Alexander Viro wrote:
 > Denied.  You can trivially do that in ext2_read_inode() and
 > ext2_new_inode().

Is this the right way to do it?  I put the lock back in the 
ext2_inode_info structure, and moved the initialization from init_once() 
to ext2_read_inode() and ext2_new_inode().

-- 
Dave Hansen
haveblue@us.ibm.com