In some email I received from Jason Thorpe, sie wrote:
> 
> WARNING: There are errors in these changes that can cause a kernel 
> crash.  In particular, union_allocvp() obtains the lwp to pass around 
> from the cnp argument unconditionally.  However, the cnp argument is 
> allowed to be NULL.  It is necessary to add an lwp * argument to 
> union_allocvp() and use that instead.
> 
> I would suggest that all of these changes need to be audited very 
> carefully, as it appears as if simple mechanical changes were made to 
> get the code to compile.

Hmmm...I replaced, for the most part, use of "proc" with "lwp".

Is there something different with the way I've passed an lwp pointer
down to how a proc pointer was passed ?

The only function I see that is at risk is union_mkwhiteout.  Am I
missing something else ?

Darren