Paul McKenney and I sat down today and hashed out just what the races are
for both  vmtruncate and the distributed filesystems.  We took Andrea's
idea of using seqlocks and came up with a simple solution that definitely
fixes the race in vmtruncate, as well as most likely the invalidate race in
distributed filesystems.  Paul is going to discuss it with the DFS folks to
verify that it's a complete fix for them, but neither of us can see a hole.

Anyway, here's the patch.

Dave McCracken

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