5.10-stable review patch. If anyone has any objections, please let me know.
------------------
From: Vincent Guittot vincent.guittot@linaro.org
[ Upstream commit 0826530de3cbdc89e60a89e86def94a5f0fc81ca ]
newidle_balance runs with both preempt and irq disabled which prevent local irq to run during this period. The duration for updating the blocked load of CPUs varies according to the number of CPU cgroups with non-decayed load and extends this critical period to an uncontrolled level.
Remove the update from newidle_balance and trigger a normal ILB that will take care of the update instead.
This reduces the IRQ latency from O(nr_cgroups * nr_nohz_cpus) to O(nr_cgroups).
Signed-off-by: Vincent Guittot vincent.guittot@linaro.org Signed-off-by: Peter Zijlstra (Intel) peterz@infradead.org Signed-off-by: Ingo Molnar mingo@kernel.org Reviewed-by: Valentin Schneider valentin.schneider@arm.com Link: https://lkml.kernel.org/r/20210224133007.28644-2-vincent.guittot@linaro.org Stable-dep-of: ff47a0acfcce ("sched/fair: Check idle_cpu() before need_resched() to detect ilb CPU turning busy") Signed-off-by: Sasha Levin sashal@kernel.org --- kernel/sched/fair.c | 33 +++++---------------------------- 1 file changed, 5 insertions(+), 28 deletions(-)
diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c index a6a755aec32b5..a6e34c58cee92 100644 --- a/kernel/sched/fair.c +++ b/kernel/sched/fair.c @@ -7621,8 +7621,6 @@ enum migration_type { #define LBF_NEED_BREAK 0x02 #define LBF_DST_PINNED 0x04 #define LBF_SOME_PINNED 0x08 -#define LBF_NOHZ_STATS 0x10 -#define LBF_NOHZ_AGAIN 0x20
struct lb_env { struct sched_domain *sd; @@ -8695,9 +8693,6 @@ static inline void update_sg_lb_stats(struct lb_env *env, for_each_cpu_and(i, sched_group_span(group), env->cpus) { struct rq *rq = cpu_rq(i);
- if ((env->flags & LBF_NOHZ_STATS) && update_nohz_stats(rq, false)) - env->flags |= LBF_NOHZ_AGAIN; - sgs->group_load += cpu_load(rq); sgs->group_util += cpu_util(i); sgs->group_runnable += cpu_runnable(rq); @@ -9230,11 +9225,6 @@ static inline void update_sd_lb_stats(struct lb_env *env, struct sd_lb_stats *sd struct sg_lb_stats tmp_sgs; int sg_status = 0;
-#ifdef CONFIG_NO_HZ_COMMON - if (env->idle == CPU_NEWLY_IDLE && READ_ONCE(nohz.has_blocked)) - env->flags |= LBF_NOHZ_STATS; -#endif - do { struct sg_lb_stats *sgs = &tmp_sgs; int local_group; @@ -9271,14 +9261,6 @@ static inline void update_sd_lb_stats(struct lb_env *env, struct sd_lb_stats *sd /* Tag domain that child domain prefers tasks go to siblings first */ sds->prefer_sibling = child && child->flags & SD_PREFER_SIBLING;
-#ifdef CONFIG_NO_HZ_COMMON - if ((env->flags & LBF_NOHZ_AGAIN) && - cpumask_subset(nohz.idle_cpus_mask, sched_domain_span(env->sd))) { - - WRITE_ONCE(nohz.next_blocked, - jiffies + msecs_to_jiffies(LOAD_AVG_PERIOD)); - } -#endif
if (env->sd->flags & SD_NUMA) env->fbq_type = fbq_classify_group(&sds->busiest_stat); @@ -10803,16 +10785,11 @@ static void nohz_newidle_balance(struct rq *this_rq) time_before(jiffies, READ_ONCE(nohz.next_blocked))) return;
- raw_spin_unlock(&this_rq->lock); /* - * This CPU is going to be idle and blocked load of idle CPUs - * need to be updated. Run the ilb locally as it is a good - * candidate for ilb instead of waking up another idle CPU. - * Kick an normal ilb if we failed to do the update. + * Blocked load of idle CPUs need to be updated. + * Kick an ILB to update statistics. */ - if (!_nohz_idle_balance(this_rq, NOHZ_STATS_KICK, CPU_NEWLY_IDLE)) - kick_ilb(NOHZ_STATS_KICK); - raw_spin_lock(&this_rq->lock); + kick_ilb(NOHZ_STATS_KICK); }
#else /* !CONFIG_NO_HZ_COMMON */ @@ -10873,8 +10850,6 @@ static int newidle_balance(struct rq *this_rq, struct rq_flags *rf) update_next_balance(sd, &next_balance); rcu_read_unlock();
- nohz_newidle_balance(this_rq); - goto out; }
@@ -10940,6 +10915,8 @@ static int newidle_balance(struct rq *this_rq, struct rq_flags *rf)
if (pulled_task) this_rq->idle_stamp = 0; + else + nohz_newidle_balance(this_rq);
rq_repin_lock(this_rq, rf);