| Message ID | 20220919101521.139727471@infradead.org (mailing list archive) |
|---|---|
| State | Handled Elsewhere, archived |
| Headers | show |
| Series | cpuidle,rcu: Clean up the mess | expand |
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
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
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
--- 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(); \ } \
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(-)