Merge branch 'sched-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip

Pull scheduler updates from Ingo Molnar:
 "The main changes in this cycle were:

   - Add the SYSTEM_SCHEDULING bootup state to move various scheduler
     debug checks earlier into the bootup. This turns silent and
     sporadically deadly bugs into nice, deterministic splats. Fix some
     of the splats that triggered. (Thomas Gleixner)

   - A round of restructuring and refactoring of the load-balancing and
     topology code (Peter Zijlstra)

   - Another round of consolidating ~20 of incremental scheduler code
     history: this time in terms of wait-queue nomenclature. (I didn't
     get much feedback on these renaming patches, and we can still
     easily change any names I might have misplaced, so if anyone hates
     a new name, please holler and I'll fix it.) (Ingo Molnar)

   - sched/numa improvements, fixes and updates (Rik van Riel)

   - Another round of x86/tsc scheduler clock code improvements, in hope
     of making it more robust (Peter Zijlstra)

   - Improve NOHZ behavior (Frederic Weisbecker)

   - Deadline scheduler improvements and fixes (Luca Abeni, Daniel
     Bristot de Oliveira)

   - Simplify and optimize the topology setup code (Lauro Ramos
     Venancio)

   - Debloat and decouple scheduler code some more (Nicolas Pitre)

   - Simplify code by making better use of llist primitives (Byungchul
     Park)

   - ... plus other fixes and improvements"

* 'sched-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (103 commits)
  sched/cputime: Refactor the cputime_adjust() code
  sched/debug: Expose the number of RT/DL tasks that can migrate
  sched/numa: Hide numa_wake_affine() from UP build
  sched/fair: Remove effective_load()
  sched/numa: Implement NUMA node level wake_affine()
  sched/fair: Simplify wake_affine() for the single socket case
  sched/numa: Override part of migrate_degrades_locality() when idle balancing
  sched/rt: Move RT related code from sched/core.c to sched/rt.c
  sched/deadline: Move DL related code from sched/core.c to sched/deadline.c
  sched/cpuset: Only offer CONFIG_CPUSETS if SMP is enabled
  sched/fair: Spare idle load balancing on nohz_full CPUs
  nohz: Move idle balancer registration to the idle path
  sched/loadavg: Generalize "_idle" naming to "_nohz"
  sched/core: Drop the unused try_get_task_struct() helper function
  sched/fair: WARN() and refuse to set buddy when !se->on_rq
  sched/debug: Fix SCHED_WARN_ON() to return a value on !CONFIG_SCHED_DEBUG as well
  sched/wait: Disambiguate wq_entry->task_list and wq_head->task_list naming
  sched/wait: Move bit_wait_table[] and related functionality from sched/core.c to sched/wait_bit.c
  sched/wait: Split out the wait_bit*() APIs from <linux/wait.h> into <linux/wait_bit.h>
  sched/wait: Re-adjust macro line continuation backslashes in <linux/wait.h>
  ...

9 files changed, 83 insertions, 170 deletions

diff --git a/arch/arm/kernel/smp.c b/arch/arm/kernel/smp.c
index 572a8df1b766..c9a0a5299827 100644
--- a/arch/arm/kernel/smp.c
+++ b/arch/arm/kernel/smp.c
@@ -555,8 +555,7 @@ static DEFINE_RAW_SPINLOCK(stop_lock);
  */
 static void ipi_cpu_stop(unsigned int cpu)
 {
-	if (system_state == SYSTEM_BOOTING ||
-	    system_state == SYSTEM_RUNNING) {
+	if (system_state <= SYSTEM_RUNNING) {
 		raw_spin_lock(&stop_lock);
 		pr_crit("CPU%u: stopping\n", cpu);
 		dump_stack();
diff --git a/arch/arm64/kernel/smp.c b/arch/arm64/kernel/smp.c
index 6e0e16a3a7d4..321119881abf 100644
--- a/arch/arm64/kernel/smp.c
+++ b/arch/arm64/kernel/smp.c
@@ -961,8 +961,7 @@ void smp_send_stop(void)
 		cpumask_copy(&mask, cpu_online_mask);
 		cpumask_clear_cpu(smp_processor_id(), &mask);
 
-		if (system_state == SYSTEM_BOOTING ||
-		    system_state == SYSTEM_RUNNING)
+		if (system_state <= SYSTEM_RUNNING)
 			pr_crit("SMP: stopping secondary CPUs\n");
 		smp_cross_call(&mask, IPI_CPU_STOP);
 	}
diff --git a/arch/metag/kernel/smp.c b/arch/metag/kernel/smp.c
index 232a12bf3f99..2dbbb7c66043 100644
--- a/arch/metag/kernel/smp.c
+++ b/arch/metag/kernel/smp.c
@@ -567,8 +567,7 @@ static void stop_this_cpu(void *data)
 {
 	unsigned int cpu = smp_processor_id();
 
-	if (system_state == SYSTEM_BOOTING ||
-	    system_state == SYSTEM_RUNNING) {
+	if (system_state <= SYSTEM_RUNNING) {
 		spin_lock(&stop_lock);
 		pr_crit("CPU%u: stopping\n", cpu);
 		dump_stack();
diff --git a/arch/powerpc/kernel/smp.c b/arch/powerpc/kernel/smp.c
index df2a41647d8e..1069f74fca47 100644
--- a/arch/powerpc/kernel/smp.c
+++ b/arch/powerpc/kernel/smp.c
@@ -97,7 +97,7 @@ int smp_generic_cpu_bootable(unsigned int nr)
 	/* Special case - we inhibit secondary thread startup
 	 * during boot if the user requests it.
 	 */
-	if (system_state == SYSTEM_BOOTING && cpu_has_feature(CPU_FTR_SMT)) {
+	if (system_state < SYSTEM_RUNNING && cpu_has_feature(CPU_FTR_SMT)) {
 		if (!smt_enabled_at_boot && cpu_thread_in_core(nr) != 0)
 			return 0;
 		if (smt_enabled_at_boot
diff --git a/arch/x86/events/core.c b/arch/x86/events/core.c
index e6f5e4b163ac..628b8c556aab 100644
--- a/arch/x86/events/core.c
+++ b/arch/x86/events/core.c
@@ -2265,7 +2265,7 @@ static struct pmu pmu = {
 void arch_perf_update_userpage(struct perf_event *event,
 			       struct perf_event_mmap_page *userpg, u64 now)
 {
-	struct cyc2ns_data *data;
+	struct cyc2ns_data data;
 	u64 offset;
 
 	userpg->cap_user_time = 0;
@@ -2277,17 +2277,17 @@ void arch_perf_update_userpage(struct perf_event *event,
 	if (!using_native_sched_clock() || !sched_clock_stable())
 		return;
 
-	data = cyc2ns_read_begin();
+	cyc2ns_read_begin(&data);
 
-	offset = data->cyc2ns_offset + __sched_clock_offset;
+	offset = data.cyc2ns_offset + __sched_clock_offset;
 
 	/*
 	 * Internal timekeeping for enabled/running/stopped times
 	 * is always in the local_clock domain.
 	 */
 	userpg->cap_user_time = 1;
-	userpg->time_mult = data->cyc2ns_mul;
-	userpg->time_shift = data->cyc2ns_shift;
+	userpg->time_mult = data.cyc2ns_mul;
+	userpg->time_shift = data.cyc2ns_shift;
 	userpg->time_offset = offset - now;
 
 	/*
@@ -2299,7 +2299,7 @@ void arch_perf_update_userpage(struct perf_event *event,
 		userpg->time_zero = offset;
 	}
 
-	cyc2ns_read_end(data);
+	cyc2ns_read_end();
 }
 
 void
diff --git a/arch/x86/include/asm/timer.h b/arch/x86/include/asm/timer.h
index 27e9f9d769b8..2016962103df 100644
--- a/arch/x86/include/asm/timer.h
+++ b/arch/x86/include/asm/timer.h
@@ -29,11 +29,9 @@ struct cyc2ns_data {
 	u32 cyc2ns_mul;
 	u32 cyc2ns_shift;
 	u64 cyc2ns_offset;
-	u32 __count;
-	/* u32 hole */
-}; /* 24 bytes -- do not grow */
+}; /* 16 bytes */
 
-extern struct cyc2ns_data *cyc2ns_read_begin(void);
-extern void cyc2ns_read_end(struct cyc2ns_data *);
+extern void cyc2ns_read_begin(struct cyc2ns_data *);
+extern void cyc2ns_read_end(void);
 
 #endif /* _ASM_X86_TIMER_H */
diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c
index f04479a8f74f..045e4f993bd2 100644
--- a/arch/x86/kernel/smpboot.c
+++ b/arch/x86/kernel/smpboot.c
@@ -863,7 +863,7 @@ static void announce_cpu(int cpu, int apicid)
 	if (cpu == 1)
 		printk(KERN_INFO "x86: Booting SMP configuration:\n");
 
-	if (system_state == SYSTEM_BOOTING) {
+	if (system_state < SYSTEM_RUNNING) {
 		if (node != current_node) {
 			if (current_node > (-1))
 				pr_cont("\n");
diff --git a/arch/x86/kernel/tsc.c b/arch/x86/kernel/tsc.c
index 714dfba6a1e7..5270fc0c2df6 100644
--- a/arch/x86/kernel/tsc.c
+++ b/arch/x86/kernel/tsc.c
@@ -51,115 +51,34 @@ static u32 art_to_tsc_denominator;
 static u64 art_to_tsc_offset;
 struct clocksource *art_related_clocksource;
 
-/*
- * Use a ring-buffer like data structure, where a writer advances the head by
- * writing a new data entry and a reader advances the tail when it observes a
- * new entry.
- *
- * Writers are made to wait on readers until there's space to write a new
- * entry.
- *
- * This means that we can always use an {offset, mul} pair to compute a ns
- * value that is 'roughly' in the right direction, even if we're writing a new
- * {offset, mul} pair during the clock read.
- *
- * The down-side is that we can no longer guarantee strict monotonicity anymore
- * (assuming the TSC was that to begin with), because while we compute the
- * intersection point of the two clock slopes and make sure the time is
- * continuous at the point of switching; we can no longer guarantee a reader is
- * strictly before or after the switch point.
- *
- * It does mean a reader no longer needs to disable IRQs in order to avoid
- * CPU-Freq updates messing with his times, and similarly an NMI reader will
- * no longer run the risk of hitting half-written state.
- */
-
 struct cyc2ns {
-	struct cyc2ns_data data[2];	/*  0 + 2*24 = 48 */
-	struct cyc2ns_data *head;	/* 48 + 8    = 56 */
-	struct cyc2ns_data *tail;	/* 56 + 8    = 64 */
-}; /* exactly fits one cacheline */
-
-static DEFINE_PER_CPU_ALIGNED(struct cyc2ns, cyc2ns);
-
-struct cyc2ns_data *cyc2ns_read_begin(void)
-{
-	struct cyc2ns_data *head;
-
-	preempt_disable();
-
-	head = this_cpu_read(cyc2ns.head);
-	/*
-	 * Ensure we observe the entry when we observe the pointer to it.
-	 * matches the wmb from cyc2ns_write_end().
-	 */
-	smp_read_barrier_depends();
-	head->__count++;
-	barrier();
+	struct cyc2ns_data data[2];	/*  0 + 2*16 = 32 */
+	seqcount_t	   seq;		/* 32 + 4    = 36 */
 
-	return head;
-}
+}; /* fits one cacheline */
 
-void cyc2ns_read_end(struct cyc2ns_data *head)
-{
-	barrier();
-	/*
-	 * If we're the outer most nested read; update the tail pointer
-	 * when we're done. This notifies possible pending writers
-	 * that we've observed the head pointer and that the other
-	 * entry is now free.
-	 */
-	if (!--head->__count) {
-		/*
-		 * x86-TSO does not reorder writes with older reads;
-		 * therefore once this write becomes visible to another
-		 * cpu, we must be finished reading the cyc2ns_data.
-		 *
-		 * matches with cyc2ns_write_begin().
-		 */
-		this_cpu_write(cyc2ns.tail, head);
-	}
-	preempt_enable();
-}
+static DEFINE_PER_CPU_ALIGNED(struct cyc2ns, cyc2ns);
 
-/*
- * Begin writing a new @data entry for @cpu.
- *
- * Assumes some sort of write side lock; currently 'provided' by the assumption
- * that cpufreq will call its notifiers sequentially.
- */
-static struct cyc2ns_data *cyc2ns_write_begin(int cpu)
+void cyc2ns_read_begin(struct cyc2ns_data *data)
 {
-	struct cyc2ns *c2n = &per_cpu(cyc2ns, cpu);
-	struct cyc2ns_data *data = c2n->data;
+	int seq, idx;
 
-	if (data == c2n->head)
-		data++;
+	preempt_disable_notrace();
 
-	/* XXX send an IPI to @cpu in order to guarantee a read? */
+	do {
+		seq = this_cpu_read(cyc2ns.seq.sequence);
+		idx = seq & 1;
 
-	/*
-	 * When we observe the tail write from cyc2ns_read_end(),
-	 * the cpu must be done with that entry and its safe
-	 * to start writing to it.
-	 */
-	while (c2n->tail == data)
-		cpu_relax();
+		data->cyc2ns_offset = this_cpu_read(cyc2ns.data[idx].cyc2ns_offset);
+		data->cyc2ns_mul    = this_cpu_read(cyc2ns.data[idx].cyc2ns_mul);
+		data->cyc2ns_shift  = this_cpu_read(cyc2ns.data[idx].cyc2ns_shift);
 
-	return data;
+	} while (unlikely(seq != this_cpu_read(cyc2ns.seq.sequence)));
 }
 
-static void cyc2ns_write_end(int cpu, struct cyc2ns_data *data)
+void cyc2ns_read_end(void)
 {
-	struct cyc2ns *c2n = &per_cpu(cyc2ns, cpu);
-
-	/*
-	 * Ensure the @data writes are visible before we publish the
-	 * entry. Matches the data-depencency in cyc2ns_read_begin().
-	 */
-	smp_wmb();
-
-	ACCESS_ONCE(c2n->head) = data;
+	preempt_enable_notrace();
 }
 
 /*
@@ -191,7 +110,6 @@ static void cyc2ns_data_init(struct cyc2ns_data *data)
 	data->cyc2ns_mul = 0;
 	data->cyc2ns_shift = 0;
 	data->cyc2ns_offset = 0;
-	data->__count = 0;
 }
 
 static void cyc2ns_init(int cpu)
@@ -201,51 +119,29 @@ static void cyc2ns_init(int cpu)
 	cyc2ns_data_init(&c2n->data[0]);
 	cyc2ns_data_init(&c2n->data[1]);
 
-	c2n->head = c2n->data;
-	c2n->tail = c2n->data;
+	seqcount_init(&c2n->seq);
 }
 
 static inline unsigned long long cycles_2_ns(unsigned long long cyc)
 {
-	struct cyc2ns_data *data, *tail;
+	struct cyc2ns_data data;
 	unsigned long long ns;
 
-	/*
-	 * See cyc2ns_read_*() for details; replicated in order to avoid
-	 * an extra few instructions that came with the abstraction.
-	 * Notable, it allows us to only do the __count and tail update
-	 * dance when its actually needed.
-	 */
-
-	preempt_disable_notrace();
-	data = this_cpu_read(cyc2ns.head);
-	tail = this_cpu_read(cyc2ns.tail);
-
-	if (likely(data == tail)) {
-		ns = data->cyc2ns_offset;
-		ns += mul_u64_u32_shr(cyc, data->cyc2ns_mul, data->cyc2ns_shift);
-	} else {
-		data->__count++;
-
-		barrier();
-
-		ns = data->cyc2ns_offset;
-		ns += mul_u64_u32_shr(cyc, data->cyc2ns_mul, data->cyc2ns_shift);
+	cyc2ns_read_begin(&data);
 
-		barrier();
+	ns = data.cyc2ns_offset;
+	ns += mul_u64_u32_shr(cyc, data.cyc2ns_mul, data.cyc2ns_shift);
 
-		if (!--data->__count)
-			this_cpu_write(cyc2ns.tail, data);
-	}
-	preempt_enable_notrace();
+	cyc2ns_read_end();
 
 	return ns;
 }
 
-static void set_cyc2ns_scale(unsigned long khz, int cpu)
+static void set_cyc2ns_scale(unsigned long khz, int cpu, unsigned long long tsc_now)
 {
-	unsigned long long tsc_now, ns_now;
-	struct cyc2ns_data *data;
+	unsigned long long ns_now;
+	struct cyc2ns_data data;
+	struct cyc2ns *c2n;
 	unsigned long flags;
 
 	local_irq_save(flags);
@@ -254,9 +150,6 @@ static void set_cyc2ns_scale(unsigned long khz, int cpu)
 	if (!khz)
 		goto done;
 
-	data = cyc2ns_write_begin(cpu);
-
-	tsc_now = rdtsc();
 	ns_now = cycles_2_ns(tsc_now);
 
 	/*
@@ -264,7 +157,7 @@ static void set_cyc2ns_scale(unsigned long khz, int cpu)
 	 * time function is continuous; see the comment near struct
 	 * cyc2ns_data.
 	 */
-	clocks_calc_mult_shift(&data->cyc2ns_mul, &data->cyc2ns_shift, khz,
+	clocks_calc_mult_shift(&data.cyc2ns_mul, &data.cyc2ns_shift, khz,
 			       NSEC_PER_MSEC, 0);
 
 	/*
@@ -273,20 +166,26 @@ static void set_cyc2ns_scale(unsigned long khz, int cpu)
 	 * conversion algorithm shifting a 32-bit value (now specifies a 64-bit
 	 * value) - refer perf_event_mmap_page documentation in perf_event.h.
 	 */
-	if (data->cyc2ns_shift == 32) {
-		data->cyc2ns_shift = 31;
-		data->cyc2ns_mul >>= 1;
+	if (data.cyc2ns_shift == 32) {
+		data.cyc2ns_shift = 31;
+		data.cyc2ns_mul >>= 1;
 	}
 
-	data->cyc2ns_offset = ns_now -
-		mul_u64_u32_shr(tsc_now, data->cyc2ns_mul, data->cyc2ns_shift);
+	data.cyc2ns_offset = ns_now -
+		mul_u64_u32_shr(tsc_now, data.cyc2ns_mul, data.cyc2ns_shift);
+
+	c2n = per_cpu_ptr(&cyc2ns, cpu);
 
-	cyc2ns_write_end(cpu, data);
+	raw_write_seqcount_latch(&c2n->seq);
+	c2n->data[0] = data;
+	raw_write_seqcount_latch(&c2n->seq);
+	c2n->data[1] = data;
 
 done:
-	sched_clock_idle_wakeup_event(0);
+	sched_clock_idle_wakeup_event();
 	local_irq_restore(flags);
 }
+
 /*
  * Scheduler clock - returns current time in nanosec units.
  */
@@ -374,6 +273,8 @@ static int __init tsc_setup(char *str)
 		tsc_clocksource_reliable = 1;
 	if (!strncmp(str, "noirqtime", 9))
 		no_sched_irq_time = 1;
+	if (!strcmp(str, "unstable"))
+		mark_tsc_unstable("boot parameter");
 	return 1;
 }
 
@@ -986,7 +887,6 @@ void tsc_restore_sched_clock_state(void)
 }
 
 #ifdef CONFIG_CPU_FREQ
-
 /* Frequency scaling support. Adjust the TSC based timer when the cpu frequency
  * changes.
  *
@@ -1027,7 +927,7 @@ static int time_cpufreq_notifier(struct notifier_block *nb, unsigned long val,
 		if (!(freq->flags & CPUFREQ_CONST_LOOPS))
 			mark_tsc_unstable("cpufreq changes");
 
-		set_cyc2ns_scale(tsc_khz, freq->cpu);
+		set_cyc2ns_scale(tsc_khz, freq->cpu, rdtsc());
 	}
 
 	return 0;
@@ -1127,6 +1027,15 @@ static void tsc_cs_mark_unstable(struct clocksource *cs)
 	pr_info("Marking TSC unstable due to clocksource watchdog\n");
 }
 
+static void tsc_cs_tick_stable(struct clocksource *cs)
+{
+	if (tsc_unstable)
+		return;
+
+	if (using_native_sched_clock())
+		sched_clock_tick_stable();
+}
+
 /*
  * .mask MUST be CLOCKSOURCE_MASK(64). See comment above read_tsc()
  */
@@ -1140,6 +1049,7 @@ static struct clocksource clocksource_tsc = {
 	.archdata               = { .vclock_mode = VCLOCK_TSC },
 	.resume			= tsc_resume,
 	.mark_unstable		= tsc_cs_mark_unstable,
+	.tick_stable		= tsc_cs_tick_stable,
 };
 
 void mark_tsc_unstable(char *reason)
@@ -1255,6 +1165,7 @@ static void tsc_refine_calibration_work(struct work_struct *work)
 	static int hpet;
 	u64 tsc_stop, ref_stop, delta;
 	unsigned long freq;
+	int cpu;
 
 	/* Don't bother refining TSC on unstable systems */
 	if (check_tsc_unstable())
@@ -1305,6 +1216,10 @@ static void tsc_refine_calibration_work(struct work_struct *work)
 	/* Inform the TSC deadline clockevent devices about the recalibration */
 	lapic_update_tsc_freq();
 
+	/* Update the sched_clock() rate to match the clocksource one */
+	for_each_possible_cpu(cpu)
+		set_cyc2ns_scale(tsc_khz, cpu, tsc_stop);
+
 out:
 	if (boot_cpu_has(X86_FEATURE_ART))
 		art_related_clocksource = &clocksource_tsc;
@@ -1350,7 +1265,7 @@ device_initcall(init_tsc_clocksource);
 
 void __init tsc_init(void)
 {
-	u64 lpj;
+	u64 lpj, cyc;
 	int cpu;
 
 	if (!boot_cpu_has(X86_FEATURE_TSC)) {
@@ -1390,9 +1305,10 @@ void __init tsc_init(void)
 	 * speed as the bootup CPU. (cpufreq notifiers will fix this
 	 * up if their speed diverges)
 	 */
+	cyc = rdtsc();
 	for_each_possible_cpu(cpu) {
 		cyc2ns_init(cpu);
-		set_cyc2ns_scale(tsc_khz, cpu);
+		set_cyc2ns_scale(tsc_khz, cpu, cyc);
 	}
 
 	if (tsc_disabled > 0)
diff --git a/arch/x86/platform/uv/tlb_uv.c b/arch/x86/platform/uv/tlb_uv.c
index 42e65fee5673..795671593528 100644
--- a/arch/x86/platform/uv/tlb_uv.c
+++ b/arch/x86/platform/uv/tlb_uv.c
@@ -456,12 +456,13 @@ static void reset_with_ipi(struct pnmask *distribution, struct bau_control *bcp)
  */
 static inline unsigned long long cycles_2_ns(unsigned long long cyc)
 {
-	struct cyc2ns_data *data = cyc2ns_read_begin();
+	struct cyc2ns_data data;
 	unsigned long long ns;
 
-	ns = mul_u64_u32_shr(cyc, data->cyc2ns_mul, data->cyc2ns_shift);
+	cyc2ns_read_begin(&data);
+	ns = mul_u64_u32_shr(cyc, data.cyc2ns_mul, data.cyc2ns_shift);
+	cyc2ns_read_end();
 
-	cyc2ns_read_end(data);
 	return ns;
 }
 
@@ -470,12 +471,13 @@ static inline unsigned long long cycles_2_ns(unsigned long long cyc)
  */
 static inline unsigned long long ns_2_cycles(unsigned long long ns)
 {
-	struct cyc2ns_data *data = cyc2ns_read_begin();
+	struct cyc2ns_data data;
 	unsigned long long cyc;
 
-	cyc = (ns << data->cyc2ns_shift) / data->cyc2ns_mul;
+	cyc2ns_read_begin(&data);
+	cyc = (ns << data.cyc2ns_shift) / data.cyc2ns_mul;
+	cyc2ns_read_end();
 
-	cyc2ns_read_end(data);
 	return cyc;
 }