See also gnumach vm map entry forward merging.
IRC, freenode, #hurd, 2012-11-16
<mcsim> hi. do I understand correct that following is true: vm_object_t a;
a->shadow->copy == a;?
<braunr> mcsim: not completely sure, but i'd say no
<braunr> but mach terminology isn't always firm, so possible
<braunr> mcsim: apparently you're right, although be careful that it may
not be the case *all* the time
<braunr> there may be inconsistent states