[S] Change in coreboot[master]: [RFC]soc/intel/cannonlake: Use regular NEM for CAR

[Arthur Heymans (Code Review)]

Attention is currently required from: Nico Huber, Tarun Tuli, Subrata Banik, Rizwan Qureshi, Lean Sheng Tan.

Arthur Heymans has posted comments on this change. ( https://review.coreboot.org/c/coreboot/+/74376 )

Change subject: [RFC]soc/intel/cannonlake: Use regular NEM for CAR
......................................................................

Patch Set 1:

(3 comments)

Patchset:

PS1:
> we shoudn't blindly check NEM over eNEM

Do you know a way at runtime to figure this out? I'm thinking SDM is either wrong or silicon has problem with cpuid being wrong.

PS1:
> seems like MSR IA32_PQR_ASSOC still exists? or could be wrong?

So not sure which MSR used in NEM_ENHANCED is actually resulting in a hang, but the bits in cpuid are clearly not working as indicator for their presence as they do work on some models where the cpuid bits are 0.

PS1:
> Maybe there's something else we could use to filter the working CPUs, e.g.
> stepping ID?

Saddly not. same stepping, but different CPU branded model have different results.