Re: [PATCH 5/7] cpuidle: Introduce CPUIDLE_FLAG_POLLING_HARD

[Rafael J. Wysocki]

On Fri, Nov 24, 2023 at 11:32 PM Frederic Weisbecker
<frederic@xxxxxxxxxx> wrote:
>
> From: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
>
> Provide a way to tell the cpuidle core about states polling/monitoring
> TIF_NEED_RESCHED on the hardware level, monitor/mwait users being the
> only examples in use.
>
> This will allow cpuidle core to manage TIF_NR_POLLING on behalf of all
> kinds of TIF_NEED_RESCHED polling states while keeping a necessary
> distinction for the governors between software loops polling on
> TIF_NEED_RESCHED and hardware monitored writes to thread flags.
>
> [fweisbec: _ Initialize flag from acpi_processor_setup_cstates() instead
>              of acpi_processor_setup_lpi_states(), as the latter seem to
>              be about arm64...
>            _ Rename CPUIDLE_FLAG_NO_IPI to CPUIDLE_FLAG_POLLING_HARD]
>
> Not-yet-signed-off-by: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
> Signed-off-by: Frederic Weisbecker <frederic@xxxxxxxxxx>
> ---
>  drivers/acpi/processor_idle.c | 3 +++
>  drivers/idle/intel_idle.c     | 5 ++++-
>  include/linux/cpuidle.h       | 3 ++-
>  3 files changed, 9 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/acpi/processor_idle.c b/drivers/acpi/processor_idle.c
> index 3a34a8c425fe..a77a4d4b0dad 100644
> --- a/drivers/acpi/processor_idle.c
> +++ b/drivers/acpi/processor_idle.c
> @@ -814,6 +814,9 @@ static int acpi_processor_setup_cstates(struct acpi_processor *pr)
>                         if (cx->type != ACPI_STATE_C3)
>                                 drv->safe_state_index = count;
>                 }
> +
> +               if (cx->entry_method == ACPI_CSTATE_FFH)
> +                       state->flags |= CPUIDLE_FLAG_POLLING_HARD;
>                 /*
>                  * Halt-induced C1 is not good for ->enter_s2idle, because it
>                  * re-enables interrupts on exit.  Moreover, C1 is generally not
> diff --git a/drivers/idle/intel_idle.c b/drivers/idle/intel_idle.c
> index 3e01a6b23e75..bc56624fe0b5 100644
> --- a/drivers/idle/intel_idle.c
> +++ b/drivers/idle/intel_idle.c
> @@ -1563,7 +1563,8 @@ static void __init intel_idle_init_cstates_acpi(struct cpuidle_driver *drv)
>                 if (cx->type > ACPI_STATE_C1)
>                         state->target_residency *= 3;
>
> -               state->flags = MWAIT2flg(cx->address);
> +               state->flags = MWAIT2flg(cx->address) | CPUIDLE_FLAG_POLLING_HARD;
> +
>                 if (cx->type > ACPI_STATE_C2)
>                         state->flags |= CPUIDLE_FLAG_TLB_FLUSHED;
>
> @@ -1836,6 +1837,8 @@ static bool __init intel_idle_verify_cstate(unsigned int mwait_hint)
>
>  static void state_update_enter_method(struct cpuidle_state *state, int cstate)
>  {
> +       state->flags |= CPUIDLE_FLAG_POLLING_HARD;
> +
>         if (state->flags & CPUIDLE_FLAG_INIT_XSTATE) {
>                 /*
>                  * Combining with XSTATE with IBRS or IRQ_ENABLE flags
> diff --git a/include/linux/cpuidle.h b/include/linux/cpuidle.h
> index 66b59868622c..873fdf200dc3 100644
> --- a/include/linux/cpuidle.h
> +++ b/include/linux/cpuidle.h
> @@ -78,13 +78,14 @@ struct cpuidle_state {
>
>  /* Idle State Flags */
>  #define CPUIDLE_FLAG_NONE              (0x00)
> -#define CPUIDLE_FLAG_POLLING_SOFT              BIT(0) /* polling state */
> +#define CPUIDLE_FLAG_POLLING_SOFT      BIT(0) /* software need_resched() polling state */
>  #define CPUIDLE_FLAG_COUPLED           BIT(1) /* state applies to multiple cpus */
>  #define CPUIDLE_FLAG_TIMER_STOP        BIT(2) /* timer is stopped on this state */
>  #define CPUIDLE_FLAG_UNUSABLE          BIT(3) /* avoid using this state */
>  #define CPUIDLE_FLAG_OFF               BIT(4) /* disable this state by default */
>  #define CPUIDLE_FLAG_TLB_FLUSHED       BIT(5) /* idle-state flushes TLBs */
>  #define CPUIDLE_FLAG_RCU_IDLE          BIT(6) /* idle-state takes care of RCU */
> +#define CPUIDLE_FLAG_POLLING_HARD      BIT(7) /* hardware need_resched() polling state */

Hardware need_resched() monitoring rather?  This doesn't do what
"polling" usually means AFAICS.

>
>  struct cpuidle_device_kobj;
>  struct cpuidle_state_kobj;
> --