Merge branch 'upstream-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mfasheh/ocfs2

* 'upstream-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mfasheh/ocfs2: (31 commits)
  ocfs2: implement i_op->permission
  configfs: make configfs_dirent_exists() static
  ocfs2: update file system paths to set atime
  ocfs2: core atime update functions
  ocfs2: Add splice support
  ocfs2: Remove ocfs2_write_should_remove_suid()
  [PATCH] Export should_remove_suid()
  configfs: mutex_lock_nested() fix
  ocfs2: Remove struct ocfs2_journal_handle in favor of handle_t
  ocfs2: remove handle argument to ocfs2_start_trans()
  ocfs2: remove ocfs2_journal_handle journal field
  ocfs2: pass ocfs2_super * into ocfs2_commit_trans()
  ocfs2: remove unused handle argument from ocfs2_meta_lock_full()
  ocfs2: make ocfs2_alloc_handle() static
  ocfs2: remove unused ocfs2_handle_add_lock()
  ocfs2: remove unused ocfs2_handle_add_inode()
  ocfs2: Don't allocate handle early in ocfs2_rename()
  ocfs2: don't use handle for locking in allocation functions
  ocfs2: don't pass handle to ocfs2_meta_lock in ocfs2_rename()
  ocfs2: don't pass handle to ocfs2_meta_lock in ocfs2_symlink()
  ...

fs/configfs/dir.c

@@ -93,8 +93,8 @@ static struct configfs_dirent *configfs_new_dirent(struct configfs_dirent * pare
  *
  * called with parent inode's i_mutex held
  */
-int configfs_dirent_exists(struct configfs_dirent *parent_sd,
-			   const unsigned char *new)
+static int configfs_dirent_exists(struct configfs_dirent *parent_sd,
+				  const unsigned char *new)
 {
 	struct configfs_dirent * sd;
 
@@ -1176,8 +1176,9 @@ void configfs_unregister_subsystem(struct configfs_subsystem *subsys)
 		return;
 	}
 
-	mutex_lock(&configfs_sb->s_root->d_inode->i_mutex);
-	mutex_lock(&dentry->d_inode->i_mutex);
+	mutex_lock_nested(&configfs_sb->s_root->d_inode->i_mutex,
+			  I_MUTEX_PARENT);
+	mutex_lock_nested(&dentry->d_inode->i_mutex, I_MUTEX_CHILD);
 	if (configfs_detach_prep(dentry)) {
 		printk(KERN_ERR "configfs: Tried to unregister non-empty subsystem!\n");
 	}

fs/ocfs2/alloc.c

@@ -52,14 +52,14 @@ static int ocfs2_extent_contig(struct inode *inode,
 			       u64 blkno);
 
 static int ocfs2_create_new_meta_bhs(struct ocfs2_super *osb,
-				     struct ocfs2_journal_handle *handle,
+				     handle_t *handle,
 				     struct inode *inode,
 				     int wanted,
 				     struct ocfs2_alloc_context *meta_ac,
 				     struct buffer_head *bhs[]);
 
 static int ocfs2_add_branch(struct ocfs2_super *osb,
-			    struct ocfs2_journal_handle *handle,
+			    handle_t *handle,
 			    struct inode *inode,
 			    struct buffer_head *fe_bh,
 			    struct buffer_head *eb_bh,
@@ -67,14 +67,14 @@ static int ocfs2_add_branch(struct ocfs2_super *osb,
 			    struct ocfs2_alloc_context *meta_ac);
 
 static int ocfs2_shift_tree_depth(struct ocfs2_super *osb,
-				  struct ocfs2_journal_handle *handle,
+				  handle_t *handle,
 				  struct inode *inode,
 				  struct buffer_head *fe_bh,
 				  struct ocfs2_alloc_context *meta_ac,
 				  struct buffer_head **ret_new_eb_bh);
 
 static int ocfs2_do_insert_extent(struct ocfs2_super *osb,
-				  struct ocfs2_journal_handle *handle,
+				  handle_t *handle,
 				  struct inode *inode,
 				  struct buffer_head *fe_bh,
 				  u64 blkno,
@@ -152,7 +152,7 @@ bail:
  * l_count for you
  */
 static int ocfs2_create_new_meta_bhs(struct ocfs2_super *osb,
-				     struct ocfs2_journal_handle *handle,
+				     handle_t *handle,
 				     struct inode *inode,
 				     int wanted,
 				     struct ocfs2_alloc_context *meta_ac,
@@ -253,7 +253,7 @@ bail:
  * contain a single record with e_clusters == 0.
  */
 static int ocfs2_add_branch(struct ocfs2_super *osb,
-			    struct ocfs2_journal_handle *handle,
+			    handle_t *handle,
 			    struct inode *inode,
 			    struct buffer_head *fe_bh,
 			    struct buffer_head *eb_bh,
@@ -418,7 +418,7 @@ bail:
  * after this call.
  */
 static int ocfs2_shift_tree_depth(struct ocfs2_super *osb,
-				  struct ocfs2_journal_handle *handle,
+				  handle_t *handle,
 				  struct inode *inode,
 				  struct buffer_head *fe_bh,
 				  struct ocfs2_alloc_context *meta_ac,
@@ -520,7 +520,7 @@ bail:
  * down.
  */
 static int ocfs2_do_insert_extent(struct ocfs2_super *osb,
-				  struct ocfs2_journal_handle *handle,
+				  handle_t *handle,
 				  struct inode *inode,
 				  struct buffer_head *fe_bh,
 				  u64 start_blk,
@@ -809,7 +809,7 @@ bail:
 
 /* the caller needs to update fe->i_clusters */
 int ocfs2_insert_extent(struct ocfs2_super *osb,
-			struct ocfs2_journal_handle *handle,
+			handle_t *handle,
 			struct inode *inode,
 			struct buffer_head *fe_bh,
 			u64 start_blk,
@@ -951,7 +951,7 @@ static int ocfs2_truncate_log_can_coalesce(struct ocfs2_truncate_log *tl,
 }
 
 static int ocfs2_truncate_log_append(struct ocfs2_super *osb,
-				     struct ocfs2_journal_handle *handle,
+				     handle_t *handle,
 				     u64 start_blk,
 				     unsigned int num_clusters)
 {
@@ -1034,7 +1034,7 @@ bail:
 }
 
 static int ocfs2_replay_truncate_records(struct ocfs2_super *osb,
-					 struct ocfs2_journal_handle *handle,
+					 handle_t *handle,
 					 struct inode *data_alloc_inode,
 					 struct buffer_head *data_alloc_bh)
 {
@@ -1113,7 +1113,7 @@ static int __ocfs2_flush_truncate_log(struct ocfs2_super *osb)
 {
 	int status;
 	unsigned int num_to_flush;
-	struct ocfs2_journal_handle *handle = NULL;
+	handle_t *handle;
 	struct inode *tl_inode = osb->osb_tl_inode;
 	struct inode *data_alloc_inode = NULL;
 	struct buffer_head *tl_bh = osb->osb_tl_bh;
@@ -1130,7 +1130,7 @@ static int __ocfs2_flush_truncate_log(struct ocfs2_super *osb)
 	if (!OCFS2_IS_VALID_DINODE(di)) {
 		OCFS2_RO_ON_INVALID_DINODE(osb->sb, di);
 		status = -EIO;
-		goto bail;
+		goto out;
 	}
 
 	num_to_flush = le16_to_cpu(tl->tl_used);
@@ -1138,14 +1138,7 @@ static int __ocfs2_flush_truncate_log(struct ocfs2_super *osb)
 	     num_to_flush, (unsigned long long)OCFS2_I(tl_inode)->ip_blkno);
 	if (!num_to_flush) {
 		status = 0;
-		goto bail;
-	}
-
-	handle = ocfs2_alloc_handle(osb);
-	if (!handle) {
-		status = -ENOMEM;
-		mlog_errno(status);
-		goto bail;
+		goto out;
 	}
 
 	data_alloc_inode = ocfs2_get_system_file_inode(osb,
@@ -1154,41 +1147,40 @@ static int __ocfs2_flush_truncate_log(struct ocfs2_super *osb)
 	if (!data_alloc_inode) {
 		status = -EINVAL;
 		mlog(ML_ERROR, "Could not get bitmap inode!\n");
-		goto bail;
+		goto out;
 	}
 
-	ocfs2_handle_add_inode(handle, data_alloc_inode);
-	status = ocfs2_meta_lock(data_alloc_inode, handle, &data_alloc_bh, 1);
+	mutex_lock(&data_alloc_inode->i_mutex);
+
+	status = ocfs2_meta_lock(data_alloc_inode, &data_alloc_bh, 1);
 	if (status < 0) {
 		mlog_errno(status);
-		goto bail;
+		goto out_mutex;
 	}
 
-	handle = ocfs2_start_trans(osb, handle, OCFS2_TRUNCATE_LOG_UPDATE);
+	handle = ocfs2_start_trans(osb, OCFS2_TRUNCATE_LOG_UPDATE);
 	if (IS_ERR(handle)) {
 		status = PTR_ERR(handle);
-		handle = NULL;
 		mlog_errno(status);
-		goto bail;
+		goto out_unlock;
 	}
 
 	status = ocfs2_replay_truncate_records(osb, handle, data_alloc_inode,
 					       data_alloc_bh);
-	if (status < 0) {
+	if (status < 0)
 		mlog_errno(status);
-		goto bail;
-	}
 
-bail:
-	if (handle)
-		ocfs2_commit_trans(handle);
+	ocfs2_commit_trans(osb, handle);
 
-	if (data_alloc_inode)
-		iput(data_alloc_inode);
+out_unlock:
+	brelse(data_alloc_bh);
+	ocfs2_meta_unlock(data_alloc_inode, 1);
 
-	if (data_alloc_bh)
-		brelse(data_alloc_bh);
+out_mutex:
+	mutex_unlock(&data_alloc_inode->i_mutex);
+	iput(data_alloc_inode);
 
+out:
 	mlog_exit(status);
 	return status;
 }
@@ -1347,7 +1339,7 @@ int ocfs2_complete_truncate_log_recovery(struct ocfs2_super *osb,
 	int i;
 	unsigned int clusters, num_recs, start_cluster;
 	u64 start_blk;
-	struct ocfs2_journal_handle *handle;
+	handle_t *handle;
 	struct inode *tl_inode = osb->osb_tl_inode;
 	struct ocfs2_truncate_log *tl;
 
@@ -1373,8 +1365,7 @@ int ocfs2_complete_truncate_log_recovery(struct ocfs2_super *osb,
 			}
 		}
 
-		handle = ocfs2_start_trans(osb, NULL,
-					   OCFS2_TRUNCATE_LOG_UPDATE);
+		handle = ocfs2_start_trans(osb, OCFS2_TRUNCATE_LOG_UPDATE);
 		if (IS_ERR(handle)) {
 			status = PTR_ERR(handle);
 			mlog_errno(status);
@@ -1387,7 +1378,7 @@ int ocfs2_complete_truncate_log_recovery(struct ocfs2_super *osb,
 
 		status = ocfs2_truncate_log_append(osb, handle,
 						   start_blk, clusters);
-		ocfs2_commit_trans(handle);
+		ocfs2_commit_trans(osb, handle);
 		if (status < 0) {
 			mlog_errno(status);
 			goto bail_up;
@@ -1543,7 +1534,7 @@ static int ocfs2_do_truncate(struct ocfs2_super *osb,
 			     struct inode *inode,
 			     struct buffer_head *fe_bh,
 			     struct buffer_head *old_last_eb_bh,
-			     struct ocfs2_journal_handle *handle,
+			     handle_t *handle,
 			     struct ocfs2_truncate_context *tc)
 {
 	int status, i, depth;
@@ -1782,7 +1773,7 @@ int ocfs2_commit_truncate(struct ocfs2_super *osb,
 	struct ocfs2_extent_block *eb;
 	struct ocfs2_extent_list *el;
 	struct buffer_head *last_eb_bh;
-	struct ocfs2_journal_handle *handle = NULL;
+	handle_t *handle = NULL;
 	struct inode *tl_inode = osb->osb_tl_inode;
 
 	mlog_entry_void();
@@ -1868,7 +1859,7 @@ start:
 
 	credits = ocfs2_calc_tree_trunc_credits(osb->sb, clusters_to_del,
 						fe, el);
-	handle = ocfs2_start_trans(osb, NULL, credits);
+	handle = ocfs2_start_trans(osb, credits);
 	if (IS_ERR(handle)) {
 		status = PTR_ERR(handle);
 		handle = NULL;
@@ -1891,7 +1882,7 @@ start:
 	mutex_unlock(&tl_inode->i_mutex);
 	tl_sem = 0;
 
-	ocfs2_commit_trans(handle);
+	ocfs2_commit_trans(osb, handle);
 	handle = NULL;
 
 	BUG_ON(le32_to_cpu(fe->i_clusters) < target_i_clusters);
@@ -1906,7 +1897,7 @@ bail:
 		mutex_unlock(&tl_inode->i_mutex);
 
 	if (handle)
-		ocfs2_commit_trans(handle);
+		ocfs2_commit_trans(osb, handle);
 
 	if (last_eb_bh)
 		brelse(last_eb_bh);
@@ -2011,10 +2002,7 @@ int ocfs2_prepare_truncate(struct ocfs2_super *osb,
 		mutex_lock(&ext_alloc_inode->i_mutex);
 		(*tc)->tc_ext_alloc_inode = ext_alloc_inode;
 
-		status = ocfs2_meta_lock(ext_alloc_inode,
-					 NULL,
-					 &ext_alloc_bh,
-					 1);
+		status = ocfs2_meta_lock(ext_alloc_inode, &ext_alloc_bh, 1);
 		if (status < 0) {
 			mlog_errno(status);
 			goto bail;

fs/ocfs2/alloc.h

@@ -28,7 +28,7 @@
 
 struct ocfs2_alloc_context;
 int ocfs2_insert_extent(struct ocfs2_super *osb,
-			struct ocfs2_journal_handle *handle,
+			handle_t *handle,
 			struct inode *inode,
 			struct buffer_head *fe_bh,
 			u64 blkno,

fs/ocfs2/aops.c

@@ -200,7 +200,7 @@ static int ocfs2_readpage(struct file *file, struct page *page)
 
 	mlog_entry("(0x%p, %lu)\n", file, (page ? page->index : 0));
 
-	ret = ocfs2_meta_lock_with_page(inode, NULL, NULL, 0, page);
+	ret = ocfs2_meta_lock_with_page(inode, NULL, 0, page);
 	if (ret != 0) {
 		if (ret == AOP_TRUNCATED_PAGE)
 			unlock = 0;
@@ -305,7 +305,7 @@ static int ocfs2_prepare_write(struct file *file, struct page *page,
 
 	mlog_entry("(0x%p, 0x%p, %u, %u)\n", file, page, from, to);
 
-	ret = ocfs2_meta_lock_with_page(inode, NULL, NULL, 0, page);
+	ret = ocfs2_meta_lock_with_page(inode, NULL, 0, page);
 	if (ret != 0) {
 		mlog_errno(ret);
 		goto out;
@@ -355,16 +355,16 @@ static int walk_page_buffers(	handle_t *handle,
 	return ret;
 }
 
-struct ocfs2_journal_handle *ocfs2_start_walk_page_trans(struct inode *inode,
+handle_t *ocfs2_start_walk_page_trans(struct inode *inode,
 							 struct page *page,
 							 unsigned from,
 							 unsigned to)
 {
 	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
-	struct ocfs2_journal_handle *handle = NULL;
+	handle_t *handle = NULL;
 	int ret = 0;
 
-	handle = ocfs2_start_trans(osb, NULL, OCFS2_INODE_UPDATE_CREDITS);
+	handle = ocfs2_start_trans(osb, OCFS2_INODE_UPDATE_CREDITS);
 	if (!handle) {
 		ret = -ENOMEM;
 		mlog_errno(ret);
@@ -372,7 +372,7 @@ struct ocfs2_journal_handle *ocfs2_start_walk_page_trans(struct inode *inode,
 	}
 
 	if (ocfs2_should_order_data(inode)) {
-		ret = walk_page_buffers(handle->k_handle,
+		ret = walk_page_buffers(handle,
 					page_buffers(page),
 					from, to, NULL,
 					ocfs2_journal_dirty_data);
@@ -382,7 +382,7 @@ struct ocfs2_journal_handle *ocfs2_start_walk_page_trans(struct inode *inode,
 out:
 	if (ret) {
 		if (handle)
-			ocfs2_commit_trans(handle);
+			ocfs2_commit_trans(osb, handle);
 		handle = ERR_PTR(ret);
 	}
 	return handle;
@@ -394,7 +394,7 @@ static int ocfs2_commit_write(struct file *file, struct page *page,
 	int ret;
 	struct buffer_head *di_bh = NULL;
 	struct inode *inode = page->mapping->host;
-	struct ocfs2_journal_handle *handle = NULL;
+	handle_t *handle = NULL;
 	struct ocfs2_dinode *di;
 
 	mlog_entry("(0x%p, 0x%p, %u, %u)\n", file, page, from, to);
@@ -412,7 +412,7 @@ static int ocfs2_commit_write(struct file *file, struct page *page,
 	 *    stale inode allocation image (i_size, i_clusters, etc).
 	 */
 
-	ret = ocfs2_meta_lock_with_page(inode, NULL, &di_bh, 1, page);
+	ret = ocfs2_meta_lock_with_page(inode, &di_bh, 1, page);
 	if (ret != 0) {
 		mlog_errno(ret);
 		goto out;
@@ -464,7 +464,7 @@ static int ocfs2_commit_write(struct file *file, struct page *page,
 	}
 
 out_commit:
-	ocfs2_commit_trans(handle);
+	ocfs2_commit_trans(OCFS2_SB(inode->i_sb), handle);
 out_unlock_data:
 	ocfs2_data_unlock(inode, 1);
 out_unlock_meta:
@@ -490,7 +490,7 @@ static sector_t ocfs2_bmap(struct address_space *mapping, sector_t block)
 	 * accessed concurrently from multiple nodes.
 	 */
 	if (!INODE_JOURNAL(inode)) {
-		err = ocfs2_meta_lock(inode, NULL, NULL, 0);
+		err = ocfs2_meta_lock(inode, NULL, 0);
 		if (err) {
 			if (err != -ENOENT)
 				mlog_errno(err);

fs/ocfs2/aops.h

@@ -25,7 +25,7 @@
 int ocfs2_prepare_write_nolock(struct inode *inode, struct page *page,
 			       unsigned from, unsigned to);
 
-struct ocfs2_journal_handle *ocfs2_start_walk_page_trans(struct inode *inode,
+handle_t *ocfs2_start_walk_page_trans(struct inode *inode,
 							 struct page *page,
 							 unsigned from,
 							 unsigned to);

fs/ocfs2/dir.c

@@ -82,6 +82,7 @@ int ocfs2_readdir(struct file * filp, void * dirent, filldir_t filldir)
 	struct inode *inode = filp->f_dentry->d_inode;
 	struct super_block * sb = inode->i_sb;
 	unsigned int ra_sectors = 16;
+	int lock_level = 0;
 
 	mlog_entry("dirino=%llu\n",
 		   (unsigned long long)OCFS2_I(inode)->ip_blkno);
@@ -89,7 +90,15 @@ int ocfs2_readdir(struct file * filp, void * dirent, filldir_t filldir)
 	stored = 0;
 	bh = NULL;
 
-	error = ocfs2_meta_lock(inode, NULL, NULL, 0);
+	error = ocfs2_meta_lock_atime(inode, filp->f_vfsmnt, &lock_level);
+	if (lock_level && error >= 0) {
+		/* We release EX lock which used to update atime
+		 * and get PR lock again to reduce contention
+		 * on commonly accessed directories. */
+		ocfs2_meta_unlock(inode, 1);
+		lock_level = 0;
+		error = ocfs2_meta_lock(inode, NULL, 0);
+	}
 	if (error < 0) {
 		if (error != -ENOENT)
 			mlog_errno(error);
@@ -198,7 +207,7 @@ revalidate:
 
 	stored = 0;
 bail:
-	ocfs2_meta_unlock(inode, 0);
+	ocfs2_meta_unlock(inode, lock_level);
 
 bail_nolock:
 	mlog_exit(stored);
@@ -340,7 +349,7 @@ int ocfs2_empty_dir(struct inode *inode)
 
 /* returns a bh of the 1st new block in the allocation. */
 int ocfs2_do_extend_dir(struct super_block *sb,
-			struct ocfs2_journal_handle *handle,
+			handle_t *handle,
 			struct inode *dir,
 			struct buffer_head *parent_fe_bh,
 			struct ocfs2_alloc_context *data_ac,
@@ -398,7 +407,7 @@ static int ocfs2_extend_dir(struct ocfs2_super *osb,
 	struct ocfs2_dinode *fe = (struct ocfs2_dinode *) parent_fe_bh->b_data;
 	struct ocfs2_alloc_context *data_ac = NULL;
 	struct ocfs2_alloc_context *meta_ac = NULL;
-	struct ocfs2_journal_handle *handle = NULL;
+	handle_t *handle = NULL;
 	struct buffer_head *new_bh = NULL;
 	struct ocfs2_dir_entry * de;
 	struct super_block *sb = osb->sb;
@@ -409,13 +418,6 @@ static int ocfs2_extend_dir(struct ocfs2_super *osb,
 	mlog(0, "extending dir %llu (i_size = %lld)\n",
 	     (unsigned long long)OCFS2_I(dir)->ip_blkno, dir_i_size);
 
-	handle = ocfs2_alloc_handle(osb);
-	if (handle == NULL) {
-		status = -ENOMEM;
-		mlog_errno(status);
-		goto bail;
-	}
-
 	/* dir->i_size is always block aligned. */
 	spin_lock(&OCFS2_I(dir)->ip_lock);
 	if (dir_i_size == ocfs2_clusters_to_bytes(sb, OCFS2_I(dir)->ip_clusters)) {
@@ -428,8 +430,7 @@ static int ocfs2_extend_dir(struct ocfs2_super *osb,
 		}
 
 		if (!num_free_extents) {
-			status = ocfs2_reserve_new_metadata(osb, handle,
-							    fe, &meta_ac);
+			status = ocfs2_reserve_new_metadata(osb, fe, &meta_ac);
 			if (status < 0) {
 				if (status != -ENOSPC)
 					mlog_errno(status);
@@ -437,7 +438,7 @@ static int ocfs2_extend_dir(struct ocfs2_super *osb,
 			}
 		}
 
-		status = ocfs2_reserve_clusters(osb, handle, 1, &data_ac);
+		status = ocfs2_reserve_clusters(osb, 1, &data_ac);
 		if (status < 0) {
 			if (status != -ENOSPC)
 				mlog_errno(status);
@@ -450,7 +451,7 @@ static int ocfs2_extend_dir(struct ocfs2_super *osb,
 		credits = OCFS2_SIMPLE_DIR_EXTEND_CREDITS;
 	}
 
-	handle = ocfs2_start_trans(osb, handle, credits);
+	handle = ocfs2_start_trans(osb, credits);
 	if (IS_ERR(handle)) {
 		status = PTR_ERR(handle);
 		handle = NULL;
@@ -496,7 +497,7 @@ static int ocfs2_extend_dir(struct ocfs2_super *osb,
 	get_bh(*new_de_bh);
 bail:
 	if (handle)
-		ocfs2_commit_trans(handle);
+		ocfs2_commit_trans(osb, handle);
 
 	if (data_ac)
 		ocfs2_free_alloc_context(data_ac);

fs/ocfs2/dir.h

@@ -45,7 +45,7 @@ int ocfs2_prepare_dir_for_insert(struct ocfs2_super *osb,
 				 struct buffer_head **ret_de_bh);
 struct ocfs2_alloc_context;
 int ocfs2_do_extend_dir(struct super_block *sb,
-			struct ocfs2_journal_handle *handle,
+			handle_t *handle,
 			struct inode *dir,
 			struct buffer_head *parent_fe_bh,
 			struct ocfs2_alloc_context *data_ac,

fs/ocfs2/dlm/dlmdomain.c

@@ -68,7 +68,8 @@ static void **dlm_alloc_pagevec(int pages)
 			goto out_free;
 
 	mlog(0, "Allocated DLM hash pagevec; %d pages (%lu expected), %lu buckets per page\n",
-	     pages, DLM_HASH_PAGES, (unsigned long)DLM_BUCKETS_PER_PAGE);
+	     pages, (unsigned long)DLM_HASH_PAGES,
+	     (unsigned long)DLM_BUCKETS_PER_PAGE);
 	return vec;
 out_free:
 	dlm_free_pagevec(vec, i);

fs/ocfs2/dlmglue.c

@@ -49,6 +49,7 @@
 #include "dcache.h"
 #include "dlmglue.h"
 #include "extent_map.h"
+#include "file.h"
 #include "heartbeat.h"
 #include "inode.h"
 #include "journal.h"
@@ -1063,10 +1064,10 @@ static void ocfs2_cluster_unlock(struct ocfs2_super *osb,
 	mlog_exit_void();
 }
 
-int ocfs2_create_new_lock(struct ocfs2_super *osb,
-			  struct ocfs2_lock_res *lockres,
-			  int ex,
-			  int local)
+static int ocfs2_create_new_lock(struct ocfs2_super *osb,
+				 struct ocfs2_lock_res *lockres,
+				 int ex,
+				 int local)
 {
 	int level =  ex ? LKM_EXMODE : LKM_PRMODE;
 	unsigned long flags;
@@ -1579,7 +1580,6 @@ static int ocfs2_assign_bh(struct inode *inode,
  * the result of the lock will be communicated via the callback.
  */
 int ocfs2_meta_lock_full(struct inode *inode,
-			 struct ocfs2_journal_handle *handle,
 			 struct buffer_head **ret_bh,
 			 int ex,
 			 int arg_flags)
@@ -1668,12 +1668,6 @@ int ocfs2_meta_lock_full(struct inode *inode,
 		}
 	}
 
-	if (handle) {
-		status = ocfs2_handle_add_lock(handle, inode);
-		if (status < 0)
-			mlog_errno(status);
-	}
-
 bail:
 	if (status < 0) {
 		if (ret_bh && (*ret_bh)) {
@@ -1713,18 +1707,16 @@ bail:
  * the lock inversion simply.
  */
 int ocfs2_meta_lock_with_page(struct inode *inode,
-			      struct ocfs2_journal_handle *handle,
 			      struct buffer_head **ret_bh,
 			      int ex,
 			      struct page *page)
 {
 	int ret;
 
-	ret = ocfs2_meta_lock_full(inode, handle, ret_bh, ex,
-				   OCFS2_LOCK_NONBLOCK);
+	ret = ocfs2_meta_lock_full(inode, ret_bh, ex, OCFS2_LOCK_NONBLOCK);
 	if (ret == -EAGAIN) {
 		unlock_page(page);
-		if (ocfs2_meta_lock(inode, handle, ret_bh, ex) == 0)
+		if (ocfs2_meta_lock(inode, ret_bh, ex) == 0)
 			ocfs2_meta_unlock(inode, ex);
 		ret = AOP_TRUNCATED_PAGE;
 	}
@@ -1732,6 +1724,44 @@ int ocfs2_meta_lock_with_page(struct inode *inode,
 	return ret;
 }
 
+int ocfs2_meta_lock_atime(struct inode *inode,
+			  struct vfsmount *vfsmnt,
+			  int *level)
+{
+	int ret;
+
+	mlog_entry_void();
+	ret = ocfs2_meta_lock(inode, NULL, 0);
+	if (ret < 0) {
+		mlog_errno(ret);
+		return ret;
+	}
+
+	/*
+	 * If we should update atime, we will get EX lock,
+	 * otherwise we just get PR lock.
+	 */
+	if (ocfs2_should_update_atime(inode, vfsmnt)) {
+		struct buffer_head *bh = NULL;
+
+		ocfs2_meta_unlock(inode, 0);
+		ret = ocfs2_meta_lock(inode, &bh, 1);
+		if (ret < 0) {
+			mlog_errno(ret);
+			return ret;
+		}
+		*level = 1;
+		if (ocfs2_should_update_atime(inode, vfsmnt))
+			ocfs2_update_inode_atime(inode, bh);
+		if (bh)
+			brelse(bh);
+	} else
+		*level = 0;
+
+	mlog_exit(ret);
+	return ret;
+}
+
 void ocfs2_meta_unlock(struct inode *inode,
 		       int ex)
 {

fs/ocfs2/dlmglue.h

@@ -68,8 +68,6 @@ void ocfs2_dentry_lock_res_init(struct ocfs2_dentry_lock *dl,
 				u64 parent, struct inode *inode);
 void ocfs2_lock_res_free(struct ocfs2_lock_res *res);
 int ocfs2_create_new_inode_locks(struct inode *inode);
-int ocfs2_create_new_lock(struct ocfs2_super *osb,
-			  struct ocfs2_lock_res *lockres, int ex, int local);
 int ocfs2_drop_inode_locks(struct inode *inode);
 int ocfs2_data_lock_full(struct inode *inode,
 			 int write,
@@ -82,19 +80,20 @@ void ocfs2_data_unlock(struct inode *inode,
 		       int write);
 int ocfs2_rw_lock(struct inode *inode, int write);
 void ocfs2_rw_unlock(struct inode *inode, int write);
+int ocfs2_meta_lock_atime(struct inode *inode,
+			  struct vfsmount *vfsmnt,
+			  int *level);
 int ocfs2_meta_lock_full(struct inode *inode,
-			 struct ocfs2_journal_handle *handle,
 			 struct buffer_head **ret_bh,
 			 int ex,
 			 int arg_flags);
 int ocfs2_meta_lock_with_page(struct inode *inode,
-			      struct ocfs2_journal_handle *handle,
 			      struct buffer_head **ret_bh,
 			      int ex,
 			      struct page *page);
 /* 99% of the time we don't want to supply any additional flags --
  * those are for very specific cases only. */
-#define ocfs2_meta_lock(i, h, b, e) ocfs2_meta_lock_full(i, h, b, e, 0)
+#define ocfs2_meta_lock(i, b, e) ocfs2_meta_lock_full(i, b, e, 0)
 void ocfs2_meta_unlock(struct inode *inode,
 		       int ex);
 int ocfs2_super_lock(struct ocfs2_super *osb,

fs/ocfs2/export.c

@@ -100,7 +100,7 @@ static struct dentry *ocfs2_get_parent(struct dentry *child)
 	mlog(0, "find parent of directory %llu\n",
 	     (unsigned long long)OCFS2_I(dir)->ip_blkno);
 
-	status = ocfs2_meta_lock(dir, NULL, NULL, 0);
+	status = ocfs2_meta_lock(dir, NULL, 0);
 	if (status < 0) {
 		if (status != -ENOENT)
 			mlog_errno(status);

fs/ocfs2/file.c

@@ -31,6 +31,8 @@
 #include <linux/pagemap.h>
 #include <linux/uio.h>
 #include <linux/sched.h>
+#include <linux/pipe_fs_i.h>
+#include <linux/mount.h>
 
 #define MLOG_MASK_PREFIX ML_INODE
 #include <cluster/masklog.h>
@@ -134,7 +136,58 @@ bail:
 	return (err < 0) ? -EIO : 0;
 }
 
-int ocfs2_set_inode_size(struct ocfs2_journal_handle *handle,
+int ocfs2_should_update_atime(struct inode *inode,
+			      struct vfsmount *vfsmnt)
+{
+	struct timespec now;
+	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
+
+	if (ocfs2_is_hard_readonly(osb) || ocfs2_is_soft_readonly(osb))
+		return 0;
+
+	if ((inode->i_flags & S_NOATIME) ||
+	    ((inode->i_sb->s_flags & MS_NODIRATIME) && S_ISDIR(inode->i_mode)))
+		return 0;
+
+	if ((vfsmnt->mnt_flags & MNT_NOATIME) ||
+	    ((vfsmnt->mnt_flags & MNT_NODIRATIME) && S_ISDIR(inode->i_mode)))
+		return 0;
+
+	now = CURRENT_TIME;
+	if ((now.tv_sec - inode->i_atime.tv_sec <= osb->s_atime_quantum))
+		return 0;
+	else
+		return 1;
+}
+
+int ocfs2_update_inode_atime(struct inode *inode,
+			     struct buffer_head *bh)
+{
+	int ret;
+	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
+	handle_t *handle;
+
+	mlog_entry_void();
+
+	handle = ocfs2_start_trans(osb, OCFS2_INODE_UPDATE_CREDITS);
+	if (handle == NULL) {
+		ret = -ENOMEM;
+		mlog_errno(ret);
+		goto out;
+	}
+
+	inode->i_atime = CURRENT_TIME;
+	ret = ocfs2_mark_inode_dirty(handle, inode, bh);
+	if (ret < 0)
+		mlog_errno(ret);
+
+	ocfs2_commit_trans(OCFS2_SB(inode->i_sb), handle);
+out:
+	mlog_exit(ret);
+	return ret;
+}
+
+int ocfs2_set_inode_size(handle_t *handle,
 			 struct inode *inode,
 			 struct buffer_head *fe_bh,
 			 u64 new_i_size)
@@ -163,10 +216,9 @@ static int ocfs2_simple_size_update(struct inode *inode,
 {
 	int ret;
 	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
-	struct ocfs2_journal_handle *handle = NULL;
+	handle_t *handle = NULL;
 
-	handle = ocfs2_start_trans(osb, NULL,
-				   OCFS2_INODE_UPDATE_CREDITS);
+	handle = ocfs2_start_trans(osb, OCFS2_INODE_UPDATE_CREDITS);
 	if (handle == NULL) {
 		ret = -ENOMEM;
 		mlog_errno(ret);
@@ -178,7 +230,7 @@ static int ocfs2_simple_size_update(struct inode *inode,
 	if (ret < 0)
 		mlog_errno(ret);
 
-	ocfs2_commit_trans(handle);
+	ocfs2_commit_trans(osb, handle);
 out:
 	return ret;
 }
@@ -189,14 +241,14 @@ static int ocfs2_orphan_for_truncate(struct ocfs2_super *osb,
 				     u64 new_i_size)
 {
 	int status;
-	struct ocfs2_journal_handle *handle;
+	handle_t *handle;
 
 	mlog_entry_void();
 
 	/* TODO: This needs to actually orphan the inode in this
 	 * transaction. */
 
-	handle = ocfs2_start_trans(osb, NULL, OCFS2_INODE_UPDATE_CREDITS);
+	handle = ocfs2_start_trans(osb, OCFS2_INODE_UPDATE_CREDITS);
 	if (IS_ERR(handle)) {
 		status = PTR_ERR(handle);
 		mlog_errno(status);
@@ -207,7 +259,7 @@ static int ocfs2_orphan_for_truncate(struct ocfs2_super *osb,
 	if (status < 0)
 		mlog_errno(status);
 
-	ocfs2_commit_trans(handle);
+	ocfs2_commit_trans(osb, handle);
 out:
 	mlog_exit(status);
 	return status;
@@ -328,7 +380,7 @@ int ocfs2_do_extend_allocation(struct ocfs2_super *osb,
 			       struct inode *inode,
 			       u32 clusters_to_add,
 			       struct buffer_head *fe_bh,
-			       struct ocfs2_journal_handle *handle,
+			       handle_t *handle,
 			       struct ocfs2_alloc_context *data_ac,
 			       struct ocfs2_alloc_context *meta_ac,
 			       enum ocfs2_alloc_restarted *reason_ret)
@@ -433,7 +485,7 @@ static int ocfs2_extend_allocation(struct inode *inode,
 	u32 prev_clusters;
 	struct buffer_head *bh = NULL;
 	struct ocfs2_dinode *fe = NULL;
-	struct ocfs2_journal_handle *handle = NULL;
+	handle_t *handle = NULL;
 	struct ocfs2_alloc_context *data_ac = NULL;
 	struct ocfs2_alloc_context *meta_ac = NULL;
 	enum ocfs2_alloc_restarted why;
@@ -463,13 +515,6 @@ restart_all:
 	     (unsigned long long)OCFS2_I(inode)->ip_blkno, i_size_read(inode),
 	     fe->i_clusters, clusters_to_add);
 
-	handle = ocfs2_alloc_handle(osb);
-	if (handle == NULL) {
-		status = -ENOMEM;
-		mlog_errno(status);
-		goto leave;
-	}
-
 	num_free_extents = ocfs2_num_free_extents(osb,
 						  inode,
 						  fe);
@@ -480,10 +525,7 @@ restart_all:
 	}
 
 	if (!num_free_extents) {
-		status = ocfs2_reserve_new_metadata(osb,
-						    handle,
-						    fe,
-						    &meta_ac);
+		status = ocfs2_reserve_new_metadata(osb, fe, &meta_ac);
 		if (status < 0) {
 			if (status != -ENOSPC)
 				mlog_errno(status);
@@ -491,10 +533,7 @@ restart_all:
 		}
 	}
 
-	status = ocfs2_reserve_clusters(osb,
-					handle,
-					clusters_to_add,
-					&data_ac);
+	status = ocfs2_reserve_clusters(osb, clusters_to_add, &data_ac);
 	if (status < 0) {
 		if (status != -ENOSPC)
 			mlog_errno(status);
@@ -509,7 +548,7 @@ restart_all:
 	drop_alloc_sem = 1;
 
 	credits = ocfs2_calc_extend_credits(osb->sb, fe, clusters_to_add);
-	handle = ocfs2_start_trans(osb, handle, credits);
+	handle = ocfs2_start_trans(osb, credits);
 	if (IS_ERR(handle)) {
 		status = PTR_ERR(handle);
 		handle = NULL;
@@ -589,7 +628,7 @@ leave:
 		drop_alloc_sem = 0;
 	}
 	if (handle) {
-		ocfs2_commit_trans(handle);
+		ocfs2_commit_trans(osb, handle);
 		handle = NULL;
 	}
 	if (data_ac) {
@@ -624,7 +663,7 @@ static int ocfs2_write_zero_page(struct inode *inode,
 	struct page *page;
 	unsigned long index;
 	unsigned int offset;
-	struct ocfs2_journal_handle *handle = NULL;
+	handle_t *handle = NULL;
 	int ret;
 
 	offset = (size & (PAGE_CACHE_SIZE-1)); /* Within page */
@@ -668,7 +707,7 @@ static int ocfs2_write_zero_page(struct inode *inode,
 		ret = 0;
 
 	if (handle)
-		ocfs2_commit_trans(handle);
+		ocfs2_commit_trans(OCFS2_SB(inode->i_sb), handle);
 out_unlock:
 	unlock_page(page);
 	page_cache_release(page);
@@ -789,7 +828,7 @@ int ocfs2_setattr(struct dentry *dentry, struct iattr *attr)
 	struct super_block *sb = inode->i_sb;
 	struct ocfs2_super *osb = OCFS2_SB(sb);
 	struct buffer_head *bh = NULL;
-	struct ocfs2_journal_handle *handle = NULL;
+	handle_t *handle = NULL;
 
 	mlog_entry("(0x%p, '%.*s')\n", dentry,
 	           dentry->d_name.len, dentry->d_name.name);
@@ -825,7 +864,7 @@ int ocfs2_setattr(struct dentry *dentry, struct iattr *attr)
 		}
 	}
 
-	status = ocfs2_meta_lock(inode, NULL, &bh, 1);
+	status = ocfs2_meta_lock(inode, &bh, 1);
 	if (status < 0) {
 		if (status != -ENOENT)
 			mlog_errno(status);
@@ -845,7 +884,7 @@ int ocfs2_setattr(struct dentry *dentry, struct iattr *attr)
 		}
 	}
 
-	handle = ocfs2_start_trans(osb, NULL, OCFS2_INODE_UPDATE_CREDITS);
+	handle = ocfs2_start_trans(osb, OCFS2_INODE_UPDATE_CREDITS);
 	if (IS_ERR(handle)) {
 		status = PTR_ERR(handle);
 		mlog_errno(status);
@@ -863,7 +902,7 @@ int ocfs2_setattr(struct dentry *dentry, struct iattr *attr)
 		mlog_errno(status);
 
 bail_commit:
-	ocfs2_commit_trans(handle);
+	ocfs2_commit_trans(osb, handle);
 bail_unlock:
 	ocfs2_meta_unlock(inode, 1);
 bail_unlock_rw:
@@ -906,19 +945,41 @@ bail:
 	return err;
 }
 
+int ocfs2_permission(struct inode *inode, int mask, struct nameidata *nd)
+{
+	int ret;
+
+	mlog_entry_void();
+
+	ret = ocfs2_meta_lock(inode, NULL, 0);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	ret = generic_permission(inode, mask, NULL);
+	if (ret)
+		mlog_errno(ret);
+
+	ocfs2_meta_unlock(inode, 0);
+out:
+	mlog_exit(ret);
+	return ret;
+}
+
 static int ocfs2_write_remove_suid(struct inode *inode)
 {
 	int ret;
 	struct buffer_head *bh = NULL;
 	struct ocfs2_inode_info *oi = OCFS2_I(inode);
-	struct ocfs2_journal_handle *handle;
+	handle_t *handle;
 	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
 	struct ocfs2_dinode *di;
 
 	mlog_entry("(Inode %llu, mode 0%o)\n",
 		   (unsigned long long)oi->ip_blkno, inode->i_mode);
 
-	handle = ocfs2_start_trans(osb, NULL, OCFS2_INODE_UPDATE_CREDITS);
+	handle = ocfs2_start_trans(osb, OCFS2_INODE_UPDATE_CREDITS);
 	if (handle == NULL) {
 		ret = -ENOMEM;
 		mlog_errno(ret);
@@ -951,75 +1012,29 @@ static int ocfs2_write_remove_suid(struct inode *inode)
 out_bh:
 	brelse(bh);
 out_trans:
-	ocfs2_commit_trans(handle);
+	ocfs2_commit_trans(osb, handle);
 out:
 	mlog_exit(ret);
 	return ret;
 }
 
-static inline int ocfs2_write_should_remove_suid(struct inode *inode)
-{
-	mode_t mode = inode->i_mode;
-
-	if (!capable(CAP_FSETID)) {
-		if (unlikely(mode & S_ISUID))
-			return 1;
-
-		if (unlikely((mode & S_ISGID) && (mode & S_IXGRP)))
-			return 1;
-	}
-	return 0;
-}
-
-static ssize_t ocfs2_file_aio_write(struct kiocb *iocb,
-				    const struct iovec *iov,
-				    unsigned long nr_segs,
-				    loff_t pos)
+static int ocfs2_prepare_inode_for_write(struct dentry *dentry,
+					 loff_t *ppos,
+					 size_t count,
+					 int appending)
 {
-	int ret, rw_level = -1, meta_level = -1, have_alloc_sem = 0;
+	int ret = 0, meta_level = appending;
+	struct inode *inode = dentry->d_inode;
 	u32 clusters;
-	struct file *filp = iocb->ki_filp;
-	struct inode *inode = filp->f_dentry->d_inode;
 	loff_t newsize, saved_pos;
 
-	mlog_entry("(0x%p, %u, '%.*s')\n", filp,
-		   (unsigned int)nr_segs,
-		   filp->f_dentry->d_name.len,
-		   filp->f_dentry->d_name.name);
-
-	/* happy write of zero bytes */
-	if (iocb->ki_left == 0)
-		return 0;
-
-	if (!inode) {
-		mlog(0, "bad inode\n");
-		return -EIO;
-	}
-
-	mutex_lock(&inode->i_mutex);
-	/* to match setattr's i_mutex -> i_alloc_sem -> rw_lock ordering */
-	if (filp->f_flags & O_DIRECT) {
-		have_alloc_sem = 1;
-		down_read(&inode->i_alloc_sem);
-	}
-
-	/* concurrent O_DIRECT writes are allowed */
-	rw_level = (filp->f_flags & O_DIRECT) ? 0 : 1;
-	ret = ocfs2_rw_lock(inode, rw_level);
-	if (ret < 0) {
-		rw_level = -1;
-		mlog_errno(ret);
-		goto out;
-	}
-
 	/* 
 	 * We sample i_size under a read level meta lock to see if our write
 	 * is extending the file, if it is we back off and get a write level
 	 * meta lock.
 	 */
-	meta_level = (filp->f_flags & O_APPEND) ? 1 : 0;
 	for(;;) {
-		ret = ocfs2_meta_lock(inode, NULL, NULL, meta_level);
+		ret = ocfs2_meta_lock(inode, NULL, meta_level);
 		if (ret < 0) {
 			meta_level = -1;
 			mlog_errno(ret);
@@ -1035,7 +1050,7 @@ static ssize_t ocfs2_file_aio_write(struct kiocb *iocb,
 		 * inode. There's also the dinode i_size state which
 		 * can be lost via setattr during extending writes (we
 		 * set inode->i_size at the end of a write. */
-		if (ocfs2_write_should_remove_suid(inode)) {
+		if (should_remove_suid(dentry)) {
 			if (meta_level == 0) {
 				ocfs2_meta_unlock(inode, meta_level);
 				meta_level = 1;
@@ -1045,19 +1060,19 @@ static ssize_t ocfs2_file_aio_write(struct kiocb *iocb,
 			ret = ocfs2_write_remove_suid(inode);
 			if (ret < 0) {
 				mlog_errno(ret);
-				goto out;
+				goto out_unlock;
 			}
 		}
 
 		/* work on a copy of ppos until we're sure that we won't have
 		 * to recalculate it due to relocking. */
-		if (filp->f_flags & O_APPEND) {
+		if (appending) {
 			saved_pos = i_size_read(inode);
 			mlog(0, "O_APPEND: inode->i_size=%llu\n", saved_pos);
 		} else {
-			saved_pos = iocb->ki_pos;
+			saved_pos = *ppos;
 		}
-		newsize = iocb->ki_left + saved_pos;
+		newsize = count + saved_pos;
 
 		mlog(0, "pos=%lld newsize=%lld cursize=%lld\n",
 		     (long long) saved_pos, (long long) newsize,
@@ -1090,19 +1105,66 @@ static ssize_t ocfs2_file_aio_write(struct kiocb *iocb,
 		if (!clusters)
 			break;
 
-		ret = ocfs2_extend_file(inode, NULL, newsize, iocb->ki_left);
+		ret = ocfs2_extend_file(inode, NULL, newsize, count);
 		if (ret < 0) {
 			if (ret != -ENOSPC)
 				mlog_errno(ret);
-			goto out;
+			goto out_unlock;
 		}
 		break;
 	}
 
-	/* ok, we're done with i_size and alloc work */
-	iocb->ki_pos = saved_pos;
+	if (appending)
+		*ppos = saved_pos;
+
+out_unlock:
 	ocfs2_meta_unlock(inode, meta_level);
-	meta_level = -1;
+
+out:
+	return ret;
+}
+
+static ssize_t ocfs2_file_aio_write(struct kiocb *iocb,
+				    const struct iovec *iov,
+				    unsigned long nr_segs,
+				    loff_t pos)
+{
+	int ret, rw_level, have_alloc_sem = 0;
+	struct file *filp = iocb->ki_filp;
+	struct inode *inode = filp->f_dentry->d_inode;
+	int appending = filp->f_flags & O_APPEND ? 1 : 0;
+
+	mlog_entry("(0x%p, %u, '%.*s')\n", filp,
+		   (unsigned int)nr_segs,
+		   filp->f_dentry->d_name.len,
+		   filp->f_dentry->d_name.name);
+
+	/* happy write of zero bytes */
+	if (iocb->ki_left == 0)
+		return 0;
+
+	mutex_lock(&inode->i_mutex);
+	/* to match setattr's i_mutex -> i_alloc_sem -> rw_lock ordering */
+	if (filp->f_flags & O_DIRECT) {
+		have_alloc_sem = 1;
+		down_read(&inode->i_alloc_sem);
+	}
+
+	/* concurrent O_DIRECT writes are allowed */
+	rw_level = (filp->f_flags & O_DIRECT) ? 0 : 1;
+	ret = ocfs2_rw_lock(inode, rw_level);
+	if (ret < 0) {
+		rw_level = -1;
+		mlog_errno(ret);
+		goto out;
+	}
+
+	ret = ocfs2_prepare_inode_for_write(filp->f_dentry, &iocb->ki_pos,
+					    iocb->ki_left, appending);
+	if (ret < 0) {
+		mlog_errno(ret);
+		goto out;
+	}
 
 	/* communicate with ocfs2_dio_end_io */
 	ocfs2_iocb_set_rw_locked(iocb);
@@ -1128,8 +1190,6 @@ static ssize_t ocfs2_file_aio_write(struct kiocb *iocb,
 	}
 
 out:
-	if (meta_level != -1)
-		ocfs2_meta_unlock(inode, meta_level);
 	if (have_alloc_sem)
 		up_read(&inode->i_alloc_sem);
 	if (rw_level != -1) 
@@ -1140,12 +1200,83 @@ out:
 	return ret;
 }
 
+static ssize_t ocfs2_file_splice_write(struct pipe_inode_info *pipe,
+				       struct file *out,
+				       loff_t *ppos,
+				       size_t len,
+				       unsigned int flags)
+{
+	int ret;
+	struct inode *inode = out->f_dentry->d_inode;
+
+	mlog_entry("(0x%p, 0x%p, %u, '%.*s')\n", out, pipe,
+		   (unsigned int)len,
+		   out->f_dentry->d_name.len,
+		   out->f_dentry->d_name.name);
+
+	inode_double_lock(inode, pipe->inode);
+
+	ret = ocfs2_rw_lock(inode, 1);
+	if (ret < 0) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+	ret = ocfs2_prepare_inode_for_write(out->f_dentry, ppos, len, 0);
+	if (ret < 0) {
+		mlog_errno(ret);
+		goto out_unlock;
+	}
+
+	/* ok, we're done with i_size and alloc work */
+	ret = generic_file_splice_write_nolock(pipe, out, ppos, len, flags);
+
+out_unlock:
+	ocfs2_rw_unlock(inode, 1);
+out:
+	inode_double_unlock(inode, pipe->inode);
+
+	mlog_exit(ret);
+	return ret;
+}
+
+static ssize_t ocfs2_file_splice_read(struct file *in,
+				      loff_t *ppos,
+				      struct pipe_inode_info *pipe,
+				      size_t len,
+				      unsigned int flags)
+{
+	int ret = 0;
+	struct inode *inode = in->f_dentry->d_inode;
+
+	mlog_entry("(0x%p, 0x%p, %u, '%.*s')\n", in, pipe,
+		   (unsigned int)len,
+		   in->f_dentry->d_name.len,
+		   in->f_dentry->d_name.name);
+
+	/*
+	 * See the comment in ocfs2_file_aio_read()
+	 */
+	ret = ocfs2_meta_lock(inode, NULL, 0);
+	if (ret < 0) {
+		mlog_errno(ret);
+		goto bail;
+	}
+	ocfs2_meta_unlock(inode, 0);
+
+	ret = generic_file_splice_read(in, ppos, pipe, len, flags);
+
+bail:
+	mlog_exit(ret);
+	return ret;
+}
+
 static ssize_t ocfs2_file_aio_read(struct kiocb *iocb,
 				   const struct iovec *iov,
 				   unsigned long nr_segs,
 				   loff_t pos)
 {
-	int ret = 0, rw_level = -1, have_alloc_sem = 0;
+	int ret = 0, rw_level = -1, have_alloc_sem = 0, lock_level = 0;
 	struct file *filp = iocb->ki_filp;
 	struct inode *inode = filp->f_dentry->d_inode;
 
@@ -1187,12 +1318,12 @@ static ssize_t ocfs2_file_aio_read(struct kiocb *iocb,
 	 * like i_size. This allows the checks down below
 	 * generic_file_aio_read() a chance of actually working. 
 	 */
-	ret = ocfs2_meta_lock(inode, NULL, NULL, 0);
+	ret = ocfs2_meta_lock_atime(inode, filp->f_vfsmnt, &lock_level);
 	if (ret < 0) {
 		mlog_errno(ret);
 		goto bail;
 	}
-	ocfs2_meta_unlock(inode, 0);
+	ocfs2_meta_unlock(inode, lock_level);
 
 	ret = generic_file_aio_read(iocb, iov, nr_segs, iocb->ki_pos);
 	if (ret == -EINVAL)
@@ -1220,11 +1351,13 @@ bail:
 struct inode_operations ocfs2_file_iops = {
 	.setattr	= ocfs2_setattr,
 	.getattr	= ocfs2_getattr,
+	.permission	= ocfs2_permission,
 };
 
 struct inode_operations ocfs2_special_file_iops = {
 	.setattr	= ocfs2_setattr,
 	.getattr	= ocfs2_getattr,
+	.permission	= ocfs2_permission,
 };
 
 const struct file_operations ocfs2_fops = {
@@ -1238,6 +1371,8 @@ const struct file_operations ocfs2_fops = {
 	.aio_read	= ocfs2_file_aio_read,
 	.aio_write	= ocfs2_file_aio_write,
 	.ioctl		= ocfs2_ioctl,
+	.splice_read	= ocfs2_file_splice_read,
+	.splice_write	= ocfs2_file_splice_write,
 };
 
 const struct file_operations ocfs2_dops = {

fs/ocfs2/file.h

@@ -41,17 +41,24 @@ int ocfs2_do_extend_allocation(struct ocfs2_super *osb,
 			       struct inode *inode,
 			       u32 clusters_to_add,
 			       struct buffer_head *fe_bh,
-			       struct ocfs2_journal_handle *handle,
+			       handle_t *handle,
 			       struct ocfs2_alloc_context *data_ac,
 			       struct ocfs2_alloc_context *meta_ac,
 			       enum ocfs2_alloc_restarted *reason);
 int ocfs2_setattr(struct dentry *dentry, struct iattr *attr);
 int ocfs2_getattr(struct vfsmount *mnt, struct dentry *dentry,
 		  struct kstat *stat);
+int ocfs2_permission(struct inode *inode, int mask,
+		     struct nameidata *nd);
 
-int ocfs2_set_inode_size(struct ocfs2_journal_handle *handle,
+int ocfs2_set_inode_size(handle_t *handle,
 			 struct inode *inode,
 			 struct buffer_head *fe_bh,
 			 u64 new_i_size);
 
+int ocfs2_should_update_atime(struct inode *inode,
+			      struct vfsmount *vfsmnt);
+int ocfs2_update_inode_atime(struct inode *inode,
+			     struct buffer_head *bh);
+
 #endif /* OCFS2_FILE_H */

fs/ocfs2/inode.c

@@ -360,7 +360,6 @@ int ocfs2_populate_inode(struct inode *inode, struct ocfs2_dinode *fe,
 				  inode);
 
 	ocfs2_set_inode_flags(inode);
-	inode->i_flags |= S_NOATIME;
 
 	status = 0;
 bail:
@@ -441,7 +440,7 @@ static int ocfs2_read_locked_inode(struct inode *inode,
 				  generation, inode);
 
 	if (can_lock) {
-		status = ocfs2_meta_lock(inode, NULL, NULL, 0);
+		status = ocfs2_meta_lock(inode, NULL, 0);
 		if (status) {
 			make_bad_inode(inode);
 			mlog_errno(status);
@@ -512,7 +511,7 @@ static int ocfs2_truncate_for_delete(struct ocfs2_super *osb,
 				     struct buffer_head *fe_bh)
 {
 	int status = 0;
-	struct ocfs2_journal_handle *handle = NULL;
+	handle_t *handle = NULL;
 	struct ocfs2_truncate_context *tc = NULL;
 	struct ocfs2_dinode *fe;
 
@@ -524,7 +523,7 @@ static int ocfs2_truncate_for_delete(struct ocfs2_super *osb,
 	if (!fe->i_clusters)
 		goto bail;
 
-	handle = ocfs2_start_trans(osb, handle, OCFS2_INODE_UPDATE_CREDITS);
+	handle = ocfs2_start_trans(osb, OCFS2_INODE_UPDATE_CREDITS);
 	if (IS_ERR(handle)) {
 		status = PTR_ERR(handle);
 		handle = NULL;
@@ -538,7 +537,7 @@ static int ocfs2_truncate_for_delete(struct ocfs2_super *osb,
 		goto bail;
 	}
 
-	ocfs2_commit_trans(handle);
+	ocfs2_commit_trans(osb, handle);
 	handle = NULL;
 
 	status = ocfs2_prepare_truncate(osb, inode, fe_bh, &tc);
@@ -554,7 +553,7 @@ static int ocfs2_truncate_for_delete(struct ocfs2_super *osb,
 	}
 bail:
 	if (handle)
-		ocfs2_commit_trans(handle);
+		ocfs2_commit_trans(osb, handle);
 
 	mlog_exit(status);
 	return status;
@@ -568,7 +567,7 @@ static int ocfs2_remove_inode(struct inode *inode,
 	int status;
 	struct inode *inode_alloc_inode = NULL;
 	struct buffer_head *inode_alloc_bh = NULL;
-	struct ocfs2_journal_handle *handle;
+	handle_t *handle;
 	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
 	struct ocfs2_dinode *di = (struct ocfs2_dinode *) di_bh->b_data;
 
@@ -582,7 +581,7 @@ static int ocfs2_remove_inode(struct inode *inode,
 	}
 
 	mutex_lock(&inode_alloc_inode->i_mutex);
-	status = ocfs2_meta_lock(inode_alloc_inode, NULL, &inode_alloc_bh, 1);
+	status = ocfs2_meta_lock(inode_alloc_inode, &inode_alloc_bh, 1);
 	if (status < 0) {
 		mutex_unlock(&inode_alloc_inode->i_mutex);
 
@@ -590,7 +589,7 @@ static int ocfs2_remove_inode(struct inode *inode,
 		goto bail;
 	}
 
-	handle = ocfs2_start_trans(osb, NULL, OCFS2_DELETE_INODE_CREDITS);
+	handle = ocfs2_start_trans(osb, OCFS2_DELETE_INODE_CREDITS);
 	if (IS_ERR(handle)) {
 		status = PTR_ERR(handle);
 		mlog_errno(status);
@@ -629,7 +628,7 @@ static int ocfs2_remove_inode(struct inode *inode,
 		mlog_errno(status);
 
 bail_commit:
-	ocfs2_commit_trans(handle);
+	ocfs2_commit_trans(osb, handle);
 bail_unlock:
 	ocfs2_meta_unlock(inode_alloc_inode, 1);
 	mutex_unlock(&inode_alloc_inode->i_mutex);
@@ -705,7 +704,7 @@ static int ocfs2_wipe_inode(struct inode *inode,
 	 * 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_meta_lock(orphan_dir_inode, NULL, &orphan_dir_bh, 1);
+	status = ocfs2_meta_lock(orphan_dir_inode, &orphan_dir_bh, 1);
 	if (status < 0) {
 		mutex_unlock(&orphan_dir_inode->i_mutex);
 
@@ -933,7 +932,7 @@ void ocfs2_delete_inode(struct inode *inode)
 	 * allocation lock here as it won't be needed - nobody will
 	 * have the file open.
 	 */
-	status = ocfs2_meta_lock(inode, NULL, &di_bh, 1);
+	status = ocfs2_meta_lock(inode, &di_bh, 1);
 	if (status < 0) {
 		if (status != -ENOENT)
 			mlog_errno(status);
@@ -1067,12 +1066,6 @@ void ocfs2_clear_inode(struct inode *inode)
 	mlog_bug_on_msg(oi->ip_open_count,
 			"Clear inode of %llu has open count %d\n",
 			(unsigned long long)oi->ip_blkno, oi->ip_open_count);
-	mlog_bug_on_msg(!list_empty(&oi->ip_handle_list),
-			"Clear inode of %llu has non empty handle list\n",
-			(unsigned long long)oi->ip_blkno);
-	mlog_bug_on_msg(oi->ip_handle,
-			"Clear inode of %llu has non empty handle pointer\n",
-			(unsigned long long)oi->ip_blkno);
 
 	/* Clear all other flags. */
 	oi->ip_flags = OCFS2_INODE_CACHE_INLINE;
@@ -1186,7 +1179,7 @@ int ocfs2_inode_revalidate(struct dentry *dentry)
 
 	/* Let ocfs2_meta_lock do the work of updating our struct
 	 * inode for us. */
-	status = ocfs2_meta_lock(inode, NULL, NULL, 0);
+	status = ocfs2_meta_lock(inode, NULL, 0);
 	if (status < 0) {
 		if (status != -ENOENT)
 			mlog_errno(status);
@@ -1204,7 +1197,7 @@ bail:
  * struct inode.
  * Only takes ip_lock.
  */
-int ocfs2_mark_inode_dirty(struct ocfs2_journal_handle *handle,
+int ocfs2_mark_inode_dirty(handle_t *handle,
 			   struct inode *inode,
 			   struct buffer_head *bh)
 {

fs/ocfs2/inode.h

@@ -48,13 +48,6 @@ struct ocfs2_inode_info
 
 	struct mutex			ip_io_mutex;
 
-	/* Used by the journalling code to attach an inode to a
-	 * handle.  These are protected by ip_io_mutex in order to lock
-	 * out other I/O to the inode until we either commit or
-	 * abort. */
-	struct list_head		ip_handle_list;
-	struct ocfs2_journal_handle	*ip_handle;
-
 	u32				ip_flags; /* see below */
 	u32				ip_attr; /* inode attributes */
 
@@ -143,7 +136,7 @@ ssize_t ocfs2_rw_direct(int rw, struct file *filp, char *buf,
 void ocfs2_sync_blockdev(struct super_block *sb);
 void ocfs2_refresh_inode(struct inode *inode,
 			 struct ocfs2_dinode *fe);
-int ocfs2_mark_inode_dirty(struct ocfs2_journal_handle *handle,
+int ocfs2_mark_inode_dirty(handle_t *handle,
 			   struct inode *inode,
 			   struct buffer_head *bh);
 int ocfs2_aio_read(struct file *file, struct kiocb *req, struct iocb *iocb);

fs/ocfs2/ioctl.c

@@ -26,7 +26,7 @@ static int ocfs2_get_inode_attr(struct inode *inode, unsigned *flags)
 {
 	int status;
 
-	status = ocfs2_meta_lock(inode, NULL, NULL, 0);
+	status = ocfs2_meta_lock(inode, NULL, 0);
 	if (status < 0) {
 		mlog_errno(status);
 		return status;
@@ -43,14 +43,14 @@ static int ocfs2_set_inode_attr(struct inode *inode, unsigned flags,
 {
 	struct ocfs2_inode_info *ocfs2_inode = OCFS2_I(inode);
 	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
-	struct ocfs2_journal_handle *handle = NULL;
+	handle_t *handle = NULL;
 	struct buffer_head *bh = NULL;
 	unsigned oldflags;
 	int status;
 
 	mutex_lock(&inode->i_mutex);
 
-	status = ocfs2_meta_lock(inode, NULL, &bh, 1);
+	status = ocfs2_meta_lock(inode, &bh, 1);
 	if (status < 0) {
 		mlog_errno(status);
 		goto bail;
@@ -67,7 +67,7 @@ static int ocfs2_set_inode_attr(struct inode *inode, unsigned flags,
 	if (!S_ISDIR(inode->i_mode))
 		flags &= ~OCFS2_DIRSYNC_FL;
 
-	handle = ocfs2_start_trans(osb, NULL, OCFS2_INODE_UPDATE_CREDITS);
+	handle = ocfs2_start_trans(osb, OCFS2_INODE_UPDATE_CREDITS);
 	if (IS_ERR(handle)) {
 		status = PTR_ERR(handle);
 		mlog_errno(status);
@@ -96,7 +96,7 @@ static int ocfs2_set_inode_attr(struct inode *inode, unsigned flags,
 	if (status < 0)
 		mlog_errno(status);
 
-	ocfs2_commit_trans(handle);
+	ocfs2_commit_trans(osb, handle);
 bail_unlock:
 	ocfs2_meta_unlock(inode, 1);
 bail:

fs/ocfs2/journal.c

@@ -57,9 +57,6 @@ static int ocfs2_recover_node(struct ocfs2_super *osb,
 static int __ocfs2_recovery_thread(void *arg);
 static int ocfs2_commit_cache(struct ocfs2_super *osb);
 static int ocfs2_wait_on_mount(struct ocfs2_super *osb);
-static void ocfs2_handle_cleanup_locks(struct ocfs2_journal *journal,
-				       struct ocfs2_journal_handle *handle);
-static void ocfs2_commit_unstarted_handle(struct ocfs2_journal_handle *handle);
 static int ocfs2_journal_toggle_dirty(struct ocfs2_super *osb,
 				      int dirty);
 static int ocfs2_trylock_journal(struct ocfs2_super *osb,
@@ -113,46 +110,18 @@ finally:
 	return status;
 }
 
-struct ocfs2_journal_handle *ocfs2_alloc_handle(struct ocfs2_super *osb)
-{
-	struct ocfs2_journal_handle *retval = NULL;
-
-	retval = kcalloc(1, sizeof(*retval), GFP_NOFS);
-	if (!retval) {
-		mlog(ML_ERROR, "Failed to allocate memory for journal "
-		     "handle!\n");
-		return NULL;
-	}
-
-	retval->max_buffs = 0;
-	retval->num_locks = 0;
-	retval->k_handle = NULL;
-
-	INIT_LIST_HEAD(&retval->locks);
-	INIT_LIST_HEAD(&retval->inode_list);
-	retval->journal = osb->journal;
-
-	return retval;
-}
-
 /* pass it NULL and it will allocate a new handle object for you.  If
  * you pass it a handle however, it may still return error, in which
  * case it has free'd the passed handle for you. */
-struct ocfs2_journal_handle *ocfs2_start_trans(struct ocfs2_super *osb,
-					       struct ocfs2_journal_handle *handle,
-					       int max_buffs)
+handle_t *ocfs2_start_trans(struct ocfs2_super *osb, int max_buffs)
 {
-	int ret;
 	journal_t *journal = osb->journal->j_journal;
-
-	mlog_entry("(max_buffs = %d)\n", max_buffs);
+	handle_t *handle;
 
 	BUG_ON(!osb || !osb->journal->j_journal);
 
-	if (ocfs2_is_hard_readonly(osb)) {
-		ret = -EROFS;
-		goto done_free;
-	}
+	if (ocfs2_is_hard_readonly(osb))
+		return ERR_PTR(-EROFS);
 
 	BUG_ON(osb->journal->j_state == OCFS2_JOURNAL_FREE);
 	BUG_ON(max_buffs <= 0);
@@ -163,154 +132,39 @@ struct ocfs2_journal_handle *ocfs2_start_trans(struct ocfs2_super *osb,
 		BUG();
 	}
 
-	if (!handle)
-		handle = ocfs2_alloc_handle(osb);
-	if (!handle) {
-		ret = -ENOMEM;
-		mlog(ML_ERROR, "Failed to allocate memory for journal "
-		     "handle!\n");
-		goto done_free;
-	}
-
-	handle->max_buffs = max_buffs;
-
 	down_read(&osb->journal->j_trans_barrier);
 
-	/* actually start the transaction now */
-	handle->k_handle = journal_start(journal, max_buffs);
-	if (IS_ERR(handle->k_handle)) {
+	handle = journal_start(journal, max_buffs);
+	if (IS_ERR(handle)) {
 		up_read(&osb->journal->j_trans_barrier);
 
-		ret = PTR_ERR(handle->k_handle);
-		handle->k_handle = NULL;
-		mlog_errno(ret);
+		mlog_errno(PTR_ERR(handle));
 
 		if (is_journal_aborted(journal)) {
 			ocfs2_abort(osb->sb, "Detected aborted journal");
-			ret = -EROFS;
+			handle = ERR_PTR(-EROFS);
 		}
-		goto done_free;
-	}
-
-	atomic_inc(&(osb->journal->j_num_trans));
-	handle->flags |= OCFS2_HANDLE_STARTED;
+	} else
+		atomic_inc(&(osb->journal->j_num_trans));
 
-	mlog_exit_ptr(handle);
 	return handle;
-
-done_free:
-	if (handle)
-		ocfs2_commit_unstarted_handle(handle); /* will kfree handle */
-
-	mlog_exit(ret);
-	return ERR_PTR(ret);
-}
-
-void ocfs2_handle_add_inode(struct ocfs2_journal_handle *handle,
-			    struct inode *inode)
-{
-	BUG_ON(!handle);
-	BUG_ON(!inode);
-
-	atomic_inc(&inode->i_count);
-
-	/* we're obviously changing it... */
-	mutex_lock(&inode->i_mutex);
-
-	/* sanity check */
-	BUG_ON(OCFS2_I(inode)->ip_handle);
-	BUG_ON(!list_empty(&OCFS2_I(inode)->ip_handle_list));
-
-	OCFS2_I(inode)->ip_handle = handle;
-	list_move_tail(&(OCFS2_I(inode)->ip_handle_list), &(handle->inode_list));
-}
-
-static void ocfs2_handle_unlock_inodes(struct ocfs2_journal_handle *handle)
-{
-	struct list_head *p, *n;
-	struct inode *inode;
-	struct ocfs2_inode_info *oi;
-
-	list_for_each_safe(p, n, &handle->inode_list) {
-		oi = list_entry(p, struct ocfs2_inode_info,
-				ip_handle_list);
-		inode = &oi->vfs_inode;
-
-		OCFS2_I(inode)->ip_handle = NULL;
-		list_del_init(&OCFS2_I(inode)->ip_handle_list);
-
-		mutex_unlock(&inode->i_mutex);
-		iput(inode);
-	}
-}
-
-/* This is trivial so we do it out of the main commit
- * paths. Beware, it can be called from start_trans too! */
-static void ocfs2_commit_unstarted_handle(struct ocfs2_journal_handle *handle)
-{
-	mlog_entry_void();
-
-	BUG_ON(handle->flags & OCFS2_HANDLE_STARTED);
-
-	ocfs2_handle_unlock_inodes(handle);
-	/* You are allowed to add journal locks before the transaction
-	 * has started. */
-	ocfs2_handle_cleanup_locks(handle->journal, handle);
-
-	kfree(handle);
-
-	mlog_exit_void();
 }
 
-void ocfs2_commit_trans(struct ocfs2_journal_handle *handle)
+int ocfs2_commit_trans(struct ocfs2_super *osb,
+		       handle_t *handle)
 {
-	handle_t *jbd_handle;
-	int retval;
-	struct ocfs2_journal *journal = handle->journal;
-
-	mlog_entry_void();
+	int ret;
+	struct ocfs2_journal *journal = osb->journal;
 
 	BUG_ON(!handle);
 
-	if (!(handle->flags & OCFS2_HANDLE_STARTED)) {
-		ocfs2_commit_unstarted_handle(handle);
-		mlog_exit_void();
-		return;
-	}
-
-	/* release inode semaphores we took during this transaction */
-	ocfs2_handle_unlock_inodes(handle);
-
-	/* ocfs2_extend_trans may have had to call journal_restart
-	 * which will always commit the transaction, but may return
-	 * error for any number of reasons. If this is the case, we
-	 * clear k_handle as it's not valid any more. */
-	if (handle->k_handle) {
-		jbd_handle = handle->k_handle;
-
-		if (handle->flags & OCFS2_HANDLE_SYNC)
-			jbd_handle->h_sync = 1;
-		else
-			jbd_handle->h_sync = 0;
-
-		/* actually stop the transaction. if we've set h_sync,
-		 * it'll have been committed when we return */
-		retval = journal_stop(jbd_handle);
-		if (retval < 0) {
-			mlog_errno(retval);
-			mlog(ML_ERROR, "Could not commit transaction\n");
-			BUG();
-		}
-
-		handle->k_handle = NULL; /* it's been free'd in journal_stop */
-	}
-
-	ocfs2_handle_cleanup_locks(journal, handle);
+	ret = journal_stop(handle);
+	if (ret < 0)
+		mlog_errno(ret);
 
 	up_read(&journal->j_trans_barrier);
 
-	kfree(handle);
-	mlog_exit_void();
+	return ret;
 }
 
 /*
@@ -326,20 +180,18 @@ void ocfs2_commit_trans(struct ocfs2_journal_handle *handle)
  * good because transaction ids haven't yet been recorded on the
  * cluster locks associated with this handle.
  */
-int ocfs2_extend_trans(struct ocfs2_journal_handle *handle,
-		       int nblocks)
+int ocfs2_extend_trans(handle_t *handle, int nblocks)
 {
 	int status;
 
 	BUG_ON(!handle);
-	BUG_ON(!(handle->flags & OCFS2_HANDLE_STARTED));
 	BUG_ON(!nblocks);
 
 	mlog_entry_void();
 
 	mlog(0, "Trying to extend transaction by %d blocks\n", nblocks);
 
-	status = journal_extend(handle->k_handle, nblocks);
+	status = journal_extend(handle, nblocks);
 	if (status < 0) {
 		mlog_errno(status);
 		goto bail;
@@ -347,15 +199,12 @@ int ocfs2_extend_trans(struct ocfs2_journal_handle *handle,
 
 	if (status > 0) {
 		mlog(0, "journal_extend failed, trying journal_restart\n");
-		status = journal_restart(handle->k_handle, nblocks);
+		status = journal_restart(handle, nblocks);
 		if (status < 0) {
-			handle->k_handle = NULL;
 			mlog_errno(status);
 			goto bail;
 		}
-		handle->max_buffs = nblocks;
-	} else
-		handle->max_buffs += nblocks;
+	}
 
 	status = 0;
 bail:
@@ -364,7 +213,7 @@ bail:
 	return status;
 }
 
-int ocfs2_journal_access(struct ocfs2_journal_handle *handle,
+int ocfs2_journal_access(handle_t *handle,
 			 struct inode *inode,
 			 struct buffer_head *bh,
 			 int type)
@@ -374,7 +223,6 @@ int ocfs2_journal_access(struct ocfs2_journal_handle *handle,
 	BUG_ON(!inode);
 	BUG_ON(!handle);
 	BUG_ON(!bh);
-	BUG_ON(!(handle->flags & OCFS2_HANDLE_STARTED));
 
 	mlog_entry("bh->b_blocknr=%llu, type=%d (\"%s\"), bh->b_size = %zu\n",
 		   (unsigned long long)bh->b_blocknr, type,
@@ -403,11 +251,11 @@ int ocfs2_journal_access(struct ocfs2_journal_handle *handle,
 	switch (type) {
 	case OCFS2_JOURNAL_ACCESS_CREATE:
 	case OCFS2_JOURNAL_ACCESS_WRITE:
-		status = journal_get_write_access(handle->k_handle, bh);
+		status = journal_get_write_access(handle, bh);
 		break;
 
 	case OCFS2_JOURNAL_ACCESS_UNDO:
-		status = journal_get_undo_access(handle->k_handle, bh);
+		status = journal_get_undo_access(handle, bh);
 		break;
 
 	default:
@@ -424,17 +272,15 @@ int ocfs2_journal_access(struct ocfs2_journal_handle *handle,
 	return status;
 }
 
-int ocfs2_journal_dirty(struct ocfs2_journal_handle *handle,
+int ocfs2_journal_dirty(handle_t *handle,
 			struct buffer_head *bh)
 {
 	int status;
 
-	BUG_ON(!(handle->flags & OCFS2_HANDLE_STARTED));
-
 	mlog_entry("(bh->b_blocknr=%llu)\n",
 		   (unsigned long long)bh->b_blocknr);
 
-	status = journal_dirty_metadata(handle->k_handle, bh);
+	status = journal_dirty_metadata(handle, bh);
 	if (status < 0)
 		mlog(ML_ERROR, "Could not dirty metadata buffer. "
 		     "(bh->b_blocknr=%llu)\n",
@@ -456,59 +302,6 @@ int ocfs2_journal_dirty_data(handle_t *handle,
 	return err;
 }
 
-/* We always assume you're adding a metadata lock at level 'ex' */
-int ocfs2_handle_add_lock(struct ocfs2_journal_handle *handle,
-			  struct inode *inode)
-{
-	int status;
-	struct ocfs2_journal_lock *lock;
-
-	BUG_ON(!inode);
-
-	lock = kmem_cache_alloc(ocfs2_lock_cache, GFP_NOFS);
-	if (!lock) {
-		status = -ENOMEM;
-		mlog_errno(-ENOMEM);
-		goto bail;
-	}
-
-	if (!igrab(inode))
-		BUG();
-	lock->jl_inode = inode;
-
-	list_add_tail(&(lock->jl_lock_list), &(handle->locks));
-	handle->num_locks++;
-
-	status = 0;
-bail:
-	mlog_exit(status);
-	return status;
-}
-
-static void ocfs2_handle_cleanup_locks(struct ocfs2_journal *journal,
-				       struct ocfs2_journal_handle *handle)
-{
-	struct list_head *p, *n;
-	struct ocfs2_journal_lock *lock;
-	struct inode *inode;
-
-	list_for_each_safe(p, n, &(handle->locks)) {
-		lock = list_entry(p, struct ocfs2_journal_lock,
-				  jl_lock_list);
-		list_del(&lock->jl_lock_list);
-		handle->num_locks--;
-
-		inode = lock->jl_inode;
-		ocfs2_meta_unlock(inode, 1);
-		if (atomic_read(&inode->i_count) == 1)
-			mlog(ML_ERROR,
-			     "Inode %llu, I'm doing a last iput for!",
-			     (unsigned long long)OCFS2_I(inode)->ip_blkno);
-		iput(inode);
-		kmem_cache_free(ocfs2_lock_cache, lock);
-	}
-}
-
 #define OCFS2_DEFAULT_COMMIT_INTERVAL 	(HZ * 5)
 
 void ocfs2_set_journal_params(struct ocfs2_super *osb)
@@ -562,8 +355,7 @@ int ocfs2_journal_init(struct ocfs2_journal *journal, int *dirty)
 	/* Skip recovery waits here - journal inode metadata never
 	 * changes in a live cluster so it can be considered an
 	 * exception to the rule. */
-	status = ocfs2_meta_lock_full(inode, NULL, &bh, 1,
-				      OCFS2_META_LOCK_RECOVERY);
+	status = ocfs2_meta_lock_full(inode, &bh, 1, OCFS2_META_LOCK_RECOVERY);
 	if (status < 0) {
 		if (status != -ERESTARTSYS)
 			mlog(ML_ERROR, "Could not get lock on journal!\n");
@@ -1160,8 +952,7 @@ static int ocfs2_replay_journal(struct ocfs2_super *osb,
 	}
 	SET_INODE_JOURNAL(inode);
 
-	status = ocfs2_meta_lock_full(inode, NULL, &bh, 1,
-				      OCFS2_META_LOCK_RECOVERY);
+	status = ocfs2_meta_lock_full(inode, &bh, 1, OCFS2_META_LOCK_RECOVERY);
 	if (status < 0) {
 		mlog(0, "status returned from ocfs2_meta_lock=%d\n", status);
 		if (status != -ERESTARTSYS)
@@ -1350,7 +1141,7 @@ static int ocfs2_trylock_journal(struct ocfs2_super *osb,
 	SET_INODE_JOURNAL(inode);
 
 	flags = OCFS2_META_LOCK_RECOVERY | OCFS2_META_LOCK_NOQUEUE;
-	status = ocfs2_meta_lock_full(inode, NULL, NULL, 1, flags);
+	status = ocfs2_meta_lock_full(inode, NULL, 1, flags);
 	if (status < 0) {
 		if (status != -EAGAIN)
 			mlog_errno(status);
@@ -1433,7 +1224,7 @@ static int ocfs2_queue_orphans(struct ocfs2_super *osb,
 	}	
 
 	mutex_lock(&orphan_dir_inode->i_mutex);
-	status = ocfs2_meta_lock(orphan_dir_inode, NULL, NULL, 0);
+	status = ocfs2_meta_lock(orphan_dir_inode, NULL, 0);
 	if (status < 0) {
 		mlog_errno(status);
 		goto out;

fs/ocfs2/journal.h

@@ -37,7 +37,6 @@ enum ocfs2_journal_state {
 
 struct ocfs2_super;
 struct ocfs2_dinode;
-struct ocfs2_journal_handle;
 
 struct ocfs2_journal {
 	enum ocfs2_journal_state   j_state;    /* Journals current state   */
@@ -133,44 +132,6 @@ static inline void ocfs2_inode_set_new(struct ocfs2_super *osb,
 	spin_unlock(&trans_inc_lock);
 }
 
-extern kmem_cache_t *ocfs2_lock_cache;
-
-struct ocfs2_journal_lock {
-	struct inode     *jl_inode;
-	struct list_head  jl_lock_list;
-};
-
-struct ocfs2_journal_handle {
-	handle_t            *k_handle; /* kernel handle.                */
-	struct ocfs2_journal        *journal;
-	u32                 flags;     /* see flags below.              */
-	int                 max_buffs; /* Buffs reserved by this handle */
-
-	/* The following two fields are for ocfs2_handle_add_lock */
-	int                 num_locks;
-	struct list_head    locks;     /* A bunch of locks to
-					* release on commit. This
-					* should be a list_head */
-
-	struct list_head     inode_list;
-};
-
-#define OCFS2_HANDLE_STARTED			1
-/* should we sync-commit this handle? */
-#define OCFS2_HANDLE_SYNC			2
-static inline int ocfs2_handle_started(struct ocfs2_journal_handle *handle)
-{
-	return handle->flags & OCFS2_HANDLE_STARTED;
-}
-
-static inline void ocfs2_handle_set_sync(struct ocfs2_journal_handle *handle, int sync)
-{
-	if (sync)
-		handle->flags |= OCFS2_HANDLE_SYNC;
-	else
-		handle->flags &= ~OCFS2_HANDLE_SYNC;
-}
-
 /* Exported only for the journal struct init code in super.c. Do not call. */
 void ocfs2_complete_recovery(void *data);
 
@@ -231,15 +192,14 @@ static inline void ocfs2_checkpoint_inode(struct inode *inode)
  *  Transaction Handling:
  *  Manage the lifetime of a transaction handle.
  *
- *  ocfs2_alloc_handle     - Only allocate a handle so we can start putting
- *                          cluster locks on it. To actually change blocks,
- *                          call ocfs2_start_trans with the handle returned
- *                          from this function. You may call ocfs2_commit_trans
- *                           at any time in the lifetime of a handle.
  *  ocfs2_start_trans      - Begin a transaction. Give it an upper estimate of
  *                          the number of blocks that will be changed during
  *                          this handle.
- *  ocfs2_commit_trans     - Complete a handle.
+ *  ocfs2_commit_trans - Complete a handle. It might return -EIO if
+ *                       the journal was aborted. The majority of paths don't
+ *                       check the return value as an error there comes too
+ *                       late to do anything (and will be picked up in a
+ *                       later transaction).
  *  ocfs2_extend_trans     - Extend a handle by nblocks credits. This may
  *                          commit the handle to disk in the process, but will
  *                          not release any locks taken during the transaction.
@@ -249,24 +209,16 @@ static inline void ocfs2_checkpoint_inode(struct inode *inode)
  *  ocfs2_journal_dirty    - Mark a journalled buffer as having dirty data.
  *  ocfs2_journal_dirty_data - Indicate that a data buffer should go out before
  *                             the current handle commits.
- *  ocfs2_handle_add_lock  - Sometimes we need to delay lock release
- *                          until after a transaction has been completed. Use
- *                          ocfs2_handle_add_lock to indicate that a lock needs
- *                          to be released at the end of that handle. Locks
- *                          will be released in the order that they are added.
- *  ocfs2_handle_add_inode - Add a locked inode to a transaction.
  */
 
 /* You must always start_trans with a number of buffs > 0, but it's
  * perfectly legal to go through an entire transaction without having
  * dirtied any buffers. */
-struct ocfs2_journal_handle *ocfs2_alloc_handle(struct ocfs2_super *osb);
-struct ocfs2_journal_handle *ocfs2_start_trans(struct ocfs2_super *osb,
-					       struct ocfs2_journal_handle *handle,
+handle_t		    *ocfs2_start_trans(struct ocfs2_super *osb,
 					       int max_buffs);
-void			     ocfs2_commit_trans(struct ocfs2_journal_handle *handle);
-int			     ocfs2_extend_trans(struct ocfs2_journal_handle *handle,
-						int nblocks);
+int			     ocfs2_commit_trans(struct ocfs2_super *osb,
+						handle_t *handle);
+int			     ocfs2_extend_trans(handle_t *handle, int nblocks);
 
 /*
  * Create access is for when we get a newly created buffer and we're
@@ -283,7 +235,7 @@ int			     ocfs2_extend_trans(struct ocfs2_journal_handle *handle,
 #define OCFS2_JOURNAL_ACCESS_WRITE  1
 #define OCFS2_JOURNAL_ACCESS_UNDO   2
 
-int                  ocfs2_journal_access(struct ocfs2_journal_handle *handle,
+int                  ocfs2_journal_access(handle_t *handle,
 					  struct inode *inode,
 					  struct buffer_head *bh,
 					  int type);
@@ -306,18 +258,10 @@ int                  ocfs2_journal_access(struct ocfs2_journal_handle *handle,
  *	<modify the bh>
  * 	ocfs2_journal_dirty(handle, bh);
  */
-int                  ocfs2_journal_dirty(struct ocfs2_journal_handle *handle,
+int                  ocfs2_journal_dirty(handle_t *handle,
 					 struct buffer_head *bh);
 int                  ocfs2_journal_dirty_data(handle_t *handle,
 					      struct buffer_head *bh);
-int                  ocfs2_handle_add_lock(struct ocfs2_journal_handle *handle,
-					   struct inode *inode);
-/*
- * Use this to protect from other processes reading buffer state while
- * it's in flight.
- */
-void                 ocfs2_handle_add_inode(struct ocfs2_journal_handle *handle,
-					    struct inode *inode);
 
 /*
  *  Credit Macros:

fs/ocfs2/localalloc.c

@@ -58,19 +58,18 @@ static int ocfs2_local_alloc_find_clear_bits(struct ocfs2_super *osb,
 static void ocfs2_clear_local_alloc(struct ocfs2_dinode *alloc);
 
 static int ocfs2_sync_local_to_main(struct ocfs2_super *osb,
-				    struct ocfs2_journal_handle *handle,
+				    handle_t *handle,
 				    struct ocfs2_dinode *alloc,
 				    struct inode *main_bm_inode,
 				    struct buffer_head *main_bm_bh);
 
 static int ocfs2_local_alloc_reserve_for_window(struct ocfs2_super *osb,
-						struct ocfs2_journal_handle *handle,
 						struct ocfs2_alloc_context **ac,
 						struct inode **bitmap_inode,
 						struct buffer_head **bitmap_bh);
 
 static int ocfs2_local_alloc_new_window(struct ocfs2_super *osb,
-					struct ocfs2_journal_handle *handle,
+					handle_t *handle,
 					struct ocfs2_alloc_context *ac);
 
 static int ocfs2_local_alloc_slide_window(struct ocfs2_super *osb,
@@ -196,7 +195,7 @@ bail:
 void ocfs2_shutdown_local_alloc(struct ocfs2_super *osb)
 {
 	int status;
-	struct ocfs2_journal_handle *handle = NULL;
+	handle_t *handle;
 	struct inode *local_alloc_inode = NULL;
 	struct buffer_head *bh = NULL;
 	struct buffer_head *main_bm_bh = NULL;
@@ -207,7 +206,7 @@ void ocfs2_shutdown_local_alloc(struct ocfs2_super *osb)
 	mlog_entry_void();
 
 	if (osb->local_alloc_state == OCFS2_LA_UNUSED)
-		goto bail;
+		goto out;
 
 	local_alloc_inode =
 		ocfs2_get_system_file_inode(osb,
@@ -216,40 +215,34 @@ void ocfs2_shutdown_local_alloc(struct ocfs2_super *osb)
 	if (!local_alloc_inode) {
 		status = -ENOENT;
 		mlog_errno(status);
-		goto bail;
+		goto out;
 	}
 
 	osb->local_alloc_state = OCFS2_LA_DISABLED;
 
-	handle = ocfs2_alloc_handle(osb);
-	if (!handle) {
-		status = -ENOMEM;
-		mlog_errno(status);
-		goto bail;
-	}
-
 	main_bm_inode = ocfs2_get_system_file_inode(osb,
 						    GLOBAL_BITMAP_SYSTEM_INODE,
 						    OCFS2_INVALID_SLOT);
 	if (!main_bm_inode) {
 		status = -EINVAL;
 		mlog_errno(status);
-		goto bail;
+		goto out;
 	}
 
-	ocfs2_handle_add_inode(handle, main_bm_inode);
-	status = ocfs2_meta_lock(main_bm_inode, handle, &main_bm_bh, 1);
+	mutex_lock(&main_bm_inode->i_mutex);
+
+	status = ocfs2_meta_lock(main_bm_inode, &main_bm_bh, 1);
 	if (status < 0) {
 		mlog_errno(status);
-		goto bail;
+		goto out_mutex;
 	}
 
 	/* WINDOW_MOVE_CREDITS is a bit heavy... */
-	handle = ocfs2_start_trans(osb, handle, OCFS2_WINDOW_MOVE_CREDITS);
+	handle = ocfs2_start_trans(osb, OCFS2_WINDOW_MOVE_CREDITS);
 	if (IS_ERR(handle)) {
 		mlog_errno(PTR_ERR(handle));
 		handle = NULL;
-		goto bail;
+		goto out_unlock;
 	}
 
 	bh = osb->local_alloc_bh;
@@ -258,7 +251,7 @@ void ocfs2_shutdown_local_alloc(struct ocfs2_super *osb)
 	alloc_copy = kmalloc(bh->b_size, GFP_KERNEL);
 	if (!alloc_copy) {
 		status = -ENOMEM;
-		goto bail;
+		goto out_commit;
 	}
 	memcpy(alloc_copy, alloc, bh->b_size);
 
@@ -266,7 +259,7 @@ void ocfs2_shutdown_local_alloc(struct ocfs2_super *osb)
 				      OCFS2_JOURNAL_ACCESS_WRITE);
 	if (status < 0) {
 		mlog_errno(status);
-		goto bail;
+		goto out_commit;
 	}
 
 	ocfs2_clear_local_alloc(alloc);
@@ -274,7 +267,7 @@ void ocfs2_shutdown_local_alloc(struct ocfs2_super *osb)
 	status = ocfs2_journal_dirty(handle, bh);
 	if (status < 0) {
 		mlog_errno(status);
-		goto bail;
+		goto out_commit;
 	}
 
 	brelse(bh);
@@ -286,16 +279,20 @@ void ocfs2_shutdown_local_alloc(struct ocfs2_super *osb)
 	if (status < 0)
 		mlog_errno(status);
 
-bail:
-	if (handle)
-		ocfs2_commit_trans(handle);
+out_commit:
+	ocfs2_commit_trans(osb, handle);
 
+out_unlock:
 	if (main_bm_bh)
 		brelse(main_bm_bh);
 
-	if (main_bm_inode)
-		iput(main_bm_inode);
+	ocfs2_meta_unlock(main_bm_inode, 1);
 
+out_mutex:
+	mutex_unlock(&main_bm_inode->i_mutex);
+	iput(main_bm_inode);
+
+out:
 	if (local_alloc_inode)
 		iput(local_alloc_inode);
 
@@ -385,61 +382,59 @@ int ocfs2_complete_local_alloc_recovery(struct ocfs2_super *osb,
 					struct ocfs2_dinode *alloc)
 {
 	int status;
-	struct ocfs2_journal_handle *handle = NULL;
+	handle_t *handle;
 	struct buffer_head *main_bm_bh = NULL;
-	struct inode *main_bm_inode = NULL;
+	struct inode *main_bm_inode;
 
 	mlog_entry_void();
 
-	handle = ocfs2_alloc_handle(osb);
-	if (!handle) {
-		status = -ENOMEM;
-		mlog_errno(status);
-		goto bail;
-	}
-
 	main_bm_inode = ocfs2_get_system_file_inode(osb,
 						    GLOBAL_BITMAP_SYSTEM_INODE,
 						    OCFS2_INVALID_SLOT);
 	if (!main_bm_inode) {
 		status = -EINVAL;
 		mlog_errno(status);
-		goto bail;
+		goto out;
 	}
 
-	ocfs2_handle_add_inode(handle, main_bm_inode);
-	status = ocfs2_meta_lock(main_bm_inode, handle, &main_bm_bh, 1);
+	mutex_lock(&main_bm_inode->i_mutex);
+
+	status = ocfs2_meta_lock(main_bm_inode, &main_bm_bh, 1);
 	if (status < 0) {
 		mlog_errno(status);
-		goto bail;
+		goto out_mutex;
 	}
 
-	handle = ocfs2_start_trans(osb, handle, OCFS2_WINDOW_MOVE_CREDITS);
+	handle = ocfs2_start_trans(osb, OCFS2_WINDOW_MOVE_CREDITS);
 	if (IS_ERR(handle)) {
 		status = PTR_ERR(handle);
 		handle = NULL;
 		mlog_errno(status);
-		goto bail;
+		goto out_unlock;
 	}
 
 	/* we want the bitmap change to be recorded on disk asap */
-	ocfs2_handle_set_sync(handle, 1);
+	handle->h_sync = 1;
 
 	status = ocfs2_sync_local_to_main(osb, handle, alloc,
 					  main_bm_inode, main_bm_bh);
 	if (status < 0)
 		mlog_errno(status);
 
-bail:
-	if (handle)
-		ocfs2_commit_trans(handle);
+	ocfs2_commit_trans(osb, handle);
+
+out_unlock:
+	ocfs2_meta_unlock(main_bm_inode, 1);
+
+out_mutex:
+	mutex_unlock(&main_bm_inode->i_mutex);
 
 	if (main_bm_bh)
 		brelse(main_bm_bh);
 
-	if (main_bm_inode)
-		iput(main_bm_inode);
+	iput(main_bm_inode);
 
+out:
 	mlog_exit(status);
 	return status;
 }
@@ -452,7 +447,6 @@ bail:
  * our own in order to shift windows.
  */
 int ocfs2_reserve_local_alloc_bits(struct ocfs2_super *osb,
-				   struct ocfs2_journal_handle *passed_handle,
 				   u32 bits_wanted,
 				   struct ocfs2_alloc_context *ac)
 {
@@ -463,9 +457,7 @@ int ocfs2_reserve_local_alloc_bits(struct ocfs2_super *osb,
 
 	mlog_entry_void();
 
-	BUG_ON(!passed_handle);
 	BUG_ON(!ac);
-	BUG_ON(passed_handle->flags & OCFS2_HANDLE_STARTED);
 
 	local_alloc_inode =
 		ocfs2_get_system_file_inode(osb,
@@ -476,7 +468,11 @@ int ocfs2_reserve_local_alloc_bits(struct ocfs2_super *osb,
 		mlog_errno(status);
 		goto bail;
 	}
-	ocfs2_handle_add_inode(passed_handle, local_alloc_inode);
+
+	mutex_lock(&local_alloc_inode->i_mutex);
+
+	ac->ac_inode = local_alloc_inode;
+	ac->ac_which = OCFS2_AC_USE_LOCAL;
 
 	if (osb->local_alloc_state != OCFS2_LA_ENABLED) {
 		status = -ENOSPC;
@@ -515,21 +511,17 @@ int ocfs2_reserve_local_alloc_bits(struct ocfs2_super *osb,
 		}
 	}
 
-	ac->ac_inode = igrab(local_alloc_inode);
 	get_bh(osb->local_alloc_bh);
 	ac->ac_bh = osb->local_alloc_bh;
-	ac->ac_which = OCFS2_AC_USE_LOCAL;
 	status = 0;
 bail:
-	if (local_alloc_inode)
-		iput(local_alloc_inode);
 
 	mlog_exit(status);
 	return status;
 }
 
 int ocfs2_claim_local_alloc_bits(struct ocfs2_super *osb,
-				 struct ocfs2_journal_handle *handle,
+				 handle_t *handle,
 				 struct ocfs2_alloc_context *ac,
 				 u32 min_bits,
 				 u32 *bit_off,
@@ -707,7 +699,7 @@ static void ocfs2_verify_zero_bits(unsigned long *bitmap,
  * passed is used for caching.
  */
 static int ocfs2_sync_local_to_main(struct ocfs2_super *osb,
-				    struct ocfs2_journal_handle *handle,
+				    handle_t *handle,
 				    struct ocfs2_dinode *alloc,
 				    struct inode *main_bm_inode,
 				    struct buffer_head *main_bm_bh)
@@ -778,7 +770,6 @@ bail:
 }
 
 static int ocfs2_local_alloc_reserve_for_window(struct ocfs2_super *osb,
-						struct ocfs2_journal_handle *handle,
 						struct ocfs2_alloc_context **ac,
 						struct inode **bitmap_inode,
 						struct buffer_head **bitmap_bh)
@@ -792,7 +783,6 @@ static int ocfs2_local_alloc_reserve_for_window(struct ocfs2_super *osb,
 		goto bail;
 	}
 
-	(*ac)->ac_handle = handle;
 	(*ac)->ac_bits_wanted = ocfs2_local_alloc_window_bits(osb);
 
 	status = ocfs2_reserve_cluster_bitmap_bits(osb, *ac);
@@ -821,7 +811,7 @@ bail:
  * pass it the bitmap lock in lock_bh if you have it.
  */
 static int ocfs2_local_alloc_new_window(struct ocfs2_super *osb,
-					struct ocfs2_journal_handle *handle,
+					handle_t *handle,
 					struct ocfs2_alloc_context *ac)
 {
 	int status = 0;
@@ -888,23 +878,15 @@ static int ocfs2_local_alloc_slide_window(struct ocfs2_super *osb,
 	int status = 0;
 	struct buffer_head *main_bm_bh = NULL;
 	struct inode *main_bm_inode = NULL;
-	struct ocfs2_journal_handle *handle = NULL;
+	handle_t *handle = NULL;
 	struct ocfs2_dinode *alloc;
 	struct ocfs2_dinode *alloc_copy = NULL;
 	struct ocfs2_alloc_context *ac = NULL;
 
 	mlog_entry_void();
 
-	handle = ocfs2_alloc_handle(osb);
-	if (!handle) {
-		status = -ENOMEM;
-		mlog_errno(status);
-		goto bail;
-	}
-
 	/* This will lock the main bitmap for us. */
 	status = ocfs2_local_alloc_reserve_for_window(osb,
-						      handle,
 						      &ac,
 						      &main_bm_inode,
 						      &main_bm_bh);
@@ -914,7 +896,7 @@ static int ocfs2_local_alloc_slide_window(struct ocfs2_super *osb,
 		goto bail;
 	}
 
-	handle = ocfs2_start_trans(osb, handle, OCFS2_WINDOW_MOVE_CREDITS);
+	handle = ocfs2_start_trans(osb, OCFS2_WINDOW_MOVE_CREDITS);
 	if (IS_ERR(handle)) {
 		status = PTR_ERR(handle);
 		handle = NULL;
@@ -972,7 +954,7 @@ static int ocfs2_local_alloc_slide_window(struct ocfs2_super *osb,
 	status = 0;
 bail:
 	if (handle)
-		ocfs2_commit_trans(handle);
+		ocfs2_commit_trans(osb, handle);
 
 	if (main_bm_bh)
 		brelse(main_bm_bh);

fs/ocfs2/localalloc.h

@@ -42,12 +42,11 @@ int ocfs2_alloc_should_use_local(struct ocfs2_super *osb,
 
 struct ocfs2_alloc_context;
 int ocfs2_reserve_local_alloc_bits(struct ocfs2_super *osb,
-				   struct ocfs2_journal_handle *passed_handle,
 				   u32 bits_wanted,
 				   struct ocfs2_alloc_context *ac);
 
 int ocfs2_claim_local_alloc_bits(struct ocfs2_super *osb,
-				 struct ocfs2_journal_handle *handle,
+				 handle_t *handle,
 				 struct ocfs2_alloc_context *ac,
 				 u32 min_bits,
 				 u32 *bit_off,

fs/ocfs2/mmap.c

@@ -82,6 +82,8 @@ static struct vm_operations_struct ocfs2_file_vm_ops = {
 
 int ocfs2_mmap(struct file *file, struct vm_area_struct *vma)
 {
+	int ret = 0, lock_level = 0;
+
 	/* We don't want to support shared writable mappings yet. */
 	if (((vma->vm_flags & VM_SHARED) || (vma->vm_flags & VM_MAYSHARE))
 	    && ((vma->vm_flags & VM_WRITE) || (vma->vm_flags & VM_MAYWRITE))) {
@@ -91,7 +93,14 @@ int ocfs2_mmap(struct file *file, struct vm_area_struct *vma)
 		return -EINVAL;
 	}
 
-	file_accessed(file);
+	ret = ocfs2_meta_lock_atime(file->f_dentry->d_inode,
+				    file->f_vfsmnt, &lock_level);
+	if (ret < 0) {
+		mlog_errno(ret);
+		goto out;
+	}
+	ocfs2_meta_unlock(file->f_dentry->d_inode, lock_level);
+out:
 	vma->vm_ops = &ocfs2_file_vm_ops;
 	return 0;
 }

fs/ocfs2/namei.c

@@ -75,12 +75,12 @@ static int inline ocfs2_search_dirblock(struct buffer_head *bh,
 					unsigned long offset,
 					struct ocfs2_dir_entry **res_dir);
 
-static int ocfs2_delete_entry(struct ocfs2_journal_handle *handle,
+static int ocfs2_delete_entry(handle_t *handle,
 			      struct inode *dir,
 			      struct ocfs2_dir_entry *de_del,
 			      struct buffer_head *bh);
 
-static int __ocfs2_add_entry(struct ocfs2_journal_handle *handle,
+static int __ocfs2_add_entry(handle_t *handle,
 			     struct inode *dir,
 			     const char *name, int namelen,
 			     struct inode *inode, u64 blkno,
@@ -93,43 +93,37 @@ static int ocfs2_mknod_locked(struct ocfs2_super *osb,
 			      dev_t dev,
 			      struct buffer_head **new_fe_bh,
 			      struct buffer_head *parent_fe_bh,
-			      struct ocfs2_journal_handle *handle,
+			      handle_t *handle,
 			      struct inode **ret_inode,
 			      struct ocfs2_alloc_context *inode_ac);
 
 static int ocfs2_fill_new_dir(struct ocfs2_super *osb,
-			      struct ocfs2_journal_handle *handle,
+			      handle_t *handle,
 			      struct inode *parent,
 			      struct inode *inode,
 			      struct buffer_head *fe_bh,
 			      struct ocfs2_alloc_context *data_ac);
 
-static int ocfs2_double_lock(struct ocfs2_super *osb,
-			     struct ocfs2_journal_handle *handle,
-			     struct buffer_head **bh1,
-			     struct inode *inode1,
-			     struct buffer_head **bh2,
-			     struct inode *inode2);
-
 static int ocfs2_prepare_orphan_dir(struct ocfs2_super *osb,
-				    struct ocfs2_journal_handle *handle,
+				    struct inode **ret_orphan_dir,
 				    struct inode *inode,
 				    char *name,
 				    struct buffer_head **de_bh);
 
 static int ocfs2_orphan_add(struct ocfs2_super *osb,
-			    struct ocfs2_journal_handle *handle,
+			    handle_t *handle,
 			    struct inode *inode,
 			    struct ocfs2_dinode *fe,
 			    char *name,
-			    struct buffer_head *de_bh);
+			    struct buffer_head *de_bh,
+			    struct inode *orphan_dir_inode);
 
 static int ocfs2_create_symlink_data(struct ocfs2_super *osb,
-				     struct ocfs2_journal_handle *handle,
+				     handle_t *handle,
 				     struct inode *inode,
 				     const char *symname);
 
-static inline int ocfs2_add_entry(struct ocfs2_journal_handle *handle,
+static inline int ocfs2_add_entry(handle_t *handle,
 				  struct dentry *dentry,
 				  struct inode *inode, u64 blkno,
 				  struct buffer_head *parent_fe_bh,
@@ -165,7 +159,7 @@ static struct dentry *ocfs2_lookup(struct inode *dir, struct dentry *dentry,
 	mlog(0, "find name %.*s in directory %llu\n", dentry->d_name.len,
 	     dentry->d_name.name, (unsigned long long)OCFS2_I(dir)->ip_blkno);
 
-	status = ocfs2_meta_lock(dir, NULL, NULL, 0);
+	status = ocfs2_meta_lock(dir, NULL, 0);
 	if (status < 0) {
 		if (status != -ENOENT)
 			mlog_errno(status);
@@ -242,7 +236,7 @@ bail:
 }
 
 static int ocfs2_fill_new_dir(struct ocfs2_super *osb,
-			      struct ocfs2_journal_handle *handle,
+			      handle_t *handle,
 			      struct inode *parent,
 			      struct inode *inode,
 			      struct buffer_head *fe_bh,
@@ -317,7 +311,7 @@ static int ocfs2_mknod(struct inode *dir,
 {
 	int status = 0;
 	struct buffer_head *parent_fe_bh = NULL;
-	struct ocfs2_journal_handle *handle = NULL;
+	handle_t *handle = NULL;
 	struct ocfs2_super *osb;
 	struct ocfs2_dinode *dirfe;
 	struct buffer_head *new_fe_bh = NULL;
@@ -333,18 +327,11 @@ static int ocfs2_mknod(struct inode *dir,
 	/* get our super block */
 	osb = OCFS2_SB(dir->i_sb);
 
-	handle = ocfs2_alloc_handle(osb);
-	if (handle == NULL) {
-		status = -ENOMEM;
-		mlog_errno(status);
-		goto leave;
-	}
-
-	status = ocfs2_meta_lock(dir, handle, &parent_fe_bh, 1);
+	status = ocfs2_meta_lock(dir, &parent_fe_bh, 1);
 	if (status < 0) {
 		if (status != -ENOENT)
 			mlog_errno(status);
-		goto leave;
+		return status;
 	}
 
 	if (S_ISDIR(mode) && (dir->i_nlink >= OCFS2_LINK_MAX)) {
@@ -374,7 +361,7 @@ static int ocfs2_mknod(struct inode *dir,
 	}
 
 	/* reserve an inode spot */
-	status = ocfs2_reserve_new_inode(osb, handle, &inode_ac);
+	status = ocfs2_reserve_new_inode(osb, &inode_ac);
 	if (status < 0) {
 		if (status != -ENOSPC)
 			mlog_errno(status);
@@ -384,7 +371,7 @@ static int ocfs2_mknod(struct inode *dir,
 	/* are we making a directory? If so, reserve a cluster for his
 	 * 1st extent. */
 	if (S_ISDIR(mode)) {
-		status = ocfs2_reserve_clusters(osb, handle, 1, &data_ac);
+		status = ocfs2_reserve_clusters(osb, 1, &data_ac);
 		if (status < 0) {
 			if (status != -ENOSPC)
 				mlog_errno(status);
@@ -392,7 +379,7 @@ static int ocfs2_mknod(struct inode *dir,
 		}
 	}
 
-	handle = ocfs2_start_trans(osb, handle, OCFS2_MKNOD_CREDITS);
+	handle = ocfs2_start_trans(osb, OCFS2_MKNOD_CREDITS);
 	if (IS_ERR(handle)) {
 		status = PTR_ERR(handle);
 		handle = NULL;
@@ -453,7 +440,9 @@ static int ocfs2_mknod(struct inode *dir,
 	status = 0;
 leave:
 	if (handle)
-		ocfs2_commit_trans(handle);
+		ocfs2_commit_trans(osb, handle);
+
+	ocfs2_meta_unlock(dir, 1);
 
 	if (status == -ENOSPC)
 		mlog(0, "Disk is full\n");
@@ -487,7 +476,7 @@ static int ocfs2_mknod_locked(struct ocfs2_super *osb,
 			      dev_t dev,
 			      struct buffer_head **new_fe_bh,
 			      struct buffer_head *parent_fe_bh,
-			      struct ocfs2_journal_handle *handle,
+			      handle_t *handle,
 			      struct inode **ret_inode,
 			      struct ocfs2_alloc_context *inode_ac)
 {
@@ -653,7 +642,7 @@ static int ocfs2_link(struct dentry *old_dentry,
 		      struct inode *dir,
 		      struct dentry *dentry)
 {
-	struct ocfs2_journal_handle *handle = NULL;
+	handle_t *handle;
 	struct inode *inode = old_dentry->d_inode;
 	int err;
 	struct buffer_head *fe_bh = NULL;
@@ -666,68 +655,60 @@ static int ocfs2_link(struct dentry *old_dentry,
 		   old_dentry->d_name.len, old_dentry->d_name.name,
 		   dentry->d_name.len, dentry->d_name.name);
 
-	if (S_ISDIR(inode->i_mode)) {
-		err = -EPERM;
-		goto bail;
-	}
-
-	handle = ocfs2_alloc_handle(osb);
-	if (handle == NULL) {
-		err = -ENOMEM;
-		goto bail;
-	}
+	if (S_ISDIR(inode->i_mode))
+		return -EPERM;
 
-	err = ocfs2_meta_lock(dir, handle, &parent_fe_bh, 1);
+	err = ocfs2_meta_lock(dir, &parent_fe_bh, 1);
 	if (err < 0) {
 		if (err != -ENOENT)
 			mlog_errno(err);
-		goto bail;
+		return err;
 	}
 
 	if (!dir->i_nlink) {
 		err = -ENOENT;
-		goto bail;
+		goto out;
 	}
 
 	err = ocfs2_check_dir_for_entry(dir, dentry->d_name.name,
 					dentry->d_name.len);
 	if (err)
-		goto bail;
+		goto out;
 
 	err = ocfs2_prepare_dir_for_insert(osb, dir, parent_fe_bh,
 					   dentry->d_name.name,
 					   dentry->d_name.len, &de_bh);
 	if (err < 0) {
 		mlog_errno(err);
-		goto bail;
+		goto out;
 	}
 
-	err = ocfs2_meta_lock(inode, handle, &fe_bh, 1);
+	err = ocfs2_meta_lock(inode, &fe_bh, 1);
 	if (err < 0) {
 		if (err != -ENOENT)
 			mlog_errno(err);
-		goto bail;
+		goto out;
 	}
 
 	fe = (struct ocfs2_dinode *) fe_bh->b_data;
 	if (le16_to_cpu(fe->i_links_count) >= OCFS2_LINK_MAX) {
 		err = -EMLINK;
-		goto bail;
+		goto out_unlock_inode;
 	}
 
-	handle = ocfs2_start_trans(osb, handle, OCFS2_LINK_CREDITS);
+	handle = ocfs2_start_trans(osb, OCFS2_LINK_CREDITS);
 	if (IS_ERR(handle)) {
 		err = PTR_ERR(handle);
 		handle = NULL;
 		mlog_errno(err);
-		goto bail;
+		goto out_unlock_inode;
 	}
 
 	err = ocfs2_journal_access(handle, inode, fe_bh,
 				   OCFS2_JOURNAL_ACCESS_WRITE);
 	if (err < 0) {
 		mlog_errno(err);
-		goto bail;
+		goto out_commit;
 	}
 
 	inc_nlink(inode);
@@ -741,7 +722,7 @@ static int ocfs2_link(struct dentry *old_dentry,
 		le16_add_cpu(&fe->i_links_count, -1);
 		drop_nlink(inode);
 		mlog_errno(err);
-		goto bail;
+		goto out_commit;
 	}
 
 	err = ocfs2_add_entry(handle, dentry, inode,
@@ -751,21 +732,27 @@ static int ocfs2_link(struct dentry *old_dentry,
 		le16_add_cpu(&fe->i_links_count, -1);
 		drop_nlink(inode);
 		mlog_errno(err);
-		goto bail;
+		goto out_commit;
 	}
 
 	err = ocfs2_dentry_attach_lock(dentry, inode, OCFS2_I(dir)->ip_blkno);
 	if (err) {
 		mlog_errno(err);
-		goto bail;
+		goto out_commit;
 	}
 
 	atomic_inc(&inode->i_count);
 	dentry->d_op = &ocfs2_dentry_ops;
 	d_instantiate(dentry, inode);
-bail:
-	if (handle)
-		ocfs2_commit_trans(handle);
+
+out_commit:
+	ocfs2_commit_trans(osb, handle);
+out_unlock_inode:
+	ocfs2_meta_unlock(inode, 1);
+
+out:
+	ocfs2_meta_unlock(dir, 1);
+
 	if (de_bh)
 		brelse(de_bh);
 	if (fe_bh)
@@ -812,13 +799,15 @@ static int ocfs2_unlink(struct inode *dir,
 			struct dentry *dentry)
 {
 	int status;
+	int child_locked = 0;
 	struct inode *inode = dentry->d_inode;
+	struct inode *orphan_dir = NULL;
 	struct ocfs2_super *osb = OCFS2_SB(dir->i_sb);
 	u64 blkno;
 	struct ocfs2_dinode *fe = NULL;
 	struct buffer_head *fe_bh = NULL;
 	struct buffer_head *parent_node_bh = NULL;
-	struct ocfs2_journal_handle *handle = NULL;
+	handle_t *handle = NULL;
 	struct ocfs2_dir_entry *dirent = NULL;
 	struct buffer_head *dirent_bh = NULL;
 	char orphan_name[OCFS2_ORPHAN_NAMELEN + 1];
@@ -833,22 +822,14 @@ static int ocfs2_unlink(struct inode *dir,
 
 	if (inode == osb->root_inode) {
 		mlog(0, "Cannot delete the root directory\n");
-		status = -EPERM;
-		goto leave;
+		return -EPERM;
 	}
 
-	handle = ocfs2_alloc_handle(osb);
-	if (handle == NULL) {
-		status = -ENOMEM;
-		mlog_errno(status);
-		goto leave;
-	}
-
-	status = ocfs2_meta_lock(dir, handle, &parent_node_bh, 1);
+	status = ocfs2_meta_lock(dir, &parent_node_bh, 1);
 	if (status < 0) {
 		if (status != -ENOENT)
 			mlog_errno(status);
-		goto leave;
+		return status;
 	}
 
 	status = ocfs2_find_files_on_disk(dentry->d_name.name,
@@ -869,12 +850,13 @@ static int ocfs2_unlink(struct inode *dir,
 		goto leave;
 	}
 
-	status = ocfs2_meta_lock(inode, handle, &fe_bh, 1);
+	status = ocfs2_meta_lock(inode, &fe_bh, 1);
 	if (status < 0) {
 		if (status != -ENOENT)
 			mlog_errno(status);
 		goto leave;
 	}
+	child_locked = 1;
 
 	if (S_ISDIR(inode->i_mode)) {
 	       	if (!ocfs2_empty_dir(inode)) {
@@ -895,7 +877,7 @@ static int ocfs2_unlink(struct inode *dir,
 	}
 
 	if (inode_is_unlinkable(inode)) {
-		status = ocfs2_prepare_orphan_dir(osb, handle, inode,
+		status = ocfs2_prepare_orphan_dir(osb, &orphan_dir, inode,
 						  orphan_name,
 						  &orphan_entry_bh);
 		if (status < 0) {
@@ -904,7 +886,7 @@ static int ocfs2_unlink(struct inode *dir,
 		}
 	}
 
-	handle = ocfs2_start_trans(osb, handle, OCFS2_UNLINK_CREDITS);
+	handle = ocfs2_start_trans(osb, OCFS2_UNLINK_CREDITS);
 	if (IS_ERR(handle)) {
 		status = PTR_ERR(handle);
 		handle = NULL;
@@ -923,7 +905,7 @@ static int ocfs2_unlink(struct inode *dir,
 
 	if (inode_is_unlinkable(inode)) {
 		status = ocfs2_orphan_add(osb, handle, inode, fe, orphan_name,
-					  orphan_entry_bh);
+					  orphan_entry_bh, orphan_dir);
 		if (status < 0) {
 			mlog_errno(status);
 			goto leave;
@@ -960,7 +942,19 @@ static int ocfs2_unlink(struct inode *dir,
 
 leave:
 	if (handle)
-		ocfs2_commit_trans(handle);
+		ocfs2_commit_trans(osb, handle);
+
+	if (child_locked)
+		ocfs2_meta_unlock(inode, 1);
+
+	ocfs2_meta_unlock(dir, 1);
+
+	if (orphan_dir) {
+		/* This was locked for us in ocfs2_prepare_orphan_dir() */
+		ocfs2_meta_unlock(orphan_dir, 1);
+		mutex_unlock(&orphan_dir->i_mutex);
+		iput(orphan_dir);
+	}
 
 	if (fe_bh)
 		brelse(fe_bh);
@@ -984,7 +978,6 @@ leave:
  * if they have the same id, then the 1st one is the only one locked.
  */
 static int ocfs2_double_lock(struct ocfs2_super *osb,
-			     struct ocfs2_journal_handle *handle,
 			     struct buffer_head **bh1,
 			     struct inode *inode1,
 			     struct buffer_head **bh2,
@@ -1000,8 +993,6 @@ static int ocfs2_double_lock(struct ocfs2_super *osb,
 		   (unsigned long long)oi1->ip_blkno,
 		   (unsigned long long)oi2->ip_blkno);
 
-	BUG_ON(!handle);
-
 	if (*bh1)
 		*bh1 = NULL;
 	if (*bh2)
@@ -1021,25 +1012,41 @@ static int ocfs2_double_lock(struct ocfs2_super *osb,
 			inode1 = tmpinode;
 		}
 		/* lock id2 */
-		status = ocfs2_meta_lock(inode2, handle, bh2, 1);
+		status = ocfs2_meta_lock(inode2, bh2, 1);
 		if (status < 0) {
 			if (status != -ENOENT)
 				mlog_errno(status);
 			goto bail;
 		}
 	}
+
 	/* lock id1 */
-	status = ocfs2_meta_lock(inode1, handle, bh1, 1);
+	status = ocfs2_meta_lock(inode1, bh1, 1);
 	if (status < 0) {
+		/*
+		 * An error return must mean that no cluster locks
+		 * were held on function exit.
+		 */
+		if (oi1->ip_blkno != oi2->ip_blkno)
+			ocfs2_meta_unlock(inode2, 1);
+
 		if (status != -ENOENT)
 			mlog_errno(status);
-		goto bail;
 	}
+
 bail:
 	mlog_exit(status);
 	return status;
 }
 
+static void ocfs2_double_unlock(struct inode *inode1, struct inode *inode2)
+{
+	ocfs2_meta_unlock(inode1, 1);
+
+	if (inode1 != inode2)
+		ocfs2_meta_unlock(inode2, 1);
+}
+
 #define PARENT_INO(buffer) \
 	((struct ocfs2_dir_entry *) \
 	 ((char *)buffer + \
@@ -1050,9 +1057,11 @@ static int ocfs2_rename(struct inode *old_dir,
 			struct inode *new_dir,
 			struct dentry *new_dentry)
 {
-	int status = 0, rename_lock = 0;
+	int status = 0, rename_lock = 0, parents_locked = 0;
+	int old_child_locked = 0, new_child_locked = 0;
 	struct inode *old_inode = old_dentry->d_inode;
 	struct inode *new_inode = new_dentry->d_inode;
+	struct inode *orphan_dir = NULL;
 	struct ocfs2_dinode *newfe = NULL;
 	char orphan_name[OCFS2_ORPHAN_NAMELEN + 1];
 	struct buffer_head *orphan_entry_bh = NULL;
@@ -1060,7 +1069,7 @@ static int ocfs2_rename(struct inode *old_dir,
 	struct buffer_head *insert_entry_bh = NULL;
 	struct ocfs2_super *osb = NULL;
 	u64 newfe_blkno;
-	struct ocfs2_journal_handle *handle = NULL;
+	handle_t *handle = NULL;
 	struct buffer_head *old_dir_bh = NULL;
 	struct buffer_head *new_dir_bh = NULL;
 	struct ocfs2_dir_entry *old_de = NULL, *new_de = NULL; // dirent for old_dentry
@@ -1105,21 +1114,14 @@ static int ocfs2_rename(struct inode *old_dir,
 		rename_lock = 1;
 	}
 
-	handle = ocfs2_alloc_handle(osb);
-	if (handle == NULL) {
-		status = -ENOMEM;
-		mlog_errno(status);
-		goto bail;
-	}
-
 	/* if old and new are the same, this'll just do one lock. */
-	status = ocfs2_double_lock(osb, handle,
-				  &old_dir_bh, old_dir,
-				  &new_dir_bh, new_dir);
+	status = ocfs2_double_lock(osb, &old_dir_bh, old_dir,
+				   &new_dir_bh, new_dir);
 	if (status < 0) {
 		mlog_errno(status);
 		goto bail;
 	}
+	parents_locked = 1;
 
 	/* make sure both dirs have bhs
 	 * get an extra ref on old_dir_bh if old==new */
@@ -1140,12 +1142,13 @@ static int ocfs2_rename(struct inode *old_dir,
 	 * the vote thread on other nodes won't have to concurrently
 	 * downconvert the inode and the dentry locks.
 	 */
-	status = ocfs2_meta_lock(old_inode, handle, NULL, 1);
+	status = ocfs2_meta_lock(old_inode, NULL, 1);
 	if (status < 0) {
 		if (status != -ENOENT)
 			mlog_errno(status);
 		goto bail;
 	}
+	old_child_locked = 1;
 
 	status = ocfs2_remote_dentry_delete(old_dentry);
 	if (status < 0) {
@@ -1231,12 +1234,13 @@ static int ocfs2_rename(struct inode *old_dir,
 			goto bail;
 		}
 
-		status = ocfs2_meta_lock(new_inode, handle, &newfe_bh, 1);
+		status = ocfs2_meta_lock(new_inode, &newfe_bh, 1);
 		if (status < 0) {
 			if (status != -ENOENT)
 				mlog_errno(status);
 			goto bail;
 		}
+		new_child_locked = 1;
 
 		status = ocfs2_remote_dentry_delete(new_dentry);
 		if (status < 0) {
@@ -1252,7 +1256,7 @@ static int ocfs2_rename(struct inode *old_dir,
 		     (unsigned long long)newfe_bh->b_blocknr : 0ULL);
 
 		if (S_ISDIR(new_inode->i_mode) || (new_inode->i_nlink == 1)) {
-			status = ocfs2_prepare_orphan_dir(osb, handle,
+			status = ocfs2_prepare_orphan_dir(osb, &orphan_dir,
 							  new_inode,
 							  orphan_name,
 							  &orphan_entry_bh);
@@ -1280,7 +1284,7 @@ static int ocfs2_rename(struct inode *old_dir,
 		}
 	}
 
-	handle = ocfs2_start_trans(osb, handle, OCFS2_RENAME_CREDITS);
+	handle = ocfs2_start_trans(osb, OCFS2_RENAME_CREDITS);
 	if (IS_ERR(handle)) {
 		status = PTR_ERR(handle);
 		handle = NULL;
@@ -1307,7 +1311,7 @@ static int ocfs2_rename(struct inode *old_dir,
 		    (newfe->i_links_count == cpu_to_le16(1))){
 			status = ocfs2_orphan_add(osb, handle, new_inode,
 						  newfe, orphan_name,
-						  orphan_entry_bh);
+						  orphan_entry_bh, orphan_dir);
 			if (status < 0) {
 				mlog_errno(status);
 				goto bail;
@@ -1424,7 +1428,23 @@ bail:
 		ocfs2_rename_unlock(osb);
 
 	if (handle)
-		ocfs2_commit_trans(handle);
+		ocfs2_commit_trans(osb, handle);
+
+	if (parents_locked)
+		ocfs2_double_unlock(old_dir, new_dir);
+
+	if (old_child_locked)
+		ocfs2_meta_unlock(old_inode, 1);
+
+	if (new_child_locked)
+		ocfs2_meta_unlock(new_inode, 1);
+
+	if (orphan_dir) {
+		/* This was locked for us in ocfs2_prepare_orphan_dir() */
+		ocfs2_meta_unlock(orphan_dir, 1);
+		mutex_unlock(&orphan_dir->i_mutex);
+		iput(orphan_dir);
+	}
 
 	if (new_inode)
 		sync_mapping_buffers(old_inode->i_mapping);
@@ -1458,7 +1478,7 @@ bail:
  * data, including the null terminator.
  */
 static int ocfs2_create_symlink_data(struct ocfs2_super *osb,
-				     struct ocfs2_journal_handle *handle,
+				     handle_t *handle,
 				     struct inode *inode,
 				     const char *symname)
 {
@@ -1573,7 +1593,7 @@ static int ocfs2_symlink(struct inode *dir,
 	struct buffer_head *parent_fe_bh = NULL;
 	struct ocfs2_dinode *fe = NULL;
 	struct ocfs2_dinode *dirfe;
-	struct ocfs2_journal_handle *handle = NULL;
+	handle_t *handle = NULL;
 	struct ocfs2_alloc_context *inode_ac = NULL;
 	struct ocfs2_alloc_context *data_ac = NULL;
 
@@ -1587,19 +1607,12 @@ static int ocfs2_symlink(struct inode *dir,
 
 	credits = ocfs2_calc_symlink_credits(sb);
 
-	handle = ocfs2_alloc_handle(osb);
-	if (handle == NULL) {
-		status = -ENOMEM;
-		mlog_errno(status);
-		goto bail;
-	}
-
 	/* lock the parent directory */
-	status = ocfs2_meta_lock(dir, handle, &parent_fe_bh, 1);
+	status = ocfs2_meta_lock(dir, &parent_fe_bh, 1);
 	if (status < 0) {
 		if (status != -ENOENT)
 			mlog_errno(status);
-		goto bail;
+		return status;
 	}
 
 	dirfe = (struct ocfs2_dinode *) parent_fe_bh->b_data;
@@ -1622,7 +1635,7 @@ static int ocfs2_symlink(struct inode *dir,
 		goto bail;
 	}
 
-	status = ocfs2_reserve_new_inode(osb, handle, &inode_ac);
+	status = ocfs2_reserve_new_inode(osb, &inode_ac);
 	if (status < 0) {
 		if (status != -ENOSPC)
 			mlog_errno(status);
@@ -1631,7 +1644,7 @@ static int ocfs2_symlink(struct inode *dir,
 
 	/* don't reserve bitmap space for fast symlinks. */
 	if (l > ocfs2_fast_symlink_chars(sb)) {
-		status = ocfs2_reserve_clusters(osb, handle, 1, &data_ac);
+		status = ocfs2_reserve_clusters(osb, 1, &data_ac);
 		if (status < 0) {
 			if (status != -ENOSPC)
 				mlog_errno(status);
@@ -1639,7 +1652,7 @@ static int ocfs2_symlink(struct inode *dir,
 		}
 	}
 
-	handle = ocfs2_start_trans(osb, handle, credits);
+	handle = ocfs2_start_trans(osb, credits);
 	if (IS_ERR(handle)) {
 		status = PTR_ERR(handle);
 		handle = NULL;
@@ -1717,7 +1730,10 @@ static int ocfs2_symlink(struct inode *dir,
 	d_instantiate(dentry, inode);
 bail:
 	if (handle)
-		ocfs2_commit_trans(handle);
+		ocfs2_commit_trans(osb, handle);
+
+	ocfs2_meta_unlock(dir, 1);
+
 	if (new_fe_bh)
 		brelse(new_fe_bh);
 	if (parent_fe_bh)
@@ -1768,7 +1784,7 @@ int ocfs2_check_dir_entry(struct inode * dir,
  * If you pass me insert_bh, I'll skip the search of the other dir
  * blocks and put the record in there.
  */
-static int __ocfs2_add_entry(struct ocfs2_journal_handle *handle,
+static int __ocfs2_add_entry(handle_t *handle,
 			     struct inode *dir,
 			     const char *name, int namelen,
 			     struct inode *inode, u64 blkno,
@@ -1854,7 +1870,7 @@ bail:
  * ocfs2_delete_entry deletes a directory entry by merging it with the
  * previous entry
  */
-static int ocfs2_delete_entry(struct ocfs2_journal_handle *handle,
+static int ocfs2_delete_entry(handle_t *handle,
 			      struct inode *dir,
 			      struct ocfs2_dir_entry *de_del,
 			      struct buffer_head *bh)
@@ -2085,19 +2101,19 @@ bail:
 }
 
 static int ocfs2_prepare_orphan_dir(struct ocfs2_super *osb,
-				    struct ocfs2_journal_handle *handle,
+				    struct inode **ret_orphan_dir,
 				    struct inode *inode,
 				    char *name,
 				    struct buffer_head **de_bh)
 {
-	struct inode *orphan_dir_inode = NULL;
+	struct inode *orphan_dir_inode;
 	struct buffer_head *orphan_dir_bh = NULL;
 	int status = 0;
 
 	status = ocfs2_blkno_stringify(OCFS2_I(inode)->ip_blkno, name);
 	if (status < 0) {
 		mlog_errno(status);
-		goto leave;
+		return status;
 	}
 
 	orphan_dir_inode = ocfs2_get_system_file_inode(osb,
@@ -2106,11 +2122,12 @@ static int ocfs2_prepare_orphan_dir(struct ocfs2_super *osb,
 	if (!orphan_dir_inode) {
 		status = -ENOENT;
 		mlog_errno(status);
-		goto leave;
+		return status;
 	}
 
-	ocfs2_handle_add_inode(handle, orphan_dir_inode);
-	status = ocfs2_meta_lock(orphan_dir_inode, handle, &orphan_dir_bh, 1);
+	mutex_lock(&orphan_dir_inode->i_mutex);
+
+	status = ocfs2_meta_lock(orphan_dir_inode, &orphan_dir_bh, 1);
 	if (status < 0) {
 		mlog_errno(status);
 		goto leave;
@@ -2120,13 +2137,19 @@ static int ocfs2_prepare_orphan_dir(struct ocfs2_super *osb,
 					      orphan_dir_bh, name,
 					      OCFS2_ORPHAN_NAMELEN, de_bh);
 	if (status < 0) {
+		ocfs2_meta_unlock(orphan_dir_inode, 1);
+
 		mlog_errno(status);
 		goto leave;
 	}
 
+	*ret_orphan_dir = orphan_dir_inode;
+
 leave:
-	if (orphan_dir_inode)
+	if (status) {
+		mutex_unlock(&orphan_dir_inode->i_mutex);
 		iput(orphan_dir_inode);
+	}
 
 	if (orphan_dir_bh)
 		brelse(orphan_dir_bh);
@@ -2136,28 +2159,19 @@ leave:
 }
 
 static int ocfs2_orphan_add(struct ocfs2_super *osb,
-			    struct ocfs2_journal_handle *handle,
+			    handle_t *handle,
 			    struct inode *inode,
 			    struct ocfs2_dinode *fe,
 			    char *name,
-			    struct buffer_head *de_bh)
+			    struct buffer_head *de_bh,
+			    struct inode *orphan_dir_inode)
 {
-	struct inode *orphan_dir_inode = NULL;
 	struct buffer_head *orphan_dir_bh = NULL;
 	int status = 0;
 	struct ocfs2_dinode *orphan_fe;
 
 	mlog_entry("(inode->i_ino = %lu)\n", inode->i_ino);
 
-	orphan_dir_inode = ocfs2_get_system_file_inode(osb,
-						       ORPHAN_DIR_SYSTEM_INODE,
-						       osb->slot_num);
-	if (!orphan_dir_inode) {
-		status = -ENOENT;
-		mlog_errno(status);
-		goto leave;
-	}
-
 	status = ocfs2_read_block(osb,
 				  OCFS2_I(orphan_dir_inode)->ip_blkno,
 				  &orphan_dir_bh, OCFS2_BH_CACHED,
@@ -2209,9 +2223,6 @@ static int ocfs2_orphan_add(struct ocfs2_super *osb,
 	     (unsigned long long)OCFS2_I(inode)->ip_blkno, osb->slot_num);
 
 leave:
-	if (orphan_dir_inode)
-		iput(orphan_dir_inode);
-
 	if (orphan_dir_bh)
 		brelse(orphan_dir_bh);
 
@@ -2221,7 +2232,7 @@ leave:
 
 /* unlike orphan_add, we expect the orphan dir to already be locked here. */
 int ocfs2_orphan_del(struct ocfs2_super *osb,
-		     struct ocfs2_journal_handle *handle,
+		     handle_t *handle,
 		     struct inode *orphan_dir_inode,
 		     struct inode *inode,
 		     struct buffer_head *orphan_dir_bh)
@@ -2300,4 +2311,5 @@ struct inode_operations ocfs2_dir_iops = {
 	.rename		= ocfs2_rename,
 	.setattr	= ocfs2_setattr,
 	.getattr	= ocfs2_getattr,
+	.permission	= ocfs2_permission,
 };

fs/ocfs2/namei.h

@@ -39,7 +39,7 @@ struct buffer_head *ocfs2_find_entry(const char *name,
 				     struct inode *dir,
 				     struct ocfs2_dir_entry **res_dir);
 int ocfs2_orphan_del(struct ocfs2_super *osb,
-		     struct ocfs2_journal_handle *handle,
+		     handle_t *handle,
 		     struct inode *orphan_dir_inode,
 		     struct inode *inode,
 		     struct buffer_head *orphan_dir_bh);

fs/ocfs2/ocfs2.h

@@ -34,6 +34,7 @@
 #include <linux/workqueue.h>
 #include <linux/kref.h>
 #include <linux/mutex.h>
+#include <linux/jbd.h>
 
 #include "cluster/nodemanager.h"
 #include "cluster/heartbeat.h"
@@ -179,9 +180,9 @@ enum ocfs2_mount_options
 #define OCFS2_OSB_SOFT_RO	0x0001
 #define OCFS2_OSB_HARD_RO	0x0002
 #define OCFS2_OSB_ERROR_FS	0x0004
+#define OCFS2_DEFAULT_ATIME_QUANTUM	60
 
 struct ocfs2_journal;
-struct ocfs2_journal_handle;
 struct ocfs2_super
 {
 	struct task_struct *commit_task;
@@ -218,6 +219,7 @@ struct ocfs2_super
 	unsigned long osb_flags;
 
 	unsigned long s_mount_opt;
+	unsigned int s_atime_quantum;
 
 	u16 max_slots;
 	s16 node_num;

fs/ocfs2/suballoc.c

@@ -49,7 +49,7 @@
 static inline void ocfs2_debug_bg(struct ocfs2_group_desc *bg);
 static inline void ocfs2_debug_suballoc_inode(struct ocfs2_dinode *fe);
 static inline u16 ocfs2_find_victim_chain(struct ocfs2_chain_list *cl);
-static int ocfs2_block_group_fill(struct ocfs2_journal_handle *handle,
+static int ocfs2_block_group_fill(handle_t *handle,
 				  struct inode *alloc_inode,
 				  struct buffer_head *bg_bh,
 				  u64 group_blkno,
@@ -59,9 +59,6 @@ static int ocfs2_block_group_alloc(struct ocfs2_super *osb,
 				   struct inode *alloc_inode,
 				   struct buffer_head *bh);
 
-static int ocfs2_reserve_suballoc_bits(struct ocfs2_super *osb,
-				       struct ocfs2_alloc_context *ac);
-
 static int ocfs2_cluster_group_search(struct inode *inode,
 				      struct buffer_head *group_bh,
 				      u32 bits_wanted, u32 min_bits,
@@ -72,6 +69,7 @@ static int ocfs2_block_group_search(struct inode *inode,
 				    u16 *bit_off, u16 *bits_found);
 static int ocfs2_claim_suballoc_bits(struct ocfs2_super *osb,
 				     struct ocfs2_alloc_context *ac,
+				     handle_t *handle,
 				     u32 bits_wanted,
 				     u32 min_bits,
 				     u16 *bit_off,
@@ -79,20 +77,20 @@ static int ocfs2_claim_suballoc_bits(struct ocfs2_super *osb,
 				     u64 *bg_blkno);
 static int ocfs2_test_bg_bit_allocatable(struct buffer_head *bg_bh,
 					 int nr);
-static inline int ocfs2_block_group_set_bits(struct ocfs2_journal_handle *handle,
+static inline int ocfs2_block_group_set_bits(handle_t *handle,
 					     struct inode *alloc_inode,
 					     struct ocfs2_group_desc *bg,
 					     struct buffer_head *group_bh,
 					     unsigned int bit_off,
 					     unsigned int num_bits);
-static inline int ocfs2_block_group_clear_bits(struct ocfs2_journal_handle *handle,
+static inline int ocfs2_block_group_clear_bits(handle_t *handle,
 					       struct inode *alloc_inode,
 					       struct ocfs2_group_desc *bg,
 					       struct buffer_head *group_bh,
 					       unsigned int bit_off,
 					       unsigned int num_bits);
 
-static int ocfs2_relink_block_group(struct ocfs2_journal_handle *handle,
+static int ocfs2_relink_block_group(handle_t *handle,
 				    struct inode *alloc_inode,
 				    struct buffer_head *fe_bh,
 				    struct buffer_head *bg_bh,
@@ -100,7 +98,7 @@ static int ocfs2_relink_block_group(struct ocfs2_journal_handle *handle,
 				    u16 chain);
 static inline int ocfs2_block_group_reasonably_empty(struct ocfs2_group_desc *bg,
 						     u32 wanted);
-static int ocfs2_free_suballoc_bits(struct ocfs2_journal_handle *handle,
+static int ocfs2_free_suballoc_bits(handle_t *handle,
 				    struct inode *alloc_inode,
 				    struct buffer_head *alloc_bh,
 				    unsigned int start_bit,
@@ -120,8 +118,16 @@ static inline void ocfs2_block_to_cluster_group(struct inode *inode,
 
 void ocfs2_free_alloc_context(struct ocfs2_alloc_context *ac)
 {
-	if (ac->ac_inode)
-		iput(ac->ac_inode);
+	struct inode *inode = ac->ac_inode;
+
+	if (inode) {
+		if (ac->ac_which != OCFS2_AC_USE_LOCAL)
+			ocfs2_meta_unlock(inode, 1);
+
+		mutex_unlock(&inode->i_mutex);
+
+		iput(inode);
+	}
 	if (ac->ac_bh)
 		brelse(ac->ac_bh);
 	kfree(ac);
@@ -190,7 +196,7 @@ static int ocfs2_check_group_descriptor(struct super_block *sb,
 	return 0;
 }
 
-static int ocfs2_block_group_fill(struct ocfs2_journal_handle *handle,
+static int ocfs2_block_group_fill(handle_t *handle,
 				  struct inode *alloc_inode,
 				  struct buffer_head *bg_bh,
 				  u64 group_blkno,
@@ -273,7 +279,7 @@ static int ocfs2_block_group_alloc(struct ocfs2_super *osb,
 	struct ocfs2_dinode *fe = (struct ocfs2_dinode *) bh->b_data;
 	struct ocfs2_chain_list *cl;
 	struct ocfs2_alloc_context *ac = NULL;
-	struct ocfs2_journal_handle *handle = NULL;
+	handle_t *handle = NULL;
 	u32 bit_off, num_bits;
 	u16 alloc_rec;
 	u64 bg_blkno;
@@ -284,16 +290,8 @@ static int ocfs2_block_group_alloc(struct ocfs2_super *osb,
 
 	mlog_entry_void();
 
-	handle = ocfs2_alloc_handle(osb);
-	if (!handle) {
-		status = -ENOMEM;
-		mlog_errno(status);
-		goto bail;
-	}
-
 	cl = &fe->id2.i_chain;
 	status = ocfs2_reserve_clusters(osb,
-					handle,
 					le16_to_cpu(cl->cl_cpg),
 					&ac);
 	if (status < 0) {
@@ -304,7 +302,7 @@ static int ocfs2_block_group_alloc(struct ocfs2_super *osb,
 
 	credits = ocfs2_calc_group_alloc_credits(osb->sb,
 						 le16_to_cpu(cl->cl_cpg));
-	handle = ocfs2_start_trans(osb, handle, credits);
+	handle = ocfs2_start_trans(osb, credits);
 	if (IS_ERR(handle)) {
 		status = PTR_ERR(handle);
 		handle = NULL;
@@ -389,7 +387,7 @@ static int ocfs2_block_group_alloc(struct ocfs2_super *osb,
 	status = 0;
 bail:
 	if (handle)
-		ocfs2_commit_trans(handle);
+		ocfs2_commit_trans(osb, handle);
 
 	if (ac)
 		ocfs2_free_alloc_context(ac);
@@ -402,27 +400,38 @@ bail:
 }
 
 static int ocfs2_reserve_suballoc_bits(struct ocfs2_super *osb,
-				       struct ocfs2_alloc_context *ac)
+				       struct ocfs2_alloc_context *ac,
+				       int type,
+				       u32 slot)
 {
 	int status;
 	u32 bits_wanted = ac->ac_bits_wanted;
-	struct inode *alloc_inode = ac->ac_inode;
+	struct inode *alloc_inode;
 	struct buffer_head *bh = NULL;
-	struct ocfs2_journal_handle *handle = ac->ac_handle;
 	struct ocfs2_dinode *fe;
 	u32 free_bits;
 
 	mlog_entry_void();
 
-	BUG_ON(handle->flags & OCFS2_HANDLE_STARTED);
+	alloc_inode = ocfs2_get_system_file_inode(osb, type, slot);
+	if (!alloc_inode) {
+		mlog_errno(-EINVAL);
+		return -EINVAL;
+	}
 
-	ocfs2_handle_add_inode(handle, alloc_inode);
-	status = ocfs2_meta_lock(alloc_inode, handle, &bh, 1);
+	mutex_lock(&alloc_inode->i_mutex);
+
+	status = ocfs2_meta_lock(alloc_inode, &bh, 1);
 	if (status < 0) {
+		mutex_unlock(&alloc_inode->i_mutex);
+		iput(alloc_inode);
+
 		mlog_errno(status);
-		goto bail;
+		return status;
 	}
 
+	ac->ac_inode = alloc_inode;
+
 	fe = (struct ocfs2_dinode *) bh->b_data;
 	if (!OCFS2_IS_VALID_DINODE(fe)) {
 		OCFS2_RO_ON_INVALID_DINODE(alloc_inode->i_sb, fe);
@@ -473,12 +482,11 @@ bail:
 }
 
 int ocfs2_reserve_new_metadata(struct ocfs2_super *osb,
-			       struct ocfs2_journal_handle *handle,
 			       struct ocfs2_dinode *fe,
 			       struct ocfs2_alloc_context **ac)
 {
 	int status;
-	struct inode *alloc_inode = NULL;
+	u32 slot;
 
 	*ac = kcalloc(1, sizeof(struct ocfs2_alloc_context), GFP_KERNEL);
 	if (!(*ac)) {
@@ -488,28 +496,18 @@ int ocfs2_reserve_new_metadata(struct ocfs2_super *osb,
 	}
 
 	(*ac)->ac_bits_wanted = ocfs2_extend_meta_needed(fe);
-	(*ac)->ac_handle = handle;
 	(*ac)->ac_which = OCFS2_AC_USE_META;
 
 #ifndef OCFS2_USE_ALL_METADATA_SUBALLOCATORS
-	alloc_inode = ocfs2_get_system_file_inode(osb,
-						  EXTENT_ALLOC_SYSTEM_INODE,
-						  0);
+	slot = 0;
 #else
-	alloc_inode = ocfs2_get_system_file_inode(osb,
-						  EXTENT_ALLOC_SYSTEM_INODE,
-						  osb->slot_num);
+	slot = osb->slot_num;
 #endif
-	if (!alloc_inode) {
-		status = -ENOMEM;
-		mlog_errno(status);
-		goto bail;
-	}
 
-	(*ac)->ac_inode = igrab(alloc_inode);
 	(*ac)->ac_group_search = ocfs2_block_group_search;
 
-	status = ocfs2_reserve_suballoc_bits(osb, (*ac));
+	status = ocfs2_reserve_suballoc_bits(osb, (*ac),
+					     EXTENT_ALLOC_SYSTEM_INODE, slot);
 	if (status < 0) {
 		if (status != -ENOSPC)
 			mlog_errno(status);
@@ -523,19 +521,14 @@ bail:
 		*ac = NULL;
 	}
 
-	if (alloc_inode)
-		iput(alloc_inode);
-
 	mlog_exit(status);
 	return status;
 }
 
 int ocfs2_reserve_new_inode(struct ocfs2_super *osb,
-			    struct ocfs2_journal_handle *handle,
 			    struct ocfs2_alloc_context **ac)
 {
 	int status;
-	struct inode *alloc_inode = NULL;
 
 	*ac = kcalloc(1, sizeof(struct ocfs2_alloc_context), GFP_KERNEL);
 	if (!(*ac)) {
@@ -545,22 +538,13 @@ int ocfs2_reserve_new_inode(struct ocfs2_super *osb,
 	}
 
 	(*ac)->ac_bits_wanted = 1;
-	(*ac)->ac_handle = handle;
 	(*ac)->ac_which = OCFS2_AC_USE_INODE;
 
-	alloc_inode = ocfs2_get_system_file_inode(osb,
-						  INODE_ALLOC_SYSTEM_INODE,
-						  osb->slot_num);
-	if (!alloc_inode) {
-		status = -ENOMEM;
-		mlog_errno(status);
-		goto bail;
-	}
-
-	(*ac)->ac_inode = igrab(alloc_inode);
 	(*ac)->ac_group_search = ocfs2_block_group_search;
 
-	status = ocfs2_reserve_suballoc_bits(osb, *ac);
+	status = ocfs2_reserve_suballoc_bits(osb, *ac,
+					     INODE_ALLOC_SYSTEM_INODE,
+					     osb->slot_num);
 	if (status < 0) {
 		if (status != -ENOSPC)
 			mlog_errno(status);
@@ -574,9 +558,6 @@ bail:
 		*ac = NULL;
 	}
 
-	if (alloc_inode)
-		iput(alloc_inode);
-
 	mlog_exit(status);
 	return status;
 }
@@ -588,20 +569,17 @@ int ocfs2_reserve_cluster_bitmap_bits(struct ocfs2_super *osb,
 {
 	int status;
 
-	ac->ac_inode = ocfs2_get_system_file_inode(osb,
-						   GLOBAL_BITMAP_SYSTEM_INODE,
-						   OCFS2_INVALID_SLOT);
-	if (!ac->ac_inode) {
-		status = -EINVAL;
-		mlog(ML_ERROR, "Could not get bitmap inode!\n");
-		goto bail;
-	}
 	ac->ac_which = OCFS2_AC_USE_MAIN;
 	ac->ac_group_search = ocfs2_cluster_group_search;
 
-	status = ocfs2_reserve_suballoc_bits(osb, ac);
-	if (status < 0 && status != -ENOSPC)
+	status = ocfs2_reserve_suballoc_bits(osb, ac,
+					     GLOBAL_BITMAP_SYSTEM_INODE,
+					     OCFS2_INVALID_SLOT);
+	if (status < 0 && status != -ENOSPC) {
 		mlog_errno(status);
+		goto bail;
+	}
+
 bail:
 	return status;
 }
@@ -610,7 +588,6 @@ bail:
  * use so we figure it out for them, but unfortunately this clutters
  * things a bit. */
 int ocfs2_reserve_clusters(struct ocfs2_super *osb,
-			   struct ocfs2_journal_handle *handle,
 			   u32 bits_wanted,
 			   struct ocfs2_alloc_context **ac)
 {
@@ -618,8 +595,6 @@ int ocfs2_reserve_clusters(struct ocfs2_super *osb,
 
 	mlog_entry_void();
 
-	BUG_ON(!handle);
-
 	*ac = kcalloc(1, sizeof(struct ocfs2_alloc_context), GFP_KERNEL);
 	if (!(*ac)) {
 		status = -ENOMEM;
@@ -628,12 +603,10 @@ int ocfs2_reserve_clusters(struct ocfs2_super *osb,
 	}
 
 	(*ac)->ac_bits_wanted = bits_wanted;
-	(*ac)->ac_handle = handle;
 
 	status = -ENOSPC;
 	if (ocfs2_alloc_should_use_local(osb, bits_wanted)) {
 		status = ocfs2_reserve_local_alloc_bits(osb,
-							handle,
 							bits_wanted,
 							*ac);
 		if ((status < 0) && (status != -ENOSPC)) {
@@ -774,7 +747,7 @@ static int ocfs2_block_group_find_clear_bits(struct ocfs2_super *osb,
 	return status;
 }
 
-static inline int ocfs2_block_group_set_bits(struct ocfs2_journal_handle *handle,
+static inline int ocfs2_block_group_set_bits(handle_t *handle,
 					     struct inode *alloc_inode,
 					     struct ocfs2_group_desc *bg,
 					     struct buffer_head *group_bh,
@@ -845,7 +818,7 @@ static inline u16 ocfs2_find_victim_chain(struct ocfs2_chain_list *cl)
 	return best;
 }
 
-static int ocfs2_relink_block_group(struct ocfs2_journal_handle *handle,
+static int ocfs2_relink_block_group(handle_t *handle,
 				    struct inode *alloc_inode,
 				    struct buffer_head *fe_bh,
 				    struct buffer_head *bg_bh,
@@ -1025,7 +998,7 @@ static int ocfs2_block_group_search(struct inode *inode,
 }
 
 static int ocfs2_alloc_dinode_update_counts(struct inode *inode,
-				       struct ocfs2_journal_handle *handle,
+				       handle_t *handle,
 				       struct buffer_head *di_bh,
 				       u32 num_bits,
 				       u16 chain)
@@ -1055,6 +1028,7 @@ out:
 }
 
 static int ocfs2_search_one_group(struct ocfs2_alloc_context *ac,
+				  handle_t *handle,
 				  u32 bits_wanted,
 				  u32 min_bits,
 				  u16 *bit_off,
@@ -1067,7 +1041,6 @@ static int ocfs2_search_one_group(struct ocfs2_alloc_context *ac,
 	struct buffer_head *group_bh = NULL;
 	struct ocfs2_group_desc *gd;
 	struct inode *alloc_inode = ac->ac_inode;
-	struct ocfs2_journal_handle *handle = ac->ac_handle;
 
 	ret = ocfs2_read_block(OCFS2_SB(alloc_inode->i_sb), gd_blkno,
 			       &group_bh, OCFS2_BH_CACHED, alloc_inode);
@@ -1115,6 +1088,7 @@ out:
 }
 
 static int ocfs2_search_chain(struct ocfs2_alloc_context *ac,
+			      handle_t *handle,
 			      u32 bits_wanted,
 			      u32 min_bits,
 			      u16 *bit_off,
@@ -1126,7 +1100,6 @@ static int ocfs2_search_chain(struct ocfs2_alloc_context *ac,
 	u16 chain, tmp_bits;
 	u32 tmp_used;
 	u64 next_group;
-	struct ocfs2_journal_handle *handle = ac->ac_handle;
 	struct inode *alloc_inode = ac->ac_inode;
 	struct buffer_head *group_bh = NULL;
 	struct buffer_head *prev_group_bh = NULL;
@@ -1272,6 +1245,7 @@ bail:
 /* will give out up to bits_wanted contiguous bits. */
 static int ocfs2_claim_suballoc_bits(struct ocfs2_super *osb,
 				     struct ocfs2_alloc_context *ac,
+				     handle_t *handle,
 				     u32 bits_wanted,
 				     u32 min_bits,
 				     u16 *bit_off,
@@ -1313,8 +1287,8 @@ static int ocfs2_claim_suballoc_bits(struct ocfs2_super *osb,
 		 * by jumping straight to the most recently used
 		 * allocation group. This helps us mantain some
 		 * contiguousness across allocations. */
-		status = ocfs2_search_one_group(ac, bits_wanted, min_bits,
-						bit_off, num_bits,
+		status = ocfs2_search_one_group(ac, handle, bits_wanted,
+						min_bits, bit_off, num_bits,
 						hint_blkno, &bits_left);
 		if (!status) {
 			/* Be careful to update *bg_blkno here as the
@@ -1336,7 +1310,7 @@ static int ocfs2_claim_suballoc_bits(struct ocfs2_super *osb,
 	ac->ac_chain = victim;
 	ac->ac_allow_chain_relink = 1;
 
-	status = ocfs2_search_chain(ac, bits_wanted, min_bits, bit_off,
+	status = ocfs2_search_chain(ac, handle, bits_wanted, min_bits, bit_off,
 				    num_bits, bg_blkno, &bits_left);
 	if (!status)
 		goto set_hint;
@@ -1360,7 +1334,7 @@ static int ocfs2_claim_suballoc_bits(struct ocfs2_super *osb,
 			continue;
 
 		ac->ac_chain = i;
-		status = ocfs2_search_chain(ac, bits_wanted, min_bits,
+		status = ocfs2_search_chain(ac, handle, bits_wanted, min_bits,
 					    bit_off, num_bits, bg_blkno,
 					    &bits_left);
 		if (!status)
@@ -1388,7 +1362,7 @@ bail:
 }
 
 int ocfs2_claim_metadata(struct ocfs2_super *osb,
-			 struct ocfs2_journal_handle *handle,
+			 handle_t *handle,
 			 struct ocfs2_alloc_context *ac,
 			 u32 bits_wanted,
 			 u16 *suballoc_bit_start,
@@ -1401,10 +1375,10 @@ int ocfs2_claim_metadata(struct ocfs2_super *osb,
 	BUG_ON(!ac);
 	BUG_ON(ac->ac_bits_wanted < (ac->ac_bits_given + bits_wanted));
 	BUG_ON(ac->ac_which != OCFS2_AC_USE_META);
-	BUG_ON(ac->ac_handle != handle);
 
 	status = ocfs2_claim_suballoc_bits(osb,
 					   ac,
+					   handle,
 					   bits_wanted,
 					   1,
 					   suballoc_bit_start,
@@ -1425,7 +1399,7 @@ bail:
 }
 
 int ocfs2_claim_new_inode(struct ocfs2_super *osb,
-			  struct ocfs2_journal_handle *handle,
+			  handle_t *handle,
 			  struct ocfs2_alloc_context *ac,
 			  u16 *suballoc_bit,
 			  u64 *fe_blkno)
@@ -1440,10 +1414,10 @@ int ocfs2_claim_new_inode(struct ocfs2_super *osb,
 	BUG_ON(ac->ac_bits_given != 0);
 	BUG_ON(ac->ac_bits_wanted != 1);
 	BUG_ON(ac->ac_which != OCFS2_AC_USE_INODE);
-	BUG_ON(ac->ac_handle != handle);
 
 	status = ocfs2_claim_suballoc_bits(osb,
 					   ac,
+					   handle,
 					   1,
 					   1,
 					   suballoc_bit,
@@ -1528,7 +1502,7 @@ static inline void ocfs2_block_to_cluster_group(struct inode *inode,
  * of any size.
  */
 int ocfs2_claim_clusters(struct ocfs2_super *osb,
-			 struct ocfs2_journal_handle *handle,
+			 handle_t *handle,
 			 struct ocfs2_alloc_context *ac,
 			 u32 min_clusters,
 			 u32 *cluster_start,
@@ -1546,7 +1520,6 @@ int ocfs2_claim_clusters(struct ocfs2_super *osb,
 
 	BUG_ON(ac->ac_which != OCFS2_AC_USE_LOCAL
 	       && ac->ac_which != OCFS2_AC_USE_MAIN);
-	BUG_ON(ac->ac_handle != handle);
 
 	if (ac->ac_which == OCFS2_AC_USE_LOCAL) {
 		status = ocfs2_claim_local_alloc_bits(osb,
@@ -1572,6 +1545,7 @@ int ocfs2_claim_clusters(struct ocfs2_super *osb,
 
 		status = ocfs2_claim_suballoc_bits(osb,
 						   ac,
+						   handle,
 						   bits_wanted,
 						   min_clusters,
 						   &bg_bit_off,
@@ -1598,7 +1572,7 @@ bail:
 	return status;
 }
 
-static inline int ocfs2_block_group_clear_bits(struct ocfs2_journal_handle *handle,
+static inline int ocfs2_block_group_clear_bits(handle_t *handle,
 					       struct inode *alloc_inode,
 					       struct ocfs2_group_desc *bg,
 					       struct buffer_head *group_bh,
@@ -1653,7 +1627,7 @@ bail:
 /*
  * expects the suballoc inode to already be locked.
  */
-static int ocfs2_free_suballoc_bits(struct ocfs2_journal_handle *handle,
+static int ocfs2_free_suballoc_bits(handle_t *handle,
 				    struct inode *alloc_inode,
 				    struct buffer_head *alloc_bh,
 				    unsigned int start_bit,
@@ -1737,7 +1711,7 @@ static inline u64 ocfs2_which_suballoc_group(u64 block, unsigned int bit)
 	return group;
 }
 
-int ocfs2_free_dinode(struct ocfs2_journal_handle *handle,
+int ocfs2_free_dinode(handle_t *handle,
 		      struct inode *inode_alloc_inode,
 		      struct buffer_head *inode_alloc_bh,
 		      struct ocfs2_dinode *di)
@@ -1750,7 +1724,7 @@ int ocfs2_free_dinode(struct ocfs2_journal_handle *handle,
 					inode_alloc_bh, bit, bg_blkno, 1);
 }
 
-int ocfs2_free_extent_block(struct ocfs2_journal_handle *handle,
+int ocfs2_free_extent_block(handle_t *handle,
 			    struct inode *eb_alloc_inode,
 			    struct buffer_head *eb_alloc_bh,
 			    struct ocfs2_extent_block *eb)
@@ -1763,7 +1737,7 @@ int ocfs2_free_extent_block(struct ocfs2_journal_handle *handle,
 					bit, bg_blkno, 1);
 }
 
-int ocfs2_free_clusters(struct ocfs2_journal_handle *handle,
+int ocfs2_free_clusters(handle_t *handle,
 		       struct inode *bitmap_inode,
 		       struct buffer_head *bitmap_bh,
 		       u64 start_blk,

fs/ocfs2/suballoc.h

@@ -43,7 +43,6 @@ struct ocfs2_alloc_context {
 #define OCFS2_AC_USE_INODE 3
 #define OCFS2_AC_USE_META  4
 	u32    ac_which;
-	struct ocfs2_journal_handle *ac_handle;
 
 	/* these are used by the chain search */
 	u16    ac_chain;
@@ -60,45 +59,42 @@ static inline int ocfs2_alloc_context_bits_left(struct ocfs2_alloc_context *ac)
 }
 
 int ocfs2_reserve_new_metadata(struct ocfs2_super *osb,
-			       struct ocfs2_journal_handle *handle,
 			       struct ocfs2_dinode *fe,
 			       struct ocfs2_alloc_context **ac);
 int ocfs2_reserve_new_inode(struct ocfs2_super *osb,
-			    struct ocfs2_journal_handle *handle,
 			    struct ocfs2_alloc_context **ac);
 int ocfs2_reserve_clusters(struct ocfs2_super *osb,
-			   struct ocfs2_journal_handle *handle,
 			   u32 bits_wanted,
 			   struct ocfs2_alloc_context **ac);
 
 int ocfs2_claim_metadata(struct ocfs2_super *osb,
-			 struct ocfs2_journal_handle *handle,
+			 handle_t *handle,
 			 struct ocfs2_alloc_context *ac,
 			 u32 bits_wanted,
 			 u16 *suballoc_bit_start,
 			 u32 *num_bits,
 			 u64 *blkno_start);
 int ocfs2_claim_new_inode(struct ocfs2_super *osb,
-			  struct ocfs2_journal_handle *handle,
+			  handle_t *handle,
 			  struct ocfs2_alloc_context *ac,
 			  u16 *suballoc_bit,
 			  u64 *fe_blkno);
 int ocfs2_claim_clusters(struct ocfs2_super *osb,
-			 struct ocfs2_journal_handle *handle,
+			 handle_t *handle,
 			 struct ocfs2_alloc_context *ac,
 			 u32 min_clusters,
 			 u32 *cluster_start,
 			 u32 *num_clusters);
 
-int ocfs2_free_dinode(struct ocfs2_journal_handle *handle,
+int ocfs2_free_dinode(handle_t *handle,
 		      struct inode *inode_alloc_inode,
 		      struct buffer_head *inode_alloc_bh,
 		      struct ocfs2_dinode *di);
-int ocfs2_free_extent_block(struct ocfs2_journal_handle *handle,
+int ocfs2_free_extent_block(handle_t *handle,
 			    struct inode *eb_alloc_inode,
 			    struct buffer_head *eb_alloc_bh,
 			    struct ocfs2_extent_block *eb);
-int ocfs2_free_clusters(struct ocfs2_journal_handle *handle,
+int ocfs2_free_clusters(handle_t *handle,
 			struct inode *bitmap_inode,
 			struct buffer_head *bitmap_bh,
 			u64 start_blk,

fs/ocfs2/super.c

@@ -70,8 +70,6 @@
 
 static kmem_cache_t *ocfs2_inode_cachep = NULL;
 
-kmem_cache_t *ocfs2_lock_cache = NULL;
-
 /* OCFS2 needs to schedule several differnt types of work which
  * require cluster locking, disk I/O, recovery waits, etc. Since these
  * types of work tend to be heavy we avoid using the kernel events
@@ -141,6 +139,7 @@ enum {
 	Opt_hb_local,
 	Opt_data_ordered,
 	Opt_data_writeback,
+	Opt_atime_quantum,
 	Opt_err,
 };
 
@@ -154,6 +153,7 @@ static match_table_t tokens = {
 	{Opt_hb_local, OCFS2_HB_LOCAL},
 	{Opt_data_ordered, "data=ordered"},
 	{Opt_data_writeback, "data=writeback"},
+	{Opt_atime_quantum, "atime_quantum=%u"},
 	{Opt_err, NULL}
 };
 
@@ -707,6 +707,7 @@ static int ocfs2_parse_options(struct super_block *sb,
 	while ((p = strsep(&options, ",")) != NULL) {
 		int token, option;
 		substring_t args[MAX_OPT_ARGS];
+		struct ocfs2_super * osb = OCFS2_SB(sb);
 
 		if (!*p)
 			continue;
@@ -747,6 +748,16 @@ static int ocfs2_parse_options(struct super_block *sb,
 		case Opt_data_writeback:
 			*mount_opt |= OCFS2_MOUNT_DATA_WRITEBACK;
 			break;
+		case Opt_atime_quantum:
+			if (match_int(&args[0], &option)) {
+				status = 0;
+				goto bail;
+			}
+			if (option >= 0)
+				osb->s_atime_quantum = option;
+			else
+				osb->s_atime_quantum = OCFS2_DEFAULT_ATIME_QUANTUM;
+			break;
 		default:
 			mlog(ML_ERROR,
 			     "Unrecognized mount option \"%s\" "
@@ -867,7 +878,7 @@ static int ocfs2_statfs(struct dentry *dentry, struct kstatfs *buf)
 		goto bail;
 	}
 
-	status = ocfs2_meta_lock(inode, NULL, &bh, 0);
+	status = ocfs2_meta_lock(inode, &bh, 0);
 	if (status < 0) {
 		mlog_errno(status);
 		goto bail;
@@ -914,9 +925,7 @@ static void ocfs2_inode_init_once(void *data,
 		oi->ip_open_count = 0;
 		spin_lock_init(&oi->ip_lock);
 		ocfs2_extent_map_init(&oi->vfs_inode);
-		INIT_LIST_HEAD(&oi->ip_handle_list);
 		INIT_LIST_HEAD(&oi->ip_io_markers);
-		oi->ip_handle = NULL;
 		oi->ip_created_trans = 0;
 		oi->ip_last_trans = 0;
 		oi->ip_dir_start_lookup = 0;
@@ -948,14 +957,6 @@ static int ocfs2_initialize_mem_caches(void)
 	if (!ocfs2_inode_cachep)
 		return -ENOMEM;
 
-	ocfs2_lock_cache = kmem_cache_create("ocfs2_lock",
-					     sizeof(struct ocfs2_journal_lock),
-					     0,
-					     SLAB_HWCACHE_ALIGN,
-					     NULL, NULL);
-	if (!ocfs2_lock_cache)
-		return -ENOMEM;
-
 	return 0;
 }
 
@@ -963,11 +964,8 @@ static void ocfs2_free_mem_caches(void)
 {
 	if (ocfs2_inode_cachep)
 		kmem_cache_destroy(ocfs2_inode_cachep);
-	if (ocfs2_lock_cache)
-		kmem_cache_destroy(ocfs2_lock_cache);
 
 	ocfs2_inode_cachep = NULL;
-	ocfs2_lock_cache = NULL;
 }
 
 static int ocfs2_get_sector(struct super_block *sb,
@@ -1280,6 +1278,8 @@ static int ocfs2_initialize_super(struct super_block *sb,
 	init_waitqueue_head(&osb->checkpoint_event);
 	atomic_set(&osb->needs_checkpoint, 0);
 
+	osb->s_atime_quantum = OCFS2_DEFAULT_ATIME_QUANTUM;
+
 	osb->node_num = O2NM_INVALID_NODE_NUM;
 	osb->slot_num = OCFS2_INVALID_SLOT;
 

fs/ocfs2/symlink.c

@@ -126,6 +126,10 @@ static int ocfs2_readlink(struct dentry *dentry,
 		goto out;
 	}
 
+	/*
+	 * Without vfsmount we can't update atime now,
+	 * but we will update atime here ultimately.
+	 */
 	ret = vfs_readlink(dentry, buffer, buflen, link);
 
 	brelse(bh);

mm/filemap.c

@@ -1893,6 +1893,7 @@ int should_remove_suid(struct dentry *dentry)
 
 	return 0;
 }
+EXPORT_SYMBOL(should_remove_suid);
 
 int __remove_suid(struct dentry *dentry, int kill)
 {