hush on NONMMU for active shell

[Rob Landley]

On Friday 08 June 2007 4:16 am, Denis Vlasenko wrote:
> Hi Rob,

Hi Denis!

> On Wednesday 06 June 2007 23:10, Bernhard Fischer wrote:
> > On Wed, Jun 06, 2007 at 04:56:33PM -0400, Mike Frysinger wrote:
> > >On Tuesday 05 June 2007, Denis Vlasenko wrote:
> > >> I want to have NOMMU test environment. I really do.
> > >
> > >would a free Blackfin board be OK ?  you can find prebuilt toolchains and 
> > >bootable images with every release so you wouldnt need to build your own, 
you 
> > >could just compile busybox and rcp it to the board and test it
> >
> > Either real or emulated hardware where it is possible to test
> > combinations of noMMU and soft-float (the latter should not really be an
> > issue since it should be doable to trap on FP instructions in qemu, but
> > of course i didn't look at this yet) would be nice to have indeed.
> > 
> > Not being able to test this stuff is awkward in many ways.
> 
> Rob, judging by http://landley.net/notes.html you do a lot of 
cross-compiling
> to different CPU arches and are playing with qemu.

Yup.  http://landley.net/hg/firmware

(Note that you have to revert the Linux kernel to 2.6.20 to get it to work on 
arm under qemu because the scsi driver changed and device probing gets 
confused by QEMU's virtual scsi card.  I haven't had time to properly track 
this down, just poke at it every couple of weeks to see if it's still screwed 
up.)

> Can you give me an advice what arch(es) is(are) suitable for
> building and testing busybox for NOMMU arch.

Well, I'm partial to arm myself, but I'm influenced by once having got 
armulator working.  In theory you can switch the MMU off in the Linux kernel 
for several architectures and then if you boot the result it doesn't matter 
if the hardware has a MMU or not, the kernel won't use it if it does.

> At least which NOMMU arches
> 1. Are supported by qemu 0.9.0
> 2. Have mostly working support in binutils and gcc

QEMU doesn't specifically have NOMMU support for anything, with the possible 
exception of coldfire.  It's about the same way they don't explicitly have a 
way of disabling most floating point coprocessors.  They just tell 
you "switch it off in the kernel".

> Availability of a known working example of qemu image for an arch is a plus,
> but isn't critical. If there is no such thing on the net,
> I will try to build a kernel for it and make my own.

I can take a stab at putting one together later today.  It's basically a 
question of altering the kernel and uClibc configs.  Nothing else is explicit 
about it one way or the other...

Rob
-- 
The Google cluster became self-aware at 2:14am EDT August 29, 2007...