Martin Pieuchot <mpi@grenadille.net> wrote:

> > I do not understand any of the diff.  It seems to be trying to return a
> > child to the parent, before resources are cleared.
> 
> It doesn't.  Order of operations didn't change.  The KERNEL_LOCK() is
> there and asserted.

I did not mention the KERNEL_LOCK().

> I disagree.  I need to remove the reaper.  That *is* the problem.

You are using proc0 in an very dangerous way.

> > I've called this uvmspace_purge().  It would be a bit like
> > 
> >      uvm_unmap_remove(p->p_vmspace, 0, VM_MAXUSER_ADDRESS,
> >          &dead_entries, FALSE, TRUE, TRUE)
> > 
> > Except there are clearly optimizations.  The function can sleep.  It is
> > called quite early in process context.
> > [...] 
> 
> There's no need to add a new function.  Calling uvm_exit() from exit1()
> is enough and work but that's not buying us much.

As proc0?  Disagree strongly on this direction.  The process should free
it's own pages out of it's own map.  It can do that if uvm_exit() is split.

But you've made up your mind to ignore everything everyone says.  You don't
even reply to emails about the proposals.  Your public requests for feedback
come off as fake.