[PATCH v2 7/7]powerpc/powernv: nest pmu cpumask and cpu hotplug support

[Madhavan Srinivasan]

Adds cpumask attribute to be used by each nest pmu since nest
units are per-chip. Only one cpu (first online cpu) from each node/chip
is designated to read counters.

On cpu hotplug, dying cpu is checked to see whether it is one of the
designated cpus, if yes, next online cpu from the same node/chip is designated
as new cpu to read counters.

Cc: Michael Ellerman <mpe at ellerman.id.au>
Cc: Benjamin Herrenschmidt <benh at kernel.crashing.org>
Cc: Paul Mackerras <paulus at samba.org>
Cc: Anton Blanchard <anton at samba.org>
Cc: Sukadev Bhattiprolu <sukadev at linux.vnet.ibm.com>
Cc: Anshuman Khandual <khandual at linux.vnet.ibm.com>
Cc: Stephane Eranian <eranian at google.com>
Cc: Preeti U Murthy <preeti at linux.vnet.ibm.com>
Cc: Ingo Molnar <mingo at kernel.org>
Cc: Peter Zijlstra <peterz at infradead.org>
Signed-off-by: Madhavan Srinivasan <maddy at linux.vnet.ibm.com>
---
 arch/powerpc/perf/nest-pmu.c | 153 +++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 153 insertions(+)

diff --git a/arch/powerpc/perf/nest-pmu.c b/arch/powerpc/perf/nest-pmu.c
index a662c14..95f8ecc 100644
--- a/arch/powerpc/perf/nest-pmu.c
+++ b/arch/powerpc/perf/nest-pmu.c
@@ -12,6 +12,155 @@
 
 static struct perchip_nest_info p8_perchip_nest_info[P8_MAX_CHIP];
 static struct nest_pmu *per_nest_pmu_arr[P8_MAX_NEST_PMUS];
+static cpumask_t cpu_mask_nest_pmu;
+
+static ssize_t cpumask_nest_pmu_get_attr(struct device *dev,
+				struct device_attribute *attr, char *buf)
+{
+	return cpumap_print_to_pagebuf(true, buf, &cpu_mask_nest_pmu);
+}
+
+static DEVICE_ATTR(cpumask, S_IRUGO, cpumask_nest_pmu_get_attr, NULL);
+
+static struct attribute *cpumask_nest_pmu_attrs[] = {
+	&dev_attr_cpumask.attr,
+	NULL,
+};
+
+static struct attribute_group cpumask_nest_pmu_attr_group = {
+	.attrs = cpumask_nest_pmu_attrs,
+};
+
+static void nest_init(void *dummy)
+{
+	opal_nest_ima_control(P8_NEST_ENGINE_START);
+}
+
+static void nest_change_cpu_context(int old_cpu, int new_cpu)
+{
+	int i;
+
+	if (new_cpu >= 0) {
+		for (i = 0; per_nest_pmu_arr[i] != NULL; i++)
+			perf_pmu_migrate_context(&per_nest_pmu_arr[i]->pmu,
+							old_cpu, new_cpu);
+	}
+}
+
+static void nest_exit_cpu(int cpu)
+{
+	int i, nid, target = -1;
+	const struct cpumask *l_cpumask;
+	int src_chipid;
+
+	/*
+	 * Check in the designated list for this cpu. Dont bother
+	 * if not one of them.
+	 */
+	if (!cpumask_test_and_clear_cpu(cpu, &cpu_mask_nest_pmu))
+		return;
+
+	/*
+	 * Now that this cpu is one of the designated,
+	 * find a new cpu a) which is not dying and
+	 * b) is in same node/chip.
+	 */
+	nid = cpu_to_node(cpu);
+	src_chipid = topology_physical_package_id(cpu);
+	l_cpumask = cpumask_of_node(nid);
+	for_each_cpu(i, l_cpumask) {
+		if (i == cpu)
+			continue;
+		if (src_chipid == topology_physical_package_id(i)) {
+			target = i;
+			break;
+		}
+	}
+
+	/*
+	 * Update the cpumask with the target cpu and
+	 * migrate the context if needed
+	 */
+	if (target >= 0) {
+		cpumask_set_cpu(target, &cpu_mask_nest_pmu);
+		nest_change_cpu_context (cpu, target);
+	}
+}
+
+static void nest_init_cpu(int cpu)
+{
+	int i, src_chipid;
+
+	/*
+	 * Search for any existing designated thread from
+	 * the incoming cpu's node/chip. If found, do nothing.
+	 */
+	src_chipid = topology_physical_package_id(cpu);
+	for_each_cpu(i, &cpu_mask_nest_pmu)
+		if (src_chipid == topology_physical_package_id(i))
+			return;
+
+	/*
+	 * Make incoming cpu as a designated thread for
+	 * this node/chip
+	 */
+	cpumask_set_cpu(cpu, &cpu_mask_nest_pmu);
+}
+
+static int nest_cpu_notifier(struct notifier_block *self,
+				unsigned long action, void *hcpu)
+{
+	long cpu = (long)hcpu;
+
+	switch (action & ~CPU_TASKS_FROZEN) {
+	case CPU_DOWN_FAILED:
+	case CPU_STARTING:
+		nest_init_cpu(cpu);
+		break;
+	case CPU_DOWN_PREPARE:
+		nest_exit_cpu(cpu);
+		break;
+	default:
+		break;
+	}
+
+	return NOTIFY_OK;
+}
+
+static struct notifier_block nest_cpu_nb = {
+	.notifier_call  = nest_cpu_notifier,
+	.priority       = CPU_PRI_PERF + 1,
+};
+
+void cpumask_chip(void)
+{
+	const struct cpumask *l_cpumask;
+	int cpu, nid;
+
+	if (!cpumask_empty(&cpu_mask_nest_pmu))
+		return;
+
+	cpu_notifier_register_begin();
+
+	/*
+	 * Nest PMUs are per-chip counters. So designate a cpu
+	 * from each node/chip for counter collection.
+	 */
+	for_each_online_node(nid) {
+		l_cpumask = cpumask_of_node(nid);
+
+		/* designate first online cpu in this node */
+		cpu = cpumask_first(l_cpumask);
+		cpumask_set_cpu(cpu, &cpu_mask_nest_pmu);
+	}
+
+	/* Initialize Nest PMUs in each node using designated cpus */
+	on_each_cpu_mask(&cpu_mask_nest_pmu, (smp_call_func_t)nest_init, NULL, 1);
+
+	__register_cpu_notifier(&nest_cpu_nb);
+
+	cpu_notifier_register_done();
+}
 
 PMU_FORMAT_ATTR(event, "config:0-20");
 struct attribute *p8_nest_format_attrs[] = {
@@ -209,6 +358,7 @@ static int nest_pmu_create(struct device_node *dev, int pmu_index)
 			sprintf(buf, "Nest_%s", (char *)pp->value);
 			pmu_ptr->pmu.name = (char *)buf;
 			pmu_ptr->attr_groups[1] = &p8_nest_format_group;
+			pmu_ptr->attr_groups[2] = &cpumask_nest_pmu_attr_group;
 		}
 
 		/* Skip these, we dont need it */
@@ -362,6 +512,9 @@ static int __init nest_pmu_init(void)
 	if (ret)
 		return ret;
 
+	/* Add cpumask and register for hotplug notification */
+	cpumask_chip();
+
 	return 0;
 }
 device_initcall(nest_pmu_init);
-- 
1.9.1