[M3devel] CM3 resource access at Elego, was: Re: Web page experimental colors

[hendrik at topoi.pooq.com]

On Fri, Jul 31, 2009 at 12:36:26PM +0000, Jay K wrote:
> 
> We should be a little careful with ARM_LINUX imho. And MIPS_LINUX.
>  
>  
> Those target name might be ok, and they'd refer to Linux 2.6+ with glibc.
> Many ARM and MIPS Linux systems aren't that.
>  
>  
> ARM has old ABI and new ABI, at the kernel level.

Nokia uses the new ABI, I believe.  Maemo, its OS, is debian-derived, 
but it is *not* Debian.  Though it is possible to set it up to use 
Debian user-space in a chroot on the maemo kernel.

> It it also common to replace glibc with uclibc.
> I don't know if they are binary compatible or not.
> My Linux/arm machine/device is old ABI and uclibc.

So it's probably not.

> It seems that if you are building your own kernel,
> it is trivial to use old ABI. It isn't like it is incompatible
> with new hardware, I think. I think whoever put together
> the Western Digital MyBook World Edition just took the default.
>  
>  
> MIPS..well, I was surprised. I installed "Tomato" on my router.
> It is /very/ low end, but it does have a builtin SMB client, therefore
> it has infinite diskspace.
> It uses Linux 2.4, and I think/assume uclibc.
>  
>  
> MIPS also has big and little endian and I don't know if either is
> more common or if it is an even split.

I know the hardware has that option -- it's controlled by a bit in some 
processor register.  I remember wonderein, years ago, whether the OS 
allowed one to set that on a per-process basis, or even more finely.

The situation reminds me of the IBM 360, which had a similar bit to 
control whether its native instructino set would support decimal 
operations in ASCII or in EBCDIC.  But because changing that bit was a 
privileged operation, no one really got to set it, and everything was 
always EBCDIC.

With the 370, I believe IBM discontinued the ASCII option.  No one 
had ever used it.

-- hendrik