2022-09-19 10:22:40

by Peter Zijlstra

[permalink] [raw]
Subject: [PATCH v2 07/44] cpuidle,psci: Push RCU-idle into driver

Doing RCU-idle outside the driver, only to then temporarily enable it
again, at least twice, before going idle is daft.

Signed-off-by: Peter Zijlstra (Intel) <[email protected]>
---
drivers/cpuidle/cpuidle-psci.c | 9 +++++----
1 file changed, 5 insertions(+), 4 deletions(-)

--- a/drivers/cpuidle/cpuidle-psci.c
+++ b/drivers/cpuidle/cpuidle-psci.c
@@ -69,12 +69,12 @@ static int __psci_enter_domain_idle_stat
return -1;

/* Do runtime PM to manage a hierarchical CPU toplogy. */
- ct_irq_enter_irqson();
if (s2idle)
dev_pm_genpd_suspend(pd_dev);
else
pm_runtime_put_sync_suspend(pd_dev);
- ct_irq_exit_irqson();
+
+ ct_idle_enter();

state = psci_get_domain_state();
if (!state)
@@ -82,12 +82,12 @@ static int __psci_enter_domain_idle_stat

ret = psci_cpu_suspend_enter(state) ? -1 : idx;

- ct_irq_enter_irqson();
+ ct_idle_exit();
+
if (s2idle)
dev_pm_genpd_resume(pd_dev);
else
pm_runtime_get_sync(pd_dev);
- ct_irq_exit_irqson();

cpu_pm_exit();

@@ -240,6 +240,7 @@ static int psci_dt_cpu_init_topology(str
* of a shared state for the domain, assumes the domain states are all
* deeper states.
*/
+ drv->states[state_count - 1].flags |= CPUIDLE_FLAG_RCU_IDLE;
drv->states[state_count - 1].enter = psci_enter_domain_idle_state;
drv->states[state_count - 1].enter_s2idle = psci_enter_s2idle_domain_idle_state;
psci_cpuidle_use_cpuhp = true;



2022-09-19 14:20:25

by Frederic Weisbecker

[permalink] [raw]
Subject: Re: [PATCH v2 07/44] cpuidle,psci: Push RCU-idle into driver

On Mon, Sep 19, 2022 at 11:59:46AM +0200, Peter Zijlstra wrote:
> Doing RCU-idle outside the driver, only to then temporarily enable it
> again, at least twice, before going idle is daft.
>
> Signed-off-by: Peter Zijlstra (Intel) <[email protected]>

Reviewed-by: Frederic Weisbecker <[email protected]>

2022-09-21 22:53:06

by Kajetan Puchalski

[permalink] [raw]
Subject: Re: [PATCH v2 07/44] cpuidle,psci: Push RCU-idle into driver

On Mon, Sep 19, 2022 at 11:59:46AM +0200, Peter Zijlstra wrote:
> Doing RCU-idle outside the driver, only to then temporarily enable it
> again, at least twice, before going idle is daft.
>
> Signed-off-by: Peter Zijlstra (Intel) <[email protected]>

Tried it on Pixel 6 running psci_idle, looks good with no apparent issues.

Tested-by: Kajetan Puchalski <[email protected]>

2022-09-22 00:48:25

by Guo Ren

[permalink] [raw]
Subject: Re: [PATCH v2 07/44] cpuidle,psci: Push RCU-idle into driver

Reviewed-by: Guo Ren <[email protected]>

On Mon, Sep 19, 2022 at 6:17 PM Peter Zijlstra <[email protected]> wrote:
>
> Doing RCU-idle outside the driver, only to then temporarily enable it
> again, at least twice, before going idle is daft.
>
> Signed-off-by: Peter Zijlstra (Intel) <[email protected]>
> ---
> drivers/cpuidle/cpuidle-psci.c | 9 +++++----
> 1 file changed, 5 insertions(+), 4 deletions(-)
>
> --- a/drivers/cpuidle/cpuidle-psci.c
> +++ b/drivers/cpuidle/cpuidle-psci.c
> @@ -69,12 +69,12 @@ static int __psci_enter_domain_idle_stat
> return -1;
>
> /* Do runtime PM to manage a hierarchical CPU toplogy. */
> - ct_irq_enter_irqson();
> if (s2idle)
> dev_pm_genpd_suspend(pd_dev);
> else
> pm_runtime_put_sync_suspend(pd_dev);
> - ct_irq_exit_irqson();
> +
> + ct_idle_enter();
>
> state = psci_get_domain_state();
> if (!state)
> @@ -82,12 +82,12 @@ static int __psci_enter_domain_idle_stat
>
> ret = psci_cpu_suspend_enter(state) ? -1 : idx;
>
> - ct_irq_enter_irqson();
> + ct_idle_exit();
> +
> if (s2idle)
> dev_pm_genpd_resume(pd_dev);
> else
> pm_runtime_get_sync(pd_dev);
> - ct_irq_exit_irqson();
>
> cpu_pm_exit();
>
> @@ -240,6 +240,7 @@ static int psci_dt_cpu_init_topology(str
> * of a shared state for the domain, assumes the domain states are all
> * deeper states.
> */
> + drv->states[state_count - 1].flags |= CPUIDLE_FLAG_RCU_IDLE;
> drv->states[state_count - 1].enter = psci_enter_domain_idle_state;
> drv->states[state_count - 1].enter_s2idle = psci_enter_s2idle_domain_idle_state;
> psci_cpuidle_use_cpuhp = true;
>
>


--
Best Regards
Guo Ren