Peter Staubach wrote:

> Hi.
>
> Attached are same changes to make the NFS client side WRITE processing 
> a little
> more defensive.  Currently, a response which indicates that the server 
> wrote
> more data than actually requested could cause bad things to happen.  
> Such a
> response could come from a broken or malicious server or from network 
> corruption
> that the normal checksumming fails to catch. 


Chuck Lever correctly pointed out that this patch did not address all of the
WRITE cases.  I hadn't incuded the direct I/O case.  The new, attached patch
should address it in addition to the sync and async cases already covered.

    Thanx...

       ps