Change in coreboot[master]: Documentation/Intel: Add MultiProcessorInit documentation

[Nico Huber (Code Review)]

Nico Huber has posted comments on this change. ( https://review.coreboot.org/25921 )

Change subject: Documentation/Intel: Add MultiProcessorInit documentation
......................................................................

Patch Set 3:
> > (4 comments)
>
> I agree with you on most this comments that you have put, but my
> only point is that, if i tries to put this implementation of
> creating EDK2 interface inside Coreboot to do certain thinks might
> not justify the purpose of adding those "foreign" interface as your
> words. Today coreboot code is so independent in nature so that we
> don;t need to talk help of such EDK2 interface to do certain
> things, if we really want to increase mp_init.c capability we could
> have done the same using native c why UEFI?
>
> UEFI is because this implementation only targeted for Intel fsp 9th
> gen + designs where we might want to do few things using coreboot
> callback from fsp, hence i could see a direct dependencies between
> this implementation proposal and supporting FSP model to justify
> that callback using SkipMpInit=0. if i make this code into common
> x86 place as well, no one will use this except 9th gen + socs from
> intel with proper FSP support.
>
> whats the point of keeping certain code into common when there are
> no user of that code expect intel native FSP support?

One point would be visibility. If anybody not familiar with the
soc/intel/ code would look at our MP interfaces, he might miss
the fact that a generic implementation of your EFI interface
already exists, might even reimplement it one way or another and
coreboot would decompose a little more.

That's just a general thought. I agree that this interface might
be too FSP specific (from a coreboot point of view).
> Hope i could able to explain my concern, i have written this
> document so that i can keep it anywhere but myself not very
> convinced about why other users will interested to make use of EFI
> code interface inside x86/coreboot when mp_init.c is capable of
> doing everything without those APIs.

You are right. I wouldn't expect it to be used inside coreboot,
I was implying other potential blob use cases.