tracing/sched: add load balancer tracepoint

When doing performance analysis it can be useful to see exactly
what is going on with the load balancer - when it runs and why
exactly it may not be redistributing load.

This additional tracepoint will show the idle context of the
load balance operation (idle, not idle, newly idle), various
values from the load balancing operation, the final result,
and the new balance interval.

Change-Id: I1538c411c5f9d17d7d37d84ead6210756be2d884
Signed-off-by: Steve Muckle <smuckle@codeaurora.org>
[rameezmustafa@codeaurora.org: Initialize variables in load_balance() to
				avoid crashes and inaccurate tracing.]
Signed-off-by: Syed Rameez Mustafa <rameezmustafa@codeaurora.org>
[joonwoop@codeaurora.org: fixed minor conflict in
 include/trace/events/sched.h.]
Signed-off-by: Joonwoo Park <joonwoop@codeaurora.org>

2 files changed, 59 insertions, 3 deletions

diff --git a/include/trace/events/sched.h b/include/trace/events/sched.h
index f6ce9e64c46c..04546537ff15 100644
--- a/include/trace/events/sched.h
+++ b/include/trace/events/sched.h
@@ -254,6 +254,56 @@ TRACE_EVENT(sched_cpu_hotplug,
 		__entry->status ? "online" : "offline", __entry->error)
 );
 
+/*
+ * Tracepoint for load balancing:
+ */
+#if NR_CPUS > 32
+#error "Unsupported NR_CPUS for lb tracepoint."
+#endif
+TRACE_EVENT(sched_load_balance,
+
+	TP_PROTO(int cpu, enum cpu_idle_type idle, int balance,
+		 unsigned long group_mask, int busiest_nr_running,
+		 unsigned long imbalance, unsigned int env_flags, int ld_moved,
+		 unsigned int balance_interval),
+
+	TP_ARGS(cpu, idle, balance, group_mask, busiest_nr_running,
+		imbalance, env_flags, ld_moved, balance_interval),
+
+	TP_STRUCT__entry(
+		__field(	int,			cpu)
+		__field(	enum cpu_idle_type,	idle)
+		__field(	int,			balance)
+		__field(	unsigned long,		group_mask)
+		__field(	int,			busiest_nr_running)
+		__field(	unsigned long,		imbalance)
+		__field(	unsigned int,		env_flags)
+		__field(	int,			ld_moved)
+		__field(	unsigned int,		balance_interval)
+	),
+
+	TP_fast_assign(
+		__entry->cpu			= cpu;
+		__entry->idle			= idle;
+		__entry->balance		= balance;
+		__entry->group_mask		= group_mask;
+		__entry->busiest_nr_running	= busiest_nr_running;
+		__entry->imbalance		= imbalance;
+		__entry->env_flags		= env_flags;
+		__entry->ld_moved		= ld_moved;
+		__entry->balance_interval	= balance_interval;
+	),
+
+	TP_printk("cpu=%d state=%s balance=%d group=%#lx busy_nr=%d imbalance=%ld flags=%#x ld_moved=%d bal_int=%d",
+		  __entry->cpu,
+		  __entry->idle == CPU_IDLE ? "idle" :
+		  (__entry->idle == CPU_NEWLY_IDLE ? "newly_idle" : "busy"),
+		  __entry->balance,
+		  __entry->group_mask, __entry->busiest_nr_running,
+		  __entry->imbalance, __entry->env_flags, __entry->ld_moved,
+		  __entry->balance_interval)
+);
+
 DECLARE_EVENT_CLASS(sched_process_template,
 
 	TP_PROTO(struct task_struct *p),
diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index 8ab12e9bac9f..917d0d4eaa71 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -6966,10 +6966,10 @@ static int load_balance(int this_cpu, struct rq *this_rq,
 			struct sched_domain *sd, enum cpu_idle_type idle,
 			int *continue_balancing)
 {
-	int ld_moved, cur_ld_moved, active_balance = 0;
+	int ld_moved = 0, cur_ld_moved, active_balance = 0;
 	struct sched_domain *sd_parent = sd->parent;
-	struct sched_group *group;
-	struct rq *busiest;
+	struct sched_group *group = NULL;
+	struct rq *busiest = NULL;
 	unsigned long flags;
 	struct cpumask *cpus = this_cpu_cpumask_var_ptr(load_balance_mask);
 
@@ -6983,6 +6983,7 @@ static int load_balance(int this_cpu, struct rq *this_rq,
 		.cpus		= cpus,
 		.fbq_type	= all,
 		.tasks		= LIST_HEAD_INIT(env.tasks),
+		.imbalance	= 0,
 	};
 
 	/*
@@ -7227,6 +7228,11 @@ out_one_pinned:
 
 	ld_moved = 0;
 out:
+	trace_sched_load_balance(this_cpu, idle, *continue_balancing,
+				 group ? group->cpumask[0] : 0,
+				 busiest ? busiest->nr_running : 0,
+				 env.imbalance, env.flags, ld_moved,
+				 sd->balance_interval);
 	return ld_moved;
 }