Merge tag 'timers-nohz-2021-06-28' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip

Pull timers/nohz updates from Ingo Molnar:

 - Micro-optimize tick_nohz_full_cpu()

 - Optimize idle exit tick restarts to be less eager

 - Optimize tick_nohz_dep_set_task() to only wake up a single CPU.
   This reduces IPIs and interruptions on nohz_full CPUs.

 - Optimize tick_nohz_dep_set_signal() in a similar fashion.

 - Skip IPIs in tick_nohz_kick_task() when trying to kick a
   non-running task.

 - Micro-optimize tick_nohz_task_switch() IRQ flags handling to
   reduce context switching costs.

 - Misc cleanups and fixes

* tag 'timers-nohz-2021-06-28' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  MAINTAINERS: Add myself as context tracking maintainer
  tick/nohz: Call tick_nohz_task_switch() with interrupts disabled
  tick/nohz: Kick only _queued_ task whose tick dependency is updated
  tick/nohz: Change signal tick dependency to wake up CPUs of member tasks
  tick/nohz: Only wake up a single target cpu when kicking a task
  tick/nohz: Update nohz_full Kconfig help
  tick/nohz: Update idle_exittime on actual idle exit
  tick/nohz: Remove superflous check for CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
  tick/nohz: Conditionally restart tick on idle exit
  tick/nohz: Evaluate the CPU expression after the static key

7 files changed, 125 insertions, 60 deletions

diff --git a/MAINTAINERS b/MAINTAINERS
index 2560e8f6abc1..4289f9396f03 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -4610,6 +4610,12 @@ S:	Supported
 F:	drivers/video/console/
 F:	include/linux/console*
 
+CONTEXT TRACKING
+M:	Frederic Weisbecker <frederic@kernel.org>
+S:	Maintained
+F:	kernel/context_tracking.c
+F:	include/linux/context_tracking*
+
 CONTROL GROUP (CGROUP)
 M:	Tejun Heo <tj@kernel.org>
 M:	Zefan Li <lizefan.x@bytedance.com>
diff --git a/include/linux/sched.h b/include/linux/sched.h
index 31aaf9d7b3e9..ec8d07d88641 100644
--- a/include/linux/sched.h
+++ b/include/linux/sched.h
@@ -2028,6 +2028,8 @@ static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
 
 #endif /* CONFIG_SMP */
 
+extern bool sched_task_on_rq(struct task_struct *p);
+
 /*
  * In order to reduce various lock holder preemption latencies provide an
  * interface to see if a vCPU is currently running or not.
diff --git a/include/linux/tick.h b/include/linux/tick.h
index 1a0ff88fa107..bfd571f18cfd 100644
--- a/include/linux/tick.h
+++ b/include/linux/tick.h
@@ -186,13 +186,17 @@ static inline bool tick_nohz_full_enabled(void)
 	return tick_nohz_full_running;
 }
 
-static inline bool tick_nohz_full_cpu(int cpu)
-{
-	if (!tick_nohz_full_enabled())
-		return false;
-
-	return cpumask_test_cpu(cpu, tick_nohz_full_mask);
-}
+/*
+ * Check if a CPU is part of the nohz_full subset. Arrange for evaluating
+ * the cpu expression (typically smp_processor_id()) _after_ the static
+ * key.
+ */
+#define tick_nohz_full_cpu(_cpu) ({					\
+	bool __ret = false;						\
+	if (tick_nohz_full_enabled())					\
+		__ret = cpumask_test_cpu((_cpu), tick_nohz_full_mask);	\
+	__ret;								\
+})
 
 static inline void tick_nohz_full_add_cpus_to(struct cpumask *mask)
 {
@@ -208,7 +212,7 @@ extern void tick_nohz_dep_set_task(struct task_struct *tsk,
 				   enum tick_dep_bits bit);
 extern void tick_nohz_dep_clear_task(struct task_struct *tsk,
 				     enum tick_dep_bits bit);
-extern void tick_nohz_dep_set_signal(struct signal_struct *signal,
+extern void tick_nohz_dep_set_signal(struct task_struct *tsk,
 				     enum tick_dep_bits bit);
 extern void tick_nohz_dep_clear_signal(struct signal_struct *signal,
 				       enum tick_dep_bits bit);
@@ -253,11 +257,11 @@ static inline void tick_dep_clear_task(struct task_struct *tsk,
 	if (tick_nohz_full_enabled())
 		tick_nohz_dep_clear_task(tsk, bit);
 }
-static inline void tick_dep_set_signal(struct signal_struct *signal,
+static inline void tick_dep_set_signal(struct task_struct *tsk,
 				       enum tick_dep_bits bit)
 {
 	if (tick_nohz_full_enabled())
-		tick_nohz_dep_set_signal(signal, bit);
+		tick_nohz_dep_set_signal(tsk, bit);
 }
 static inline void tick_dep_clear_signal(struct signal_struct *signal,
 					 enum tick_dep_bits bit)
@@ -285,7 +289,7 @@ static inline void tick_dep_set_task(struct task_struct *tsk,
 				     enum tick_dep_bits bit) { }
 static inline void tick_dep_clear_task(struct task_struct *tsk,
 				       enum tick_dep_bits bit) { }
-static inline void tick_dep_set_signal(struct signal_struct *signal,
+static inline void tick_dep_set_signal(struct task_struct *tsk,
 				       enum tick_dep_bits bit) { }
 static inline void tick_dep_clear_signal(struct signal_struct *signal,
 					 enum tick_dep_bits bit) { }
diff --git a/kernel/sched/core.c b/kernel/sched/core.c
index 4a2a448d8dd4..cf16f8fda9a6 100644
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -1928,6 +1928,11 @@ static inline void uclamp_post_fork(struct task_struct *p) { }
 static inline void init_uclamp(void) { }
 #endif /* CONFIG_UCLAMP_TASK */
 
+bool sched_task_on_rq(struct task_struct *p)
+{
+	return task_on_rq_queued(p);
+}
+
 static inline void enqueue_task(struct rq *rq, struct task_struct *p, int flags)
 {
 	if (!(flags & ENQUEUE_NOCLOCK))
@@ -4546,6 +4551,7 @@ static struct rq *finish_task_switch(struct task_struct *prev)
 	vtime_task_switch(prev);
 	perf_event_task_sched_in(prev, current);
 	finish_task(prev);
+	tick_nohz_task_switch();
 	finish_lock_switch(rq);
 	finish_arch_post_lock_switch();
 	kcov_finish_switch(current);
@@ -4591,7 +4597,6 @@ static struct rq *finish_task_switch(struct task_struct *prev)
 		put_task_struct_rcu_user(prev);
 	}
 
-	tick_nohz_task_switch();
 	return rq;
 }
 
diff --git a/kernel/time/Kconfig b/kernel/time/Kconfig
index 83e158d016ba..7df71ef0e1fd 100644
--- a/kernel/time/Kconfig
+++ b/kernel/time/Kconfig
@@ -117,13 +117,14 @@ config NO_HZ_FULL
 	 the task mostly runs in userspace and has few kernel activity.
 
 	 You need to fill up the nohz_full boot parameter with the
-	 desired range of dynticks CPUs.
+	 desired range of dynticks CPUs to use it. This is implemented at
+	 the expense of some overhead in user <-> kernel transitions:
+	 syscalls, exceptions and interrupts.
 
-	 This is implemented at the expense of some overhead in user <-> kernel
-	 transitions: syscalls, exceptions and interrupts. Even when it's
-	 dynamically off.
+	 By default, without passing the nohz_full parameter, this behaves just
+	 like NO_HZ_IDLE.
 
-	 Say N.
+	 If you're a distro say Y.
 
 endchoice
 
diff --git a/kernel/time/posix-cpu-timers.c b/kernel/time/posix-cpu-timers.c
index 3bb96a8b49c9..29a5e54e6e10 100644
--- a/kernel/time/posix-cpu-timers.c
+++ b/kernel/time/posix-cpu-timers.c
@@ -523,7 +523,7 @@ static void arm_timer(struct k_itimer *timer, struct task_struct *p)
 	if (CPUCLOCK_PERTHREAD(timer->it_clock))
 		tick_dep_set_task(p, TICK_DEP_BIT_POSIX_TIMER);
 	else
-		tick_dep_set_signal(p->signal, TICK_DEP_BIT_POSIX_TIMER);
+		tick_dep_set_signal(p, TICK_DEP_BIT_POSIX_TIMER);
 }
 
 /*
@@ -1358,7 +1358,7 @@ void set_process_cpu_timer(struct task_struct *tsk, unsigned int clkid,
 	if (*newval < *nextevt)
 		*nextevt = *newval;
 
-	tick_dep_set_signal(tsk->signal, TICK_DEP_BIT_POSIX_TIMER);
+	tick_dep_set_signal(tsk, TICK_DEP_BIT_POSIX_TIMER);
 }
 
 static int do_cpu_nanosleep(const clockid_t which_clock, int flags,
diff --git a/kernel/time/tick-sched.c b/kernel/time/tick-sched.c
index 6784f27a3099..6bffe5af8cb1 100644
--- a/kernel/time/tick-sched.c
+++ b/kernel/time/tick-sched.c
@@ -323,6 +323,46 @@ void tick_nohz_full_kick_cpu(int cpu)
 	irq_work_queue_on(&per_cpu(nohz_full_kick_work, cpu), cpu);
 }
 
+static void tick_nohz_kick_task(struct task_struct *tsk)
+{
+	int cpu;
+
+	/*
+	 * If the task is not running, run_posix_cpu_timers()
+	 * has nothing to elapse, IPI can then be spared.
+	 *
+	 * activate_task()                      STORE p->tick_dep_mask
+	 *   STORE p->on_rq
+	 * __schedule() (switch to task 'p')    smp_mb() (atomic_fetch_or())
+	 *   LOCK rq->lock                      LOAD p->on_rq
+	 *   smp_mb__after_spin_lock()
+	 *   tick_nohz_task_switch()
+	 *     LOAD p->tick_dep_mask
+	 */
+	if (!sched_task_on_rq(tsk))
+		return;
+
+	/*
+	 * If the task concurrently migrates to another CPU,
+	 * we guarantee it sees the new tick dependency upon
+	 * schedule.
+	 *
+	 * set_task_cpu(p, cpu);
+	 *   STORE p->cpu = @cpu
+	 * __schedule() (switch to task 'p')
+	 *   LOCK rq->lock
+	 *   smp_mb__after_spin_lock()          STORE p->tick_dep_mask
+	 *   tick_nohz_task_switch()            smp_mb() (atomic_fetch_or())
+	 *      LOAD p->tick_dep_mask           LOAD p->cpu
+	 */
+	cpu = task_cpu(tsk);
+
+	preempt_disable();
+	if (cpu_online(cpu))
+		tick_nohz_full_kick_cpu(cpu);
+	preempt_enable();
+}
+
 /*
  * Kick all full dynticks CPUs in order to force these to re-evaluate
  * their dependency on the tick and restart it if necessary.
@@ -405,19 +445,8 @@ EXPORT_SYMBOL_GPL(tick_nohz_dep_clear_cpu);
  */
 void tick_nohz_dep_set_task(struct task_struct *tsk, enum tick_dep_bits bit)
 {
-	if (!atomic_fetch_or(BIT(bit), &tsk->tick_dep_mask)) {
-		if (tsk == current) {
-			preempt_disable();
-			tick_nohz_full_kick();
-			preempt_enable();
-		} else {
-			/*
-			 * Some future tick_nohz_full_kick_task()
-			 * should optimize this.
-			 */
-			tick_nohz_full_kick_all();
-		}
-	}
+	if (!atomic_fetch_or(BIT(bit), &tsk->tick_dep_mask))
+		tick_nohz_kick_task(tsk);
 }
 EXPORT_SYMBOL_GPL(tick_nohz_dep_set_task);
 
@@ -431,9 +460,20 @@ EXPORT_SYMBOL_GPL(tick_nohz_dep_clear_task);
  * Set a per-taskgroup tick dependency. Posix CPU timers need this in order to elapse
  * per process timers.
  */
-void tick_nohz_dep_set_signal(struct signal_struct *sig, enum tick_dep_bits bit)
+void tick_nohz_dep_set_signal(struct task_struct *tsk,
+			      enum tick_dep_bits bit)
 {
-	tick_nohz_dep_set_all(&sig->tick_dep_mask, bit);
+	int prev;
+	struct signal_struct *sig = tsk->signal;
+
+	prev = atomic_fetch_or(BIT(bit), &sig->tick_dep_mask);
+	if (!prev) {
+		struct task_struct *t;
+
+		lockdep_assert_held(&tsk->sighand->siglock);
+		__for_each_thread(sig, t)
+			tick_nohz_kick_task(t);
+	}
 }
 
 void tick_nohz_dep_clear_signal(struct signal_struct *sig, enum tick_dep_bits bit)
@@ -448,13 +488,10 @@ void tick_nohz_dep_clear_signal(struct signal_struct *sig, enum tick_dep_bits bi
  */
 void __tick_nohz_task_switch(void)
 {
-	unsigned long flags;
 	struct tick_sched *ts;
 
-	local_irq_save(flags);
-
 	if (!tick_nohz_full_cpu(smp_processor_id()))
-		goto out;
+		return;
 
 	ts = this_cpu_ptr(&tick_cpu_sched);
 
@@ -463,8 +500,6 @@ void __tick_nohz_task_switch(void)
 		    atomic_read(&current->signal->tick_dep_mask))
 			tick_nohz_full_kick();
 	}
-out:
-	local_irq_restore(flags);
 }
 
 /* Get the boot-time nohz CPU list from the kernel parameters. */
@@ -922,27 +957,31 @@ static void tick_nohz_restart_sched_tick(struct tick_sched *ts, ktime_t now)
 	 * Cancel the scheduled timer and restore the tick
 	 */
 	ts->tick_stopped  = 0;
-	ts->idle_exittime = now;
-
 	tick_nohz_restart(ts, now);
 }
 
-static void tick_nohz_full_update_tick(struct tick_sched *ts)
+static void __tick_nohz_full_update_tick(struct tick_sched *ts,
+					 ktime_t now)
 {
 #ifdef CONFIG_NO_HZ_FULL
 	int cpu = smp_processor_id();
 
-	if (!tick_nohz_full_cpu(cpu))
+	if (can_stop_full_tick(cpu, ts))
+		tick_nohz_stop_sched_tick(ts, cpu);
+	else if (ts->tick_stopped)
+		tick_nohz_restart_sched_tick(ts, now);
+#endif
+}
+
+static void tick_nohz_full_update_tick(struct tick_sched *ts)
+{
+	if (!tick_nohz_full_cpu(smp_processor_id()))
 		return;
 
 	if (!ts->tick_stopped && ts->nohz_mode == NOHZ_MODE_INACTIVE)
 		return;
 
-	if (can_stop_full_tick(cpu, ts))
-		tick_nohz_stop_sched_tick(ts, cpu);
-	else if (ts->tick_stopped)
-		tick_nohz_restart_sched_tick(ts, ktime_get());
-#endif
+	__tick_nohz_full_update_tick(ts, ktime_get());
 }
 
 static bool can_stop_idle_tick(int cpu, struct tick_sched *ts)
@@ -1189,11 +1228,13 @@ unsigned long tick_nohz_get_idle_calls(void)
 	return ts->idle_calls;
 }
 
-static void tick_nohz_account_idle_ticks(struct tick_sched *ts)
+static void tick_nohz_account_idle_time(struct tick_sched *ts,
+					ktime_t now)
 {
-#ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
 	unsigned long ticks;
 
+	ts->idle_exittime = now;
+
 	if (vtime_accounting_enabled_this_cpu())
 		return;
 	/*
@@ -1207,21 +1248,27 @@ static void tick_nohz_account_idle_ticks(struct tick_sched *ts)
 	 */
 	if (ticks && ticks < LONG_MAX)
 		account_idle_ticks(ticks);
-#endif
 }
 
-static void __tick_nohz_idle_restart_tick(struct tick_sched *ts, ktime_t now)
+void tick_nohz_idle_restart_tick(void)
 {
-	tick_nohz_restart_sched_tick(ts, now);
-	tick_nohz_account_idle_ticks(ts);
+	struct tick_sched *ts = this_cpu_ptr(&tick_cpu_sched);
+
+	if (ts->tick_stopped) {
+		ktime_t now = ktime_get();
+		tick_nohz_restart_sched_tick(ts, now);
+		tick_nohz_account_idle_time(ts, now);
+	}
 }
 
-void tick_nohz_idle_restart_tick(void)
+static void tick_nohz_idle_update_tick(struct tick_sched *ts, ktime_t now)
 {
-	struct tick_sched *ts = this_cpu_ptr(&tick_cpu_sched);
+	if (tick_nohz_full_cpu(smp_processor_id()))
+		__tick_nohz_full_update_tick(ts, now);
+	else
+		tick_nohz_restart_sched_tick(ts, now);
 
-	if (ts->tick_stopped)
-		__tick_nohz_idle_restart_tick(ts, ktime_get());
+	tick_nohz_account_idle_time(ts, now);
 }
 
 /**
@@ -1253,7 +1300,7 @@ void tick_nohz_idle_exit(void)
 		tick_nohz_stop_idle(ts, now);
 
 	if (tick_stopped)
-		__tick_nohz_idle_restart_tick(ts, now);
+		tick_nohz_idle_update_tick(ts, now);
 
 	local_irq_enable();
 }