LGTM

• Delete the conditional branching part using the EARLY_AP_STARTUP macro constant. (because the EARLY_AP_STARTUP constant definition is now mandatory.)
• Add a comment to the x2apic_mode check block in the lapic_enable_thermal function.
• Modify the lapic_handle_thermal function so that the handle function and its argument pointers are saved to the stack before being used. (improving safety in the event that the pointers are cleared between the NULL check and the handle function call).

• Fix the old style of comments at the beginning of the file (removed hyphens).
• Change the HISTORY section from "under development" to "released from FreeBSD 16.0."

LGTM

Please remove all SCHED_ULE ifs. Recently we made a change so that scheduler can be chosen on boot time, so it's now meaningless to have SCHED_ULE or SCHED_4BSD

• Squash the previous review fixes and re-create the patch commit.
• Correct the driver name insertion line to be in alphabetical order

• Squash the previous review fixes and re-create the patch commit.
• The driver description was mistakenly placed in the VM OPTIONS section, so it has been moved to the SMP OPTIONS section.

Squash the previous review fixes and re-create the patch commit.

The diff is off. Could you rebase main and create diff with git format-patch -U999999 --stdout main > change.diff?

You can squash all your previous and this one into one commit. (and that's preferred)

You can squash all your previous and this one into one commit. (and that's preferred)

The diff is off. Could you rebase main and create diff with git format-patch -U999999 --stdout main > change.diff?

I checked the latest version of the Intel Software Developer's manual (SDM) again, and it appears that the Intel Hardware Feedback Interface / Thread Director does not provide a table with performance index values ​​for each P core/E core/LP-E core, but rather a table with two values ​​for each core group: a performance index value (computation speed) and a power saving index value (power efficiency).
It does not appear that the table or columns have been expanded for LP-E cores.

• Correct the driver name insertion line to be in alphabetical order (My mistake. Sorry)

The driver description was mistakenly placed in the VM OPTIONS section, so it has been moved to the SMP OPTIONS section.
(Sorry for not noticing my carelessness. Thank you for pointing that out.)

• Correct the driver name insertion line to be in alphabetical order (My mistake. Sorry)

This one is done in D42612

LGTM

• Fix the driver name (coredirector -> intelhfi)
• Fix file name with changed the driver name.
• Fix review findings. (Copyright style, etc)

• Fix the driver name (coredirector -> intelhfi)
• Fix review findings. (__builtin_popcount, device_printf, etc)

• Fix macro define name (CPUGRP_SCORE -> HMP)
• Rebase for freebsd current source tree.

• Fix the driver name (coredirector -> intelhfi)
• Fix file name with changed the driver name.

Fix the driver name (coredirector -> intelhfi) and macro definition name (CPUGRP_SCORE -> HMP).

In D55091#1259161, @emaste wrote:

I don't think there's any expectation that the rc scripts will be portable, but when there's no material difference (e.g. in performance, clarity, maintainability, etc.) I agree we should indeed prefer the portable way.

I don't think there's any expectation that the rc scripts will be portable, but when there's no material difference (e.g. in performance, clarity, maintainability, etc.) I agree we should indeed prefer the portable way.