rcu: Add event-tracing for RCU callback invocation

There was recently some controversy about the overhead of invoking RCU
callbacks.  Add TRACE_EVENT()s to obtain fine-grained timings for the
start and stop of a batch of callbacks and also for each callback invoked.

Signed-off-by: Paul E. McKenney <paul.mckenney@linaro.org>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>

1 files changed, 12 insertions, 3 deletions

diff --git a/kernel/rcutree.c b/kernel/rcutree.c
index a7c6bce1af83..45dcc2036a1e 100644
--- a/kernel/rcutree.c
+++ b/kernel/rcutree.c
@@ -52,6 +52,9 @@
 #include <linux/prefetch.h>
 
 #include "rcutree.h"
+#include <trace/events/rcu.h>
+
+#include "rcu.h"
 
 /* Data structures. */
 
@@ -1190,17 +1193,22 @@ static void rcu_do_batch(struct rcu_state *rsp, struct rcu_data *rdp)
 {
 	unsigned long flags;
 	struct rcu_head *next, *list, **tail;
-	int count;
+	int bl, count;
 
 	/* If no callbacks are ready, just return.*/
-	if (!cpu_has_callbacks_ready_to_invoke(rdp))
+	if (!cpu_has_callbacks_ready_to_invoke(rdp)) {
+		trace_rcu_batch_start(0, 0);
+		trace_rcu_batch_end(0);
 		return;
+	}
 
 	/*
 	 * Extract the list of ready callbacks, disabling to prevent
 	 * races with call_rcu() from interrupt handlers.
 	 */
 	local_irq_save(flags);
+	bl = rdp->blimit;
+	trace_rcu_batch_start(rdp->qlen, bl);
 	list = rdp->nxtlist;
 	rdp->nxtlist = *rdp->nxttail[RCU_DONE_TAIL];
 	*rdp->nxttail[RCU_DONE_TAIL] = NULL;
@@ -1218,11 +1226,12 @@ static void rcu_do_batch(struct rcu_state *rsp, struct rcu_data *rdp)
 		debug_rcu_head_unqueue(list);
 		__rcu_reclaim(list);
 		list = next;
-		if (++count >= rdp->blimit)
+		if (++count >= bl)
 			break;
 	}
 
 	local_irq_save(flags);
+	trace_rcu_batch_end(count);
 
 	/* Update count, and requeue any remaining callbacks. */
 	rdp->qlen -= count;