[Unionfs] notify_change in inode.c:unionfs_create()

[Charles P. Wright]

On Mon, 2006-03-13 at 09:41 +0900, hooanon05 at yahoo.co.jp wrote:
> Josef Sipek:
> > Hmm..I shouldn't need any fsuid/fsgid change since we are passing
> > ATTR_FORCE which bypasses all the permission checks. So, I'm tempted to
> > blame nfs.
> 
> Actually, linux nfs client doesn't change fsuid before sending the
> request to nfs server. It might be a bug as you wrote, but I am not
> sure.
> Here is a patch for unionfs.
> 
> Junjiro Okajima
[snip]
> +	//BUG_ON(!mutex_is_locked(&inode->i_mutex)); // should be locked before this function
> +	if (IS_IMMUTABLE(inode) || IS_APPEND(inode)) // is i_flags maintenanced in unionfs?
> +		return -EPERM;
The i_flag won't be propagated to the Unionfs inode, but Unionfs won't
blow i_flag away on the lower-level inode (which is what you are
accessing here).  So this is safe.

Charles