Change in coreboot[master]: soc/intel/alderlake, soc/common: Add method to determine the cpu type...

[Arthur Heymans (Code Review)]

Attention is currently required from: Tim Wawrzynczak, Sridhar Siricilla, Arthur Heymans, Patrick Rudolph.
Arthur Heymans has posted comments on this change. ( https://review.coreboot.org/c/coreboot/+/59360 )

Change subject: soc/intel/alderlake, soc/common: Add method to determine the cpu type mask
......................................................................

Patch Set 3:

(5 comments)

File src/soc/intel/common/block/acpi/Kconfig:

https://review.coreboot.org/c/coreboot/+/59360/comment/1f2b6316_c922be04 
PS3, Line 50: Generate CPPCv3 entries for Intel hybrid processors
That is not what the code does in this CL.

File src/soc/intel/common/block/acpi/cpu_hybrid.c:

https://review.coreboot.org/c/coreboot/+/59360/comment/9aa2d3ba_11d05fce 
PS1, Line 25: return global_cpu_type_bitmask;
> The bitmask is updated in the multi core context, but accessed by only BSP. So, it's a thread-safe.

That's not necesarily threadsafe. You need to make sure the multi core code is done doing this before accessing this variable on the BSP. Where is this checked?

File src/soc/intel/common/block/acpi/cpu_hybrid.c:

https://review.coreboot.org/c/coreboot/+/59360/comment/0bc75f28_820bf204 
PS3, Line 37: set_cpu_type_bitmask
Is it really useful to encode this information in a uint32_t? It won't scale past 32 big cores? Why not add the cpu type to struct device?

https://review.coreboot.org/c/coreboot/+/59360/comment/a15d897e_fbd45c40 
PS3, Line 43: get_cpu_index
global_cpu_type_bitmaks will overflow if there are more than 32 CPUs.

File src/soc/intel/common/block/include/intelblocks/acpi.h:

https://review.coreboot.org/c/coreboot/+/59360/comment/50ed37a6_a0e9e5d4 
PS3, Line 11: 
            : #define CPUID_CORE_TYPE_INTEL_ATOM	0x20
            : #define CPUID_CORE_TYPE_INTEL_CORE	0x40
Maybe use an enum for this and also in the return value for the functions.