lustre_cfg_new() currently returns ERR_PTR(-ENOMEM) at allocation failure, but all callers except one checks its return against NULL. As there's no other error to return, I think lustre_cfg_new() should simply return NULL rather than error coded pointers.

Also, a minor issue in ptlrpc_check_status(), the console message suppression branch will never be entered and thus is rendered useless.

A patch will follow shortly to fix them.