Containers: Avoid lockdep warning

I think this is the right way to handle the lockdep false-positive in the current containers patches, but I'm not that familiar with lockdep so any suggestions for a better approach are welcomed.

In order to avoid a false-positive lockdep warning, we lock the root inode of a new filesystem mount prior to taking container_mutex, to preserve the invariant that container_mutex nests inside inode->i_mutex. In order to prevent a lockdep false positive when locking i_mutex on a newly-created container directory inode we use mutex_lock_nested(), with a nesting level of I_MUTEX_CHILD since the new inode will ultimately be a child directory of the parent whose i_mutex is nested outside of container_mutex.

I think this is the right way to handle the lockdep false-positive in the current containers patches, but I'm not that familiar with lockdep so any suggestions for a better approach are welcomed.

In order to avoid a false-positive lockdep warning, we lock the root inode of a new filesystem mount prior to taking container_mutex, to preserve the invariant that container_mutex nests inside inode->i_mutex. In order to prevent a lockdep false positive when locking i_mutex on a newly-created container directory inode we use mutex_lock_nested(), with a nesting level of I_MUTEX_CHILD since the new inode will ultimately be a child directory of the parent whose i_mutex is nested outside of container_mutex.

Signed-off-by Paul Menage <[email protected]>

---
 kernel/container.c |   17 +++++++----------
 1 file changed, 7 insertions(+), 10 deletions(-)

Index: container-2.6.23-rc3-mm1/kernel/container.c
===================================================================

kernel/container.c +6 -9

@@ -966,13 +966,16 @@ static int container_get_sb(struct file_
 	} else {
 		/* New superblock */
 		struct container *cont = &root->top_container;
+		struct inode *inode;
 
 		BUG_ON(sb->s_root != NULL);
 
 		ret = container_get_rootdir(sb);
 		if (ret)
 			goto drop_new_super;
+		inode = sb->s_root->d_inode;
 
+		mutex_lock(&inode->i_mutex);
 		mutex_lock(&container_mutex);
 
 		/*

@@ -985,12 +988,14 @@ static int container_get_sb(struct file_
 		ret = allocate_cg_links(css_group_count, &tmp_cg_links);
 		if (ret) {
 			mutex_unlock(&container_mutex);
+			mutex_unlock(&inode->i_mutex);
 			goto drop_new_super;
 		}
 
 		ret = rebind_subsystems(root, root->subsys_bits);
 		if (ret == -EBUSY) {
 			mutex_unlock(&container_mutex);
+			mutex_unlock(&inode->i_mutex);
 			goto drop_new_super;
 		}
 

@@ -1030,16 +1035,8 @@ static int container_get_sb(struct file_
 		BUG_ON(!list_empty(&cont->children));
 		BUG_ON(root->number_of_containers != 1);
 
-		/*
-		 * I believe that it's safe to nest i_mutex inside
-		 * container_mutex in this case, since no-one else can
-		 * be accessing this directory yet. But we still need
-		 * to teach lockdep that this is the case - currently
-		 * a containerfs remount triggers a lockdep warning
-		 */
-		mutex_lock(&cont->dentry->d_inode->i_mutex);
 		container_populate_dir(cont);
-		mutex_unlock(&cont->dentry->d_inode->i_mutex);
+		mutex_unlock(&inode->i_mutex);
 		mutex_unlock(&container_mutex);
 	}