(Small point: Last if currently looks slightly weird as it is calling getcwd_physical() even on !physical, would use a logical variable instead of physical.)

In D55139#1260674, @guest-seuros wrote:

Your approach is much more elegant..

Sorry for not doing another search before submitting. I had this change locally (before the diff was open).

Thanks.

New approach

"Simpler" approach with some common file holding the common sysctl knobs. As said, I anticipate that machdep.hwpstate_pkg_ctrl will be removed from hwpstate_amd, but debug.hwpstate_verbose will probably stay and start to be used in hwpstate_intel as well. There are other knobs in hwpstate_amd that might become relevant for Intel too, but I haven't spent too much time on them, they could be moved later.

I'm already working on it. May be what ShengYi proposed in D54528 already, but I'm trying a different, somewhat simpler approach.

I do not see any other such type mistake with enum power_stype involved (have checked quite a number of references, including all of acpic.c and subr_power.c).

In D54996#1258764, @kib wrote:

And yes, SAFE variants are added specifically to be treated differently from the RENDEZVOUS case. If you want/need atomicity, the RENDEZVOUS is the right choice, if atomicity is not needed, which is not rare case, SAFE is good.

In D54996#1258764, @kib wrote:

SAFE variants cannot be called while a spinlock is owned, otherwise sched_bind() cannot work, it needs to migrate thread to different CPU.

In D54996#1258511, @kib wrote:

I am fine with either case, IMO it is simpler to disable the interrupts unconditionally. It is boot code only anyway, and we should target the code size and not code speed.

Unconditionally deactivate interrupts on MSR_OP_LOCAL.

Gentle ping.

Not doing it now, but going further, in order to facilitate changing the internal type of sysctl knobs that are integers, beyond the case evoked here (int -> bool), without breaking the ABI, we should probably tolerate any type of integer in input and output (really, size of I/O buffers) and convert them to the target (internal) type, raising some errors where appropriate (e.g., if the value is out of range for the target type with the exception of booleans).

Would appreciate a:
Suggested by: olce
tag line at commit. Thanks!

• Disable interrupts on MSR_OP_LOCAL and non-atomic MSR manipulation.
• Trim and update comments.

In D54833#1256748, @kevans wrote:

rer: removing the prlabel, @csjp noted (paraphrasing, maybe poorly) that that's a MAC-philosophical thing that policy (writers?) shouldn't have to have 'that' much knowledge of kernel internals. It turns out that they often do anyways, but the historically consistent thing is that object labels get passed along with the labels to avoid the policy reaching into the object when it's something that they'll want to act on.

In D54833#1255914, @kevans wrote:

prison_created is done after we know that the jail's going to stick around because I wanted the prison state to be 'final' (post-OSD) before we do any label propagation stuff... maybe that wasn't quite the right call.

The /* Symmetry with prison_created */ comment is indeed welcome to clear the slight but apparently necessary confusion coming from the hook prison_cleanup being called in mac_prison_destroy() while prison_created is not called from mac_prison_init(). :-)

Alternative: Leave sysctl_handle_bool() as is, and created another one, as well as accompanying macros SYSCTL_ADD_xxx() and co., to define booleans compatible with ints. In addition, these could unconditionally output/input an integer, easing back and forth internal transition between int and bool as this would preserve the ABI.

Remove useless initialization of temp_int in the SYSCTL_IN() case.

Update commit message

Update commit message

Prevent users to set hw.acpi.s4bios to true if S4BIOS is not supported.

• Impacts of new prerequisite (D54926)
• Slightly tweak the description of 'hw.acpi.s4bios'