> > This turns out to be a problem with crunchgen needing to grok
> > MAKEOBJDIRPREFIX (just as it already groks obj.${MACHINE} and obj).
> > The patch below, gets us past that issue, and a make in miniroot then
> > gets as far as:
> 
> Can something smart be done with the equivalent of "make print-objdir"
> or similar?  crunchgen shouldn't have to know too much about the build
> environment itself, I would think...

The problem is that crunchgen _already_ knows too much about our make.
It knows to look for obj.${MACHINE} and obj, thus it also needs to
look for $MAKEOBJDIRPREFIX as well or it is not in sync with what make
will react to.  If it did not know about obj etc, then the Makefiles
would need to tell it where the object directory was and could do the
same in the case of MAKEOBJDIRPREFIX.  Just as the simple change I
made to etc/Makefile does.  But in this case crunchgen needs the fix.

> It depends on how a particular port installs its bootblocks.  I've never
> had "options INSECURE" on a pmax and never had a problem building a
> release on that platform.  File a bug under port-sparc :-)

Nah, I just rebuilt my kernel with INSECURE :-)

--sjg