On Fri, Aug 19, 2011 at 01:09:36PM -0400, Trond Myklebust wrote:

quoted

 void locks_remove_posix(struct file *filp, fl_owner_t owner)
 {
 	struct file_lock lock;
+	struct file_lock **before;
+	struct inode *inode;
 
 	/*
 	 * If there are no locks held on this file, we don't need to call
 	 * posix_lock_file().  Another process could be setting a lock on this
 	 * file at the same time, but we wouldn't remove that lock anyway.
 	 */
-	if (!filp->f_path.dentry->d_inode->i_flock)
+	inode = filp->f_path.dentry->d_inode;
+
+	if (!inode->i_flock)
 		return;
 
+	lock_flocks();
+	for_each_lock(inode, before) {
+		struct file_lock *fl = *before;
+		if (IS_POSIX(fl))
+			goto do_unlock;
+	}
+	unlock_flocks();
+	return;
+
+do_unlock:
+	unlock_flocks();
 	lock.fl_type = F_UNLCK;
 	lock.fl_flags = FL_POSIX | FL_CLOSE;
 	lock.fl_start = 0;

This assumes that all locks are mirrored in inode->i_flock, which may
not be the case if the filesystem implements its own f_op->lock(). The
right place for this optimisation would be in your filesystem callback.

I suppose if i_flock is being faked at the same time to avoid the null
check.  I once looked at not doing lock mirroring in vfs and couldn't for
reasons I don't remember... may the null i_flock check was the only
reason, not sure.

Dave