1 files changed, 202 insertions, 197 deletions

diff --git a/fs/ocfs2/inode.c b/fs/ocfs2/inode.c
index 0297fb8982b..437de7f768c 100644
--- a/fs/ocfs2/inode.c
+++ b/fs/ocfs2/inode.c
@@ -25,14 +25,12 @@
 
 #include <linux/fs.h>
 #include <linux/types.h>
-#include <linux/slab.h>
 #include <linux/highmem.h>
 #include <linux/pagemap.h>
 #include <linux/quotaops.h>
 
 #include <asm/byteorder.h>
 
-#define MLOG_MASK_PREFIX ML_INODE
 #include <cluster/masklog.h>
 
 #include "ocfs2.h"
@@ -54,6 +52,7 @@
 #include "uptodate.h"
 #include "xattr.h"
 #include "refcounttree.h"
+#include "ocfs2_trace.h"
 
 #include "buffer_head_io.h"
 
@@ -131,8 +130,10 @@ struct inode *ocfs2_iget(struct ocfs2_super *osb, u64 blkno, unsigned flags,
 	struct inode *inode = NULL;
 	struct super_block *sb = osb->sb;
 	struct ocfs2_find_inode_args args;
+	journal_t *journal = OCFS2_SB(sb)->journal->j_journal;
 
-	mlog_entry("(blkno = %llu)\n", (unsigned long long)blkno);
+	trace_ocfs2_iget_begin((unsigned long long)blkno, flags,
+			       sysfile_type);
 
 	/* Ok. By now we've either got the offsets passed to us by the
 	 * caller, or we just pulled them off the bh. Lets do some
@@ -153,27 +154,52 @@ struct inode *ocfs2_iget(struct ocfs2_super *osb, u64 blkno, unsigned flags,
 	/* inode was *not* in the inode cache. 2.6.x requires
 	 * us to do our own read_inode call and unlock it
 	 * afterwards. */
-	if (inode && inode->i_state & I_NEW) {
-		mlog(0, "Inode was not in inode cache, reading it.\n");
-		ocfs2_read_locked_inode(inode, &args);
-		unlock_new_inode(inode);
-	}
 	if (inode == NULL) {
 		inode = ERR_PTR(-ENOMEM);
 		mlog_errno(PTR_ERR(inode));
 		goto bail;
 	}
+	trace_ocfs2_iget5_locked(inode->i_state);
+	if (inode->i_state & I_NEW) {
+		ocfs2_read_locked_inode(inode, &args);
+		unlock_new_inode(inode);
+	}
 	if (is_bad_inode(inode)) {
 		iput(inode);
 		inode = ERR_PTR(-ESTALE);
 		goto bail;
 	}
 
+	/*
+	 * Set transaction id's of transactions that have to be committed
+	 * to finish f[data]sync. We set them to currently running transaction
+	 * as we cannot be sure that the inode or some of its metadata isn't
+	 * part of the transaction - the inode could have been reclaimed and
+	 * now it is reread from disk.
+	 */
+	if (journal) {
+		transaction_t *transaction;
+		tid_t tid;
+		struct ocfs2_inode_info *oi = OCFS2_I(inode);
+
+		read_lock(&journal->j_state_lock);
+		if (journal->j_running_transaction)
+			transaction = journal->j_running_transaction;
+		else
+			transaction = journal->j_committing_transaction;
+		if (transaction)
+			tid = transaction->t_tid;
+		else
+			tid = journal->j_commit_sequence;
+		read_unlock(&journal->j_state_lock);
+		oi->i_sync_tid = tid;
+		oi->i_datasync_tid = tid;
+	}
+
 bail:
 	if (!IS_ERR(inode)) {
-		mlog(0, "returning inode with number %llu\n",
-		     (unsigned long long)OCFS2_I(inode)->ip_blkno);
-		mlog_exit_ptr(inode);
+		trace_ocfs2_iget_end(inode, 
+			(unsigned long long)OCFS2_I(inode)->ip_blkno);
 	}
 
 	return inode;
@@ -193,18 +219,17 @@ static int ocfs2_find_actor(struct inode *inode, void *opaque)
 	struct ocfs2_inode_info *oi = OCFS2_I(inode);
 	int ret = 0;
 
-	mlog_entry("(0x%p, %lu, 0x%p)\n", inode, inode->i_ino, opaque);
-
 	args = opaque;
 
 	mlog_bug_on_msg(!inode, "No inode in find actor!\n");
 
+	trace_ocfs2_find_actor(inode, inode->i_ino, opaque, args->fi_blkno);
+
 	if (oi->ip_blkno != args->fi_blkno)
 		goto bail;
 
 	ret = 1;
 bail:
-	mlog_exit(ret);
 	return ret;
 }
 
@@ -219,8 +244,6 @@ static int ocfs2_init_locked_inode(struct inode *inode, void *opaque)
 	static struct lock_class_key ocfs2_quota_ip_alloc_sem_key,
 				     ocfs2_file_ip_alloc_sem_key;
 
-	mlog_entry("inode = %p, opaque = %p\n", inode, opaque);
-
 	inode->i_ino = args->fi_ino;
 	OCFS2_I(inode)->ip_blkno = args->fi_blkno;
 	if (args->fi_sysfile_type != 0)
@@ -236,7 +259,6 @@ static int ocfs2_init_locked_inode(struct inode *inode, void *opaque)
 		lockdep_set_class(&OCFS2_I(inode)->ip_alloc_sem,
 				  &ocfs2_file_ip_alloc_sem_key);
 
-	mlog_exit(0);
 	return 0;
 }
 
@@ -247,9 +269,6 @@ void ocfs2_populate_inode(struct inode *inode, struct ocfs2_dinode *fe,
 	struct ocfs2_super *osb;
 	int use_plocks = 1;
 
-	mlog_entry("(0x%p, size:%llu)\n", inode,
-		   (unsigned long long)le64_to_cpu(fe->i_size));
-
 	sb = inode->i_sb;
 	osb = OCFS2_SB(sb);
 
@@ -277,15 +296,17 @@ void ocfs2_populate_inode(struct inode *inode, struct ocfs2_dinode *fe,
 	inode->i_generation = le32_to_cpu(fe->i_generation);
 	inode->i_rdev = huge_decode_dev(le64_to_cpu(fe->id1.dev1.i_rdev));
 	inode->i_mode = le16_to_cpu(fe->i_mode);
-	inode->i_uid = le32_to_cpu(fe->i_uid);
-	inode->i_gid = le32_to_cpu(fe->i_gid);
+	i_uid_write(inode, le32_to_cpu(fe->i_uid));
+	i_gid_write(inode, le32_to_cpu(fe->i_gid));
 
 	/* Fast symlinks will have i_size but no allocated clusters. */
-	if (S_ISLNK(inode->i_mode) && !fe->i_clusters)
+	if (S_ISLNK(inode->i_mode) && !fe->i_clusters) {
 		inode->i_blocks = 0;
-	else
+		inode->i_mapping->a_ops = &ocfs2_fast_symlink_aops;
+	} else {
 		inode->i_blocks = ocfs2_inode_sector_count(inode);
-	inode->i_mapping->a_ops = &ocfs2_aops;
+		inode->i_mapping->a_ops = &ocfs2_aops;
+	}
 	inode->i_atime.tv_sec = le64_to_cpu(fe->i_atime);
 	inode->i_atime.tv_nsec = le32_to_cpu(fe->i_atime_nsec);
 	inode->i_mtime.tv_sec = le64_to_cpu(fe->i_mtime);
@@ -299,22 +320,22 @@ void ocfs2_populate_inode(struct inode *inode, struct ocfs2_dinode *fe,
 		     (unsigned long long)OCFS2_I(inode)->ip_blkno,
 		     (unsigned long long)le64_to_cpu(fe->i_blkno));
 
-	inode->i_nlink = ocfs2_read_links_count(fe);
+	set_nlink(inode, ocfs2_read_links_count(fe));
 
+	trace_ocfs2_populate_inode(OCFS2_I(inode)->ip_blkno,
+				   le32_to_cpu(fe->i_flags));
 	if (fe->i_flags & cpu_to_le32(OCFS2_SYSTEM_FL)) {
 		OCFS2_I(inode)->ip_flags |= OCFS2_INODE_SYSTEM_FILE;
 		inode->i_flags |= S_NOQUOTA;
 	}
-
+  
 	if (fe->i_flags & cpu_to_le32(OCFS2_LOCAL_ALLOC_FL)) {
 		OCFS2_I(inode)->ip_flags |= OCFS2_INODE_BITMAP;
-		mlog(0, "local alloc inode: i_ino=%lu\n", inode->i_ino);
 	} else if (fe->i_flags & cpu_to_le32(OCFS2_BITMAP_FL)) {
 		OCFS2_I(inode)->ip_flags |= OCFS2_INODE_BITMAP;
 	} else if (fe->i_flags & cpu_to_le32(OCFS2_QUOTA_FL)) {
 		inode->i_flags |= S_NOQUOTA;
 	} else if (fe->i_flags & cpu_to_le32(OCFS2_SUPER_BLOCK_FL)) {
-		mlog(0, "superblock inode: i_ino=%lu\n", inode->i_ino);
 		/* we can't actually hit this as read_inode can't
 		 * handle superblocks today ;-) */
 		BUG();
@@ -336,12 +357,10 @@ void ocfs2_populate_inode(struct inode *inode, struct ocfs2_dinode *fe,
 		    else
 			    inode->i_fop = &ocfs2_dops_no_plocks;
 		    i_size_write(inode, le64_to_cpu(fe->i_size));
+		    OCFS2_I(inode)->ip_dir_lock_gen = 1;
 		    break;
 	    case S_IFLNK:
-		    if (ocfs2_inode_is_fast_symlink(inode))
-			inode->i_op = &ocfs2_fast_symlink_inode_operations;
-		    else
-			inode->i_op = &ocfs2_symlink_inode_operations;
+		    inode->i_op = &ocfs2_symlink_inode_operations;
 		    i_size_write(inode, le64_to_cpu(fe->i_size));
 		    break;
 	    default:
@@ -377,7 +396,10 @@ void ocfs2_populate_inode(struct inode *inode, struct ocfs2_dinode *fe,
 
 	OCFS2_I(inode)->ip_last_used_slot = 0;
 	OCFS2_I(inode)->ip_last_used_group = 0;
-	mlog_exit_void();
+
+	if (S_ISDIR(inode->i_mode))
+		ocfs2_resv_set_type(&OCFS2_I(inode)->ip_la_data_resv,
+				    OCFS2_RESV_FLAG_DIR);
 }
 
 static int ocfs2_read_locked_inode(struct inode *inode,
@@ -390,22 +412,10 @@ static int ocfs2_read_locked_inode(struct inode *inode,
 	int status, can_lock;
 	u32 generation = 0;
 
-	mlog_entry("(0x%p, 0x%p)\n", inode, args);
-
 	status = -EINVAL;
-	if (inode == NULL || inode->i_sb == NULL) {
-		mlog(ML_ERROR, "bad inode\n");
-		return status;
-	}
 	sb = inode->i_sb;
 	osb = OCFS2_SB(sb);
 
-	if (!args) {
-		mlog(ML_ERROR, "bad inode args\n");
-		make_bad_inode(inode);
-		return status;
-	}
-
 	/*
 	 * To improve performance of cold-cache inode stats, we take
 	 * the cluster lock here if possible.
@@ -430,7 +440,7 @@ static int ocfs2_read_locked_inode(struct inode *inode,
 	 * #1 and #2 can be simply solved by never taking the lock
 	 * here for system files (which are the only type we read
 	 * during mount). It's a heavier approach, but our main
-	 * concern is user-accesible files anyway.
+	 * concern is user-accessible files anyway.
 	 *
 	 * #3 works itself out because we'll eventually take the
 	 * cluster lock before trusting anything anyway.
@@ -439,6 +449,9 @@ static int ocfs2_read_locked_inode(struct inode *inode,
 		&& !(args->fi_flags & OCFS2_FI_FLAG_ORPHAN_RECOVERY)
 		&& !ocfs2_mount_local(osb);
 
+	trace_ocfs2_read_locked_inode(
+		(unsigned long long)OCFS2_I(inode)->ip_blkno, can_lock);
+
 	/*
 	 * To maintain backwards compatibility with older versions of
 	 * ocfs2-tools, we still store the generation value for system
@@ -475,7 +488,7 @@ static int ocfs2_read_locked_inode(struct inode *inode,
 	if (args->fi_flags & OCFS2_FI_FLAG_ORPHAN_RECOVERY) {
 		status = ocfs2_try_open_lock(inode, 0);
 		if (status) {
-			make_bad_inode(inode);	
+			make_bad_inode(inode);
 			return status;
 		}
 	}
@@ -485,7 +498,11 @@ static int ocfs2_read_locked_inode(struct inode *inode,
 						     OCFS2_BH_IGNORE_CACHE);
 	} else {
 		status = ocfs2_read_blocks_sync(osb, args->fi_blkno, 1, &bh);
-		if (!status)
+		/*
+		 * If buffer is in jbd, then its checksum may not have been
+		 * computed as yet.
+		 */
+		if (!status && !buffer_jbd(bh))
 			status = ocfs2_validate_inode_block(osb->sb, bh);
 	}
 	if (status < 0) {
@@ -526,7 +543,6 @@ bail:
 	if (args && bh)
 		brelse(bh);
 
-	mlog_exit(status);
 	return status;
 }
 
@@ -540,12 +556,9 @@ static int ocfs2_truncate_for_delete(struct ocfs2_super *osb,
 				     struct buffer_head *fe_bh)
 {
 	int status = 0;
-	struct ocfs2_truncate_context *tc = NULL;
 	struct ocfs2_dinode *fe;
 	handle_t *handle = NULL;
 
-	mlog_entry_void();
-
 	fe = (struct ocfs2_dinode *) fe_bh->b_data;
 
 	/*
@@ -559,6 +572,7 @@ static int ocfs2_truncate_for_delete(struct ocfs2_super *osb,
 		handle = ocfs2_start_trans(osb, OCFS2_INODE_UPDATE_CREDITS);
 		if (IS_ERR(handle)) {
 			status = PTR_ERR(handle);
+			handle = NULL;
 			mlog_errno(status);
 			goto out;
 		}
@@ -582,13 +596,7 @@ static int ocfs2_truncate_for_delete(struct ocfs2_super *osb,
 		ocfs2_commit_trans(osb, handle);
 		handle = NULL;
 
-		status = ocfs2_prepare_truncate(osb, inode, fe_bh, &tc);
-		if (status < 0) {
-			mlog_errno(status);
-			goto out;
-		}
-
-		status = ocfs2_commit_truncate(osb, inode, fe_bh, tc);
+		status = ocfs2_commit_truncate(osb, inode, fe_bh);
 		if (status < 0) {
 			mlog_errno(status);
 			goto out;
@@ -598,7 +606,6 @@ static int ocfs2_truncate_for_delete(struct ocfs2_super *osb,
 out:
 	if (handle)
 		ocfs2_commit_trans(osb, handle);
-	mlog_exit(status);
 	return status;
 }
 
@@ -640,11 +647,13 @@ static int ocfs2_remove_inode(struct inode *inode,
 		goto bail_unlock;
 	}
 
-	status = ocfs2_orphan_del(osb, handle, orphan_dir_inode, inode,
-				  orphan_dir_bh);
-	if (status < 0) {
-		mlog_errno(status);
-		goto bail_commit;
+	if (!(OCFS2_I(inode)->ip_flags & OCFS2_INODE_SKIP_ORPHAN_DIR)) {
+		status = ocfs2_orphan_del(osb, handle, orphan_dir_inode, inode,
+					  orphan_dir_bh);
+		if (status < 0) {
+			mlog_errno(status);
+			goto bail_commit;
+		}
 	}
 
 	/* set the inodes dtime */
@@ -657,15 +666,10 @@ static int ocfs2_remove_inode(struct inode *inode,
 
 	di->i_dtime = cpu_to_le64(CURRENT_TIME.tv_sec);
 	di->i_flags &= cpu_to_le32(~(OCFS2_VALID_FL | OCFS2_ORPHANED_FL));
-
-	status = ocfs2_journal_dirty(handle, di_bh);
-	if (status < 0) {
-		mlog_errno(status);
-		goto bail_commit;
-	}
+	ocfs2_journal_dirty(handle, di_bh);
 
 	ocfs2_remove_from_cache(INODE_CACHE(inode), di_bh);
-	vfs_dq_free_inode(inode);
+	dquot_free_inode(inode);
 
 	status = ocfs2_free_dinode(handle, inode_alloc_inode,
 				   inode_alloc_bh, di);
@@ -684,7 +688,7 @@ bail:
 	return status;
 }
 
-/* 
+/*
  * Serialize with orphan dir recovery. If the process doing
  * recovery on this orphan dir does an iget() with the dir
  * i_mutex held, we'll deadlock here. Instead we detect this
@@ -697,8 +701,6 @@ static int ocfs2_check_orphan_recovery_state(struct ocfs2_super *osb,
 
 	spin_lock(&osb->osb_lock);
 	if (ocfs2_node_map_test_bit(osb, &osb->osb_recovering_orphan_dirs, slot)) {
-		mlog(0, "Recovery is happening on orphan dir %d, will skip "
-		     "this inode\n", slot);
 		ret = -EDEADLK;
 		goto out;
 	}
@@ -707,6 +709,7 @@ static int ocfs2_check_orphan_recovery_state(struct ocfs2_super *osb,
 	osb->osb_orphan_wipes[slot]++;
 out:
 	spin_unlock(&osb->osb_lock);
+	trace_ocfs2_check_orphan_recovery_state(slot, ret);
 	return ret;
 }
 
@@ -723,38 +726,39 @@ static void ocfs2_signal_wipe_completion(struct ocfs2_super *osb,
 static int ocfs2_wipe_inode(struct inode *inode,
 			    struct buffer_head *di_bh)
 {
-	int status, orphaned_slot;
+	int status, orphaned_slot = -1;
 	struct inode *orphan_dir_inode = NULL;
 	struct buffer_head *orphan_dir_bh = NULL;
 	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
-	struct ocfs2_dinode *di;
+	struct ocfs2_dinode *di = (struct ocfs2_dinode *) di_bh->b_data;
 
-	di = (struct ocfs2_dinode *) di_bh->b_data;
-	orphaned_slot = le16_to_cpu(di->i_orphaned_slot);
+	if (!(OCFS2_I(inode)->ip_flags & OCFS2_INODE_SKIP_ORPHAN_DIR)) {
+		orphaned_slot = le16_to_cpu(di->i_orphaned_slot);
 
-	status = ocfs2_check_orphan_recovery_state(osb, orphaned_slot);
-	if (status)
-		return status;
+		status = ocfs2_check_orphan_recovery_state(osb, orphaned_slot);
+		if (status)
+			return status;
 
-	orphan_dir_inode = ocfs2_get_system_file_inode(osb,
-						       ORPHAN_DIR_SYSTEM_INODE,
-						       orphaned_slot);
-	if (!orphan_dir_inode) {
-		status = -EEXIST;
-		mlog_errno(status);
-		goto bail;
-	}
+		orphan_dir_inode = ocfs2_get_system_file_inode(osb,
+							       ORPHAN_DIR_SYSTEM_INODE,
+							       orphaned_slot);
+		if (!orphan_dir_inode) {
+			status = -EEXIST;
+			mlog_errno(status);
+			goto bail;
+		}
 
-	/* Lock the orphan dir. The lock will be held for the entire
-	 * delete_inode operation. We do this now to avoid races with
-	 * recovery completion on other nodes. */
-	mutex_lock(&orphan_dir_inode->i_mutex);
-	status = ocfs2_inode_lock(orphan_dir_inode, &orphan_dir_bh, 1);
-	if (status < 0) {
-		mutex_unlock(&orphan_dir_inode->i_mutex);
+		/* Lock the orphan dir. The lock will be held for the entire
+		 * delete_inode operation. We do this now to avoid races with
+		 * recovery completion on other nodes. */
+		mutex_lock(&orphan_dir_inode->i_mutex);
+		status = ocfs2_inode_lock(orphan_dir_inode, &orphan_dir_bh, 1);
+		if (status < 0) {
+			mutex_unlock(&orphan_dir_inode->i_mutex);
 
-		mlog_errno(status);
-		goto bail;
+			mlog_errno(status);
+			goto bail;
+		}
 	}
 
 	/* we do this while holding the orphan dir lock because we
@@ -795,6 +799,9 @@ static int ocfs2_wipe_inode(struct inode *inode,
 		mlog_errno(status);
 
 bail_unlock_dir:
+	if (OCFS2_I(inode)->ip_flags & OCFS2_INODE_SKIP_ORPHAN_DIR)
+		return status;
+
 	ocfs2_inode_unlock(orphan_dir_inode, 1);
 	mutex_unlock(&orphan_dir_inode->i_mutex);
 	brelse(orphan_dir_bh);
@@ -813,6 +820,10 @@ static int ocfs2_inode_is_valid_to_delete(struct inode *inode)
 	struct ocfs2_inode_info *oi = OCFS2_I(inode);
 	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
 
+	trace_ocfs2_inode_is_valid_to_delete(current, osb->dc_task,
+					     (unsigned long long)oi->ip_blkno,
+					     oi->ip_flags);
+
 	/* We shouldn't be getting here for the root directory
 	 * inode.. */
 	if (inode == osb->root_inode) {
@@ -820,16 +831,15 @@ static int ocfs2_inode_is_valid_to_delete(struct inode *inode)
 		goto bail;
 	}
 
-	/* If we're coming from downconvert_thread we can't go into our own
-	 * voting [hello, deadlock city!], so unforuntately we just
-	 * have to skip deleting this guy. That's OK though because
-	 * the node who's doing the actual deleting should handle it
-	 * anyway. */
-	if (current == osb->dc_task) {
-		mlog(0, "Skipping delete of %lu because we're currently "
-		     "in downconvert\n", inode->i_ino);
+	/*
+	 * If we're coming from downconvert_thread we can't go into our own
+	 * voting [hello, deadlock city!] so we cannot delete the inode. But
+	 * since we dropped last inode ref when downconverting dentry lock,
+	 * we cannot have the file open and thus the node doing unlink will
+	 * take care of deleting the inode.
+	 */
+	if (current == osb->dc_task)
 		goto bail;
-	}
 
 	spin_lock(&oi->ip_lock);
 	/* OCFS2 *never* deletes system files. This should technically
@@ -841,15 +851,6 @@ static int ocfs2_inode_is_valid_to_delete(struct inode *inode)
 		goto bail_unlock;
 	}
 
-	/* If we have allowd wipe of this inode for another node, it
-	 * will be marked here so we can safely skip it. Recovery will
-	 * cleanup any inodes we might inadvertantly skip here. */
-	if (oi->ip_flags & OCFS2_INODE_SKIP_DELETE) {
-		mlog(0, "Skipping delete of %lu because another node "
-		     "has done this for us.\n", inode->i_ino);
-		goto bail_unlock;
-	}
-
 	ret = 1;
 bail_unlock:
 	spin_unlock(&oi->ip_lock);
@@ -865,32 +866,45 @@ static int ocfs2_query_inode_wipe(struct inode *inode,
 				  struct buffer_head *di_bh,
 				  int *wipe)
 {
-	int status = 0;
+	int status = 0, reason = 0;
 	struct ocfs2_inode_info *oi = OCFS2_I(inode);
 	struct ocfs2_dinode *di;
 
 	*wipe = 0;
 
+	trace_ocfs2_query_inode_wipe_begin((unsigned long long)oi->ip_blkno,
+					   inode->i_nlink);
+
 	/* While we were waiting for the cluster lock in
 	 * ocfs2_delete_inode, another node might have asked to delete
 	 * the inode. Recheck our flags to catch this. */
 	if (!ocfs2_inode_is_valid_to_delete(inode)) {
-		mlog(0, "Skipping delete of %llu because flags changed\n",
-		     (unsigned long long)oi->ip_blkno);
+		reason = 1;
 		goto bail;
 	}
 
 	/* Now that we have an up to date inode, we can double check
 	 * the link count. */
-	if (inode->i_nlink) {
-		mlog(0, "Skipping delete of %llu because nlink = %u\n",
-		     (unsigned long long)oi->ip_blkno, inode->i_nlink);
+	if (inode->i_nlink)
 		goto bail;
-	}
 
 	/* Do some basic inode verification... */
 	di = (struct ocfs2_dinode *) di_bh->b_data;
-	if (!(di->i_flags & cpu_to_le32(OCFS2_ORPHANED_FL))) {
+	if (!(di->i_flags & cpu_to_le32(OCFS2_ORPHANED_FL)) &&
+	    !(oi->ip_flags & OCFS2_INODE_SKIP_ORPHAN_DIR)) {
+		/*
+		 * Inodes in the orphan dir must have ORPHANED_FL.  The only
+		 * inodes that come back out of the orphan dir are reflink
+		 * targets. A reflink target may be moved out of the orphan
+		 * dir between the time we scan the directory and the time we
+		 * process it. This would lead to HAS_REFCOUNT_FL being set but
+		 * ORPHANED_FL not.
+		 */
+		if (di->i_dyn_features & cpu_to_le16(OCFS2_HAS_REFCOUNT_FL)) {
+			reason = 2;
+			goto bail;
+		}
+
 		/* for lack of a better error? */
 		status = -EEXIST;
 		mlog(ML_ERROR,
@@ -915,7 +929,7 @@ static int ocfs2_query_inode_wipe(struct inode *inode,
 	 * the inode open lock in ocfs2_read_locked_inode(). When we
 	 * get to ->delete_inode(), each node tries to convert it's
 	 * lock to an exclusive. Trylocks are serialized by the inode
-	 * meta data lock. If the upconvert suceeds, we know the inode
+	 * meta data lock. If the upconvert succeeds, we know the inode
 	 * is no longer live and can be deleted.
 	 *
 	 * Though we call this with the meta data lock held, the
@@ -924,8 +938,7 @@ static int ocfs2_query_inode_wipe(struct inode *inode,
 	status = ocfs2_try_open_lock(inode, 1);
 	if (status == -EAGAIN) {
 		status = 0;
-		mlog(0, "Skipping delete of %llu because it is in use on "
-		     "other nodes\n", (unsigned long long)oi->ip_blkno);
+		reason = 3;
 		goto bail;
 	}
 	if (status < 0) {
@@ -934,11 +947,10 @@ static int ocfs2_query_inode_wipe(struct inode *inode,
 	}
 
 	*wipe = 1;
-	mlog(0, "Inode %llu is ok to wipe from orphan dir %u\n",
-	     (unsigned long long)oi->ip_blkno,
-	     le16_to_cpu(di->i_orphaned_slot));
+	trace_ocfs2_query_inode_wipe_succ(le16_to_cpu(di->i_orphaned_slot));
 
 bail:
+	trace_ocfs2_query_inode_wipe_end(status, reason);
 	return status;
 }
 
@@ -948,28 +960,28 @@ bail:
 static void ocfs2_cleanup_delete_inode(struct inode *inode,
 				       int sync_data)
 {
-	mlog(0, "Cleanup inode %llu, sync = %d\n",
-	     (unsigned long long)OCFS2_I(inode)->ip_blkno, sync_data);
+	trace_ocfs2_cleanup_delete_inode(
+		(unsigned long long)OCFS2_I(inode)->ip_blkno, sync_data);
 	if (sync_data)
-		write_inode_now(inode, 1);
-	truncate_inode_pages(&inode->i_data, 0);
+		filemap_write_and_wait(inode->i_mapping);
+	truncate_inode_pages_final(&inode->i_data);
 }
 
-void ocfs2_delete_inode(struct inode *inode)
+static void ocfs2_delete_inode(struct inode *inode)
 {
 	int wipe, status;
-	sigset_t blocked, oldset;
+	sigset_t oldset;
 	struct buffer_head *di_bh = NULL;
 
-	mlog_entry("(inode->i_ino = %lu)\n", inode->i_ino);
+	trace_ocfs2_delete_inode(inode->i_ino,
+				 (unsigned long long)OCFS2_I(inode)->ip_blkno,
+				 is_bad_inode(inode));
 
 	/* When we fail in read_inode() we mark inode as bad. The second test
 	 * catches the case when inode allocation fails before allocating
 	 * a block for inode. */
-	if (is_bad_inode(inode) || !OCFS2_I(inode)->ip_blkno) {
-		mlog(0, "Skipping delete of bad inode\n");
+	if (is_bad_inode(inode) || !OCFS2_I(inode)->ip_blkno)
 		goto bail;
-	}
 
 	if (!ocfs2_inode_is_valid_to_delete(inode)) {
 		/* It's probably not necessary to truncate_inode_pages
@@ -979,17 +991,13 @@ void ocfs2_delete_inode(struct inode *inode)
 		goto bail;
 	}
 
+	dquot_initialize(inode);
+
 	/* We want to block signals in delete_inode as the lock and
 	 * messaging paths may return us -ERESTARTSYS. Which would
 	 * cause us to exit early, resulting in inodes being orphaned
 	 * forever. */
-	sigfillset(&blocked);
-	status = sigprocmask(SIG_BLOCK, &blocked, &oldset);
-	if (status < 0) {
-		mlog_errno(status);
-		ocfs2_cleanup_delete_inode(inode, 1);
-		goto bail;
-	}
+	ocfs2_block_signals(&oldset);
 
 	/*
 	 * Synchronize us against ocfs2_get_dentry. We take this in
@@ -1063,39 +1071,39 @@ bail_unlock_nfs_sync:
 	ocfs2_nfs_sync_unlock(OCFS2_SB(inode->i_sb), 0);
 
 bail_unblock:
-	status = sigprocmask(SIG_SETMASK, &oldset, NULL);
-	if (status < 0)
-		mlog_errno(status);
+	ocfs2_unblock_signals(&oldset);
 bail:
-	clear_inode(inode);
-	mlog_exit_void();
+	return;
 }
 
-void ocfs2_clear_inode(struct inode *inode)
+static void ocfs2_clear_inode(struct inode *inode)
 {
 	int status;
 	struct ocfs2_inode_info *oi = OCFS2_I(inode);
+	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
 
-	mlog_entry_void();
-
-	if (!inode)
-		goto bail;
-
-	mlog(0, "Clearing inode: %llu, nlink = %u\n",
-	     (unsigned long long)OCFS2_I(inode)->ip_blkno, inode->i_nlink);
+	clear_inode(inode);
+	trace_ocfs2_clear_inode((unsigned long long)oi->ip_blkno,
+				inode->i_nlink);
 
 	mlog_bug_on_msg(OCFS2_SB(inode->i_sb) == NULL,
 			"Inode=%lu\n", inode->i_ino);
 
+	dquot_drop(inode);
+
 	/* To preven remote deletes we hold open lock before, now it
 	 * is time to unlock PR and EX open locks. */
 	ocfs2_open_unlock(inode);
 
 	/* Do these before all the other work so that we don't bounce
 	 * the downconvert thread while waiting to destroy the locks. */
-	ocfs2_mark_lockres_freeing(&oi->ip_rw_lockres);
-	ocfs2_mark_lockres_freeing(&oi->ip_inode_lockres);
-	ocfs2_mark_lockres_freeing(&oi->ip_open_lockres);
+	ocfs2_mark_lockres_freeing(osb, &oi->ip_rw_lockres);
+	ocfs2_mark_lockres_freeing(osb, &oi->ip_inode_lockres);
+	ocfs2_mark_lockres_freeing(osb, &oi->ip_open_lockres);
+
+	ocfs2_resv_discard(&OCFS2_SB(inode->i_sb)->osb_la_resmap,
+			   &oi->ip_la_data_resv);
+	ocfs2_resv_init_once(&oi->ip_la_data_resv);
 
 	/* We very well may get a clear_inode before all an inodes
 	 * metadata has hit disk. Of course, we can't drop any cluster
@@ -1165,29 +1173,36 @@ void ocfs2_clear_inode(struct inode *inode)
 	 */
 	jbd2_journal_release_jbd_inode(OCFS2_SB(inode->i_sb)->journal->j_journal,
 				       &oi->ip_jinode);
+}
 
-bail:
-	mlog_exit_void();
+void ocfs2_evict_inode(struct inode *inode)
+{
+	if (!inode->i_nlink ||
+	    (OCFS2_I(inode)->ip_flags & OCFS2_INODE_MAYBE_ORPHANED)) {
+		ocfs2_delete_inode(inode);
+	} else {
+		truncate_inode_pages_final(&inode->i_data);
+	}
+	ocfs2_clear_inode(inode);
 }
 
 /* Called under inode_lock, with no more references on the
  * struct inode, so it's safe here to check the flags field
  * and to manipulate i_nlink without any other locks. */
-void ocfs2_drop_inode(struct inode *inode)
+int ocfs2_drop_inode(struct inode *inode)
 {
 	struct ocfs2_inode_info *oi = OCFS2_I(inode);
+	int res;
 
-	mlog_entry_void();
-
-	mlog(0, "Drop inode %llu, nlink = %u, ip_flags = 0x%x\n",
-	     (unsigned long long)oi->ip_blkno, inode->i_nlink, oi->ip_flags);
+	trace_ocfs2_drop_inode((unsigned long long)oi->ip_blkno,
+				inode->i_nlink, oi->ip_flags);
 
 	if (oi->ip_flags & OCFS2_INODE_MAYBE_ORPHANED)
-		generic_delete_inode(inode);
+		res = 1;
 	else
-		generic_drop_inode(inode);
+		res = generic_drop_inode(inode);
 
-	mlog_exit_void();
+	return res;
 }
 
 /*
@@ -1198,11 +1213,11 @@ int ocfs2_inode_revalidate(struct dentry *dentry)
 	struct inode *inode = dentry->d_inode;
 	int status = 0;
 
-	mlog_entry("(inode = 0x%p, ino = %llu)\n", inode,
-		   inode ? (unsigned long long)OCFS2_I(inode)->ip_blkno : 0ULL);
+	trace_ocfs2_inode_revalidate(inode,
+		inode ? (unsigned long long)OCFS2_I(inode)->ip_blkno : 0ULL,
+		inode ? (unsigned long long)OCFS2_I(inode)->ip_flags : 0);
 
 	if (!inode) {
-		mlog(0, "eep, no inode!\n");
 		status = -ENOENT;
 		goto bail;
 	}
@@ -1210,7 +1225,6 @@ int ocfs2_inode_revalidate(struct dentry *dentry)
 	spin_lock(&OCFS2_I(inode)->ip_lock);
 	if (OCFS2_I(inode)->ip_flags & OCFS2_INODE_DELETED) {
 		spin_unlock(&OCFS2_I(inode)->ip_lock);
-		mlog(0, "inode deleted!\n");
 		status = -ENOENT;
 		goto bail;
 	}
@@ -1226,8 +1240,6 @@ int ocfs2_inode_revalidate(struct dentry *dentry)
 	}
 	ocfs2_inode_unlock(inode, 0);
 bail:
-	mlog_exit(status);
-
 	return status;
 }
 
@@ -1243,8 +1255,7 @@ int ocfs2_mark_inode_dirty(handle_t *handle,
 	int status;
 	struct ocfs2_dinode *fe = (struct ocfs2_dinode *) bh->b_data;
 
-	mlog_entry("(inode %llu)\n",
-		   (unsigned long long)OCFS2_I(inode)->ip_blkno);
+	trace_ocfs2_mark_inode_dirty((unsigned long long)OCFS2_I(inode)->ip_blkno);
 
 	status = ocfs2_journal_access_di(handle, INODE_CACHE(inode), bh,
 					 OCFS2_JOURNAL_ACCESS_WRITE);
@@ -1262,8 +1273,8 @@ int ocfs2_mark_inode_dirty(handle_t *handle,
 
 	fe->i_size = cpu_to_le64(i_size_read(inode));
 	ocfs2_set_links_count(fe, inode->i_nlink);
-	fe->i_uid = cpu_to_le32(inode->i_uid);
-	fe->i_gid = cpu_to_le32(inode->i_gid);
+	fe->i_uid = cpu_to_le32(i_uid_read(inode));
+	fe->i_gid = cpu_to_le32(i_gid_read(inode));
 	fe->i_mode = cpu_to_le16(inode->i_mode);
 	fe->i_atime = cpu_to_le64(inode->i_atime.tv_sec);
 	fe->i_atime_nsec = cpu_to_le32(inode->i_atime.tv_nsec);
@@ -1272,14 +1283,9 @@ int ocfs2_mark_inode_dirty(handle_t *handle,
 	fe->i_mtime = cpu_to_le64(inode->i_mtime.tv_sec);
 	fe->i_mtime_nsec = cpu_to_le32(inode->i_mtime.tv_nsec);
 
-	status = ocfs2_journal_dirty(handle, bh);
-	if (status < 0)
-		mlog_errno(status);
-
-	status = 0;
+	ocfs2_journal_dirty(handle, bh);
+	ocfs2_update_inode_fsync_trans(handle, inode, 1);
 leave:
-
-	mlog_exit(status);
 	return status;
 }
 
@@ -1298,9 +1304,9 @@ void ocfs2_refresh_inode(struct inode *inode,
 	OCFS2_I(inode)->ip_dyn_features = le16_to_cpu(fe->i_dyn_features);
 	ocfs2_set_inode_flags(inode);
 	i_size_write(inode, le64_to_cpu(fe->i_size));
-	inode->i_nlink = ocfs2_read_links_count(fe);
-	inode->i_uid = le32_to_cpu(fe->i_uid);
-	inode->i_gid = le32_to_cpu(fe->i_gid);
+	set_nlink(inode, ocfs2_read_links_count(fe));
+	i_uid_write(inode, le32_to_cpu(fe->i_uid));
+	i_gid_write(inode, le32_to_cpu(fe->i_gid));
 	inode->i_mode = le16_to_cpu(fe->i_mode);
 	if (S_ISLNK(inode->i_mode) && le32_to_cpu(fe->i_clusters) == 0)
 		inode->i_blocks = 0;
@@ -1322,8 +1328,7 @@ int ocfs2_validate_inode_block(struct super_block *sb,
 	int rc;
 	struct ocfs2_dinode *di = (struct ocfs2_dinode *)bh->b_data;
 
-	mlog(0, "Validating dinode %llu\n",
-	     (unsigned long long)bh->b_blocknr);
+	trace_ocfs2_validate_inode_block((unsigned long long)bh->b_blocknr);
 
 	BUG_ON(!buffer_uptodate(bh));