Subject: Re: MACHINE_ARCH on mips
To: Jason Thorpe <>
From: Jonathan Stone <jonathan@DSG.Stanford.EDU>
List: tech-toolchain
Date: 07/26/1998 14:08:14
On  Sun, 26 Jul 1998 13:36:43 -0700,  : Jason Thorpe
<> writes:

>It's just plain STUPID to say that we'll try and do every MIPS ABI
>convention in NetBSD.  We should stick with ILP32 "mipsel" and "mipseb"
>and maybe the LP64 "mipsel64" and "mipseb64" on those systems where it
>makes sense.

People keep asking for it, though.  It's even in an ARC wishlist which
has your name on it.

>Adding "find-grained" checking via strings in the kernel is just silly,
>is incorrect (see my previous example), and overly complex.

But "mipsel" vs "mipseb" is "fine-grained", and with 64-bit, we have four.  

I think having to check for each of four mips ${MACHINE_ARCH}es in
makefiles is silly.  I think having to kludge up special machinery to
deal with ${MACHINE_ARCH} and arch/${MACHINE_ARCH} subdirs is silly. 
I think having to kludge up ${MACHINE_ARCH} when building set lists
based on ``is this CPU a mips or not'' is silly.

There really are two different levels of description.  Being
condescending to people who disagree doesn't make that go away.