Attention is currently required from: Cliff Huang, Jamie Ryu, Patrick Rudolph, Sean Rhodes, Zhixing Ma.
Jérémy Compostella has posted comments on this change by Jérémy Compostella. ( https://review.coreboot.org/c/coreboot/+/86004?usp=email )
Change subject: cpu/x86/topology: Fix FSP-S crash caused by shared core ID ......................................................................
Patch Set 5:
(3 comments)
Commit Message:
https://review.coreboot.org/c/coreboot/+/86004/comment/31ebbf58_7048cd96?usp... : PS2, Line 18: GetProcessorInfo
name the coreboot function: ` […]
Done
https://review.coreboot.org/c/coreboot/+/86004/comment/57d7494d_5261ab86?usp... : PS2, Line 21: complete path
what is a complete path? […]
Done
File src/cpu/x86/topology.c:
https://review.coreboot.org/c/coreboot/+/86004/comment/2c72ce5d_5b424ae0?usp... : PS2, Line 156: EFI_CPU_PHYSICAL_LOCATION
FSP quirks should be handled in FSP wrapper code, for example in ` […]
`mp_get_processor_info` does not know how many bit are reserved per domain and therefore cannot reconstitute the full path even if it was supplied with the die group, die, tile, module and core.
As an alternative to the current approach, I can introduce a new field `complete_path_core_id`, fill that one and make MP service use it instead of `core_id`. It should help lift your concern. What do you think?