sched_clock: fix cpu_clock()

This patch fixes 3 issues:

a) it removes the dependency on jiffies, because jiffies are incremented
   by a single CPU, and the tick is not synchronized between CPUs. Therefore
   relying on it to calculate a window to clip whacky TSC values doesn't work
   as it can drift around.

   So instead use [GTOD, GTOD+TICK_NSEC) as the window.

b) __update_sched_clock() did (roughly speaking):

   delta = sched_clock() - scd->tick_raw;
   clock += delta;

   Which gives exponential growth, instead of linear.

c) allows the sched_clock_cpu() value to warp the u64 without breaking.

the results are more reliable sched_clock() deltas:

           before       after   sched_clock

cpu_clock: 15750        51312   51488
cpu_clock: 59719        51052   50947
cpu_clock: 15879        51249   51061
cpu_clock: 1            50933   51198
cpu_clock: 1            50931   51039
cpu_clock: 1            51093   50981
cpu_clock: 1            51043   51040
cpu_clock: 1            50959   50938
cpu_clock: 1            50981   51011
cpu_clock: 1            51364   51212
cpu_clock: 1            51219   51273
cpu_clock: 1            51389   51048
cpu_clock: 1            51285   51611
cpu_clock: 1            50964   51137
cpu_clock: 1            50973   50968
cpu_clock: 1            50967   50972
cpu_clock: 1            58910   58485
cpu_clock: 1            51082   51025
cpu_clock: 1            50957   50958
cpu_clock: 1            50958   50957
cpu_clock: 1006128      51128   50971
cpu_clock: 1            51107   51155
cpu_clock: 1            51371   51081
cpu_clock: 1            51104   51365
cpu_clock: 1            51363   51309
cpu_clock: 1            51107   51160
cpu_clock: 1            51139   51100
cpu_clock: 1            51216   51136
cpu_clock: 1            51207   51215
cpu_clock: 1            51087   51263
cpu_clock: 1            51249   51177
cpu_clock: 1            51519   51412
cpu_clock: 1            51416   51255
cpu_clock: 1            51591   51594
cpu_clock: 1            50966   51374
cpu_clock: 1            50966   50966
cpu_clock: 1            51291   50948
cpu_clock: 1            50973   50867
cpu_clock: 1            50970   50970
cpu_clock: 998306       50970   50971
cpu_clock: 1            50971   50970
cpu_clock: 1            50970   50970
cpu_clock: 1            50971   50971
cpu_clock: 1            50970   50970
cpu_clock: 1            51351   50970
cpu_clock: 1            50970   51352
cpu_clock: 1            50971   50970
cpu_clock: 1            50970   50970
cpu_clock: 1            51321   50971
cpu_clock: 1            50974   51324
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

kernel/sched_clock.c

@@ -12,19 +12,17 @@
  *
  * Create a semi stable clock from a mixture of other events, including:
  *  - gtod
- *  - jiffies
  *  - sched_clock()
  *  - explicit idle events
  *
  * We use gtod as base and the unstable clock deltas. The deltas are filtered,
- * making it monotonic and keeping it within an expected window.  This window
- * is set up using jiffies.
+ * making it monotonic and keeping it within an expected window.
  *
  * Furthermore, explicit sleep and wakeup hooks allow us to account for time
  * that is otherwise invisible (TSC gets stopped).
  *
  * The clock: sched_clock_cpu() is monotonic per cpu, and should be somewhat
- * consistent between cpus (never more than 1 jiffies difference).
+ * consistent between cpus (never more than 2 jiffies difference).
  */
 #include <linux/sched.h>
 #include <linux/percpu.h>
@@ -54,7 +52,6 @@ struct sched_clock_data {
 	 */
 	raw_spinlock_t		lock;
 
-	unsigned long		tick_jiffies;
 	u64			tick_raw;
 	u64			tick_gtod;
 	u64			clock;
@@ -75,14 +72,12 @@ static inline struct sched_clock_data *cpu_sdc(int cpu)
 void sched_clock_init(void)
 {
 	u64 ktime_now = ktime_to_ns(ktime_get());
-	unsigned long now_jiffies = jiffies;
 	int cpu;
 
 	for_each_possible_cpu(cpu) {
 		struct sched_clock_data *scd = cpu_sdc(cpu);
 
 		scd->lock = (raw_spinlock_t)__RAW_SPIN_LOCK_UNLOCKED;
-		scd->tick_jiffies = now_jiffies;
 		scd->tick_raw = 0;
 		scd->tick_gtod = ktime_now;
 		scd->clock = ktime_now;
@@ -91,47 +86,52 @@ void sched_clock_init(void)
 	sched_clock_running = 1;
 }
 
+/*
+ * min,max except they take wrapping into account
+ */
+
+static inline u64 wrap_min(u64 x, u64 y)
+{
+	return (s64)(x - y) < 0 ? x : y;
+}
+
+static inline u64 wrap_max(u64 x, u64 y)
+{
+	return (s64)(x - y) > 0 ? x : y;
+}
+
 /*
  * update the percpu scd from the raw @now value
  *
  *  - filter out backward motion
- *  - use jiffies to generate a min,max window to clip the raw values
+ *  - use the GTOD tick value to create a window to filter crazy TSC values
  */
 static u64 __update_sched_clock(struct sched_clock_data *scd, u64 now)
 {
-	unsigned long now_jiffies = jiffies;
-	long delta_jiffies = now_jiffies - scd->tick_jiffies;
-	u64 clock = scd->clock;
-	u64 min_clock, max_clock;
 	s64 delta = now - scd->tick_raw;
+	u64 clock, min_clock, max_clock;
 
 	WARN_ON_ONCE(!irqs_disabled());
-	min_clock = scd->tick_gtod + delta_jiffies * TICK_NSEC;
 
-	if (unlikely(delta < 0)) {
-		clock++;
-		goto out;
-	}
+	if (unlikely(delta < 0))
+		delta = 0;
 
-	max_clock = min_clock + TICK_NSEC;
+	/*
+	 * scd->clock = clamp(scd->tick_gtod + delta,
+	 * 		      max(scd->tick_gtod, scd->clock),
+	 * 		      scd->tick_gtod + TICK_NSEC);
+	 */
 
-	if (unlikely(clock + delta > max_clock)) {
-		if (clock < max_clock)
-			clock = max_clock;
-		else
-			clock++;
-	} else {
-		clock += delta;
-	}
+	clock = scd->tick_gtod + delta;
+	min_clock = wrap_max(scd->tick_gtod, scd->clock);
+	max_clock = scd->tick_gtod + TICK_NSEC;
 
- out:
-	if (unlikely(clock < min_clock))
-		clock = min_clock;
+	clock = wrap_max(clock, min_clock);
+	clock = wrap_min(clock, max_clock);
 
-	scd->tick_jiffies = now_jiffies;
 	scd->clock = clock;
 
-	return clock;
+	return scd->clock;
 }
 
 static void lock_double_clock(struct sched_clock_data *data1,
@@ -171,7 +171,7 @@ u64 sched_clock_cpu(int cpu)
 		 * larger time as the latest time for both
 		 * runqueues. (this creates monotonic movement)
 		 */
-		if (likely(remote_clock < this_clock)) {
+		if (likely((s64)(remote_clock - this_clock) < 0)) {
 			clock = this_clock;
 			scd->clock = clock;
 		} else {
@@ -207,14 +207,9 @@ void sched_clock_tick(void)
 	now = sched_clock();
 
 	__raw_spin_lock(&scd->lock);
-	__update_sched_clock(scd, now);
-	/*
-	 * update tick_gtod after __update_sched_clock() because that will
-	 * already observe 1 new jiffy; adding a new tick_gtod to that would
-	 * increase the clock 2 jiffies.
-	 */
 	scd->tick_raw = now;
 	scd->tick_gtod = now_gtod;
+	__update_sched_clock(scd, now);
 	__raw_spin_unlock(&scd->lock);
 }
 
@@ -232,18 +227,7 @@ EXPORT_SYMBOL_GPL(sched_clock_idle_sleep_event);
  */
 void sched_clock_idle_wakeup_event(u64 delta_ns)
 {
-	struct sched_clock_data *scd = this_scd();
-
-	/*
-	 * Override the previous timestamp and ignore all
-	 * sched_clock() deltas that occured while we idled,
-	 * and use the PM-provided delta_ns to advance the
-	 * rq clock:
-	 */
-	__raw_spin_lock(&scd->lock);
-	scd->clock += delta_ns;
-	__raw_spin_unlock(&scd->lock);
-
+	sched_clock_tick();
 	touch_softlockup_watchdog();
 }
 EXPORT_SYMBOL_GPL(sched_clock_idle_wakeup_event);