[PATCH v2 2/2] kernel/events: Add option to enable counting sideband events in wakeup_events

[Naveen N. Rao]

Many sideband events are interesting by themselves. When profiling only
for sideband events, it is useful to be able to control process wakeup
(wakeup_events) based on sideband events alone. Add a new option
'count_sb_events' to do the same.

IOC_REFRESH won't be supported with this, so disable that.

Signed-off-by: Naveen N. Rao <naveen.n.rao@xxxxxxxxxxxxxxxxxx>
---
 include/uapi/linux/perf_event.h |  3 ++-
 kernel/events/core.c            | 23 +++++++----------------
 kernel/events/internal.h        |  1 +
 kernel/events/ring_buffer.c     | 18 ++++++++++++++++++
 4 files changed, 28 insertions(+), 17 deletions(-)

diff --git a/include/uapi/linux/perf_event.h b/include/uapi/linux/perf_event.h
index e5810b1d74a4..ab4dc9a02151 100644
--- a/include/uapi/linux/perf_event.h
+++ b/include/uapi/linux/perf_event.h
@@ -346,7 +346,8 @@ struct perf_event_attr {
 				write_backward :  1, /* Write ring buffer from end to beginning */
 				namespaces     :  1, /* include namespaces data */
 				signal_on_wakeup : 1, /* send signal on wakeup */
-				__reserved_1   : 34;
+				count_sb_events  : 1, /* wakeup_events also counts sideband events */
+				__reserved_1   : 33;
 
 	union {
 		__u32		wakeup_events;	  /* wakeup every n events */
diff --git a/kernel/events/core.c b/kernel/events/core.c
index 4fe708a4fdee..118a100108b1 100644
--- a/kernel/events/core.c
+++ b/kernel/events/core.c
@@ -2680,11 +2680,13 @@ EXPORT_SYMBOL_GPL(perf_event_addr_filters_sync);
 static int _perf_event_refresh(struct perf_event *event, int refresh)
 {
 	/*
-	 * not supported on inherited events or if user has requested for
-	 * signals on ring buffer wakeup.
+	 * not supported on:
+	 * - inherited events
+	 * - if user has requested for signals on ring buffer wakeup, or
+	 * - if counting sideband events
 	 */
 	if (event->attr.inherit || event->attr.signal_on_wakeup ||
-			!is_sampling_event(event))
+			event->attr.count_sb_events || !is_sampling_event(event))
 		return -EINVAL;
 
 	atomic_add(refresh, &event->event_limit);
@@ -5974,19 +5976,8 @@ void perf_output_sample(struct perf_output_handle *handle,
 		}
 	}
 
-	if (!event->attr.watermark) {
-		int wakeup_events = event->attr.wakeup_events;
-
-		if (wakeup_events) {
-			struct ring_buffer *rb = handle->rb;
-			int events = local_inc_return(&rb->events);
-
-			if (events >= wakeup_events) {
-				local_sub(wakeup_events, &rb->events);
-				local_inc(&rb->wakeup);
-			}
-		}
-	}
+	if (!event->attr.count_sb_events)
+		rb_handle_wakeup_events(event, handle->rb);
 }
 
 void perf_prepare_sample(struct perf_event_header *header,
diff --git a/kernel/events/internal.h b/kernel/events/internal.h
index 486fd78eb8d5..b75137ebe9f7 100644
--- a/kernel/events/internal.h
+++ b/kernel/events/internal.h
@@ -81,6 +81,7 @@ extern int rb_alloc_aux(struct ring_buffer *rb, struct perf_event *event,
 extern void rb_free_aux(struct ring_buffer *rb);
 extern struct ring_buffer *ring_buffer_get(struct perf_event *event);
 extern void ring_buffer_put(struct ring_buffer *rb);
+extern void rb_handle_wakeup_events(struct perf_event *event, struct ring_buffer *rb);
 
 static inline bool rb_has_aux(struct ring_buffer *rb)
 {
diff --git a/kernel/events/ring_buffer.c b/kernel/events/ring_buffer.c
index e7a558cfcadb..a34f5c2e7ed1 100644
--- a/kernel/events/ring_buffer.c
+++ b/kernel/events/ring_buffer.c
@@ -116,6 +116,21 @@ ring_buffer_has_space(unsigned long head, unsigned long tail,
 		return CIRC_SPACE(tail, head, data_size) >= size;
 }
 
+void __always_inline
+rb_handle_wakeup_events(struct perf_event *event, struct ring_buffer *rb)
+{
+	int wakeup_events = event->attr.wakeup_events;
+
+	if (!event->attr.watermark && wakeup_events) {
+		int events = local_inc_return(&rb->events);
+
+		if (events >= wakeup_events) {
+			local_sub(wakeup_events, &rb->events);
+			local_inc(&rb->wakeup);
+		}
+	}
+}
+
 static int __always_inline
 __perf_output_begin(struct perf_output_handle *handle,
 		    struct perf_event *event, unsigned int size,
@@ -197,6 +212,9 @@ __perf_output_begin(struct perf_output_handle *handle,
 	 * none of the data stores below can be lifted up by the compiler.
 	 */
 
+	if (unlikely(event->attr.count_sb_events))
+		rb_handle_wakeup_events(event, rb);
+
 	if (unlikely(head - local_read(&rb->wakeup) > rb->watermark))
 		local_add(rb->watermark, &rb->wakeup);
 
-- 
2.13.3