CFS bandwidth limits and NOHZ full don't play well together. Tasks
can easily run well past their quotas before a remote tick does
accounting. This leads to long, multi-period stalls before such
tasks can run again. Currently, when presented with these conflicting
requirements the scheduler is favoring nohz_full and letting the tick
be stopped. However, nohz tick stopping is already best-effort, there
are a number of conditions that can prevent it, whereas cfs runtime
bandwidth is expected to be enforced.
Make the scheduler favor bandwidth over stopping the tick by setting
TICK_DEP_BIT_SCHED when the only running task is a cfs task with
runtime limit enabled. We use cfs_b->hierarchical_quota to
determine if the task requires the tick.
Add check in pick_next_task_fair() as well since that is where
we have a handle on the task that is actually going to be running.
Add check in sched_can_stop_tick() to cover some edge cases such
as nr_running going from 2->1 and the 1 remains the running task.
Add sched_feat HZ_BW (off by default) to control the tick_stop
behavior.
Signed-off-by: Phil Auld <pauld@redhat.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Vincent Guittot <vincent.guittot@linaro.org>
Cc: Juri Lelli <juri.lelli@redhat.com>
Cc: Dietmar Eggemann <dietmar.eggemann@arm.com>
Cc: Valentin Schneider <vschneid@redhat.com>
Cc: Ben Segall <bsegall@google.com>
Cc: Frederic Weisbecker <frederic@kernel.org>
---
v6: restore check for fair_sched_class
v5: Reworked checks to use newly-fixed cfs_b->hierarchical_quota to
check for bw
constraints.
v4: Made checks for runtime_enabled hierarchical.
v3: Moved sched_cfs_bandwidth_active() prototype to sched.h outside of
guards to
silence -Wmissing-prototypes.
v2: Ben pointed out that the bit could get cleared in the dequeue path
if we migrate a newly enqueued task without preempting curr. Added a
check for that edge case to sched_can_stop_tick. Removed the call to
sched_can_stop_tick from sched_fair_update_stop_tick since it was
redundant.
kernel/sched/core.c | 26 ++++++++++++++++++++++
kernel/sched/fair.c | 49 +++++++++++++++++++++++++++++++++++++++++
kernel/sched/features.h | 2 ++
kernel/sched/sched.h | 1 +
4 files changed, 78 insertions(+)
diff --git a/kernel/sched/core.c b/kernel/sched/core.c
index f80697a79baf..8a2ed4c0b709 100644
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -1194,6 +1194,20 @@ static void nohz_csd_func(void *info)
#endif /* CONFIG_NO_HZ_COMMON */
#ifdef CONFIG_NO_HZ_FULL
+static inline bool __need_bw_check(struct rq *rq, struct task_struct *p)
+{
+ if (rq->nr_running != 1)
+ return false;
+
+ if (p->sched_class != &fair_sched_class)
+ return false;
+
+ if (!task_on_rq_queued(p))
+ return false;
+
+ return true;
+}
+
bool sched_can_stop_tick(struct rq *rq)
{
int fifo_nr_running;
@@ -1229,6 +1243,18 @@ bool sched_can_stop_tick(struct rq *rq)
if (rq->nr_running > 1)
return false;
+ /*
+ * If there is one task and it has CFS runtime bandwidth constraints
+ * and it's on the cpu now we don't want to stop the tick.
+ * This check prevents clearing the bit if a newly enqueued task here is
+ * dequeued by migrating while the constrained task continues to run.
+ * E.g. going from 2->1 without going through pick_next_task().
+ */
+ if (sched_feat(HZ_BW) && __need_bw_check(rq, rq->curr)) {
+ if (cfs_task_bw_constrained(rq->curr))
+ return false;
+ }
+
return true;
}
#endif /* CONFIG_NO_HZ_FULL */
diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index d9b3d4617e16..acd9f317aad1 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -6140,6 +6140,46 @@ static void __maybe_unused unthrottle_offline_cfs_rqs(struct rq *rq)
rcu_read_unlock();
}
+bool cfs_task_bw_constrained(struct task_struct *p)
+{
+ struct cfs_rq *cfs_rq = task_cfs_rq(p);
+
+ if (!cfs_bandwidth_used())
+ return false;
+
+ if (cfs_rq->runtime_enabled ||
+ tg_cfs_bandwidth(cfs_rq->tg)->hierarchical_quota != RUNTIME_INF)
+ return true;
+
+ return false;
+}
+
+#ifdef CONFIG_NO_HZ_FULL
+/* called from pick_next_task_fair() */
+static void sched_fair_update_stop_tick(struct rq *rq, struct task_struct *p)
+{
+ int cpu = cpu_of(rq);
+
+ if (!sched_feat(HZ_BW) || !cfs_bandwidth_used())
+ return;
+
+ if (!tick_nohz_full_cpu(cpu))
+ return;
+
+ if (rq->nr_running != 1)
+ return;
+
+ /*
+ * We know there is only one task runnable and we've just picked it. The
+ * normal enqueue path will have cleared TICK_DEP_BIT_SCHED if we will
+ * be otherwise able to stop the tick. Just need to check if we are using
+ * bandwidth control.
+ */
+ if (cfs_task_bw_constrained(p))
+ tick_nohz_dep_set_cpu(cpu, TICK_DEP_BIT_SCHED);
+}
+#endif
+
#else /* CONFIG_CFS_BANDWIDTH */
static inline bool cfs_bandwidth_used(void)
@@ -6182,9 +6222,17 @@ static inline struct cfs_bandwidth *tg_cfs_bandwidth(struct task_group *tg)
static inline void destroy_cfs_bandwidth(struct cfs_bandwidth *cfs_b) {}
static inline void update_runtime_enabled(struct rq *rq) {}
static inline void unthrottle_offline_cfs_rqs(struct rq *rq) {}
+bool cfs_task_bw_constrained(struct task_struct *p)
+{
+ return false;
+}
#endif /* CONFIG_CFS_BANDWIDTH */
+#if !defined(CONFIG_CFS_BANDWIDTH) || !defined(CONFIG_NO_HZ_FULL)
+static inline void sched_fair_update_stop_tick(struct rq *rq, struct task_struct *p) {}
+#endif
+
/**************************************************
* CFS operations on tasks:
*/
@@ -8098,6 +8146,7 @@ done: __maybe_unused;
hrtick_start_fair(rq, p);
update_misfit_status(p, rq);
+ sched_fair_update_stop_tick(rq, p);
return p;
diff --git a/kernel/sched/features.h b/kernel/sched/features.h
index ee7f23c76bd3..6fdf1fdf6b17 100644
--- a/kernel/sched/features.h
+++ b/kernel/sched/features.h
@@ -101,3 +101,5 @@ SCHED_FEAT(LATENCY_WARN, false)
SCHED_FEAT(ALT_PERIOD, true)
SCHED_FEAT(BASE_SLICE, true)
+
+SCHED_FEAT(HZ_BW, false)
diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h
index 63822c9238cc..d6d346bc78aa 100644
--- a/kernel/sched/sched.h
+++ b/kernel/sched/sched.h
@@ -465,6 +465,7 @@ extern void init_cfs_bandwidth(struct cfs_bandwidth *cfs_b, struct cfs_bandwidth
extern void __refill_cfs_bandwidth_runtime(struct cfs_bandwidth *cfs_b);
extern void start_cfs_bandwidth(struct cfs_bandwidth *cfs_b);
extern void unthrottle_cfs_rq(struct cfs_rq *cfs_rq);
+extern bool cfs_task_bw_constrained(struct task_struct *p);
extern void init_tg_rt_entry(struct task_group *tg, struct rt_rq *rt_rq,
struct sched_rt_entity *rt_se, int cpu,
--
2.31.1On Wed, Jul 12, 2023 at 09:33:57AM -0400, Phil Auld wrote: > CFS bandwidth limits and NOHZ full don't play well together. Tasks > can easily run well past their quotas before a remote tick does > accounting. This leads to long, multi-period stalls before such > tasks can run again. Currently, when presented with these conflicting > requirements the scheduler is favoring nohz_full and letting the tick > be stopped. However, nohz tick stopping is already best-effort, there > are a number of conditions that can prevent it, whereas cfs runtime > bandwidth is expected to be enforced. > > Make the scheduler favor bandwidth over stopping the tick by setting > TICK_DEP_BIT_SCHED when the only running task is a cfs task with > runtime limit enabled. We use cfs_b->hierarchical_quota to > determine if the task requires the tick. > > Add check in pick_next_task_fair() as well since that is where > we have a handle on the task that is actually going to be running. > > Add check in sched_can_stop_tick() to cover some edge cases such > as nr_running going from 2->1 and the 1 remains the running task. These appear fine to me, except: > Add sched_feat HZ_BW (off by default) to control the tick_stop > behavior. What was the thinking here? This means nobody will be using this -- why would you want this default disabled?
On Tue, Aug 01, 2023 at 12:49:34AM +0200 Peter Zijlstra wrote: > On Wed, Jul 12, 2023 at 09:33:57AM -0400, Phil Auld wrote: > > CFS bandwidth limits and NOHZ full don't play well together. Tasks > > can easily run well past their quotas before a remote tick does > > accounting. This leads to long, multi-period stalls before such > > tasks can run again. Currently, when presented with these conflicting > > requirements the scheduler is favoring nohz_full and letting the tick > > be stopped. However, nohz tick stopping is already best-effort, there > > are a number of conditions that can prevent it, whereas cfs runtime > > bandwidth is expected to be enforced. > > > > Make the scheduler favor bandwidth over stopping the tick by setting > > TICK_DEP_BIT_SCHED when the only running task is a cfs task with > > runtime limit enabled. We use cfs_b->hierarchical_quota to > > determine if the task requires the tick. > > > > Add check in pick_next_task_fair() as well since that is where > > we have a handle on the task that is actually going to be running. > > > > Add check in sched_can_stop_tick() to cover some edge cases such > > as nr_running going from 2->1 and the 1 remains the running task. > > These appear fine to me, except: > > > Add sched_feat HZ_BW (off by default) to control the tick_stop > > behavior. > > What was the thinking here? This means nobody will be using this -- why > would you want this default disabled? > That was just a hedge in case it caused issues. I'd probably have had to enable it in RHEL anyway. Using a feature was to make it inocuous when disabled. Would you prefer me to enable it or remove the sched_feat entirely? (or do you want to just switch that to true when you apply it?) Thanks, Phil --
On Tue, Aug 01, 2023 at 07:13:42AM -0400, Phil Auld wrote: > On Tue, Aug 01, 2023 at 12:49:34AM +0200 Peter Zijlstra wrote: > > On Wed, Jul 12, 2023 at 09:33:57AM -0400, Phil Auld wrote: > > > CFS bandwidth limits and NOHZ full don't play well together. Tasks > > > can easily run well past their quotas before a remote tick does > > > accounting. This leads to long, multi-period stalls before such > > > tasks can run again. Currently, when presented with these conflicting > > > requirements the scheduler is favoring nohz_full and letting the tick > > > be stopped. However, nohz tick stopping is already best-effort, there > > > are a number of conditions that can prevent it, whereas cfs runtime > > > bandwidth is expected to be enforced. > > > > > > Make the scheduler favor bandwidth over stopping the tick by setting > > > TICK_DEP_BIT_SCHED when the only running task is a cfs task with > > > runtime limit enabled. We use cfs_b->hierarchical_quota to > > > determine if the task requires the tick. > > > > > > Add check in pick_next_task_fair() as well since that is where > > > we have a handle on the task that is actually going to be running. > > > > > > Add check in sched_can_stop_tick() to cover some edge cases such > > > as nr_running going from 2->1 and the 1 remains the running task. > > > > These appear fine to me, except: > > > > > Add sched_feat HZ_BW (off by default) to control the tick_stop > > > behavior. > > > > What was the thinking here? This means nobody will be using this -- why > > would you want this default disabled? > > > > That was just a hedge in case it caused issues. I'd probably have had to > enable it in RHEL anyway. Using a feature was to make it inocuous when > disabled. Would you prefer me to enable it or remove the sched_feat > entirely? (or do you want to just switch that to true when you apply it?) I've edited it to default enabled -- we can pull the feature flag eventually I suppose. Things didn't readily apply, so I've kicked at it a little. Should be in queue/sched/core for the robots to chew on.
On Tue, Aug 01, 2023 at 05:37:31PM +0200 Peter Zijlstra wrote: > On Tue, Aug 01, 2023 at 07:13:42AM -0400, Phil Auld wrote: > > On Tue, Aug 01, 2023 at 12:49:34AM +0200 Peter Zijlstra wrote: > > > On Wed, Jul 12, 2023 at 09:33:57AM -0400, Phil Auld wrote: > > > > CFS bandwidth limits and NOHZ full don't play well together. Tasks > > > > can easily run well past their quotas before a remote tick does > > > > accounting. This leads to long, multi-period stalls before such > > > > tasks can run again. Currently, when presented with these conflicting > > > > requirements the scheduler is favoring nohz_full and letting the tick > > > > be stopped. However, nohz tick stopping is already best-effort, there > > > > are a number of conditions that can prevent it, whereas cfs runtime > > > > bandwidth is expected to be enforced. > > > > > > > > Make the scheduler favor bandwidth over stopping the tick by setting > > > > TICK_DEP_BIT_SCHED when the only running task is a cfs task with > > > > runtime limit enabled. We use cfs_b->hierarchical_quota to > > > > determine if the task requires the tick. > > > > > > > > Add check in pick_next_task_fair() as well since that is where > > > > we have a handle on the task that is actually going to be running. > > > > > > > > Add check in sched_can_stop_tick() to cover some edge cases such > > > > as nr_running going from 2->1 and the 1 remains the running task. > > > > > > These appear fine to me, except: > > > > > > > Add sched_feat HZ_BW (off by default) to control the tick_stop > > > > behavior. > > > > > > What was the thinking here? This means nobody will be using this -- why > > > would you want this default disabled? > > > > > > > That was just a hedge in case it caused issues. I'd probably have had to > > enable it in RHEL anyway. Using a feature was to make it inocuous when > > disabled. Would you prefer me to enable it or remove the sched_feat > > entirely? (or do you want to just switch that to true when you apply it?) > > I've edited it to default enabled -- we can pull the feature flag > eventually I suppose. > > Things didn't readily apply, so I've kicked at it a little. Should be in > queue/sched/core for the robots to chew on. > Or "choke on" as the case may be :) I sent you something that will hopefully clean that up. Thanks! Cheers, Phil --
Phil Auld <pauld@redhat.com> writes:
> CFS bandwidth limits and NOHZ full don't play well together. Tasks
> can easily run well past their quotas before a remote tick does
> accounting. This leads to long, multi-period stalls before such
> tasks can run again. Currently, when presented with these conflicting
> requirements the scheduler is favoring nohz_full and letting the tick
> be stopped. However, nohz tick stopping is already best-effort, there
> are a number of conditions that can prevent it, whereas cfs runtime
> bandwidth is expected to be enforced.
>
> Make the scheduler favor bandwidth over stopping the tick by setting
> TICK_DEP_BIT_SCHED when the only running task is a cfs task with
> runtime limit enabled. We use cfs_b->hierarchical_quota to
> determine if the task requires the tick.
>
> Add check in pick_next_task_fair() as well since that is where
> we have a handle on the task that is actually going to be running.
>
> Add check in sched_can_stop_tick() to cover some edge cases such
> as nr_running going from 2->1 and the 1 remains the running task.
>
> Add sched_feat HZ_BW (off by default) to control the tick_stop
> behavior.
I think this looks good now.
Reviewed-By: Ben Segall <bsegall@google.com>
>
> Signed-off-by: Phil Auld <pauld@redhat.com>
> Cc: Ingo Molnar <mingo@redhat.com>
> Cc: Peter Zijlstra <peterz@infradead.org>
> Cc: Vincent Guittot <vincent.guittot@linaro.org>
> Cc: Juri Lelli <juri.lelli@redhat.com>
> Cc: Dietmar Eggemann <dietmar.eggemann@arm.com>
> Cc: Valentin Schneider <vschneid@redhat.com>
> Cc: Ben Segall <bsegall@google.com>
> Cc: Frederic Weisbecker <frederic@kernel.org>
> ---
> v6: restore check for fair_sched_class
>
> v5: Reworked checks to use newly-fixed cfs_b->hierarchical_quota to
> check for bw
> constraints.
>
> v4: Made checks for runtime_enabled hierarchical.
>
> v3: Moved sched_cfs_bandwidth_active() prototype to sched.h outside of
> guards to
> silence -Wmissing-prototypes.
>
> v2: Ben pointed out that the bit could get cleared in the dequeue path
> if we migrate a newly enqueued task without preempting curr. Added a
> check for that edge case to sched_can_stop_tick. Removed the call to
> sched_can_stop_tick from sched_fair_update_stop_tick since it was
> redundant.
>
> kernel/sched/core.c | 26 ++++++++++++++++++++++
> kernel/sched/fair.c | 49 +++++++++++++++++++++++++++++++++++++++++
> kernel/sched/features.h | 2 ++
> kernel/sched/sched.h | 1 +
> 4 files changed, 78 insertions(+)
>
> diff --git a/kernel/sched/core.c b/kernel/sched/core.c
> index f80697a79baf..8a2ed4c0b709 100644
> --- a/kernel/sched/core.c
> +++ b/kernel/sched/core.c
> @@ -1194,6 +1194,20 @@ static void nohz_csd_func(void *info)
> #endif /* CONFIG_NO_HZ_COMMON */
>
> #ifdef CONFIG_NO_HZ_FULL
> +static inline bool __need_bw_check(struct rq *rq, struct task_struct *p)
> +{
> + if (rq->nr_running != 1)
> + return false;
> +
> + if (p->sched_class != &fair_sched_class)
> + return false;
> +
> + if (!task_on_rq_queued(p))
> + return false;
> +
> + return true;
> +}
> +
> bool sched_can_stop_tick(struct rq *rq)
> {
> int fifo_nr_running;
> @@ -1229,6 +1243,18 @@ bool sched_can_stop_tick(struct rq *rq)
> if (rq->nr_running > 1)
> return false;
>
> + /*
> + * If there is one task and it has CFS runtime bandwidth constraints
> + * and it's on the cpu now we don't want to stop the tick.
> + * This check prevents clearing the bit if a newly enqueued task here is
> + * dequeued by migrating while the constrained task continues to run.
> + * E.g. going from 2->1 without going through pick_next_task().
> + */
> + if (sched_feat(HZ_BW) && __need_bw_check(rq, rq->curr)) {
> + if (cfs_task_bw_constrained(rq->curr))
> + return false;
> + }
> +
> return true;
> }
> #endif /* CONFIG_NO_HZ_FULL */
> diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
> index d9b3d4617e16..acd9f317aad1 100644
> --- a/kernel/sched/fair.c
> +++ b/kernel/sched/fair.c
> @@ -6140,6 +6140,46 @@ static void __maybe_unused unthrottle_offline_cfs_rqs(struct rq *rq)
> rcu_read_unlock();
> }
>
> +bool cfs_task_bw_constrained(struct task_struct *p)
> +{
> + struct cfs_rq *cfs_rq = task_cfs_rq(p);
> +
> + if (!cfs_bandwidth_used())
> + return false;
> +
> + if (cfs_rq->runtime_enabled ||
> + tg_cfs_bandwidth(cfs_rq->tg)->hierarchical_quota != RUNTIME_INF)
> + return true;
> +
> + return false;
> +}
> +
> +#ifdef CONFIG_NO_HZ_FULL
> +/* called from pick_next_task_fair() */
> +static void sched_fair_update_stop_tick(struct rq *rq, struct task_struct *p)
> +{
> + int cpu = cpu_of(rq);
> +
> + if (!sched_feat(HZ_BW) || !cfs_bandwidth_used())
> + return;
> +
> + if (!tick_nohz_full_cpu(cpu))
> + return;
> +
> + if (rq->nr_running != 1)
> + return;
> +
> + /*
> + * We know there is only one task runnable and we've just picked it. The
> + * normal enqueue path will have cleared TICK_DEP_BIT_SCHED if we will
> + * be otherwise able to stop the tick. Just need to check if we are using
> + * bandwidth control.
> + */
> + if (cfs_task_bw_constrained(p))
> + tick_nohz_dep_set_cpu(cpu, TICK_DEP_BIT_SCHED);
> +}
> +#endif
> +
> #else /* CONFIG_CFS_BANDWIDTH */
>
> static inline bool cfs_bandwidth_used(void)
> @@ -6182,9 +6222,17 @@ static inline struct cfs_bandwidth *tg_cfs_bandwidth(struct task_group *tg)
> static inline void destroy_cfs_bandwidth(struct cfs_bandwidth *cfs_b) {}
> static inline void update_runtime_enabled(struct rq *rq) {}
> static inline void unthrottle_offline_cfs_rqs(struct rq *rq) {}
> +bool cfs_task_bw_constrained(struct task_struct *p)
> +{
> + return false;
> +}
>
> #endif /* CONFIG_CFS_BANDWIDTH */
>
> +#if !defined(CONFIG_CFS_BANDWIDTH) || !defined(CONFIG_NO_HZ_FULL)
> +static inline void sched_fair_update_stop_tick(struct rq *rq, struct task_struct *p) {}
> +#endif
> +
> /**************************************************
> * CFS operations on tasks:
> */
> @@ -8098,6 +8146,7 @@ done: __maybe_unused;
> hrtick_start_fair(rq, p);
>
> update_misfit_status(p, rq);
> + sched_fair_update_stop_tick(rq, p);
>
> return p;
>
> diff --git a/kernel/sched/features.h b/kernel/sched/features.h
> index ee7f23c76bd3..6fdf1fdf6b17 100644
> --- a/kernel/sched/features.h
> +++ b/kernel/sched/features.h
> @@ -101,3 +101,5 @@ SCHED_FEAT(LATENCY_WARN, false)
>
> SCHED_FEAT(ALT_PERIOD, true)
> SCHED_FEAT(BASE_SLICE, true)
> +
> +SCHED_FEAT(HZ_BW, false)
> diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h
> index 63822c9238cc..d6d346bc78aa 100644
> --- a/kernel/sched/sched.h
> +++ b/kernel/sched/sched.h
> @@ -465,6 +465,7 @@ extern void init_cfs_bandwidth(struct cfs_bandwidth *cfs_b, struct cfs_bandwidth
> extern void __refill_cfs_bandwidth_runtime(struct cfs_bandwidth *cfs_b);
> extern void start_cfs_bandwidth(struct cfs_bandwidth *cfs_b);
> extern void unthrottle_cfs_rq(struct cfs_rq *cfs_rq);
> +extern bool cfs_task_bw_constrained(struct task_struct *p);
>
> extern void init_tg_rt_entry(struct task_group *tg, struct rt_rq *rt_rq,
> struct sched_rt_entity *rt_se, int cpu,
On Wed, Jul 12, 2023 at 03:11:32PM -0700 Benjamin Segall wrote:
> Phil Auld <pauld@redhat.com> writes:
>
> > CFS bandwidth limits and NOHZ full don't play well together. Tasks
> > can easily run well past their quotas before a remote tick does
> > accounting. This leads to long, multi-period stalls before such
> > tasks can run again. Currently, when presented with these conflicting
> > requirements the scheduler is favoring nohz_full and letting the tick
> > be stopped. However, nohz tick stopping is already best-effort, there
> > are a number of conditions that can prevent it, whereas cfs runtime
> > bandwidth is expected to be enforced.
> >
> > Make the scheduler favor bandwidth over stopping the tick by setting
> > TICK_DEP_BIT_SCHED when the only running task is a cfs task with
> > runtime limit enabled. We use cfs_b->hierarchical_quota to
> > determine if the task requires the tick.
> >
> > Add check in pick_next_task_fair() as well since that is where
> > we have a handle on the task that is actually going to be running.
> >
> > Add check in sched_can_stop_tick() to cover some edge cases such
> > as nr_running going from 2->1 and the 1 remains the running task.
> >
> > Add sched_feat HZ_BW (off by default) to control the tick_stop
> > behavior.
>
> I think this looks good now.
> Reviewed-By: Ben Segall <bsegall@google.com>
>
Thanks for your help and patience!
Cheers,
Phil
> >
> > Signed-off-by: Phil Auld <pauld@redhat.com>
> > Cc: Ingo Molnar <mingo@redhat.com>
> > Cc: Peter Zijlstra <peterz@infradead.org>
> > Cc: Vincent Guittot <vincent.guittot@linaro.org>
> > Cc: Juri Lelli <juri.lelli@redhat.com>
> > Cc: Dietmar Eggemann <dietmar.eggemann@arm.com>
> > Cc: Valentin Schneider <vschneid@redhat.com>
> > Cc: Ben Segall <bsegall@google.com>
> > Cc: Frederic Weisbecker <frederic@kernel.org>
> > ---
> > v6: restore check for fair_sched_class
> >
> > v5: Reworked checks to use newly-fixed cfs_b->hierarchical_quota to
> > check for bw
> > constraints.
> >
> > v4: Made checks for runtime_enabled hierarchical.
> >
> > v3: Moved sched_cfs_bandwidth_active() prototype to sched.h outside of
> > guards to
> > silence -Wmissing-prototypes.
> >
> > v2: Ben pointed out that the bit could get cleared in the dequeue path
> > if we migrate a newly enqueued task without preempting curr. Added a
> > check for that edge case to sched_can_stop_tick. Removed the call to
> > sched_can_stop_tick from sched_fair_update_stop_tick since it was
> > redundant.
> >
> > kernel/sched/core.c | 26 ++++++++++++++++++++++
> > kernel/sched/fair.c | 49 +++++++++++++++++++++++++++++++++++++++++
> > kernel/sched/features.h | 2 ++
> > kernel/sched/sched.h | 1 +
> > 4 files changed, 78 insertions(+)
> >
> > diff --git a/kernel/sched/core.c b/kernel/sched/core.c
> > index f80697a79baf..8a2ed4c0b709 100644
> > --- a/kernel/sched/core.c
> > +++ b/kernel/sched/core.c
> > @@ -1194,6 +1194,20 @@ static void nohz_csd_func(void *info)
> > #endif /* CONFIG_NO_HZ_COMMON */
> >
> > #ifdef CONFIG_NO_HZ_FULL
> > +static inline bool __need_bw_check(struct rq *rq, struct task_struct *p)
> > +{
> > + if (rq->nr_running != 1)
> > + return false;
> > +
> > + if (p->sched_class != &fair_sched_class)
> > + return false;
> > +
> > + if (!task_on_rq_queued(p))
> > + return false;
> > +
> > + return true;
> > +}
> > +
> > bool sched_can_stop_tick(struct rq *rq)
> > {
> > int fifo_nr_running;
> > @@ -1229,6 +1243,18 @@ bool sched_can_stop_tick(struct rq *rq)
> > if (rq->nr_running > 1)
> > return false;
> >
> > + /*
> > + * If there is one task and it has CFS runtime bandwidth constraints
> > + * and it's on the cpu now we don't want to stop the tick.
> > + * This check prevents clearing the bit if a newly enqueued task here is
> > + * dequeued by migrating while the constrained task continues to run.
> > + * E.g. going from 2->1 without going through pick_next_task().
> > + */
> > + if (sched_feat(HZ_BW) && __need_bw_check(rq, rq->curr)) {
> > + if (cfs_task_bw_constrained(rq->curr))
> > + return false;
> > + }
> > +
> > return true;
> > }
> > #endif /* CONFIG_NO_HZ_FULL */
> > diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
> > index d9b3d4617e16..acd9f317aad1 100644
> > --- a/kernel/sched/fair.c
> > +++ b/kernel/sched/fair.c
> > @@ -6140,6 +6140,46 @@ static void __maybe_unused unthrottle_offline_cfs_rqs(struct rq *rq)
> > rcu_read_unlock();
> > }
> >
> > +bool cfs_task_bw_constrained(struct task_struct *p)
> > +{
> > + struct cfs_rq *cfs_rq = task_cfs_rq(p);
> > +
> > + if (!cfs_bandwidth_used())
> > + return false;
> > +
> > + if (cfs_rq->runtime_enabled ||
> > + tg_cfs_bandwidth(cfs_rq->tg)->hierarchical_quota != RUNTIME_INF)
> > + return true;
> > +
> > + return false;
> > +}
> > +
> > +#ifdef CONFIG_NO_HZ_FULL
> > +/* called from pick_next_task_fair() */
> > +static void sched_fair_update_stop_tick(struct rq *rq, struct task_struct *p)
> > +{
> > + int cpu = cpu_of(rq);
> > +
> > + if (!sched_feat(HZ_BW) || !cfs_bandwidth_used())
> > + return;
> > +
> > + if (!tick_nohz_full_cpu(cpu))
> > + return;
> > +
> > + if (rq->nr_running != 1)
> > + return;
> > +
> > + /*
> > + * We know there is only one task runnable and we've just picked it. The
> > + * normal enqueue path will have cleared TICK_DEP_BIT_SCHED if we will
> > + * be otherwise able to stop the tick. Just need to check if we are using
> > + * bandwidth control.
> > + */
> > + if (cfs_task_bw_constrained(p))
> > + tick_nohz_dep_set_cpu(cpu, TICK_DEP_BIT_SCHED);
> > +}
> > +#endif
> > +
> > #else /* CONFIG_CFS_BANDWIDTH */
> >
> > static inline bool cfs_bandwidth_used(void)
> > @@ -6182,9 +6222,17 @@ static inline struct cfs_bandwidth *tg_cfs_bandwidth(struct task_group *tg)
> > static inline void destroy_cfs_bandwidth(struct cfs_bandwidth *cfs_b) {}
> > static inline void update_runtime_enabled(struct rq *rq) {}
> > static inline void unthrottle_offline_cfs_rqs(struct rq *rq) {}
> > +bool cfs_task_bw_constrained(struct task_struct *p)
> > +{
> > + return false;
> > +}
> >
> > #endif /* CONFIG_CFS_BANDWIDTH */
> >
> > +#if !defined(CONFIG_CFS_BANDWIDTH) || !defined(CONFIG_NO_HZ_FULL)
> > +static inline void sched_fair_update_stop_tick(struct rq *rq, struct task_struct *p) {}
> > +#endif
> > +
> > /**************************************************
> > * CFS operations on tasks:
> > */
> > @@ -8098,6 +8146,7 @@ done: __maybe_unused;
> > hrtick_start_fair(rq, p);
> >
> > update_misfit_status(p, rq);
> > + sched_fair_update_stop_tick(rq, p);
> >
> > return p;
> >
> > diff --git a/kernel/sched/features.h b/kernel/sched/features.h
> > index ee7f23c76bd3..6fdf1fdf6b17 100644
> > --- a/kernel/sched/features.h
> > +++ b/kernel/sched/features.h
> > @@ -101,3 +101,5 @@ SCHED_FEAT(LATENCY_WARN, false)
> >
> > SCHED_FEAT(ALT_PERIOD, true)
> > SCHED_FEAT(BASE_SLICE, true)
> > +
> > +SCHED_FEAT(HZ_BW, false)
> > diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h
> > index 63822c9238cc..d6d346bc78aa 100644
> > --- a/kernel/sched/sched.h
> > +++ b/kernel/sched/sched.h
> > @@ -465,6 +465,7 @@ extern void init_cfs_bandwidth(struct cfs_bandwidth *cfs_b, struct cfs_bandwidth
> > extern void __refill_cfs_bandwidth_runtime(struct cfs_bandwidth *cfs_b);
> > extern void start_cfs_bandwidth(struct cfs_bandwidth *cfs_b);
> > extern void unthrottle_cfs_rq(struct cfs_rq *cfs_rq);
> > +extern bool cfs_task_bw_constrained(struct task_struct *p);
> >
> > extern void init_tg_rt_entry(struct task_group *tg, struct rt_rq *rt_rq,
> > struct sched_rt_entity *rt_se, int cpu,
>
--
The following commit has been merged into the sched/core branch of tip:
Commit-ID: 88c56cfeaec4642aee8aac58b38d5708c6aae0d3
Gitweb: https://git.kernel.org/tip/88c56cfeaec4642aee8aac58b38d5708c6aae0d3
Author: Phil Auld <pauld@redhat.com>
AuthorDate: Wed, 12 Jul 2023 09:33:57 -04:00
Committer: Peter Zijlstra <peterz@infradead.org>
CommitterDate: Wed, 02 Aug 2023 16:19:26 +02:00
sched/fair: Block nohz tick_stop when cfs bandwidth in use
CFS bandwidth limits and NOHZ full don't play well together. Tasks
can easily run well past their quotas before a remote tick does
accounting. This leads to long, multi-period stalls before such
tasks can run again. Currently, when presented with these conflicting
requirements the scheduler is favoring nohz_full and letting the tick
be stopped. However, nohz tick stopping is already best-effort, there
are a number of conditions that can prevent it, whereas cfs runtime
bandwidth is expected to be enforced.
Make the scheduler favor bandwidth over stopping the tick by setting
TICK_DEP_BIT_SCHED when the only running task is a cfs task with
runtime limit enabled. We use cfs_b->hierarchical_quota to
determine if the task requires the tick.
Add check in pick_next_task_fair() as well since that is where
we have a handle on the task that is actually going to be running.
Add check in sched_can_stop_tick() to cover some edge cases such
as nr_running going from 2->1 and the 1 remains the running task.
Reviewed-By: Ben Segall <bsegall@google.com>
Signed-off-by: Phil Auld <pauld@redhat.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/r/20230712133357.381137-3-pauld@redhat.com
---
kernel/sched/core.c | 26 ++++++++++++++++++++-
kernel/sched/fair.c | 52 +++++++++++++++++++++++++++++++++++++++-
kernel/sched/features.h | 2 ++-
kernel/sched/sched.h | 2 ++-
4 files changed, 81 insertions(+), 1 deletion(-)
diff --git a/kernel/sched/core.c b/kernel/sched/core.c
index 3af25ca..614271a 100644
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -1194,6 +1194,20 @@ static void nohz_csd_func(void *info)
#endif /* CONFIG_NO_HZ_COMMON */
#ifdef CONFIG_NO_HZ_FULL
+static inline bool __need_bw_check(struct rq *rq, struct task_struct *p)
+{
+ if (rq->nr_running != 1)
+ return false;
+
+ if (p->sched_class != &fair_sched_class)
+ return false;
+
+ if (!task_on_rq_queued(p))
+ return false;
+
+ return true;
+}
+
bool sched_can_stop_tick(struct rq *rq)
{
int fifo_nr_running;
@@ -1229,6 +1243,18 @@ bool sched_can_stop_tick(struct rq *rq)
if (rq->nr_running > 1)
return false;
+ /*
+ * If there is one task and it has CFS runtime bandwidth constraints
+ * and it's on the cpu now we don't want to stop the tick.
+ * This check prevents clearing the bit if a newly enqueued task here is
+ * dequeued by migrating while the constrained task continues to run.
+ * E.g. going from 2->1 without going through pick_next_task().
+ */
+ if (sched_feat(HZ_BW) && __need_bw_check(rq, rq->curr)) {
+ if (cfs_task_bw_constrained(rq->curr))
+ return false;
+ }
+
return true;
}
#endif /* CONFIG_NO_HZ_FULL */
diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index 26bfbb6..c282064 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -6189,6 +6189,46 @@ static void __maybe_unused unthrottle_offline_cfs_rqs(struct rq *rq)
rq_clock_stop_loop_update(rq);
}
+bool cfs_task_bw_constrained(struct task_struct *p)
+{
+ struct cfs_rq *cfs_rq = task_cfs_rq(p);
+
+ if (!cfs_bandwidth_used())
+ return false;
+
+ if (cfs_rq->runtime_enabled ||
+ tg_cfs_bandwidth(cfs_rq->tg)->hierarchical_quota != RUNTIME_INF)
+ return true;
+
+ return false;
+}
+
+#ifdef CONFIG_NO_HZ_FULL
+/* called from pick_next_task_fair() */
+static void sched_fair_update_stop_tick(struct rq *rq, struct task_struct *p)
+{
+ int cpu = cpu_of(rq);
+
+ if (!sched_feat(HZ_BW) || !cfs_bandwidth_used())
+ return;
+
+ if (!tick_nohz_full_cpu(cpu))
+ return;
+
+ if (rq->nr_running != 1)
+ return;
+
+ /*
+ * We know there is only one task runnable and we've just picked it. The
+ * normal enqueue path will have cleared TICK_DEP_BIT_SCHED if we will
+ * be otherwise able to stop the tick. Just need to check if we are using
+ * bandwidth control.
+ */
+ if (cfs_task_bw_constrained(p))
+ tick_nohz_dep_set_cpu(cpu, TICK_DEP_BIT_SCHED);
+}
+#endif
+
#else /* CONFIG_CFS_BANDWIDTH */
static inline bool cfs_bandwidth_used(void)
@@ -6231,9 +6271,18 @@ static inline struct cfs_bandwidth *tg_cfs_bandwidth(struct task_group *tg)
static inline void destroy_cfs_bandwidth(struct cfs_bandwidth *cfs_b) {}
static inline void update_runtime_enabled(struct rq *rq) {}
static inline void unthrottle_offline_cfs_rqs(struct rq *rq) {}
-
+#ifdef CONFIG_CGROUP_SCHED
+bool cfs_task_bw_constrained(struct task_struct *p)
+{
+ return false;
+}
+#endif
#endif /* CONFIG_CFS_BANDWIDTH */
+#if !defined(CONFIG_CFS_BANDWIDTH) || !defined(CONFIG_NO_HZ_FULL)
+static inline void sched_fair_update_stop_tick(struct rq *rq, struct task_struct *p) {}
+#endif
+
/**************************************************
* CFS operations on tasks:
*/
@@ -8201,6 +8250,7 @@ done: __maybe_unused;
hrtick_start_fair(rq, p);
update_misfit_status(p, rq);
+ sched_fair_update_stop_tick(rq, p);
return p;
diff --git a/kernel/sched/features.h b/kernel/sched/features.h
index ee7f23c..e10074c 100644
--- a/kernel/sched/features.h
+++ b/kernel/sched/features.h
@@ -101,3 +101,5 @@ SCHED_FEAT(LATENCY_WARN, false)
SCHED_FEAT(ALT_PERIOD, true)
SCHED_FEAT(BASE_SLICE, true)
+
+SCHED_FEAT(HZ_BW, true)
diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h
index 602de71..19af176 100644
--- a/kernel/sched/sched.h
+++ b/kernel/sched/sched.h
@@ -459,6 +459,7 @@ extern void init_cfs_bandwidth(struct cfs_bandwidth *cfs_b, struct cfs_bandwidth
extern void __refill_cfs_bandwidth_runtime(struct cfs_bandwidth *cfs_b);
extern void start_cfs_bandwidth(struct cfs_bandwidth *cfs_b);
extern void unthrottle_cfs_rq(struct cfs_rq *cfs_rq);
+extern bool cfs_task_bw_constrained(struct task_struct *p);
extern void init_tg_rt_entry(struct task_group *tg, struct rt_rq *rt_rq,
struct sched_rt_entity *rt_se, int cpu,
@@ -494,6 +495,7 @@ static inline void set_task_rq_fair(struct sched_entity *se,
#else /* CONFIG_CGROUP_SCHED */
struct cfs_bandwidth { };
+static inline bool cfs_task_bw_constrained(struct task_struct *p) { return false; }
#endif /* CONFIG_CGROUP_SCHED */© 2016 - 2026 Red Hat, Inc.