[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [N8VEM-S100:245] Bootstrapping Lunux on the S100/68K...

Hash: SHA1


That is true, the MC68000 cannot do demand paging because of the reason
you state. However the MC68010 which is pin compatible can.

Yes, normally Linux depends on demand paging being available, however,
several years ago a fork of linux branched off called uClinux and is
popular on embedded 68K architecture (I used it to develop some
Lantronix X-Port applications on Colfire uC for example).

Many of the 68K controllers don't have an MMU so there was a need to
develop an MMU-less mode of operation. Since then the -nommu option is
being branched into the ARM tree, and is coming full circle and is now
being integrated into the mainstream kernel source. It does have some
limitations, such as no swap file, and no memory protection other than a
thin barrier of a few protected instructions between supervisor mode and
user mode, but user mode can 'see' kernel memory, so an unstable user
app can bring down the kernel.

I will be building my initial FLASH based kernel from the 2.4 linux
source stream from a uClinux SDK simply because that's what I have
already up and running, and can build a XIP (eXecute In Place) kernel
from it.

In any case, the initial build will happily run from FLASH on the 68000.
The plan from then on is to get the MMU
http://8bit.zapto.org/index.php/mc68010-mmu card up and running and plug
in a MC68010 in place of the MC68000. Of coarse I will have to write
some kernel code to support this MMU since Linux has never seen one
before. I've been studying the sun3 and the Mac/68k MMU implementations
and I have a pan of attack taking shape on that front.

Once I have the MMU up and running, then I will set my sights on booting
from a fixed disk, do developing a fixed disk device driver for the S100
bus, etc...

So anyway, yes, you are entirely correct with what you said.

Kind Regards,

Mike Sharkey

On 11-06-03 01:24 AM, Bill Lewis wrote:
> I was under the impression a 68000 can't run true Unix because of problems
> with restarting instructions after page faults.  And that was fixed in the
> 68010.  Not true?
> Not sure if Linux's architecture had this issue or not.
> Bill

Version: GnuPG v1.4.10 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/