[rtems commit] score: Move _SMP_Handler_initialize()

[Chris Johns]

On 23/02/2014 1:51 pm, Gedare Bloom wrote:
> On Sat, Feb 22, 2014 at 5:02 PM, Chris Johns <chrisj at rtems.org> wrote:
>>
>> I do not agree with this approach. All cores need to be present before
>> continuing and it forces cpu 0 to be a "boot" or "primary" processor. This
>> design is not symmetric.
>>
>> Chris
> Chris, this patch is just moving around the code that exists. Feel
> free to file a PR if you think there is a bug in the existing code.
>

I know the origin of the code and I am not rejecting the change. It is 
part of the various bits that make up initialising SMP and I just wish 
to draw attention to what we have. I do not think a PR will help here.

My concern is this approach of synced initialisation of all cores being 
so deeply embedded in the code it will be impossible to remove and I do 
not remember any discussion about this being the design for RTEMS.

Chris