1 files changed, 967 insertions, 37 deletions

diff --git a/fs/xfs/xfs_inode.c b/fs/xfs/xfs_inode.c
index 1029f56..73a7a1d 100644
--- a/fs/xfs/xfs_inode.c
+++ b/fs/xfs/xfs_inode.c
@@ -23,14 +23,19 @@
 #include "xfs_log.h"
 #include "xfs_inum.h"
 #include "xfs_trans.h"
+#include "xfs_trans_space.h"
 #include "xfs_trans_priv.h"
 #include "xfs_sb.h"
 #include "xfs_ag.h"
 #include "xfs_mount.h"
+#include "xfs_da_btree.h"
+#include "xfs_dir2_format.h"
+#include "xfs_dir2.h"
 #include "xfs_bmap_btree.h"
 #include "xfs_alloc_btree.h"
 #include "xfs_ialloc_btree.h"
 #include "xfs_attr_sf.h"
+#include "xfs_attr.h"
 #include "xfs_dinode.h"
 #include "xfs_inode.h"
 #include "xfs_buf_item.h"
@@ -44,10 +49,10 @@
 #include "xfs_utils.h"
 #include "xfs_quota.h"
 #include "xfs_filestream.h"
-#include "xfs_vnodeops.h"
 #include "xfs_cksum.h"
 #include "xfs_trace.h"
 #include "xfs_icache.h"
+#include "xfs_symlink.h"
 
 kmem_zone_t *xfs_inode_zone;
 
@@ -307,6 +312,188 @@ xfs_isilocked(
 }
 #endif
 
+#ifdef DEBUG
+int xfs_locked_n;
+int xfs_small_retries;
+int xfs_middle_retries;
+int xfs_lots_retries;
+int xfs_lock_delays;
+#endif
+
+/*
+ * Bump the subclass so xfs_lock_inodes() acquires each lock with
+ * a different value
+ */
+static inline int
+xfs_lock_inumorder(int lock_mode, int subclass)
+{
+	if (lock_mode & (XFS_IOLOCK_SHARED|XFS_IOLOCK_EXCL))
+		lock_mode |= (subclass + XFS_LOCK_INUMORDER) << XFS_IOLOCK_SHIFT;
+	if (lock_mode & (XFS_ILOCK_SHARED|XFS_ILOCK_EXCL))
+		lock_mode |= (subclass + XFS_LOCK_INUMORDER) << XFS_ILOCK_SHIFT;
+
+	return lock_mode;
+}
+
+/*
+ * 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.
+ */
+void
+xfs_lock_inodes(
+	xfs_inode_t	**ips,
+	int		inodes,
+	uint		lock_mode)
+{
+	int		attempts = 0, i, j, try_lock;
+	xfs_log_item_t	*lp;
+
+	ASSERT(ips && (inodes >= 2)); /* we need at least two */
+
+	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 = (xfs_log_item_t *)ips[j]->i_itemp;
+				if (lp && (lp->li_flags & XFS_LI_IN_AIL)) {
+					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) {
+			/* try_lock must be 0 if i is 0. */
+			/*
+			 * 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))) {
+				attempts++;
+
+				/*
+				 * Unlock all previous guys and try again.
+				 * xfs_iunlock will try to push the tail
+				 * if the inode is in the AIL.
+				 */
+
+				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 */
+#ifdef DEBUG
+					xfs_lock_delays++;
+#endif
+				}
+				i = 0;
+				try_lock = 0;
+				goto again;
+			}
+		} else {
+			xfs_ilock(ips[i], xfs_lock_inumorder(lock_mode, i));
+		}
+	}
+
+#ifdef DEBUG
+	if (attempts) {
+		if (attempts < 5) xfs_small_retries++;
+		else if (attempts < 100) xfs_middle_retries++;
+		else xfs_lots_retries++;
+	} else {
+		xfs_locked_n++;
+	}
+#endif
+}
+
+/*
+ * xfs_lock_two_inodes() can only be used to lock one type of lock
+ * at a time - the iolock or the ilock, but not both at once. If
+ * we lock both at once, lockdep will report false positives saying
+ * we have violated locking orders.
+ */
+void
+xfs_lock_two_inodes(
+	xfs_inode_t		*ip0,
+	xfs_inode_t		*ip1,
+	uint			lock_mode)
+{
+	xfs_inode_t		*temp;
+	int			attempts = 0;
+	xfs_log_item_t		*lp;
+
+	if (lock_mode & (XFS_IOLOCK_SHARED|XFS_IOLOCK_EXCL))
+		ASSERT((lock_mode & (XFS_ILOCK_SHARED|XFS_ILOCK_EXCL)) == 0);
+	ASSERT(ip0->i_ino != ip1->i_ino);
+
+	if (ip0->i_ino > ip1->i_ino) {
+		temp = ip0;
+		ip0 = ip1;
+		ip1 = temp;
+	}
+
+ again:
+	xfs_ilock(ip0, xfs_lock_inumorder(lock_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 = (xfs_log_item_t *)ip0->i_itemp;
+	if (lp && (lp->li_flags & XFS_LI_IN_AIL)) {
+		if (!xfs_ilock_nowait(ip1, xfs_lock_inumorder(lock_mode, 1))) {
+			xfs_iunlock(ip0, lock_mode);
+			if ((++attempts % 5) == 0)
+				delay(1); /* Don't just spin the CPU */
+			goto again;
+		}
+	} else {
+		xfs_ilock(ip1, xfs_lock_inumorder(lock_mode, 1));
+	}
+}
+
+
 void
 __xfs_iflock(
 	struct xfs_inode	*ip)
@@ -382,6 +569,49 @@ xfs_dic2xflags(
 }
 
 /*
+ * 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;
+	uint			lock_mode;
+
+	trace_xfs_lookup(dp, name);
+
+	if (XFS_FORCED_SHUTDOWN(dp->i_mount))
+		return XFS_ERROR(EIO);
+
+	lock_mode = xfs_ilock_map_shared(dp);
+	error = xfs_dir_lookup(NULL, dp, name, &inum, ci_name);
+	xfs_iunlock_map_shared(dp, lock_mode);
+
+	if (error)
+		goto out;
+
+	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:
+	*ipp = NULL;
+	return error;
+}
+
+/*
  * Allocate an inode on disk and return a copy of its in-core version.
  * The in-core inode is locked exclusively.  Set mode, nlink, and rdev
  * appropriately within the inode.  The uid and gid for the inode are
@@ -632,6 +862,308 @@ xfs_ialloc(
 	return 0;
 }
 
+int
+xfs_create(
+	xfs_inode_t		*dp,
+	struct xfs_name		*name,
+	umode_t			mode,
+	xfs_dev_t		rdev,
+	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;
+	xfs_bmap_free_t		free_list;
+	xfs_fsblock_t		first_block;
+	bool                    unlock_dp_on_error = false;
+	uint			cancel_flags;
+	int			committed;
+	prid_t			prid;
+	struct xfs_dquot	*udqp = NULL;
+	struct xfs_dquot	*gdqp = NULL;
+	struct xfs_dquot	*pdqp = NULL;
+	uint			resblks;
+	uint			log_res;
+	uint			log_count;
+
+	trace_xfs_create(dp, name);
+
+	if (XFS_FORCED_SHUTDOWN(mp))
+		return XFS_ERROR(EIO);
+
+	if (dp->i_d.di_flags & XFS_DIFLAG_PROJINHERIT)
+		prid = xfs_get_projid(dp);
+	else
+		prid = XFS_PROJID_DEFAULT;
+
+	/*
+	 * Make sure that we have allocated dquot(s) on disk.
+	 */
+	error = xfs_qm_vop_dqalloc(dp, current_fsuid(), current_fsgid(), prid,
+					XFS_QMOPT_QUOTALL | XFS_QMOPT_INHERIT,
+					&udqp, &gdqp, &pdqp);
+	if (error)
+		return error;
+
+	if (is_dir) {
+		rdev = 0;
+		resblks = XFS_MKDIR_SPACE_RES(mp, name->len);
+		log_res = XFS_MKDIR_LOG_RES(mp);
+		log_count = XFS_MKDIR_LOG_COUNT;
+		tp = xfs_trans_alloc(mp, XFS_TRANS_MKDIR);
+	} else {
+		resblks = XFS_CREATE_SPACE_RES(mp, name->len);
+		log_res = XFS_CREATE_LOG_RES(mp);
+		log_count = XFS_CREATE_LOG_COUNT;
+		tp = xfs_trans_alloc(mp, XFS_TRANS_CREATE);
+	}
+
+	cancel_flags = XFS_TRANS_RELEASE_LOG_RES;
+
+	/*
+	 * 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_reserve(tp, resblks, log_res, 0,
+			XFS_TRANS_PERM_LOG_RES, log_count);
+	if (error == ENOSPC) {
+		/* flush outstanding delalloc blocks and retry */
+		xfs_flush_inodes(mp);
+		error = xfs_trans_reserve(tp, resblks, log_res, 0,
+				XFS_TRANS_PERM_LOG_RES, log_count);
+	}
+	if (error == ENOSPC) {
+		/* No space at all so try a "no-allocation" reservation */
+		resblks = 0;
+		error = xfs_trans_reserve(tp, 0, log_res, 0,
+				XFS_TRANS_PERM_LOG_RES, log_count);
+	}
+	if (error) {
+		cancel_flags = 0;
+		goto out_trans_cancel;
+	}
+
+	xfs_ilock(dp, XFS_ILOCK_EXCL | XFS_ILOCK_PARENT);
+	unlock_dp_on_error = true;
+
+	xfs_bmap_init(&free_list, &first_block);
+
+	/*
+	 * Reserve disk quota and the inode.
+	 */
+	error = xfs_trans_reserve_quota(tp, mp, udqp, gdqp,
+						pdqp, resblks, 1, 0);
+	if (error)
+		goto out_trans_cancel;
+
+	error = xfs_dir_canenter(tp, dp, name, resblks);
+	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_dir_ialloc(&tp, dp, mode, is_dir ? 2 : 1, rdev,
+			       prid, resblks > 0, &ip, &committed);
+	if (error) {
+		if (error == ENOSPC)
+			goto out_trans_cancel;
+		goto out_trans_abort;
+	}
+
+	/*
+	 * Now we join the directory inode to the transaction.  We do not do it
+	 * earlier because xfs_dir_ialloc 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,
+					&first_block, &free_list, resblks ?
+					resblks - XFS_IALLOC_SPACE_RES(mp) : 0);
+	if (error) {
+		ASSERT(error != ENOSPC);
+		goto out_trans_abort;
+	}
+	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_bmap_cancel;
+
+		error = xfs_bumplink(tp, dp);
+		if (error)
+			goto out_bmap_cancel;
+	}
+
+	/*
+	 * If this is a synchronous mount, make sure that the
+	 * create transaction goes to disk before returning to
+	 * the user.
+	 */
+	if (mp->m_flags & (XFS_MOUNT_WSYNC|XFS_MOUNT_DIRSYNC))
+		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_bmap_finish(&tp, &free_list, &committed);
+	if (error)
+		goto out_bmap_cancel;
+
+	error = xfs_trans_commit(tp, XFS_TRANS_RELEASE_LOG_RES);
+	if (error)
+		goto out_release_inode;
+
+	xfs_qm_dqrele(udqp);
+	xfs_qm_dqrele(gdqp);
+	xfs_qm_dqrele(pdqp);
+
+	*ipp = ip;
+	return 0;
+
+ out_bmap_cancel:
+	xfs_bmap_cancel(&free_list);
+ out_trans_abort:
+	cancel_flags |= XFS_TRANS_ABORT;
+ out_trans_cancel:
+	xfs_trans_cancel(tp, cancel_flags);
+ out_release_inode:
+	/*
+	 * Wait until after the current transaction is aborted to
+	 * release the inode.  This prevents recursive transactions
+	 * and deadlocks from xfs_inactive.
+	 */
+	if (ip)
+		IRELE(ip);
+
+	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_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;
+	xfs_bmap_free_t         free_list;
+	xfs_fsblock_t           first_block;
+	int			cancel_flags;
+	int			committed;
+	int			resblks;
+
+	trace_xfs_link(tdp, target_name);
+
+	ASSERT(!S_ISDIR(sip->i_d.di_mode));
+
+	if (XFS_FORCED_SHUTDOWN(mp))
+		return XFS_ERROR(EIO);
+
+	error = xfs_qm_dqattach(sip, 0);
+	if (error)
+		goto std_return;
+
+	error = xfs_qm_dqattach(tdp, 0);
+	if (error)
+		goto std_return;
+
+	tp = xfs_trans_alloc(mp, XFS_TRANS_LINK);
+	cancel_flags = XFS_TRANS_RELEASE_LOG_RES;
+	resblks = XFS_LINK_SPACE_RES(mp, target_name->len);
+	error = xfs_trans_reserve(tp, resblks, XFS_LINK_LOG_RES(mp), 0,
+			XFS_TRANS_PERM_LOG_RES, XFS_LINK_LOG_COUNT);
+	if (error == ENOSPC) {
+		resblks = 0;
+		error = xfs_trans_reserve(tp, 0, XFS_LINK_LOG_RES(mp), 0,
+				XFS_TRANS_PERM_LOG_RES, XFS_LINK_LOG_COUNT);
+	}
+	if (error) {
+		cancel_flags = 0;
+		goto error_return;
+	}
+
+	xfs_lock_two_inodes(sip, tdp, XFS_ILOCK_EXCL);
+
+	xfs_trans_ijoin(tp, sip, XFS_ILOCK_EXCL);
+	xfs_trans_ijoin(tp, tdp, XFS_ILOCK_EXCL);
+
+	/*
+	 * 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_d.di_flags & XFS_DIFLAG_PROJINHERIT) &&
+		     (xfs_get_projid(tdp) != xfs_get_projid(sip)))) {
+		error = XFS_ERROR(EXDEV);
+		goto error_return;
+	}
+
+	error = xfs_dir_canenter(tp, tdp, target_name, resblks);
+	if (error)
+		goto error_return;
+
+	xfs_bmap_init(&free_list, &first_block);
+
+	error = xfs_dir_createname(tp, tdp, target_name, sip->i_ino,
+					&first_block, &free_list, resblks);
+	if (error)
+		goto abort_return;
+	xfs_trans_ichgtime(tp, tdp, XFS_ICHGTIME_MOD | XFS_ICHGTIME_CHG);
+	xfs_trans_log_inode(tp, tdp, XFS_ILOG_CORE);
+
+	error = xfs_bumplink(tp, sip);
+	if (error)
+		goto abort_return;
+
+	/*
+	 * If this is a synchronous mount, make sure that the
+	 * link transaction goes to disk before returning to
+	 * the user.
+	 */
+	if (mp->m_flags & (XFS_MOUNT_WSYNC|XFS_MOUNT_DIRSYNC)) {
+		xfs_trans_set_sync(tp);
+	}
+
+	error = xfs_bmap_finish (&tp, &free_list, &committed);
+	if (error) {
+		xfs_bmap_cancel(&free_list);
+		goto abort_return;
+	}
+
+	return xfs_trans_commit(tp, XFS_TRANS_RELEASE_LOG_RES);
+
+ abort_return:
+	cancel_flags |= XFS_TRANS_ABORT;
+ error_return:
+	xfs_trans_cancel(tp, cancel_flags);
+ std_return:
+	return error;
+}
+
 /*
  * 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
@@ -771,6 +1303,276 @@ out_bmap_cancel:
 	goto out;
 }
 
+int
+xfs_release(
+	xfs_inode_t	*ip)
+{
+	xfs_mount_t	*mp = ip->i_mount;
+	int		error;
+
+	if (!S_ISREG(ip->i_d.di_mode) || (ip->i_d.di_mode == 0))
+		return 0;
+
+	/* If this is a read-only mount, don't do this (would generate I/O) */
+	if (mp->m_flags & XFS_MOUNT_RDONLY)
+		return 0;
+
+	if (!XFS_FORCED_SHUTDOWN(mp)) {
+		int truncated;
+
+		/*
+		 * If we are using filestreams, and we have an unlinked
+		 * file that we are processing the last close on, then nothing
+		 * will be able to reopen and write to this file. Purge this
+		 * inode from the filestreams cache so that it doesn't delay
+		 * teardown of the inode.
+		 */
+		if ((ip->i_d.di_nlink == 0) && xfs_inode_is_filestream(ip))
+			xfs_filestream_deassociate(ip);
+
+		/*
+		 * 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 (VN_DIRTY(VFS_I(ip)) && ip->i_delayed_blks > 0) {
+				error = -filemap_flush(VFS_I(ip)->i_mapping);
+				if (error)
+					return error;
+			}
+		}
+	}
+
+	if (ip->i_d.di_nlink == 0)
+		return 0;
+
+	if (xfs_can_free_eofblocks(ip, false)) {
+
+		/*
+		 * If we can't get the iolock just skip truncating the blocks
+		 * past EOF because we could deadlock with the mmap_sem
+		 * 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.
+		 *
+		 * Further, 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, either, 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))
+			return 0;
+
+		error = xfs_free_eofblocks(mp, ip, true);
+		if (error && error != EAGAIN)
+			return error;
+
+		/* delalloc blocks after truncation means it really is dirty */
+		if (ip->i_delayed_blks)
+			xfs_iflags_set(ip, XFS_IDIRTY_RELEASE);
+	}
+	return 0;
+}
+
+/*
+ * 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.
+ */
+int
+xfs_inactive(
+	xfs_inode_t	*ip)
+{
+	xfs_bmap_free_t	free_list;
+	xfs_fsblock_t	first_block;
+	int		committed;
+	xfs_trans_t	*tp;
+	xfs_mount_t	*mp;
+	int		error;
+	int		truncate = 0;
+
+	/*
+	 * If the inode is already free, then there can be nothing
+	 * to clean up here.
+	 */
+	if (ip->i_d.di_mode == 0 || is_bad_inode(VFS_I(ip))) {
+		ASSERT(ip->i_df.if_real_bytes == 0);
+		ASSERT(ip->i_df.if_broot_bytes == 0);
+		return VN_INACTIVE_CACHE;
+	}
+
+	mp = ip->i_mount;
+
+	error = 0;
+
+	/* If this is a read-only mount, don't do this (would generate I/O) */
+	if (mp->m_flags & XFS_MOUNT_RDONLY)
+		goto out;
+
+	if (ip->i_d.di_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.
+		 */
+		if (xfs_can_free_eofblocks(ip, true)) {
+			error = xfs_free_eofblocks(mp, ip, false);
+			if (error)
+				return VN_INACTIVE_CACHE;
+		}
+		goto out;
+	}
+
+	if (S_ISREG(ip->i_d.di_mode) &&
+	    (ip->i_d.di_size != 0 || XFS_ISIZE(ip) != 0 ||
+	     ip->i_d.di_nextents > 0 || ip->i_delayed_blks > 0))
+		truncate = 1;
+
+	error = xfs_qm_dqattach(ip, 0);
+	if (error)
+		return VN_INACTIVE_CACHE;
+
+	tp = xfs_trans_alloc(mp, XFS_TRANS_INACTIVE);
+	error = xfs_trans_reserve(tp, 0,
+			(truncate || S_ISLNK(ip->i_d.di_mode)) ?
+				XFS_ITRUNCATE_LOG_RES(mp) :
+				XFS_IFREE_LOG_RES(mp),
+			0,
+			XFS_TRANS_PERM_LOG_RES,
+			XFS_ITRUNCATE_LOG_COUNT);
+	if (error) {
+		ASSERT(XFS_FORCED_SHUTDOWN(mp));
+		xfs_trans_cancel(tp, 0);
+		return VN_INACTIVE_CACHE;
+	}
+
+	xfs_ilock(ip, XFS_ILOCK_EXCL);
+	xfs_trans_ijoin(tp, ip, 0);
+
+	if (S_ISLNK(ip->i_d.di_mode)) {
+		error = xfs_inactive_symlink(ip, &tp);
+		if (error)
+			goto out_cancel;
+	} else if (truncate) {
+		ip->i_d.di_size = 0;
+		xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
+
+		error = xfs_itruncate_extents(&tp, ip, XFS_DATA_FORK, 0);
+		if (error)
+			goto out_cancel;
+
+		ASSERT(ip->i_d.di_nextents == 0);
+	}
+
+	/*
+	 * If there are attributes associated with the file then blow them away
+	 * now.  The code calls a routine that recursively deconstructs the
+	 * attribute fork.  We need to just commit the current transaction
+	 * because we can't use it for xfs_attr_inactive().
+	 */
+	if (ip->i_d.di_anextents > 0) {
+		ASSERT(ip->i_d.di_forkoff != 0);
+
+		error = xfs_trans_commit(tp, XFS_TRANS_RELEASE_LOG_RES);
+		if (error)
+			goto out_unlock;
+
+		xfs_iunlock(ip, XFS_ILOCK_EXCL);
+
+		error = xfs_attr_inactive(ip);
+		if (error)
+			goto out;
+
+		tp = xfs_trans_alloc(mp, XFS_TRANS_INACTIVE);
+		error = xfs_trans_reserve(tp, 0,
+					  XFS_IFREE_LOG_RES(mp),
+					  0, XFS_TRANS_PERM_LOG_RES,
+					  XFS_INACTIVE_LOG_COUNT);
+		if (error) {
+			xfs_trans_cancel(tp, 0);
+			goto out;
+		}
+
+		xfs_ilock(ip, XFS_ILOCK_EXCL);
+		xfs_trans_ijoin(tp, ip, 0);
+	}
+
+	if (ip->i_afp)
+		xfs_idestroy_fork(ip, XFS_ATTR_FORK);
+
+	ASSERT(ip->i_d.di_anextents == 0);
+
+	/*
+	 * Free the inode.
+	 */
+	xfs_bmap_init(&free_list, &first_block);
+	error = xfs_ifree(tp, ip, &free_list);
+	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_FORCED_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, XFS_TRANS_RELEASE_LOG_RES|XFS_TRANS_ABORT);
+	} else {
+		/*
+		 * 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_bmap_finish(&tp,  &free_list, &committed);
+		if (error)
+			xfs_notice(mp, "%s: xfs_bmap_finish returned error %d",
+				__func__, error);
+		error = xfs_trans_commit(tp, XFS_TRANS_RELEASE_LOG_RES);
+		if (error)
+			xfs_notice(mp, "%s: xfs_trans_commit returned error %d",
+				__func__, error);
+	}
+
+	/*
+	 * Release the dquots held by inode, if any.
+	 */
+	xfs_qm_dqdetach(ip);
+out_unlock:
+	xfs_iunlock(ip, XFS_ILOCK_EXCL);
+out:
+	return VN_INACTIVE_CACHE;
+out_cancel:
+	xfs_trans_cancel(tp, XFS_TRANS_RELEASE_LOG_RES | XFS_TRANS_ABORT);
+	goto out_unlock;
+}
+
 /*
  * This is called when the inode's link count goes to 0.
  * We place the on-disk inode on a list in the AGI.  It
@@ -1302,6 +2104,170 @@ xfs_iunpin_wait(
 		__xfs_iunpin_wait(ip);
 }
 
+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(ip->i_d.di_mode);
+	int                     error = 0;
+	xfs_bmap_free_t         free_list;
+	xfs_fsblock_t           first_block;
+	int			cancel_flags;
+	int			committed;
+	int			link_zero;
+	uint			resblks;
+	uint			log_count;
+
+	trace_xfs_remove(dp, name);
+
+	if (XFS_FORCED_SHUTDOWN(mp))
+		return XFS_ERROR(EIO);
+
+	error = xfs_qm_dqattach(dp, 0);
+	if (error)
+		goto std_return;
+
+	error = xfs_qm_dqattach(ip, 0);
+	if (error)
+		goto std_return;
+
+	if (is_dir) {
+		tp = xfs_trans_alloc(mp, XFS_TRANS_RMDIR);
+		log_count = XFS_DEFAULT_LOG_COUNT;
+	} else {
+		tp = xfs_trans_alloc(mp, XFS_TRANS_REMOVE);
+		log_count = XFS_REMOVE_LOG_COUNT;
+	}
+	cancel_flags = XFS_TRANS_RELEASE_LOG_RES;
+
+	/*
+	 * 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_reserve(tp, resblks, XFS_REMOVE_LOG_RES(mp), 0,
+				  XFS_TRANS_PERM_LOG_RES, log_count);
+	if (error == ENOSPC) {
+		resblks = 0;
+		error = xfs_trans_reserve(tp, 0, XFS_REMOVE_LOG_RES(mp), 0,
+					  XFS_TRANS_PERM_LOG_RES, log_count);
+	}
+	if (error) {
+		ASSERT(error != ENOSPC);
+		cancel_flags = 0;
+		goto out_trans_cancel;
+	}
+
+	xfs_lock_two_inodes(dp, 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(ip->i_d.di_nlink >= 2);
+		if (ip->i_d.di_nlink != 2) {
+			error = XFS_ERROR(ENOTEMPTY);
+			goto out_trans_cancel;
+		}
+		if (!xfs_dir_isempty(ip)) {
+			error = XFS_ERROR(ENOTEMPTY);
+			goto out_trans_cancel;
+		}
+	}
+
+	xfs_bmap_init(&free_list, &first_block);
+	error = xfs_dir_removename(tp, dp, name, ip->i_ino,
+					&first_block, &free_list, resblks);
+	if (error) {
+		ASSERT(error != ENOENT);
+		goto out_bmap_cancel;
+	}
+	xfs_trans_ichgtime(tp, dp, XFS_ICHGTIME_MOD | XFS_ICHGTIME_CHG);
+
+	if (is_dir) {
+		/*
+		 * Drop the link from ip's "..".
+		 */
+		error = xfs_droplink(tp, dp);
+		if (error)
+			goto out_bmap_cancel;
+
+		/*
+		 * Drop the "." link from ip to self.
+		 */
+		error = xfs_droplink(tp, ip);
+		if (error)
+			goto out_bmap_cancel;
+	} 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);
+	}
+
+	/*
+	 * Drop the link from dp to ip.
+	 */
+	error = xfs_droplink(tp, ip);
+	if (error)
+		goto out_bmap_cancel;
+
+	/*
+	 * Determine if this is the last link while
+	 * we are in the transaction.
+	 */
+	link_zero = (ip->i_d.di_nlink == 0);
+
+	/*
+	 * If this is a synchronous mount, make sure that the
+	 * remove transaction goes to disk before returning to
+	 * the user.
+	 */
+	if (mp->m_flags & (XFS_MOUNT_WSYNC|XFS_MOUNT_DIRSYNC))
+		xfs_trans_set_sync(tp);
+
+	error = xfs_bmap_finish(&tp, &free_list, &committed);
+	if (error)
+		goto out_bmap_cancel;
+
+	error = xfs_trans_commit(tp, XFS_TRANS_RELEASE_LOG_RES);
+	if (error)
+		goto std_return;
+
+	/*
+	 * If we are using filestreams, kill the stream association.
+	 * If the file is still open it may get a new one but that
+	 * will get killed on last close in xfs_close() so we don't
+	 * have to worry about that.
+	 */
+	if (!is_dir && link_zero && xfs_inode_is_filestream(ip))
+		xfs_filestream_deassociate(ip);
+
+	return 0;
+
+ out_bmap_cancel:
+	xfs_bmap_cancel(&free_list);
+	cancel_flags |= XFS_TRANS_ABORT;
+ out_trans_cancel:
+	xfs_trans_cancel(tp, cancel_flags);
+ std_return:
+	return error;
+}
+
 STATIC int
 xfs_iflush_cluster(
 	xfs_inode_t	*ip,
@@ -1746,39 +2712,3 @@ xfs_iflush_int(
 corrupt_out:
 	return XFS_ERROR(EFSCORRUPTED);
 }
-
-/*
- * Test whether it is appropriate to check an inode for and free post EOF
- * blocks. The 'force' parameter determines whether we should also consider
- * regular files that are marked preallocated or append-only.
- */
-bool
-xfs_can_free_eofblocks(struct xfs_inode *ip, bool force)
-{
-	/* prealloc/delalloc exists only on regular files */
-	if (!S_ISREG(ip->i_d.di_mode))
-		return false;
-
-	/*
-	 * Zero sized files with no cached pages and delalloc blocks will not
-	 * have speculative prealloc/delalloc blocks to remove.
-	 */
-	if (VFS_I(ip)->i_size == 0 &&
-	    VN_CACHED(VFS_I(ip)) == 0 &&
-	    ip->i_delayed_blks == 0)
-		return false;
-
-	/* If we haven't read in the extent list, then don't do it now. */
-	if (!(ip->i_df.if_flags & XFS_IFEXTENTS))
-		return false;
-
-	/*
-	 * Do not free real preallocated or append-only files unless the file
-	 * has delalloc blocks and we are forced to remove them.
-	 */
-	if (ip->i_d.di_flags & (XFS_DIFLAG_PREALLOC | XFS_DIFLAG_APPEND))
-		if (!force || ip->i_delayed_blks == 0)
-			return false;
-
-	return true;
-}