2023-11-29 14:32:33

by Jiang Yihe

Subject: [PATCH] intel_idle: add Cometlake support

Since the Cometlake C-State is supported, support for Cometlake should
be added to intel_idle. Just use Kabylake C-State table for Cometlake
because they share the same table in intel_cstate.

Signed-off-by: Jiang Yihe <[email protected]>
---
drivers/idle/intel_idle.c | 2 ++
1 file changed, 2 insertions(+)

diff --git a/drivers/idle/intel_idle.c b/drivers/idle/intel_idle.c
index dcda0afec..f83f78037 100644
--- a/drivers/idle/intel_idle.c
+++ b/drivers/idle/intel_idle.c
@@ -1418,6 +1418,8 @@ static const struct x86_cpu_id intel_idle_ids[] __initconst = {
X86_MATCH_INTEL_FAM6_MODEL(SKYLAKE, &idle_cpu_skl),
X86_MATCH_INTEL_FAM6_MODEL(KABYLAKE_L, &idle_cpu_skl),
X86_MATCH_INTEL_FAM6_MODEL(KABYLAKE, &idle_cpu_skl),
+ X86_MATCH_INTEL_FAM6_MODEL(COMETLAKE_L, &idle_cpu_skl),
+ X86_MATCH_INTEL_FAM6_MODEL(COMETLAKE, &idle_cpu_skl),
X86_MATCH_INTEL_FAM6_MODEL(SKYLAKE_X, &idle_cpu_skx),
X86_MATCH_INTEL_FAM6_MODEL(ICELAKE_X, &idle_cpu_icx),
X86_MATCH_INTEL_FAM6_MODEL(ICELAKE_D, &idle_cpu_icx),
--
2.43.0

2023-11-29 15:19:55

by Rafael J. Wysocki

Subject: Re: [PATCH] intel_idle: add Cometlake support

+Rui and Artem

On Wed, Nov 29, 2023 at 3:32 PM Jiang Yihe <[email protected]> wrote:
>
> Since the Cometlake C-State is supported, support for Cometlake should
> be added to intel_idle. Just use Kabylake C-State table for Cometlake
> because they share the same table in intel_cstate.
>
> Signed-off-by: Jiang Yihe <[email protected]>
> ---
> drivers/idle/intel_idle.c | 2 ++
> 1 file changed, 2 insertions(+)
>
> diff --git a/drivers/idle/intel_idle.c b/drivers/idle/intel_idle.c
> index dcda0afec..f83f78037 100644
> --- a/drivers/idle/intel_idle.c
> +++ b/drivers/idle/intel_idle.c
> @@ -1418,6 +1418,8 @@ static const struct x86_cpu_id intel_idle_ids[] __initconst = {
> X86_MATCH_INTEL_FAM6_MODEL(SKYLAKE, &idle_cpu_skl),
> X86_MATCH_INTEL_FAM6_MODEL(KABYLAKE_L, &idle_cpu_skl),
> X86_MATCH_INTEL_FAM6_MODEL(KABYLAKE, &idle_cpu_skl),
> + X86_MATCH_INTEL_FAM6_MODEL(COMETLAKE_L, &idle_cpu_skl),
> + X86_MATCH_INTEL_FAM6_MODEL(COMETLAKE, &idle_cpu_skl),
> X86_MATCH_INTEL_FAM6_MODEL(SKYLAKE_X, &idle_cpu_skx),
> X86_MATCH_INTEL_FAM6_MODEL(ICELAKE_X, &idle_cpu_icx),
> X86_MATCH_INTEL_FAM6_MODEL(ICELAKE_D, &idle_cpu_icx),
> --

2023-11-29 15:46:14

by Doug Smythies

Subject: RE: [PATCH] intel_idle: add Cometlake support

Hi All,

This proposed patch has been submitted at least once before.
(I think more than once, but only found 1)
Reference:
https://lore.kernel.org/linux-pm/20200227013411.1.Ica3bb9fa898499d94e0b0a2bfa08ec46c89d84fa@changeid/

I have a Cometlake processor and am not in favor of this patch.
Reference at to why not:
https://bugzilla.kernel.org/show_bug.cgi?id=210741

... Doug

On 2023.11.29 07:19 Rafael wrote:
> +Rui and Artem
>On Wed, Nov 29, 2023 at 3:32 PM Jiang Yihe <[email protected]> wrote:
>>
>> Since the Cometlake C-State is supported, support for Cometlake should
>> be added to intel_idle. Just use Kabylake C-State table for Cometlake
>> because they share the same table in intel_cstate.
>>
>> Signed-off-by: Jiang Yihe <[email protected]>
>> ---
>> drivers/idle/intel_idle.c | 2 ++
>> 1 file changed, 2 insertions(+)
>>
>> diff --git a/drivers/idle/intel_idle.c b/drivers/idle/intel_idle.c
>> index dcda0afec..f83f78037 100644
>> --- a/drivers/idle/intel_idle.c
>> +++ b/drivers/idle/intel_idle.c
>> @@ -1418,6 +1418,8 @@ static const struct x86_cpu_id intel_idle_ids[] __initconst = {
>> X86_MATCH_INTEL_FAM6_MODEL(SKYLAKE, &idle_cpu_skl),
>> X86_MATCH_INTEL_FAM6_MODEL(KABYLAKE_L, &idle_cpu_skl),
>> X86_MATCH_INTEL_FAM6_MODEL(KABYLAKE, &idle_cpu_skl),
>> + X86_MATCH_INTEL_FAM6_MODEL(COMETLAKE_L, &idle_cpu_skl),
>> + X86_MATCH_INTEL_FAM6_MODEL(COMETLAKE, &idle_cpu_skl),
>> X86_MATCH_INTEL_FAM6_MODEL(SKYLAKE_X, &idle_cpu_skx),
>> X86_MATCH_INTEL_FAM6_MODEL(ICELAKE_X, &idle_cpu_icx),
>> X86_MATCH_INTEL_FAM6_MODEL(ICELAKE_D, &idle_cpu_icx),
>> --

2023-11-30 08:24:51

by Artem Bityutskiy

Subject: Re: [PATCH] intel_idle: add Cometlake support

On Wed, 2023-11-29 at 07:45 -0800, Doug Smythies wrote:
> Hi All,
>
> This proposed patch has been submitted at least once before.
> (I think more than once, but only found 1)
> Reference:
> https://lore.kernel.org/linux-pm/20200227013411.1.Ica3bb9fa898499d94e0b0a2bfa08ec46c89d84fa@changeid/
>
> I have a Cometlake processor and am not in favor of this patch.
> Reference at to why not:
> https://bugzilla.kernel.org/show_bug.cgi?id=210741

CometLake is a platform from 2020, so I'd classified it as "legacy" and would
not apply this patch without a plausible justification in form of measurement
data.

Comet lake has been used with ACPI C-states for quite a long time, and switching
it to intel_idle now may cause unwelcome changes in user workloads.

Artem.