sched: convert local_cpu_mask to cpumask_var_t.

Impact: (future) size reduction for large NR_CPUS.

Dynamically allocating cpumasks (when CONFIG_CPUMASK_OFFSTACK) saves
space for small nr_cpu_ids but big CONFIG_NR_CPUS.  cpumask_var_t
is just a struct cpumask for !CONFIG_CPUMASK_OFFSTACK.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
diff --git a/kernel/sched_rt.c b/kernel/sched_rt.c
index 1fa1362..1f0e99d 100644
--- a/kernel/sched_rt.c
+++ b/kernel/sched_rt.c
@@ -962,7 +962,7 @@
 	return next;
 }
 
-static DEFINE_PER_CPU(cpumask_t, local_cpu_mask);
+static DEFINE_PER_CPU(cpumask_var_t, local_cpu_mask);
 
 static inline int pick_optimal_cpu(int this_cpu, cpumask_t *mask)
 {
@@ -982,7 +982,7 @@
 static int find_lowest_rq(struct task_struct *task)
 {
 	struct sched_domain *sd;
-	cpumask_t *lowest_mask = &__get_cpu_var(local_cpu_mask);
+	cpumask_t *lowest_mask = __get_cpu_var(local_cpu_mask);
 	int this_cpu = smp_processor_id();
 	int cpu      = task_cpu(task);
 
@@ -1551,3 +1551,12 @@
 	rcu_read_unlock();
 }
 #endif /* CONFIG_SCHED_DEBUG */
+
+/* Note that this is never called for !SMP, but that's OK. */
+static inline void init_sched_rt_class(void)
+{
+	unsigned int i;
+
+	for_each_possible_cpu(i)
+		alloc_cpumask_var(&per_cpu(local_cpu_mask, i), GFP_KERNEL);
+}