[PATCH v2 00/03] ARM: shmobile: r8a7790 APMU SMP support

[Simon Horman]

On Tue, Sep 10, 2013 at 01:19:21PM +0100, Sudeep KarkadaNagesha wrote:
> On 04/09/13 09:39, Simon Horman wrote:
> > On Thu, Aug 29, 2013 at 01:34:28PM +0900, Simon Horman wrote:
> >> On Thu, Aug 29, 2013 at 01:11:37PM +0900, Magnus Damm wrote:
> >>> On Thu, Aug 29, 2013 at 9:18 AM, Simon Horman <horms at verge.net.au> wrote:
> >>>> On Thu, Aug 29, 2013 at 08:21:48AM +0900, Magnus Damm wrote:
> >>>>> ARM: shmobile: r8a7790 APMU SMP support
> >>>>>
> >>>>> [PATCH v2 01/03] ARM: shmobile: Shared APMU SMP support code without DT
> >>>>> [PATCH v2 02/03] ARM: shmobile: Add r8a7790 SMP support using APMU code
> >>>>> [PATCH v2 03/03] ARM: shmobile: Add r8a7790 CA15 CPU cores
> >>>>>
> >>>>> This series adds SMP support to r8a7790 using the included APMU code.
> >>>>> The previous version of r8a7790 support depended on the APMU code in
> >>>>> "[PATCH] ARM: shmobile: Shared APMU SMP support code", but in this
> >>>>> version the APMU code has been reworked to not use DT. In the future
> >>>>> r8a73a4 and other SoCs may make use of the APMU as well, but the
> >>>>> r8a73a4 code first needs to be adjusted to work with the CCI driver.
> >>>>>
> >>>>> With these patches applied r8a7790 CA15 support for SMP boot
> >>>>> and CPU Hotplug is known to be working.
> >>>>>
> >>>>> Thanks to Sudeep for his feedback.
> >>>>>
> >>>>> Signed-off-by: Magnus Damm <damm at opensource.se>
> >>>>
> >>>> Hi Magnus,
> >>>>
> >>>> can I clarify that these patches are targeted at mainline?
> >>>
> >>> Hi Simon, yes, that's correct. It's possible to back port too if needed.
> >>
> >> Thanks for the clarification.
> >>
> >>> Actual merge is not rushing though, so please don't interpret this as
> >>> any last minute v3.12 merge attempt.
> >>
> >> No problem, I didn't interpret it that way.
> > 
> > I have queued these up for v3.13 and they are present
> > in the freshly baked renesas-devel-20130904 tag.
> > 
> As I already mentioned I asked not to add any custom bindings to support
> multiple cluster. But this version removed APMU node from DT completely
> which is not correct.
> 
> Also you removed multi cluster support to avoid CCI and MCPM reuse. I
> think you may have to rework these completely when adding multi-cluster
> support. I strongly recommend to reuse the CCI and MPCM code from first
> instead of reworking again, but I leave it SH platform maintainer to
> decide.

Thanks for your feedback,

I plan to meet with Magnus in person on Friday and I will
discuss this issue with him in our capacities as shmobile
co-maintainers.