cpufreq: Call transition notifier only once for each policy

Currently, the notifiers are called once for each CPU of the policy->cpus
cpumask. It would be more optimal if the notifier can be called only
once and all the relevant information be provided to it. Out of the 23
drivers that register for the transition notifiers today, only 4 of them
do per-cpu updates and the callback for the rest can be called only once
for the policy without any impact.

This would also avoid multiple function calls to the notifier callbacks
and reduce multiple iterations of notifier core's code (which does
locking as well).

This patch adds pointer to the cpufreq policy to the struct
cpufreq_freqs, so the notifier callback has all the information
available to it with a single call. The five drivers which perform
per-cpu updates are updated to use the cpufreq policy. The freqs->cpu
field is redundant now and is removed.

Acked-by: David S. Miller <davem@davemloft.net> (sparc)
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>

arch/arm/kernel/smp.c

@@ -754,15 +754,20 @@ static int cpufreq_callback(struct notifier_block *nb,
 					unsigned long val, void *data)
 {
 	struct cpufreq_freqs *freq = data;
-	int cpu = freq->cpu;
+	struct cpumask *cpus = freq->policy->cpus;
+	int cpu, first = cpumask_first(cpus);
+	unsigned int lpj;
 
 	if (freq->flags & CPUFREQ_CONST_LOOPS)
 		return NOTIFY_OK;
 
-	if (!per_cpu(l_p_j_ref, cpu)) {
+	if (!per_cpu(l_p_j_ref, first)) {
+		for_each_cpu(cpu, cpus) {
 			per_cpu(l_p_j_ref, cpu) =
 				per_cpu(cpu_data, cpu).loops_per_jiffy;
 			per_cpu(l_p_j_ref_freq, cpu) = freq->old;
+		}
+
 		if (!global_l_p_j_ref) {
 			global_l_p_j_ref = loops_per_jiffy;
 			global_l_p_j_ref_freq = freq->old;
@@ -774,10 +779,11 @@ static int cpufreq_callback(struct notifier_block *nb,
 		loops_per_jiffy = cpufreq_scale(global_l_p_j_ref,
 						global_l_p_j_ref_freq,
 						freq->new);
-		per_cpu(cpu_data, cpu).loops_per_jiffy =
-			cpufreq_scale(per_cpu(l_p_j_ref, cpu),
-					per_cpu(l_p_j_ref_freq, cpu),
-					freq->new);
+
+		lpj = cpufreq_scale(per_cpu(l_p_j_ref, first),
+				    per_cpu(l_p_j_ref_freq, first), freq->new);
+		for_each_cpu(cpu, cpus)
+			per_cpu(cpu_data, cpu).loops_per_jiffy = lpj;
 	}
 	return NOTIFY_OK;
 }

arch/sparc/kernel/time_64.c

@@ -653,20 +653,24 @@ static int sparc64_cpufreq_notifier(struct notifier_block *nb, unsigned long val
 				    void *data)
 {
 	struct cpufreq_freqs *freq = data;
-	unsigned int cpu = freq->cpu;
-	struct freq_table *ft = &per_cpu(sparc64_freq_table, cpu);
+	unsigned int cpu;
+	struct freq_table *ft;
+
+	for_each_cpu(cpu, freq->policy->cpus) {
+		ft = &per_cpu(sparc64_freq_table, cpu);
 
 		if (!ft->ref_freq) {
 			ft->ref_freq = freq->old;
 			ft->clock_tick_ref = cpu_data(cpu).clock_tick;
 		}
+
 		if ((val == CPUFREQ_PRECHANGE  && freq->old < freq->new) ||
 		    (val == CPUFREQ_POSTCHANGE && freq->old > freq->new)) {
 			cpu_data(cpu).clock_tick =
-			cpufreq_scale(ft->clock_tick_ref,
-				      ft->ref_freq,
+				cpufreq_scale(ft->clock_tick_ref, ft->ref_freq,
 					      freq->new);
 		}
+	}
 
 	return 0;
 }

arch/x86/kernel/tsc.c

@@ -979,7 +979,7 @@ 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, rdtsc());
+		set_cyc2ns_scale(tsc_khz, freq->policy->cpu, rdtsc());
 	}
 
 	return 0;

arch/x86/kvm/x86.c

@@ -6698,10 +6698,8 @@ static void kvm_hyperv_tsc_notifier(void)
 }
 #endif
 
-static int kvmclock_cpufreq_notifier(struct notifier_block *nb, unsigned long val,
-				     void *data)
+static void __kvmclock_cpufreq_notifier(struct cpufreq_freqs *freq, int cpu)
 {
-	struct cpufreq_freqs *freq = data;
 	struct kvm *kvm;
 	struct kvm_vcpu *vcpu;
 	int i, send_ipi = 0;
@@ -6745,17 +6743,12 @@ static int kvmclock_cpufreq_notifier(struct notifier_block *nb, unsigned long va
 	 *
 	 */
 
-	if (val == CPUFREQ_PRECHANGE && freq->old > freq->new)
-		return 0;
-	if (val == CPUFREQ_POSTCHANGE && freq->old < freq->new)
-		return 0;
-
-	smp_call_function_single(freq->cpu, tsc_khz_changed, freq, 1);
+	smp_call_function_single(cpu, tsc_khz_changed, freq, 1);
 
 	spin_lock(&kvm_lock);
 	list_for_each_entry(kvm, &vm_list, vm_list) {
 		kvm_for_each_vcpu(i, vcpu, kvm) {
-			if (vcpu->cpu != freq->cpu)
+			if (vcpu->cpu != cpu)
 				continue;
 			kvm_make_request(KVM_REQ_CLOCK_UPDATE, vcpu);
 			if (vcpu->cpu != smp_processor_id())
@@ -6777,8 +6770,24 @@ static int kvmclock_cpufreq_notifier(struct notifier_block *nb, unsigned long va
 		 * guest context is entered kvmclock will be updated,
 		 * so the guest will not see stale values.
 		 */
-		smp_call_function_single(freq->cpu, tsc_khz_changed, freq, 1);
+		smp_call_function_single(cpu, tsc_khz_changed, freq, 1);
 	}
+}
+
+static int kvmclock_cpufreq_notifier(struct notifier_block *nb, unsigned long val,
+				     void *data)
+{
+	struct cpufreq_freqs *freq = data;
+	int cpu;
+
+	if (val == CPUFREQ_PRECHANGE && freq->old > freq->new)
+		return 0;
+	if (val == CPUFREQ_POSTCHANGE && freq->old < freq->new)
+		return 0;
+
+	for_each_cpu(cpu, freq->policy->cpus)
+		__kvmclock_cpufreq_notifier(freq, cpu);
+
 	return 0;
 }
 

drivers/cpufreq/cpufreq.c

@@ -340,11 +340,14 @@ static void cpufreq_notify_transition(struct cpufreq_policy *policy,
 				      struct cpufreq_freqs *freqs,
 				      unsigned int state)
 {
+	int cpu;
+
 	BUG_ON(irqs_disabled());
 
 	if (cpufreq_disabled())
 		return;
 
+	freqs->policy = policy;
 	freqs->flags = cpufreq_driver->flags;
 	pr_debug("notification %u of frequency transition to %u kHz\n",
 		 state, freqs->new);
@@ -364,10 +367,8 @@ static void cpufreq_notify_transition(struct cpufreq_policy *policy,
 			}
 		}
 
-		for_each_cpu(freqs->cpu, policy->cpus) {
 		srcu_notifier_call_chain(&cpufreq_transition_notifier_list,
 					 CPUFREQ_PRECHANGE, freqs);
-		}
 
 		adjust_jiffies(CPUFREQ_PRECHANGE, freqs);
 		break;
@@ -377,11 +378,11 @@ static void cpufreq_notify_transition(struct cpufreq_policy *policy,
 		pr_debug("FREQ: %u - CPUs: %*pbl\n", freqs->new,
 			 cpumask_pr_args(policy->cpus));
 
-		for_each_cpu(freqs->cpu, policy->cpus) {
-			trace_cpu_frequency(freqs->new, freqs->cpu);
+		for_each_cpu(cpu, policy->cpus)
+			trace_cpu_frequency(freqs->new, cpu);
+
 		srcu_notifier_call_chain(&cpufreq_transition_notifier_list,
 					 CPUFREQ_POSTCHANGE, freqs);
-		}
 
 		cpufreq_stats_record_transition(policy, freqs->new);
 		policy->cur = freqs->new;

include/linux/cpufreq.h

@@ -42,13 +42,6 @@ enum cpufreq_table_sorting {
 	CPUFREQ_TABLE_SORTED_DESCENDING
 };
 
-struct cpufreq_freqs {
-	unsigned int cpu;	/* cpu nr */
-	unsigned int old;
-	unsigned int new;
-	u8 flags;		/* flags of cpufreq_driver, see below. */
-};
-
 struct cpufreq_cpuinfo {
 	unsigned int		max_freq;
 	unsigned int		min_freq;
@@ -156,6 +149,13 @@ struct cpufreq_policy {
 	struct thermal_cooling_device *cdev;
 };
 
+struct cpufreq_freqs {
+	struct cpufreq_policy *policy;
+	unsigned int old;
+	unsigned int new;
+	u8 flags;		/* flags of cpufreq_driver, see below. */
+};
+
 /* Only for ACPI */
 #define CPUFREQ_SHARED_TYPE_NONE (0) /* None */
 #define CPUFREQ_SHARED_TYPE_HW	 (1) /* HW does needed coordination */