blk-throttle: Make dispatch stats per cpu

Currently we take blkg_stat lock for even updating the stats. So even if
a group has no throttling rules (common case for root group), we end
up taking blkg_lock, for updating the stats.

Make dispatch stats per cpu so that these can be updated without taking
blkg lock.

If cpu goes offline, these stats simply disappear. No protection has
been provided for that yet. Do we really need anything for that?

Signed-off-by: Vivek Goyal <vgoyal@redhat.com>
Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
diff --git a/block/cfq-iosched.c b/block/cfq-iosched.c
index 606020f..d646b27 100644
--- a/block/cfq-iosched.c
+++ b/block/cfq-iosched.c
@@ -1051,7 +1051,7 @@
 static struct cfq_group * cfq_alloc_cfqg(struct cfq_data *cfqd)
 {
 	struct cfq_group *cfqg = NULL;
-	int i, j;
+	int i, j, ret;
 	struct cfq_rb_root *st;
 
 	cfqg = kzalloc_node(sizeof(*cfqg), GFP_ATOMIC, cfqd->queue->node);
@@ -1069,6 +1069,13 @@
 	 * or cgroup deletion path depending on who is exiting first.
 	 */
 	cfqg->ref = 1;
+
+	ret = blkio_alloc_blkg_stats(&cfqg->blkg);
+	if (ret) {
+		kfree(cfqg);
+		return NULL;
+	}
+
 	return cfqg;
 }
 
@@ -1183,6 +1190,7 @@
 		return;
 	for_each_cfqg_st(cfqg, i, j, st)
 		BUG_ON(!RB_EMPTY_ROOT(&st->rb));
+	free_percpu(cfqg->blkg.stats_cpu);
 	kfree(cfqg);
 }
 
@@ -3995,7 +4003,15 @@
 	 * throtl_data goes away.
 	 */
 	cfqg->ref = 2;
+
+	if (blkio_alloc_blkg_stats(&cfqg->blkg)) {
+		kfree(cfqg);
+		kfree(cfqd);
+		return NULL;
+	}
+
 	rcu_read_lock();
+
 	cfq_blkiocg_add_blkio_group(&blkio_root_cgroup, &cfqg->blkg,
 					(void *)cfqd, 0);
 	rcu_read_unlock();