fs/xfs/xfs_inode.c

// SPDX-License-Identifier: GPL-2.0
/*
 * Copyright (c) 2000-2006 Silicon Graphics, Inc.
 * All Rights Reserved.
 */
#include <linux/iversion.h>

#include "xfs.h"
#include "xfs_fs.h"
#include "xfs_shared.h"
#include "xfs_format.h"
#include "xfs_log_format.h"
#include "xfs_trans_resv.h"
#include "xfs_mount.h"
#include "xfs_defer.h"
#include "xfs_inode.h"
#include "xfs_dir2.h"
#include "xfs_attr.h"
#include "xfs_trans_space.h"
#include "xfs_trans.h"
#include "xfs_buf_item.h"
#include "xfs_inode_item.h"
#include "xfs_ialloc.h"
#include "xfs_bmap.h"
#include "xfs_bmap_util.h"
#include "xfs_errortag.h"
#include "xfs_error.h"
#include "xfs_quota.h"
#include "xfs_filestream.h"
#include "xfs_trace.h"
#include "xfs_icache.h"
#include "xfs_symlink.h"
#include "xfs_trans_priv.h"
#include "xfs_log.h"
#include "xfs_bmap_btree.h"
#include "xfs_reflink.h"
#include "xfs_ag.h"

struct kmem_cache *xfs_inode_zone;

/*
 * Used in xfs_itruncate_extents().  This is the maximum number of extents
 * freed from a file in a single transaction.
 */
#define	XFS_ITRUNC_MAX_EXTENTS	2

STATIC int xfs_iunlink(struct xfs_trans *, struct xfs_inode *);
STATIC int xfs_iunlink_remove(struct xfs_trans *tp, struct xfs_perag *pag,
	struct xfs_inode *);

/*
 * helper function to extract extent size hint from inode
 */
xfs_extlen_t
xfs_get_extsz_hint(
	struct xfs_inode	*ip)
{
	/*
	 * No point in aligning allocations if we need to COW to actually
	 * write to them.
	 */
	if (xfs_is_always_cow_inode(ip))
		return 0;
	if ((ip->i_diflags & XFS_DIFLAG_EXTSIZE) && ip->i_extsize)
		return ip->i_extsize;
	if (XFS_IS_REALTIME_INODE(ip))
		return ip->i_mount->m_sb.sb_rextsize;
	return 0;
}

/*
 * Helper function to extract CoW extent size hint from inode.
 * Between the extent size hint and the CoW extent size hint, we
 * return the greater of the two.  If the value is zero (automatic),
 * use the default size.
 */
xfs_extlen_t
xfs_get_cowextsz_hint(
	struct xfs_inode	*ip)
{
	xfs_extlen_t		a, b;

	a = 0;
	if (ip->i_diflags2 & XFS_DIFLAG2_COWEXTSIZE)
		a = ip->i_cowextsize;
	b = xfs_get_extsz_hint(ip);

	a = max(a, b);
	if (a == 0)
		return XFS_DEFAULT_COWEXTSZ_HINT;
	return a;
}

/*
 * These two are wrapper routines around the xfs_ilock() routine used to
 * centralize some grungy code.  They are used in places that wish to lock the
 * inode solely for reading the extents.  The reason these places can't just
 * call xfs_ilock(ip, XFS_ILOCK_SHARED) is that the inode lock also guards to
 * bringing in of the extents from disk for a file in b-tree format.  If the
 * inode is in b-tree format, then we need to lock the inode exclusively until
 * the extents are read in.  Locking it exclusively all the time would limit
 * our parallelism unnecessarily, though.  What we do instead is check to see
 * if the extents have been read in yet, and only lock the inode exclusively
 * if they have not.
 *
 * The functions return a value which should be given to the corresponding
 * xfs_iunlock() call.
 */
uint
xfs_ilock_data_map_shared(
	struct xfs_inode	*ip)
{
	uint			lock_mode = XFS_ILOCK_SHARED;

	if (xfs_need_iread_extents(&ip->i_df))
		lock_mode = XFS_ILOCK_EXCL;
	xfs_ilock(ip, lock_mode);
	return lock_mode;
}

uint
xfs_ilock_attr_map_shared(
	struct xfs_inode	*ip)
{
	uint			lock_mode = XFS_ILOCK_SHARED;

	if (ip->i_afp && xfs_need_iread_extents(ip->i_afp))
		lock_mode = XFS_ILOCK_EXCL;
	xfs_ilock(ip, lock_mode);
	return lock_mode;
}

/*
 * In addition to i_rwsem in the VFS inode, the xfs inode contains 2
 * multi-reader locks: invalidate_lock and the i_lock.  This routine allows
 * various combinations of the locks to be obtained.
 *
 * The 3 locks should always be ordered so that the IO lock is obtained first,
 * the mmap lock second and the ilock last in order to prevent deadlock.
 *
 * Basic locking order:
 *
 * i_rwsem -> invalidate_lock -> page_lock -> i_ilock
 *
 * mmap_lock locking order:
 *
 * i_rwsem -> page lock -> mmap_lock
 * mmap_lock -> invalidate_lock -> page_lock
 *
 * The difference in mmap_lock locking order mean that we cannot hold the
 * invalidate_lock over syscall based read(2)/write(2) based IO. These IO paths
 * can fault in pages during copy in/out (for buffered IO) or require the
 * mmap_lock in get_user_pages() to map the user pages into the kernel address
 * space for direct IO. Similarly the i_rwsem cannot be taken inside a page
 * fault because page faults already hold the mmap_lock.
 *
 * Hence to serialise fully against both syscall and mmap based IO, we need to
 * take both the i_rwsem and the invalidate_lock. These locks should *only* be
 * both taken in places where we need to invalidate the page cache in a race
 * free manner (e.g. truncate, hole punch and other extent manipulation
 * functions).
 */
void
xfs_ilock(
	xfs_inode_t		*ip,
	uint			lock_flags)
{
	trace_xfs_ilock(ip, lock_flags, _RET_IP_);

	/*
	 * You can't set both SHARED and EXCL for the same lock,
	 * and only XFS_IOLOCK_SHARED, XFS_IOLOCK_EXCL, XFS_ILOCK_SHARED,
	 * and XFS_ILOCK_EXCL are valid values to set in lock_flags.
	 */
	ASSERT((lock_flags & (XFS_IOLOCK_SHARED | XFS_IOLOCK_EXCL)) !=
	       (XFS_IOLOCK_SHARED | XFS_IOLOCK_EXCL));
	ASSERT((lock_flags & (XFS_MMAPLOCK_SHARED | XFS_MMAPLOCK_EXCL)) !=
	       (XFS_MMAPLOCK_SHARED | XFS_MMAPLOCK_EXCL));
	ASSERT((lock_flags & (XFS_ILOCK_SHARED | XFS_ILOCK_EXCL)) !=
	       (XFS_ILOCK_SHARED | XFS_ILOCK_EXCL));
	ASSERT((lock_flags & ~(XFS_LOCK_MASK | XFS_LOCK_SUBCLASS_MASK)) == 0);

	if (lock_flags & XFS_IOLOCK_EXCL) {
		down_write_nested(&VFS_I(ip)->i_rwsem,
				  XFS_IOLOCK_DEP(lock_flags));
	} else if (lock_flags & XFS_IOLOCK_SHARED) {
		down_read_nested(&VFS_I(ip)->i_rwsem,
				 XFS_IOLOCK_DEP(lock_flags));
	}

	if (lock_flags & XFS_MMAPLOCK_EXCL) {
		down_write_nested(&VFS_I(ip)->i_mapping->invalidate_lock,
				  XFS_MMAPLOCK_DEP(lock_flags));
	} else if (lock_flags & XFS_MMAPLOCK_SHARED) {
		down_read_nested(&VFS_I(ip)->i_mapping->invalidate_lock,
				 XFS_MMAPLOCK_DEP(lock_flags));
	}

	if (lock_flags & XFS_ILOCK_EXCL)
		mrupdate_nested(&ip->i_lock, XFS_ILOCK_DEP(lock_flags));
	else if (lock_flags & XFS_ILOCK_SHARED)
		mraccess_nested(&ip->i_lock, XFS_ILOCK_DEP(lock_flags));
}

/*
 * This is just like xfs_ilock(), except that the caller
 * is guaranteed not to sleep.  It returns 1 if it gets
 * the requested locks and 0 otherwise.  If the IO lock is
 * obtained but the inode lock cannot be, then the IO lock
 * is dropped before returning.
 *
 * ip -- the inode being locked
 * lock_flags -- this parameter indicates the inode's locks to be
 *       to be locked.  See the comment for xfs_ilock() for a list
 *	 of valid values.
 */
int
xfs_ilock_nowait(
	xfs_inode_t		*ip,
	uint			lock_flags)
{
	trace_xfs_ilock_nowait(ip, lock_flags, _RET_IP_);

	/*
	 * You can't set both SHARED and EXCL for the same lock,
	 * and only XFS_IOLOCK_SHARED, XFS_IOLOCK_EXCL, XFS_ILOCK_SHARED,
	 * and XFS_ILOCK_EXCL are valid values to set in lock_flags.
	 */
	ASSERT((lock_flags & (XFS_IOLOCK_SHARED | XFS_IOLOCK_EXCL)) !=
	       (XFS_IOLOCK_SHARED | XFS_IOLOCK_EXCL));
	ASSERT((lock_flags & (XFS_MMAPLOCK_SHARED | XFS_MMAPLOCK_EXCL)) !=
	       (XFS_MMAPLOCK_SHARED | XFS_MMAPLOCK_EXCL));
	ASSERT((lock_flags & (XFS_ILOCK_SHARED | XFS_ILOCK_EXCL)) !=
	       (XFS_ILOCK_SHARED | XFS_ILOCK_EXCL));
	ASSERT((lock_flags & ~(XFS_LOCK_MASK | XFS_LOCK_SUBCLASS_MASK)) == 0);

	if (lock_flags & XFS_IOLOCK_EXCL) {
		if (!down_write_trylock(&VFS_I(ip)->i_rwsem))
			goto out;
	} else if (lock_flags & XFS_IOLOCK_SHARED) {
		if (!down_read_trylock(&VFS_I(ip)->i_rwsem))
			goto out;
	}

	if (lock_flags & XFS_MMAPLOCK_EXCL) {
		if (!down_write_trylock(&VFS_I(ip)->i_mapping->invalidate_lock))
			goto out_undo_iolock;
	} else if (lock_flags & XFS_MMAPLOCK_SHARED) {
		if (!down_read_trylock(&VFS_I(ip)->i_mapping->invalidate_lock))
			goto out_undo_iolock;
	}

	if (lock_flags & XFS_ILOCK_EXCL) {
		if (!mrtryupdate(&ip->i_lock))
			goto out_undo_mmaplock;
	} else if (lock_flags & XFS_ILOCK_SHARED) {
		if (!mrtryaccess(&ip->i_lock))
			goto out_undo_mmaplock;
	}
	return 1;

out_undo_mmaplock:
	if (lock_flags & XFS_MMAPLOCK_EXCL)
		up_write(&VFS_I(ip)->i_mapping->invalidate_lock);
	else if (lock_flags & XFS_MMAPLOCK_SHARED)
		up_read(&VFS_I(ip)->i_mapping->invalidate_lock);
out_undo_iolock:
	if (lock_flags & XFS_IOLOCK_EXCL)
		up_write(&VFS_I(ip)->i_rwsem);
	else if (lock_flags & XFS_IOLOCK_SHARED)
		up_read(&VFS_I(ip)->i_rwsem);
out:
	return 0;
}

/*
 * xfs_iunlock() is used to drop the inode locks acquired with
 * xfs_ilock() and xfs_ilock_nowait().  The caller must pass
 * in the flags given to xfs_ilock() or xfs_ilock_nowait() so
 * that we know which locks to drop.
 *
 * ip -- the inode being unlocked
 * lock_flags -- this parameter indicates the inode's locks to be
 *       to be unlocked.  See the comment for xfs_ilock() for a list
 *	 of valid values for this parameter.
 *
 */
void
xfs_iunlock(
	xfs_inode_t		*ip,
	uint			lock_flags)
{
	/*
	 * You can't set both SHARED and EXCL for the same lock,
	 * and only XFS_IOLOCK_SHARED, XFS_IOLOCK_EXCL, XFS_ILOCK_SHARED,
	 * and XFS_ILOCK_EXCL are valid values to set in lock_flags.
	 */
	ASSERT((lock_flags & (XFS_IOLOCK_SHARED | XFS_IOLOCK_EXCL)) !=
	       (XFS_IOLOCK_SHARED | XFS_IOLOCK_EXCL));
	ASSERT((lock_flags & (XFS_MMAPLOCK_SHARED | XFS_MMAPLOCK_EXCL)) !=
	       (XFS_MMAPLOCK_SHARED | XFS_MMAPLOCK_EXCL));
	ASSERT((lock_flags & (XFS_ILOCK_SHARED | XFS_ILOCK_EXCL)) !=
	       (XFS_ILOCK_SHARED | XFS_ILOCK_EXCL));
	ASSERT((lock_flags & ~(XFS_LOCK_MASK | XFS_LOCK_SUBCLASS_MASK)) == 0);
	ASSERT(lock_flags != 0);

	if (lock_flags & XFS_IOLOCK_EXCL)
		up_write(&VFS_I(ip)->i_rwsem);
	else if (lock_flags & XFS_IOLOCK_SHARED)
		up_read(&VFS_I(ip)->i_rwsem);

	if (lock_flags & XFS_MMAPLOCK_EXCL)
		up_write(&VFS_I(ip)->i_mapping->invalidate_lock);
	else if (lock_flags & XFS_MMAPLOCK_SHARED)
		up_read(&VFS_I(ip)->i_mapping->invalidate_lock);

	if (lock_flags & XFS_ILOCK_EXCL)
		mrunlock_excl(&ip->i_lock);
	else if (lock_flags & XFS_ILOCK_SHARED)
		mrunlock_shared(&ip->i_lock);

	trace_xfs_iunlock(ip, lock_flags, _RET_IP_);
}

/*
 * give up write locks.  the i/o lock cannot be held nested
 * if it is being demoted.
 */
void
xfs_ilock_demote(
	xfs_inode_t		*ip,
	uint			lock_flags)
{
	ASSERT(lock_flags & (XFS_IOLOCK_EXCL|XFS_MMAPLOCK_EXCL|XFS_ILOCK_EXCL));
	ASSERT((lock_flags &
		~(XFS_IOLOCK_EXCL|XFS_MMAPLOCK_EXCL|XFS_ILOCK_EXCL)) == 0);

	if (lock_flags & XFS_ILOCK_EXCL)
		mrdemote(&ip->i_lock);
	if (lock_flags & XFS_MMAPLOCK_EXCL)
		downgrade_write(&VFS_I(ip)->i_mapping->invalidate_lock);
	if (lock_flags & XFS_IOLOCK_EXCL)
		downgrade_write(&VFS_I(ip)->i_rwsem);

	trace_xfs_ilock_demote(ip, lock_flags, _RET_IP_);
}

#if defined(DEBUG) || defined(XFS_WARN)
static inline bool
__xfs_rwsem_islocked(
	struct rw_semaphore	*rwsem,
	bool			shared)
{
	if (!debug_locks)
		return rwsem_is_locked(rwsem);

	if (!shared)
		return lockdep_is_held_type(rwsem, 0);

	/*
	 * We are checking that the lock is held at least in shared
	 * mode but don't care that it might be held exclusively
	 * (i.e. shared | excl). Hence we check if the lock is held
	 * in any mode rather than an explicit shared mode.
	 */
	return lockdep_is_held_type(rwsem, -1);
}

bool
xfs_isilocked(
	struct xfs_inode	*ip,
	uint			lock_flags)
{
	if (lock_flags & (XFS_ILOCK_EXCL|XFS_ILOCK_SHARED)) {
		if (!(lock_flags & XFS_ILOCK_SHARED))
			return !!ip->i_lock.mr_writer;
		return rwsem_is_locked(&ip->i_lock.mr_lock);
	}

	if (lock_flags & (XFS_MMAPLOCK_EXCL|XFS_MMAPLOCK_SHARED)) {
		return __xfs_rwsem_islocked(&VFS_I(ip)->i_rwsem,
				(lock_flags & XFS_IOLOCK_SHARED));
	}

	if (lock_flags & (XFS_IOLOCK_EXCL | XFS_IOLOCK_SHARED)) {
		return __xfs_rwsem_islocked(&VFS_I(ip)->i_rwsem,
				(lock_flags & XFS_IOLOCK_SHARED));
	}

	ASSERT(0);
	return false;
}
#endif

/*
 * xfs_lockdep_subclass_ok() is only used in an ASSERT, so is only called when
 * DEBUG or XFS_WARN is set. And MAX_LOCKDEP_SUBCLASSES is then only defined
 * when CONFIG_LOCKDEP is set. Hence the complex define below to avoid build
 * errors and warnings.
 */
#if (defined(DEBUG) || defined(XFS_WARN)) && defined(CONFIG_LOCKDEP)
static bool
xfs_lockdep_subclass_ok(
	int subclass)
{
	return subclass < MAX_LOCKDEP_SUBCLASSES;
}
#else
#define xfs_lockdep_subclass_ok(subclass)	(true)
#endif

/*
 * Bump the subclass so xfs_lock_inodes() acquires each lock with a different
 * value. This can be called for any type of inode lock combination, including
 * parent locking. Care must be taken to ensure we don't overrun the subclass
 * storage fields in the class mask we build.
 */
static inline int
xfs_lock_inumorder(int lock_mode, int subclass)
{
	int	class = 0;

	ASSERT(!(lock_mode & (XFS_ILOCK_PARENT | XFS_ILOCK_RTBITMAP |
			      XFS_ILOCK_RTSUM)));
	ASSERT(xfs_lockdep_subclass_ok(subclass));

	if (lock_mode & (XFS_IOLOCK_SHARED|XFS_IOLOCK_EXCL)) {
		ASSERT(subclass <= XFS_IOLOCK_MAX_SUBCLASS);
		class += subclass << XFS_IOLOCK_SHIFT;
	}

	if (lock_mode & (XFS_MMAPLOCK_SHARED|XFS_MMAPLOCK_EXCL)) {
		ASSERT(subclass <= XFS_MMAPLOCK_MAX_SUBCLASS);
		class += subclass << XFS_MMAPLOCK_SHIFT;
	}

	if (lock_mode & (XFS_ILOCK_SHARED|XFS_ILOCK_EXCL)) {
		ASSERT(subclass <= XFS_ILOCK_MAX_SUBCLASS);
		class += subclass << XFS_ILOCK_SHIFT;
	}

	return (lock_mode & ~XFS_LOCK_SUBCLASS_MASK) | class;
}

/*
 * The following routine will lock n inodes in exclusive mode.  We assume the
 * caller calls us with the inodes in i_ino order.
 *
 * We need to detect deadlock where an inode that we lock is in the AIL and we
 * start waiting for another inode that is locked by a thread in a long running
 * transaction (such as truncate). This can result in deadlock since the long
 * running trans might need to wait for the inode we just locked in order to
 * push the tail and free space in the log.
 *
 * xfs_lock_inodes() can only be used to lock one type of lock at a time -
 * the iolock, the mmaplock or the ilock, but not more than one at a time. If we
 * lock more than one at a time, lockdep will report false positives saying we
 * have violated locking orders.
 */
static void
xfs_lock_inodes(
	struct xfs_inode	**ips,
	int			inodes,
	uint			lock_mode)
{
	int			attempts = 0, i, j, try_lock;
	struct xfs_log_item	*lp;

	/*
	 * Currently supports between 2 and 5 inodes with exclusive locking.  We
	 * support an arbitrary depth of locking here, but absolute limits on
	 * inodes depend on the type of locking and the limits placed by
	 * lockdep annotations in xfs_lock_inumorder.  These are all checked by
	 * the asserts.
	 */
	ASSERT(ips && inodes >= 2 && inodes <= 5);
	ASSERT(lock_mode & (XFS_IOLOCK_EXCL | XFS_MMAPLOCK_EXCL |
			    XFS_ILOCK_EXCL));
	ASSERT(!(lock_mode & (XFS_IOLOCK_SHARED | XFS_MMAPLOCK_SHARED |
			      XFS_ILOCK_SHARED)));
	ASSERT(!(lock_mode & XFS_MMAPLOCK_EXCL) ||
		inodes <= XFS_MMAPLOCK_MAX_SUBCLASS + 1);
	ASSERT(!(lock_mode & XFS_ILOCK_EXCL) ||
		inodes <= XFS_ILOCK_MAX_SUBCLASS + 1);

	if (lock_mode & XFS_IOLOCK_EXCL) {
		ASSERT(!(lock_mode & (XFS_MMAPLOCK_EXCL | XFS_ILOCK_EXCL)));
	} else if (lock_mode & XFS_MMAPLOCK_EXCL)
		ASSERT(!(lock_mode & XFS_ILOCK_EXCL));

	try_lock = 0;
	i = 0;
again:
	for (; i < inodes; i++) {
		ASSERT(ips[i]);

		if (i && (ips[i] == ips[i - 1]))	/* Already locked */
			continue;

		/*
		 * If try_lock is not set yet, make sure all locked inodes are
		 * not in the AIL.  If any are, set try_lock to be used later.
		 */
		if (!try_lock) {
			for (j = (i - 1); j >= 0 && !try_lock; j--) {
				lp = &ips[j]->i_itemp->ili_item;
				if (lp && test_bit(XFS_LI_IN_AIL, &lp->li_flags))
					try_lock++;
			}
		}

		/*
		 * If any of the previous locks we have locked is in the AIL,
		 * we must TRY to get the second and subsequent locks. If
		 * we can't get any, we must release all we have
		 * and try again.
		 */
		if (!try_lock) {
			xfs_ilock(ips[i], xfs_lock_inumorder(lock_mode, i));
			continue;
		}

		/* try_lock means we have an inode locked that is in the AIL. */
		ASSERT(i != 0);
		if (xfs_ilock_nowait(ips[i], xfs_lock_inumorder(lock_mode, i)))
			continue;

		/*
		 * Unlock all previous guys and try again.  xfs_iunlock will try
		 * to push the tail if the inode is in the AIL.
		 */
		attempts++;
		for (j = i - 1; j >= 0; j--) {
			/*
			 * Check to see if we've already unlocked this one.  Not
			 * the first one going back, and the inode ptr is the
			 * same.
			 */
			if (j != (i - 1) && ips[j] == ips[j + 1])
				continue;

			xfs_iunlock(ips[j], lock_mode);
		}

		if ((attempts % 5) == 0) {
			delay(1); /* Don't just spin the CPU */
		}
		i = 0;
		try_lock = 0;
		goto again;
	}
}

/*
 * xfs_lock_two_inodes() can only be used to lock ilock. The iolock and
 * mmaplock must be double-locked separately since we use i_rwsem and
 * invalidate_lock for that. We now support taking one lock EXCL and the
 * other SHARED.
 */
void
xfs_lock_two_inodes(
	struct xfs_inode	*ip0,
	uint			ip0_mode,
	struct xfs_inode	*ip1,
	uint			ip1_mode)
{
	struct xfs_inode	*temp;
	uint			mode_temp;
	int			attempts = 0;
	struct xfs_log_item	*lp;

	ASSERT(hweight32(ip0_mode) == 1);
	ASSERT(hweight32(ip1_mode) == 1);
	ASSERT(!(ip0_mode & (XFS_IOLOCK_SHARED|XFS_IOLOCK_EXCL)));
	ASSERT(!(ip1_mode & (XFS_IOLOCK_SHARED|XFS_IOLOCK_EXCL)));
	ASSERT(!(ip0_mode & (XFS_MMAPLOCK_SHARED|XFS_MMAPLOCK_EXCL)));
	ASSERT(!(ip1_mode & (XFS_MMAPLOCK_SHARED|XFS_MMAPLOCK_EXCL)));
	ASSERT(ip0->i_ino != ip1->i_ino);

	if (ip0->i_ino > ip1->i_ino) {
		temp = ip0;
		ip0 = ip1;
		ip1 = temp;
		mode_temp = ip0_mode;
		ip0_mode = ip1_mode;
		ip1_mode = mode_temp;
	}

 again:
	xfs_ilock(ip0, xfs_lock_inumorder(ip0_mode, 0));

	/*
	 * If the first lock we have locked is in the AIL, we must TRY to get
	 * the second lock. If we can't get it, we must release the first one
	 * and try again.
	 */
	lp = &ip0->i_itemp->ili_item;
	if (lp && test_bit(XFS_LI_IN_AIL, &lp->li_flags)) {
		if (!xfs_ilock_nowait(ip1, xfs_lock_inumorder(ip1_mode, 1))) {
			xfs_iunlock(ip0, ip0_mode);
			if ((++attempts % 5) == 0)
				delay(1); /* Don't just spin the CPU */
			goto again;
		}
	} else {
		xfs_ilock(ip1, xfs_lock_inumorder(ip1_mode, 1));
	}
}

uint
xfs_ip2xflags(
	struct xfs_inode	*ip)
{
	uint			flags = 0;

	if (ip->i_diflags & XFS_DIFLAG_ANY) {
		if (ip->i_diflags & XFS_DIFLAG_REALTIME)
			flags |= FS_XFLAG_REALTIME;
		if (ip->i_diflags & XFS_DIFLAG_PREALLOC)
			flags |= FS_XFLAG_PREALLOC;
		if (ip->i_diflags & XFS_DIFLAG_IMMUTABLE)
			flags |= FS_XFLAG_IMMUTABLE;
		if (ip->i_diflags & XFS_DIFLAG_APPEND)
			flags |= FS_XFLAG_APPEND;
		if (ip->i_diflags & XFS_DIFLAG_SYNC)
			flags |= FS_XFLAG_SYNC;
		if (ip->i_diflags & XFS_DIFLAG_NOATIME)
			flags |= FS_XFLAG_NOATIME;
		if (ip->i_diflags & XFS_DIFLAG_NODUMP)
			flags |= FS_XFLAG_NODUMP;
		if (ip->i_diflags & XFS_DIFLAG_RTINHERIT)
			flags |= FS_XFLAG_RTINHERIT;
		if (ip->i_diflags & XFS_DIFLAG_PROJINHERIT)
			flags |= FS_XFLAG_PROJINHERIT;
		if (ip->i_diflags & XFS_DIFLAG_NOSYMLINKS)
			flags |= FS_XFLAG_NOSYMLINKS;
		if (ip->i_diflags & XFS_DIFLAG_EXTSIZE)
			flags |= FS_XFLAG_EXTSIZE;
		if (ip->i_diflags & XFS_DIFLAG_EXTSZINHERIT)
			flags |= FS_XFLAG_EXTSZINHERIT;
		if (ip->i_diflags & XFS_DIFLAG_NODEFRAG)
			flags |= FS_XFLAG_NODEFRAG;
		if (ip->i_diflags & XFS_DIFLAG_FILESTREAM)
			flags |= FS_XFLAG_FILESTREAM;
	}

	if (ip->i_diflags2 & XFS_DIFLAG2_ANY) {
		if (ip->i_diflags2 & XFS_DIFLAG2_DAX)
			flags |= FS_XFLAG_DAX;
		if (ip->i_diflags2 & XFS_DIFLAG2_COWEXTSIZE)
			flags |= FS_XFLAG_COWEXTSIZE;
	}

	if (XFS_IFORK_Q(ip))
		flags |= FS_XFLAG_HASATTR;
	return flags;
}

/*
 * Lookups up an inode from "name". If ci_name is not NULL, then a CI match
 * is allowed, otherwise it has to be an exact match. If a CI match is found,
 * ci_name->name will point to a the actual name (caller must free) or
 * will be set to NULL if an exact match is found.
 */
int
xfs_lookup(
	xfs_inode_t		*dp,
	struct xfs_name		*name,
	xfs_inode_t		**ipp,
	struct xfs_name		*ci_name)
{
	xfs_ino_t		inum;
	int			error;

	trace_xfs_lookup(dp, name);

	if (xfs_is_shutdown(dp->i_mount))
		return -EIO;

	error = xfs_dir_lookup(NULL, dp, name, &inum, ci_name);
	if (error)
		goto out_unlock;

	error = xfs_iget(dp->i_mount, NULL, inum, 0, 0, ipp);
	if (error)
		goto out_free_name;

	return 0;

out_free_name:
	if (ci_name)
		kmem_free(ci_name->name);
out_unlock:
	*ipp = NULL;
	return error;
}

/* Propagate di_flags from a parent inode to a child inode. */
static void
xfs_inode_inherit_flags(
	struct xfs_inode	*ip,
	const struct xfs_inode	*pip)
{
	unsigned int		di_flags = 0;
	xfs_failaddr_t		failaddr;
	umode_t			mode = VFS_I(ip)->i_mode;

	if (S_ISDIR(mode)) {
		if (pip->i_diflags & XFS_DIFLAG_RTINHERIT)
			di_flags |= XFS_DIFLAG_RTINHERIT;
		if (pip->i_diflags & XFS_DIFLAG_EXTSZINHERIT) {
			di_flags |= XFS_DIFLAG_EXTSZINHERIT;
			ip->i_extsize = pip->i_extsize;
		}
		if (pip->i_diflags & XFS_DIFLAG_PROJINHERIT)
			di_flags |= XFS_DIFLAG_PROJINHERIT;
	} else if (S_ISREG(mode)) {
		if ((pip->i_diflags & XFS_DIFLAG_RTINHERIT) &&
		    xfs_has_realtime(ip->i_mount))
			di_flags |= XFS_DIFLAG_REALTIME;
		if (pip->i_diflags & XFS_DIFLAG_EXTSZINHERIT) {
			di_flags |= XFS_DIFLAG_EXTSIZE;
			ip->i_extsize = pip->i_extsize;
		}
	}
	if ((pip->i_diflags & XFS_DIFLAG_NOATIME) &&
	    xfs_inherit_noatime)
		di_flags |= XFS_DIFLAG_NOATIME;
	if ((pip->i_diflags & XFS_DIFLAG_NODUMP) &&
	    xfs_inherit_nodump)
		di_flags |= XFS_DIFLAG_NODUMP;
	if ((pip->i_diflags & XFS_DIFLAG_SYNC) &&
	    xfs_inherit_sync)
		di_flags |= XFS_DIFLAG_SYNC;
	if ((pip->i_diflags & XFS_DIFLAG_NOSYMLINKS) &&
	    xfs_inherit_nosymlinks)
		di_flags |= XFS_DIFLAG_NOSYMLINKS;
	if ((pip->i_diflags & XFS_DIFLAG_NODEFRAG) &&
	    xfs_inherit_nodefrag)
		di_flags |= XFS_DIFLAG_NODEFRAG;
	if (pip->i_diflags & XFS_DIFLAG_FILESTREAM)
		di_flags |= XFS_DIFLAG_FILESTREAM;

	ip->i_diflags |= di_flags;

	/*
	 * Inode verifiers on older kernels only check that the extent size
	 * hint is an integer multiple of the rt extent size on realtime files.
	 * They did not check the hint alignment on a directory with both
	 * rtinherit and extszinherit flags set.  If the misaligned hint is
	 * propagated from a directory into a new realtime file, new file
	 * allocations will fail due to math errors in the rt allocator and/or
	 * trip the verifiers.  Validate the hint settings in the new file so
	 * that we don't let broken hints propagate.
	 */
	failaddr = xfs_inode_validate_extsize(ip->i_mount, ip->i_extsize,
			VFS_I(ip)->i_mode, ip->i_diflags);
	if (failaddr) {
		ip->i_diflags &= ~(XFS_DIFLAG_EXTSIZE |
				   XFS_DIFLAG_EXTSZINHERIT);
		ip->i_extsize = 0;
	}
}

/* Propagate di_flags2 from a parent inode to a child inode. */
static void
xfs_inode_inherit_flags2(
	struct xfs_inode	*ip,
	const struct xfs_inode	*pip)
{
	xfs_failaddr_t		failaddr;

	if (pip->i_diflags2 & XFS_DIFLAG2_COWEXTSIZE) {
		ip->i_diflags2 |= XFS_DIFLAG2_COWEXTSIZE;
		ip->i_cowextsize = pip->i_cowextsize;
	}
	if (pip->i_diflags2 & XFS_DIFLAG2_DAX)
		ip->i_diflags2 |= XFS_DIFLAG2_DAX;

	/* Don't let invalid cowextsize hints propagate. */
	failaddr = xfs_inode_validate_cowextsize(ip->i_mount, ip->i_cowextsize,
			VFS_I(ip)->i_mode, ip->i_diflags, ip->i_diflags2);
	if (failaddr) {
		ip->i_diflags2 &= ~XFS_DIFLAG2_COWEXTSIZE;
		ip->i_cowextsize = 0;
	}
}

/*
 * Initialise a newly allocated inode and return the in-core inode to the
 * caller locked exclusively.
 */
int
xfs_init_new_inode(
	struct user_namespace	*mnt_userns,
	struct xfs_trans	*tp,
	struct xfs_inode	*pip,
	xfs_ino_t		ino,
	umode_t			mode,
	xfs_nlink_t		nlink,
	dev_t			rdev,
	prid_t			prid,
	bool			init_xattrs,
	struct xfs_inode	**ipp)
{
	struct inode		*dir = pip ? VFS_I(pip) : NULL;
	struct xfs_mount	*mp = tp->t_mountp;
	struct xfs_inode	*ip;
	unsigned int		flags;
	int			error;
	struct timespec64	tv;
	struct inode		*inode;

	/*
	 * Protect against obviously corrupt allocation btree records. Later
	 * xfs_iget checks will catch re-allocation of other active in-memory
	 * and on-disk inodes. If we don't catch reallocating the parent inode
	 * here we will deadlock in xfs_iget() so we have to do these checks
	 * first.
	 */
	if ((pip && ino == pip->i_ino) || !xfs_verify_dir_ino(mp, ino)) {
		xfs_alert(mp, "Allocated a known in-use inode 0x%llx!", ino);
		return -EFSCORRUPTED;
	}

	/*
	 * Get the in-core inode with the lock held exclusively to prevent
	 * others from looking at until we're done.
	 */
	error = xfs_iget(mp, tp, ino, XFS_IGET_CREATE, XFS_ILOCK_EXCL, &ip);
	if (error)
		return error;

	ASSERT(ip != NULL);
	inode = VFS_I(ip);
	set_nlink(inode, nlink);
	inode->i_rdev = rdev;
	ip->i_projid = prid;

	if (dir && !(dir->i_mode & S_ISGID) && xfs_has_grpid(mp)) {
		inode_fsuid_set(inode, mnt_userns);
		inode->i_gid = dir->i_gid;
		inode->i_mode = mode;
	} else {
		inode_init_owner(mnt_userns, inode, dir, mode);
	}

	/*
	 * 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 &&
	    (inode->i_mode & S_ISGID) &&
	    !in_group_p(i_gid_into_mnt(mnt_userns, inode)))
		inode->i_mode &= ~S_ISGID;

	ip->i_disk_size = 0;
	ip->i_df.if_nextents = 0;
	ASSERT(ip->i_nblocks == 0);

	tv = current_time(inode);
	inode->i_mtime = tv;
	inode->i_atime = tv;
	inode->i_ctime = tv;

	ip->i_extsize = 0;
	ip->i_diflags = 0;

	if (xfs_has_v3inodes(mp)) {
		inode_set_iversion(inode, 1);
		ip->i_cowextsize = 0;
		ip->i_crtime = tv;
	}

	flags = XFS_ILOG_CORE;
	switch (mode & S_IFMT) {
	case S_IFIFO:
	case S_IFCHR:
	case S_IFBLK:
	case S_IFSOCK:
		ip->i_df.if_format = XFS_DINODE_FMT_DEV;
		flags |= XFS_ILOG_DEV;
		break;
	case S_IFREG:
	case S_IFDIR:
		if (pip && (pip->i_diflags & XFS_DIFLAG_ANY))
			xfs_inode_inherit_flags(ip, pip);
		if (pip && (pip->i_diflags2 & XFS_DIFLAG2_ANY))
			xfs_inode_inherit_flags2(ip, pip);
		fallthrough;
	case S_IFLNK:
		ip->i_df.if_format = XFS_DINODE_FMT_EXTENTS;
		ip->i_df.if_bytes = 0;
		ip->i_df.if_u1.if_root = NULL;
		break;
	default:
		ASSERT(0);
	}

	/*
	 * If we need to create attributes immediately after allocating the
	 * inode, initialise an empty attribute fork right now. We use the
	 * default fork offset for attributes here as we don't know exactly what
	 * size or how many attributes we might be adding. We can do this
	 * safely here because we know the data fork is completely empty and
	 * this saves us from needing to run a separate transaction to set the
	 * fork offset in the immediate future.
	 */
	if (init_xattrs && xfs_has_attr(mp)) {
		ip->i_forkoff = xfs_default_attroffset(ip) >> 3;
		ip->i_afp = xfs_ifork_alloc(XFS_DINODE_FMT_EXTENTS, 0);
	}

	/*
	 * Log the new values stuffed into the inode.
	 */
	xfs_trans_ijoin(tp, ip, XFS_ILOCK_EXCL);
	xfs_trans_log_inode(tp, ip, flags);

	/* now that we have an i_mode we can setup the inode structure */
	xfs_setup_inode(ip);

	*ipp = ip;
	return 0;
}

/*
 * Decrement the link count on an inode & log the change.  If this causes the
 * link count to go to zero, move the inode to AGI unlinked list so that it can
 * be freed when the last active reference goes away via xfs_inactive().
 */
static int			/* error */
xfs_droplink(
	xfs_trans_t *tp,
	xfs_inode_t *ip)
{
	xfs_trans_ichgtime(tp, ip, XFS_ICHGTIME_CHG);

	drop_nlink(VFS_I(ip));
	xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);

	if (VFS_I(ip)->i_nlink)
		return 0;

	return xfs_iunlink(tp, ip);
}

/*
 * Increment the link count on an inode & log the change.
 */
static void
xfs_bumplink(
	xfs_trans_t *tp,
	xfs_inode_t *ip)
{
	xfs_trans_ichgtime(tp, ip, XFS_ICHGTIME_CHG);

	inc_nlink(VFS_I(ip));
	xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
}

int
xfs_create(
	struct user_namespace	*mnt_userns,
	xfs_inode_t		*dp,
	struct xfs_name		*name,
	umode_t			mode,
	dev_t			rdev,
	bool			init_xattrs,
	xfs_inode_t		**ipp)
{
	int			is_dir = S_ISDIR(mode);
	struct xfs_mount	*mp = dp->i_mount;
	struct xfs_inode	*ip = NULL;
	struct xfs_trans	*tp = NULL;
	int			error;
	bool                    unlock_dp_on_error = false;
	prid_t			prid;
	struct xfs_dquot	*udqp = NULL;
	struct xfs_dquot	*gdqp = NULL;
	struct xfs_dquot	*pdqp = NULL;
	struct xfs_trans_res	*tres;
	uint			resblks;
	xfs_ino_t		ino;

	trace_xfs_create(dp, name);

	if (xfs_is_shutdown(mp))
		return -EIO;

	prid = xfs_get_initial_prid(dp);

	/*
	 * Make sure that we have allocated dquot(s) on disk.
	 */
	error = xfs_qm_vop_dqalloc(dp, mapped_fsuid(mnt_userns),
			mapped_fsgid(mnt_userns), prid,
			XFS_QMOPT_QUOTALL | XFS_QMOPT_INHERIT,
			&udqp, &gdqp, &pdqp);
	if (error)
		return error;

	if (is_dir) {
		resblks = XFS_MKDIR_SPACE_RES(mp, name->len);
		tres = &M_RES(mp)->tr_mkdir;
	} else {
		resblks = XFS_CREATE_SPACE_RES(mp, name->len);
		tres = &M_RES(mp)->tr_create;
	}

	/*
	 * Initially assume that the file does not exist and
	 * reserve the resources for that case.  If that is not
	 * the case we'll drop the one we have and get a more
	 * appropriate transaction later.
	 */
	error = xfs_trans_alloc_icreate(mp, tres, udqp, gdqp, pdqp, resblks,
			&tp);
	if (error == -ENOSPC) {
		/* flush outstanding delalloc blocks and retry */
		xfs_flush_inodes(mp);
		error = xfs_trans_alloc_icreate(mp, tres, udqp, gdqp, pdqp,
				resblks, &tp);
	}
	if (error)
		goto out_release_dquots;

	xfs_ilock(dp, XFS_ILOCK_EXCL | XFS_ILOCK_PARENT);
	unlock_dp_on_error = true;

	error = xfs_iext_count_may_overflow(dp, XFS_DATA_FORK,
			XFS_IEXT_DIR_MANIP_CNT(mp));
	if (error)
		goto out_trans_cancel;

	/*
	 * A newly created regular or special file just has one directory
	 * entry pointing to them, but a directory also the "." entry
	 * pointing to itself.
	 */
	error = xfs_dialloc(&tp, dp->i_ino, mode, &ino);
	if (!error)
		error = xfs_init_new_inode(mnt_userns, tp, dp, ino, mode,
				is_dir ? 2 : 1, rdev, prid, init_xattrs, &ip);
	if (error)
		goto out_trans_cancel;

	/*
	 * Now we join the directory inode to the transaction.  We do not do it
	 * earlier because xfs_dialloc might commit the previous transaction
	 * (and release all the locks).  An error from here on will result in
	 * the transaction cancel unlocking dp so don't do it explicitly in the
	 * error path.
	 */
	xfs_trans_ijoin(tp, dp, XFS_ILOCK_EXCL);
	unlock_dp_on_error = false;

	error = xfs_dir_createname(tp, dp, name, ip->i_ino,
					resblks - XFS_IALLOC_SPACE_RES(mp));
	if (error) {
		ASSERT(error != -ENOSPC);
		goto out_trans_cancel;
	}
	xfs_trans_ichgtime(tp, dp, XFS_ICHGTIME_MOD | XFS_ICHGTIME_CHG);
	xfs_trans_log_inode(tp, dp, XFS_ILOG_CORE);

	if (is_dir) {
		error = xfs_dir_init(tp, ip, dp);
		if (error)
			goto out_trans_cancel;

		xfs_bumplink(tp, dp);
	}

	/*
	 * If this is a synchronous mount, make sure that the
	 * create transaction goes to disk before returning to
	 * the user.
	 */
	if (xfs_has_wsync(mp) || xfs_has_dirsync(mp))
		xfs_trans_set_sync(tp);

	/*
	 * Attach the dquot(s) to the inodes and modify them incore.
	 * These ids of the inode couldn't have changed since the new
	 * inode has been locked ever since it was created.
	 */
	xfs_qm_vop_create_dqattach(tp, ip, udqp, gdqp, pdqp);

	error = xfs_trans_commit(tp);
	if (error)
		goto out_release_inode;

	xfs_qm_dqrele(udqp);
	xfs_qm_dqrele(gdqp);
	xfs_qm_dqrele(pdqp);

	*ipp = ip;
	return 0;

 out_trans_cancel:
	xfs_trans_cancel(tp);
 out_release_inode:
	/*
	 * Wait until after the current transaction is aborted to finish the
	 * setup of the inode and release the inode.  This prevents recursive
	 * transactions and deadlocks from xfs_inactive.
	 */
	if (ip) {
		xfs_finish_inode_setup(ip);
		xfs_irele(ip);
	}
 out_release_dquots:
	xfs_qm_dqrele(udqp);
	xfs_qm_dqrele(gdqp);
	xfs_qm_dqrele(pdqp);

	if (unlock_dp_on_error)
		xfs_iunlock(dp, XFS_ILOCK_EXCL);
	return error;
}

int
xfs_create_tmpfile(
	struct user_namespace	*mnt_userns,
	struct xfs_inode	*dp,
	umode_t			mode,
	struct xfs_inode	**ipp)
{
	struct xfs_mount	*mp = dp->i_mount;
	struct xfs_inode	*ip = NULL;
	struct xfs_trans	*tp = NULL;
	int			error;
	prid_t                  prid;
	struct xfs_dquot	*udqp = NULL;
	struct xfs_dquot	*gdqp = NULL;
	struct xfs_dquot	*pdqp = NULL;
	struct xfs_trans_res	*tres;
	uint			resblks;
	xfs_ino_t		ino;

	if (xfs_is_shutdown(mp))
		return -EIO;

	prid = xfs_get_initial_prid(dp);

	/*
	 * Make sure that we have allocated dquot(s) on disk.
	 */
	error = xfs_qm_vop_dqalloc(dp, mapped_fsuid(mnt_userns),
			mapped_fsgid(mnt_userns), prid,
			XFS_QMOPT_QUOTALL | XFS_QMOPT_INHERIT,
			&udqp, &gdqp, &pdqp);
	if (error)
		return error;

	resblks = XFS_IALLOC_SPACE_RES(mp);
	tres = &M_RES(mp)->tr_create_tmpfile;

	error = xfs_trans_alloc_icreate(mp, tres, udqp, gdqp, pdqp, resblks,
			&tp);
	if (error)
		goto out_release_dquots;

	error = xfs_dialloc(&tp, dp->i_ino, mode, &ino);
	if (!error)
		error = xfs_init_new_inode(mnt_userns, tp, dp, ino, mode,
				0, 0, prid, false, &ip);
	if (error)
		goto out_trans_cancel;

	if (xfs_has_wsync(mp))
		xfs_trans_set_sync(tp);

	/*
	 * Attach the dquot(s) to the inodes and modify them incore.
	 * These ids of the inode couldn't have changed since the new
	 * inode has been locked ever since it was created.
	 */
	xfs_qm_vop_create_dqattach(tp, ip, udqp, gdqp, pdqp);

	error = xfs_iunlink(tp, ip);
	if (error)
		goto out_trans_cancel;

	error = xfs_trans_commit(tp);
	if (error)
		goto out_release_inode;

	xfs_qm_dqrele(udqp);
	xfs_qm_dqrele(gdqp);
	xfs_qm_dqrele(pdqp);

	*ipp = ip;
	return 0;

 out_trans_cancel:
	xfs_trans_cancel(tp);
 out_release_inode:
	/*
	 * Wait until after the current transaction is aborted to finish the
	 * setup of the inode and release the inode.  This prevents recursive
	 * transactions and deadlocks from xfs_inactive.
	 */
	if (ip) {
		xfs_finish_inode_setup(ip);
		xfs_irele(ip);
	}
 out_release_dquots:
	xfs_qm_dqrele(udqp);
	xfs_qm_dqrele(gdqp);
	xfs_qm_dqrele(pdqp);

	return error;
}

int
xfs_link(
	xfs_inode_t		*tdp,
	xfs_inode_t		*sip,
	struct xfs_name		*target_name)
{
	xfs_mount_t		*mp = tdp->i_mount;
	xfs_trans_t		*tp;
	int			error;
	int			resblks;

	trace_xfs_link(tdp, target_name);

	ASSERT(!S_ISDIR(VFS_I(sip)->i_mode));

	if (xfs_is_shutdown(mp))
		return -EIO;

	error = xfs_qm_dqattach(sip);
	if (error)
		goto std_return;

	error = xfs_qm_dqattach(tdp);
	if (error)
		goto std_return;

	resblks = XFS_LINK_SPACE_RES(mp, target_name->len);
	error = xfs_trans_alloc(mp, &M_RES(mp)->tr_link, resblks, 0, 0, &tp);
	if (error == -ENOSPC) {
		resblks = 0;
		error = xfs_trans_alloc(mp, &M_RES(mp)->tr_link, 0, 0, 0, &tp);
	}
	if (error)
		goto std_return;

	xfs_lock_two_inodes(sip, XFS_ILOCK_EXCL, tdp, XFS_ILOCK_EXCL);

	xfs_trans_ijoin(tp, sip, XFS_ILOCK_EXCL);
	xfs_trans_ijoin(tp, tdp, XFS_ILOCK_EXCL);

	error = xfs_iext_count_may_overflow(tdp, XFS_DATA_FORK,
			XFS_IEXT_DIR_MANIP_CNT(mp));
	if (error)
		goto error_return;

	/*
	 * If we are using project inheritance, we only allow hard link
	 * creation in our tree when the project IDs are the same; else
	 * the tree quota mechanism could be circumvented.
	 */
	if (unlikely((tdp->i_diflags & XFS_DIFLAG_PROJINHERIT) &&
		     tdp->i_projid != sip->i_projid)) {
		error = -EXDEV;
		goto error_return;
	}

	if (!resblks) {
		error = xfs_dir_canenter(tp, tdp, target_name);
		if (error)
			goto error_return;
	}

	/*
	 * Handle initial link state of O_TMPFILE inode
	 */
	if (VFS_I(sip)->i_nlink == 0) {
		struct xfs_perag	*pag;

		pag = xfs_perag_get(mp, XFS_INO_TO_AGNO(mp, sip->i_ino));
		error = xfs_iunlink_remove(tp, pag, sip);
		xfs_perag_put(pag);
		if (error)
			goto error_return;
	}

	error = xfs_dir_createname(tp, tdp, target_name, sip->i_ino,
				   resblks);
	if (error)
		goto error_return;
	xfs_trans_ichgtime(tp, tdp, XFS_ICHGTIME_MOD | XFS_ICHGTIME_CHG);
	xfs_trans_log_inode(tp, tdp, XFS_ILOG_CORE);

	xfs_bumplink(tp, sip);

	/*
	 * If this is a synchronous mount, make sure that the
	 * link transaction goes to disk before returning to
	 * the user.
	 */
	if (xfs_has_wsync(mp) || xfs_has_dirsync(mp))
		xfs_trans_set_sync(tp);

	return xfs_trans_commit(tp);

 error_return:
	xfs_trans_cancel(tp);
 std_return:
	return error;
}

/* Clear the reflink flag and the cowblocks tag if possible. */
static void
xfs_itruncate_clear_reflink_flags(
	struct xfs_inode	*ip)
{
	struct xfs_ifork	*dfork;
	struct xfs_ifork	*cfork;

	if (!xfs_is_reflink_inode(ip))
		return;
	dfork = XFS_IFORK_PTR(ip, XFS_DATA_FORK);
	cfork = XFS_IFORK_PTR(ip, XFS_COW_FORK);
	if (dfork->if_bytes == 0 && cfork->if_bytes == 0)
		ip->i_diflags2 &= ~XFS_DIFLAG2_REFLINK;
	if (cfork->if_bytes == 0)
		xfs_inode_clear_cowblocks_tag(ip);
}

/*
 * Free up the underlying blocks past new_size.  The new size must be smaller
 * than the current size.  This routine can be used both for the attribute and
 * data fork, and does not modify the inode size, which is left to the caller.
 *
 * 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.
 *
 * If we get an error, we must return with the inode locked and linked into the
 * current transaction. 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 on error so that transactions can be easily aborted if possible.
 */
int
xfs_itruncate_extents_flags(
	struct xfs_trans	**tpp,
	struct xfs_inode	*ip,
	int			whichfork,
	xfs_fsize_t		new_size,
	int			flags)
{
	struct xfs_mount	*mp = ip->i_mount;
	struct xfs_trans	*tp = *tpp;
	xfs_fileoff_t		first_unmap_block;
	xfs_filblks_t		unmap_len;
	int			error = 0;

	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));
	ASSERT(!atomic_read(&VFS_I(ip)->i_count) ||
	       xfs_isilocked(ip, XFS_IOLOCK_EXCL));
	ASSERT(new_size <= XFS_ISIZE(ip));
	ASSERT(tp->t_flags & XFS_TRANS_PERM_LOG_RES);
	ASSERT(ip->i_itemp != NULL);
	ASSERT(ip->i_itemp->ili_lock_flags == 0);
	ASSERT(!XFS_NOT_DQATTACHED(mp, ip));

	trace_xfs_itruncate_extents_start(ip, new_size);

	flags |= xfs_bmapi_aflag(whichfork);

	/*
	 * 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.
	 *
	 * We have to free all the blocks to the bmbt maximum offset, even if
	 * the page cache can't scale that far.
	 */
	first_unmap_block = XFS_B_TO_FSB(mp, (xfs_ufsize_t)new_size);
	if (!xfs_verify_fileoff(mp, first_unmap_block)) {
		WARN_ON_ONCE(first_unmap_block > XFS_MAX_FILEOFF);
		return 0;
	}

	unmap_len = XFS_MAX_FILEOFF - first_unmap_block + 1;
	while (unmap_len > 0) {
		ASSERT(tp->t_firstblock == NULLFSBLOCK);
		error = __xfs_bunmapi(tp, ip, first_unmap_block, &unmap_len,
				flags, XFS_ITRUNC_MAX_EXTENTS);
		if (error)
			goto out;

		/* free the just unmapped extents */
		error = xfs_defer_finish(&tp);
		if (error)
			goto out;
	}

	if (whichfork == XFS_DATA_FORK) {
		/* Remove all pending CoW reservations. */
		error = xfs_reflink_cancel_cow_blocks(ip, &tp,
				first_unmap_block, XFS_MAX_FILEOFF, true);
		if (error)
			goto out;

		xfs_itruncate_clear_reflink_flags(ip);
	}

	/*
	 * Always re-log the inode so that our permanent transaction can keep
	 * on rolling it forward in the log.
	 */
	xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);

	trace_xfs_itruncate_extents_end(ip, new_size);

out:
	*tpp = tp;
	return error;
}

int
xfs_release(
	xfs_inode_t	*ip)
{
	xfs_mount_t	*mp = ip->i_mount;
	int		error = 0;

	if (!S_ISREG(VFS_I(ip)->i_mode) || (VFS_I(ip)->i_mode == 0))
		return 0;

	/* If this is a read-only mount, don't do this (would generate I/O) */
	if (xfs_is_readonly(mp))
		return 0;

	if (!xfs_is_shutdown(mp)) {
		int truncated;

		/*
		 * If we previously truncated this file and removed old data
		 * in the process, we want to initiate "early" writeout on
		 * the last close.  This is an attempt to combat the notorious
		 * NULL files problem which is particularly noticeable from a
		 * truncate down, buffered (re-)write (delalloc), followed by
		 * a crash.  What we are effectively doing here is
		 * significantly reducing the time window where we'd otherwise
		 * be exposed to that problem.
		 */
		truncated = xfs_iflags_test_and_clear(ip, XFS_ITRUNCATED);
		if (truncated) {
			xfs_iflags_clear(ip, XFS_IDIRTY_RELEASE);
			if (ip->i_delayed_blks > 0) {
				error = filemap_flush(VFS_I(ip)->i_mapping);
				if (error)
					return error;
			}
		}
	}

	if (VFS_I(ip)->i_nlink == 0)
		return 0;

	/*
	 * If we can't get the iolock just skip truncating the blocks past EOF
	 * because we could deadlock with the mmap_lock otherwise. We'll get
	 * another chance to drop them once the last reference to the inode is
	 * dropped, so we'll never leak blocks permanently.
	 */
	if (!xfs_ilock_nowait(ip, XFS_IOLOCK_EXCL))
		return 0;

	if (xfs_can_free_eofblocks(ip, false)) {
		/*
		 * Check if the inode is being opened, written and closed
		 * frequently and we have delayed allocation blocks outstanding
		 * (e.g. streaming writes from the NFS server), truncating the
		 * blocks past EOF will cause fragmentation to occur.
		 *
		 * In this case don't do the truncation, but we have to be
		 * careful how we detect this case. Blocks beyond EOF show up as
		 * i_delayed_blks even when the inode is clean, so we need to
		 * truncate them away first before checking for a dirty release.
		 * Hence on the first dirty close we will still remove the
		 * speculative allocation, but after that we will leave it in
		 * place.
		 */
		if (xfs_iflags_test(ip, XFS_IDIRTY_RELEASE))
			goto out_unlock;

		error = xfs_free_eofblocks(ip);
		if (error)
			goto out_unlock;

		/* delalloc blocks after truncation means it really is dirty */
		if (ip->i_delayed_blks)
			xfs_iflags_set(ip, XFS_IDIRTY_RELEASE);
	}

out_unlock:
	xfs_iunlock(ip, XFS_IOLOCK_EXCL);
	return error;
}

/*
 * xfs_inactive_truncate
 *
 * Called to perform a truncate when an inode becomes unlinked.
 */
STATIC int
xfs_inactive_truncate(
	struct xfs_inode *ip)
{
	struct xfs_mount	*mp = ip->i_mount;
	struct xfs_trans	*tp;
	int			error;

	error = xfs_trans_alloc(mp, &M_RES(mp)->tr_itruncate, 0, 0, 0, &tp);
	if (error) {
		ASSERT(xfs_is_shutdown(mp));
		return error;
	}
	xfs_ilock(ip, XFS_ILOCK_EXCL);
	xfs_trans_ijoin(tp, ip, 0);

	/*
	 * Log the inode size first to prevent stale data exposure in the event
	 * of a system crash before the truncate completes. See the related
	 * comment in xfs_vn_setattr_size() for details.
	 */
	ip->i_disk_size = 0;
	xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);

	error = xfs_itruncate_extents(&tp, ip, XFS_DATA_FORK, 0);
	if (error)
		goto error_trans_cancel;

	ASSERT(ip->i_df.if_nextents == 0);

	error = xfs_trans_commit(tp);
	if (error)
		goto error_unlock;

	xfs_iunlock(ip, XFS_ILOCK_EXCL);
	return 0;

error_trans_cancel:
	xfs_trans_cancel(tp);
error_unlock:
	xfs_iunlock(ip, XFS_ILOCK_EXCL);
	return error;
}

/*
 * xfs_inactive_ifree()
 *
 * Perform the inode free when an inode is unlinked.
 */
STATIC int
xfs_inactive_ifree(
	struct xfs_inode *ip)
{
	struct xfs_mount	*mp = ip->i_mount;
	struct xfs_trans	*tp;
	int			error;

	/*
	 * We try to use a per-AG reservation for any block needed by the finobt
	 * tree, but as the finobt feature predates the per-AG reservation
	 * support a degraded file system might not have enough space for the
	 * reservation at mount time.  In that case try to dip into the reserved
	 * pool and pray.
	 *
	 * Send a warning if the reservation does happen to fail, as the inode
	 * now remains allocated and sits on the unlinked list until the fs is
	 * repaired.
	 */
	if (unlikely(mp->m_finobt_nores)) {
		error = xfs_trans_alloc(mp, &M_RES(mp)->tr_ifree,
				XFS_IFREE_SPACE_RES(mp), 0, XFS_TRANS_RESERVE,
				&tp);
	} else {
		error = xfs_trans_alloc(mp, &M_RES(mp)->tr_ifree, 0, 0, 0, &tp);
	}
	if (error) {
		if (error == -ENOSPC) {
			xfs_warn_ratelimited(mp,
			"Failed to remove inode(s) from unlinked list. "
			"Please free space, unmount and run xfs_repair.");
		} else {
			ASSERT(xfs_is_shutdown(mp));
		}
		return error;
	}

	/*
	 * We do not hold the inode locked across the entire rolling transaction
	 * here. We only need to hold it for the first transaction that
	 * xfs_ifree() builds, which may mark the inode XFS_ISTALE if the
	 * underlying cluster buffer is freed. Relogging an XFS_ISTALE inode
	 * here breaks the relationship between cluster buffer invalidation and
	 * stale inode invalidation on cluster buffer item journal commit
	 * completion, and can result in leaving dirty stale inodes hanging
	 * around in memory.
	 *
	 * We have no need for serialising this inode operation against other
	 * operations - we freed the inode and hence reallocation is required
	 * and that will serialise on reallocating the space the deferops need
	 * to free. Hence we can unlock the inode on the first commit of
	 * the transaction rather than roll it right through the deferops. This
	 * avoids relogging the XFS_ISTALE inode.
	 *
	 * We check that xfs_ifree() hasn't grown an internal transaction roll
	 * by asserting that the inode is still locked when it returns.
	 */
	xfs_ilock(ip, XFS_ILOCK_EXCL);
	xfs_trans_ijoin(tp, ip, XFS_ILOCK_EXCL);

	error = xfs_ifree(tp, ip);
	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));
	if (error) {
		/*
		 * If we fail to free the inode, shut down.  The cancel
		 * might do that, we need to make sure.  Otherwise the
		 * inode might be lost for a long time or forever.
		 */
		if (!xfs_is_shutdown(mp)) {
			xfs_notice(mp, "%s: xfs_ifree returned error %d",
				__func__, error);
			xfs_force_shutdown(mp, SHUTDOWN_META_IO_ERROR);
		}
		xfs_trans_cancel(tp);
		return error;
	}

	/*
	 * Credit the quota account(s). The inode is gone.
	 */
	xfs_trans_mod_dquot_byino(tp, ip, XFS_TRANS_DQ_ICOUNT, -1);

	/*
	 * Just ignore errors at this point.  There is nothing we can do except
	 * to try to keep going. Make sure it's not a silent error.
	 */
	error = xfs_trans_commit(tp);
	if (error)
		xfs_notice(mp, "%s: xfs_trans_commit returned error %d",
			__func__, error);

	return 0;
}

/*
 * Returns true if we need to update the on-disk metadata before we can free
 * the memory used by this inode.  Updates include freeing post-eof
 * preallocations; freeing COW staging extents; and marking the inode free in
 * the inobt if it is on the unlinked list.
 */
bool
xfs_inode_needs_inactive(
	struct xfs_inode	*ip)
{
	struct xfs_mount	*mp = ip->i_mount;
	struct xfs_ifork	*cow_ifp = XFS_IFORK_PTR(ip, XFS_COW_FORK);

	/*
	 * If the inode is already free, then there can be nothing
	 * to clean up here.
	 */
	if (VFS_I(ip)->i_mode == 0)
		return false;

	/* If this is a read-only mount, don't do this (would generate I/O) */
	if (xfs_is_readonly(mp))
		return false;

	/* If the log isn't running, push inodes straight to reclaim. */
	if (xfs_is_shutdown(mp) || xfs_has_norecovery(mp))
		return false;

	/* Metadata inodes require explicit resource cleanup. */
	if (xfs_is_metadata_inode(ip))
		return false;

	/* Want to clean out the cow blocks if there are any. */
	if (cow_ifp && cow_ifp->if_bytes > 0)
		return true;

	/* Unlinked files must be freed. */
	if (VFS_I(ip)->i_nlink == 0)
		return true;

	/*
	 * This file isn't being freed, so check if there are post-eof blocks
	 * to free.  @force is true because we are evicting an inode from the
	 * cache.  Post-eof blocks must be freed, lest we end up with broken
	 * free space accounting.
	 *
	 * Note: don't bother with iolock here since lockdep complains about
	 * acquiring it in reclaim context. We have the only reference to the
	 * inode at this point anyways.
	 */
	return xfs_can_free_eofblocks(ip, true);
}

/*
 * xfs_inactive
 *
 * This is called when the vnode reference count for the vnode
 * goes to zero.  If the file has been unlinked, then it must
 * now be truncated.  Also, we clear all of the read-ahead state
 * kept for the inode here since the file is now closed.
 */
void
xfs_inactive(
	xfs_inode_t	*ip)
{
	struct xfs_mount	*mp;
	int			error;
	int			truncate = 0;

	/*
	 * If the inode is already free, then there can be nothing
	 * to clean up here.
	 */
	if (VFS_I(ip)->i_mode == 0) {
		ASSERT(ip->i_df.if_broot_bytes == 0);
		goto out;
	}

	mp = ip->i_mount;
	ASSERT(!xfs_iflags_test(ip, XFS_IRECOVERY));

	/* If this is a read-only mount, don't do this (would generate I/O) */
	if (xfs_is_readonly(mp))
		goto out;

	/* Metadata inodes require explicit resource cleanup. */
	if (xfs_is_metadata_inode(ip))
		goto out;

	/* Try to clean out the cow blocks if there are any. */
	if (xfs_inode_has_cow_data(ip))
		xfs_reflink_cancel_cow_range(ip, 0, NULLFILEOFF, true);

	if (VFS_I(ip)->i_nlink != 0) {
		/*
		 * force is true because we are evicting an inode from the
		 * cache. Post-eof blocks must be freed, lest we end up with
		 * broken free space accounting.
		 *
		 * Note: don't bother with iolock here since lockdep complains
		 * about acquiring it in reclaim context. We have the only
		 * reference to the inode at this point anyways.
		 */
		if (xfs_can_free_eofblocks(ip, true))
			xfs_free_eofblocks(ip);

		goto out;
	}

	if (S_ISREG(VFS_I(ip)->i_mode) &&
	    (ip->i_disk_size != 0 || XFS_ISIZE(ip) != 0 ||
	     ip->i_df.if_nextents > 0 || ip->i_delayed_blks > 0))
		truncate = 1;

	error = xfs_qm_dqattach(ip);
	if (error)
		goto out;

	if (S_ISLNK(VFS_I(ip)->i_mode))
		error = xfs_inactive_symlink(ip);
	else if (truncate)
		error = xfs_inactive_truncate(ip);
	if (error)
		goto out;

	/*
	 * If there are attributes associated with the file then blow them away
	 * now.  The code calls a routine that recursively deconstructs the
	 * attribute fork. If also blows away the in-core attribute fork.
	 */
	if (XFS_IFORK_Q(ip)) {
		error = xfs_attr_inactive(ip);
		if (error)
			goto out;
	}

	ASSERT(!ip->i_afp);
	ASSERT(ip->i_forkoff == 0);

	/*
	 * Free the inode.
	 */
	xfs_inactive_ifree(ip);

out:
	/*
	 * We're done making metadata updates for this inode, so we can release
	 * the attached dquots.
	 */
	xfs_qm_dqdetach(ip);
}

/*
 * In-Core Unlinked List Lookups
 * =============================
 *
 * Every inode is supposed to be reachable from some other piece of metadata
 * with the exception of the root directory.  Inodes with a connection to a
 * file descriptor but not linked from anywhere in the on-disk directory tree
 * are collectively known as unlinked inodes, though the filesystem itself
 * maintains links to these inodes so that on-disk metadata are consistent.
 *
 * XFS implements a per-AG on-disk hash table of unlinked inodes.  The AGI
 * header contains a number of buckets that point to an inode, and each inode
 * record has a pointer to the next inode in the hash chain.  This
 * singly-linked list causes scaling problems in the iunlink remove function
 * because we must walk that list to find the inode that points to the inode
 * being removed from the unlinked hash bucket list.
 *
 * What if we modelled the unlinked list as a collection of records capturing
 * "X.next_unlinked = Y" relations?  If we indexed those records on Y, we'd
 * have a fast way to look up unlinked list predecessors, which avoids the
 * slow list walk.  That's exactly what we do here (in-core) with a per-AG
 * rhashtable.
 *
 * Because this is a backref cache, we ignore operational failures since the
 * iunlink code can fall back to the slow bucket walk.  The only errors that
 * should bubble out are for obviously incorrect situations.
 *
 * All users of the backref cache MUST hold the AGI buffer lock to serialize
 * access or have otherwise provided for concurrency control.
 */

/* Capture a "X.next_unlinked = Y" relationship. */
struct xfs_iunlink {
	struct rhash_head	iu_rhash_head;
	xfs_agino_t		iu_agino;		/* X */
	xfs_agino_t		iu_next_unlinked;	/* Y */
};

/* Unlinked list predecessor lookup hashtable construction */
static int
xfs_iunlink_obj_cmpfn(
	struct rhashtable_compare_arg	*arg,
	const void			*obj)
{
	const xfs_agino_t		*key = arg->key;
	const struct xfs_iunlink	*iu = obj;

	if (iu->iu_next_unlinked != *key)
		return 1;
	return 0;
}

static const struct rhashtable_params xfs_iunlink_hash_params = {
	.min_size		= XFS_AGI_UNLINKED_BUCKETS,
	.key_len		= sizeof(xfs_agino_t),
	.key_offset		= offsetof(struct xfs_iunlink,
					   iu_next_unlinked),
	.head_offset		= offsetof(struct xfs_iunlink, iu_rhash_head),
	.automatic_shrinking	= true,
	.obj_cmpfn		= xfs_iunlink_obj_cmpfn,
};

/*
 * Return X, where X.next_unlinked == @agino.  Returns NULLAGINO if no such
 * relation is found.
 */
static xfs_agino_t
xfs_iunlink_lookup_backref(
	struct xfs_perag	*pag,
	xfs_agino_t		agino)
{
	struct xfs_iunlink	*iu;

	iu = rhashtable_lookup_fast(&pag->pagi_unlinked_hash, &agino,
			xfs_iunlink_hash_params);
	return iu ? iu->iu_agino : NULLAGINO;
}

/*
 * Take ownership of an iunlink cache entry and insert it into the hash table.
 * If successful, the entry will be owned by the cache; if not, it is freed.
 * Either way, the caller does not own @iu after this call.
 */
static int
xfs_iunlink_insert_backref(
	struct xfs_perag	*pag,
	struct xfs_iunlink	*iu)
{
	int			error;

	error = rhashtable_insert_fast(&pag->pagi_unlinked_hash,
			&iu->iu_rhash_head, xfs_iunlink_hash_params);
	/*
	 * Fail loudly if there already was an entry because that's a sign of
	 * corruption of in-memory data.  Also fail loudly if we see an error
	 * code we didn't anticipate from the rhashtable code.  Currently we
	 * only anticipate ENOMEM.
	 */
	if (error) {
		WARN(error != -ENOMEM, "iunlink cache insert error %d", error);
		kmem_free(iu);
	}
	/*
	 * Absorb any runtime errors that aren't a result of corruption because
	 * this is a cache and we can always fall back to bucket list scanning.
	 */
	if (error != 0 && error != -EEXIST)
		error = 0;
	return error;
}

/* Remember that @prev_agino.next_unlinked = @this_agino. */
static int
xfs_iunlink_add_backref(
	struct xfs_perag	*pag,
	xfs_agino_t		prev_agino,
	xfs_agino_t		this_agino)
{
	struct xfs_iunlink	*iu;

	if (XFS_TEST_ERROR(false, pag->pag_mount, XFS_ERRTAG_IUNLINK_FALLBACK))
		return 0;

	iu = kmem_zalloc(sizeof(*iu), KM_NOFS);
	iu->iu_agino = prev_agino;
	iu->iu_next_unlinked = this_agino;

	return xfs_iunlink_insert_backref(pag, iu);
}

/*
 * Replace X.next_unlinked = @agino with X.next_unlinked = @next_unlinked.
 * If @next_unlinked is NULLAGINO, we drop the backref and exit.  If there
 * wasn't any such entry then we don't bother.
 */
static int
xfs_iunlink_change_backref(
	struct xfs_perag	*pag,
	xfs_agino_t		agino,
	xfs_agino_t		next_unlinked)
{
	struct xfs_iunlink	*iu;
	int			error;

	/* Look up the old entry; if there wasn't one then exit. */
	iu = rhashtable_lookup_fast(&pag->pagi_unlinked_hash, &agino,
			xfs_iunlink_hash_params);
	if (!iu)
		return 0;

	/*
	 * Remove the entry.  This shouldn't ever return an error, but if we
	 * couldn't remove the old entry we don't want to add it again to the
	 * hash table, and if the entry disappeared on us then someone's
	 * violated the locking rules and we need to fail loudly.  Either way
	 * we cannot remove the inode because internal state is or would have
	 * been corrupt.
	 */
	error = rhashtable_remove_fast(&pag->pagi_unlinked_hash,
			&iu->iu_rhash_head, xfs_iunlink_hash_params);
	if (error)
		return error;

	/* If there is no new next entry just free our item and return. */
	if (next_unlinked == NULLAGINO) {
		kmem_free(iu);
		return 0;
	}

	/* Update the entry and re-add it to the hash table. */
	iu->iu_next_unlinked = next_unlinked;
	return xfs_iunlink_insert_backref(pag, iu);
}

/* Set up the in-core predecessor structures. */
int
xfs_iunlink_init(
	struct xfs_perag	*pag)
{
	return rhashtable_init(&pag->pagi_unlinked_hash,
			&xfs_iunlink_hash_params);
}

/* Free the in-core predecessor structures. */
static void
xfs_iunlink_free_item(
	void			*ptr,
	void			*arg)
{
	struct xfs_iunlink	*iu = ptr;
	bool			*freed_anything = arg;

	*freed_anything = true;
	kmem_free(iu);
}

void
xfs_iunlink_destroy(
	struct xfs_perag	*pag)
{
	bool			freed_anything = false;

	rhashtable_free_and_destroy(&pag->pagi_unlinked_hash,
			xfs_iunlink_free_item, &freed_anything);

	ASSERT(freed_anything == false || xfs_is_shutdown(pag->pag_mount));
}

/*
 * Point the AGI unlinked bucket at an inode and log the results.  The caller
 * is responsible for validating the old value.
 */
STATIC int
xfs_iunlink_update_bucket(
	struct xfs_trans	*tp,
	struct xfs_perag	*pag,
	struct xfs_buf		*agibp,
	unsigned int		bucket_index,
	xfs_agino_t		new_agino)
{
	struct xfs_agi		*agi = agibp->b_addr;
	xfs_agino_t		old_value;
	int			offset;

	ASSERT(xfs_verify_agino_or_null(tp->t_mountp, pag->pag_agno, new_agino));

	old_value = be32_to_cpu(agi->agi_unlinked[bucket_index]);
	trace_xfs_iunlink_update_bucket(tp->t_mountp, pag->pag_agno, bucket_index,
			old_value, new_agino);

	/*
	 * We should never find the head of the list already set to the value
	 * passed in because either we're adding or removing ourselves from the
	 * head of the list.
	 */
	if (old_value == new_agino) {
		xfs_buf_mark_corrupt(agibp);
		return -EFSCORRUPTED;
	}

	agi->agi_unlinked[bucket_index] = cpu_to_be32(new_agino);
	offset = offsetof(struct xfs_agi, agi_unlinked) +
			(sizeof(xfs_agino_t) * bucket_index);
	xfs_trans_log_buf(tp, agibp, offset, offset + sizeof(xfs_agino_t) - 1);
	return 0;
}

/* Set an on-disk inode's next_unlinked pointer. */
STATIC void
xfs_iunlink_update_dinode(
	struct xfs_trans	*tp,
	struct xfs_perag	*pag,
	xfs_agino_t		agino,
	struct xfs_buf		*ibp,
	struct xfs_dinode	*dip,
	struct xfs_imap		*imap,
	xfs_agino_t		next_agino)
{
	struct xfs_mount	*mp = tp->t_mountp;
	int			offset;

	ASSERT(xfs_verify_agino_or_null(mp, pag->pag_agno, next_agino));

	trace_xfs_iunlink_update_dinode(mp, pag->pag_agno, agino,
			be32_to_cpu(dip->di_next_unlinked), next_agino);

	dip->di_next_unlinked = cpu_to_be32(next_agino);
	offset = imap->im_boffset +
			offsetof(struct xfs_dinode, di_next_unlinked);

	/* need to recalc the inode CRC if appropriate */
	xfs_dinode_calc_crc(mp, dip);
	xfs_trans_inode_buf(tp, ibp);
	xfs_trans_log_buf(tp, ibp, offset, offset + sizeof(xfs_agino_t) - 1);
}

/* Set an in-core inode's unlinked pointer and return the old value. */
STATIC int
xfs_iunlink_update_inode(
	struct xfs_trans	*tp,
	struct xfs_inode	*ip,
	struct xfs_perag	*pag,
	xfs_agino_t		next_agino,
	xfs_agino_t		*old_next_agino)
{
	struct xfs_mount	*mp = tp->t_mountp;
	struct xfs_dinode	*dip;
	struct xfs_buf		*ibp;
	xfs_agino_t		old_value;
	int			error;

	ASSERT(xfs_verify_agino_or_null(mp, pag->pag_agno, next_agino));

	error = xfs_imap_to_bp(mp, tp, &ip->i_imap, &ibp);
	if (error)
		return error;
	dip = xfs_buf_offset(ibp, ip->i_imap.im_boffset);

	/* Make sure the old pointer isn't garbage. */
	old_value = be32_to_cpu(dip->di_next_unlinked);
	if (!xfs_verify_agino_or_null(mp, pag->pag_agno, old_value)) {
		xfs_inode_verifier_error(ip, -EFSCORRUPTED, __func__, dip,
				sizeof(*dip), __this_address);
		error = -EFSCORRUPTED;
		goto out;
	}

	/*
	 * Since we're updating a linked list, we should never find that the
	 * current pointer is the same as the new value, unless we're
	 * terminating the list.
	 */
	*old_next_agino = old_value;
	if (old_value == next_agino) {
		if (next_agino != NULLAGINO) {
			xfs_inode_verifier_error(ip, -EFSCORRUPTED, __func__,
					dip, sizeof(*dip), __this_address);
			error = -EFSCORRUPTED;
		}
		goto out;
	}

	/* Ok, update the new pointer. */
	xfs_iunlink_update_dinode(tp, pag, XFS_INO_TO_AGINO(mp, ip->i_ino),
			ibp, dip, &ip->i_imap, next_agino);
	return 0;
out:
	xfs_trans_brelse(tp, ibp);
	return error;
}

/*
 * This is called when the inode's link count has gone to 0 or we are creating
 * a tmpfile via O_TMPFILE.  The inode @ip must have nlink == 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.
 */
STATIC int
xfs_iunlink(
	struct xfs_trans	*tp,
	struct xfs_inode	*ip)
{
	struct xfs_mount	*mp = tp->t_mountp;
	struct xfs_perag	*pag;
	struct xfs_agi		*agi;
	struct xfs_buf		*agibp;
	xfs_agino_t		next_agino;
	xfs_agino_t		agino = XFS_INO_TO_AGINO(mp, ip->i_ino);
	short			bucket_index = agino % XFS_AGI_UNLINKED_BUCKETS;
	int			error;

	ASSERT(VFS_I(ip)->i_nlink == 0);
	ASSERT(VFS_I(ip)->i_mode != 0);
	trace_xfs_iunlink(ip);

	pag = xfs_perag_get(mp, XFS_INO_TO_AGNO(mp, ip->i_ino));

	/* Get the agi buffer first.  It ensures lock ordering on the list. */
	error = xfs_read_agi(mp, tp, pag->pag_agno, &agibp);
	if (error)
		goto out;
	agi = agibp->b_addr;

	/*
	 * Get the index into the agi hash table for the list this inode will
	 * go on.  Make sure the pointer isn't garbage and that this inode
	 * isn't already on the list.
	 */
	next_agino = be32_to_cpu(agi->agi_unlinked[bucket_index]);
	if (next_agino == agino ||
	    !xfs_verify_agino_or_null(mp, pag->pag_agno, next_agino)) {
		xfs_buf_mark_corrupt(agibp);
		error = -EFSCORRUPTED;
		goto out;
	}

	if (next_agino != NULLAGINO) {
		xfs_agino_t		old_agino;

		/*
		 * There is already another inode in the bucket, so point this
		 * inode to the current head of the list.
		 */
		error = xfs_iunlink_update_inode(tp, ip, pag, next_agino,
				&old_agino);
		if (error)
			goto out;
		ASSERT(old_agino == NULLAGINO);

		/*
		 * agino has been unlinked, add a backref from the next inode
		 * back to agino.
		 */
		error = xfs_iunlink_add_backref(pag, agino, next_agino);
		if (error)
			goto out;
	}

	/* Point the head of the list to point to this inode. */
	error = xfs_iunlink_update_bucket(tp, pag, agibp, bucket_index, agino);
out:
	xfs_perag_put(pag);
	return error;
}

/* Return the imap, dinode pointer, and buffer for an inode. */
STATIC int
xfs_iunlink_map_ino(
	struct xfs_trans	*tp,
	xfs_agnumber_t		agno,
	xfs_agino_t		agino,
	struct xfs_imap		*imap,
	struct xfs_dinode	**dipp,
	struct xfs_buf		**bpp)
{
	struct xfs_mount	*mp = tp->t_mountp;
	int			error;

	imap->im_blkno = 0;
	error = xfs_imap(mp, tp, XFS_AGINO_TO_INO(mp, agno, agino), imap, 0);
	if (error) {
		xfs_warn(mp, "%s: xfs_imap returned error %d.",
				__func__, error);
		return error;
	}

	error = xfs_imap_to_bp(mp, tp, imap, bpp);
	if (error) {
		xfs_warn(mp, "%s: xfs_imap_to_bp returned error %d.",
				__func__, error);
		return error;
	}

	*dipp = xfs_buf_offset(*bpp, imap->im_boffset);
	return 0;
}

/*
 * Walk the unlinked chain from @head_agino until we find the inode that
 * points to @target_agino.  Return the inode number, map, dinode pointer,
 * and inode cluster buffer of that inode as @agino, @imap, @dipp, and @bpp.
 *
 * @tp, @pag, @head_agino, and @target_agino are input parameters.
 * @agino, @imap, @dipp, and @bpp are all output parameters.
 *
 * Do not call this function if @target_agino is the head of the list.
 */
STATIC int
xfs_iunlink_map_prev(
	struct xfs_trans	*tp,
	struct xfs_perag	*pag,
	xfs_agino_t		head_agino,
	xfs_agino_t		target_agino,
	xfs_agino_t		*agino,
	struct xfs_imap		*imap,
	struct xfs_dinode	**dipp,
	struct xfs_buf		**bpp)
{
	struct xfs_mount	*mp = tp->t_mountp;
	xfs_agino_t		next_agino;
	int			error;

	ASSERT(head_agino != target_agino);
	*bpp = NULL;

	/* See if our backref cache can find it faster. */
	*agino = xfs_iunlink_lookup_backref(pag, target_agino);
	if (*agino != NULLAGINO) {
		error = xfs_iunlink_map_ino(tp, pag->pag_agno, *agino, imap,
				dipp, bpp);
		if (error)
			return error;

		if (be32_to_cpu((*dipp)->di_next_unlinked) == target_agino)
			return 0;

		/*
		 * If we get here the cache contents were corrupt, so drop the
		 * buffer and fall back to walking the bucket list.
		 */
		xfs_trans_brelse(tp, *bpp);
		*bpp = NULL;
		WARN_ON_ONCE(1);
	}

	trace_xfs_iunlink_map_prev_fallback(mp, pag->pag_agno);

	/* Otherwise, walk the entire bucket until we find it. */
	next_agino = head_agino;
	while (next_agino != target_agino) {
		xfs_agino_t	unlinked_agino;

		if (*bpp)
			xfs_trans_brelse(tp, *bpp);

		*agino = next_agino;
		error = xfs_iunlink_map_ino(tp, pag->pag_agno, next_agino, imap,
				dipp, bpp);
		if (error)
			return error;

		unlinked_agino = be32_to_cpu((*dipp)->di_next_unlinked);
		/*
		 * Make sure this pointer is valid and isn't an obvious
		 * infinite loop.
		 */
		if (!xfs_verify_agino(mp, pag->pag_agno, unlinked_agino) ||
		    next_agino == unlinked_agino) {
			XFS_CORRUPTION_ERROR(__func__,
					XFS_ERRLEVEL_LOW, mp,
					*dipp, sizeof(**dipp));
			error = -EFSCORRUPTED;
			return error;
		}
		next_agino = unlinked_agino;
	}

	return 0;
}

/*
 * Pull the on-disk inode from the AGI unlinked list.
 */
STATIC int
xfs_iunlink_remove(
	struct xfs_trans	*tp,
	struct xfs_perag	*pag,
	struct xfs_inode	*ip)
{
	struct xfs_mount	*mp = tp->t_mountp;
	struct xfs_agi		*agi;
	struct xfs_buf		*agibp;
	struct xfs_buf		*last_ibp;
	struct xfs_dinode	*last_dip = NULL;
	xfs_agino_t		agino = XFS_INO_TO_AGINO(mp, ip->i_ino);
	xfs_agino_t		next_agino;
	xfs_agino_t		head_agino;
	short			bucket_index = agino % XFS_AGI_UNLINKED_BUCKETS;
	int			error;

	trace_xfs_iunlink_remove(ip);

	/* Get the agi buffer first.  It ensures lock ordering on the list. */
	error = xfs_read_agi(mp, tp, pag->pag_agno, &agibp);
	if (error)
		return error;
	agi = agibp->b_addr;

	/*
	 * Get the index into the agi hash table for the list this inode will
	 * go on.  Make sure the head pointer isn't garbage.
	 */
	head_agino = be32_to_cpu(agi->agi_unlinked[bucket_index]);
	if (!xfs_verify_agino(mp, pag->pag_agno, head_agino)) {
		XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp,
				agi, sizeof(*agi));
		return -EFSCORRUPTED;
	}

	/*
	 * Set our inode's next_unlinked pointer to NULL and then return
	 * the old pointer value so that we can update whatever was previous
	 * to us in the list to point to whatever was next in the list.
	 */
	error = xfs_iunlink_update_inode(tp, ip, pag, NULLAGINO, &next_agino);
	if (error)
		return error;

	/*
	 * If there was a backref pointing from the next inode back to this
	 * one, remove it because we've removed this inode from the list.
	 *
	 * Later, if this inode was in the middle of the list we'll update
	 * this inode's backref to point from the next inode.
	 */
	if (next_agino != NULLAGINO) {
		error = xfs_iunlink_change_backref(pag, next_agino, NULLAGINO);
		if (error)
			return error;
	}

	if (head_agino != agino) {
		struct xfs_imap	imap;
		xfs_agino_t	prev_agino;

		/* We need to search the list for the inode being freed. */
		error = xfs_iunlink_map_prev(tp, pag, head_agino, agino,
				&prev_agino, &imap, &last_dip, &last_ibp);
		if (error)
			return error;

		/* Point the previous inode on the list to the next inode. */
		xfs_iunlink_update_dinode(tp, pag, prev_agino, last_ibp,
				last_dip, &imap, next_agino);

		/*
		 * Now we deal with the backref for this inode.  If this inode
		 * pointed at a real inode, change the backref that pointed to
		 * us to point to our old next.  If this inode was the end of
		 * the list, delete the backref that pointed to us.  Note that
		 * change_backref takes care of deleting the backref if
		 * next_agino is NULLAGINO.
		 */
		return xfs_iunlink_change_backref(agibp->b_pag, agino,
				next_agino);
	}

	/* Point the head of the list to the next unlinked inode. */
	return xfs_iunlink_update_bucket(tp, pag, agibp, bucket_index,
			next_agino);
}

/*
 * Look up the inode number specified and if it is not already marked XFS_ISTALE
 * mark it stale. We should only find clean inodes in this lookup that aren't
 * already stale.
 */
static void
xfs_ifree_mark_inode_stale(
	struct xfs_perag	*pag,
	struct xfs_inode	*free_ip,
	xfs_ino_t		inum)
{
	struct xfs_mount	*mp = pag->pag_mount;
	struct xfs_inode_log_item *iip;
	struct xfs_inode	*ip;

retry:
	rcu_read_lock();
	ip = radix_tree_lookup(&pag->pag_ici_root, XFS_INO_TO_AGINO(mp, inum));

	/* Inode not in memory, nothing to do */
	if (!ip) {
		rcu_read_unlock();
		return;
	}

	/*
	 * because this is an RCU protected lookup, we could find a recently
	 * freed or even reallocated inode during the lookup. We need to check
	 * under the i_flags_lock for a valid inode here. Skip it if it is not
	 * valid, the wrong inode or stale.
	 */
	spin_lock(&ip->i_flags_lock);
	if (ip->i_ino != inum || __xfs_iflags_test(ip, XFS_ISTALE))
		goto out_iflags_unlock;

	/*
	 * Don't try to lock/unlock the current inode, but we _cannot_ skip the
	 * other inodes that we did not find in the list attached to the buffer
	 * and are not already marked stale. If we can't lock it, back off and
	 * retry.
	 */
	if (ip != free_ip) {
		if (!xfs_ilock_nowait(ip, XFS_ILOCK_EXCL)) {
			spin_unlock(&ip->i_flags_lock);
			rcu_read_unlock();
			delay(1);
			goto retry;
		}
	}
	ip->i_flags |= XFS_ISTALE;

	/*
	 * If the inode is flushing, it is already attached to the buffer.  All
	 * we needed to do here is mark the inode stale so buffer IO completion
	 * will remove it from the AIL.
	 */
	iip = ip->i_itemp;
	if (__xfs_iflags_test(ip, XFS_IFLUSHING)) {
		ASSERT(!list_empty(&iip->ili_item.li_bio_list));
		ASSERT(iip->ili_last_fields);
		goto out_iunlock;
	}

	/*
	 * Inodes not attached to the buffer can be released immediately.
	 * Everything else has to go through xfs_iflush_abort() on journal
	 * commit as the flock synchronises removal of the inode from the
	 * cluster buffer against inode reclaim.
	 */
	if (!iip || list_empty(&iip->ili_item.li_bio_list))
		goto out_iunlock;

	__xfs_iflags_set(ip, XFS_IFLUSHING);
	spin_unlock(&ip->i_flags_lock);
	rcu_read_unlock();

	/* we have a dirty inode in memory that has not yet been flushed. */
	spin_lock(&iip->ili_lock);
	iip->ili_last_fields = iip->ili_fields;
	iip->ili_fields = 0;
	iip->ili_fsync_fields = 0;
	spin_unlock(&iip->ili_lock);
	ASSERT(iip->ili_last_fields);

	if (ip != free_ip)
		xfs_iunlock(ip, XFS_ILOCK_EXCL);
	return;

out_iunlock:
	if (ip != free_ip)
		xfs_iunlock(ip, XFS_ILOCK_EXCL);
out_iflags_unlock:
	spin_unlock(&ip->i_flags_lock);
	rcu_read_unlock();
}

/*
 * A big issue when freeing the inode cluster is that we _cannot_ skip any
 * inodes that are in memory - they all must be marked stale and attached to
 * the cluster buffer.
 */
static int
xfs_ifree_cluster(
	struct xfs_trans	*tp,
	struct xfs_perag	*pag,
	struct xfs_inode	*free_ip,
	struct xfs_icluster	*xic)
{
	struct xfs_mount	*mp = free_ip->i_mount;
	struct xfs_ino_geometry	*igeo = M_IGEO(mp);
	struct xfs_buf		*bp;
	xfs_daddr_t		blkno;
	xfs_ino_t		inum = xic->first_ino;
	int			nbufs;
	int			i, j;
	int			ioffset;
	int			error;

	nbufs = igeo->ialloc_blks / igeo->blocks_per_cluster;

	for (j = 0; j < nbufs; j++, inum += igeo->inodes_per_cluster) {
		/*
		 * The allocation bitmap tells us which inodes of the chunk were
		 * physically allocated. Skip the cluster if an inode falls into
		 * a sparse region.
		 */
		ioffset = inum - xic->first_ino;
		if ((xic->alloc & XFS_INOBT_MASK(ioffset)) == 0) {
			ASSERT(ioffset % igeo->inodes_per_cluster == 0);
			continue;
		}

		blkno = XFS_AGB_TO_DADDR(mp, XFS_INO_TO_AGNO(mp, inum),
					 XFS_INO_TO_AGBNO(mp, inum));

		/*
		 * We obtain and lock the backing buffer first in the process
		 * here to ensure dirty inodes attached to the buffer remain in
		 * the flushing state while we mark them stale.
		 *
		 * If we scan the in-memory inodes first, then buffer IO can
		 * complete before we get a lock on it, and hence we may fail
		 * to mark all the active inodes on the buffer stale.
		 */
		error = xfs_trans_get_buf(tp, mp->m_ddev_targp, blkno,
				mp->m_bsize * igeo->blocks_per_cluster,
				XBF_UNMAPPED, &bp);
		if (error)
			return error;

		/*
		 * This buffer may not have been correctly initialised as we
		 * didn't read it from disk. That's not important because we are
		 * only using to mark the buffer as stale in the log, and to
		 * attach stale cached inodes on it. That means it will never be
		 * dispatched for IO. If it is, we want to know about it, and we
		 * want it to fail. We can acheive this by adding a write
		 * verifier to the buffer.
		 */
		bp->b_ops = &xfs_inode_buf_ops;

		/*
		 * Now we need to set all the cached clean inodes as XFS_ISTALE,
		 * too. This requires lookups, and will skip inodes that we've
		 * already marked XFS_ISTALE.
		 */
		for (i = 0; i < igeo->inodes_per_cluster; i++)
			xfs_ifree_mark_inode_stale(pag, free_ip, inum + i);

		xfs_trans_stale_inode_buf(tp, bp);
		xfs_trans_binval(tp, bp);
	}
	return 0;
}

/*
 * 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(
	struct xfs_trans	*tp,
	struct xfs_inode	*ip)
{
	struct xfs_mount	*mp = ip->i_mount;
	struct xfs_perag	*pag;
	struct xfs_icluster	xic = { 0 };
	struct xfs_inode_log_item *iip = ip->i_itemp;
	int			error;

	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));
	ASSERT(VFS_I(ip)->i_nlink == 0);
	ASSERT(ip->i_df.if_nextents == 0);
	ASSERT(ip->i_disk_size == 0 || !S_ISREG(VFS_I(ip)->i_mode));
	ASSERT(ip->i_nblocks == 0);

	pag = xfs_perag_get(mp, XFS_INO_TO_AGNO(mp, ip->i_ino));

	/*
	 * Pull the on-disk inode from the AGI unlinked list.
	 */
	error = xfs_iunlink_remove(tp, pag, ip);
	if (error)
		goto out;

	error = xfs_difree(tp, pag, ip->i_ino, &xic);
	if (error)
		goto out;

	/*
	 * Free any local-format data sitting around before we reset the
	 * data fork to extents format.  Note that the attr fork data has
	 * already been freed by xfs_attr_inactive.
	 */
	if (ip->i_df.if_format == XFS_DINODE_FMT_LOCAL) {
		kmem_free(ip->i_df.if_u1.if_data);
		ip->i_df.if_u1.if_data = NULL;
		ip->i_df.if_bytes = 0;
	}

	VFS_I(ip)->i_mode = 0;		/* mark incore inode as free */
	ip->i_diflags = 0;
	ip->i_diflags2 = mp->m_ino_geo.new_diflags2;
	ip->i_forkoff = 0;		/* mark the attr fork not in use */
	ip->i_df.if_format = XFS_DINODE_FMT_EXTENTS;
	if (xfs_iflags_test(ip, XFS_IPRESERVE_DM_FIELDS))
		xfs_iflags_clear(ip, XFS_IPRESERVE_DM_FIELDS);

	/* Don't attempt to replay owner changes for a deleted inode */
	spin_lock(&iip->ili_lock);
	iip->ili_fields &= ~(XFS_ILOG_AOWNER | XFS_ILOG_DOWNER);
	spin_unlock(&iip->ili_lock);

	/*
	 * Bump the generation count so no one will be confused
	 * by reincarnations of this inode.
	 */
	VFS_I(ip)->i_generation++;
	xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);

	if (xic.deleted)
		error = xfs_ifree_cluster(tp, pag, ip, &xic);
out:
	xfs_perag_put(pag);
	return error;
}

/*
 * This is called to unpin an inode.  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(
	struct xfs_inode	*ip)
{
	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL|XFS_ILOCK_SHARED));

	trace_xfs_inode_unpin_nowait(ip, _RET_IP_);

	/* Give the log a push to start the unpinning I/O */
	xfs_log_force_seq(ip->i_mount, ip->i_itemp->ili_commit_seq, 0, NULL);

}

static void
__xfs_iunpin_wait(
	struct xfs_inode	*ip)
{
	wait_queue_head_t *wq = bit_waitqueue(&ip->i_flags, __XFS_IPINNED_BIT);
	DEFINE_WAIT_BIT(wait, &ip->i_flags, __XFS_IPINNED_BIT);

	xfs_iunpin(ip);

	do {
		prepare_to_wait(wq, &wait.wq_entry, TASK_UNINTERRUPTIBLE);
		if (xfs_ipincount(ip))
			io_schedule();
	} while (xfs_ipincount(ip));
	finish_wait(wq, &wait.wq_entry);
}

void
xfs_iunpin_wait(
	struct xfs_inode	*ip)
{
	if (xfs_ipincount(ip))
		__xfs_iunpin_wait(ip);
}

/*
 * Removing an inode from the namespace involves removing the directory entry
 * and dropping the link count on the inode. Removing the directory entry can
 * result in locking an AGF (directory blocks were freed) and removing a link
 * count can result in placing the inode on an unlinked list which results in
 * locking an AGI.
 *
 * The big problem here is that we have an ordering constraint on AGF and AGI
 * locking - inode allocation locks the AGI, then can allocate a new extent for
 * new inodes, locking the AGF after the AGI. Similarly, freeing the inode
 * removes the inode from the unlinked list, requiring that we lock the AGI
 * first, and then freeing the inode can result in an inode chunk being freed
 * and hence freeing disk space requiring that we lock an AGF.
 *
 * Hence the ordering that is imposed by other parts of the code is AGI before
 * AGF. This means we cannot remove the directory entry before we drop the inode
 * reference count and put it on the unlinked list as this results in a lock
 * order of AGF then AGI, and this can deadlock against inode allocation and
 * freeing. Therefore we must drop the link counts before we remove the
 * directory entry.
 *
 * This is still safe from a transactional point of view - it is not until we
 * get to xfs_defer_finish() that we have the possibility of multiple
 * transactions in this operation. Hence as long as we remove the directory
 * entry and drop the link count in the first transaction of the remove
 * operation, there are no transactional constraints on the ordering here.
 */
int
xfs_remove(
	xfs_inode_t             *dp,
	struct xfs_name		*name,
	xfs_inode_t		*ip)
{
	xfs_mount_t		*mp = dp->i_mount;
	xfs_trans_t             *tp = NULL;
	int			is_dir = S_ISDIR(VFS_I(ip)->i_mode);
	int                     error = 0;
	uint			resblks;

	trace_xfs_remove(dp, name);

	if (xfs_is_shutdown(mp))
		return -EIO;

	error = xfs_qm_dqattach(dp);
	if (error)
		goto std_return;

	error = xfs_qm_dqattach(ip);
	if (error)
		goto std_return;

	/*
	 * We try to get the real space reservation first,
	 * allowing for directory btree deletion(s) implying
	 * possible bmap insert(s).  If we can't get the space
	 * reservation then we use 0 instead, and avoid the bmap
	 * btree insert(s) in the directory code by, if the bmap
	 * insert tries to happen, instead trimming the LAST
	 * block from the directory.
	 */
	resblks = XFS_REMOVE_SPACE_RES(mp);
	error = xfs_trans_alloc(mp, &M_RES(mp)->tr_remove, resblks, 0, 0, &tp);
	if (error == -ENOSPC) {
		resblks = 0;
		error = xfs_trans_alloc(mp, &M_RES(mp)->tr_remove, 0, 0, 0,
				&tp);
	}
	if (error) {
		ASSERT(error != -ENOSPC);
		goto std_return;
	}

	xfs_lock_two_inodes(dp, XFS_ILOCK_EXCL, ip, XFS_ILOCK_EXCL);

	xfs_trans_ijoin(tp, dp, XFS_ILOCK_EXCL);
	xfs_trans_ijoin(tp, ip, XFS_ILOCK_EXCL);

	/*
	 * If we're removing a directory perform some additional validation.
	 */
	if (is_dir) {
		ASSERT(VFS_I(ip)->i_nlink >= 2);
		if (VFS_I(ip)->i_nlink != 2) {
			error = -ENOTEMPTY;
			goto out_trans_cancel;
		}
		if (!xfs_dir_isempty(ip)) {
			error = -ENOTEMPTY;
			goto out_trans_cancel;
		}

		/* Drop the link from ip's "..".  */
		error = xfs_droplink(tp, dp);
		if (error)
			goto out_trans_cancel;

		/* Drop the "." link from ip to self.  */
		error = xfs_droplink(tp, ip);
		if (error)
			goto out_trans_cancel;

		/*
		 * Point the unlinked child directory's ".." entry to the root
		 * directory to eliminate back-references to inodes that may
		 * get freed before the child directory is closed.  If the fs
		 * gets shrunk, this can lead to dirent inode validation errors.
		 */
		if (dp->i_ino != tp->t_mountp->m_sb.sb_rootino) {
			error = xfs_dir_replace(tp, ip, &xfs_name_dotdot,
					tp->t_mountp->m_sb.sb_rootino, 0);
			if (error)
				return error;
		}
	} else {
		/*
		 * When removing a non-directory we need to log the parent
		 * inode here.  For a directory this is done implicitly
		 * by the xfs_droplink call for the ".." entry.
		 */
		xfs_trans_log_inode(tp, dp, XFS_ILOG_CORE);
	}
	xfs_trans_ichgtime(tp, dp, XFS_ICHGTIME_MOD | XFS_ICHGTIME_CHG);

	/* Drop the link from dp to ip. */
	error = xfs_droplink(tp, ip);
	if (error)
		goto out_trans_cancel;

	error = xfs_dir_removename(tp, dp, name, ip->i_ino, resblks);
	if (error) {
		ASSERT(error != -ENOENT);
		goto out_trans_cancel;
	}

	/*
	 * If this is a synchronous mount, make sure that the
	 * remove transaction goes to disk before returning to
	 * the user.
	 */
	if (xfs_has_wsync(mp) || xfs_has_dirsync(mp))
		xfs_trans_set_sync(tp);

	error = xfs_trans_commit(tp);
	if (error)
		goto std_return;

	if (is_dir && xfs_inode_is_filestream(ip))
		xfs_filestream_deassociate(ip);

	return 0;

 out_trans_cancel:
	xfs_trans_cancel(tp);
 std_return:
	return error;
}

/*
 * Enter all inodes for a rename transaction into a sorted array.
 */
#define __XFS_SORT_INODES	5
STATIC void
xfs_sort_for_rename(
	struct xfs_inode	*dp1,	/* in: old (source) directory inode */
	struct xfs_inode	*dp2,	/* in: new (target) directory inode */
	struct xfs_inode	*ip1,	/* in: inode of old entry */
	struct xfs_inode	*ip2,	/* in: inode of new entry */
	struct xfs_inode	*wip,	/* in: whiteout inode */
	struct xfs_inode	**i_tab,/* out: sorted array of inodes */
	int			*num_inodes)  /* in/out: inodes in array */
{
	int			i, j;

	ASSERT(*num_inodes == __XFS_SORT_INODES);
	memset(i_tab, 0, *num_inodes * sizeof(struct xfs_inode *));

	/*
	 * i_tab contains a list of pointers to inodes.  We initialize
	 * the table here & we'll sort it.  We will then use it to
	 * order the acquisition of the inode locks.
	 *
	 * Note that the table may contain duplicates.  e.g., dp1 == dp2.
	 */
	i = 0;
	i_tab[i++] = dp1;
	i_tab[i++] = dp2;
	i_tab[i++] = ip1;
	if (ip2)
		i_tab[i++] = ip2;
	if (wip)
		i_tab[i++] = wip;
	*num_inodes = i;

	/*
	 * Sort the elements via bubble sort.  (Remember, there are at
	 * most 5 elements to sort, so this is adequate.)
	 */
	for (i = 0; i < *num_inodes; i++) {
		for (j = 1; j < *num_inodes; j++) {
			if (i_tab[j]->i_ino < i_tab[j-1]->i_ino) {
				struct xfs_inode *temp = i_tab[j];
				i_tab[j] = i_tab[j-1];
				i_tab[j-1] = temp;
			}
		}
	}
}

static int
xfs_finish_rename(
	struct xfs_trans	*tp)
{
	/*
	 * If this is a synchronous mount, make sure that the rename transaction
	 * goes to disk before returning to the user.
	 */
	if (xfs_has_wsync(tp->t_mountp) || xfs_has_dirsync(tp->t_mountp))
		xfs_trans_set_sync(tp);

	return xfs_trans_commit(tp);
}

/*
 * xfs_cross_rename()
 *
 * responsible for handling RENAME_EXCHANGE flag in renameat2() syscall
 */
STATIC int
xfs_cross_rename(
	struct xfs_trans	*tp,
	struct xfs_inode	*dp1,
	struct xfs_name		*name1,
	struct xfs_inode	*ip1,
	struct xfs_inode	*dp2,
	struct xfs_name		*name2,
	struct xfs_inode	*ip2,
	int			spaceres)
{
	int		error = 0;
	int		ip1_flags = 0;
	int		ip2_flags = 0;
	int		dp2_flags = 0;

	/* Swap inode number for dirent in first parent */
	error = xfs_dir_replace(tp, dp1, name1, ip2->i_ino, spaceres);
	if (error)
		goto out_trans_abort;

	/* Swap inode number for dirent in second parent */
	error = xfs_dir_replace(tp, dp2, name2, ip1->i_ino, spaceres);
	if (error)
		goto out_trans_abort;

	/*
	 * If we're renaming one or more directories across different parents,
	 * update the respective ".." entries (and link counts) to match the new
	 * parents.
	 */
	if (dp1 != dp2) {
		dp2_flags = XFS_ICHGTIME_MOD | XFS_ICHGTIME_CHG;

		if (S_ISDIR(VFS_I(ip2)->i_mode)) {
			error = xfs_dir_replace(tp, ip2, &xfs_name_dotdot,
						dp1->i_ino, spaceres);
			if (error)
				goto out_trans_abort;

			/* transfer ip2 ".." reference to dp1 */
			if (!S_ISDIR(VFS_I(ip1)->i_mode)) {
				error = xfs_droplink(tp, dp2);
				if (error)
					goto out_trans_abort;
				xfs_bumplink(tp, dp1);
			}

			/*
			 * Although ip1 isn't changed here, userspace needs
			 * to be warned about the change, so that applications
			 * relying on it (like backup ones), will properly
			 * notify the change
			 */
			ip1_flags |= XFS_ICHGTIME_CHG;
			ip2_flags |= XFS_ICHGTIME_MOD | XFS_ICHGTIME_CHG;
		}

		if (S_ISDIR(VFS_I(ip1)->i_mode)) {
			error = xfs_dir_replace(tp, ip1, &xfs_name_dotdot,
						dp2->i_ino, spaceres);
			if (error)
				goto out_trans_abort;

			/* transfer ip1 ".." reference to dp2 */
			if (!S_ISDIR(VFS_I(ip2)->i_mode)) {
				error = xfs_droplink(tp, dp1);
				if (error)
					goto out_trans_abort;
				xfs_bumplink(tp, dp2);
			}

			/*
			 * Although ip2 isn't changed here, userspace needs
			 * to be warned about the change, so that applications
			 * relying on it (like backup ones), will properly
			 * notify the change
			 */
			ip1_flags |= XFS_ICHGTIME_MOD | XFS_ICHGTIME_CHG;
			ip2_flags |= XFS_ICHGTIME_CHG;
		}
	}

	if (ip1_flags) {
		xfs_trans_ichgtime(tp, ip1, ip1_flags);
		xfs_trans_log_inode(tp, ip1, XFS_ILOG_CORE);
	}
	if (ip2_flags) {
		xfs_trans_ichgtime(tp, ip2, ip2_flags);
		xfs_trans_log_inode(tp, ip2, XFS_ILOG_CORE);
	}
	if (dp2_flags) {
		xfs_trans_ichgtime(tp, dp2, dp2_flags);
		xfs_trans_log_inode(tp, dp2, XFS_ILOG_CORE);
	}
	xfs_trans_ichgtime(tp, dp1, XFS_ICHGTIME_MOD | XFS_ICHGTIME_CHG);
	xfs_trans_log_inode(tp, dp1, XFS_ILOG_CORE);
	return xfs_finish_rename(tp);

out_trans_abort:
	xfs_trans_cancel(tp);
	return error;
}

/*
 * xfs_rename_alloc_whiteout()
 *
 * Return a referenced, unlinked, unlocked inode that can be used as a
 * whiteout in a rename transaction. We use a tmpfile inode here so that if we
 * crash between allocating the inode and linking it into the rename transaction
 * recovery will free the inode and we won't leak it.
 */
static int
xfs_rename_alloc_whiteout(
	struct user_namespace	*mnt_userns,
	struct xfs_inode	*dp,
	struct xfs_inode	**wip)
{
	struct xfs_inode	*tmpfile;
	int			error;

	error = xfs_create_tmpfile(mnt_userns, dp, S_IFCHR | WHITEOUT_MODE,
				   &tmpfile);
	if (error)
		return error;

	/*
	 * Prepare the tmpfile inode as if it were created through the VFS.
	 * Complete the inode setup and flag it as linkable.  nlink is already
	 * zero, so we can skip the drop_nlink.
	 */
	xfs_setup_iops(tmpfile);
	xfs_finish_inode_setup(tmpfile);
	VFS_I(tmpfile)->i_state |= I_LINKABLE;

	*wip = tmpfile;
	return 0;
}

/*
 * xfs_rename
 */
int
xfs_rename(
	struct user_namespace	*mnt_userns,
	struct xfs_inode	*src_dp,
	struct xfs_name		*src_name,
	struct xfs_inode	*src_ip,
	struct xfs_inode	*target_dp,
	struct xfs_name		*target_name,
	struct xfs_inode	*target_ip,
	unsigned int		flags)
{
	struct xfs_mount	*mp = src_dp->i_mount;
	struct xfs_trans	*tp;
	struct xfs_inode	*wip = NULL;		/* whiteout inode */
	struct xfs_inode	*inodes[__XFS_SORT_INODES];
	int			i;
	int			num_inodes = __XFS_SORT_INODES;
	bool			new_parent = (src_dp != target_dp);
	bool			src_is_directory = S_ISDIR(VFS_I(src_ip)->i_mode);
	int			spaceres;
	int			error;

	trace_xfs_rename(src_dp, target_dp, src_name, target_name);

	if ((flags & RENAME_EXCHANGE) && !target_ip)
		return -EINVAL;

	/*
	 * If we are doing a whiteout operation, allocate the whiteout inode
	 * we will be placing at the target and ensure the type is set
	 * appropriately.
	 */
	if (flags & RENAME_WHITEOUT) {
		ASSERT(!(flags & (RENAME_NOREPLACE | RENAME_EXCHANGE)));
		error = xfs_rename_alloc_whiteout(mnt_userns, target_dp, &wip);
		if (error)
			return error;

		/* setup target dirent info as whiteout */
		src_name->type = XFS_DIR3_FT_CHRDEV;
	}

	xfs_sort_for_rename(src_dp, target_dp, src_ip, target_ip, wip,
				inodes, &num_inodes);

	spaceres = XFS_RENAME_SPACE_RES(mp, target_name->len);
	error = xfs_trans_alloc(mp, &M_RES(mp)->tr_rename, spaceres, 0, 0, &tp);
	if (error == -ENOSPC) {
		spaceres = 0;
		error = xfs_trans_alloc(mp, &M_RES(mp)->tr_rename, 0, 0, 0,
				&tp);
	}
	if (error)
		goto out_release_wip;

	/*
	 * Attach the dquots to the inodes
	 */
	error = xfs_qm_vop_rename_dqattach(inodes);
	if (error)
		goto out_trans_cancel;

	/*
	 * Lock all the participating inodes. Depending upon whether
	 * the target_name exists in the target directory, and
	 * whether the target directory is the same as the source
	 * directory, we can lock from 2 to 4 inodes.
	 */
	xfs_lock_inodes(inodes, num_inodes, XFS_ILOCK_EXCL);

	/*
	 * Join all the inodes to the transaction. From this point on,
	 * we can rely on either trans_commit or trans_cancel to unlock
	 * them.
	 */
	xfs_trans_ijoin(tp, src_dp, XFS_ILOCK_EXCL);
	if (new_parent)
		xfs_trans_ijoin(tp, target_dp, XFS_ILOCK_EXCL);
	xfs_trans_ijoin(tp, src_ip, XFS_ILOCK_EXCL);
	if (target_ip)
		xfs_trans_ijoin(tp, target_ip, XFS_ILOCK_EXCL);
	if (wip)
		xfs_trans_ijoin(tp, wip, XFS_ILOCK_EXCL);

	/*
	 * If we are using project inheritance, we only allow renames
	 * into our tree when the project IDs are the same; else the
	 * tree quota mechanism would be circumvented.
	 */
	if (unlikely((target_dp->i_diflags & XFS_DIFLAG_PROJINHERIT) &&
		     target_dp->i_projid != src_ip->i_projid)) {
		error = -EXDEV;
		goto out_trans_cancel;
	}

	/* RENAME_EXCHANGE is unique from here on. */
	if (flags & RENAME_EXCHANGE)
		return xfs_cross_rename(tp, src_dp, src_name, src_ip,
					target_dp, target_name, target_ip,
					spaceres);

	/*
	 * Check for expected errors before we dirty the transaction
	 * so we can return an error without a transaction abort.
	 *
	 * Extent count overflow check:
	 *
	 * From the perspective of src_dp, a rename operation is essentially a
	 * directory entry remove operation. Hence the only place where we check
	 * for extent count overflow for src_dp is in
	 * xfs_bmap_del_extent_real(). xfs_bmap_del_extent_real() returns
	 * -ENOSPC when it detects a possible extent count overflow and in
	 * response, the higher layers of directory handling code do the
	 * following:
	 * 1. Data/Free blocks: XFS lets these blocks linger until a
	 *    future remove operation removes them.
	 * 2. Dabtree blocks: XFS swaps the blocks with the last block in the
	 *    Leaf space and unmaps the last block.
	 *
	 * For target_dp, there are two cases depending on whether the
	 * destination directory entry exists or not.
	 *
	 * When destination directory entry does not exist (i.e. target_ip ==
	 * NULL), extent count overflow check is performed only when transaction
	 * has a non-zero sized space reservation associated with it.  With a
	 * zero-sized space reservation, XFS allows a rename operation to
	 * continue only when the directory has sufficient free space in its
	 * data/leaf/free space blocks to hold the new entry.
	 *
	 * When destination directory entry exists (i.e. target_ip != NULL), all
	 * we need to do is change the inode number associated with the already
	 * existing entry. Hence there is no need to perform an extent count
	 * overflow check.
	 */
	if (target_ip == NULL) {
		/*
		 * If there's no space reservation, check the entry will
		 * fit before actually inserting it.
		 */
		if (!spaceres) {
			error = xfs_dir_canenter(tp, target_dp, target_name);
			if (error)
				goto out_trans_cancel;
		} else {
			error = xfs_iext_count_may_overflow(target_dp,
					XFS_DATA_FORK,
					XFS_IEXT_DIR_MANIP_CNT(mp));
			if (error)
				goto out_trans_cancel;
		}
	} else {
		/*
		 * If target exists and it's a directory, check that whether
		 * it can be destroyed.
		 */
		if (S_ISDIR(VFS_I(target_ip)->i_mode) &&
		    (!xfs_dir_isempty(target_ip) ||
		     (VFS_I(target_ip)->i_nlink > 2))) {
			error = -EEXIST;
			goto out_trans_cancel;
		}
	}

	/*
	 * Lock the AGI buffers we need to handle bumping the nlink of the
	 * whiteout inode off the unlinked list and to handle dropping the
	 * nlink of the target inode.  Per locking order rules, do this in
	 * increasing AG order and before directory block allocation tries to
	 * grab AGFs because we grab AGIs before AGFs.
	 *
	 * The (vfs) caller must ensure that if src is a directory then
	 * target_ip is either null or an empty directory.
	 */
	for (i = 0; i < num_inodes && inodes[i] != NULL; i++) {
		if (inodes[i] == wip ||
		    (inodes[i] == target_ip &&
		     (VFS_I(target_ip)->i_nlink == 1 || src_is_directory))) {
			struct xfs_buf	*bp;
			xfs_agnumber_t	agno;

			agno = XFS_INO_TO_AGNO(mp, inodes[i]->i_ino);
			error = xfs_read_agi(mp, tp, agno, &bp);
			if (error)
				goto out_trans_cancel;
		}
	}

	/*
	 * Directory entry creation below may acquire the AGF. Remove
	 * the whiteout from the unlinked list first to preserve correct
	 * AGI/AGF locking order. This dirties the transaction so failures
	 * after this point will abort and log recovery will clean up the
	 * mess.
	 *
	 * For whiteouts, we need to bump the link count on the whiteout
	 * inode. After this point, we have a real link, clear the tmpfile
	 * state flag from the inode so it doesn't accidentally get misused
	 * in future.
	 */
	if (wip) {
		struct xfs_perag	*pag;

		ASSERT(VFS_I(wip)->i_nlink == 0);

		pag = xfs_perag_get(mp, XFS_INO_TO_AGNO(mp, wip->i_ino));
		error = xfs_iunlink_remove(tp, pag, wip);
		xfs_perag_put(pag);
		if (error)
			goto out_trans_cancel;

		xfs_bumplink(tp, wip);
		VFS_I(wip)->i_state &= ~I_LINKABLE;
	}

	/*
	 * Set up the target.
	 */
	if (target_ip == NULL) {
		/*
		 * If target does not exist and the rename crosses
		 * directories, adjust the target directory link count
		 * to account for the ".." reference from the new entry.
		 */
		error = xfs_dir_createname(tp, target_dp, target_name,
					   src_ip->i_ino, spaceres);
		if (error)
			goto out_trans_cancel;

		xfs_trans_ichgtime(tp, target_dp,
					XFS_ICHGTIME_MOD | XFS_ICHGTIME_CHG);

		if (new_parent && src_is_directory) {
			xfs_bumplink(tp, target_dp);
		}
	} else { /* target_ip != NULL */
		/*
		 * Link the source inode under the target name.
		 * If the source inode is a directory and we are moving
		 * it across directories, its ".." entry will be
		 * inconsistent until we replace that down below.
		 *
		 * In case there is already an entry with the same
		 * name at the destination directory, remove it first.
		 */
		error = xfs_dir_replace(tp, target_dp, target_name,
					src_ip->i_ino, spaceres);
		if (error)
			goto out_trans_cancel;

		xfs_trans_ichgtime(tp, target_dp,
					XFS_ICHGTIME_MOD | XFS_ICHGTIME_CHG);

		/*
		 * Decrement the link count on the target since the target
		 * dir no longer points to it.
		 */
		error = xfs_droplink(tp, target_ip);
		if (error)
			goto out_trans_cancel;

		if (src_is_directory) {
			/*
			 * Drop the link from the old "." entry.
			 */
			error = xfs_droplink(tp, target_ip);
			if (error)
				goto out_trans_cancel;
		}
	} /* target_ip != NULL */

	/*
	 * Remove the source.
	 */
	if (new_parent && src_is_directory) {
		/*
		 * Rewrite the ".." entry to point to the new
		 * directory.
		 */
		error = xfs_dir_replace(tp, src_ip, &xfs_name_dotdot,
					target_dp->i_ino, spaceres);
		ASSERT(error != -EEXIST);
		if (error)
			goto out_trans_cancel;
	}

	/*
	 * We always want to hit the ctime on the source inode.
	 *
	 * This isn't strictly required by the standards since the source
	 * inode isn't really being changed, but old unix file systems did
	 * it and some incremental backup programs won't work without it.
	 */
	xfs_trans_ichgtime(tp, src_ip, XFS_ICHGTIME_CHG);
	xfs_trans_log_inode(tp, src_ip, XFS_ILOG_CORE);

	/*
	 * Adjust the link count on src_dp.  This is necessary when
	 * renaming a directory, either within one parent when
	 * the target existed, or across two parent directories.
	 */
	if (src_is_directory && (new_parent || target_ip != NULL)) {

		/*
		 * Decrement link count on src_directory since the
		 * entry that's moved no longer points to it.
		 */
		error = xfs_droplink(tp, src_dp);
		if (error)
			goto out_trans_cancel;
	}

	/*
	 * For whiteouts, we only need to update the source dirent with the
	 * inode number of the whiteout inode rather than removing it
	 * altogether.
	 */
	if (wip) {
		error = xfs_dir_replace(tp, src_dp, src_name, wip->i_ino,
					spaceres);
	} else {
		/*
		 * NOTE: We don't need to check for extent count overflow here
		 * because the dir remove name code will leave the dir block in
		 * place if the extent count would overflow.
		 */
		error = xfs_dir_removename(tp, src_dp, src_name, src_ip->i_ino,
					   spaceres);
	}

	if (error)
		goto out_trans_cancel;

	xfs_trans_ichgtime(tp, src_dp, XFS_ICHGTIME_MOD | XFS_ICHGTIME_CHG);
	xfs_trans_log_inode(tp, src_dp, XFS_ILOG_CORE);
	if (new_parent)
		xfs_trans_log_inode(tp, target_dp, XFS_ILOG_CORE);

	error = xfs_finish_rename(tp);
	if (wip)
		xfs_irele(wip);
	return error;

out_trans_cancel:
	xfs_trans_cancel(tp);
out_release_wip:
	if (wip)
		xfs_irele(wip);
	return error;
}

static int
xfs_iflush(
	struct xfs_inode	*ip,
	struct xfs_buf		*bp)
{
	struct xfs_inode_log_item *iip = ip->i_itemp;
	struct xfs_dinode	*dip;
	struct xfs_mount	*mp = ip->i_mount;
	int			error;

	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL|XFS_ILOCK_SHARED));
	ASSERT(xfs_iflags_test(ip, XFS_IFLUSHING));
	ASSERT(ip->i_df.if_format != XFS_DINODE_FMT_BTREE ||
	       ip->i_df.if_nextents > XFS_IFORK_MAXEXT(ip, XFS_DATA_FORK));
	ASSERT(iip->ili_item.li_buf == bp);

	dip = xfs_buf_offset(bp, ip->i_imap.im_boffset);

	/*
	 * We don't flush the inode if any of the following checks fail, but we
	 * do still update the log item and attach to the backing buffer as if
	 * the flush happened. This is a formality to facilitate predictable
	 * error handling as the caller will shutdown and fail the buffer.
	 */
	error = -EFSCORRUPTED;
	if (XFS_TEST_ERROR(dip->di_magic != cpu_to_be16(XFS_DINODE_MAGIC),
			       mp, XFS_ERRTAG_IFLUSH_1)) {
		xfs_alert_tag(mp, XFS_PTAG_IFLUSH,
			"%s: Bad inode %Lu magic number 0x%x, ptr "PTR_FMT,
			__func__, ip->i_ino, be16_to_cpu(dip->di_magic), dip);
		goto flush_out;
	}
	if (S_ISREG(VFS_I(ip)->i_mode)) {
		if (XFS_TEST_ERROR(
		    ip->i_df.if_format != XFS_DINODE_FMT_EXTENTS &&
		    ip->i_df.if_format != XFS_DINODE_FMT_BTREE,
		    mp, XFS_ERRTAG_IFLUSH_3)) {
			xfs_alert_tag(mp, XFS_PTAG_IFLUSH,
				"%s: Bad regular inode %Lu, ptr "PTR_FMT,
				__func__, ip->i_ino, ip);
			goto flush_out;
		}
	} else if (S_ISDIR(VFS_I(ip)->i_mode)) {
		if (XFS_TEST_ERROR(
		    ip->i_df.if_format != XFS_DINODE_FMT_EXTENTS &&
		    ip->i_df.if_format != XFS_DINODE_FMT_BTREE &&
		    ip->i_df.if_format != XFS_DINODE_FMT_LOCAL,
		    mp, XFS_ERRTAG_IFLUSH_4)) {
			xfs_alert_tag(mp, XFS_PTAG_IFLUSH,
				"%s: Bad directory inode %Lu, ptr "PTR_FMT,
				__func__, ip->i_ino, ip);
			goto flush_out;
		}
	}
	if (XFS_TEST_ERROR(ip->i_df.if_nextents + xfs_ifork_nextents(ip->i_afp) >
				ip->i_nblocks, mp, XFS_ERRTAG_IFLUSH_5)) {
		xfs_alert_tag(mp, XFS_PTAG_IFLUSH,
			"%s: detected corrupt incore inode %Lu, "
			"total extents = %d, nblocks = %Ld, ptr "PTR_FMT,
			__func__, ip->i_ino,
			ip->i_df.if_nextents + xfs_ifork_nextents(ip->i_afp),
			ip->i_nblocks, ip);
		goto flush_out;
	}
	if (XFS_TEST_ERROR(ip->i_forkoff > mp->m_sb.sb_inodesize,
				mp, XFS_ERRTAG_IFLUSH_6)) {
		xfs_alert_tag(mp, XFS_PTAG_IFLUSH,
			"%s: bad inode %Lu, forkoff 0x%x, ptr "PTR_FMT,
			__func__, ip->i_ino, ip->i_forkoff, ip);
		goto flush_out;
	}

	/*
	 * Inode item log recovery for v2 inodes are dependent on the flushiter
	 * count for correct sequencing.  We bump the flush iteration count so
	 * we can detect flushes which postdate a log record during recovery.
	 * This is redundant as we now log every change and hence this can't
	 * happen but we need to still do it to ensure backwards compatibility
	 * with old kernels that predate logging all inode changes.
	 */
	if (!xfs_has_v3inodes(mp))
		ip->i_flushiter++;

	/*
	 * If there are inline format data / attr forks attached to this inode,
	 * make sure they are not corrupt.
	 */
	if (ip->i_df.if_format == XFS_DINODE_FMT_LOCAL &&
	    xfs_ifork_verify_local_data(ip))
		goto flush_out;
	if (ip->i_afp && ip->i_afp->if_format == XFS_DINODE_FMT_LOCAL &&
	    xfs_ifork_verify_local_attr(ip))
		goto flush_out;

	/*
	 * Copy the dirty parts of the inode into the on-disk inode.  We always
	 * copy out the core of the inode, because if the inode is dirty at all
	 * the core must be.
	 */
	xfs_inode_to_disk(ip, dip, iip->ili_item.li_lsn);

	/* Wrap, we never let the log put out DI_MAX_FLUSH */
	if (!xfs_has_v3inodes(mp)) {
		if (ip->i_flushiter == DI_MAX_FLUSH)
			ip->i_flushiter = 0;
	}

	xfs_iflush_fork(ip, dip, iip, XFS_DATA_FORK);
	if (XFS_IFORK_Q(ip))
		xfs_iflush_fork(ip, dip, iip, XFS_ATTR_FORK);

	/*
	 * We've recorded everything logged in the inode, so we'd like to clear
	 * the ili_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 ili_fields field.
	 * In the xfs_buf_inode_iodone() 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 ili_fields and ili_last_fields will be cleared.
	 */
	error = 0;
flush_out:
	spin_lock(&iip->ili_lock);
	iip->ili_last_fields = iip->ili_fields;
	iip->ili_fields = 0;
	iip->ili_fsync_fields = 0;
	spin_unlock(&iip->ili_lock);

	/*
	 * Store the current LSN of the inode so that we can tell whether the
	 * item has moved in the AIL from xfs_buf_inode_iodone().
	 */
	xfs_trans_ail_copy_lsn(mp->m_ail, &iip->ili_flush_lsn,
				&iip->ili_item.li_lsn);

	/* generate the checksum. */
	xfs_dinode_calc_crc(mp, dip);
	return error;
}

/*
 * Non-blocking flush of dirty inode metadata into the backing buffer.
 *
 * The caller must have a reference to the inode and hold the cluster buffer
 * locked. The function will walk across all the inodes on the cluster buffer it
 * can find and lock without blocking, and flush them to the cluster buffer.
 *
 * On successful flushing of at least one inode, the caller must write out the
 * buffer and release it. If no inodes are flushed, -EAGAIN will be returned and
 * the caller needs to release the buffer. On failure, the filesystem will be
 * shut down, the buffer will have been unlocked and released, and EFSCORRUPTED
 * will be returned.
 */
int
xfs_iflush_cluster(
	struct xfs_buf		*bp)
{
	struct xfs_mount	*mp = bp->b_mount;
	struct xfs_log_item	*lip, *n;
	struct xfs_inode	*ip;
	struct xfs_inode_log_item *iip;
	int			clcount = 0;
	int			error = 0;

	/*
	 * We must use the safe variant here as on shutdown xfs_iflush_abort()
	 * can remove itself from the list.
	 */
	list_for_each_entry_safe(lip, n, &bp->b_li_list, li_bio_list) {
		iip = (struct xfs_inode_log_item *)lip;
		ip = iip->ili_inode;

		/*
		 * Quick and dirty check to avoid locks if possible.
		 */
		if (__xfs_iflags_test(ip, XFS_IRECLAIM | XFS_IFLUSHING))
			continue;
		if (xfs_ipincount(ip))
			continue;

		/*
		 * The inode is still attached to the buffer, which means it is
		 * dirty but reclaim might try to grab it. Check carefully for
		 * that, and grab the ilock while still holding the i_flags_lock
		 * to guarantee reclaim will not be able to reclaim this inode
		 * once we drop the i_flags_lock.
		 */
		spin_lock(&ip->i_flags_lock);
		ASSERT(!__xfs_iflags_test(ip, XFS_ISTALE));
		if (__xfs_iflags_test(ip, XFS_IRECLAIM | XFS_IFLUSHING)) {
			spin_unlock(&ip->i_flags_lock);
			continue;
		}

		/*
		 * ILOCK will pin the inode against reclaim and prevent
		 * concurrent transactions modifying the inode while we are
		 * flushing the inode. If we get the lock, set the flushing
		 * state before we drop the i_flags_lock.
		 */
		if (!xfs_ilock_nowait(ip, XFS_ILOCK_SHARED)) {
			spin_unlock(&ip->i_flags_lock);
			continue;
		}
		__xfs_iflags_set(ip, XFS_IFLUSHING);
		spin_unlock(&ip->i_flags_lock);

		/*
		 * Abort flushing this inode if we are shut down because the
		 * inode may not currently be in the AIL. This can occur when
		 * log I/O failure unpins the inode without inserting into the
		 * AIL, leaving a dirty/unpinned inode attached to the buffer
		 * that otherwise looks like it should be flushed.
		 */
		if (xfs_is_shutdown(mp)) {
			xfs_iunpin_wait(ip);
			xfs_iflush_abort(ip);
			xfs_iunlock(ip, XFS_ILOCK_SHARED);
			error = -EIO;
			continue;
		}

		/* don't block waiting on a log force to unpin dirty inodes */
		if (xfs_ipincount(ip)) {
			xfs_iflags_clear(ip, XFS_IFLUSHING);
			xfs_iunlock(ip, XFS_ILOCK_SHARED);
			continue;
		}

		if (!xfs_inode_clean(ip))
			error = xfs_iflush(ip, bp);
		else
			xfs_iflags_clear(ip, XFS_IFLUSHING);
		xfs_iunlock(ip, XFS_ILOCK_SHARED);
		if (error)
			break;
		clcount++;
	}

	if (error) {
		bp->b_flags |= XBF_ASYNC;
		xfs_buf_ioend_fail(bp);
		xfs_force_shutdown(mp, SHUTDOWN_CORRUPT_INCORE);
		return error;
	}

	if (!clcount)
		return -EAGAIN;

	XFS_STATS_INC(mp, xs_icluster_flushcnt);
	XFS_STATS_ADD(mp, xs_icluster_flushinode, clcount);
	return 0;

}

/* Release an inode. */
void
xfs_irele(
	struct xfs_inode	*ip)
{
	trace_xfs_irele(ip, _RET_IP_);
	iput(VFS_I(ip));
}

/*
 * Ensure all commited transactions touching the inode are written to the log.
 */
int
xfs_log_force_inode(
	struct xfs_inode	*ip)
{
	xfs_csn_t		seq = 0;

	xfs_ilock(ip, XFS_ILOCK_SHARED);
	if (xfs_ipincount(ip))
		seq = ip->i_itemp->ili_commit_seq;
	xfs_iunlock(ip, XFS_ILOCK_SHARED);

	if (!seq)
		return 0;
	return xfs_log_force_seq(ip->i_mount, seq, XFS_LOG_SYNC, NULL);
}

/*
 * Grab the exclusive iolock for a data copy from src to dest, making sure to
 * abide vfs locking order (lowest pointer value goes first) and breaking the
 * layout leases before proceeding.  The loop is needed because we cannot call
 * the blocking break_layout() with the iolocks held, and therefore have to
 * back out both locks.
 */
static int
xfs_iolock_two_inodes_and_break_layout(
	struct inode		*src,
	struct inode		*dest)
{
	int			error;

	if (src > dest)
		swap(src, dest);

retry:
	/* Wait to break both inodes' layouts before we start locking. */
	error = break_layout(src, true);
	if (error)
		return error;
	if (src != dest) {
		error = break_layout(dest, true);
		if (error)
			return error;
	}

	/* Lock one inode and make sure nobody got in and leased it. */
	inode_lock(src);
	error = break_layout(src, false);
	if (error) {
		inode_unlock(src);
		if (error == -EWOULDBLOCK)
			goto retry;
		return error;
	}

	if (src == dest)
		return 0;

	/* Lock the other inode and make sure nobody got in and leased it. */
	inode_lock_nested(dest, I_MUTEX_NONDIR2);
	error = break_layout(dest, false);
	if (error) {
		inode_unlock(src);
		inode_unlock(dest);
		if (error == -EWOULDBLOCK)
			goto retry;
		return error;
	}

	return 0;
}

/*
 * Lock two inodes so that userspace cannot initiate I/O via file syscalls or
 * mmap activity.
 */
int
xfs_ilock2_io_mmap(
	struct xfs_inode	*ip1,
	struct xfs_inode	*ip2)
{
	int			ret;

	ret = xfs_iolock_two_inodes_and_break_layout(VFS_I(ip1), VFS_I(ip2));
	if (ret)
		return ret;
	filemap_invalidate_lock_two(VFS_I(ip1)->i_mapping,
				    VFS_I(ip2)->i_mapping);
	return 0;
}

/* Unlock both inodes to allow IO and mmap activity. */
void
xfs_iunlock2_io_mmap(
	struct xfs_inode	*ip1,
	struct xfs_inode	*ip2)
{
	filemap_invalidate_unlock_two(VFS_I(ip1)->i_mapping,
				      VFS_I(ip2)->i_mapping);
	inode_unlock(VFS_I(ip2));
	if (ip1 != ip2)
		inode_unlock(VFS_I(ip1));
}