Subject: Re: CVS commit: syssrc
To: Izumi Tsutsui <tsutsui@ceres.dti.ne.jp>
From: Matt Thomas <matt@3am-software.com>
List: source-changes
Date: 06/16/2001 07:39:42
At 03:42 PM 6/16/2001 +0900, Izumi Tsutsui wrote:
>In article <4.3.2.7.2.20010615231513.088b3e78@3am-software.com>
>matt@3am-software.com writes:
>
> > With arm, we trying a new technique.  The Makfile.arm will automagically
> > allow <machine/asm.h> to be picked up from <arm/asm.h> without having to
> > have a #include <arm/asm.h> in <port>/include.  And for user space,
> > usr/include/machine will be a symlink to arm, not the port.
>
>Hmm, if usr/include/machine can be shared, why the ports are split?
>At least, machine/param.h defines MACHINE, which cannot be shared
>among different ports.

User programs should *NEVER* look at MACHINE in param.h.  They should get
the value from uname(2).

>My current problem is that config(8) creates a symlink of
>"machine -> ../../../../arch/cats/include" in compile dirs.

Yes it does.  But Makefile.arm creates an include subdir
which contains a machine -> ../../../../../arch/arm/include
and a arm/cats -> ../../../../../../arch/cats/include.

I've built both NETWINDER & CATS GENERIC kernels using the current
source pool with no problem.
--
Matt Thomas               Internet:   matt@3am-software.com
3am Software Foundry      WWW URL:    http://www.3am-software.com/bio/matt/
Cupertino, CA             Disclaimer: I avow all knowledge of this message