Re: earmhf issues on Beaglebone Black

[Manuel Bouyer <bouyer%antioche.eu.org@localhost>]

On Fri, Jul 25, 2014 at 04:40:26PM +0100, Nick Hudson wrote:
> On 07/25/14 16:35, Manuel Bouyer wrote:
> >On Fri, Jul 25, 2014 at 03:57:22PM +0100, Nick Hudson wrote:
> >>On 07/25/14 15:43, Manuel Bouyer wrote:
> >>
> >>>The RPI CPU has a VIVT L1 data cache, which could explain the
> >>>behavior difference (pmap_zero_page_generic() will behaves differently).
> >>>But I'm still surprised that it didn't cause more problems.
> >>RPI L1 is VIPT.
> >Right (that's what I meant I guess), and the A8 is PIPT, so
> >and pmap_zero_page_generic() checks for PIPT. So it's still
> >valid to explain why the RPI seems more stable.
> >But theorically, the RPI could still see this problem in
> >some case too, I guess.
> >
> Glad you found the problem before RPI did :)
> 
> I think the main reason that the RPI didn't see the problem is that it isn't
> ARM_MMU_EXTENDED yet, but I hope to fix that soon.

ha yes (it's hard to follow where all theses #define comes from :)
in this case, pmap_zero_page_generic() would always write back the zeroed
page, so no problem.

-- 
Manuel Bouyer <bouyer%antioche.eu.org@localhost>
     NetBSD: 26 ans d'experience feront toujours la difference
--