acpi(4) corrections, clarifications

[b. f.-2]

While this documentation cleanup is going on, I'd like to point out
that acpi(4) could use some improvements:

(1)In the "SYSCTL VARIABLES" section, under hw.acpi.cpu.cx_lowest,
acpi(4) states "To enable ACPI CPU idling control,
machdep.cpu_idle_hlt must be set to 1".  However, apparently only the
pc98 port still uses this OID, while amd64, i386, etc., use something
like machdep.idle=hlt, so this remark should be changed.  And in any
case, is this statement correct?  Can't idling now be used with other
values of machdep.idle, like mwait, mwait_hlt, amdc1e, hlt, and acpi?

(2)Under the "BUGS" section, acpi(4) states:

"The acpi CPU idle power management drive conflicts with the local
APIC (LAPIC) timer. Disable APIC mode with hint.apic.0.disabled or do
not use the C2 and C3 states if APIC mode is enabled."

However, mav@, in his recent series of recommendations on
power-saving, advocated the use of hints like hint.apic.0.clock="0"
instead.  Is simply turning off the timer sufficient to use the deeper
sleep states, or must all of the local apic be disabled?  Some
machines won't boot with hint.apic.0.disabled, so it would be useful
to determine whether this is really necessary.


Regards,
                b.
_______________________________________________
freebsd-acpi@... mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-acpi
To unsubscribe, send any mail to "freebsd-acpi-unsubscribe@..."

[Alexander Motin-3]

b. f. wrote:

> While this documentation cleanup is going on, I'd like to point out
> that acpi(4) could use some improvements:
>
> (1)In the "SYSCTL VARIABLES" section, under hw.acpi.cpu.cx_lowest,
> acpi(4) states "To enable ACPI CPU idling control,
> machdep.cpu_idle_hlt must be set to 1".  However, apparently only the
> pc98 port still uses this OID, while amd64, i386, etc., use something
> like machdep.idle=hlt, so this remark should be changed.  And in any
> case, is this statement correct?  Can't idling now be used with other
> values of machdep.idle, like mwait, mwait_hlt, amdc1e, hlt, and acpi?
>
> (2)Under the "BUGS" section, acpi(4) states:
>
> "The acpi CPU idle power management drive conflicts with the local
> APIC (LAPIC) timer. Disable APIC mode with hint.apic.0.disabled or do
> not use the C2 and C3 states if APIC mode is enabled."
>
> However, mav@, in his recent series of recommendations on
> power-saving, advocated the use of hints like hint.apic.0.clock="0"
> instead.  Is simply turning off the timer sufficient to use the deeper
> sleep states, or must all of the local apic be disabled?  Some
> machines won't boot with hint.apic.0.disabled, so it would be useful
> to determine whether this is really necessary.

It is enough to disable APIC timer to use C2/C3. No need to disable
whole APIC. There were some objections that legacy timers may not route
their interrupts properly on some new systems, but I am not sure whether
disabling whole APIC help them somehow, but SMP it kill definitely.

--
Alexander Motin
_______________________________________________
freebsd-acpi@... mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-acpi
To unsubscribe, send any mail to "freebsd-acpi-unsubscribe@..."