It occurred to me that a simple way to improve objrmap performance would be
to sort the vma chains off address_space by offset.  Here's a patch that
does it.  Tests show no measureable cost, and it could significantly reduce
the impact of the worst case scenario (100 mappings * 100 processes) we've
all worried about.

Dave McCracken

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