diff mbox

platform/x86: intel_turbo_max_3: Add Skylake platform

Message ID 1508604720-7261-1-git-send-email-srinivas.pandruvada@linux.intel.com (mailing list archive)
State Accepted, archived
Delegated to: Andy Shevchenko
Headers show

Commit Message

Srinivas Pandruvada Oct. 21, 2017, 4:52 p.m. UTC
Ev Kontsevoy reported that he can't see the presence of
"/proc/sys/kernel/sched_itmt_enabled" on i9-7900x with Asrock x299
Taichi system even if he enabled "Turbo 3.0" in the BIOS.

The problem is that even if one core max is 200MHz more than others, the
current implementation couldn't enumerate that with the way the system
is configured.

The system by default configured for legacy mode (no HWP or speed shift
technology), in this mode only way we can enumerate via the mail box
interface as implemented in this driver. We were planing to only use
this driver for Broadwell, but we need to extend this because some
Skylake system has same issue as Braodwell systems.

On this system BIOS allows to change to HWP mode, where we expect that
we can enumerate favored core with ACPI-CPPC. But on this system the
core priority is 0xff for all cores in CPPC object. So this is not an
option.

Hence this change allows Skylake systems to be enumerate favored core
similar to Broadwell in legacy mode.

Reported-and-tested-by: Ev Kontsevoy <ev@kontsevoy.com>
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
---
 drivers/platform/x86/intel_turbo_max_3.c | 1 +
 1 file changed, 1 insertion(+)

Comments

Andy Shevchenko Oct. 27, 2017, 4:32 p.m. UTC | #1
On Sat, Oct 21, 2017 at 7:52 PM, Srinivas Pandruvada
<srinivas.pandruvada@linux.intel.com> wrote:
> Ev Kontsevoy reported that he can't see the presence of
> "/proc/sys/kernel/sched_itmt_enabled" on i9-7900x with Asrock x299
> Taichi system even if he enabled "Turbo 3.0" in the BIOS.
>
> The problem is that even if one core max is 200MHz more than others, the
> current implementation couldn't enumerate that with the way the system
> is configured.
>
> The system by default configured for legacy mode (no HWP or speed shift
> technology), in this mode only way we can enumerate via the mail box
> interface as implemented in this driver. We were planing to only use
> this driver for Broadwell, but we need to extend this because some
> Skylake system has same issue as Braodwell systems.
>
> On this system BIOS allows to change to HWP mode, where we expect that
> we can enumerate favored core with ACPI-CPPC. But on this system the
> core priority is 0xff for all cores in CPPC object. So this is not an
> option.
>
> Hence this change allows Skylake systems to be enumerate favored core
> similar to Broadwell in legacy mode.

Pushed to my review and testing queue, thanks!

>
> Reported-and-tested-by: Ev Kontsevoy <ev@kontsevoy.com>
> Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
> ---
>  drivers/platform/x86/intel_turbo_max_3.c | 1 +
>  1 file changed, 1 insertion(+)
>
> diff --git a/drivers/platform/x86/intel_turbo_max_3.c b/drivers/platform/x86/intel_turbo_max_3.c
> index 4f60d8e..d4ea018 100644
> --- a/drivers/platform/x86/intel_turbo_max_3.c
> +++ b/drivers/platform/x86/intel_turbo_max_3.c
> @@ -125,6 +125,7 @@ static int itmt_legacy_cpu_online(unsigned int cpu)
>
>  static const struct x86_cpu_id itmt_legacy_cpu_ids[] = {
>         ICPU(INTEL_FAM6_BROADWELL_X),
> +       ICPU(INTEL_FAM6_SKYLAKE_X),
>         {}
>  };
>
> --
> 2.5.5
>
diff mbox

Patch

diff --git a/drivers/platform/x86/intel_turbo_max_3.c b/drivers/platform/x86/intel_turbo_max_3.c
index 4f60d8e..d4ea018 100644
--- a/drivers/platform/x86/intel_turbo_max_3.c
+++ b/drivers/platform/x86/intel_turbo_max_3.c
@@ -125,6 +125,7 @@  static int itmt_legacy_cpu_online(unsigned int cpu)
 
 static const struct x86_cpu_id itmt_legacy_cpu_ids[] = {
 	ICPU(INTEL_FAM6_BROADWELL_X),
+	ICPU(INTEL_FAM6_SKYLAKE_X),
 	{}
 };