inode_t	*ip;
 	bhv_vnode_t	*vp;
 	uint		flags;
 	int		error;
 	/*
 	 * Call the space management code to pick
 	 * the on-disk inode to be allocated.
 	 */
 	error = xfs_dialloc(tp, pip->i_ino, mode, okalloc,
 			    ialloc_context, call_again, &ino);
 	if (error != 0) {
 		return error;
 	}
 	if (*call_again || ino == NULLFSINO) {
 		*ipp = NULL;
 		return 0;
 	}
 	ASSERT(*ialloc_context == NULL);
 	/*
 	 * Get the in-core inode with the lock held exclusively.
 	 * This is because we're setting fields here we need
 	 * to prevent others from looking at until we're done.
 	 */
 	error = xfs_trans_iget(tp->t_mountp, tp, ino,
-			IGET_CREATE, XFS_ILOCK_EXCL, &ip);
+				XFS_IGET_CREATE, XFS_ILOCK_EXCL, &ip);
 	if (error != 0) {
 		return error;
 	}
 	ASSERT(ip != NULL);
 	vp = XFS_ITOV(ip);
 	ip->i_d.di_mode = (__uint16_t)mode;
 	ip->i_d.di_onlink = 0;
 	ip->i_d.di_nlink = nlink;
 	ASSERT(ip->i_d.di_nlink == nlink);
 	ip->i_d.di_uid = current_fsuid(cr);
 	ip->i_d.di_gid = current_fsgid(cr);
 	ip->i_d.di_projid = prid;
 	memset(&(ip->i_d.di_pad[0]), 0, sizeof(ip->i_d.di_pad));
 	/*
 	 * If the superblock version is up to where we support new format
 	 * inodes and this is currently an old format inode, then change
 	 * the inode version number now.  This way we only do the conversion
 	 * here rather than here and in the flush/logging code.
 	 */
 	if (XFS_SB_VERSION_HASNLINK(&tp->t_mountp->m_sb) &&
 	    ip->i_d.di_version == XFS_DINODE_VERSION_1) {
 		ip->i_d.di_version = XFS_DINODE_VERSION_2;
 		/*
 		 * We've already zeroed the old link count, the projid field,
 		 * and the pad field.
 		 */
 	}
 	/*
 	 * Project ids won't be stored on disk if we are using a version 1 inode.
 	 */
 	if ( (prid != 0) && (ip->i_d.di_version == XFS_DINODE_VERSION_1))
 		xfs_bump_ino_vers2(tp, ip);
 	if (XFS_INHERIT_GID(pip, vp->v_vfsp)) {
 		ip->i_d.di_gid = pip->i_d.di_gid;
 		if ((pip->i_d.di_mode & S_ISGID) && (mode & S_IFMT) == S_IFDIR) {
 			ip->i_d.di_mode |= S_ISGID;
 		}
 	}
 	/*
 	 * If the group ID of the new file does not match the effective group
 	 * ID or one of the supplementary group IDs, the S_ISGID bit is cleared
 	 * (and only if the irix_sgid_inherit compatibility variable is set).
 	 */
 	if ((irix_sgid_inherit) &&
 	    (ip->i_d.di_mode & S_ISGID) &&
 	    (!in_group_p((gid_t)ip->i_d.di_gid))) {
 		ip->i_d.di_mode &= ~S_ISGID;
 	}
 	ip->i_d.di_size = 0;
 	ip->i_d.di_nextents = 0;
 	ASSERT(ip->i_d.di_nblocks == 0);
 	xfs_ichgtime(ip, XFS_ICHGTIME_CHG|XFS_ICHGTIME_ACC|XFS_ICHGTIME_MOD);
 	/*
 	 * di_gen will have been taken care of in xfs_iread.
 	 */
 	ip->i_d.di_extsize = 0;
 	ip->i_d.di_dmevmask = 0;
 	ip->i_d.di_dmstate = 0;
 	ip->i_d.di_flags = 0;
 	flags = XFS_ILOG_CORE;
 	switch (mode & S_IFMT) {
 	case S_IFIFO:
 	case S_IFCHR:
 	case S_IFBLK:
 	case S_IFSOCK:
 		ip->i_d.di_format = XFS_DINODE_FMT_DEV;
 		ip->i_df.if_u2.if_rdev = rdev;
 		ip->i_df.if_flags = 0;
 		flags |= XFS_ILOG_DEV;
 		break;
 	case S_IFREG:
 	case S_IFDIR:
 		if (unlikely(pip->i_d.di_flags & XFS_DIFLAG_ANY)) {
 			uint	di_flags = 0;
 			if ((mode & S_IFMT) == S_IFDIR) {
 				if (pip->i_d.di_flags & XFS_DIFLAG_RTINHERIT)
 					di_flags |= XFS_DIFLAG_RTINHERIT;
 				if (pip->i_d.di_flags & XFS_DIFLAG_EXTSZINHERIT) {
 					di_flags |= XFS_DIFLAG_EXTSZINHERIT;
 					ip->i_d.di_extsize = pip->i_d.di_extsize;
 				}
 			} else if ((mode & S_IFMT) == S_IFREG) {
 				if (pip->i_d.di_flags & XFS_DIFLAG_RTINHERIT) {
 					di_flags |= XFS_DIFLAG_REALTIME;
 					ip->i_iocore.io_flags |= XFS_IOCORE_RT;
 				}
 				if (pip->i_d.di_flags & XFS_DIFLAG_EXTSZINHERIT) {
 					di_flags |= XFS_DIFLAG_EXTSIZE;
 					ip->i_d.di_extsize = pip->i_d.di_extsize;
 				}
 			}
 			if ((pip->i_d.di_flags & XFS_DIFLAG_NOATIME) &&
 			    xfs_inherit_noatime)
 				di_flags |= XFS_DIFLAG_NOATIME;
 			if ((pip->i_d.di_flags & XFS_DIFLAG_NODUMP) &&
 			    xfs_inherit_nodump)
 				di_flags |= XFS_DIFLAG_NODUMP;
 			if ((pip->i_d.di_flags & XFS_DIFLAG_SYNC) &&
 			    xfs_inherit_sync)
 				di_flags |= XFS_DIFLAG_SYNC;
 			if ((pip->i_d.di_flags & XFS_DIFLAG_NOSYMLINKS) &&
 			    xfs_inherit_nosymlinks)
 				di_flags |= XFS_DIFLAG_NOSYMLINKS;
 			if (pip->i_d.di_flags & XFS_DIFLAG_PROJINHERIT)
 				di_flags |= XFS_DIFLAG_PROJINHERIT;
 			if ((pip->i_d.di_flags & XFS_DIFLAG_NODEFRAG) &&
 			    xfs_inherit_nodefrag)
 				di_flags |= XFS_DIFLAG_NODEFRAG;
 			ip->i_d.di_flags |= di_flags;
 		}
 		/* FALLTHROUGH */
 	case S_IFLNK:
 		ip->i_d.di_format = XFS_DINODE_FMT_EXTENTS;
 		ip->i_df.if_flags = XFS_IFEXTENTS;
 		ip->i_df.if_bytes = ip->i_df.if_real_bytes = 0;
 		ip->i_df.if_u1.if_extents = NULL;
 		break;
 	default:
 		ASSERT(0);
 	}
 	/*
 	 * Attribute fork settings for new inode.
 	 */
 	ip->i_d.di_aformat = XFS_DINODE_FMT_EXTENTS;
 	ip->i_d.di_anextents = 0;
 	/*
 	 * Log the new values stuffed into the inode.
 	 */
 	xfs_trans_log_inode(tp, ip, flags);
 	/* now that we have an i_mode we can setup inode ops and unlock */
 	bhv_vfs_init_vnode(XFS_MTOVFS(tp->t_mountp), vp, XFS_ITOBHV(ip), 1);
 	*ipp = ip;
 	return 0;
 }
 /*
  * Check to make sure that there are no blocks allocated to the
  * file beyond the size of the file.  We don't check this for
  * files with fixed size extents or real time extents, but we
  * at least do it for regular files.
  */
 #ifdef DEBUG
 void
 xfs_isize_check(
 	xfs_mount_t	*mp,
 	xfs_inode_t	*ip,
 	xfs_fsize_t	isize)
 {
 	xfs_fileoff_t	map_first;
 	int		nimaps;
 	xfs_bmbt_irec_t	imaps[2];
 	if ((ip->i_d.di_mode & S_IFMT) != S_IFREG)
 		return;
 	if (ip->i_d.di_flags & (XFS_DIFLAG_REALTIME | XFS_DIFLAG_EXTSIZE))
 		return;
 	nimaps = 2;
 	map_first = XFS_B_TO_FSB(mp, (xfs_ufsize_t)isize);
 	/*
 	 * The filesystem could be shutting down, so bmapi may return
 	 * an error.
 	 */
 	if (xfs_bmapi(NULL, ip, map_first,
 			 (XFS_B_TO_FSB(mp,
 				       (xfs_ufsize_t)XFS_MAXIOFFSET(mp)) -
 			  map_first),
 			 XFS_BMAPI_ENTIRE, NULL, 0, imaps, &nimaps,
 			 NULL, NULL))
 	    return;
 	ASSERT(nimaps == 1);
 	ASSERT(imaps[0].br_startblock == HOLESTARTBLOCK);
 }
 #endif	/* DEBUG */
 /*
  * Calculate the last possible buffered byte in a file.  This must
  * include data that was buffered beyond the EOF by the write code.
  * This also needs to deal with overflowing the xfs_fsize_t type
  * which can happen for sizes near the limit.
  *
  * We also need to take into account any blocks beyond the EOF.  It
  * may be the case that they were buffered by a write which failed.
  * In that case the pages will still be in memory, but the inode size
  * will never have been updated.
  */
 xfs_fsize_t
 xfs_file_last_byte(
 	xfs_inode_t	*ip)
 {
 	xfs_mount_t	*mp;
 	xfs_fsize_t	last_byte;
 	xfs_fileoff_t	last_block;
 	xfs_fileoff_t	size_last_block;
 	int		error;
 	ASSERT(ismrlocked(&(ip->i_iolock), MR_UPDATE | MR_ACCESS));
 	mp = ip->i_mount;
 	/*
 	 * Only check for blocks beyond the EOF if the extents have
 	 * been read in.  This eliminates the need for the inode lock,
 	 * and it also saves us from looking when it really isn't
 	 * necessary.
 	 */
 	if (ip->i_df.if_flags & XFS_IFEXTENTS) {
 		error = xfs_bmap_last_offset(NULL, ip, &last_block,
 			XFS_DATA_FORK);
 		if (error) {
 			last_block = 0;
 		}
 	} else {
 		last_block = 0;
 	}
 	size_last_block = XFS_B_TO_FSB(mp, (xfs_ufsize_t)ip->i_d.di_size);
 	last_block = XFS_FILEOFF_MAX(last_block, size_last_block);
 	last_byte = XFS_FSB_TO_B(mp, last_block);
 	if (last_byte < 0) {
 		return XFS_MAXIOFFSET(mp);
 	}
 	last_byte += (1 << mp->m_writeio_log);
 	if (last_byte < 0) {
 		return XFS_MAXIOFFSET(mp);
 	}
 	return last_byte;
 }
 #if defined(XFS_RW_TRACE)
 STATIC void
 xfs_itrunc_trace(
 	int		tag,
 	xfs_inode_t	*ip,
 	int		flag,
 	xfs_fsize_t	new_size,
 	xfs_off_t	toss_start,
 	xfs_off_t	toss_finish)
 {
 	if (ip->i_rwtrace == NULL) {
 		return;
 	}
 	ktrace_enter(ip->i_rwtrace,
 		     (void*)((long)tag),
 		     (void*)ip,
 		     (void*)(unsigned long)((ip->i_d.di_size >> 32) & 0xffffffff),
 		     (void*)(unsigned long)(ip->i_d.di_size & 0xffffffff),
 		     (void*)((long)flag),
 		     (void*)(unsigned long)((new_size >> 32) & 0xffffffff),
 		     (void*)(unsigned long)(new_size & 0xffffffff),
 		     (void*)(unsigned long)((toss_start >> 32) & 0xffffffff),
 		     (void*)(unsigned long)(toss_start & 0xffffffff),
 		     (void*)(unsigned long)((toss_finish >> 32) & 0xffffffff),
 		     (void*)(unsigned long)(toss_finish & 0xffffffff),
 		     (void*)(unsigned long)current_cpu(),
 		     (void*)(unsigned long)current_pid(),
 		     (void*)NULL,
 		     (void*)NULL,
 		     (void*)NULL);
 }
 #else
 #define	xfs_itrunc_trace(tag, ip, flag, new_size, toss_start, toss_finish)
 #endif
 /*
  * Start the truncation of the file to new_size.  The new size
  * must be smaller than the current size.  This routine will
  * clear the buffer and page caches of file data in the removed
  * range, and xfs_itruncate_finish() will remove the underlying
  * disk blocks.
  *
  * The inode must have its I/O lock locked EXCLUSIVELY, and it
  * must NOT have the inode lock held at all.  This is because we're
  * calling into the buffer/page cache code and we can't hold the
  * inode lock when we do so.
  *
  * We need to wait for any direct I/Os in flight to complete before we
  * proceed with the truncate. This is needed to prevent the extents
  * being read or written by the direct I/Os from being removed while the
  * I/O is in flight as there is no other method of synchronising
  * direct I/O with the truncate operation.  Also, because we hold
  * the IOLOCK in exclusive mode, we prevent new direct I/Os from being
  * started until the truncate completes and drops the lock. Essentially,
  * the vn_iowait() call forms an I/O barrier that provides strict ordering
  * between direct I/Os and the truncate operation.
  *
  * The flags parameter can have either the value XFS_ITRUNC_DEFINITE
  * or XFS_ITRUNC_MAYBE.  The XFS_ITRUNC_MAYBE value should be used
  * in the case that the caller is locking things out of order and
  * may not be able to call xfs_itruncate_finish() with the inode lock
  * held without dropping the I/O lock.  If the caller must drop the
  * I/O lock before calling xfs_itruncate_finish(), then xfs_itruncate_start()
  * must be called again with all the same restrictions as the initial
  * call.
  */
 void
 xfs_itruncate_start(
 	xfs_inode_t	*ip,
 	uint		flags,
 	xfs_fsize_t	new_size)
 {
 	xfs_fsize_t	last_byte;
 	xfs_off_t	toss_start;
 	xfs_mount_t	*mp;
 	bhv_vnode_t	*vp;
 	ASSERT(ismrlocked(&ip->i_iolock, MR_UPDATE) != 0);
 	ASSERT((new_size == 0) || (new_size <= ip->i_d.di_size));
 	ASSERT((flags == XFS_ITRUNC_DEFINITE) ||
 	       (flags == XFS_ITRUNC_MAYBE));
 	mp = ip->i_mount;
 	vp = XFS_ITOV(ip);
 	vn_iowait(vp);  /* wait for the completion of any pending DIOs */
 	/*
 	 * Call toss_pages or flushinval_pages to get rid of pages
 	 * overlapping the region being removed.  We have to use
 	 * the less efficient flushinval_pages in the case that the
 	 * caller may not be able to finish the truncate without
 	 * dropping the inode's I/O lock.  Make sure
 	 * to catch any pages brought in by buffers overlapping
 	 * the EOF by searching out beyond the isize by our
 	 * block size. We round new_size up to a block boundary
 	 * so that we don't toss things on the same block as
 	 * new_size but before it.
 	 *
 	 * Before calling toss_page or flushinval_pages, make sure to
 	 * call remapf() over the same region if the file is mapped.
 	 * This frees up mapped file references to the pages in the
 	 * given range and for the flushinval_pages case it ensures
 	 * that we get the latest mapped changes flushed out.
 	 */
 	toss_start = XFS_B_TO_FSB(mp, (xfs_ufsize_t)new_size);
 	toss_start = XFS_FSB_TO_B(mp, toss_start);
 	if (toss_start < 0) {
 		/*
 		 * The place to start tossing is beyond our maximum
 		 * file size, so there is no way that the data extended
 		 * out there.
 		 */
 		return;
 	}
 	last_byte = xfs_file_last_byte(ip);
 	xfs_itrunc_trace(XFS_ITRUNC_START, ip, flags, new_size, toss_start,
 			 last_byte);
 	if (last_byte > toss_start) {
 		if (flags & XFS_ITRUNC_DEFINITE) {
 			bhv_vop_toss_pages(vp, toss_start, -1, FI_REMAPF_LOCKED);
 		} else {
 			bhv_vop_flushinval_pages(vp, toss_start, -1, FI_REMAPF_LOCKED);
 		}
 	}
 #ifdef DEBUG
 	if (new_size == 0) {
 		ASSERT(VN_CACHED(vp) == 0);
 	}
 #endif
 }
 /*
  * Shrink the file to the given new_size.  The new
  * size must be smaller than the current size.
  * This will free up the underlying blocks
  * in the removed range after a call to xfs_itruncate_start()
  * or xfs_atruncate_start().
  *
  * The transaction passed to this routine must have made
  * a permanent log reservation of at least XFS_ITRUNCATE_LOG_RES.
  * This routine may commit the given transaction and
  * start new ones, so make sure everything involved in
  * the transaction is tidy before calling here.
  * Some transaction will be returned to the caller to be
  * committed.  The incoming transaction must already include
  * the inode, and both inode locks must be held exclusively.
  * The inode must also be "held" within the transaction.  On
  * return the inode will be "held" within the returned transaction.
  * This routine does NOT require any disk space to be reserved
  * for it within the transaction.
  *
  * The fork parameter must be either xfs_attr_fork or xfs_data_fork,
  * and it indicates the fork which is to be truncated.  For the
  * attribute fork we only support truncation to size 0.
  *
  * We use the sync parameter to indicate whether or not the first
  * transaction we perform might have to be synchronous.  For the attr fork,
  * it needs to be so if the unlink of the inode is not yet known to be
  * permanent in the log.  This keeps us from freeing and reusing the
  * blocks of the attribute fork before the unlink of the inode becomes
  * permanent.
  *
  * For the data fork, we normally have to run synchronously if we're
  * being called out of the inactive path or we're being called
  * out of the create path where we're truncating an existing file.
  * Either way, the truncate needs to be sync so blocks don't reappear
  * in the file with altered data in case of a crash.  wsync filesystems
  * can run the first case async because anything that shrinks the inode
  * has to run sync so by the time we're called here from inactive, the
  * inode size is permanently set to 0.
  *
  * Calls from the truncate path always need to be sync unless we're
  * in a wsync filesystem and the file has already been unlinked.
  *
  * The caller is responsible for correctly setting the sync parameter.
  * It gets too hard for us to guess here which path we're being called
  * out of just based on inode state.
  */
 int
 xfs_itruncate_finish(
 	xfs_trans_t	**tp,
 	xfs_inode_t	*ip,
 	xfs_fsize_t	new_size,
 	int		fork,
 	int		sync)
 {
 	xfs_fsblock_t	first_block;
 	xfs_fileoff_t	first_unmap_block;
 	xfs_fileoff_t	last_block;
 	xfs_filblks_t	unmap_len=0;
 	xfs_mount_t	*mp;
 	xfs_trans_t	*ntp;
 	int		done;
 	int		committed;
 	xfs_bmap_free_t	free_list;
 	int		error;
 	ASSERT(ismrlocked(&ip->i_iolock, MR_UPDATE) != 0);
 	ASSERT(ismrlocked(&ip->i_lock, MR_UPDATE) != 0);
 	ASSERT((new_size == 0) || (new_size <= ip->i_d.di_size));
 	ASSERT(*tp != NULL);
 	ASSERT((*tp)->t_flags & XFS_TRANS_PERM_LOG_RES);
 	ASSERT(ip->i_transp == *tp);
 	ASSERT(ip->i_itemp != NULL);
 	ASSERT(ip->i_itemp->ili_flags & XFS_ILI_HOLD);
 	ntp = *tp;
 	mp = (ntp)->t_mountp;
 	ASSERT(! XFS_NOT_DQATTACHED(mp, ip));
 	/*
 	 * We only support truncating the entire attribute fork.
 	 */
 	if (fork == XFS_ATTR_FORK) {
 		new_size = 0LL;
 	}
 	first_unmap_block = XFS_B_TO_FSB(mp, (xfs_ufsize_t)new_size);
 	xfs_itrunc_trace(XFS_ITRUNC_FINISH1, ip, 0, new_size, 0, 0);
 	/*
 	 * The first thing we do is set the size to new_size permanently
 	 * on disk.  This way we don't have to worry about anyone ever
 	 * being able to look at the data being freed even in the face
 	 * of a crash.  What we're getting around here is the case where
 	 * we free a block, it is allocated to another file, it is written
 	 * to, and then we crash.  If the new data gets written to the
 	 * file but the log buffers containing the free and reallocation
 	 * don't, then we'd end up with garbage in the blocks being freed.
 	 * As long as we make the new_size permanent before actually
 	 * freeing any blocks it doesn't matter if they get writtten to.
 	 *
 	 * The callers must signal into us whether or not the size
 	 * setting here must be synchronous.  There are a few cases
 	 * where it doesn't have to be synchronous.  Those cases
 	 * occur if the file is unlinked and we know the unlink is
 	 * permanent or if the blocks being truncated are guaranteed
 	 * to be beyond the inode eof (regardless of the link count)
 	 * and the eof value is permanent.  Both of these cases occur
 	 * only on wsync-mounted filesystems.  In those cases, we're
 	 * guaranteed that no user will ever see the data in the blocks
 	 * that are being truncated so the truncate can run async.
 	 * In the free beyond eof case, the file may wind up with
 	 * more blocks allocated to it than it needs if we crash
 	 * and that won't get fixed until the next time the file
 	 * is re-opened and closed but that's ok as that shouldn't
 	 * be too many blocks.
 	 *
 	 * However, we can't just make all wsync xactions run async
 	 * because there's one call out of the create path that needs
 	 * to run sync where it's truncating an existing file to size
 	 * 0 whose size is > 0.
 	 *
 	 * It's probably possible to come up with a test in this
 	 * routine that would correctly distinguish all the above
 	 * cases from the values of the function parameters and the
 	 * inode state but for sanity's sake, I've decided to let the
 	 * layers above just tell us.  It's simpler to correctly figure
 	 * out in the layer above exactly under what conditions we
 	 * can run async and I think it's easier for others read and
 	 * follow the logic in case something has to be changed.
 	 * cscope is your friend -- rcc.
 	 *
 	 * The attribute fork is much simpler.
 	 *
 	 * For the attribute fork we allow the caller to tell us whether
 	 * the unlink of the inode that led to this call is yet permanent
 	 * in the on disk log.  If it is not and we will be freeing extents
 	 * in this inode then we make the first transaction synchronous
 	 * to make sure that the unlink is permanent by the time we free
 	 * the blocks.
 	 */
 	if (fork == XFS_DATA_FORK) {
 		if (ip->i_d.di_nextents > 0) {
 			ip->i_d.di_size = new_size;
 			xfs_trans_log_inode(ntp, ip, XFS_ILOG_CORE);
 		}
 	} else if (sync) {
 		ASSERT(!(mp->m_flags & XFS_MOUNT_WSYNC));
 		if (ip->i_d.di_anextents > 0)
 			xfs_trans_set_sync(ntp);
 	}
 	ASSERT(fork == XFS_DATA_FORK ||
 		(fork == XFS_ATTR_FORK &&
 			((sync && !(mp->m_flags & XFS_MOUNT_WSYNC)) ||
 			 (sync == 0 && (mp->m_flags & XFS_MOUNT_WSYNC)))));
 	/*
 	 * Since it is possible for space to become allocated beyond
 	 * the end of the file (in a crash where the space is allocated
 	 * but the inode size is not yet updated), simply remove any
 	 * blocks which show up between the new EOF and the maximum
 	 * possible file size.  If the first block to be removed is
 	 * beyond the maximum file size (ie it is the same as last_block),
 	 * then there is nothing to do.
 	 */
 	last_block = XFS_B_TO_FSB(mp, (xfs_ufsize_t)XFS_MAXIOFFSET(mp));
 	ASSERT(first_unmap_block <= last_block);
 	done = 0;
 	if (last_block == first_unmap_block) {
 		done = 1;
 	} else {
 		unmap_len = last_block - first_unmap_block + 1;
 	}
 	while (!done) {
 		/*
 		 * Free up up to XFS_ITRUNC_MAX_EXTENTS.  xfs_bunmapi()
 		 * will tell us whether it freed the entire range or
 		 * not.  If this is a synchronous mount (wsync),
 		 * then we can tell bunmapi to keep all the
 		 * transactions asynchronous since the unlink
 		 * transaction that made this inode inactive has
 		 * already hit the disk.  There's no danger of
 		 * the freed blocks being reused, there being a
 		 * crash, and the reused blocks suddenly reappearing
 		 * in this file with garbage in them once recovery
 		 * runs.
 		 */
 		XFS_BMAP_INIT(&free_list, &first_block);
 		error = XFS_BUNMAPI(mp, ntp, &ip->i_iocore,
 				    first_unmap_block, unmap_len,
 				    XFS_BMAPI_AFLAG(fork) |
 				      (sync ? 0 : XFS_BMAPI_ASYNC),
 				    XFS_ITRUNC_MAX_EXTENTS,
 				    &first_block, &free_list,
 				    NULL, &done);
 		if (error) {
 			/*
 			 * If the bunmapi call encounters an error,
 			 * return to the caller where the transaction
 			 * can be properly aborted.  We just need to
 			 * make sure we're not holding any resources
 			 * that we were not when we came in.
 			 */
 			xfs_bmap_cancel(&free_list);
 			return error;
 		}
 		/*
 		 * Duplicate the transaction that has the permanent
 		 * reservation and commit the old transaction.
 		 */
 		error = xfs_bmap_finish(tp, &free_list, first_block,
 					&committed);
 		ntp = *tp;
 		if (error) {
 			/*
 			 * If the bmap finish call encounters an error,
 			 * return to the caller where the transaction
 			 * can be properly aborted.  We just need to
 			 * make sure we're not holding any resources
 			 * that we were not when we came in.
 			 *
 			 * Aborting from this point might lose some
 			 * blocks in the file system, but oh well.
 			 */
 			xfs_bmap_cancel(&free_list);
 			if (committed) {
 				/*
 				 * If the passed in transaction committed
 				 * in xfs_bmap_finish(), then we want to
 				 * add the inode to this one before returning.
 				 * This keeps things simple for the higher
 				 * level code, because it always knows that
 				 * the inode is locked and held in the
 				 * transaction that returns to it whether
 				 * errors occur or not.  We don't mark the
 				 * inode dirty so that this transaction can
 				 * be easily aborted if possible.
 				 */
 				xfs_trans_ijoin(ntp, ip,
 					XFS_ILOCK_EXCL | XFS_IOLOCK_EXCL);
 				xfs_trans_ihold(ntp, ip);
 			}
 			return error;
 		}
 		if (committed) {
 			/*
 			 * The first xact was committed,
 			 * so add the inode to the new one.
 			 * Mark it dirty so it will be logged
 			 * and moved forward in the log as
 			 * part of every commit.
 			 */
 			xfs_trans_ijoin(ntp, ip,
 					XFS_ILOCK_EXCL | XFS_IOLOCK_EXCL);
 			xfs_trans_ihold(ntp, ip);
 			xfs_trans_log_inode(ntp, ip, XFS_ILOG_CORE);
 		}
 		ntp = xfs_trans_dup(ntp);
 		(void) xfs_trans_commit(*tp, 0, NULL);
 		*tp = ntp;
 		error = xfs_trans_reserve(ntp, 0, XFS_ITRUNCATE_LOG_RES(mp), 0,
 					  XFS_TRANS_PERM_LOG_RES,
 					  XFS_ITRUNCATE_LOG_COUNT);
 		/*
 		 * Add the inode being truncated to the next chained
 		 * transaction.
 		 */
 		xfs_trans_ijoin(ntp, ip, XFS_ILOCK_EXCL | XFS_IOLOCK_EXCL);
 		xfs_trans_ihold(ntp, ip);
 		if (error)
 			return (error);
 	}
 	/*
 	 * Only update the size in the case of the data fork, but
 	 * always re-log the inode so that our permanent transaction
 	 * can keep on rolling it forward in the log.
 	 */
 	if (fork == XFS_DATA_FORK) {
 		xfs_isize_check(mp, ip, new_size);
 		ip->i_d.di_size = new_size;
 	}
 	xfs_trans_log_inode(ntp, ip, XFS_ILOG_CORE);
 	ASSERT((new_size != 0) ||
 	       (fork == XFS_ATTR_FORK) ||
 	       (ip->i_delayed_blks == 0));
 	ASSERT((new_size != 0) ||
 	       (fork == XFS_ATTR_FORK) ||
 	       (ip->i_d.di_nextents == 0));
 	xfs_itrunc_trace(XFS_ITRUNC_FINISH2, ip, 0, new_size, 0, 0);
 	return 0;
 }
 /*
  * xfs_igrow_start
  *
  * Do the first part of growing a file: zero any data in the last
  * block that is beyond the old EOF.  We need to do this before
  * the inode is joined to the transaction to modify the i_size.
  * That way we can drop the inode lock and call into the buffer
  * cache to get the buffer mapping the EOF.
  */
 int
 xfs_igrow_start(
 	xfs_inode_t	*ip,
 	xfs_fsize_t	new_size,
 	cred_t		*credp)
 {
 	int		error;
 	ASSERT(ismrlocked(&(ip->i_lock), MR_UPDATE) != 0);
 	ASSERT(ismrlocked(&(ip->i_iolock), MR_UPDATE) != 0);
 	ASSERT(new_size > ip->i_d.di_size);
 	/*
 	 * Zero any pages that may have been created by
 	 * xfs_write_file() beyond the end of the file
 	 * and any blocks between the old and new file sizes.
 	 */
 	error = xfs_zero_eof(XFS_ITOV(ip), &ip->i_iocore, new_size,
 			     ip->i_d.di_size, new_size);
 	return error;
 }
 /*
  * xfs_igrow_finish
  *
  * This routine is called to extend the size of a file.
  * The inode must have both the iolock and the ilock locked
  * for update and it must be a part of the current transaction.
  * The xfs_igrow_start() function must have been called previously.
  * If the change_flag is not zero, the inode change timestamp will
  * be updated.
  */
 void
 xfs_igrow_finish(
 	xfs_trans_t	*tp,
 	xfs_inode_t	*ip,
 	xfs_fsize_t	new_size,
 	int		change_flag)
 {
 	ASSERT(ismrlocked(&(ip->i_lock), MR_UPDATE) != 0);
 	ASSERT(ismrlocked(&(ip->i_iolock), MR_UPDATE) != 0);
 	ASSERT(ip->i_transp == tp);
 	ASSERT(new_size > ip->i_d.di_size);
 	/*
 	 * Update the file size.  Update the inode change timestamp
 	 * if change_flag set.
 	 */
 	ip->i_d.di_size = new_size;
 	if (change_flag)
 		xfs_ichgtime(ip, XFS_ICHGTIME_CHG);
 	xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
 }
 /*
  * This is called when the inode's link count goes to 0.
  * We place the on-disk inode on a list in the AGI.  It
  * will be pulled from this list when the inode is freed.
  */
 int
 xfs_iunlink(
 	xfs_trans_t	*tp,
 	xfs_inode_t	*ip)
 {
 	xfs_mount_t	*mp;
 	xfs_agi_t	*agi;
 	xfs_dinode_t	*dip;
 	xfs_buf_t	*agibp;
 	xfs_buf_t	*ibp;
 	xfs_agnumber_t	agno;
 	xfs_daddr_t	agdaddr;
 	xfs_agino_t	agino;
 	short		bucket_index;
 	int		offset;
 	int		error;
 	int		agi_ok;
 	ASSERT(ip->i_d.di_nlink == 0);
 	ASSERT(ip->i_d.di_mode != 0);
 	ASSERT(ip->i_transp == tp);
 	mp = tp->t_mountp;
 	agno = XFS_INO_TO_AGNO(mp, ip->i_ino);
 	agdaddr = XFS_AG_DADDR(mp, agno, XFS_AGI_DADDR(mp));
 	/*
 	 * Get the agi buffer first.  It ensures lock ordering
 	 * on the list.
 	 */
 	error = xfs_trans_read_buf(mp, tp, mp->m_ddev_targp, agdaddr,
 				   XFS_FSS_TO_BB(mp, 1), 0, &agibp);
 	if (error) {
 		return error;
 	}
 	/*
 	 * Validate the magic number of the agi block.
 	 */
 	agi = XFS_BUF_TO_AGI(agibp);
 	agi_ok =
 		be32_to_cpu(agi->agi_magicnum) == XFS_AGI_MAGIC &&
 		XFS_AGI_GOOD_VERSION(be32_to_cpu(agi->agi_versionnum));
 	if (unlikely(XFS_TEST_ERROR(!agi_ok, mp, XFS_ERRTAG_IUNLINK,
 			XFS_RANDOM_IUNLINK))) {
 		XFS_CORRUPTION_ERROR("xfs_iunlink", XFS_ERRLEVEL_LOW, mp, agi);
 		xfs_trans_brelse(tp, agibp);
 		return XFS_ERROR(EFSCORRUPTED);
 	}
 	/*
 	 * Get the index into the agi hash table for the
 	 * list this inode will go on.
 	 */
 	agino = XFS_INO_TO_AGINO(mp, ip->i_ino);
 	ASSERT(agino != 0);
 	bucket_index = agino % XFS_AGI_UNLINKED_BUCKETS;
 	ASSERT(agi->agi_unlinked[bucket_index]);
 	ASSERT(be32_to_cpu(agi->agi_unlinked[bucket_index]) != agino);
 	if (be32_to_cpu(agi->agi_unlinked[bucket_index]) != NULLAGINO) {
 		/*
 		 * There is already another inode in the bucket we need
 		 * to add ourselves to.  Add us at the front of the list.
 		 * Here we put the head pointer into our next pointer,
 		 * and then we fall through to point the head at us.
 		 */
 		error = xfs_itobp(mp, tp, ip, &dip, &ibp, 0, 0);
 		if (error) {
 			return error;
 		}
 		ASSERT(INT_GET(dip->di_next_unlinked, ARCH_CONVERT) == NULLAGINO);
 		ASSERT(dip->di_next_unlinked);
 		/* both on-disk, don't endian flip twice */
 		dip->di_next_unlinked = agi->agi_unlinked[bucket_index];
 		offset = ip->i_boffset +
 			offsetof(xfs_dinode_t, di_next_unlinked);
 		xfs_trans_inode_buf(tp, ibp);
 		xfs_trans_log_buf(tp, ibp, offset,
 				  (offset + sizeof(xfs_agino_t) - 1));
 		xfs_inobp_check(mp, ibp);
 	}
 	/*
 	 * Point the bucket head pointer at the inode being inserted.
 	 */
 	ASSERT(agino != 0);
 	agi->agi_unlinked[bucket_index] = cpu_to_be32(agino);
 	offset = offsetof(xfs_agi_t, agi_unlinked) +
 		(sizeof(xfs_agino_t) * bucket_index);
 	xfs_trans_log_buf(tp, agibp, offset,
 			  (offset + sizeof(xfs_agino_t) - 1));
 	return 0;
 }
 /*
  * Pull the on-disk inode from the AGI unlinked list.
  */
 STATIC int
 xfs_iunlink_remove(
 	xfs_trans_t	*tp,
 	xfs_inode_t	*ip)
 {
 	xfs_ino_t	next_ino;
 	xfs_mount_t	*mp;
 	xfs_agi_t	*agi;
 	xfs_dinode_t	*dip;
 	xfs_buf_t	*agibp;
 	xfs_buf_t	*ibp;
 	xfs_agnumber_t	agno;
 	xfs_daddr_t	agdaddr;
 	xfs_agino_t	agino;
 	xfs_agino_t	next_agino;
 	xfs_buf_t	*last_ibp;
 	xfs_dinode_t	*last_dip = NULL;
 	short		bucket_index;
 	int		offset, last_offset = 0;
 	int		error;
 	int		agi_ok;
 	/*
 	 * First pull the on-disk inode from the AGI unlinked list.
 	 */
 	mp = tp->t_mountp;
 	agno = XFS_INO_TO_AGNO(mp, ip->i_ino);
 	agdaddr = XFS_AG_DADDR(mp, agno, XFS_AGI_DADDR(mp));
 	/*
 	 * Get the agi buffer first.  It ensures lock ordering
 	 * on the list.
 	 */
 	error = xfs_trans_read_buf(mp, tp, mp->m_ddev_targp, agdaddr,
 				   XFS_FSS_TO_BB(mp, 1), 0, &agibp);
 	if (error) {
 		cmn_err(CE_WARN,
 			"xfs_iunlink_remove: xfs_trans_read_buf()  returned an error %d on %s.  Returning error.",
 			error, mp->m_fsname);
 		return error;
 	}
 	/*
 	 * Validate the magic number of the agi block.
 	 */
 	agi = XFS_BUF_TO_AGI(agibp);
 	agi_ok =
 		be32_to_cpu(agi->agi_magicnum) == XFS_AGI_MAGIC &&
 		XFS_AGI_GOOD_VERSION(be32_to_cpu(agi->agi_versionnum));
 	if (unlikely(XFS_TEST_ERROR(!agi_ok, mp, XFS_ERRTAG_IUNLINK_REMOVE,
 			XFS_RANDOM_IUNLINK_REMOVE))) {
 		XFS_CORRUPTION_ERROR("xfs_iunlink_remove", XFS_ERRLEVEL_LOW,
 				     mp, agi);
 		xfs_trans_brelse(tp, agibp);
 		cmn_err(CE_WARN,
 			"xfs_iunlink_remove: XFS_TEST_ERROR()  returned an error on %s.  Returning EFSCORRUPTED.",
 			 mp->m_fsname);
 		return XFS_ERROR(EFSCORRUPTED);
 	}
 	/*
 	 * Get the index into the agi hash table for the
 	 * list this inode will go on.
 	 */
 	agino = XFS_INO_TO_AGINO(mp, ip->i_ino);
 	ASSERT(agino != 0);
 	bucket_index = agino % XFS_AGI_UNLINKED_BUCKETS;
 	ASSERT(be32_to_cpu(agi->agi_unlinked[bucket_index]) != NULLAGINO);
 	ASSERT(agi->agi_unlinked[bucket_index]);
 	if (be32_to_cpu(agi->agi_unlinked[bucket_index]) == agino) {
 		/*
 		 * We're at the head of the list.  Get the inode's
 		 * on-disk buffer to see if there is anyone after us
 		 * on the list.  Only modify our next pointer if it
 		 * is not already NULLAGINO.  This saves us the overhead
 		 * of dealing with the buffer when there is no need to
 		 * change it.
 		 */
 		error = xfs_itobp(mp, tp, ip, &dip, &ibp, 0, 0);
 		if (error) {
 			cmn_err(CE_WARN,
 				"xfs_iunlink_remove: xfs_itobp()  returned an error %d on %s.  Returning error.",
 				error, mp->m_fsname);
 			return error;
 		}
 		next_agino = INT_GET(dip->di_next_unlinked, ARCH_CONVERT);
 		ASSERT(next_agino != 0);
 		if (next_agino != NULLAGINO) {
 			INT_SET(dip->di_next_unlinked, ARCH_CONVERT, NULLAGINO);
 			offset = ip->i_boffset +
 				offsetof(xfs_dinode_t, di_next_unlinked);
 			xfs_trans_inode_buf(tp, ibp);
 			xfs_trans_log_buf(tp, ibp, offset,
 					  (offset + sizeof(xfs_agino_t) - 1));
 			xfs_inobp_check(mp, ibp);
 		} else {
 			xfs_trans_brelse(tp, ibp);
 		}
 		/*
 		 * Point the bucket head pointer at the next inode.
 		 */
 		ASSERT(next_agino != 0);
 		ASSERT(next_agino != agino);
 		agi->agi_unlinked[bucket_index] = cpu_to_be32(next_agino);
 		offset = offsetof(xfs_agi_t, agi_unlinked) +
 			(sizeof(xfs_agino_t) * bucket_index);
 		xfs_trans_log_buf(tp, agibp, offset,
 				  (offset + sizeof(xfs_agino_t) - 1));
 	} else {
 		/*
 		 * We need to search the list for the inode being freed.
 		 */
 		next_agino = be32_to_cpu(agi->agi_unlinked[bucket_index]);
 		last_ibp = NULL;
 		while (next_agino != agino) {
 			/*
 			 * If the last inode wasn't the one pointing to
 			 * us, then release its buffer since we're not
 			 * going to do anything with it.
 			 */
 			if (last_ibp != NULL) {
 				xfs_trans_brelse(tp, last_ibp);
 			}
 			next_ino = XFS_AGINO_TO_INO(mp, agno, next_agino);
 			error = xfs_inotobp(mp, tp, next_ino, &last_dip,
 					    &last_ibp, &last_offset);
 			if (error) {
 				cmn_err(CE_WARN,
 			"xfs_iunlink_remove: xfs_inotobp()  returned an error %d on %s.  Returning error.",
 					error, mp->m_fsname);
 				return error;
 			}
 			next_agino = INT_GET(last_dip->di_next_unlinked, ARCH_CONVERT);
 			ASSERT(next_agino != NULLAGINO);
 			ASSERT(next_agino != 0);
 		}
 		/*
 		 * Now last_ibp points to the buffer previous to us on
 		 * the unlinked list.  Pull us from the list.
 		 */
 		error = xfs_itobp(mp, tp, ip, &dip, &ibp, 0, 0);
 		if (error) {
 			cmn_err(CE_WARN,
 				"xfs_iunlink_remove: xfs_itobp()  returned an error %d on %s.  Returning error.",
 				error, mp->m_fsname);
 			return error;
 		}
 		next_agino = INT_GET(dip->di_next_unlinked, ARCH_CONVERT);
 		ASSERT(next_agino != 0);
 		ASSERT(next_agino != agino);
 		if (next_agino != NULLAGINO) {
 			INT_SET(dip->di_next_unlinked, ARCH_CONVERT, NULLAGINO);
 			offset = ip->i_boffset +
 				offsetof(xfs_dinode_t, di_next_unlinked);
 			xfs_trans_inode_buf(tp, ibp);
 			xfs_trans_log_buf(tp, ibp, offset,
 					  (offset + sizeof(xfs_agino_t) - 1));
 			xfs_inobp_check(mp, ibp);
 		} else {
 			xfs_trans_brelse(tp, ibp);
 		}
 		/*
 		 * Point the previous inode on the list to the next inode.
 		 */
 		INT_SET(last_dip->di_next_unlinked, ARCH_CONVERT, next_agino);
 		ASSERT(next_agino != 0);
 		offset = last_offset + offsetof(xfs_dinode_t, di_next_unlinked);
 		xfs_trans_inode_buf(tp, last_ibp);
 		xfs_trans_log_buf(tp, last_ibp, offset,
 				  (offset + sizeof(xfs_agino_t) - 1));
 		xfs_inobp_check(mp, last_ibp);
 	}
 	return 0;
 }
 static __inline__ int xfs_inode_clean(xfs_inode_t *ip)
 {
 	return (((ip->i_itemp == NULL) ||
 		!(ip->i_itemp->ili_format.ilf_fields & XFS_ILOG_ALL)) &&
 		(ip->i_update_core == 0));
 }
 STATIC void
 xfs_ifree_cluster(
 	xfs_inode_t	*free_ip,
 	xfs_trans_t	*tp,
 	xfs_ino_t	inum)
 {
 	xfs_mount_t		*mp = free_ip->i_mount;
 	int			blks_per_cluster;
 	int			nbufs;
 	int			ninodes;
 	int			i, j, found, pre_flushed;
 	xfs_daddr_t		blkno;
 	xfs_buf_t		*bp;
 	xfs_ihash_t		*ih;
 	xfs_inode_t		*ip, **ip_found;
 	xfs_inode_log_item_t	*iip;
 	xfs_log_item_t		*lip;
 	SPLDECL(s);
 	if (mp->m_sb.sb_blocksize >= XFS_INODE_CLUSTER_SIZE(mp)) {
 		blks_per_cluster = 1;
 		ninodes = mp->m_sb.sb_inopblock;
 		nbufs = XFS_IALLOC_BLOCKS(mp);
 	} else {
 		blks_per_cluster = XFS_INODE_CLUSTER_SIZE(mp) /
 					mp->m_sb.sb_blocksize;
 		ninodes = blks_per_cluster * mp->m_sb.sb_inopblock;
 		nbufs = XFS_IALLOC_BLOCKS(mp) / blks_per_cluster;
 	}
 	ip_found = kmem_alloc(ninodes * sizeof(xfs_inode_t *), KM_NOFS);
 	for (j = 0; j < nbufs; j++, inum += ninodes) {
 		blkno = XFS_AGB_TO_DADDR(mp, XFS_INO_TO_AGNO(mp, inum),
 					 XFS_INO_TO_AGBNO(mp, inum));
 		/*
 		 * Look for each inode in memory and attempt to lock it,
 		 * we can be racing with flush and tail pushing here.
 		 * any inode we get the locks on, add to an array of
 		 * inode items to process later.
 		 *
 		 * The get the buffer lock, we could beat a flush
 		 * or tail pushing thread to the lock here, in which
 		 * case they will go looking for the inode buffer
 		 * and fail, we need some other form of interlock
 		 * here.
 		 */
 		found = 0;
 		for (i = 0; i < ninodes; i++) {
 			ih = XFS_IHASH(mp, inum + i);
 			read_lock(&ih->ih_lock);
 			for (ip = ih->ih_next; ip != NULL; ip = ip->i_next) {
 				if (ip->i_ino == inum + i)
 					break;
 			}
 			/* Inode not in memory or we found it already,
 			 * nothing to do
 			 */
 			if (!ip || (ip->i_flags & XFS_ISTALE)) {
 				read_unlock(&ih->ih_lock);
 				continue;
 			}
 			if (xfs_inode_clean(ip)) {
 				read_unlock(&ih->ih_lock);
 				continue;
 			}
 			/* If we can get the locks then add it to the
 			 * list, otherwise by the time we get the bp lock
 			 * below it will already be attached to the
 			 * inode buffer.
 			 */
 			/* This inode will already be locked - by us, lets
 			 * keep it that way.
 			 */
 			if (ip == free_ip) {
 				if (xfs_iflock_nowait(ip)) {
 					ip->i_flags |= XFS_ISTALE;
 					if (xfs_inode_clean(ip)) {
 						xfs_ifunlock(ip);
 					} else {
 						ip_found[found++] = ip;
 					}
 				}
 				read_unlock(&ih->ih_lock);
 				continue;
 			}
 			if (xfs_ilock_nowait(ip, XFS_ILOCK_EXCL)) {
 				if (xfs_iflock_nowait(ip)) {
 					ip->i_flags |= XFS_ISTALE;
 					if (xfs_inode_clean(ip)) {
 						xfs_ifunlock(ip);
 						xfs_iunlock(ip, XFS_ILOCK_EXCL);
 					} else {
 						ip_found[found++] = ip;
 					}
 				} else {
 					xfs_iunlock(ip, XFS_ILOCK_EXCL);
 				}
 			}
 			read_unlock(&ih->ih_lock);
 		}
 		bp = xfs_trans_get_buf(tp, mp->m_ddev_targp, blkno,
 					mp->m_bsize * blks_per_cluster,
 					XFS_BUF_LOCK);
 		pre_flushed = 0;
 		lip = XFS_BUF_FSPRIVATE(bp, xfs_log_item_t *);
 		while (lip) {
 			if (lip->li_type == XFS_LI_INODE) {
 				iip = (xfs_inode_log_item_t *)lip;
 				ASSERT(iip->ili_logged == 1);
 				lip->li_cb = (void(*)(xfs_buf_t*,xfs_log_item_t*)) xfs_istale_done;
 				AIL_LOCK(mp,s);
 				iip->ili_flush_lsn = iip->ili_item.li_lsn;
 				AIL_UNLOCK(mp, s);
 				iip->ili_inode->i_flags |= XFS_ISTALE;
 				pre_flushed++;
 			}
 			lip = lip->li_bio_list;
 		}
 		for (i = 0; i < found; i++) {
 			ip = ip_found[i];
 			iip = ip->i_itemp;
 			if (!iip) {
 				ip->i_update_core = 0;
 				xfs_ifunlock(ip);
 				xfs_iunlock(ip, XFS_ILOCK_EXCL);
 				continue;
 			}
 			iip->ili_last_fields = iip->ili_format.ilf_fields;
 			iip->ili_format.ilf_fields = 0;
 			iip->ili_logged = 1;
 			AIL_LOCK(mp,s);
 			iip->ili_flush_lsn = iip->ili_item.li_lsn;
 			AIL_UNLOCK(mp, s);
 			xfs_buf_attach_iodone(bp,
 				(void(*)(xfs_buf_t*,xfs_log_item_t*))
 				xfs_istale_done, (xfs_log_item_t *)iip);
 			if (ip != free_ip) {
 				xfs_iunlock(ip, XFS_ILOCK_EXCL);
 			}
 		}
 		if (found || pre_flushed)
 			xfs_trans_stale_inode_buf(tp, bp);
 		xfs_trans_binval(tp, bp);
 	}
 	kmem_free(ip_found, ninodes * sizeof(xfs_inode_t *));
 }
 /*
  * This is called to return an inode to the inode free list.
  * The inode should already be truncated to 0 length and have
  * no pages associated with it.  This routine also assumes that
  * the inode is already a part of the transaction.
  *
  * The on-disk copy of the inode will have been added to the list
  * of unlinked inodes in the AGI. We need to remove the inode from
  * that list atomically with respect to freeing it here.
  */
 int
 xfs_ifree(
 	xfs_trans_t	*tp,
 	xfs_inode_t	*ip,
 	xfs_bmap_free_t	*flist)
 {
 	int			error;
 	int			delete;
 	xfs_ino_t		first_ino;
 	ASSERT(ismrlocked(&ip->i_lock, MR_UPDATE));
 	ASSERT(ip->i_transp == tp);
 	ASSERT(ip->i_d.di_nlink == 0);
 	ASSERT(ip->i_d.di_nextents == 0);
 	ASSERT(ip->i_d.di_anextents == 0);
 	ASSERT((ip->i_d.di_size == 0) ||
 	       ((ip->i_d.di_mode & S_IFMT) != S_IFREG));
 	ASSERT(ip->i_d.di_nblocks == 0);
 	/*
 	 * Pull the on-disk inode from the AGI unlinked list.
 	 */
 	error = xfs_iunlink_remove(tp, ip);
 	if (error != 0) {
 		return error;
 	}
 	error = xfs_difree(tp, ip->i_ino, flist, &delete, &first_ino);
 	if (error != 0) {
 		return error;
 	}
 	ip->i_d.di_mode = 0;		/* mark incore inode as free */
 	ip->i_d.di_flags = 0;
 	ip->i_d.di_dmevmask = 0;
 	ip->i_d.di_forkoff = 0;		/* mark the attr fork not in use */
 	ip->i_df.if_ext_max =
 		XFS_IFORK_DSIZE(ip) / (uint)sizeof(xfs_bmbt_rec_t);
 	ip->i_d.di_format = XFS_DINODE_FMT_EXTENTS;
 	ip->i_d.di_aformat = XFS_DINODE_FMT_EXTENTS;
 	/*
 	 * Bump the generation count so no one will be confused
 	 * by reincarnations of this inode.
 	 */
 	ip->i_d.di_gen++;
 	xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
 	if (delete) {
 		xfs_ifree_cluster(ip, tp, first_ino);
 	}
 	return 0;
 }
 /*
  * Reallocate the space for if_broot based on the number of records
  * being added or deleted as indicated in rec_diff.  Move the records
  * and pointers in if_broot to fit the new size.  When shrinking this
  * will eliminate holes between the records and pointers created by
  * the caller.  When growing this will create holes to be filled in
  * by the caller.
  *
  * The caller must not request to add more records than would fit in
  * the on-disk inode root.  If the if_broot is currently NULL, then
  * if we adding records one will be allocated.  The caller must also
  * not request that the number of records go below zero, although
  * it can go to zero.
  *
  * ip -- the inode whose if_broot area is changing
  * ext_diff -- the change in the number of records, positive or negative,
  *	 requested for the if_broot array.
  */
 void
 xfs_iroot_realloc(
 	xfs_inode_t		*ip,
 	int			rec_diff,
 	int			whichfork)
 {
 	int			cur_max;
 	xfs_ifork_t		*ifp;
 	xfs_bmbt_block_t	*new_broot;
 	int			new_max;
 	size_t			new_size;
 	char			*np;
 	char			*op;
 	/*
 	 * Handle the degenerate case quietly.
 	 */
 	if (rec_diff == 0) {
 		return;
 	}
 	ifp = XFS_IFORK_PTR(ip, whichfork);
 	if (rec_diff > 0) {
 		/*
 		 * If there wasn't any memory allocated before, just
 		 * allocate it now and get out.
 		 */
 		if (ifp->if_broot_bytes == 0) {
 			new_size = (size_t)XFS_BMAP_BROOT_SPACE_CALC(rec_diff);
 			ifp->if_broot = (xfs_bmbt_block_t*)kmem_alloc(new_size,
 								     KM_SLEEP);
 			ifp->if_broot_bytes = (int)new_size;
 			return;
 		}
 		/*
 		 * If there is already an existing if_broot, then we need
 		 * to realloc() it and shift the pointers to their new
 		 * location.  The records don't change location because
 		 * they are kept butted up against the btree block header.
 		 */
 		cur_max = XFS_BMAP_BROOT_MAXRECS(ifp->if_broot_bytes);
 		new_max = cur_max + rec_diff;
 		new_size = (size_t)XFS_BMAP_BROOT_SPACE_CALC(new_max);
 		ifp->if_broot = (xfs_bmbt_block_t *)
 		  kmem_realloc(ifp->if_broot,
 				new_size,
 				(size_t)XFS_BMAP_BROOT_SPACE_CALC(cur_max), /* old size */
 				KM_SLEEP);
 		op = (char *)XFS_BMAP_BROOT_PTR_ADDR(ifp->if_broot, 1,
 						      ifp->if_broot_bytes);
 		np = (char *)XFS_BMAP_BROOT_PTR_ADDR(ifp->if_broot, 1,
 						      (int)new_size);
 		ifp->if_broot_bytes = (int)new_size;
 		ASSERT(ifp->if_broot_bytes <=
 			XFS_IFORK_SIZE(ip, whichfork) + XFS_BROOT_SIZE_ADJ);
 		memmove(np, op, cur_max * (uint)sizeof(xfs_dfsbno_t));
 		return;
 	}
 	/*
 	 * rec_diff is less than 0.  In this case, we are shrinking the
 	 * if_broot buffer.  It must already exist.  If we go to zero
 	 * records, just get rid of the root and clear the status bit.
 	 */
 	ASSERT((ifp->if_broot != NULL) && (ifp->if_broot_bytes > 0));
 	cur_max = XFS_BMAP_BROOT_MAXRECS(ifp->if_broot_bytes);
 	new_max = cur_max + rec_diff;
 	ASSERT(new_max >= 0);
 	if (new_max > 0)
 		new_size = (size_t)XFS_BMAP_BROOT_SPACE_CALC(new_max);
 	else
 		new_size = 0;
 	if (new_size > 0) {
 		new_broot = (xfs_bmbt_block_t *)kmem_alloc(new_size, KM_SLEEP);
 		/*
 		 * First copy over the btree block header.
 		 */
 		memcpy(new_broot, ifp->if_broot, sizeof(xfs_bmbt_block_t));
 	} else {
 		new_broot = NULL;
 		ifp->if_flags &= ~XFS_IFBROOT;
 	}
 	/*
 	 * Only copy the records and pointers if there are any.
 	 */
 	if (new_max > 0) {
 		/*
 		 * First copy the records.
 		 */
 		op = (char *)XFS_BMAP_BROOT_REC_ADDR(ifp->if_broot, 1,
 						     ifp->if_broot_bytes);
 		np = (char *)XFS_BMAP_BROOT_REC_ADDR(new_broot, 1,
 						     (int)new_size);
 		memcpy(np, op, new_max * (uint)sizeof(xfs_bmbt_rec_t));
 		/*
 		 * Then copy the pointers.
 		 */
 		op = (char *)XFS_BMAP_BROOT_PTR_ADDR(ifp->if_broot, 1,
 						     ifp->if_broot_bytes);
 		np = (char *)XFS_BMAP_BROOT_PTR_ADDR(new_broot, 1,
 						     (int)new_size);
 		memcpy(np, op, new_max * (uint)sizeof(xfs_dfsbno_t));
 	}
 	kmem_free(ifp->if_broot, ifp->if_broot_bytes);
 	ifp->if_broot = new_broot;
 	ifp->if_broot_bytes = (int)new_size;
 	ASSERT(ifp->if_broot_bytes <=
 		XFS_IFORK_SIZE(ip, whichfork) + XFS_BROOT_SIZE_ADJ);
 	return;
 }
 /*
  * This is called when the amount of space needed for if_data
  * is increased or decreased.  The change in size is indicated by
  * the number of bytes that need to be added or deleted in the
  * byte_diff parameter.
  *
  * If the amount of space needed has decreased below the size of the
  * inline buffer, then switch to using the inline buffer.  Otherwise,
  * use kmem_realloc() or kmem_alloc() to adjust the size of the buffer
  * to what is needed.
  *
  * ip -- the inode whose if_data area is changing
  * byte_diff -- the change in the number of bytes, positive or negative,
  *	 requested for the if_data array.
  */
 void
 xfs_idata_realloc(
 	xfs_inode_t	*ip,
 	int		byte_diff,
 	int		whichfork)
 {
 	xfs_ifork_t	*ifp;
 	int		new_size;
 	int		real_size;
 	if (byte_diff == 0) {
 		return;
 	}
 	ifp = XFS_IFORK_PTR(ip, whichfork);
 	new_size = (int)ifp->if_bytes + byte_diff;
 	ASSERT(new_size >= 0);
 	if (new_size == 0) {
 		if (ifp->if_u1.if_data != ifp->if_u2.if_inline_data) {
 			kmem_free(ifp->if_u1.if_data, ifp->if_real_bytes);
 		}
 		ifp->if_u1.if_data = NULL;
 		real_size = 0;
 	} else if (new_size <= sizeof(ifp->if_u2.if_inline_data)) {
 		/*
 		 * If the valid extents/data can fit in if_inline_ext/data,
 		 * copy them from the malloc'd vector and free it.
 		 */
 		if (ifp->if_u1.if_data == NULL) {
 			ifp->if_u1.if_data = ifp->if_u2.if_inline_data;
 		} else if (ifp->if_u1.if_data != ifp->if_u2.if_inline_data) {
 			ASSERT(ifp->if_real_bytes != 0);
 			memcpy(ifp->if_u2.if_inline_data, ifp->if_u1.if_data,
 			      new_size);
 			kmem_free(ifp->if_u1.if_data, ifp->if_real_bytes);
 			ifp->if_u1.if_data = ifp->if_u2.if_inline_data;
 		}
 		real_size = 0;
 	} else {
 		/*
 		 * Stuck with malloc/realloc.
 		 * For inline data, the underlying buffer must be
 		 * a multiple of 4 bytes in size so that it can be
 		 * logged and stay on word boundaries.  We enforce
 		 * that here.
 		 */
 		real_size = roundup(new_size, 4);
 		if (ifp->if_u1.if_data == NULL) {
 			ASSERT(ifp->if_real_bytes == 0);
 			ifp->if_u1.if_data = kmem_alloc(real_size, KM_SLEEP);
 		} else if (ifp->if_u1.if_data != ifp->if_u2.if_inline_data) {
 			/*
 			 * Only do the realloc if the underlying size
 			 * is really changing.
 			 */
 			if (ifp->if_real_bytes != real_size) {
 				ifp->if_u1.if_data =
 					kmem_realloc(ifp->if_u1.if_data,
 							real_size,
 							ifp->if_real_bytes,
 							KM_SLEEP);
 			}
 		} else {
 			ASSERT(ifp->if_real_bytes == 0);
 			ifp->if_u1.if_data = kmem_alloc(real_size, KM_SLEEP);
 			memcpy(ifp->if_u1.if_data, ifp->if_u2.if_inline_data,
 				ifp->if_bytes);
 		}
 	}
 	ifp->if_real_bytes = real_size;
 	ifp->if_bytes = new_size;
 	ASSERT(ifp->if_bytes <= XFS_IFORK_SIZE(ip, whichfork));
 }
 /*
  * Map inode to disk block and offset.
  *
  * mp -- the mount point structure for the current file system
  * tp -- the current transaction
  * ino -- the inode number of the inode to be located
  * imap -- this structure is filled in with the information necessary
  *	 to retrieve the given inode from disk
  * flags -- flags to pass to xfs_dilocate indicating whether or not
  *	 lookups in the inode btree were OK or not
  */
 int
 xfs_imap(
 	xfs_mount_t	*mp,
 	xfs_trans_t	*tp,
 	xfs_ino_t	ino,
 	xfs_imap_t	*imap,
 	uint		flags)
 {
 	xfs_fsblock_t	fsbno;
 	int		len;
 	int		off;
 	int		error;
 	fsbno = imap->im_blkno ?
 		XFS_DADDR_TO_FSB(mp, imap->im_blkno) : NULLFSBLOCK;
 	error = xfs_dilocate(mp, tp, ino, &fsbno, &len, &off, flags);
 	if (error != 0) {
 		return error;
 	}
 	imap->im_blkno = XFS_FSB_TO_DADDR(mp, fsbno);
 	imap->im_len = XFS_FSB_TO_BB(mp, len);
 	imap->im_agblkno = XFS_FSB_TO_AGBNO(mp, fsbno);
 	imap->im_ioffset = (ushort)off;
 	imap->im_boffset = (ushort)(off << mp->m_sb.sb_inodelog);
 	return 0;
 }
 void
 xfs_idestroy_fork(
 	xfs_inode_t	*ip,
 	int		whichfork)
 {
 	xfs_ifork_t	*ifp;
 	ifp = XFS_IFORK_PTR(ip, whichfork);
 	if (ifp->if_broot != NULL) {
 		kmem_free(ifp->if_broot, ifp->if_broot_bytes);
 		ifp->if_broot = NULL;
 	}
 	/*
 	 * If the format is local, then we can't have an extents
 	 * array so just look for an inline data array.  If we're
 	 * not local then we may or may not have an extents list,
 	 * so check and free it up if we do.
 	 */
 	if (XFS_IFORK_FORMAT(ip, whichfork) == XFS_DINODE_FMT_LOCAL) {
 		if ((ifp->if_u1.if_data != ifp->if_u2.if_inline_data) &&
 		    (ifp->if_u1.if_data != NULL)) {
 			ASSERT(ifp->if_real_bytes != 0);
 			kmem_free(ifp->if_u1.if_data, ifp->if_real_bytes);
 			ifp->if_u1.if_data = NULL;
 			ifp->if_real_bytes = 0;
 		}
 	} else if ((ifp->if_flags & XFS_IFEXTENTS) &&
 		   ((ifp->if_flags & XFS_IFEXTIREC) ||
 		    ((ifp->if_u1.if_extents != NULL) &&
 		     (ifp->if_u1.if_extents != ifp->if_u2.if_inline_ext)))) {
 		ASSERT(ifp->if_real_bytes != 0);
 		xfs_iext_destroy(ifp);
 	}
 	ASSERT(ifp->if_u1.if_extents == NULL ||
 	       ifp->if_u1.if_extents == ifp->if_u2.if_inline_ext);
 	ASSERT(ifp->if_real_bytes == 0);
 	if (whichfork == XFS_ATTR_FORK) {
 		kmem_zone_free(xfs_ifork_zone, ip->i_afp);
 		ip->i_afp = NULL;
 	}
 }
 /*
  * This is called free all the memory associated with an inode.
  * It must free the inode itself and any buffers allocated for
  * if_extents/if_data and if_broot.  It must also free the lock
  * associated with the inode.
  */
 void
 xfs_idestroy(
 	xfs_inode_t	*ip)
 {
 	switch (ip->i_d.di_mode & S_IFMT) {
 	case S_IFREG:
 	case S_IFDIR:
 	case S_IFLNK:
 		xfs_idestroy_fork(ip, XFS_DATA_FORK);
 		break;
 	}
 	if (ip->i_afp)
 		xfs_idestroy_fork(ip, XFS_ATTR_FORK);
 	mrfree(&ip->i_lock);
 	mrfree(&ip->i_iolock);
 	freesema(&ip->i_flock);
 #ifdef XFS_BMAP_TRACE
 	ktrace_free(ip->i_xtrace);
 #endif
 #ifdef XFS_BMBT_TRACE
 	ktrace_free(ip->i_btrace);
 #endif
 #ifdef XFS_RW_TRACE
 	ktrace_free(ip->i_rwtrace);
 #endif
 #ifdef XFS_ILOCK_TRACE
 	ktrace_free(ip->i_lock_trace);
 #endif
 #ifdef XFS_DIR2_TRACE
 	ktrace_free(ip->i_dir_trace);
 #endif
 	if (ip->i_itemp) {
 		/* XXXdpd should be able to assert this but shutdown
 		 * is leaving the AIL behind. */
 		ASSERT(((ip->i_itemp->ili_item.li_flags & XFS_LI_IN_AIL) == 0) ||
 		       XFS_FORCED_SHUTDOWN(ip->i_mount));
 		xfs_inode_item_destroy(ip);
 	}
 	kmem_zone_free(xfs_inode_zone, ip);
 }
 /*
  * Increment the pin count of the given buffer.
  * This value is protected by ipinlock spinlock in the mount structure.
  */
 void
 xfs_ipin(
 	xfs_inode_t	*ip)
 {
 	ASSERT(ismrlocked(&ip->i_lock, MR_UPDATE));
 	atomic_inc(&ip->i_pincount);
 }
 /*
  * Decrement the pin count of the given inode, and wake up
  * anyone in xfs_iwait_unpin() if the count goes to 0.  The
  * inode must have been previously pinned with a call to xfs_ipin().
  */
 void
 xfs_iunpin(
 	xfs_inode_t	*ip)
 {
 	ASSERT(atomic_read(&ip->i_pincount) > 0);
 	if (atomic_dec_and_test(&ip->i_pincount)) {
 		/*
 		 * If the inode is currently being reclaimed, the
 		 * linux inode _and_ the xfs vnode may have been
 		 * freed so we cannot reference either of them safely.
 		 * Hence we should not try to do anything to them
 		 * if the xfs inode is currently in the reclaim
 		 * path.
 		 *
 		 * However, we still need to issue the unpin wakeup
 		 * call as the inode reclaim may be blocked waiting for
 		 * the inode to become unpinned.
 		 */
 		if (!(ip->i_flags & (XFS_IRECLAIM|XFS_IRECLAIMABLE))) {
 			bhv_vnode_t	*vp = XFS_ITOV_NULL(ip);
 			/* make sync come back and flush this inode */
 			if (vp) {
 				struct inode	*inode = vn_to_inode(vp);
 				if (!(inode->i_state &
 						(I_NEW|I_FREEING|I_CLEAR)))
 					mark_inode_dirty_sync(inode);
 			}
 		}
 		wake_up(&ip->i_ipin_wait);
 	}
 }
 /*
  * This is called to wait for the given inode to be unpinned.
  * It will sleep until this happens.  The caller must have the
  * inode locked in at least shared mode so that the buffer cannot
  * be subsequently pinned once someone is waiting for it to be
  * unpinned.
  */
 STATIC void
 xfs_iunpin_wait(
 	xfs_inode_t	*ip)
 {
 	xfs_inode_log_item_t	*iip;
 	xfs_lsn_t	lsn;
 	ASSERT(ismrlocked(&ip->i_lock, MR_UPDATE | MR_ACCESS));
 	if (atomic_read(&ip->i_pincount) == 0) {
 		return;
 	}
 	iip = ip->i_itemp;
 	if (iip && iip->ili_last_lsn) {
 		lsn = iip->ili_last_lsn;
 	} else {
 		lsn = (xfs_lsn_t)0;
 	}
 	/*
 	 * Give the log a push so we don't wait here too long.
 	 */
 	xfs_log_force(ip->i_mount, lsn, XFS_LOG_FORCE);
 	wait_event(ip->i_ipin_wait, (atomic_read(&ip->i_pincount) == 0));
 }
 /*
  * xfs_iextents_copy()
  *
  * This is called to copy the REAL extents (as opposed to the delayed
  * allocation extents) from the inode into the given buffer.  It
  * returns the number of bytes copied into the buffer.
  *
  * If there are no delayed allocation extents, then we can just
  * memcpy() the extents into the buffer.  Otherwise, we need to
  * examine each extent in turn and skip those which are delayed.
  */
 int
 xfs_iextents_copy(
 	xfs_inode_t		*ip,
 	xfs_bmbt_rec_t		*buffer,
 	int			whichfork)
 {
 	int			copied;
 	xfs_bmbt_rec_t		*dest_ep;
 	xfs_bmbt_rec_t		*ep;
 #ifdef XFS_BMAP_TRACE
 	static char		fname[] = "xfs_iextents_copy";
 #endif
 	int			i;
 	xfs_ifork_t		*ifp;
 	int			nrecs;
 	xfs_fsblock_t		start_block;
 	ifp = XFS_IFORK_PTR(ip, whichfork);
 	ASSERT(ismrlocked(&ip->i_lock, MR_UPDATE|MR_ACCESS));
 	ASSERT(ifp->if_bytes > 0);
 	nrecs = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t);
 	xfs_bmap_trace_exlist(fname, ip, nrecs, whichfork);
 	ASSERT(nrecs > 0);
 	/*
 	 * There are some delayed allocation extents in the
 	 * inode, so copy the extents one at a time and skip
 	 * the delayed ones.  There must be at least one
 	 * non-delayed extent.
 	 */
 	dest_ep = buffer;
 	copied = 0;
 	for (i = 0; i < nrecs; i++) {
 		ep = xfs_iext_get_ext(ifp, i);
 		start_block = xfs_bmbt_get_startblock(ep);
 		if (ISNULLSTARTBLOCK(start_block)) {
 			/*
 			 * It's a delayed allocation extent, so skip it.
 			 */
 			continue;
 		}
 		/* Translate to on disk format */
 		put_unaligned(INT_GET(ep->l0, ARCH_CONVERT),
 			      (__uint64_t*)&dest_ep->l0);
 		put_unaligned(INT_GET(ep->l1, ARCH_CONVERT),
 			      (__uint64_t*)&dest_ep->l1);
 		dest_ep++;
 		copied++;
 	}
 	ASSERT(copied != 0);
 	xfs_validate_extents(ifp, copied, 1, XFS_EXTFMT_INODE(ip));
 	return (copied * (uint)sizeof(xfs_bmbt_rec_t));
 }
 /*
  * Each of the following cases stores data into the same region
  * of the on-disk inode, so only one of them can be valid at
  * any given time. While it is possible to have conflicting formats
  * and log flags, e.g. having XFS_ILOG_?DATA set when the fork is
  * in EXTENTS format, this can only happen when the fork has
  * changed formats after being modified but before being flushed.
  * In these cases, the format always takes precedence, because the
  * format indicates the current state of the fork.
  */
 /*ARGSUSED*/
 STATIC int
 xfs_iflush_fork(
 	xfs_inode_t		*ip,
 	xfs_dinode_t		*dip,
 	xfs_inode_log_item_t	*iip,
 	int			whichfork,
 	xfs_buf_t		*bp)
 {
 	char			*cp;
 	xfs_ifork_t		*ifp;
 	xfs_mount_t		*mp;
 #ifdef XFS_TRANS_DEBUG
 	int			first;
 #endif
 	static const short	brootflag[2] =
 		{ XFS_ILOG_DBROOT, XFS_ILOG_ABROOT };
 	static const short	dataflag[2] =
 		{ XFS_ILOG_DDATA, XFS_ILOG_ADATA };
 	static const short	extflag[2] =
 		{ XFS_ILOG_DEXT, XFS_ILOG_AEXT };
 	if (iip == NULL)
 		return 0;
 	ifp = XFS_IFORK_PTR(ip, whichfork);
 	/*
 	 * This can happen if we gave up in iformat in an error path,
 	 * for the attribute fork.
 	 */
 	if (ifp == NULL) {
 		ASSERT(whichfork == XFS_ATTR_FORK);
 		return 0;
 	}
 	cp = XFS_DFORK_PTR(dip, whichfork);
 	mp = ip->i_mount;
 	switch (XFS_IFORK_FORMAT(ip, whichfork)) {
 	case XFS_DINODE_FMT_LOCAL:
 		if ((iip->ili_format.ilf_fields & dataflag[whichfork]) &&
 		    (ifp->if_bytes > 0)) {
 			ASSERT(ifp->if_u1.if_data != NULL);
 			ASSERT(ifp->if_bytes <= XFS_IFORK_SIZE(ip, whichfork));
 			memcpy(cp, ifp->if_u1.if_data, ifp->if_bytes);
 		}
 		break;
 	case XFS_DINODE_FMT_EXTENTS:
 		ASSERT((ifp->if_flags & XFS_IFEXTENTS) ||
 		       !(iip->ili_format.ilf_fields & extflag[whichfork]));
 		ASSERT((xfs_iext_get_ext(ifp, 0) != NULL) ||
 			(ifp->if_bytes == 0));
 		ASSERT((xfs_iext_get_ext(ifp, 0) == NULL) ||
 			(ifp->if_bytes > 0));
 		if ((iip->ili_format.ilf_fields & extflag[whichfork]) &&
 		    (ifp->if_bytes > 0)) {
 			ASSERT(XFS_IFORK_NEXTENTS(ip, whichfork) > 0);
 			(void)xfs_iextents_copy(ip, (xfs_bmbt_rec_t *)cp,
 				whichfork);
 		}
 		break;
 	case XFS_DINODE_FMT_BTREE:
 		if ((iip->ili_format.ilf_fields & brootflag[whichfork]) &&
 		    (ifp->if_broot_bytes > 0)) {
 			ASSERT(ifp->if_broot != NULL);
 			ASSERT(ifp->if_broot_bytes <=
 			       (XFS_IFORK_SIZE(ip, whichfork) +
 				XFS_BROOT_SIZE_ADJ));
 			xfs_bmbt_to_bmdr(ifp->if_broot, ifp->if_broot_bytes,
 				(xfs_bmdr_block_t *)cp,
 				XFS_DFORK_SIZE(dip, mp, whichfork));
 		}
 		break;
 	case XFS_DINODE_FMT_DEV:
 		if (iip->ili_format.ilf_fields & XFS_ILOG_DEV) {
 			ASSERT(whichfork == XFS_DATA_FORK);
 			INT_SET(dip->di_u.di_dev, ARCH_CONVERT, ip->i_df.if_u2.if_rdev);
 		}
 		break;
 	case XFS_DINODE_FMT_UUID:
 		if (iip->ili_format.ilf_fields & XFS_ILOG_UUID) {
 			ASSERT(whichfork == XFS_DATA_FORK);
 			memcpy(&dip->di_u.di_muuid, &ip->i_df.if_u2.if_uuid,
 				sizeof(uuid_t));
 		}
 		break;
 	default:
 		ASSERT(0);
 		break;
 	}
 	return 0;
 }
 /*
  * xfs_iflush() will write a modified inode's changes out to the
  * inode's on disk home.  The caller must have the inode lock held
  * in at least shared mode and the inode flush semaphore must be
  * held as well.  The inode lock will still be held upon return from
  * the call and the caller is free to unlock it.
  * The inode flush lock will be unlocked when the inode reaches the disk.
  * The flags indicate how the inode's buffer should be written out.
  */
 int
 xfs_iflush(
 	xfs_inode_t		*ip,
 	uint			flags)
 {
 	xfs_inode_log_item_t	*iip;
 	xfs_buf_t		*bp;
 	xfs_dinode_t		*dip;
 	xfs_mount_t		*mp;
 	int			error;
 	/* REFERENCED */
 	xfs_chash_t		*ch;
 	xfs_inode_t		*iq;
 	int			clcount;	/* count of inodes clustered */
 	int			bufwasdelwri;
 	enum { INT_DELWRI = (1 << 0), INT_ASYNC = (1 << 1) };
 	SPLDECL(s);
 	XFS_STATS_INC(xs_iflush_count);
 	ASSERT(ismrlocked(&ip->i_lock, MR_UPDATE|MR_ACCESS));
 	ASSERT(issemalocked(&(ip->i_flock)));
 	ASSERT(ip->i_d.di_format != XFS_DINODE_FMT_BTREE ||
 	       ip->i_d.di_nextents > ip->i_df.if_ext_max);
 	iip = ip->i_itemp;
 	mp = ip->i_mount;
 	/*
 	 * If the inode isn't dirty, then just release the inode
 	 * flush lock and do nothing.
 	 */
 	if ((ip->i_update_core == 0) &&
 	    ((iip == NULL) || !(iip->ili_format.ilf_fields & XFS_ILOG_ALL))) {
 		ASSERT((iip != NULL) ?
 			 !(iip->ili_item.li_flags & XFS_LI_IN_AIL) : 1);
 		xfs_ifunlock(ip);
 		return 0;
 	}
 	/*
 	 * We can't flush the inode until it is unpinned, so
 	 * wait for it.  We know noone new can pin it, because
 	 * we are holding the inode lock shared and you need
 	 * to hold it exclusively to pin the inode.
 	 */
 	xfs_iunpin_wait(ip);
 	/*
 	 * This may have been unpinned because the filesystem is shutting
 	 * down forcibly. If that's the case we must not write this inode
 	 * to disk, because the log record didn't make it to disk!
 	 */
 	if (XFS_FORCED_SHUTDOWN(mp)) {
 		ip->i_update_core = 0;
 		if (iip)
 			iip->ili_format.ilf_fields = 0;
 		xfs_ifunlock(ip);
 		return XFS_ERROR(EIO);
 	}
 	/*
 	 * Get the buffer containing the on-disk inode.
 	 */
 	error = xfs_itobp(mp, NULL, ip, &dip, &bp, 0, 0);
 	if (error) {
 		xfs_ifunlock(ip);
 		return error;
 	}
 	/*
 	 * Decide how buffer will be flushed out.  This is done before
 	 * the call to xfs_iflush_int because this field is zeroed by it.
 	 */
 	if (iip != NULL && iip->ili_format.ilf_fields != 0) {
 		/*
 		 * Flush out the inode buffer according to the directions
 		 * of the caller.  In the cases where the caller has given
 		 * us a choice choose the non-delwri case.  This is because
 		 * the inode is in the AIL and we need to get it out soon.
 		 */
 		switch (flags) {
 		case XFS_IFLUSH_SYNC:
 		case XFS_IFLUSH_DELWRI_ELSE_SYNC:
 			flags = 0;
 			break;
 		case XFS_IFLUSH_ASYNC:
 		case XFS_IFLUSH_DELWRI_ELSE_ASYNC:
 			flags = INT_ASYNC;
 			break;
 		case XFS_IFLUSH_DELWRI:
 			flags = INT_DELWRI;
 			break;
 		default:
 			ASSERT(0);
 			flags = 0;
 			break;
 		}
 	} else {
 		switch (flags) {
 		case XFS_IFLUSH_DELWRI_ELSE_SYNC:
 		case XFS_IFLUSH_DELWRI_ELSE_ASYNC:
 		case XFS_IFLUSH_DELWRI:
 			flags = INT_DELWRI;
 			break;
 		case XFS_IFLUSH_ASYNC:
 			flags = INT_ASYNC;
 			break;
 		case XFS_IFLUSH_SYNC:
 			flags = 0;
 			break;
 		default:
 			ASSERT(0);
 			flags = 0;
 			break;
 		}
 	}
 	/*
 	 * First flush out the inode that xfs_iflush was called with.
 	 */
 	error = xfs_iflush_int(ip, bp);
 	if (error) {
 		goto corrupt_out;
 	}
 	/*
 	 * inode clustering:
 	 * see if other inodes can be gathered into this write
 	 */
 	ip->i_chash->chl_buf = bp;
 	ch = XFS_CHASH(mp, ip->i_blkno);
 	s = mutex_spinlock(&ch->ch_lock);
 	clcount = 0;
 	for (iq = ip->i_cnext; iq != ip; iq = iq->i_cnext) {
 		/*
 		 * Do an un-protected check to see if the inode is dirty and
 		 * is a candidate for flushing.  These checks will be repeated
 		 * later after the appropriate locks are acquired.
 		 */
 		iip = iq->i_itemp;
 		if ((iq->i_update_core == 0) &&
 		    ((iip == NULL) ||
 		     !(iip->ili_format.ilf_fields & XFS_ILOG_ALL)) &&
 		      xfs_ipincount(iq) == 0) {
 			continue;
 		}
 		/*
 		 * Try to get locks.  If any are unavailable,
 		 * then this inode cannot be flushed and is skipped.
 		 */
 		/* get inode locks (just i_lock) */
 		if (xfs_ilock_nowait(iq, XFS_ILOCK_SHARED)) {
 			/* get inode flush lock */
 			if (xfs_iflock_nowait(iq)) {
 				/* check if pinned */
 				if (xfs_ipincount(iq) == 0) {
 					/* arriving here means that
 					 * this inode can be flushed.
 					 * first re-check that it's
 					 * dirty
 					 */
 					iip = iq->i_itemp;
 					if ((iq->i_update_core != 0)||
 					    ((iip != NULL) &&
 					     (iip->ili_format.ilf_fields & XFS_ILOG_ALL))) {
 						clcount++;
 						error = xfs_iflush_int(iq, bp);
 						if (error) {
 							xfs_iunlock(iq,
 								    XFS_ILOCK_SHARED);
 							goto cluster_corrupt_out;
 						}
 					} else {
 						xfs_ifunlock(iq);
 					}
 				} else {
 					xfs_ifunlock(iq);
 				}
 			}
 			xfs_iunlock(iq, XFS_ILOCK_SHARED);
 		}
 	}
 	mutex_spinunlock(&ch->ch_lock, s);
 	if (clcount) {
 		XFS_STATS_INC(xs_icluster_flushcnt);
 		XFS_STATS_ADD(xs_icluster_flushinode, clcount);
 	}
 	/*
 	 * If the buffer is pinned then push on the log so we won't
 	 * get stuck waiting in the write for too long.
 	 */
 	if (XFS_BUF_ISPINNED(bp)){
 		xfs_log_force(mp, (xfs_lsn_t)0, XFS_LOG_FORCE);
 	}
 	if (flags & INT_DELWRI) {
 		xfs_bdwrite(mp, bp);
 	} else if (flags & INT_ASYNC) {
 		xfs_bawrite(mp, bp);
 	} else {
 		error = xfs_bwrite(mp, bp);
 	}
 	return error;
 corrupt_out:
 	xfs_buf_relse(bp);
 	xfs_force_shutdown(mp, SHUTDOWN_CORRUPT_INCORE);
 	xfs_iflush_abort(ip);
 	/*
 	 * Unlocks the flush lock
 	 */
 	return XFS_ERROR(EFSCORRUPTED);
 cluster_corrupt_out:
 	/* Corruption detected in the clustering loop.  Invalidate the
 	 * inode buffer and shut down the filesystem.
 	 */
 	mutex_spinunlock(&ch->ch_lock, s);
 	/*
 	 * Clean up the buffer.  If it was B_DELWRI, just release it --
 	 * brelse can handle it with no problems.  If not, shut down the
 	 * filesystem before releasing the buffer.
 	 */
 	if ((bufwasdelwri= XFS_BUF_ISDELAYWRITE(bp))) {
 		xfs_buf_relse(bp);
 	}
 	xfs_force_shutdown(mp, SHUTDOWN_CORRUPT_INCORE);
 	if(!bufwasdelwri)  {
 		/*
 		 * Just like incore_relse: if we have b_iodone functions,
 		 * mark the buffer as an error and call them.  Otherwise
 		 * mark it as stale and brelse.
 		 */
 		if (XFS_BUF_IODONE_FUNC(bp)) {
 			XFS_BUF_CLR_BDSTRAT_FUNC(bp);
 			XFS_BUF_UNDONE(bp);
 			XFS_BUF_STALE(bp);
 			XFS_BUF_SHUT(bp);
 			XFS_BUF_ERROR(bp,EIO);
 			xfs_biodone(bp);
 		} else {
 			XFS_BUF_STALE(bp);
 			xfs_buf_relse(bp);
 		}
 	}
 	xfs_iflush_abort(iq);
 	/*
 	 * Unlocks the flush lock
 	 */
 	return XFS_ERROR(EFSCORRUPTED);
 }
 STATIC int
 xfs_iflush_int(
 	xfs_inode_t		*ip,
 	xfs_buf_t		*bp)
 {
 	xfs_inode_log_item_t	*iip;
 	xfs_dinode_t		*dip;
 	xfs_mount_t		*mp;
 #ifdef XFS_TRANS_DEBUG
 	int			first;
 #endif
 	SPLDECL(s);
 	ASSERT(ismrlocked(&ip->i_lock, MR_UPDATE|MR_ACCESS));
 	ASSERT(issemalocked(&(ip->i_flock)));
 	ASSERT(ip->i_d.di_format != XFS_DINODE_FMT_BTREE ||
 	       ip->i_d.di_nextents > ip->i_df.if_ext_max);
 	iip = ip->i_itemp;
 	mp = ip->i_mount;
 	/*
 	 * If the inode isn't dirty, then just release the inode
 	 * flush lock and do nothing.
 	 */
 	if ((ip->i_update_core == 0) &&
 	    ((iip == NULL) || !(iip->ili_format.ilf_fields & XFS_ILOG_ALL))) {
 		xfs_ifunlock(ip);
 		return 0;
 	}
 	/* set *dip = inode's place in the buffer */
 	dip = (xfs_dinode_t *)xfs_buf_offset(bp, ip->i_boffset);
 	/*
 	 * Clear i_update_core before copying out the data.
 	 * This is for coordination with our timestamp updates
 	 * that don't hold the inode lock. They will always
 	 * update the timestamps BEFORE setting i_update_core,
 	 * so if we clear i_update_core after they set it we
 	 * are guaranteed to see their updates to the timestamps.
 	 * I believe that this depends on strongly ordered memory
 	 * semantics, but we have that.  We use the SYNCHRONIZE
 	 * macro to make sure that the compiler does not reorder
 	 * the i_update_core access below the data copy below.
 	 */
 	ip->i_update_core = 0;
 	SYNCHRONIZE();
 	/*
 	 * Make sure to get the latest atime from the Linux inode.
 	 */
 	xfs_synchronize_atime(ip);
 	if (XFS_TEST_ERROR(INT_GET(dip->di_core.di_magic,ARCH_CONVERT) != XFS_DINODE_MAGIC,
 			       mp, XFS_ERRTAG_IFLUSH_1, XFS_RANDOM_IFLUSH_1)) {
 		xfs_cmn_err(XFS_PTAG_IFLUSH, CE_ALERT, mp,
 		    "xfs_iflush: Bad inode %Lu magic number 0x%x, ptr 0x%p",
 			ip->i_ino, (int) INT_GET(dip->di_core.di_magic, ARCH_CONVERT), dip);
 		goto corrupt_out;
 	}
 	if (XFS_TEST_ERROR(ip->i_d.di_magic != XFS_DINODE_MAGIC,
 				mp, XFS_ERRTAG_IFLUSH_2, XFS_RANDOM_IFLUSH_2)) {
 		xfs_cmn_err(XFS_PTAG_IFLUSH, CE_ALERT, mp,
 			"xfs_iflush: Bad inode %Lu, ptr 0x%p, magic number 0x%x",
 			ip->i_ino, ip, ip->i_d.di_magic);
 		goto corrupt_out;
 	}
 	if ((ip->i_d.di_mode & S_IFMT) == S_IFREG) {
 		if (XFS_TEST_ERROR(
 		    (ip->i_d.di_format != XFS_DINODE_FMT_EXTENTS) &&
 		    (ip->i_d.di_format != XFS_DINODE_FMT_BTREE),
 		    mp, XFS_ERRTAG_IFLUSH_3, XFS_RANDOM_IFLUSH_3)) {
 			xfs_cmn_err(XFS_PTAG_IFLUSH, CE_ALERT, mp,
 				"xfs_iflush: Bad regular inode %Lu, ptr 0x%p",
 				ip->i_ino, ip);
 			goto corrupt_out;
 		}
 	} else if ((ip->i_d.di_mode & S_IFMT) == S_IFDIR) {
 		if (XFS_TEST_ERROR(
 		    (ip->i_d.di_format != XFS_DINODE_FMT_EXTENTS) &&
 		    (ip->i_d.di_format != XFS_DINODE_FMT_BTREE) &&
 		    (ip->i_d.di_format != XFS_DINODE_FMT_LOCAL),
 		    mp, XFS_ERRTAG_IFLUSH_4, XFS_RANDOM_IFLUSH_4)) {
 			xfs_cmn_err(XFS_PTAG_IFLUSH, CE_ALERT, mp,
 				"xfs_iflush: Bad directory inode %Lu, ptr 0x%p",
 				ip->i_ino, ip);
 			goto corrupt_out;
 		}
 	}
 	if (XFS_TEST_ERROR(ip->i_d.di_nextents + ip->i_d.di_anextents >
 				ip->i_d.di_nblocks, mp, XFS_ERRTAG_IFLUSH_5,
 				XFS_RANDOM_IFLUSH_5)) {
 		xfs_cmn_err(XFS_PTAG_IFLUSH, CE_ALERT, mp,
 			"xfs_iflush: detected corrupt incore inode %Lu, total extents = %d, nblocks = %Ld, ptr 0x%p",
 			ip->i_ino,
 			ip->i_d.di_nextents + ip->i_d.di_anextents,
 			ip->i_d.di_nblocks,
 			ip);
 		goto corrupt_out;
 	}
 	if (XFS_TEST_ERROR(ip->i_d.di_forkoff > mp->m_sb.sb_inodesize,
 				mp, XFS_ERRTAG_IFLUSH_6, XFS_RANDOM_IFLUSH_6)) {
 		xfs_cmn_err(XFS_PTAG_IFLUSH, CE_ALERT, mp,
 			"xfs_iflush: bad inode %Lu, forkoff 0x%x, ptr 0x%p",
 			ip->i_ino, ip->i_d.di_forkoff, ip);
 		goto corrupt_out;
 	}
 	/*
 	 * bump the flush iteration count, used to detect flushes which
 	 * postdate a log record during recovery.
 	 */
 	ip->i_d.di_flushiter++;
 	/*
 	 * Copy the dirty parts of the inode into the on-disk
 	 * inode.  We always copy out the core of the inode,
 	 * because if the inode is dirty at all the core must
 	 * be.
 	 */
 	xfs_xlate_dinode_core((xfs_caddr_t)&(dip->di_core), &(ip->i_d), -1);
 	/* Wrap, we never let the log put out DI_MAX_FLUSH */
 	if (ip->i_d.di_flushiter == DI_MAX_FLUSH)
 		ip->i_d.di_flushiter = 0;
 	/*
 	 * If this is really an old format inode and the superblock version
 	 * has not been updated to support only new format inodes, then
 	 * convert back to the old inode format.  If the superblock version
 	 * has been updated, then make the conversion permanent.
 	 */
 	ASSERT(ip->i_d.di_version == XFS_DINODE_VERSION_1 ||
 	       XFS_SB_VERSION_HASNLINK(&mp->m_sb));
 	if (ip->i_d.di_version == XFS_DINODE_VERSION_1) {
 		if (!XFS_SB_VERSION_HASNLINK(&mp->m_sb)) {
 			/*
 			 * Convert it back.
 			 */
 			ASSERT(ip->i_d.di_nlink <= XFS_MAXLINK_1);
 			INT_SET(dip->di_core.di_onlink, ARCH_CONVERT, ip->i_d.di_nlink);
 		} else {
 			/*
 			 * The superblock version has already been bumped,
 			 * so just make the conversion to the new inode
 			 * format permanent.
 			 */
 			ip->i_d.di_version = XFS_DINODE_VERSION_2;
 			INT_SET(dip->di_core.di_version, ARCH_CONVERT, XFS_DINODE_VERSION_2);
 			ip->i_d.di_onlink = 0;
 			dip->di_core.di_onlink = 0;
 			memset(&(ip->i_d.di_pad[0]), 0, sizeof(ip->i_d.di_pad));
 			memset(&(dip->di_core.di_pad[0]), 0,
 			      sizeof(dip->di_core.di_pad));
 			ASSERT(ip->i_d.di_projid == 0);
 		}
 	}
 	if (xfs_iflush_fork(ip, dip, iip, XFS_DATA_FORK, bp) == EFSCORRUPTED) {
 		goto corrupt_out;
 	}
 	if (XFS_IFORK_Q(ip)) {
 		/*
 		 * The only error from xfs_iflush_fork is on the data fork.
 		 */
 		(void) xfs_iflush_fork(ip, dip, iip, XFS_ATTR_FORK, bp);
 	}
 	xfs_inobp_check(mp, bp);
 	/*
 	 * We've recorded everything logged in the inode, so we'd
 	 * like to clear the ilf_fields bits so we don't log and
 	 * flush things unnecessarily.  However, we can't stop
 	 * logging all this information until the data we've copied
 	 * into the disk buffer is written to disk.  If we did we might
 	 * overwrite the copy of the inode in the log with all the
 	 * data after re-logging only part of it, and in the face of
 	 * a crash we wouldn't have all the data we need to recover.
 	 *
 	 * What we do is move the bits to the ili_last_fields field.
 	 * When logging the inode, these bits are moved back to the
 	 * ilf_fields field.  In the xfs_iflush_done() routine we
 	 * clear ili_last_fields, since we know that the information
 	 * those bits represent is permanently on disk.  As long as
 	 * the flush completes before the inode is logged again, then
 	 * both ilf_fields and ili_last_fields will be cleared.
 	 *
 	 * We can play with the ilf_fields bits here, because the inode
 	 * lock must be held exclusively in order to set bits there
 	 * and the flush lock protects the ili_last_fields bits.
 	 * Set ili_logged so the flush done
 	 * routine can tell whether or not to look in the AIL.
 	 * Also, store the current LSN of the inode so that we can tell
 	 * whether the item has moved in the AIL from xfs_iflush_done().
 	 * In order to read the lsn we need the AIL lock, because
 	 * it is a 64 bit value that cannot be read atomically.
 	 */
 	if (iip != NULL && iip->ili_format.ilf_fields != 0) {
 		iip->ili_last_fields = iip->ili_format.ilf_fields;
 		iip->ili_format.ilf_fields = 0;
 		iip->ili_logged = 1;
 		ASSERT(sizeof(xfs_lsn_t) == 8);	/* don't lock if it shrinks */
 		AIL_LOCK(mp,s);
 		iip->ili_flush_lsn = iip->ili_item.li_lsn;
 		AIL_UNLOCK(mp, s);
 		/*
 		 * Attach the function xfs_iflush_done to the inode's
 		 * buffer.  This will remove the inode from the AIL
 		 * and unlock the inode's flush lock when the inode is
 		 * completely written to disk.
 		 */
 		xfs_buf_attach_iodone(bp, (void(*)(xfs_buf_t*,xfs_log_item_t*))
 				      xfs_iflush_done, (xfs_log_item_t *)iip);
 		ASSERT(XFS_BUF_FSPRIVATE(bp, void *) != NULL);
 		ASSERT(XFS_BUF_IODONE_FUNC(bp) != NULL);
 	} else {
 		/*
 		 * We're flushing an inode which is not in the AIL and has
 		 * not been logged but has i_update_core set.  For this
 		 * case we can use a B_DELWRI flush and immediately drop
 		 * the inode flush lock because we can avoid the whole
 		 * AIL state thing.  It's OK to drop the flush lock now,
 		 * because we've already locked the buffer and to do anything
 		 * you really need both.
 		 */
 		if (iip != NULL) {
 			ASSERT(iip->ili_logged == 0);
 			ASSERT(iip->ili_last_fields == 0);
 			ASSERT((iip->ili_item.li_flags & XFS_LI_IN_AIL) == 0);
 		}
 		xfs_ifunlock(ip);
 	}
 	return 0;
 corrupt_out:
 	return XFS_ERROR(EFSCORRUPTED);
 }
 /*
  * Flush all inactive inodes in mp.
  */
 void
 xfs_iflush_all(
 	xfs_mount_t	*mp)
 {
 	xfs_inode_t	*ip;
 	bhv_vnode_t	*vp;
  again:
 	XFS_MOUNT_ILOCK(mp);
 	ip = mp->m_inodes;
 	if (ip == NULL)
 		goto out;
 	do {
 		/* Make sure we skip markers inserted by sync */
 		if (ip->i_mount == NULL) {
 			ip = ip->i_mnext;
 			continue;
 		}
 		vp = XFS_ITOV_NULL(ip);
 		if (!vp) {
 			XFS_MOUNT_IUNLOCK(mp);
 			xfs_finish_reclaim(ip, 0, XFS_IFLUSH_ASYNC);
 			goto again;
 		}
 		ASSERT(vn_count(vp) == 0);
 		ip = ip->i_mnext;
 	} while (ip != mp->m_inodes);
  out:
 	XFS_MOUNT_IUNLOCK(mp);
 }
 /*
  * xfs_iaccess: check accessibility of inode for mode.
  */
 int
 xfs_iaccess(
 	xfs_inode_t	*ip,
 	mode_t		mode,
 	cred_t		*cr)
 {
 	int		error;
 	mode_t		orgmode = mode;
 	struct inode	*inode = vn_to_inode(XFS_ITOV(ip));
 	if (mode & S_IWUSR) {
 		umode_t		imode = inode->i_mode;
 		if (IS_RDONLY(inode) &&
 		    (S_ISREG(imode) || S_ISDIR(imode) || S_ISLNK(imode)))
 			return XFS_ERROR(EROFS);
 		if (IS_IMMUTABLE(inode))
 			return XFS_ERROR(EACCES);
 	}
 	/*
 	 * If there's an Access Control List it's used instead of
 	 * the mode bits.
 	 */
 	if ((error = _ACL_XFS_IACCESS(ip, mode, cr)) != -1)
 		return error ? XFS_ERROR(error) : 0;
 	if (current_fsuid(cr) != ip->i_d.di_uid) {
 		mode >>= 3;
 		if (!in_group_p((gid_t)ip->i_d.di_gid))
 			mode >>= 3;
 	}
 	/*
 	 * If the DACs are ok we don't need any capability check.
 	 */
 	if ((ip->i_d.di_mode & mode) == mode)
 		return 0;
 	/*
 	 * Read/write DACs are always overridable.
 	 * Executable DACs are overridable if at least one exec bit is set.
 	 */
 	if (!(orgmode & S_IXUSR) ||
 	    (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode))
 		if (capable_cred(cr, CAP_DAC_OVERRIDE))
 			return 0;
 	if ((orgmode == S_IRUSR) ||
 	    (S_ISDIR(inode->i_mode) && (!(orgmode & S_IWUSR)))) {
 		if (capable_cred(cr, CAP_DAC_READ_SEARCH))
 			return 0;
 #ifdef	NOISE
 		cmn_err(CE_NOTE, "Ick: mode=%o, orgmode=%o", mode, orgmode);
 #endif	/* NOISE */
 		return XFS_ERROR(EACCES);
 	}
 	return XFS_ERROR(EACCES);
 }
 /*
  * xfs_iroundup: round up argument to next power of two
  */
 uint
 xfs_iroundup(
 	uint	v)
 {
 	int i;
 	uint m;
 	if ((v & (v - 1)) == 0)
 		return v;
 	ASSERT((v & 0x80000000) == 0);
 	if ((v & (v + 1)) == 0)
 		return v + 1;
 	for (i = 0, m = 1; i < 31; i++, m <<= 1) {
 		if (v & m)
 			continue;
 		v |= m;
 		if ((v & (v + 1)) == 0)
 			return v + 1;
 	}
 	ASSERT(0);
 	return( 0 );
 }
 #ifdef XFS_ILOCK_TRACE
 ktrace_t	*xfs_ilock_trace_buf;
 void
 xfs_ilock_trace(xfs_inode_t *ip, int lock, unsigned int lockflags, inst_t *ra)
 {
 	ktrace_enter(ip->i_lock_trace,
 		     (void *)ip,
 		     (void *)(unsigned long)lock, /* 1 = LOCK, 3=UNLOCK, etc */
 		     (void *)(unsigned long)lockflags, /* XFS_ILOCK_EXCL etc */
 		     (void *)ra,		/* caller of ilock */
 		     (void *)(unsigned long)current_cpu(),
 		     (void *)(unsigned long)current_pid(),
 		     NULL,NULL,NULL,NULL,NULL,NULL,NULL,NULL,NULL,NULL);
 }
 #endif
 /*
  * Return a pointer to the extent record at file index idx.
  */
 xfs_bmbt_rec_t *
 xfs_iext_get_ext(
 	xfs_ifork_t	*ifp,		/* inode fork pointer */
 	xfs_extnum_t	idx)		/* index of target extent */
 {
 	ASSERT(idx >= 0);
 	if ((ifp->if_flags & XFS_IFEXTIREC) && (idx == 0)) {
 		return ifp->if_u1.if_ext_irec->er_extbuf;
 	} else if (ifp->if_flags & XFS_IFEXTIREC) {
 		xfs_ext_irec_t	*erp;		/* irec pointer */
 		int		erp_idx = 0;	/* irec index */
 		xfs_extnum_t	page_idx = idx;	/* ext index in target list */
 		erp = xfs_iext_idx_to_irec(ifp, &page_idx, &erp_idx, 0);
 		return &erp->er_extbuf[page_idx];
 	} else if (ifp->if_bytes) {
 		return &ifp->if_u1.if_extents[idx];
 	} else {
 		return NULL;
 	}
 }
 /*
  * Insert new item(s) into the extent records for incore inode
  * fork 'ifp'.  'count' new items are inserted at index 'idx'.
  */
 void
 xfs_iext_insert(
 	xfs_ifork_t	*ifp,		/* inode fork pointer */
 	xfs_extnum_t	idx,		/* starting index of new items */
 	xfs_extnum_t	count,		/* number of inserted items */
 	xfs_bmbt_irec_t	*new)		/* items to insert */
 {
 	xfs_bmbt_rec_t	*ep;		/* extent record pointer */
 	xfs_extnum_t	i;		/* extent record index */
 	ASSERT(ifp->if_flags & XFS_IFEXTENTS);
 	xfs_iext_add(ifp, idx, count);
 	for (i = idx; i < idx + count; i++, new++) {
 		ep = xfs_iext_get_ext(ifp, i);
 		xfs_bmbt_set_all(ep, new);
 	}
 }
 /*
  * This is called when the amount of space required for incore file
  * extents needs to be increased. The ext_diff parameter stores the
  * number of new extents being added and the idx parameter contains
  * the extent index where the new extents will be added. If the new
  * extents are being appended, then we just need to (re)allocate and
  * initialize the space. Otherwise, if the new extents are being
  * inserted into the middle of the existing entries, a bit more work
  * is required to make room for the new extents to be inserted. The
  * caller is responsible for filling in the new extent entries upon
  * return.
  */
 void
 xfs_iext_add(
 	xfs_ifork_t	*ifp,		/* inode fork pointer */
 	xfs_extnum_t	idx,		/* index to begin adding exts */
 	int		ext_diff)	/* number of extents to add */
 {
 	int		byte_diff;	/* new bytes being added */
 	int		new_size;	/* size of extents after adding */
 	xfs_extnum_t	nextents;	/* number of extents in file */
 	nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t);
 	ASSERT((idx >= 0) && (idx <= nextents));
 	byte_diff = ext_diff * sizeof(xfs_bmbt_rec_t);
 	new_size = ifp->if_bytes + byte_diff;
 	/*
 	 * If the new number of extents (nextents + ext_diff)
 	 * fits inside the inode, then continue to use the inline
 	 * extent buffer.
 	 */
 	if (nextents + ext_diff <= XFS_INLINE_EXTS) {
 		if (idx < nextents) {
 			memmove(&ifp->if_u2.if_inline_ext[idx + ext_diff],
 				&ifp->if_u2.if_inline_ext[idx],
 				(nextents - idx) * sizeof(xfs_bmbt_rec_t));
 			memset(&ifp->if_u2.if_inline_ext[idx], 0, byte_diff);
 		}
 		ifp->if_u1.if_extents = ifp->if_u2.if_inline_ext;
 		ifp->if_real_bytes = 0;
 		ifp->if_lastex = nextents + ext_diff;
 	}
 	/*
 	 * Otherwise use a linear (direct) extent list.
 	 * If the extents are currently inside the inode,
 	 * xfs_iext_realloc_direct will switch us from
 	 * inline to direct extent allocation mode.
 	 */
 	else if (nextents + ext_diff <= XFS_LINEAR_EXTS) {
 		xfs_iext_realloc_direct(ifp, new_size);
 		if (idx < nextents) {
 			memmove(&ifp->if_u1.if_extents[idx + ext_diff],
 				&ifp->if_u1.if_extents[idx],
 				(nextents - idx) * sizeof(xfs_bmbt_rec_t));
 			memset(&ifp->if_u1.if_extents[idx], 0, byte_diff);
 		}
 	}
 	/* Indirection array */
 	else {
 		xfs_ext_irec_t	*erp;
 		int		erp_idx = 0;
 		int		page_idx = idx;
 		ASSERT(nextents + ext_diff > XFS_LINEAR_EXTS);
 		if (ifp->if_flags & XFS_IFEXTIREC) {
 			erp = xfs_iext_idx_to_irec(ifp, &page_idx, &erp_idx, 1);
 		} else {
 			xfs_iext_irec_init(ifp);
 			ASSERT(ifp->if_flags & XFS_IFEXTIREC);
 			erp = ifp->if_u1.if_ext_irec;
 		}
 		/* Extents fit in target extent page */
 		if (erp && erp->er_extcount + ext_diff <= XFS_LINEAR_EXTS) {
 			if (page_idx < erp->er_extcount) {
 				memmove(&erp->er_extbuf[page_idx + ext_diff],
 					&erp->er_extbuf[page_idx],
 					(erp->er_extcount - page_idx) *
 					sizeof(xfs_bmbt_rec_t));
 				memset(&erp->er_extbuf[page_idx], 0, byte_diff);
 			}
 			erp->er_extcount += ext_diff;
 			xfs_iext_irec_update_extoffs(ifp, erp_idx + 1, ext_diff);
 		}
 		/* Insert a new extent page */
 		else if (erp) {
 			xfs_iext_add_indirect_multi(ifp,
 				erp_idx, page_idx, ext_diff);
 		}
 		/*
 		 * If extent(s) are being appended to the last page in
 		 * the indirection array and the new extent(s) don't fit
 		 * in the page, then erp is NULL and erp_idx is set to
 		 * the next index needed in the indirection array.
 		 */
 		else {
 			int	count = ext_diff;
 			while (count) {
 				erp = xfs_iext_irec_new(ifp, erp_idx);
 				erp->er_extcount = count;
 				count -= MIN(count, (int)XFS_LINEAR_EXTS);
 				if (count) {
 					erp_idx++;
 				}
 			}
 		}
 	}
 	ifp->if_bytes = new_size;
 }
 /*
  * This is called when incore extents are being added to the indirection
  * array and the new extents do not fit in the target extent list. The
  * erp_idx parameter contains the irec index for the target extent list
  * in the indirection array, and the idx parameter contains the extent
  * index within the list. The number of extents being added is stored
  * in the count parameter.
  *
  *    |-------|   |-------|
  *    |       |   |       |    idx - number of extents before idx
  *    |  idx  |   | count |
  *    |       |   |       |    count - number of extents being inserted at idx
  *    |-------|   |-------|
  *    | count |   | nex2  |    nex2 - number of extents after idx + count
  *    |-------|   |-------|
  */
 void
 xfs_iext_add_indirect_multi(
 	xfs_ifork_t	*ifp,			/* inode fork pointer */
 	int		erp_idx,		/* target extent irec index */
 	xfs_extnum_t	idx,			/* index within target list */
 	int		count)			/* new extents being added */
 {
 	int		byte_diff;		/* new bytes being added */
 	xfs_ext_irec_t	*erp;			/* pointer to irec entry */
 	xfs_extnum_t	ext_diff;		/* number of extents to add */
 	xfs_extnum_t	ext_cnt;		/* new extents still needed */
 	xfs_extnum_t	nex2;			/* extents after idx + count */
 	xfs_bmbt_rec_t	*nex2_ep = NULL;	/* temp list for nex2 extents */
 	int		nlists;			/* number of irec's (lists) */
 	ASSERT(ifp->if_flags & XFS_IFEXTIREC);
 	erp = &ifp->if_u1.if_ext_irec[erp_idx];
 	nex2 = erp->er_extcount - idx;
 	nlists = ifp->if_real_bytes / XFS_IEXT_BUFSZ;
 	/*
 	 * Save second part of target extent list
 	 * (all extents past */
 	if (nex2) {
 		byte_diff = nex2 * sizeof(xfs_bmbt_rec_t);
 		nex2_ep = (xfs_bmbt_rec_t *) kmem_alloc(byte_diff, KM_SLEEP);
 		memmove(nex2_ep, &erp->er_extbuf[idx], byte_diff);
 		erp->er_extcount -= nex2;
 		xfs_iext_irec_update_extoffs(ifp, erp_idx + 1, -nex2);
 		memset(&erp->er_extbuf[idx], 0, byte_diff);
 	}
 	/*
 	 * Add the new extents to the end of the target
 	 * list, then allocate new irec record(s) and
 	 * extent buffer(s) as needed to store the rest
 	 * of the new extents.
 	 */
 	ext_cnt = count;
 	ext_diff = MIN(ext_cnt, (int)XFS_LINEAR_EXTS - erp->er_extcount);
 	if (ext_diff) {
 		erp->er_extcount += ext_diff;
 		xfs_iext_irec_update_extoffs(ifp, erp_idx + 1, ext_diff);
 		ext_cnt -= ext_diff;
 	}
 	while (ext_cnt) {
 		erp_idx++;
 		erp = xfs_iext_irec_new(ifp, erp_idx);
 		ext_diff = MIN(ext_cnt, (int)XFS_LINEAR_EXTS);
 		erp->er_extcount = ext_diff;
 		xfs_iext_irec_update_extoffs(ifp, erp_idx + 1, ext_diff);
 		ext_cnt -= ext_diff;
 	}
 	/* Add nex2 extents back to indirection array */
 	if (nex2) {
 		xfs_extnum_t	ext_avail;
 		int		i;
 		byte_diff = nex2 * sizeof(xfs_bmbt_rec_t);
 		ext_avail = XFS_LINEAR_EXTS - erp->er_extcount;
 		i = 0;
 		/*
 		 * If nex2 extents fit in the current page, append
 		 * nex2_ep after the new extents.
 		 */
 		if (nex2 <= ext_avail) {
 			i = erp->er_extcount;
 		}
 		/*
 		 * Otherwise, check if space is available in the
 		 * next page.
 		 */
 		else if ((erp_idx < nlists - 1) &&
 			 (nex2 <= (ext_avail = XFS_LINEAR_EXTS -
 			  ifp->if_u1.if_ext_irec[erp_idx+1].er_extcount))) {
 			erp_idx++;
 			erp++;
 			/* Create a hole for nex2 extents */
 			memmove(&erp->er_extbuf[nex2], erp->er_extbuf,
 				erp->er_extcount * sizeof(xfs_bmbt_rec_t));
 		}
 		/*
 		 * Final choice, create a new extent page for
 		 * nex2 extents.
 		 */
 		else {
 			erp_idx++;
 			erp = xfs_iext_irec_new(ifp, erp_idx);
 		}
 		memmove(&erp->er_extbuf[i], nex2_ep, byte_diff);
 		kmem_free(nex2_ep, byte_diff);
 		erp->er_extcount += nex2;
 		xfs_iext_irec_update_extoffs(ifp, erp_idx + 1, nex2);
 	}
 }
 /*
  * This is called when the amount of space required for incore file
  * extents needs to be decreased. The ext_diff parameter stores the
  * number of extents to be removed and the idx parameter contains
  * the extent index where the extents will be removed from.
  *
  * If the amount of space needed has decreased below the linear
  * limit, XFS_IEXT_BUFSZ, then switch to using the contiguous
  * extent array.  Otherwise, use kmem_realloc() to adjust the
  * size to what is needed.
  */
 void
 xfs_iext_remove(
 	xfs_ifork_t	*ifp,		/* inode fork pointer */
 	xfs_extnum_t	idx,		/* index to begin removing exts */
 	int		ext_diff)	/* number of extents to remove */
 {
 	xfs_extnum_t	nextents;	/* number of extents in file */
 	int		new_size;	/* size of extents after removal */
 	ASSERT(ext_diff > 0);
 	nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t);
 	new_size = (nextents - ext_diff) * sizeof(xfs_bmbt_rec_t);
 	if (new_size == 0) {
 		xfs_iext_destroy(ifp);
 	} else if (ifp->if_flags & XFS_IFEXTIREC) {
 		xfs_iext_remove_indirect(ifp, idx, ext_diff);
 	} else if (ifp->if_real_bytes) {
 		xfs_iext_remove_direct(ifp, idx, ext_diff);
 	} else {
 		xfs_iext_remove_inline(ifp, idx, ext_diff);
 	}
 	ifp->if_bytes = new_size;
 }
 /*
  * This removes ext_diff extents from the inline buffer, beginning
  * at extent index idx.
  */
 void
 xfs_iext_remove_inline(
 	xfs_ifork_t	*ifp,		/* inode fork pointer */
 	xfs_extnum_t	idx,		/* index to begin removing exts */
 	int		ext_diff)	/* number of extents to remove */
 {
 	int		nextents;	/* number of extents in file */
 	ASSERT(!(ifp->if_flags & XFS_IFEXTIREC));
 	ASSERT(idx < XFS_INLINE_EXTS);
 	nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t);
 	ASSERT(((nextents - ext_diff) > 0) &&
 		(nextents - ext_diff) < XFS_INLINE_EXTS);
 	if (idx + ext_diff < nextents) {
 		memmove(&ifp->if_u2.if_inline_ext[idx],
 			&ifp->if_u2.if_inline_ext[idx + ext_diff],
 			(nextents - (idx + ext_diff)) *
 			 sizeof(xfs_bmbt_rec_t));
 		memset(&ifp->if_u2.if_inline_ext[nextents - ext_diff],
 			0, ext_diff * sizeof(xfs_bmbt_rec_t));
 	} else {
 		memset(&ifp->if_u2.if_inline_ext[idx], 0,
 			ext_diff * sizeof(xfs_bmbt_rec_t));
 	}
 }
 /*
  * This removes ext_diff extents from a linear (direct) extent list,
  * beginning at extent index idx. If the extents are being removed
  * from the end of the list (ie. truncate) then we just need to re-
  * allocate the list to remove the extra space. Otherwise, if the
  * extents are being removed from the middle of the existing extent
  * entries, then we first need to move the extent records beginning
  * at idx + ext_diff up in the list to overwrite the records being
  * removed, then remove the extra space via kmem_realloc.
  */
 void
 xfs_iext_remove_direct(
 	xfs_ifork_t	*ifp,		/* inode fork pointer */
 	xfs_extnum_t	idx,		/* index to begin removing exts */
 	int		ext_diff)	/* number of extents to remove */
 {
 	xfs_extnum_t	nextents;	/* number of extents in file */
 	int		new_size;	/* size of extents after removal */
 	ASSERT(!(ifp->if_flags & XFS_IFEXTIREC));
 	new_size = ifp->if_bytes -
 		(ext_diff * sizeof(xfs_bmbt_rec_t));
 	nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t);
 	if (new_size == 0) {
 		xfs_iext_destroy(ifp);
 		return;
 	}
 	/* Move extents up in the list (if needed) */
 	if (idx + ext_diff < nextents) {
 		memmove(&ifp->if_u1.if_extents[idx],
 			&ifp->if_u1.if_extents[idx + ext_diff],
 			(nextents - (idx + ext_diff)) *
 			 sizeof(xfs_bmbt_rec_t));
 	}
 	memset(&ifp->if_u1.if_extents[nextents - ext_diff],
 		0, ext_diff * sizeof(xfs_bmbt_rec_t));
 	/*
 	 * Reallocate the direct extent list. If the extents
 	 * will fit inside the inode then xfs_iext_realloc_direct
 	 * will switch from direct to inline extent allocation
 	 * mode for us.
 	 */
 	xfs_iext_realloc_direct(ifp, new_size);
 	ifp->if_bytes = new_size;
 }
 /*
  * This is called when incore extents are being removed from the
  * indirection array and the extents being removed span multiple extent
  * buffers. The idx parameter contains the file extent index where we
  * want to begin removing extents, and the count parameter contains
  * how many extents need to be removed.
  *
  *    |-------|   |-------|
  *    | nex1  |   |       |    nex1 - number of extents before idx
  *    |-------|   | count |
  *    |       |   |       |    count - number of extents being removed at idx
  *    | count |   |-------|
  *    |       |   | nex2  |    nex2 - number of extents after idx + count
  *    |-------|   |-------|
  */
 void
 xfs_iext_remove_indirect(
 	xfs_ifork_t	*ifp,		/* inode fork pointer */
 	xfs_extnum_t	idx,		/* index to begin removing extents */
 	int		count)		/* number of extents to remove */
 {
 	xfs_ext_irec_t	*erp;		/* indirection array pointer */
 	int		erp_idx = 0;	/* indirection array index */
 	xfs_extnum_t	ext_cnt;	/* extents left to remove */
 	xfs_extnum_t	ext_diff;	/* extents to remove in current list */
 	xfs_extnum_t	nex1;		/* number of extents before idx */
 	xfs_extnum_t	nex2;		/* extents after idx + count */
 	int		nlists;		/* entries in indirection array */
 	int		page_idx = idx;	/* index in target extent list */
 	ASSERT(ifp->if_flags & XFS_IFEXTIREC);
 	erp = xfs_iext_idx_to_irec(ifp,  &page_idx, &erp_idx, 0);
 	ASSERT(erp != NULL);
 	nlists = ifp->if_real_bytes / XFS_IEXT_BUFSZ;
 	nex1 = page_idx;
 	ext_cnt = count;
 	while (ext_cnt) {
 		nex2 = MAX((erp->er_extcount - (nex1 + ext_cnt)), 0);
 		ext_diff = MIN(ext_cnt, (erp->er_extcount - nex1));
 		/*
 		 * Check for deletion of entire list;
 		 * xfs_iext_irec_remove() updates extent offsets.
 		 */
 		if (ext_diff == erp->er_extcount) {
 			xfs_iext_irec_remove(ifp, erp_idx);
 			ext_cnt -= ext_diff;
 			nex1 = 0;
 			if (ext_cnt) {
 				ASSERT(erp_idx < ifp->if_real_bytes /
 					XFS_IEXT_BUFSZ);
 				erp = &ifp->if_u1.if_ext_irec[erp_idx];
 				nex1 = 0;
 				continue;
 			} else {
 				break;
 			}
 		}
 		/* Move extents up (if needed) */
 		if (nex2) {
 			memmove(&erp->er_extbuf[nex1],
 				&erp->er_extbuf[nex1 + ext_diff],
 				nex2 * sizeof(xfs_bmbt_rec_t));
 		}
 		/* Zero out rest of page */
 		memset(&erp->er_extbuf[nex1 + nex2], 0, (XFS_IEXT_BUFSZ -
 			((nex1 + nex2) * sizeof(xfs_bmbt_rec_t))));
 		/* Update remaining counters */
 		erp->er_extcount -= ext_diff;
 		xfs_iext_irec_update_extoffs(ifp, erp_idx + 1, -ext_diff);
 		ext_cnt -= ext_diff;
 		nex1 = 0;
 		erp_idx++;
 		erp++;
 	}
 	ifp->if_bytes -= count * sizeof(xfs_bmbt_rec_t);
 	xfs_iext_irec_compact(ifp);
 }
 /*
  * Create, destroy, or resize a linear (direct) block of extents.
  */
 void
 xfs_iext_realloc_direct(
 	xfs_ifork_t	*ifp,		/* inode fork pointer */
 	int		new_size)	/* new size of extents */
 {
 	int		rnew_size;	/* real new size of extents */
 	rnew_size = new_size;
 	ASSERT(!(ifp->if_flags & XFS_IFEXTIREC) ||
 		((new_size >= 0) && (new_size <= XFS_IEXT_BUFSZ) &&
 		 (new_size != ifp->if_real_bytes)));
 	/* Free extent records */
 	if (new_size == 0) {
 		xfs_iext_destroy(ifp);
 	}
 	/* Resize direct extent list and zero any new bytes */
 	else if (ifp->if_real_bytes) {
 		/* Check if extents will fit inside the inode */
 		if (new_size <= XFS_INLINE_EXTS * sizeof(xfs_bmbt_rec_t)) {
 			xfs_iext_direct_to_inline(ifp, new_size /
 				(uint)sizeof(xfs_bmbt_rec_t));
 			ifp->if_bytes = new_size;
 			return;
 		}
 		if ((new_size & (new_size - 1)) != 0) {
 			rnew_size = xfs_iroundup(new_size);
 		}
 		if (rnew_size != ifp->if_real_bytes) {
 			ifp->if_u1.if_extents = (xfs_bmbt_rec_t *)
 				kmem_realloc(ifp->if_u1.if_extents,
 						rnew_size,
 						ifp->if_real_bytes,
 						KM_SLEEP);
 		}
 		if (rnew_size > ifp->if_real_bytes) {
 			memset(&ifp->if_u1.if_extents[ifp->if_bytes /
 				(uint)sizeof(xfs_bmbt_rec_t)], 0,
 				rnew_size - ifp->if_real_bytes);
 		}
 	}
 	/*
 	 * Switch from the inline extent buffer to a direct
 	 * extent list. Be sure to include the inline extent
 	 * bytes in new_size.
 	 */
 	else {
 		new_size += ifp->if_bytes;
 		if ((new_size & (new_size - 1)) != 0) {
 			rnew_size = xfs_iroundup(new_size);
 		}
 		xfs_iext_inline_to_direct(ifp, rnew_size);
 	}
 	ifp->if_real_bytes = rnew_size;
 	ifp->if_bytes = new_size;
 }
 /*
  * Switch from linear (direct) extent records to inline buffer.
  */
 void
 xfs_iext_direct_to_inline(
 	xfs_ifork_t	*ifp,		/* inode fork pointer */
 	xfs_extnum_t	nextents)	/* number of extents in file */
 {
 	ASSERT(ifp->if_flags & XFS_IFEXTENTS);
 	ASSERT(nextents <= XFS_INLINE_EXTS);
 	/*
 	 * The inline buffer was zeroed when we switched
 	 * from inline to direct extent allocation mode,
 	 * so we don't need to clear it here.
 	 */
 	memcpy(ifp->if_u2.if_inline_ext, ifp->if_u1.if_extents,
 		nextents * sizeof(xfs_bmbt_rec_t));
 	kmem_free(ifp->if_u1.if_extents, ifp->if_real_bytes);
 	ifp->if_u1.if_extents = ifp->if_u2.if_inline_ext;
 	ifp->if_real_bytes = 0;
 }
 /*
  * Switch from inline buffer to linear (direct) extent records.
  * new_size should already be rounded up to the next power of 2
  * by the caller (when appropriate), so use new_size as it is.
  * However, since new_size may be rounded up, we can't update
  * if_bytes here. It is the caller's responsibility to update
  * if_bytes upon return.
  */
 void
 xfs_iext_inline_to_direct(
 	xfs_ifork_t	*ifp,		/* inode fork pointer */
 	int		new_size)	/* number of extents in file */
 {
 	ifp->if_u1.if_extents = (xfs_bmbt_rec_t *)
 		kmem_alloc(new_size, KM_SLEEP);
 	memset(ifp->if_u1.if_extents, 0, new_size);
 	if (ifp->if_bytes) {
 		memcpy(ifp->if_u1.if_extents, ifp->if_u2.if_inline_ext,
 			ifp->if_bytes);
 		memset(ifp->if_u2.if_inline_ext, 0, XFS_INLINE_EXTS *
 			sizeof(xfs_bmbt_rec_t));
 	}
 	ifp->if_real_bytes = new_size;
 }
 /*
  * Resize an extent indirection array to new_size bytes.
  */
 void
 xfs_iext_realloc_indirect(
 	xfs_ifork_t	*ifp,		/* inode fork pointer */
 	int		new_size)	/* new indirection array size */
 {
 	int		nlists;		/* number of irec's (ex lists) */
 	int		size;		/* current indirection array size */
 	ASSERT(ifp->if_flags & XFS_IFEXTIREC);
 	nlists = ifp->if_real_bytes / XFS_IEXT_BUFSZ;
 	size = nlists * sizeof(xfs_ext_irec_t);
 	ASSERT(ifp->if_real_bytes);
 	ASSERT((new_size >= 0) && (new_size != size));
 	if (new_size == 0) {
 		xfs_iext_destroy(ifp);
 	} else {
 		ifp->if_u1.if_ext_irec = (xfs_ext_irec_t *)
 			kmem_realloc(ifp->if_u1.if_ext_irec,
 				new_size, size, KM_SLEEP);
 	}
 }
 /*
  * Switch from indirection array to linear (direct) extent allocations.
  */
 void
 xfs_iext_indirect_to_direct(
 	 xfs_ifork_t	*ifp)		/* inode fork pointer */
 {
 	xfs_bmbt_rec_t	*ep;		/* extent record pointer */
 	xfs_extnum_t	nextents;	/* number of extents in file */
 	int		size;		/* size of file extents */
 	ASSERT(ifp->if_flags & XFS_IFEXTIREC);
 	nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t);
 	ASSERT(nextents <= XFS_LINEAR_EXTS);
 	size = nextents * sizeof(xfs_bmbt_rec_t);
 	xfs_iext_irec_compact_full(ifp);
 	ASSERT(ifp->if_real_bytes == XFS_IEXT_BUFSZ);
 	ep = ifp->if_u1.if_ext_irec->er_extbuf;
 	kmem_free(ifp->if_u1.if_ext_irec, sizeof(xfs_ext_irec_t));
 	ifp->if_flags &= ~XFS_IFEXTIREC;
 	ifp->if_u1.if_extents = ep;
 	ifp->if_bytes = size;
 	if (nextents < XFS_LINEAR_EXTS) {
 		xfs_iext_realloc_direct(ifp, size);
 	}
 }
 /*
  * Free incore file extents.
  */
 void
 xfs_iext_destroy(
 	xfs_ifork_t	*ifp)		/* inode fork pointer */
 {
 	if (ifp->if_flags & XFS_IFEXTIREC) {
 		int	erp_idx;
 		int	nlists;
 		nlists = ifp->if_real_bytes / XFS_IEXT_BUFSZ;
 		for (erp_idx = nlists - 1; erp_idx >= 0 ; erp_idx--) {
 			xfs_iext_irec_remove(ifp, erp_idx);
 		}
 		ifp->if_flags &= ~XFS_IFEXTIREC;
 	} else if (ifp->if_real_bytes) {
 		kmem_free(ifp->if_u1.if_extents, ifp->if_real_bytes);
 	} else if (ifp->if_bytes) {
 		memset(ifp->if_u2.if_inline_ext, 0, XFS_INLINE_EXTS *
 			sizeof(xfs_bmbt_rec_t));
 	}
 	ifp->if_u1.if_extents = NULL;
 	ifp->if_real_bytes = 0;
 	ifp->if_bytes = 0;
 }
 /*
  * Return a pointer to the extent record for file system block bno.
  */
 xfs_bmbt_rec_t *			/* pointer to found extent record */
 xfs_iext_bno_to_ext(
 	xfs_ifork_t	*ifp,		/* inode fork pointer */
 	xfs_fileoff_t	bno,		/* block number to search for */
 	xfs_extnum_t	*idxp)		/* index of target extent */
 {
 	xfs_bmbt_rec_t	*base;		/* pointer to first extent */
 	xfs_filblks_t	blockcount = 0;	/* number of blocks in extent */
 	xfs_bmbt_rec_t	*ep = NULL;	/* pointer to target extent */
 	xfs_ext_irec_t	*erp = NULL;	/* indirection array pointer */
 	int		high;		/* upper boundary in search */
 	xfs_extnum_t	idx = 0;	/* index of target extent */
 	int		low;		/* lower boundary in search */
 	xfs_extnum_t	nextents;	/* number of file extents */
 	xfs_fileoff_t	startoff = 0;	/* start offset of extent */
 	nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t);
 	if (nextents == 0) {
 		*idxp = 0;
 		return NULL;
 	}
 	low = 0;
 	if (ifp->if_flags & XFS_IFEXTIREC) {
 		/* Find target extent list */
 		int	erp_idx = 0;
 		erp = xfs_iext_bno_to_irec(ifp, bno, &erp_idx);
 		base = erp->er_extbuf;
 		high = erp->er_extcount - 1;
 	} else {
 		base = ifp->if_u1.if_extents;
 		high = nextents - 1;
 	}
 	/* Binary search extent records */
 	while (low <= high) {
 		idx = (low + high) >> 1;
 		ep = base + idx;
 		startoff = xfs_bmbt_get_startoff(ep);
 		blockcount = xfs_bmbt_get_blockcount(ep);
 		if (bno < startoff) {
 			high = idx - 1;
 		} else if (bno >= startoff + blockcount) {
 			low = idx + 1;
 		} else {
 			/* Convert back to file-based extent index */
 			if (ifp->if_flags & XFS_IFEXTIREC) {
 				idx += erp->er_extoff;
 			}
 			*idxp = idx;
 			return ep;
 		}
 	}
 	/* Convert back to file-based extent index */
 	if (ifp->if_flags & XFS_IFEXTIREC) {
 		idx += erp->er_extoff;
 	}
 	if (bno >= startoff + blockcount) {
 		if (++idx == nextents) {
 			ep = NULL;
 		} else {
 			ep = xfs_iext_get_ext(ifp, idx);
 		}
 	}
 	*idxp = idx;
 	return ep;
 }
 /*
  * Return a pointer to the indirection array entry containing the
  * extent record for filesystem block bno. Store the index of the
  * target irec in *erp_idxp.
  */
 xfs_ext_irec_t *			/* pointer to found extent record */
 xfs_iext_bno_to_irec(
 	xfs_ifork_t	*ifp,		/* inode fork pointer */
 	xfs_fileoff_t	bno,		/* block number to search for */
 	int		*erp_idxp)	/* irec index of target ext list */
 {
 	xfs_ext_irec_t	*erp = NULL;	/* indirection array pointer */
 	xfs_ext_irec_t	*erp_next;	/* next indirection array entry */
 	int		erp_idx;	/* indirection array index */
 	int		nlists;		/* number of extent irec's (lists) */
 	int		high;		/* binary search upper limit */
 	int		low;		/* binary search lower limit */
 	ASSERT(ifp->if_flags & XFS_IFEXTIREC);
 	nlists = ifp->if_real_bytes / XFS_IEXT_BUFSZ;
 	erp_idx = 0;
 	low = 0;
 	high = nlists - 1;
 	while (low <= high) {
 		erp_idx = (low + high) >> 1;
 		erp = &ifp->if_u1.if_ext_irec[erp_idx];
 		erp_next = erp_idx < nlists - 1 ? erp + 1 : NULL;
 		if (bno < xfs_bmbt_get_startoff(erp->er_extbuf)) {
 			high = erp_idx - 1;
 		} else if (erp_next && bno >=
 			   xfs_bmbt_get_startoff(erp_next->er_extbuf)) {
 			low = erp_idx + 1;
 		} else {
 			break;
 		}
 	}
 	*erp_idxp = erp_idx;
 	return erp;
 }
 /*
  * Return a pointer to the indirection array entry containing the
  * extent record at file extent index *idxp. Store the index of the
  * target irec in *erp_idxp and store the page index of the target
  * extent record in *idxp.
  */
 xfs_ext_irec_t *
 xfs_iext_idx_to_irec(
 	xfs_ifork_t	*ifp,		/* inode fork pointer */
 	xfs_extnum_t	*idxp,		/* extent index (file -> page) */
 	int		*erp_idxp,	/* pointer to target irec */
 	int		realloc)	/* new bytes were just added */
 {
 	xfs_ext_irec_t	*prev;		/* pointer to previous irec */
 	xfs_ext_irec_t	*erp = NULL;	/* pointer to current irec */
 	int		erp_idx;	/* indirection array index */
 	int		nlists;		/* number of irec's (ex lists) */
 	int		high;		/* binary search upper limit */
 	int		low;		/* binary search lower limit */
 	xfs_extnum_t	page_idx = *idxp; /* extent index in target list */
 	ASSERT(ifp->if_flags & XFS_IFEXTIREC);
 	ASSERT(page_idx >= 0 && page_idx <=
 		ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t));
 	nlists = ifp->if_real_bytes / XFS_IEXT_BUFSZ;
 	erp_idx = 0;
 	low = 0;
 	high = nlists - 1;
 	/* Binary search extent irec's */
 	while (low <= high) {
 		erp_idx = (low + high) >> 1;
 		erp = &ifp->if_u1.if_ext_irec[erp_idx];
 		prev = erp_idx > 0 ? erp - 1 : NULL;
 		if (page_idx < erp->er_extoff || (page_idx == erp->er_extoff &&
 		     realloc && prev && prev->er_extcount < XFS_LINEAR_EXTS)) {
 			high = erp_idx - 1;
 		} else if (page_idx > erp->er_extoff + erp->er_extcount ||
 			   (page_idx == erp->er_extoff + erp->er_extcount &&
 			    !realloc)) {
 			low = erp_idx + 1;
 		} else if (page_idx == erp->er_extoff + erp->er_extcount &&
 			   erp->er_extcount == XFS_LINEAR_EXTS) {
 			ASSERT(realloc);
 			page_idx = 0;
 			erp_idx++;
 			erp = erp_idx < nlists ? erp + 1 : NULL;
 			break;
 		} else {
 			page_idx -= erp->er_extoff;
 			break;
 		}
 	}
 	*idxp = page_idx;
 	*erp_idxp = erp_idx;
 	return(erp);
 }
 /*
  * Allocate and initialize an indirection array once the space needed
  * for incore extents increases above XFS_IEXT_BUFSZ.
  */
 void
 xfs_iext_irec_init(
 	xfs_ifork_t	*ifp)		/* inode fork pointer */
 {
 	xfs_ext_irec_t	*erp;		/* indirection array pointer */
 	xfs_extnum_t	nextents;	/* number of extents in file */
 	ASSERT(!(ifp->if_flags & XFS_IFEXTIREC));
 	nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t);
 	ASSERT(nextents <= XFS_LINEAR_EXTS);
 	erp = (xfs_ext_irec_t *)
 		kmem_alloc(sizeof(xfs_ext_irec_t), KM_SLEEP);
 	if (nextents == 0) {
 		ifp->if_u1.if_extents = (xfs_bmbt_rec_t *)
 			kmem_alloc(XFS_IEXT_BUFSZ, KM_SLEEP);
 	} else if (!ifp->if_real_bytes) {
 		xfs_iext_inline_to_direct(ifp, XFS_IEXT_BUFSZ);
 	} else if (ifp->if_real_bytes < XFS_IEXT_BUFSZ) {
 		xfs_iext_realloc_direct(ifp, XFS_IEXT_BUFSZ);
 	}
 	erp->er_extbuf = ifp->if_u1.if_extents;
 	erp->er_extcount = nextents;
 	erp->er_extoff = 0;
 	ifp->if_flags |= XFS_IFEXTIREC;
 	ifp->if_real_bytes = XFS_IEXT_BUFSZ;
 	ifp->if_bytes = nextents * sizeof(xfs_bmbt_rec_t);
 	ifp->if_u1.if_ext_irec = erp;
 	return;
 }
 /*
  * Allocate and initialize a new entry in the indirection array.
  */
 xfs_ext_irec_t *
 xfs_iext_irec_new(
 	xfs_ifork_t	*ifp,		/* inode fork pointer */
 	int		erp_idx)	/* index for new irec */
 {
 	xfs_ext_irec_t	*erp;		/* indirection array pointer */
 	int		i;		/* loop counter */
 	int		nlists;		/* number of irec's (ex lists) */
 	ASSERT(ifp->if_flags & XFS_IFEXTIREC);
 	nlists = ifp->if_real_bytes / XFS_IEXT_BUFSZ;
 	/* Resize indirection array */
 	xfs_iext_realloc_indirect(ifp, ++nlists *
 				  sizeof(xfs_ext_irec_t));
 	/*
 	 * Move records down in the array so the
 	 * new page can use erp_idx.
 	 */
 	erp = ifp->if_u1.if_ext_irec;
 	for (i = nlists - 1; i > erp_idx; i--) {
 		memmove(&erp[i], &erp[i-1], sizeof(xfs_ext_irec_t));
 	}
 	ASSERT(i == erp_idx);
 	/* Initialize new extent record */
 	erp = ifp->if_u1.if_ext_irec;
 	erp[erp_idx].er_extbuf = (xfs_bmbt_rec_t *)
 		kmem_alloc(XFS_IEXT_BUFSZ, KM_SLEEP);
 	ifp->if_real_bytes = nlists * XFS_IEXT_BUFSZ;
 	memset(erp[erp_idx].er_extbuf, 0, XFS_IEXT_BUFSZ);
 	erp[erp_idx].er_extcount = 0;
 	erp[erp_idx].er_extoff = erp_idx > 0 ?
 		erp[erp_idx-1].er_extoff + erp[erp_idx-1].er_extcount : 0;
 	return (&erp[erp_idx]);
 }
 /*
  * Remove a record from the indirection array.
  */
 void
 xfs_iext_irec_remove(
 	xfs_ifork_t	*ifp,		/* inode fork pointer */
 	int		erp_idx)	/* irec index to remove */
 {
 	xfs_ext_irec_t	*erp;		/* indirection array pointer */
 	int		i;		/* loop counter */
 	int		nlists;		/* number of irec's (ex lists) */
 	ASSERT(ifp->if_flags & XFS_IFEXTIREC);
 	nlists = ifp->if_real_bytes / XFS_IEXT_BUFSZ;
 	erp = &ifp->if_u1.if_ext_irec[erp_idx];
 	if (erp->er_extbuf) {
 		xfs_iext_irec_update_extoffs(ifp, erp_idx + 1,
 			-erp->er_extcount);
 		kmem_free(erp->er_extbuf, XFS_IEXT_BUFSZ);
 	}
 	/* Compact extent records */
 	erp = ifp->if_u1.if_ext_irec;
 	for (i = erp_idx; i < nlists - 1; i++) {
 		memmove(&erp[i], &erp[i+1], sizeof(xfs_ext_irec_t));
 	}
 	/*
 	 * Manually free the last extent record from the indirection
 	 * array.  A call to xfs_iext_realloc_indirect() with a size
 	 * of zero would result in a call to xfs_iext_destroy() which
 	 * would in turn call this function again, creating a nasty
 	 * infinite loop.
 	 */
 	if (--nlists) {
 		xfs_iext_realloc_indirect(ifp,
 			nlists * sizeof(xfs_ext_irec_t));
 	} else {
 		kmem_free(ifp->if_u1.if_ext_irec,
 			sizeof(xfs_ext_irec_t));
 	}
 	ifp->if_real_bytes = nlists * XFS_IEXT_BUFSZ;
 }
 /*
  * This is called to clean up large amounts of unused memory allocated
  * by the indirection array.  Before compacting anything though, verify
  * that the indirection array is still needed and switch back to the
  * linear extent list (or even the inline buffer) if possible.  The
  * compaction policy is as follows:
  *
  *    Full Compaction: Extents fit into a single page (or inline buffer)
  *    Full Compaction: Extents occupy less than 10% of allocated space
  * Partial Compaction: Extents occupy > 10% and < 50% of allocated space
  *      No Compaction: Extents occupy at least 50% of allocated space
  */
 void
 xfs_iext_irec_compact(
 	xfs_ifork_t	*ifp)		/* inode fork pointer */
 {
 	xfs_extnum_t	nextents;	/* number of extents in file */
 	int		nlists;		/* number of irec's (ex lists) */
 	ASSERT(ifp->if_flags & XFS_IFEXTIREC);
 	nlists = ifp->if_real_bytes / XFS_IEXT_BUFSZ;
 	nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t);
 	if (nextents == 0) {
 		xfs_iext_destroy(ifp);
 	} else if (nextents <= XFS_INLINE_EXTS) {
 		xfs_iext_indirect_to_direct(ifp);
 		xfs_iext_direct_to_inline(ifp, nextents);
 	} else if (nextents <= XFS_LINEAR_EXTS) {
 		xfs_iext_indirect_to_direct(ifp);
 	} else if (nextents < (nlists * XFS_LINEAR_EXTS) >> 3) {
 		xfs_iext_irec_compact_full(ifp);
 	} else if (nextents < (nlists * XFS_LINEAR_EXTS) >> 1) {
 		xfs_iext_irec_compact_pages(ifp);
 	}
 }
 /*
  * Combine extents from neighboring extent pages.
  */
 void
 xfs_iext_irec_compact_pages(
 	xfs_ifork_t	*ifp)		/* inode fork pointer */
 {
 	xfs_ext_irec_t	*erp, *erp_next;/* pointers to irec entries */
 	int		erp_idx = 0;	/* indirection array index */
 	int		nlists;		/* number of irec's (ex lists) */
 	ASSERT(ifp->if_flags & XFS_IFEXTIREC);
 	nlists = ifp->if_real_bytes / XFS_IEXT_BUFSZ;
 	while (erp_idx < nlists - 1) {
 		erp = &ifp->if_u1.if_ext_irec[erp_idx];
 		erp_next = erp + 1;
 		if (erp_next->er_extcount <=
 		    (XFS_LINEAR_EXTS - erp->er_extcount)) {
 			memmove(&erp->er_extbuf[erp->er_extcount],
 				erp_next->er_extbuf, erp_next->er_extcount *
 				sizeof(xfs_bmbt_rec_t));
 			erp->er_extcount += erp_next->er_extcount;
 			/*
 			 * Free page before removing extent record
 			 * so er_extoffs don't get modified in
 			 * xfs_iext_irec_remove.
 			 */
 			kmem_free(erp_next->er_extbuf, XFS_IEXT_BUFSZ);
 			erp_next->er_extbuf = NULL;
 			xfs_iext_irec_remove(ifp, erp_idx + 1);
 			nlists = ifp->if_real_bytes / XFS_IEXT_BUFSZ;
 		} else {
 			erp_idx++;
 		}
 	}
 }
 /*
  * Fully compact the extent records managed by the indirection array.
  */
 void
 xfs_iext_irec_compact_full(
 	xfs_ifork_t	*ifp)			/* inode fork pointer */
 {
 	xfs_bmbt_rec_t	*ep, *ep_next;		/* extent record pointers */
 	xfs_ext_irec_t	*erp, *erp_next;	/* extent irec pointers */
 	int		erp_idx = 0;		/* extent irec index */
 	int		ext_avail;		/* empty entries in ex list */
 	int		ext_diff;		/* number of exts to add */
 	int		nlists;			/* number of irec's (ex lists) */
 	ASSERT(ifp->if_flags & XFS_IFEXTIREC);
 	nlists = ifp->if_real_bytes / XFS_IEXT_BUFSZ;
 	erp = ifp->if_u1.if_ext_irec;
 	ep = &erp->er_extbuf[erp->er_extcount];
 	erp_next = erp + 1;
 	ep_next = erp_next->er_extbuf;
 	while (erp_idx < nlists - 1) {
 		ext_avail = XFS_LINEAR_EXTS - erp->er_extcount;
 		ext_diff = MIN(ext_avail, erp_next->er_extcount);
 		memcpy(ep, ep_next, ext_diff * sizeof(xfs_bmbt_rec_t));
 		erp->er_extcount += ext_diff;
 		erp_next->er_extcount -= ext_diff;
 		/* Remove next page */
 		if (erp_next->er_extcount == 0) {
 			/*
 			 * Free page before removing extent record
 			 * so er_extoffs don't get modified in
 			 * xfs_iext_irec_remove.
 			 */
 			kmem_free(erp_next->er_extbuf,
 				erp_next->er_extcount * sizeof(xfs_bmbt_rec_t));
 			erp_next->er_extbuf = NULL;
 			xfs_iext_irec_remove(ifp, erp_idx + 1);
 			erp = &ifp->if_u1.if_ext_irec[erp_idx];
 			nlists = ifp->if_real_bytes / XFS_IEXT_BUFSZ;
 		/* Update next page */
 		} else {
 			/* Move rest of page up to become next new page */
 			memmove(erp_next->er_extbuf, ep_next,
 				erp_next->er_extcount * sizeof(xfs_bmbt_rec_t));
 			ep_next = erp_next->er_extbuf;
 			memset(&ep_next[erp_next->er_extcount], 0,
 				(XFS_LINEAR_EXTS - erp_next->er_extcount) *
 				sizeof(xfs_bmbt_rec_t));
 		}
 		if (erp->er_extcount == XFS_LINEAR_EXTS) {
 			erp_idx++;
 			if (erp_idx < nlists)
 				erp = &ifp->if_u1.if_ext_irec[erp_idx];
 			else
 				break;
 		}
 		ep = &erp->er_extbuf[erp->er_extcount];
 		erp_next = erp + 1;
 		ep_next = erp_next->er_extbuf;
 	}
 }
 /*
  * This is called to update the er_extoff field in the indirection
  * array when extents have been added or removed from one of the
  * extent lists. erp_idx contains the irec index to begin updating
  * at and ext_diff contains the number of extents that were added
  * or removed.
  */
 void
 xfs_iext_irec_update_extoffs(
 	xfs_ifork_t	*ifp,		/* inode fork pointer */
 	int		erp_idx,	/* irec index to update */
 	int		ext_diff)	/* number of new extents */
 {
 	int		i;		/* loop counter */
 	int		nlists;		/* number of irec's (ex lists */
 	ASSERT(ifp->if_flags & XFS_IFEXTIREC);
 	nlists = ifp->if_real_bytes / XFS_IEXT_BUFSZ;
 	for (i = erp_idx; i < nlists; i++) {
 		ifp->if_u1.if_ext_irec[i].er_extoff += ext_diff;
 	}
 }

fs/xfs/xfs_inode.h

Diff comments View file @ 745b1f4

fs/xfs/xfs_itable.c

Diff comments View file @ 745b1f4

 /*
  * Copyright (c) 2000-2002,2005 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_ialloc.h"
 #include "xfs_itable.h"
 #include "xfs_error.h"
 #include "xfs_btree.h"
 STATIC int
 xfs_bulkstat_one_iget(
 	xfs_mount_t	*mp,		/* mount point for filesystem */
 	xfs_ino_t	ino,		/* inode number to get data for */
 	xfs_daddr_t	bno,		/* starting bno of inode cluster */
 	xfs_bstat_t	*buf,		/* return buffer */
 	int		*stat)		/* BULKSTAT_RV_... */
 {
 	xfs_dinode_core_t *dic;		/* dinode core info pointer */
 	xfs_inode_t	*ip;		/* incore inode pointer */
 	bhv_vnode_t	*vp;
 	int		error;
-	error = xfs_iget(mp, NULL, ino, 0, XFS_ILOCK_SHARED, &ip, bno);
+	error = xfs_iget(mp, NULL, ino,
+			 XFS_IGET_BULKSTAT, XFS_ILOCK_SHARED, &ip, bno);
 	if (error) {
 		*stat = BULKSTAT_RV_NOTHING;
 		return error;
 	}
 	ASSERT(ip != NULL);
 	ASSERT(ip->i_blkno != (xfs_daddr_t)0);
 	if (ip->i_d.di_mode == 0) {
 		*stat = BULKSTAT_RV_NOTHING;
 		error = XFS_ERROR(ENOENT);
 		goto out_iput;
 	}
 	vp = XFS_ITOV(ip);
 	dic = &ip->i_d;
 	/* xfs_iget returns the following without needing
 	 * further change.
 	 */
 	buf->bs_nlink = dic->di_nlink;
 	buf->bs_projid = dic->di_projid;
 	buf->bs_ino = ino;
 	buf->bs_mode = dic->di_mode;
 	buf->bs_uid = dic->di_uid;
 	buf->bs_gid = dic->di_gid;
 	buf->bs_size = dic->di_size;
 	vn_atime_to_bstime(vp, &buf->bs_atime);
 	buf->bs_mtime.tv_sec = dic->di_mtime.t_sec;
 	buf->bs_mtime.tv_nsec = dic->di_mtime.t_nsec;
 	buf->bs_ctime.tv_sec = dic->di_ctime.t_sec;
 	buf->bs_ctime.tv_nsec = dic->di_ctime.t_nsec;
 	buf->bs_xflags = xfs_ip2xflags(ip);
 	buf->bs_extsize = dic->di_extsize << mp->m_sb.sb_blocklog;
 	buf->bs_extents = dic->di_nextents;
 	buf->bs_gen = dic->di_gen;
 	memset(buf->bs_pad, 0, sizeof(buf->bs_pad));
 	buf->bs_dmevmask = dic->di_dmevmask;
 	buf->bs_dmstate = dic->di_dmstate;
 	buf->bs_aextents = dic->di_anextents;
 	switch (dic->di_format) {
 	case XFS_DINODE_FMT_DEV:
 		buf->bs_rdev = ip->i_df.if_u2.if_rdev;
 		buf->bs_blksize = BLKDEV_IOSIZE;
 		buf->bs_blocks = 0;
 		break;
 	case XFS_DINODE_FMT_LOCAL:
 	case XFS_DINODE_FMT_UUID:
 		buf->bs_rdev = 0;
 		buf->bs_blksize = mp->m_sb.sb_blocksize;
 		buf->bs_blocks = 0;
 		break;
 	case XFS_DINODE_FMT_EXTENTS:
 	case XFS_DINODE_FMT_BTREE:
 		buf->bs_rdev = 0;
 		buf->bs_blksize = mp->m_sb.sb_blocksize;
 		buf->bs_blocks = dic->di_nblocks + ip->i_delayed_blks;
 		break;
 	}
  out_iput:
 	xfs_iput(ip, XFS_ILOCK_SHARED);
 	return error;
 }
 STATIC int
 xfs_bulkstat_one_dinode(
 	xfs_mount_t	*mp,		/* mount point for filesystem */
 	xfs_ino_t	ino,		/* inode number to get data for */
 	xfs_dinode_t	*dip,		/* dinode inode pointer */
 	xfs_bstat_t	*buf)		/* return buffer */
 {
 	xfs_dinode_core_t *dic;		/* dinode core info pointer */
 	dic = &dip->di_core;
 	/*
 	 * The inode format changed when we moved the link count and
 	 * made it 32 bits long.  If this is an old format inode,
 	 * convert it in memory to look like a new one.  If it gets
 	 * flushed to disk we will convert back before flushing or
 	 * logging it.  We zero out the new projid field and the old link
 	 * count field.  We'll handle clearing the pad field (the remains
 	 * of the old uuid field) when we actually convert the inode to
 	 * the new format. We don't change the version number so that we
 	 * can distinguish this from a real new format inode.
 	 */
 	if (INT_GET(dic->di_version, ARCH_CONVERT) == XFS_DINODE_VERSION_1) {
 		buf->bs_nlink = INT_GET(dic->di_onlink, ARCH_CONVERT);
 		buf->bs_projid = 0;
 	} else {
 		buf->bs_nlink = INT_GET(dic->di_nlink, ARCH_CONVERT);
 		buf->bs_projid = INT_GET(dic->di_projid, ARCH_CONVERT);
 	}
 	buf->bs_ino = ino;
 	buf->bs_mode = INT_GET(dic->di_mode, ARCH_CONVERT);
 	buf->bs_uid = INT_GET(dic->di_uid, ARCH_CONVERT);
 	buf->bs_gid = INT_GET(dic->di_gid, ARCH_CONVERT);
 	buf->bs_size = INT_GET(dic->di_size, ARCH_CONVERT);
 	buf->bs_atime.tv_sec = INT_GET(dic->di_atime.t_sec, ARCH_CONVERT);
 	buf->bs_atime.tv_nsec = INT_GET(dic->di_atime.t_nsec, ARCH_CONVERT);
 	buf->bs_mtime.tv_sec = INT_GET(dic->di_mtime.t_sec, ARCH_CONVERT);
 	buf->bs_mtime.tv_nsec = INT_GET(dic->di_mtime.t_nsec, ARCH_CONVERT);
 	buf->bs_ctime.tv_sec = INT_GET(dic->di_ctime.t_sec, ARCH_CONVERT);
 	buf->bs_ctime.tv_nsec = INT_GET(dic->di_ctime.t_nsec, ARCH_CONVERT);
 	buf->bs_xflags = xfs_dic2xflags(dic);
 	buf->bs_extsize = INT_GET(dic->di_extsize, ARCH_CONVERT) << mp->m_sb.sb_blocklog;
 	buf->bs_extents = INT_GET(dic->di_nextents, ARCH_CONVERT);
 	buf->bs_gen = INT_GET(dic->di_gen, ARCH_CONVERT);
 	memset(buf->bs_pad, 0, sizeof(buf->bs_pad));
 	buf->bs_dmevmask = INT_GET(dic->di_dmevmask, ARCH_CONVERT);
 	buf->bs_dmstate = INT_GET(dic->di_dmstate, ARCH_CONVERT);
 	buf->bs_aextents = INT_GET(dic->di_anextents, ARCH_CONVERT);
 	switch (INT_GET(dic->di_format, ARCH_CONVERT)) {
 	case XFS_DINODE_FMT_DEV:
 		buf->bs_rdev = INT_GET(dip->di_u.di_dev, ARCH_CONVERT);
 		buf->bs_blksize = BLKDEV_IOSIZE;
 		buf->bs_blocks = 0;
 		break;
 	case XFS_DINODE_FMT_LOCAL:
 	case XFS_DINODE_FMT_UUID:
 		buf->bs_rdev = 0;
 		buf->bs_blksize = mp->m_sb.sb_blocksize;
 		buf->bs_blocks = 0;
 		break;
 	case XFS_DINODE_FMT_EXTENTS:
 	case XFS_DINODE_FMT_BTREE:
 		buf->bs_rdev = 0;
 		buf->bs_blksize = mp->m_sb.sb_blocksize;
 		buf->bs_blocks = INT_GET(dic->di_nblocks, ARCH_CONVERT);
 		break;
 	}
 	return 0;
 }
 /*
  * Return stat information for one inode.
  * Return 0 if ok, else errno.
  */
 int		       		/* error status */
 xfs_bulkstat_one(
 	xfs_mount_t	*mp,		/* mount point for filesystem */
 	xfs_ino_t	ino,		/* inode number to get data for */
 	void		__user *buffer,	/* buffer to place output in */
 	int		ubsize,		/* size of buffer */
 	void		*private_data,	/* my private data */
 	xfs_daddr_t	bno,		/* starting bno of inode cluster */
 	int		*ubused,	/* bytes used by me */
 	void		*dibuff,	/* on-disk inode buffer */
 	int		*stat)		/* BULKSTAT_RV_... */
 {
 	xfs_bstat_t	*buf;		/* return buffer */
 	int		error = 0;	/* error value */
 	xfs_dinode_t	*dip;		/* dinode inode pointer */
 	dip = (xfs_dinode_t *)dibuff;
 	if (!buffer || ino == mp->m_sb.sb_rbmino || ino == mp->m_sb.sb_rsumino ||
 	    (XFS_SB_VERSION_HASQUOTA(&mp->m_sb) &&
 	     (ino == mp->m_sb.sb_uquotino || ino == mp->m_sb.sb_gquotino))) {
 		*stat = BULKSTAT_RV_NOTHING;
 		return XFS_ERROR(EINVAL);
 	}
 	if (ubsize < sizeof(*buf)) {
 		*stat = BULKSTAT_RV_NOTHING;
 		return XFS_ERROR(ENOMEM);
 	}
 	buf = kmem_alloc(sizeof(*buf), KM_SLEEP);
 	if (dip == NULL) {
 		/* We're not being passed a pointer to a dinode.  This happens
 		 * if BULKSTAT_FG_IGET is selected.  Do the iget.
 		 */
 		error = xfs_bulkstat_one_iget(mp, ino, bno, buf, stat);
 		if (error)
 			goto out_free;
 	} else {
 		xfs_bulkstat_one_dinode(mp, ino, dip, buf);
 	}
 	if (copy_to_user(buffer, buf, sizeof(*buf)))  {
 		*stat = BULKSTAT_RV_NOTHING;
 		error =  EFAULT;
 		goto out_free;
 	}
 	*stat = BULKSTAT_RV_DIDONE;
 	if (ubused)
 		*ubused = sizeof(*buf);
  out_free:
 	kmem_free(buf, sizeof(*buf));
 	return error;
 }
 /*
  * Test to see whether we can use the ondisk inode directly, based
  * on the given bulkstat flags, filling in dipp accordingly.
  * Returns zero if the inode is dodgey.
  */
 STATIC int
 xfs_bulkstat_use_dinode(
 	xfs_mount_t	*mp,
 	int		flags,
 	xfs_buf_t	*bp,
 	int		clustidx,
 	xfs_dinode_t	**dipp)
 {
 	xfs_dinode_t	*dip;
 	unsigned int	aformat;
 	*dipp = NULL;
 	if (!bp || (flags & BULKSTAT_FG_IGET))
 		return 1;
 	dip = (xfs_dinode_t *)
 			xfs_buf_offset(bp, clustidx << mp->m_sb.sb_inodelog);
 	if (INT_GET(dip->di_core.di_magic, ARCH_CONVERT) != XFS_DINODE_MAGIC ||
 	    !XFS_DINODE_GOOD_VERSION(
 			INT_GET(dip->di_core.di_version, ARCH_CONVERT)))
 		return 0;
 	if (flags & BULKSTAT_FG_QUICK) {
 		*dipp = dip;
 		return 1;
 	}
 	/* BULKSTAT_FG_INLINE: if attr fork is local, or not there, use it */
 	aformat = INT_GET(dip->di_core.di_aformat, ARCH_CONVERT);
 	if ((XFS_CFORK_Q(&dip->di_core) == 0) ||
 	    (aformat == XFS_DINODE_FMT_LOCAL) ||
 	    (aformat == XFS_DINODE_FMT_EXTENTS && !dip->di_core.di_anextents)) {
 		*dipp = dip;
 		return 1;
 	}
 	return 1;
 }
 /*
  * Return stat information in bulk (by-inode) for the filesystem.
  */
 int					/* error status */
 xfs_bulkstat(
 	xfs_mount_t		*mp,	/* mount point for filesystem */
 	xfs_ino_t		*lastinop, /* last inode returned */
 	int			*ubcountp, /* size of buffer/count returned */
 	bulkstat_one_pf		formatter, /* func that'd fill a single buf */
 	void			*private_data,/* private data for formatter */
 	size_t			statstruct_size, /* sizeof struct filling */
 	char			__user *ubuffer, /* buffer with inode stats */
 	int			flags,	/* defined in xfs_itable.h */
 	int			*done)	/* 1 if there are more stats to get */
 {
 	xfs_agblock_t		agbno=0;/* allocation group block number */
 	xfs_buf_t		*agbp;	/* agi header buffer */
 	xfs_agi_t		*agi;	/* agi header data */
 	xfs_agino_t		agino;	/* inode # in allocation group */
 	xfs_agnumber_t		agno;	/* allocation group number */
 	xfs_daddr_t		bno;	/* inode cluster start daddr */
 	int			chunkidx; /* current index into inode chunk */
 	int			clustidx; /* current index into inode cluster */
 	xfs_btree_cur_t		*cur;	/* btree cursor for ialloc btree */
 	int			end_of_ag; /* set if we've seen the ag end */
 	int			error;	/* error code */
 	int                     fmterror;/* bulkstat formatter result */
 	__int32_t		gcnt;	/* current btree rec's count */
 	xfs_inofree_t		gfree;	/* current btree rec's free mask */
 	xfs_agino_t		gino;	/* current btree rec's start inode */
 	int			i;	/* loop index */
 	int			icount;	/* count of inodes good in irbuf */
 	int			irbsize; /* size of irec buffer in bytes */
 	unsigned int		kmflags; /* flags for allocating irec buffer */
 	xfs_ino_t		ino;	/* inode number (filesystem) */
 	xfs_inobt_rec_incore_t	*irbp;	/* current irec buffer pointer */
 	xfs_inobt_rec_incore_t	*irbuf;	/* start of irec buffer */
 	xfs_inobt_rec_incore_t	*irbufend; /* end of good irec buffer entries */
 	xfs_ino_t		lastino=0; /* last inode number returned */
 	int			nbcluster; /* # of blocks in a cluster */
 	int			nicluster; /* # of inodes in a cluster */
 	int			nimask;	/* mask for inode clusters */
 	int			nirbuf;	/* size of irbuf */
 	int			rval;	/* return value error code */
 	int			tmp;	/* result value from btree calls */
 	int			ubcount; /* size of user's buffer */
 	int			ubleft;	/* bytes left in user's buffer */
 	char			__user *ubufp;	/* pointer into user's buffer */
 	int			ubelem;	/* spaces used in user's buffer */
 	int			ubused;	/* bytes used by formatter */
 	xfs_buf_t		*bp;	/* ptr to on-disk inode cluster buf */
 	xfs_dinode_t		*dip;	/* ptr into bp for specific inode */
 	xfs_inode_t		*ip;	/* ptr to in-core inode struct */
 	/*
 	 * Get the last inode value, see if there's nothing to do.
 	 */
 	ino = (xfs_ino_t)*lastinop;
 	dip = NULL;
 	agno = XFS_INO_TO_AGNO(mp, ino);
 	agino = XFS_INO_TO_AGINO(mp, ino);
 	if (agno >= mp->m_sb.sb_agcount ||
 	    ino != XFS_AGINO_TO_INO(mp, agno, agino)) {
 		*done = 1;
 		*ubcountp = 0;
 		return 0;
 	}
 	ubcount = *ubcountp; /* statstruct's */
 	ubleft = ubcount * statstruct_size; /* bytes */
 	*ubcountp = ubelem = 0;
 	*done = 0;
 	fmterror = 0;
 	ubufp = ubuffer;
 	nicluster = mp->m_sb.sb_blocksize >= XFS_INODE_CLUSTER_SIZE(mp) ?
 		mp->m_sb.sb_inopblock :
 		(XFS_INODE_CLUSTER_SIZE(mp) >> mp->m_sb.sb_inodelog);
 	nimask = ~(nicluster - 1);
 	nbcluster = nicluster >> mp->m_sb.sb_inopblog;
 	/*
 	 * Allocate a local buffer for inode cluster btree records.
 	 * This caps our maximum readahead window (so don't be stingy)
 	 * but we must handle the case where we can't get a contiguous
 	 * multi-page buffer, so we drop back toward pagesize; the end
 	 * case we ensure succeeds, via appropriate allocation flags.
 	 */
 	irbsize = NBPP * 4;
 	kmflags = KM_SLEEP | KM_MAYFAIL;
 	while (!(irbuf = kmem_alloc(irbsize, kmflags))) {
 		if ((irbsize >>= 1) <= NBPP)
 			kmflags = KM_SLEEP;
 	}
 	nirbuf = irbsize / sizeof(*irbuf);
 	/*
 	 * Loop over the allocation groups, starting from the last
 	 * inode returned; 0 means start of the allocation group.
 	 */
 	rval = 0;
 	while (ubleft >= statstruct_size && agno < mp->m_sb.sb_agcount) {
 		bp = NULL;
 		down_read(&mp->m_peraglock);
 		error = xfs_ialloc_read_agi(mp, NULL, agno, &agbp);
 		up_read(&mp->m_peraglock);
 		if (error) {
 			/*
 			 * Skip this allocation group and go to the next one.
 			 */
 			agno++;
 			agino = 0;
 			continue;
 		}
 		agi = XFS_BUF_TO_AGI(agbp);
 		/*
 		 * Allocate and initialize a btree cursor for ialloc btree.
 		 */
 		cur = xfs_btree_init_cursor(mp, NULL, agbp, agno, XFS_BTNUM_INO,
 						(xfs_inode_t *)0, 0);
 		irbp = irbuf;
 		irbufend = irbuf + nirbuf;
 		end_of_ag = 0;
 		/*
 		 * If we're returning in the middle of an allocation group,
 		 * we need to get the remainder of the chunk we're in.
 		 */
 		if (agino > 0) {
 			/*
 			 * Lookup the inode chunk that this inode lives in.
 			 */
 			error = xfs_inobt_lookup_le(cur, agino, 0, 0, &tmp);
 			if (!error &&	/* no I/O error */
 			    tmp &&	/* lookup succeeded */
 					/* got the record, should always work */
 			    !(error = xfs_inobt_get_rec(cur, &gino, &gcnt,
 				    &gfree, &i)) &&
 			    i == 1 &&
 					/* this is the right chunk */
 			    agino < gino + XFS_INODES_PER_CHUNK &&
 					/* lastino was not last in chunk */
 			    (chunkidx = agino - gino + 1) <
 				    XFS_INODES_PER_CHUNK &&
 					/* there are some left allocated */
 			    XFS_INOBT_MASKN(chunkidx,
 				    XFS_INODES_PER_CHUNK - chunkidx) & ~gfree) {
 				/*
 				 * Grab the chunk record.  Mark all the
 				 * uninteresting inodes (because they're
 				 * before our start point) free.
 				 */
 				for (i = 0; i < chunkidx; i++) {
 					if (XFS_INOBT_MASK(i) & ~gfree)
 						gcnt++;
 				}
 				gfree |= XFS_INOBT_MASKN(0, chunkidx);
 				irbp->ir_startino = gino;
 				irbp->ir_freecount = gcnt;
 				irbp->ir_free = gfree;
 				irbp++;
 				agino = gino + XFS_INODES_PER_CHUNK;
 				icount = XFS_INODES_PER_CHUNK - gcnt;
 			} else {
 				/*
 				 * If any of those tests failed, bump the
 				 * inode number (just in case).
 				 */
 				agino++;
 				icount = 0;
 			}
 			/*
 			 * In any case, increment to the next record.
 			 */
 			if (!error)
 				error = xfs_inobt_increment(cur, 0, &tmp);
 		} else {
 			/*
 			 * Start of ag.  Lookup the first inode chunk.
 			 */
 			error = xfs_inobt_lookup_ge(cur, 0, 0, 0, &tmp);
 			icount = 0;
 		}
 		/*
 		 * Loop through inode btree records in this ag,
 		 * until we run out of inodes or space in the buffer.
 		 */
 		while (irbp < irbufend && icount < ubcount) {
 			/*
 			 * Loop as long as we're unable to read the
 			 * inode btree.
 			 */
 			while (error) {
 				agino += XFS_INODES_PER_CHUNK;
 				if (XFS_AGINO_TO_AGBNO(mp, agino) >=
 						be32_to_cpu(agi->agi_length))
 					break;
 				error = xfs_inobt_lookup_ge(cur, agino, 0, 0,
 							    &tmp);
 			}
 			/*
 			 * If ran off the end of the ag either with an error,
 			 * or the normal way, set end and stop collecting.
 			 */
 			if (error ||
 			    (error = xfs_inobt_get_rec(cur, &gino, &gcnt,
 				    &gfree, &i)) ||
 			    i == 0) {
 				end_of_ag = 1;
 				break;
 			}
 			/*
 			 * If this chunk has any allocated inodes, save it.
 			 * Also start read-ahead now for this chunk.
 			 */
 			if (gcnt < XFS_INODES_PER_CHUNK) {
 				/*
 				 * Loop over all clusters in the next chunk.
 				 * Do a readahead if there are any allocated
 				 * inodes in that cluster.
 				 */
 				for (agbno = XFS_AGINO_TO_AGBNO(mp, gino),
 				     chunkidx = 0;
 				     chunkidx < XFS_INODES_PER_CHUNK;
 				     chunkidx += nicluster,
 				     agbno += nbcluster) {
 					if (XFS_INOBT_MASKN(chunkidx,
 							    nicluster) & ~gfree)
 						xfs_btree_reada_bufs(mp, agno,
 							agbno, nbcluster);
 				}
 				irbp->ir_startino = gino;
 				irbp->ir_freecount = gcnt;
 				irbp->ir_free = gfree;
 				irbp++;
 				icount += XFS_INODES_PER_CHUNK - gcnt;
 			}
 			/*
 			 * Set agino to after this chunk and bump the cursor.
 			 */
 			agino = gino + XFS_INODES_PER_CHUNK;
 			error = xfs_inobt_increment(cur, 0, &tmp);
 		}
 		/*
 		 * Drop the btree buffers and the agi buffer.
 		 * We can't hold any of the locks these represent
 		 * when calling iget.
 		 */
 		xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
 		xfs_buf_relse(agbp);
 		/*
 		 * Now format all the good inodes into the user's buffer.
 		 */
 		irbufend = irbp;
 		for (irbp = irbuf;
 		     irbp < irbufend && ubleft >= statstruct_size; irbp++) {
 			/*
 			 * Now process this chunk of inodes.
 			 */
 			for (agino = irbp->ir_startino, chunkidx = clustidx = 0;
 			     ubleft > 0 &&
 				irbp->ir_freecount < XFS_INODES_PER_CHUNK;
 			     chunkidx++, clustidx++, agino++) {
 				ASSERT(chunkidx < XFS_INODES_PER_CHUNK);
 				/*
 				 * Recompute agbno if this is the
 				 * first inode of the cluster.
 				 *
 				 * Careful with clustidx.   There can be
 				 * multple clusters per chunk, a single
 				 * cluster per chunk or a cluster that has
 				 * inodes represented from several different
 				 * chunks (if blocksize is large).
 				 *
 				 * Because of this, the starting clustidx is
 				 * initialized to zero in this loop but must
 				 * later be reset after reading in the cluster
 				 * buffer.
 				 */
 				if ((chunkidx & (nicluster - 1)) == 0) {
 					agbno = XFS_AGINO_TO_AGBNO(mp,
 							irbp->ir_startino) +
 						((chunkidx & nimask) >>
 						 mp->m_sb.sb_inopblog);
 					if (flags & (BULKSTAT_FG_QUICK |
 						     BULKSTAT_FG_INLINE)) {
 						ino = XFS_AGINO_TO_INO(mp, agno,
 								       agino);
 						bno = XFS_AGB_TO_DADDR(mp, agno,
 								       agbno);
 						/*
 						 * Get the inode cluster buffer
 						 */
 						ASSERT(xfs_inode_zone != NULL);
 						ip = kmem_zone_zalloc(xfs_inode_zone,
 								      KM_SLEEP);
 						ip->i_ino = ino;
 						ip->i_mount = mp;
 						if (bp)
 							xfs_buf_relse(bp);
 						error = xfs_itobp(mp, NULL, ip,
 								&dip, &bp, bno,
 								XFS_IMAP_BULKSTAT);
 						if (!error)
 							clustidx = ip->i_boffset / mp->m_sb.sb_inodesize;
 						kmem_zone_free(xfs_inode_zone, ip);
 						if (XFS_TEST_ERROR(error != 0,
 								   mp, XFS_ERRTAG_BULKSTAT_READ_CHUNK,
 								   XFS_RANDOM_BULKSTAT_READ_CHUNK)) {
 							bp = NULL;
 							ubleft = 0;
 							rval = error;
 							break;
 						}
 					}
 				}
 				/*
 				 * Skip if this inode is free.
 				 */
 				if (XFS_INOBT_MASK(chunkidx) & irbp->ir_free)
 					continue;
 				/*
 				 * Count used inodes as free so we can tell
 				 * when the chunk is used up.
 				 */
 				irbp->ir_freecount++;
 				ino = XFS_AGINO_TO_INO(mp, agno, agino);
 				bno = XFS_AGB_TO_DADDR(mp, agno, agbno);
 				if (!xfs_bulkstat_use_dinode(mp, flags, bp,
 							     clustidx, &dip))
 					continue;
 				/*
 				 * If we need to do an iget, cannot hold bp.
 				 * Drop it, until starting the next cluster.
 				 */
 				if ((flags & BULKSTAT_FG_INLINE) && !dip) {
 					if (bp)
 						xfs_buf_relse(bp);
 					bp = NULL;
 				}
 				/*
 				 * Get the inode and fill in a single buffer.
 				 * BULKSTAT_FG_QUICK uses dip to fill it in.
 				 * BULKSTAT_FG_IGET uses igets.
 				 * BULKSTAT_FG_INLINE uses dip if we have an
 				 * inline attr fork, else igets.
 				 * See: xfs_bulkstat_one & xfs_dm_bulkstat_one.
 				 * This is also used to count inodes/blks, etc
 				 * in xfs_qm_quotacheck.
 				 */
 				ubused = statstruct_size;
 				error = formatter(mp, ino, ubufp,
 						ubleft, private_data,
 						bno, &ubused, dip, &fmterror);
 				if (fmterror == BULKSTAT_RV_NOTHING) {
 					if (error == ENOMEM)
 						ubleft = 0;
 					continue;
 				}
 				if (fmterror == BULKSTAT_RV_GIVEUP) {
 					ubleft = 0;
 					ASSERT(error);
 					rval = error;
 					break;
 				}
 				if (ubufp)
 					ubufp += ubused;
 				ubleft -= ubused;
 				ubelem++;
 				lastino = ino;
 			}
 		}
 		if (bp)
 			xfs_buf_relse(bp);
 		/*
 		 * Set up for the next loop iteration.
 		 */
 		if (ubleft > 0) {
 			if (end_of_ag) {
 				agno++;
 				agino = 0;
 			} else
 				agino = XFS_INO_TO_AGINO(mp, lastino);
 		} else
 			break;
 	}
 	/*
 	 * Done, we're either out of filesystem or space to put the data.
 	 */
 	kmem_free(irbuf, irbsize);
 	*ubcountp = ubelem;
 	if (agno >= mp->m_sb.sb_agcount) {
 		/*
 		 * If we ran out of filesystem, mark lastino as off
 		 * the end of the filesystem, so the next call
 		 * will return immediately.
 		 */
 		*lastinop = (xfs_ino_t)XFS_AGINO_TO_INO(mp, agno, 0);
 		*done = 1;
 	} else
 		*lastinop = (xfs_ino_t)lastino;
 	return rval;
 }
 /*
  * Return stat information in bulk (by-inode) for the filesystem.
  * Special case for non-sequential one inode bulkstat.
  */
 int					/* error status */
 xfs_bulkstat_single(
 	xfs_mount_t		*mp,	/* mount point for filesystem */
 	xfs_ino_t		*lastinop, /* inode to return */
 	char			__user *buffer, /* buffer with inode stats */
 	int			*done)	/* 1 if there are more stats to get */
 {
 	int			count;	/* count value for bulkstat call */
 	int			error;	/* return value */
 	xfs_ino_t		ino;	/* filesystem inode number */
 	int			res;	/* result from bs1 */
 	/*
 	 * note that requesting valid inode numbers which are not allocated
 	 * to inodes will most likely cause xfs_itobp to generate warning
 	 * messages about bad magic numbers. This is ok. The fact that
 	 * the inode isn't actually an inode is handled by the
 	 * error check below. Done this way to make the usual case faster
 	 * at the expense of the error case.
 	 */
 	ino = (xfs_ino_t)*lastinop;
 	error = xfs_bulkstat_one(mp, ino, buffer, sizeof(xfs_bstat_t),
 				 NULL, 0, NULL, NULL, &res);
 	if (error) {
 		/*
 		 * Special case way failed, do it the "long" way
 		 * to see if that works.
 		 */
 		(*lastinop)--;
 		count = 1;
 		if (xfs_bulkstat(mp, lastinop, &count, xfs_bulkstat_one,
 				NULL, sizeof(xfs_bstat_t), buffer,
 				BULKSTAT_FG_IGET, done))
 			return error;
 		if (count == 0 || (xfs_ino_t)*lastinop != ino)
 			return error == EFSCORRUPTED ?
 				XFS_ERROR(EINVAL) : error;
 		else
 			return 0;
 	}
 	*done = 0;
 	return 0;
 }
 /*
  * Return inode number table for the filesystem.
  */
 int					/* error status */
 xfs_inumbers(
 	xfs_mount_t	*mp,		/* mount point for filesystem */
 	xfs_ino_t	*lastino,	/* last inode returned */
 	int		*count,		/* size of buffer/count returned */
 	xfs_inogrp_t	__user *ubuffer)/* buffer with inode descriptions */
 {
 	xfs_buf_t	*agbp;
 	xfs_agino_t	agino;
 	xfs_agnumber_t	agno;
 	int		bcount;
 	xfs_inogrp_t	*buffer;
 	int		bufidx;
 	xfs_btree_cur_t	*cur;
 	int		error;
 	__int32_t	gcnt;
 	xfs_inofree_t	gfree;
 	xfs_agino_t	gino;
 	int		i;
 	xfs_ino_t	ino;
 	int		left;
 	int		tmp;
 	ino = (xfs_ino_t)*lastino;
 	agno = XFS_INO_TO_AGNO(mp, ino);
 	agino = XFS_INO_TO_AGINO(mp, ino);
 	left = *count;
 	*count = 0;
 	bcount = MIN(left, (int)(NBPP / sizeof(*buffer)));
 	buffer = kmem_alloc(bcount * sizeof(*buffer), KM_SLEEP);
 	error = bufidx = 0;
 	cur = NULL;
 	agbp = NULL;
 	while (left > 0 && agno < mp->m_sb.sb_agcount) {
 		if (agbp == NULL) {
 			down_read(&mp->m_peraglock);
 			error = xfs_ialloc_read_agi(mp, NULL, agno, &agbp);
 			up_read(&mp->m_peraglock);
 			if (error) {
 				/*
 				 * If we can't read the AGI of this ag,
 				 * then just skip to the next one.
 				 */
 				ASSERT(cur == NULL);
 				agbp = NULL;
 				agno++;
 				agino = 0;
 				continue;
 			}
 			cur = xfs_btree_init_cursor(mp, NULL, agbp, agno,
 				XFS_BTNUM_INO, (xfs_inode_t *)0, 0);
 			error = xfs_inobt_lookup_ge(cur, agino, 0, 0, &tmp);
 			if (error) {
 				xfs_btree_del_cursor(cur, XFS_BTREE_ERROR);
 				cur = NULL;
 				xfs_buf_relse(agbp);
 				agbp = NULL;
 				/*
 				 * Move up the the last inode in the current
 				 * chunk.  The lookup_ge will always get
 				 * us the first inode in the next chunk.
 				 */
 				agino += XFS_INODES_PER_CHUNK - 1;
 				continue;
 			}
 		}
 		if ((error = xfs_inobt_get_rec(cur, &gino, &gcnt, &gfree,
 			&i)) ||
 		    i == 0) {
 			xfs_buf_relse(agbp);
 			agbp = NULL;
 			xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
 			cur = NULL;
 			agno++;
 			agino = 0;
 			continue;
 		}
 		agino = gino + XFS_INODES_PER_CHUNK - 1;
 		buffer[bufidx].xi_startino = XFS_AGINO_TO_INO(mp, agno, gino);
 		buffer[bufidx].xi_alloccount = XFS_INODES_PER_CHUNK - gcnt;
 		buffer[bufidx].xi_allocmask = ~gfree;
 		bufidx++;
 		left--;
 		if (bufidx == bcount) {
 			if (copy_to_user(ubuffer, buffer,
 					bufidx * sizeof(*buffer))) {
 				error = XFS_ERROR(EFAULT);
 				break;
 			}
 			ubuffer += bufidx;
 			*count += bufidx;
 			bufidx = 0;
 		}
 		if (left) {
 			error = xfs_inobt_increment(cur, 0, &tmp);
 			if (error) {
 				xfs_btree_del_cursor(cur, XFS_BTREE_ERROR);
 				cur = NULL;
 				xfs_buf_relse(agbp);
 				agbp = NULL;
 				/*
 				 * The agino value has already been bumped.
 				 * Just try to skip up to it.
 				 */
 				agino += XFS_INODES_PER_CHUNK;
 				continue;
 			}
 		}
 	}
 	if (!error) {
 		if (bufidx) {
 			if (copy_to_user(ubuffer, buffer,
 					bufidx * sizeof(*buffer)))
 				error = XFS_ERROR(EFAULT);
 			else
 				*count += bufidx;
 		}
 		*lastino = XFS_AGINO_TO_INO(mp, agno, agino);
 	}
 	kmem_free(buffer, bcount * sizeof(*buffer));
 	if (cur)
 		xfs_btree_del_cursor(cur, (error ? XFS_BTREE_ERROR :
 					   XFS_BTREE_NOERROR));
 	if (agbp)
 		xfs_buf_relse(agbp);
 	return error;
 }