Re: GNUMach 1.x fixed (autoconf+kb)

[Roland McGrath]

> Also, I think the _LOCAL functions may have to be preserved to permit
> bootstrapping.  GCC faced the same problem, so they may have gotten the
> fixes into upstream autoconf.

I don't think there should be any kind of bootstrapping constraint for the
microkernel.  It really doesn't matter what kind of host libraries you use
to build the microkernel.  That's why taking out the $host_os check is ok.
Just doing "configure; make" on GNU/Linux builds a working kernel.