Re: x96 cannot build without MULTIPROCESSOR

[Joerg Sonnenberger <joerg%britannica.bec.de@localhost>]

On Wed, Jul 01, 2015 at 06:57:26PM +0200, Emmanuel Dreyfus wrote:
> Martin Husemann <martin%duskware.de@localhost> wrote:
> 
> > > Then it should not be an option.
> > It is a real option on some other architecturs, so this gets tricky.
> 
> At least it should be documented as such: options(4) can note it is
> implicit x86 and whatever other architectures.
> 
> Then MULTIPROCESSOR can be set by a #define on x86, and config(1) can be
> modified to reject the option as deprecated on x86.

Why do you believe this is a problem that needs to solved? std.XXX
generally has options that are required for one reason or another.
Should mainbus0 at root be hard-coded in the kernel as well because it
is not optional on x86? It's an option because that's the way config(8)
works and switch itself is mostly MI.

Joerg