Re: [PATCH v6 04/10] arm/tegra: prepare early init for multiple tegra variants

[Peter De Schrijver <pdeschrijver@nvidia.com>]

On Fri, Dec 09, 2011 at 07:35:09PM +0100, Colin Cross wrote:
> On Fri, Dec 9, 2011 at 3:19 AM, Peter De Schrijver
> <pdeschrijver@nvidia.com> wrote:
> > On Thu, Dec 08, 2011 at 07:29:43PM +0100, Colin Cross wrote:
> >> On Thu, Dec 8, 2011 at 9:57 AM, Stephen Warren <swarren@nvidia.com> wrote:
> >> > Peter De Schrijver wrote at Thursday, December 08, 2011 5:44 AM:
> >> >> This patch splits the early init code in a common and a tegra20 specific part.
> >> >> L2 cache initialization is generalized and discovers the cache associativity
> >> >> at runtime. Also use arm_pm_restart instead of arm_arch_reset and reset the
> >> >> the system using the PMC reset feature rather then the CAR system reset.
> >> >
> >> > This one really should be at least 3 separate patches:
> >> > * s/tegra_init_early/tegra20_init_early/ in machine descriptions
> >> > * Switch reset mechanism from CAR to PMC
> >> > * tegra_init_cache modifications
> >> > * Not sure if all the ifdef additions go in the above, or if some should be
> >> >  separate?
> >> >
> >> > Still, I think it's probably OK to go in as it is even if it isn't optimal.
> >>
> >> Please at least split out the reset change.  Does using the PMC reset
> >> instead of the CAR reset still result in a warm reset, or does it
> >> change to a cold reset?
> >
> > What do you consider a cold reset in the tegra world?
> 
> Power rails turn off and on.  Some of our debugging tools rely on
> memory contents surviving across a reset, and that is generally true
> of a warm reset, but not a cold reset.

Neither of those methods result in a warm reset. In both cases the memory
controller is reset so the SDRAM won't be refreshed until the memory
controller is reintialized. So there is no guarantee the memory contents
will survive the reset.

Cheers,

Peter.