From: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Date: Fri, 3 Jul 2009 08:44:31 -0500
Subject: hrtimer: by timers by default into the softirq context

We can't have hrtimers callbacks running in hardirq context on RT. Therefore
the timers are deferred to the softirq context by default.
There are few timers which expect to be run in hardirq context even on RT.
Those are:
- very short running where low latency is critical (kvm lapic)
- timers which take raw locks and need run in hard-irq context (perf, sched)
- wake up related timer (kernel side of clock_nanosleep() and so on)

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
---
 arch/x86/kvm/lapic.c                 |    2 +-
 include/linux/hrtimer.h              |    6 ++++++
 kernel/events/core.c                 |    4 ++--
 kernel/sched/core.c                  |    2 +-
 kernel/sched/deadline.c              |    2 +-
 kernel/sched/rt.c                    |    4 ++--
 kernel/time/hrtimer.c                |   21 +++++++++++++++++++--
 kernel/time/tick-broadcast-hrtimer.c |    2 +-
 kernel/time/tick-sched.c             |    2 +-
 kernel/watchdog.c                    |    2 +-
 10 files changed, 35 insertions(+), 12 deletions(-)

--- a/arch/x86/kvm/lapic.c
+++ b/arch/x86/kvm/lapic.c
@@ -2158,7 +2158,7 @@ int kvm_create_lapic(struct kvm_vcpu *vc
 	apic->vcpu = vcpu;
 
 	hrtimer_init(&apic->lapic_timer.timer, CLOCK_MONOTONIC,
-		     HRTIMER_MODE_ABS_PINNED);
+		     HRTIMER_MODE_ABS_PINNED_HARD);
 	apic->lapic_timer.timer.function = apic_timer_fn;
 
 	/*
--- a/include/linux/hrtimer.h
+++ b/include/linux/hrtimer.h
@@ -42,6 +42,7 @@ enum hrtimer_mode {
 	HRTIMER_MODE_REL	= 0x01,
 	HRTIMER_MODE_PINNED	= 0x02,
 	HRTIMER_MODE_SOFT	= 0x04,
+	HRTIMER_MODE_HARD	= 0x08,
 
 	HRTIMER_MODE_ABS_PINNED = HRTIMER_MODE_ABS | HRTIMER_MODE_PINNED,
 	HRTIMER_MODE_REL_PINNED = HRTIMER_MODE_REL | HRTIMER_MODE_PINNED,
@@ -52,6 +53,11 @@ enum hrtimer_mode {
 	HRTIMER_MODE_ABS_PINNED_SOFT = HRTIMER_MODE_ABS_PINNED | HRTIMER_MODE_SOFT,
 	HRTIMER_MODE_REL_PINNED_SOFT = HRTIMER_MODE_REL_PINNED | HRTIMER_MODE_SOFT,
 
+	HRTIMER_MODE_ABS_HARD	= HRTIMER_MODE_ABS | HRTIMER_MODE_HARD,
+	HRTIMER_MODE_REL_HARD	= HRTIMER_MODE_REL | HRTIMER_MODE_HARD,
+
+	HRTIMER_MODE_ABS_PINNED_HARD = HRTIMER_MODE_ABS_PINNED | HRTIMER_MODE_HARD,
+	HRTIMER_MODE_REL_PINNED_HARD = HRTIMER_MODE_REL_PINNED | HRTIMER_MODE_HARD,
 };
 
 /*
--- a/kernel/events/core.c
+++ b/kernel/events/core.c
@@ -1090,7 +1090,7 @@ static void __perf_mux_hrtimer_init(stru
 	cpuctx->hrtimer_interval = ns_to_ktime(NSEC_PER_MSEC * interval);
 
 	raw_spin_lock_init(&cpuctx->hrtimer_lock);
-	hrtimer_init(timer, CLOCK_MONOTONIC, HRTIMER_MODE_ABS_PINNED);
+	hrtimer_init(timer, CLOCK_MONOTONIC, HRTIMER_MODE_ABS_PINNED_HARD);
 	timer->function = perf_mux_hrtimer_handler;
 }
 
@@ -8683,7 +8683,7 @@ static void perf_swevent_init_hrtimer(st
 	if (!is_sampling_event(event))
 		return;
 
-	hrtimer_init(&hwc->hrtimer, CLOCK_MONOTONIC, HRTIMER_MODE_REL);
+	hrtimer_init(&hwc->hrtimer, CLOCK_MONOTONIC, HRTIMER_MODE_REL_HARD);
 	hwc->hrtimer.function = perf_swevent_hrtimer;
 
 	/*
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -344,7 +344,7 @@ static void init_rq_hrtick(struct rq *rq
 	rq->hrtick_csd.info = rq;
 #endif
 
-	hrtimer_init(&rq->hrtick_timer, CLOCK_MONOTONIC, HRTIMER_MODE_REL);
+	hrtimer_init(&rq->hrtick_timer, CLOCK_MONOTONIC, HRTIMER_MODE_REL_HARD);
 	rq->hrtick_timer.function = hrtick;
 }
 #else	/* CONFIG_SCHED_HRTICK */
--- a/kernel/sched/deadline.c
+++ b/kernel/sched/deadline.c
@@ -1057,7 +1057,7 @@ void init_dl_task_timer(struct sched_dl_
 {
 	struct hrtimer *timer = &dl_se->dl_timer;
 
-	hrtimer_init(timer, CLOCK_MONOTONIC, HRTIMER_MODE_REL);
+	hrtimer_init(timer, CLOCK_MONOTONIC, HRTIMER_MODE_REL_HARD);
 	timer->function = dl_task_timer;
 }
 
--- a/kernel/sched/rt.c
+++ b/kernel/sched/rt.c
@@ -47,8 +47,8 @@ void init_rt_bandwidth(struct rt_bandwid
 
 	raw_spin_lock_init(&rt_b->rt_runtime_lock);
 
-	hrtimer_init(&rt_b->rt_period_timer,
-			CLOCK_MONOTONIC, HRTIMER_MODE_REL);
+	hrtimer_init(&rt_b->rt_period_timer, CLOCK_MONOTONIC,
+		     HRTIMER_MODE_REL_HARD);
 	rt_b->rt_period_timer.function = sched_rt_period_timer;
 }
 
--- a/kernel/time/hrtimer.c
+++ b/kernel/time/hrtimer.c
@@ -1119,7 +1119,9 @@ void hrtimer_start_range_ns(struct hrtim
 	 * Check whether the HRTIMER_MODE_SOFT bit and hrtimer.is_soft
 	 * match.
 	 */
+#ifndef CONFIG_PREEMPT_RT_BASE
 	WARN_ON_ONCE(!(mode & HRTIMER_MODE_SOFT) ^ !timer->is_soft);
+#endif
 
 	base = lock_hrtimer_base(timer, &flags);
 
@@ -1246,10 +1248,17 @@ static inline int hrtimer_clockid_to_bas
 static void __hrtimer_init(struct hrtimer *timer, clockid_t clock_id,
 			   enum hrtimer_mode mode)
 {
-	bool softtimer = !!(mode & HRTIMER_MODE_SOFT);
-	int base = softtimer ? HRTIMER_MAX_CLOCK_BASES / 2 : 0;
+	bool softtimer;
+	int base;
 	struct hrtimer_cpu_base *cpu_base;
 
+	softtimer = !!(mode & HRTIMER_MODE_SOFT);
+#ifdef CONFIG_PREEMPT_RT_FULL
+	if (!softtimer && !(mode & HRTIMER_MODE_HARD))
+		softtimer = true;
+#endif
+	base = softtimer ? HRTIMER_MAX_CLOCK_BASES / 2 : 0;
+
 	memset(timer, 0, sizeof(struct hrtimer));
 
 	cpu_base = raw_cpu_ptr(&hrtimer_bases);
@@ -1633,6 +1642,14 @@ static void __hrtimer_init_sleeper(struc
 				   enum hrtimer_mode mode,
 				   struct task_struct *task)
 {
+#ifdef CONFIG_PREEMPT_RT_FULL
+	if (!(mode & (HRTIMER_MODE_SOFT | HRTIMER_MODE_HARD))) {
+		if (task_is_realtime(current) || system_state != SYSTEM_RUNNING)
+			mode |= HRTIMER_MODE_HARD;
+		else
+			mode |= HRTIMER_MODE_SOFT;
+	}
+#endif
 	__hrtimer_init(&sl->timer, clock_id, mode);
 	sl->timer.function = hrtimer_wakeup;
 	sl->task = task;
--- a/kernel/time/tick-broadcast-hrtimer.c
+++ b/kernel/time/tick-broadcast-hrtimer.c
@@ -106,7 +106,7 @@ static enum hrtimer_restart bc_handler(s
 
 void tick_setup_hrtimer_broadcast(void)
 {
-	hrtimer_init(&bctimer, CLOCK_MONOTONIC, HRTIMER_MODE_ABS);
+	hrtimer_init(&bctimer, CLOCK_MONOTONIC, HRTIMER_MODE_ABS_HARD);
 	bctimer.function = bc_handler;
 	clockevents_register_device(&ce_broadcast_hrtimer);
 }
--- a/kernel/time/tick-sched.c
+++ b/kernel/time/tick-sched.c
@@ -1231,7 +1231,7 @@ void tick_setup_sched_timer(void)
 	/*
 	 * Emulate tick processing via per-CPU hrtimers:
 	 */
-	hrtimer_init(&ts->sched_timer, CLOCK_MONOTONIC, HRTIMER_MODE_ABS);
+	hrtimer_init(&ts->sched_timer, CLOCK_MONOTONIC, HRTIMER_MODE_ABS_HARD);
 	ts->sched_timer.function = tick_sched_timer;
 
 	/* Get the next period (per-CPU) */
--- a/kernel/watchdog.c
+++ b/kernel/watchdog.c
@@ -463,7 +463,7 @@ static void watchdog_enable(unsigned int
 	 * Start the timer first to prevent the NMI watchdog triggering
 	 * before the timer has a chance to fire.
 	 */
-	hrtimer_init(hrtimer, CLOCK_MONOTONIC, HRTIMER_MODE_REL);
+	hrtimer_init(hrtimer, CLOCK_MONOTONIC, HRTIMER_MODE_REL_HARD);
 	hrtimer->function = watchdog_timer_fn;
 	hrtimer_start(hrtimer, ns_to_ktime(sample_period),
 		      HRTIMER_MODE_REL_PINNED);