--xHFwDpU9dbj6ez1V
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline

Dear Gavan,

On Tue, Jul 03, 2007 at 05:23:55PM +0100, Gavan Fantom wrote:
> >i'm confused about the correct procedure concerning MMU disabling on ARMv3 
> >and ARMv4. The data sheets i have for ARM610/ARM710 and SA don't mention 
> >what effect it will have on the instruction fetching. AFAIK its two 
> >instructions that are in the pipeline before the translation becomes 
> >effective.
> >
> >The offending code is in src/sys/arch/acorn32/stand/boot32/start.S around 
> >line 83. Will the following patch fix the problem? or is something really 
> >wrong and should i split off ARMv3 and ARMv4 mmu disable code?
> 
> What is the problem you are trying to fix?

erradic booting problems on various machines; some work and bootup NetBSD 
fine and others never make it into kernel code it appears. I think it might 
be due to interaction between MMU, D&I caches etc. that it sometimes works 
and otherwise not and might be as simple as does the code start in a cache 
line or not etc.

So i'd rather fix it for once and for all :)

Cheers,
Reinoud

--xHFwDpU9dbj6ez1V
Content-Type: application/pgp-signature
Content-Disposition: inline

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (NetBSD)

iQEVAwUBRop5hoKcNwBDyKpoAQIdfwf6A4wV+VBqSMZQ53mAZE+kL6Ix1jLkCayD
yMLgCfJ047+PeKAesHWLZ+vl571ne9TJ0hYi2ijpAA3DNz9R3OlI4fLuHShw3kfL
D3q82GfOFYogXHfg4E2VcWVkBdzX4z5m8YGFk/tyU+bcp9hADMb9i04wtx3z7gm+
zosI4ucZbFwe5006pE9RQjN1Ya04TLD+xAm3YvPXD5RxWoHdGWA0mpd09jdnB1TB
s4xF8gZ3cE6z57cTBK+DxZd5DjEcQPFGG1FdJLLt2LsrpYVtfuPCf75FeQTf6wG2
10TsGpG6nqAb83wQrB001GvpNpXAFEql8Yz4Va6mAxHV2I8tDkQJ4g==
=D6Ek
-----END PGP SIGNATURE-----

--xHFwDpU9dbj6ez1V--