Doing RCU-idle outside the driver, only to then temporarily enable it
again before going idle is daft.
Notably: this converts all dt_init_idle_driver() and
__CPU_PM_CPU_IDLE_ENTER() users for they are inextrably intertwined.
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
---
arch/arm/mach-omap2/cpuidle34xx.c | 4 ++--
drivers/acpi/processor_idle.c | 2 ++
drivers/cpuidle/cpuidle-arm.c | 1 +
drivers/cpuidle/cpuidle-big_little.c | 8 ++++++--
drivers/cpuidle/cpuidle-psci.c | 1 +
drivers/cpuidle/cpuidle-qcom-spm.c | 1 +
drivers/cpuidle/cpuidle-riscv-sbi.c | 1 +
drivers/cpuidle/dt_idle_states.c | 2 +-
include/linux/cpuidle.h | 4 ++++
9 files changed, 19 insertions(+), 5 deletions(-)
--- a/drivers/acpi/processor_idle.c
+++ b/drivers/acpi/processor_idle.c
@@ -1200,6 +1200,8 @@ static int acpi_processor_setup_lpi_stat
state->target_residency = lpi->min_residency;
if (lpi->arch_flags)
state->flags |= CPUIDLE_FLAG_TIMER_STOP;
+ if (lpi->entry_method == ACPI_CSTATE_FFH)
+ state->flags |= CPUIDLE_FLAG_RCU_IDLE;
state->enter = acpi_idle_lpi_enter;
drv->safe_state_index = i;
}
--- a/drivers/cpuidle/cpuidle-arm.c
+++ b/drivers/cpuidle/cpuidle-arm.c
@@ -53,6 +53,7 @@ static struct cpuidle_driver arm_idle_dr
* handler for idle state index 0.
*/
.states[0] = {
+ .flags = CPUIDLE_FLAG_RCU_IDLE,
.enter = arm_enter_idle_state,
.exit_latency = 1,
.target_residency = 1,
--- a/drivers/cpuidle/cpuidle-big_little.c
+++ b/drivers/cpuidle/cpuidle-big_little.c
@@ -64,7 +64,8 @@ static struct cpuidle_driver bl_idle_lit
.enter = bl_enter_powerdown,
.exit_latency = 700,
.target_residency = 2500,
- .flags = CPUIDLE_FLAG_TIMER_STOP,
+ .flags = CPUIDLE_FLAG_TIMER_STOP |
+ CPUIDLE_FLAG_RCU_IDLE,
.name = "C1",
.desc = "ARM little-cluster power down",
},
@@ -85,7 +86,8 @@ static struct cpuidle_driver bl_idle_big
.enter = bl_enter_powerdown,
.exit_latency = 500,
.target_residency = 2000,
- .flags = CPUIDLE_FLAG_TIMER_STOP,
+ .flags = CPUIDLE_FLAG_TIMER_STOP |
+ CPUIDLE_FLAG_RCU_IDLE,
.name = "C1",
.desc = "ARM big-cluster power down",
},
@@ -124,11 +126,13 @@ static int bl_enter_powerdown(struct cpu
struct cpuidle_driver *drv, int idx)
{
cpu_pm_enter();
+ ct_idle_enter();
cpu_suspend(0, bl_powerdown_finisher);
/* signals the MCPM core that CPU is out of low power state */
mcpm_cpu_powered_up();
+ ct_idle_exit();
cpu_pm_exit();
--- a/drivers/cpuidle/cpuidle-psci.c
+++ b/drivers/cpuidle/cpuidle-psci.c
@@ -357,6 +357,7 @@ static int psci_idle_init_cpu(struct dev
* PSCI idle states relies on architectural WFI to be represented as
* state index 0.
*/
+ drv->states[0].flags = CPUIDLE_FLAG_RCU_IDLE;
drv->states[0].enter = psci_enter_idle_state;
drv->states[0].exit_latency = 1;
drv->states[0].target_residency = 1;
--- a/drivers/cpuidle/cpuidle-qcom-spm.c
+++ b/drivers/cpuidle/cpuidle-qcom-spm.c
@@ -72,6 +72,7 @@ static struct cpuidle_driver qcom_spm_id
.owner = THIS_MODULE,
.states[0] = {
.enter = spm_enter_idle_state,
+ .flags = CPUIDLE_FLAG_RCU_IDLE,
.exit_latency = 1,
.target_residency = 1,
.power_usage = UINT_MAX,
--- a/drivers/cpuidle/cpuidle-riscv-sbi.c
+++ b/drivers/cpuidle/cpuidle-riscv-sbi.c
@@ -332,6 +332,7 @@ static int sbi_cpuidle_init_cpu(struct d
drv->cpumask = (struct cpumask *)cpumask_of(cpu);
/* RISC-V architectural WFI to be represented as state index 0. */
+ drv->states[0].flags = CPUIDLE_FLAG_RCU_IDLE;
drv->states[0].enter = sbi_cpuidle_enter_state;
drv->states[0].exit_latency = 1;
drv->states[0].target_residency = 1;
--- a/drivers/cpuidle/dt_idle_states.c
+++ b/drivers/cpuidle/dt_idle_states.c
@@ -77,7 +77,7 @@ static int init_state_node(struct cpuidl
if (err)
desc = state_node->name;
- idle_state->flags = 0;
+ idle_state->flags = CPUIDLE_FLAG_RCU_IDLE;
if (of_property_read_bool(state_node, "local-timer-stop"))
idle_state->flags |= CPUIDLE_FLAG_TIMER_STOP;
/*
--- a/include/linux/cpuidle.h
+++ b/include/linux/cpuidle.h
@@ -282,14 +282,18 @@ extern s64 cpuidle_governor_latency_req(
int __ret = 0; \
\
if (!idx) { \
+ ct_idle_enter(); \
cpu_do_idle(); \
+ ct_idle_exit(); \
return idx; \
} \
\
if (!is_retention) \
__ret = cpu_pm_enter(); \
if (!__ret) { \
+ ct_idle_enter(); \
__ret = low_level_idle_enter(state); \
+ ct_idle_exit(); \
if (!is_retention) \
cpu_pm_exit(); \
} \On Mon, 19 Sept 2022 at 12:18, Peter Zijlstra <peterz@infradead.org> wrote:
>
> Doing RCU-idle outside the driver, only to then temporarily enable it
> again before going idle is daft.
>
> Notably: this converts all dt_init_idle_driver() and
> __CPU_PM_CPU_IDLE_ENTER() users for they are inextrably intertwined.
>
> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Ulf Hansson <ulf.hansson@linaro.org>
Kind regards
Uffe
> ---
> arch/arm/mach-omap2/cpuidle34xx.c | 4 ++--
> drivers/acpi/processor_idle.c | 2 ++
> drivers/cpuidle/cpuidle-arm.c | 1 +
> drivers/cpuidle/cpuidle-big_little.c | 8 ++++++--
> drivers/cpuidle/cpuidle-psci.c | 1 +
> drivers/cpuidle/cpuidle-qcom-spm.c | 1 +
> drivers/cpuidle/cpuidle-riscv-sbi.c | 1 +
> drivers/cpuidle/dt_idle_states.c | 2 +-
> include/linux/cpuidle.h | 4 ++++
> 9 files changed, 19 insertions(+), 5 deletions(-)
>
> --- a/drivers/acpi/processor_idle.c
> +++ b/drivers/acpi/processor_idle.c
> @@ -1200,6 +1200,8 @@ static int acpi_processor_setup_lpi_stat
> state->target_residency = lpi->min_residency;
> if (lpi->arch_flags)
> state->flags |= CPUIDLE_FLAG_TIMER_STOP;
> + if (lpi->entry_method == ACPI_CSTATE_FFH)
> + state->flags |= CPUIDLE_FLAG_RCU_IDLE;
I assume the state index here will never be 0?
If not, it may lead to that acpi_processor_ffh_lpi_enter() may trigger
CPU_PM_CPU_IDLE_ENTER_PARAM() to call ct_cpuidle_enter|exit() for an
idle-state that doesn't have the CPUIDLE_FLAG_RCU_IDLE bit set.
> state->enter = acpi_idle_lpi_enter;
> drv->safe_state_index = i;
> }
> --- a/drivers/cpuidle/cpuidle-arm.c
> +++ b/drivers/cpuidle/cpuidle-arm.c
> @@ -53,6 +53,7 @@ static struct cpuidle_driver arm_idle_dr
> * handler for idle state index 0.
> */
> .states[0] = {
> + .flags = CPUIDLE_FLAG_RCU_IDLE,
Comparing arm64 and arm32 idle-states/idle-drivers, the $subject
series ends up setting the CPUIDLE_FLAG_RCU_IDLE for the ARM WFI idle
state (state zero), but only for the arm64 and psci cases (mostly
arm64). For arm32 we would need to update the ARM_CPUIDLE_WFI_STATE
too, as that is what most arm32 idle-drivers are using. My point is,
the code becomes a bit inconsistent.
Perhaps it's easier to avoid setting the CPUIDLE_FLAG_RCU_IDLE bit for
all of the ARM WFI idle states, for both arm64 and arm32?
> .enter = arm_enter_idle_state,
> .exit_latency = 1,
> .target_residency = 1,
> --- a/drivers/cpuidle/cpuidle-big_little.c
> +++ b/drivers/cpuidle/cpuidle-big_little.c
> @@ -64,7 +64,8 @@ static struct cpuidle_driver bl_idle_lit
> .enter = bl_enter_powerdown,
> .exit_latency = 700,
> .target_residency = 2500,
> - .flags = CPUIDLE_FLAG_TIMER_STOP,
> + .flags = CPUIDLE_FLAG_TIMER_STOP |
> + CPUIDLE_FLAG_RCU_IDLE,
> .name = "C1",
> .desc = "ARM little-cluster power down",
> },
> @@ -85,7 +86,8 @@ static struct cpuidle_driver bl_idle_big
> .enter = bl_enter_powerdown,
> .exit_latency = 500,
> .target_residency = 2000,
> - .flags = CPUIDLE_FLAG_TIMER_STOP,
> + .flags = CPUIDLE_FLAG_TIMER_STOP |
> + CPUIDLE_FLAG_RCU_IDLE,
> .name = "C1",
> .desc = "ARM big-cluster power down",
> },
> @@ -124,11 +126,13 @@ static int bl_enter_powerdown(struct cpu
> struct cpuidle_driver *drv, int idx)
> {
> cpu_pm_enter();
> + ct_idle_enter();
>
> cpu_suspend(0, bl_powerdown_finisher);
>
> /* signals the MCPM core that CPU is out of low power state */
> mcpm_cpu_powered_up();
> + ct_idle_exit();
>
> cpu_pm_exit();
>
> --- a/drivers/cpuidle/cpuidle-psci.c
> +++ b/drivers/cpuidle/cpuidle-psci.c
> @@ -357,6 +357,7 @@ static int psci_idle_init_cpu(struct dev
> * PSCI idle states relies on architectural WFI to be represented as
> * state index 0.
> */
> + drv->states[0].flags = CPUIDLE_FLAG_RCU_IDLE;
> drv->states[0].enter = psci_enter_idle_state;
> drv->states[0].exit_latency = 1;
> drv->states[0].target_residency = 1;
> --- a/drivers/cpuidle/cpuidle-qcom-spm.c
> +++ b/drivers/cpuidle/cpuidle-qcom-spm.c
> @@ -72,6 +72,7 @@ static struct cpuidle_driver qcom_spm_id
> .owner = THIS_MODULE,
> .states[0] = {
> .enter = spm_enter_idle_state,
> + .flags = CPUIDLE_FLAG_RCU_IDLE,
> .exit_latency = 1,
> .target_residency = 1,
> .power_usage = UINT_MAX,
> --- a/drivers/cpuidle/cpuidle-riscv-sbi.c
> +++ b/drivers/cpuidle/cpuidle-riscv-sbi.c
> @@ -332,6 +332,7 @@ static int sbi_cpuidle_init_cpu(struct d
> drv->cpumask = (struct cpumask *)cpumask_of(cpu);
>
> /* RISC-V architectural WFI to be represented as state index 0. */
> + drv->states[0].flags = CPUIDLE_FLAG_RCU_IDLE;
> drv->states[0].enter = sbi_cpuidle_enter_state;
> drv->states[0].exit_latency = 1;
> drv->states[0].target_residency = 1;
> --- a/drivers/cpuidle/dt_idle_states.c
> +++ b/drivers/cpuidle/dt_idle_states.c
> @@ -77,7 +77,7 @@ static int init_state_node(struct cpuidl
> if (err)
> desc = state_node->name;
>
> - idle_state->flags = 0;
> + idle_state->flags = CPUIDLE_FLAG_RCU_IDLE;
> if (of_property_read_bool(state_node, "local-timer-stop"))
> idle_state->flags |= CPUIDLE_FLAG_TIMER_STOP;
> /*
> --- a/include/linux/cpuidle.h
> +++ b/include/linux/cpuidle.h
> @@ -282,14 +282,18 @@ extern s64 cpuidle_governor_latency_req(
> int __ret = 0; \
> \
> if (!idx) { \
> + ct_idle_enter(); \
According to my comment above, we should then drop these calls to
ct_idle_enter and ct_idle_exit() here. Right?
> cpu_do_idle(); \
> + ct_idle_exit(); \
> return idx; \
> } \
> \
> if (!is_retention) \
> __ret = cpu_pm_enter(); \
> if (!__ret) { \
> + ct_idle_enter(); \
> __ret = low_level_idle_enter(state); \
> + ct_idle_exit(); \
> if (!is_retention) \
> cpu_pm_exit(); \
> } \
>
Kind regards
Uffe
Sorry; things keep getting in the way of finishing this :/
As such, I need a bit of time to get on-track again..
On Tue, Oct 04, 2022 at 01:03:57PM +0200, Ulf Hansson wrote:
> > --- a/drivers/acpi/processor_idle.c
> > +++ b/drivers/acpi/processor_idle.c
> > @@ -1200,6 +1200,8 @@ static int acpi_processor_setup_lpi_stat
> > state->target_residency = lpi->min_residency;
> > if (lpi->arch_flags)
> > state->flags |= CPUIDLE_FLAG_TIMER_STOP;
> > + if (lpi->entry_method == ACPI_CSTATE_FFH)
> > + state->flags |= CPUIDLE_FLAG_RCU_IDLE;
>
> I assume the state index here will never be 0?
>
> If not, it may lead to that acpi_processor_ffh_lpi_enter() may trigger
> CPU_PM_CPU_IDLE_ENTER_PARAM() to call ct_cpuidle_enter|exit() for an
> idle-state that doesn't have the CPUIDLE_FLAG_RCU_IDLE bit set.
I'm not quite sure I see how. AFAICT this condition above implies
acpi_processor_ffh_lpi_enter() gets called, no?
Which in turn is an unconditional __CPU_PM_CPU_IDLE_ENTER() user, so
even if idx==0, it ends up in ct_idle_{enter,exit}().
>
> > state->enter = acpi_idle_lpi_enter;
> > drv->safe_state_index = i;
> > }
> > --- a/drivers/cpuidle/cpuidle-arm.c
> > +++ b/drivers/cpuidle/cpuidle-arm.c
> > @@ -53,6 +53,7 @@ static struct cpuidle_driver arm_idle_dr
> > * handler for idle state index 0.
> > */
> > .states[0] = {
> > + .flags = CPUIDLE_FLAG_RCU_IDLE,
>
> Comparing arm64 and arm32 idle-states/idle-drivers, the $subject
> series ends up setting the CPUIDLE_FLAG_RCU_IDLE for the ARM WFI idle
> state (state zero), but only for the arm64 and psci cases (mostly
> arm64). For arm32 we would need to update the ARM_CPUIDLE_WFI_STATE
> too, as that is what most arm32 idle-drivers are using. My point is,
> the code becomes a bit inconsistent.
True.
> Perhaps it's easier to avoid setting the CPUIDLE_FLAG_RCU_IDLE bit for
> all of the ARM WFI idle states, for both arm64 and arm32?
As per the below?
>
> > .enter = arm_enter_idle_state,
> > .exit_latency = 1,
> > .target_residency = 1,
> > --- a/include/linux/cpuidle.h
> > +++ b/include/linux/cpuidle.h
> > @@ -282,14 +282,18 @@ extern s64 cpuidle_governor_latency_req(
> > int __ret = 0; \
> > \
> > if (!idx) { \
> > + ct_idle_enter(); \
>
> According to my comment above, we should then drop these calls to
> ct_idle_enter and ct_idle_exit() here. Right?
Yes, if we ensure idx==0 never has RCU_IDLE set then these must be
removed.
> > cpu_do_idle(); \
> > + ct_idle_exit(); \
> > return idx; \
> > } \
> > \
> > if (!is_retention) \
> > __ret = cpu_pm_enter(); \
> > if (!__ret) { \
> > + ct_idle_enter(); \
> > __ret = low_level_idle_enter(state); \
> > + ct_idle_exit(); \
> > if (!is_retention) \
> > cpu_pm_exit(); \
> > } \
> >
So the basic premise is that everything that needs RCU inside the idle
callback must set CPUIDLE_FLAG_RCU_IDLE and by doing that promise to
call ct_idle_{enter,exit}() themselves.
Setting RCU_IDLE is required when there is RCU usage, however even if
there is no RCU usage, setting RCU_IDLE is fine, as long as
ct_idle_{enter,exit}() then get called.
So does the below (delta) look better to you?
--- a/drivers/acpi/processor_idle.c
+++ b/drivers/acpi/processor_idle.c
@@ -1218,7 +1218,7 @@ static int acpi_processor_setup_lpi_stat
state->target_residency = lpi->min_residency;
if (lpi->arch_flags)
state->flags |= CPUIDLE_FLAG_TIMER_STOP;
- if (lpi->entry_method == ACPI_CSTATE_FFH)
+ if (i != 0 && lpi->entry_method == ACPI_CSTATE_FFH)
state->flags |= CPUIDLE_FLAG_RCU_IDLE;
state->enter = acpi_idle_lpi_enter;
drv->safe_state_index = i;
--- a/drivers/cpuidle/cpuidle-arm.c
+++ b/drivers/cpuidle/cpuidle-arm.c
@@ -53,7 +53,7 @@ static struct cpuidle_driver arm_idle_dr
* handler for idle state index 0.
*/
.states[0] = {
- .flags = CPUIDLE_FLAG_RCU_IDLE,
+ .flags = 0,
.enter = arm_enter_idle_state,
.exit_latency = 1,
.target_residency = 1,
--- a/drivers/cpuidle/cpuidle-psci.c
+++ b/drivers/cpuidle/cpuidle-psci.c
@@ -357,7 +357,7 @@ static int psci_idle_init_cpu(struct dev
* PSCI idle states relies on architectural WFI to be represented as
* state index 0.
*/
- drv->states[0].flags = CPUIDLE_FLAG_RCU_IDLE;
+ drv->states[0].flags = 0;
drv->states[0].enter = psci_enter_idle_state;
drv->states[0].exit_latency = 1;
drv->states[0].target_residency = 1;
--- a/drivers/cpuidle/cpuidle-qcom-spm.c
+++ b/drivers/cpuidle/cpuidle-qcom-spm.c
@@ -72,7 +72,7 @@ static struct cpuidle_driver qcom_spm_id
.owner = THIS_MODULE,
.states[0] = {
.enter = spm_enter_idle_state,
- .flags = CPUIDLE_FLAG_RCU_IDLE,
+ .flags = 0,
.exit_latency = 1,
.target_residency = 1,
.power_usage = UINT_MAX,
--- a/drivers/cpuidle/cpuidle-riscv-sbi.c
+++ b/drivers/cpuidle/cpuidle-riscv-sbi.c
@@ -337,7 +337,7 @@ static int sbi_cpuidle_init_cpu(struct d
drv->cpumask = (struct cpumask *)cpumask_of(cpu);
/* RISC-V architectural WFI to be represented as state index 0. */
- drv->states[0].flags = CPUIDLE_FLAG_RCU_IDLE;
+ drv->states[0].flags = 0;
drv->states[0].enter = sbi_cpuidle_enter_state;
drv->states[0].exit_latency = 1;
drv->states[0].target_residency = 1;
--- a/include/linux/cpuidle.h
+++ b/include/linux/cpuidle.h
@@ -282,9 +282,7 @@ extern s64 cpuidle_governor_latency_req(
int __ret = 0; \
\
if (!idx) { \
- ct_idle_enter(); \
cpu_do_idle(); \
- ct_idle_exit(); \
return idx; \
} \
\
On Wed, 16 Nov 2022 at 16:29, Peter Zijlstra <peterz@infradead.org> wrote:
>
>
> Sorry; things keep getting in the way of finishing this :/
>
> As such, I need a bit of time to get on-track again..
>
> On Tue, Oct 04, 2022 at 01:03:57PM +0200, Ulf Hansson wrote:
>
> > > --- a/drivers/acpi/processor_idle.c
> > > +++ b/drivers/acpi/processor_idle.c
> > > @@ -1200,6 +1200,8 @@ static int acpi_processor_setup_lpi_stat
> > > state->target_residency = lpi->min_residency;
> > > if (lpi->arch_flags)
> > > state->flags |= CPUIDLE_FLAG_TIMER_STOP;
> > > + if (lpi->entry_method == ACPI_CSTATE_FFH)
> > > + state->flags |= CPUIDLE_FLAG_RCU_IDLE;
> >
> > I assume the state index here will never be 0?
> >
> > If not, it may lead to that acpi_processor_ffh_lpi_enter() may trigger
> > CPU_PM_CPU_IDLE_ENTER_PARAM() to call ct_cpuidle_enter|exit() for an
> > idle-state that doesn't have the CPUIDLE_FLAG_RCU_IDLE bit set.
>
> I'm not quite sure I see how. AFAICT this condition above implies
> acpi_processor_ffh_lpi_enter() gets called, no?
>
> Which in turn is an unconditional __CPU_PM_CPU_IDLE_ENTER() user, so
> even if idx==0, it ends up in ct_idle_{enter,exit}().
Seems like I was overlooking something here, you are right, this
shouldn't really be a problem.
>
> >
> > > state->enter = acpi_idle_lpi_enter;
> > > drv->safe_state_index = i;
> > > }
> > > --- a/drivers/cpuidle/cpuidle-arm.c
> > > +++ b/drivers/cpuidle/cpuidle-arm.c
> > > @@ -53,6 +53,7 @@ static struct cpuidle_driver arm_idle_dr
> > > * handler for idle state index 0.
> > > */
> > > .states[0] = {
> > > + .flags = CPUIDLE_FLAG_RCU_IDLE,
> >
> > Comparing arm64 and arm32 idle-states/idle-drivers, the $subject
> > series ends up setting the CPUIDLE_FLAG_RCU_IDLE for the ARM WFI idle
> > state (state zero), but only for the arm64 and psci cases (mostly
> > arm64). For arm32 we would need to update the ARM_CPUIDLE_WFI_STATE
> > too, as that is what most arm32 idle-drivers are using. My point is,
> > the code becomes a bit inconsistent.
>
> True.
>
> > Perhaps it's easier to avoid setting the CPUIDLE_FLAG_RCU_IDLE bit for
> > all of the ARM WFI idle states, for both arm64 and arm32?
>
> As per the below?
>
> >
> > > .enter = arm_enter_idle_state,
> > > .exit_latency = 1,
> > > .target_residency = 1,
>
> > > --- a/include/linux/cpuidle.h
> > > +++ b/include/linux/cpuidle.h
> > > @@ -282,14 +282,18 @@ extern s64 cpuidle_governor_latency_req(
> > > int __ret = 0; \
> > > \
> > > if (!idx) { \
> > > + ct_idle_enter(); \
> >
> > According to my comment above, we should then drop these calls to
> > ct_idle_enter and ct_idle_exit() here. Right?
>
> Yes, if we ensure idx==0 never has RCU_IDLE set then these must be
> removed.
>
> > > cpu_do_idle(); \
> > > + ct_idle_exit(); \
> > > return idx; \
> > > } \
> > > \
> > > if (!is_retention) \
> > > __ret = cpu_pm_enter(); \
> > > if (!__ret) { \
> > > + ct_idle_enter(); \
> > > __ret = low_level_idle_enter(state); \
> > > + ct_idle_exit(); \
> > > if (!is_retention) \
> > > cpu_pm_exit(); \
> > > } \
> > >
>
> So the basic premise is that everything that needs RCU inside the idle
> callback must set CPUIDLE_FLAG_RCU_IDLE and by doing that promise to
> call ct_idle_{enter,exit}() themselves.
>
> Setting RCU_IDLE is required when there is RCU usage, however even if
> there is no RCU usage, setting RCU_IDLE is fine, as long as
> ct_idle_{enter,exit}() then get called.
Right, I was thinking that it could make sense to shrink the window
for users getting this wrong. In other words, we shouldn't set the
CPUIDLE_FLAG_RCU_IDLE unless we really need to.
And as I said, consistent behaviour is also nice to have.
>
>
> So does the below (delta) look better to you?
Yes, it does!
Although, one minor comment below.
>
> --- a/drivers/acpi/processor_idle.c
> +++ b/drivers/acpi/processor_idle.c
> @@ -1218,7 +1218,7 @@ static int acpi_processor_setup_lpi_stat
> state->target_residency = lpi->min_residency;
> if (lpi->arch_flags)
> state->flags |= CPUIDLE_FLAG_TIMER_STOP;
> - if (lpi->entry_method == ACPI_CSTATE_FFH)
> + if (i != 0 && lpi->entry_method == ACPI_CSTATE_FFH)
> state->flags |= CPUIDLE_FLAG_RCU_IDLE;
> state->enter = acpi_idle_lpi_enter;
> drv->safe_state_index = i;
> --- a/drivers/cpuidle/cpuidle-arm.c
> +++ b/drivers/cpuidle/cpuidle-arm.c
> @@ -53,7 +53,7 @@ static struct cpuidle_driver arm_idle_dr
> * handler for idle state index 0.
> */
> .states[0] = {
> - .flags = CPUIDLE_FLAG_RCU_IDLE,
> + .flags = 0,
Nitpick: I don't think we need to explicitly clear the flag, as it
should already be zeroed by the compiler from its static declaration.
Right?
> .enter = arm_enter_idle_state,
> .exit_latency = 1,
> .target_residency = 1,
> --- a/drivers/cpuidle/cpuidle-psci.c
> +++ b/drivers/cpuidle/cpuidle-psci.c
> @@ -357,7 +357,7 @@ static int psci_idle_init_cpu(struct dev
> * PSCI idle states relies on architectural WFI to be represented as
> * state index 0.
> */
> - drv->states[0].flags = CPUIDLE_FLAG_RCU_IDLE;
> + drv->states[0].flags = 0;
> drv->states[0].enter = psci_enter_idle_state;
> drv->states[0].exit_latency = 1;
> drv->states[0].target_residency = 1;
> --- a/drivers/cpuidle/cpuidle-qcom-spm.c
> +++ b/drivers/cpuidle/cpuidle-qcom-spm.c
> @@ -72,7 +72,7 @@ static struct cpuidle_driver qcom_spm_id
> .owner = THIS_MODULE,
> .states[0] = {
> .enter = spm_enter_idle_state,
> - .flags = CPUIDLE_FLAG_RCU_IDLE,
> + .flags = 0,
> .exit_latency = 1,
> .target_residency = 1,
> .power_usage = UINT_MAX,
> --- a/drivers/cpuidle/cpuidle-riscv-sbi.c
> +++ b/drivers/cpuidle/cpuidle-riscv-sbi.c
> @@ -337,7 +337,7 @@ static int sbi_cpuidle_init_cpu(struct d
> drv->cpumask = (struct cpumask *)cpumask_of(cpu);
>
> /* RISC-V architectural WFI to be represented as state index 0. */
> - drv->states[0].flags = CPUIDLE_FLAG_RCU_IDLE;
> + drv->states[0].flags = 0;
> drv->states[0].enter = sbi_cpuidle_enter_state;
> drv->states[0].exit_latency = 1;
> drv->states[0].target_residency = 1;
> --- a/include/linux/cpuidle.h
> +++ b/include/linux/cpuidle.h
> @@ -282,9 +282,7 @@ extern s64 cpuidle_governor_latency_req(
> int __ret = 0; \
> \
> if (!idx) { \
> - ct_idle_enter(); \
> cpu_do_idle(); \
> - ct_idle_exit(); \
> return idx; \
> } \
> \
Kind regards
Uffe
On Tue, 4 Oct 2022 at 13:03, Ulf Hansson <ulf.hansson@linaro.org> wrote:
>
> On Mon, 19 Sept 2022 at 12:18, Peter Zijlstra <peterz@infradead.org> wrote:
> >
> > Doing RCU-idle outside the driver, only to then temporarily enable it
> > again before going idle is daft.
> >
> > Notably: this converts all dt_init_idle_driver() and
> > __CPU_PM_CPU_IDLE_ENTER() users for they are inextrably intertwined.
> >
> > Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
>
> Reviewed-by: Ulf Hansson <ulf.hansson@linaro.org>
This was not (yet) my intention. Please have a look at the comments I
provided below.
Kind regards
Uffe
>
> > ---
> > arch/arm/mach-omap2/cpuidle34xx.c | 4 ++--
> > drivers/acpi/processor_idle.c | 2 ++
> > drivers/cpuidle/cpuidle-arm.c | 1 +
> > drivers/cpuidle/cpuidle-big_little.c | 8 ++++++--
> > drivers/cpuidle/cpuidle-psci.c | 1 +
> > drivers/cpuidle/cpuidle-qcom-spm.c | 1 +
> > drivers/cpuidle/cpuidle-riscv-sbi.c | 1 +
> > drivers/cpuidle/dt_idle_states.c | 2 +-
> > include/linux/cpuidle.h | 4 ++++
> > 9 files changed, 19 insertions(+), 5 deletions(-)
> >
> > --- a/drivers/acpi/processor_idle.c
> > +++ b/drivers/acpi/processor_idle.c
> > @@ -1200,6 +1200,8 @@ static int acpi_processor_setup_lpi_stat
> > state->target_residency = lpi->min_residency;
> > if (lpi->arch_flags)
> > state->flags |= CPUIDLE_FLAG_TIMER_STOP;
> > + if (lpi->entry_method == ACPI_CSTATE_FFH)
> > + state->flags |= CPUIDLE_FLAG_RCU_IDLE;
>
> I assume the state index here will never be 0?
>
> If not, it may lead to that acpi_processor_ffh_lpi_enter() may trigger
> CPU_PM_CPU_IDLE_ENTER_PARAM() to call ct_cpuidle_enter|exit() for an
> idle-state that doesn't have the CPUIDLE_FLAG_RCU_IDLE bit set.
>
> > state->enter = acpi_idle_lpi_enter;
> > drv->safe_state_index = i;
> > }
> > --- a/drivers/cpuidle/cpuidle-arm.c
> > +++ b/drivers/cpuidle/cpuidle-arm.c
> > @@ -53,6 +53,7 @@ static struct cpuidle_driver arm_idle_dr
> > * handler for idle state index 0.
> > */
> > .states[0] = {
> > + .flags = CPUIDLE_FLAG_RCU_IDLE,
>
> Comparing arm64 and arm32 idle-states/idle-drivers, the $subject
> series ends up setting the CPUIDLE_FLAG_RCU_IDLE for the ARM WFI idle
> state (state zero), but only for the arm64 and psci cases (mostly
> arm64). For arm32 we would need to update the ARM_CPUIDLE_WFI_STATE
> too, as that is what most arm32 idle-drivers are using. My point is,
> the code becomes a bit inconsistent.
>
> Perhaps it's easier to avoid setting the CPUIDLE_FLAG_RCU_IDLE bit for
> all of the ARM WFI idle states, for both arm64 and arm32?
>
> > .enter = arm_enter_idle_state,
> > .exit_latency = 1,
> > .target_residency = 1,
> > --- a/drivers/cpuidle/cpuidle-big_little.c
> > +++ b/drivers/cpuidle/cpuidle-big_little.c
> > @@ -64,7 +64,8 @@ static struct cpuidle_driver bl_idle_lit
> > .enter = bl_enter_powerdown,
> > .exit_latency = 700,
> > .target_residency = 2500,
> > - .flags = CPUIDLE_FLAG_TIMER_STOP,
> > + .flags = CPUIDLE_FLAG_TIMER_STOP |
> > + CPUIDLE_FLAG_RCU_IDLE,
> > .name = "C1",
> > .desc = "ARM little-cluster power down",
> > },
> > @@ -85,7 +86,8 @@ static struct cpuidle_driver bl_idle_big
> > .enter = bl_enter_powerdown,
> > .exit_latency = 500,
> > .target_residency = 2000,
> > - .flags = CPUIDLE_FLAG_TIMER_STOP,
> > + .flags = CPUIDLE_FLAG_TIMER_STOP |
> > + CPUIDLE_FLAG_RCU_IDLE,
> > .name = "C1",
> > .desc = "ARM big-cluster power down",
> > },
> > @@ -124,11 +126,13 @@ static int bl_enter_powerdown(struct cpu
> > struct cpuidle_driver *drv, int idx)
> > {
> > cpu_pm_enter();
> > + ct_idle_enter();
> >
> > cpu_suspend(0, bl_powerdown_finisher);
> >
> > /* signals the MCPM core that CPU is out of low power state */
> > mcpm_cpu_powered_up();
> > + ct_idle_exit();
> >
> > cpu_pm_exit();
> >
> > --- a/drivers/cpuidle/cpuidle-psci.c
> > +++ b/drivers/cpuidle/cpuidle-psci.c
> > @@ -357,6 +357,7 @@ static int psci_idle_init_cpu(struct dev
> > * PSCI idle states relies on architectural WFI to be represented as
> > * state index 0.
> > */
> > + drv->states[0].flags = CPUIDLE_FLAG_RCU_IDLE;
> > drv->states[0].enter = psci_enter_idle_state;
> > drv->states[0].exit_latency = 1;
> > drv->states[0].target_residency = 1;
> > --- a/drivers/cpuidle/cpuidle-qcom-spm.c
> > +++ b/drivers/cpuidle/cpuidle-qcom-spm.c
> > @@ -72,6 +72,7 @@ static struct cpuidle_driver qcom_spm_id
> > .owner = THIS_MODULE,
> > .states[0] = {
> > .enter = spm_enter_idle_state,
> > + .flags = CPUIDLE_FLAG_RCU_IDLE,
> > .exit_latency = 1,
> > .target_residency = 1,
> > .power_usage = UINT_MAX,
> > --- a/drivers/cpuidle/cpuidle-riscv-sbi.c
> > +++ b/drivers/cpuidle/cpuidle-riscv-sbi.c
> > @@ -332,6 +332,7 @@ static int sbi_cpuidle_init_cpu(struct d
> > drv->cpumask = (struct cpumask *)cpumask_of(cpu);
> >
> > /* RISC-V architectural WFI to be represented as state index 0. */
> > + drv->states[0].flags = CPUIDLE_FLAG_RCU_IDLE;
> > drv->states[0].enter = sbi_cpuidle_enter_state;
> > drv->states[0].exit_latency = 1;
> > drv->states[0].target_residency = 1;
> > --- a/drivers/cpuidle/dt_idle_states.c
> > +++ b/drivers/cpuidle/dt_idle_states.c
> > @@ -77,7 +77,7 @@ static int init_state_node(struct cpuidl
> > if (err)
> > desc = state_node->name;
> >
> > - idle_state->flags = 0;
> > + idle_state->flags = CPUIDLE_FLAG_RCU_IDLE;
> > if (of_property_read_bool(state_node, "local-timer-stop"))
> > idle_state->flags |= CPUIDLE_FLAG_TIMER_STOP;
> > /*
> > --- a/include/linux/cpuidle.h
> > +++ b/include/linux/cpuidle.h
> > @@ -282,14 +282,18 @@ extern s64 cpuidle_governor_latency_req(
> > int __ret = 0; \
> > \
> > if (!idx) { \
> > + ct_idle_enter(); \
>
> According to my comment above, we should then drop these calls to
> ct_idle_enter and ct_idle_exit() here. Right?
>
> > cpu_do_idle(); \
> > + ct_idle_exit(); \
> > return idx; \
> > } \
> > \
> > if (!is_retention) \
> > __ret = cpu_pm_enter(); \
> > if (!__ret) { \
> > + ct_idle_enter(); \
> > __ret = low_level_idle_enter(state); \
> > + ct_idle_exit(); \
> > if (!is_retention) \
> > cpu_pm_exit(); \
> > } \
> >
>
> Kind regards
> Uffe
© 2016 - 2026 Red Hat, Inc.