--On Thursday, April 03, 2003 14:24:41 -0800 Andrew Morton <akpm@digeo.com>
wrote:

> It's more conventional to lock the page in the caller.  And we forgot the
> whole reason for locking it: to keep truncate away.  We need to check that
> the page is still on the address_space after the page lock has been
> acquired.
> 
> This applies on top of your first.

Ok, that all makes sense.  Here's a patch with all your changes applied.

Dave

======================================================================
Dave McCracken          IBM Linux Base Kernel Team      1-512-838-3059
dmccr@us.ibm.com                                        T/L   678-3059