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

@@ -93,7 +93,7 @@ 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,
+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 @@ int ocfs2_num_free_extents(struct ocfs2_super *osb,
  * 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 @@ static int ocfs2_create_new_meta_bhs(struct ocfs2_super *osb,
  * 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 @@ static int ocfs2_add_branch(struct ocfs2_super *osb,
  * 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 @@ static int ocfs2_shift_tree_depth(struct ocfs2_super *osb,
  * 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 @@ static int ocfs2_find_branch_target(struct ocfs2_super *osb,
 
 /* 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 @@ static int ocfs2_truncate_log_append(struct ocfs2_super *osb,
 }
 
 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);
-
-	if (data_alloc_inode)
-		iput(data_alloc_inode);
+	ocfs2_commit_trans(osb, handle);
 
-	if (data_alloc_bh)
+out_unlock:
 	brelse(data_alloc_bh);
+	ocfs2_meta_unlock(data_alloc_inode, 1);
+
+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 @@ int ocfs2_commit_truncate(struct ocfs2_super *osb,
 
 	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 @@ int ocfs2_commit_truncate(struct ocfs2_super *osb,
 	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 @@ int ocfs2_commit_truncate(struct ocfs2_super *osb,
 		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 @@ int ocfs2_readdir(struct file * filp, void * dirent, filldir_t filldir)
 
 	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"
@@ -1063,7 +1064,7 @@ static void ocfs2_cluster_unlock(struct ocfs2_super *osb,
 	mlog_exit_void();
 }
 
-int ocfs2_create_new_lock(struct ocfs2_super *osb,
+static int ocfs2_create_new_lock(struct ocfs2_super *osb,
 				 struct ocfs2_lock_res *lockres,
 				 int ex,
 				 int local)
@@ -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 @@ int ocfs2_meta_lock_full(struct inode *inode,
  * 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.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 @@ int ocfs2_inode_revalidate(struct dentry *dentry)
  * 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.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 @@ int ocfs2_load_local_alloc(struct ocfs2_super *osb)
 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)
+	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);
 
+out:
 	mlog_exit(status);
 	return status;
 }
@@ -452,7 +447,6 @@ int ocfs2_complete_local_alloc_recovery(struct ocfs2_super *osb,
  * 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 @@ static int ocfs2_sync_local_to_main(struct ocfs2_super *osb,
 }
 
 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 @@ static int ocfs2_local_alloc_reserve_for_window(struct ocfs2_super *osb,
  * 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.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.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)
 {