Re: [PATCH] cpufreq: intel_pstate: Allow EPP 0x80 setting by the firmware

[Rafael J. Wysocki]

On Thu, Oct 27, 2022 at 7:19 PM Srinivas Pandruvada
<srinivas.pandruvada@xxxxxxxxxxxxxxx> wrote:
>
> With the
> "commit 3d13058ed2a6 ("cpufreq: intel_pstate: Use firmware default EPP")"
> the firmware can set an EPP, which driver will not overwrite. But the
> driver has a valid range check for:
> 0x40 > firmware epp < 0x80.
> Hence firmware can't specify EPP of 0x80.
>
> If the firmware didn't specify in the valid range, the driver has a
> hard coded EPP of 102. But some Chrome hardware vendors don't want
> this overwrite and wants to boot with chipset default EPP of 0x80 as
> this improves battery life.
>
> In this case they want to have capability to specify EPP of 0x80 via
> the firmware. This require the valid range to include 0x80 also.
> But here the valid range can't be simply extended to include 0x80 as
> this is the chipset default EPP. Even without any firmware specifying
> EPP, the chipset will always boot with EPP of 0x80.
>
> To make sure that firmware specified EPP of 0x80 and not by the
> chipset default, it will require additional check to make sure HWP
> was enabled by the firmware before boot. Only way the firmware can
> update EPP, is to enable HWP and update EPP via MSR_HWP_REQUEST.
>
> This driver already checks, if the HWP is enabled by the firmware.
> Use the same flag and extend valid range to include 0x80.
>
> Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@xxxxxxxxxxxxxxx>
> ---
>  drivers/cpufreq/intel_pstate.c | 9 +++++----
>  1 file changed, 5 insertions(+), 4 deletions(-)
>
> diff --git a/drivers/cpufreq/intel_pstate.c b/drivers/cpufreq/intel_pstate.c
> index fc3ebeb0bbe5..c41875abd8be 100644
> --- a/drivers/cpufreq/intel_pstate.c
> +++ b/drivers/cpufreq/intel_pstate.c
> @@ -297,6 +297,7 @@ static int hwp_active __read_mostly;
>  static int hwp_mode_bdw __read_mostly;
>  static bool per_cpu_limits __read_mostly;
>  static bool hwp_boost __read_mostly;
> +static bool hwp_forced __read_mostly;
>
>  static struct cpufreq_driver *intel_pstate_driver __read_mostly;
>
> @@ -1705,12 +1706,12 @@ static void intel_pstate_update_epp_defaults(struct cpudata *cpudata)
>                 return;
>
>         /*
> -        * If powerup EPP is something other than chipset default 0x80 and
> -        * - is more performance oriented than 0x80 (default balance_perf EPP)
> +        * If the EPP is set by firmware, which means that firmware enabled HWP
> +        * - Is equal or less than 0x80 (default balance_perf EPP)
>          * - But less performance oriented than performance EPP
>          *   then use this as new balance_perf EPP.
>          */
> -       if (cpudata->epp_default < HWP_EPP_BALANCE_PERFORMANCE &&
> +       if (hwp_forced && cpudata->epp_default <= HWP_EPP_BALANCE_PERFORMANCE &&
>             cpudata->epp_default > HWP_EPP_PERFORMANCE) {
>                 epp_values[EPP_INDEX_BALANCE_PERFORMANCE] = cpudata->epp_default;
>                 return;
> @@ -3423,7 +3424,7 @@ static int __init intel_pstate_init(void)
>
>         id = x86_match_cpu(hwp_support_ids);
>         if (id) {
> -               bool hwp_forced = intel_pstate_hwp_is_enabled();
> +               hwp_forced = intel_pstate_hwp_is_enabled();
>
>                 if (hwp_forced)
>                         pr_info("HWP enabled by BIOS\n");
> --

Applied as 6.2 material, thanks!