A system-wakeup QoS limit may have been requested by user-space. To avoid
entering a too deep state during s2idle, let's start to take into account
the QoS limit when selecting a suitable low-power-state.

Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
---
 drivers/cpuidle/cpuidle.c | 9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)

diff --git a/drivers/cpuidle/cpuidle.c b/drivers/cpuidle/cpuidle.c
index 0835da449db8..5f6dacb5b134 100644
--- a/drivers/cpuidle/cpuidle.c
+++ b/drivers/cpuidle/cpuidle.c
@@ -190,14 +190,15 @@ static noinstr void enter_s2idle_proper(struct cpuidle_driver *drv,
  */
 int cpuidle_enter_s2idle(struct cpuidle_driver *drv, struct cpuidle_device *dev)
 {
+	u64 constraint_ns = system_wakeup_latency_qos_limit() * NSEC_PER_USEC;
 	int index;
 
 	/*
-	 * Find the deepest state with ->enter_s2idle present, which guarantees
-	 * that interrupts won't be enabled when it exits and allows the tick to
-	 * be frozen safely.
+	 * Find the deepest state with ->enter_s2idle present that meets the
+	 * system-wakeup QoS limit, which guarantees that interrupts won't be
+	 * enabled when it exits and allows the tick to be frozen safely.
 	 */
-	index = find_deepest_state(drv, dev, U64_MAX, 0, true);
+	index = find_deepest_state(drv, dev, constraint_ns, 0, true);
 	if (index > 0) {
 		enter_s2idle_proper(drv, dev, index);
 		local_irq_enable();
-- 
2.43.0

On Wed, Jul 16, 2025 at 2:33 PM Ulf Hansson <ulf.hansson@linaro.org> wrote:
>
> A system-wakeup QoS limit may have been requested by user-space. To avoid
> entering a too deep state during s2idle, let's start to take into account
> the QoS limit when selecting a suitable low-power-state.
>
> Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
> ---
>  drivers/cpuidle/cpuidle.c | 9 +++++----
>  1 file changed, 5 insertions(+), 4 deletions(-)
>
> diff --git a/drivers/cpuidle/cpuidle.c b/drivers/cpuidle/cpuidle.c
> index 0835da449db8..5f6dacb5b134 100644
> --- a/drivers/cpuidle/cpuidle.c
> +++ b/drivers/cpuidle/cpuidle.c
> @@ -190,14 +190,15 @@ static noinstr void enter_s2idle_proper(struct cpuidle_driver *drv,
>   */
>  int cpuidle_enter_s2idle(struct cpuidle_driver *drv, struct cpuidle_device *dev)
>  {
> +       u64 constraint_ns = system_wakeup_latency_qos_limit() * NSEC_PER_USEC;
>         int index;
>
>         /*
> -        * Find the deepest state with ->enter_s2idle present, which guarantees
> -        * that interrupts won't be enabled when it exits and allows the tick to
> -        * be frozen safely.
> +        * Find the deepest state with ->enter_s2idle present that meets the
> +        * system-wakeup QoS limit, which guarantees that interrupts won't be
> +        * enabled when it exits and allows the tick to be frozen safely.
>          */
> -       index = find_deepest_state(drv, dev, U64_MAX, 0, true);
> +       index = find_deepest_state(drv, dev, constraint_ns, 0, true);
>         if (index > 0) {
>                 enter_s2idle_proper(drv, dev, index);
>                 local_irq_enable();

This is not the only place that needs to be patched this way.
cpuidle_idle_call() is another one AFAICS.