diff --git a/drivers/gpu/nvgpu/gk20a/channel_gk20a.c b/drivers/gpu/nvgpu/gk20a/channel_gk20a.c
index b8f769a45..ab0333d62 100644
--- a/drivers/gpu/nvgpu/gk20a/channel_gk20a.c
+++ b/drivers/gpu/nvgpu/gk20a/channel_gk20a.c
@@ -1572,9 +1572,11 @@ int gk20a_submit_channel_gpfifo(struct channel_gk20a *c,
 	/* Make sure we have enough space for gpfifo entries. If not,
 	 * wait for signals from completed submits */
 	if (gp_free_count(c) < num_entries + extra_entries) {
+		trace_gk20a_gpfifo_submit_wait_for_space(c->g->dev->name);
 		err = wait_event_interruptible(c->submit_wq,
 			get_gp_free_count(c) >= num_entries + extra_entries ||
 			c->has_timedout);
+		trace_gk20a_gpfifo_submit_wait_for_space_done(c->g->dev->name);
 	}
 
 	if (c->has_timedout) {
diff --git a/include/trace/events/gk20a.h b/include/trace/events/gk20a.h
index 8fa8cb947..32e3100a5 100644
--- a/include/trace/events/gk20a.h
+++ b/include/trace/events/gk20a.h
@@ -55,6 +55,16 @@ DEFINE_EVENT(gk20a, gk20a_finalize_poweron_done,
 	TP_ARGS(name)
 );
 
+DEFINE_EVENT(gk20a, gk20a_gpfifo_submit_wait_for_space,
+	TP_PROTO(const char *name),
+	TP_ARGS(name)
+);
+
+DEFINE_EVENT(gk20a, gk20a_gpfifo_submit_wait_for_space_done,
+	TP_PROTO(const char *name),
+	TP_ARGS(name)
+);
+
 TRACE_EVENT(gk20a_channel_update,
 	TP_PROTO(const void *channel),
 	TP_ARGS(channel),