This minor patch enables config option GENERIC_IDLE_POLL_SETUP for arch
powerpc. This should add support for kernel param 'nohlt'.

Powerpc kernel also supports another kernel boot-time param called
'powersave' which can also be used to disable all cpu idle-states and
forces CPU to an idle-loop similar to what cpu_idle_poll() does. This
patch however makes powerpc kernel-parameters better aligned to the
generic boot-time parameters.

Signed-off-by: Vaibhav Jain<vaibhav@linux.ibm.com>
---
 Documentation/admin-guide/kernel-parameters.txt | 2 +-
 arch/powerpc/Kconfig                            | 1 +
 2 files changed, 2 insertions(+), 1 deletion(-)

diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt
index 722b6eca2e93..6b657ebafcfb 100644
--- a/Documentation/admin-guide/kernel-parameters.txt
+++ b/Documentation/admin-guide/kernel-parameters.txt
@@ -3738,7 +3738,7 @@
 
 	nohibernate	[HIBERNATION] Disable hibernation and resume.
 
-	nohlt		[ARM,ARM64,MICROBLAZE,MIPS,SH] Forces the kernel to
+	nohlt		[ARM,ARM64,MICROBLAZE,MIPS,PPC,SH] Forces the kernel to
 			busy wait in do_idle() and not use the arch_cpu_idle()
 			implementation; requires CONFIG_GENERIC_IDLE_POLL_SETUP
 			to be effective. This is useful on platforms where the
diff --git a/arch/powerpc/Kconfig b/arch/powerpc/Kconfig
index 0b1172cbeccb..574661403800 100644
--- a/arch/powerpc/Kconfig
+++ b/arch/powerpc/Kconfig
@@ -193,6 +193,7 @@ config PPC
 	select GENERIC_CPU_VULNERABILITIES	if PPC_BARRIER_NOSPEC
 	select GENERIC_EARLY_IOREMAP
 	select GENERIC_GETTIMEOFDAY
+	select GENERIC_IDLE_POLL_SETUP
 	select GENERIC_IRQ_SHOW
 	select GENERIC_IRQ_SHOW_LEVEL
 	select GENERIC_PCI_IOMAP		if PCI
-- 
2.41.0

On Fri, 18 Aug 2023 10:37:37 +0530, Vaibhav Jain wrote:
> This minor patch enables config option GENERIC_IDLE_POLL_SETUP for arch
> powerpc. This should add support for kernel param 'nohlt'.
> 
> Powerpc kernel also supports another kernel boot-time param called
> 'powersave' which can also be used to disable all cpu idle-states and
> forces CPU to an idle-loop similar to what cpu_idle_poll() does. This
> patch however makes powerpc kernel-parameters better aligned to the
> generic boot-time parameters.
> 
> [...]

Applied to powerpc/next.

[1/1] powerpc: Enable generic cpu idle-loop
      https://git.kernel.org/powerpc/c/0ceef6e99cc3a57f33b84dafc6df5dfb3b28278d

cheers

Vaibhav Jain <vaibhav@linux.ibm.com> writes:
> This minor patch enables config option GENERIC_IDLE_POLL_SETUP for arch
> powerpc. This should add support for kernel param 'nohlt'.

Which is named after an 8086 instruction :), but oh well.

> Powerpc kernel also supports another kernel boot-time param called
> 'powersave' which can also be used to disable all cpu idle-states and
> forces CPU to an idle-loop similar to what cpu_idle_poll() does. This
> patch however makes powerpc kernel-parameters better aligned to the
> generic boot-time parameters.

It would be nice if we could make our powersave=off parameter just
enable this generic logic.

Have you looked at if that's possible? At a glance it looks like it
should be, when cpu_idle_force_poll is true do_idle() never calls
cpuidle_idle_call(), so the cpuidle drivers are never invoked.

cheers

Thanks for looking at this patch Mpe.

Michael Ellerman <mpe@ellerman.id.au> writes:

> Vaibhav Jain <vaibhav@linux.ibm.com> writes:
>> This minor patch enables config option GENERIC_IDLE_POLL_SETUP for arch
>> powerpc. This should add support for kernel param 'nohlt'.
>
> Which is named after an 8086 instruction :), but oh well.
>
Thanks. This is an interesting trivia :)

>> Powerpc kernel also supports another kernel boot-time param called
>> 'powersave' which can also be used to disable all cpu idle-states and
>> forces CPU to an idle-loop similar to what cpu_idle_poll() does. This
>> patch however makes powerpc kernel-parameters better aligned to the
>> generic boot-time parameters.
>
> It would be nice if we could make our powersave=off parameter just
> enable this generic logic.
>
> Have you looked at if that's possible? At a glance it looks like it
> should be, when cpu_idle_force_poll is true do_idle() never calls
> cpuidle_idle_call(), so the cpuidle drivers are never invoked.

Thanks for suggesting this and it looks like a simple fix to arch/idle.c
code. I have posted a v2 with the needed change on the mailing list at
https://lore.kernel.org/all/20230821045928.1350893-1-vaibhav@linux.ibm.com

-- 
Cheers
~ Vaibhav