[Intel-gfx] [RFC 10/10] drm/i915: Gate engine stats collection with a static key

[Tvrtko Ursulin]

From: Tvrtko Ursulin <tvrtko.ursulin at intel.com>

This reduces the cost of the software engine busyness tracking
to a single no-op instruction when there are no listeners.

Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin at intel.com>
---
 drivers/gpu/drm/i915/intel_engine_cs.c  | 17 +++++++++++
 drivers/gpu/drm/i915/intel_ringbuffer.h | 54 ++++++++++++++++++---------------
 2 files changed, 47 insertions(+), 24 deletions(-)

diff --git a/drivers/gpu/drm/i915/intel_engine_cs.c b/drivers/gpu/drm/i915/intel_engine_cs.c
index e709f4edef90..fb61df3f770c 100644
--- a/drivers/gpu/drm/i915/intel_engine_cs.c
+++ b/drivers/gpu/drm/i915/intel_engine_cs.c
@@ -21,6 +21,7 @@
  * IN THE SOFTWARE.
  *
  */
+#include <linux/static_key.h>
 
 #include "i915_drv.h"
 #include "intel_ringbuffer.h"
@@ -1347,6 +1348,10 @@ void intel_engines_mark_idle(struct drm_i915_private *i915)
 	}
 }
 
+DEFINE_STATIC_KEY_FALSE(i915_engine_stats_key);
+static DEFINE_MUTEX(i915_engine_stats_mutex);
+static int i915_engine_stats_ref;
+
 int intel_enable_engine_stats(struct intel_engine_cs *engine)
 {
 	unsigned long flags;
@@ -1354,6 +1359,8 @@ int intel_enable_engine_stats(struct intel_engine_cs *engine)
 	if (!i915.enable_execlists)
 		return -ENODEV;
 
+	mutex_lock(&i915_engine_stats_mutex);
+
 	spin_lock_irqsave(&engine->stats.lock, flags);
 	if (engine->stats.enabled++ == 0) {
 		engine->stats.ref = 0;
@@ -1363,10 +1370,16 @@ int intel_enable_engine_stats(struct intel_engine_cs *engine)
 	}
 	spin_unlock_irqrestore(&engine->stats.lock, flags);
 
+	if (i915_engine_stats_ref++ == 0)
+		static_branch_enable(&i915_engine_stats_key);
+
+	mutex_unlock(&i915_engine_stats_mutex);
+
 	return 0;
 
 busy:
 	spin_unlock_irqrestore(&engine->stats.lock, flags);
+	mutex_unlock(&i915_engine_stats_mutex);
 
 	return -EBUSY;
 }
@@ -1378,10 +1391,14 @@ void intel_disable_engine_stats(struct intel_engine_cs *engine)
 	if (!i915.enable_execlists)
 		return;
 
+	mutex_lock(&i915_engine_stats_mutex);
 	spin_lock_irqsave(&engine->stats.lock, flags);
 	WARN_ON_ONCE(engine->stats.enabled == 0);
 	engine->stats.enabled--;
 	spin_unlock_irqrestore(&engine->stats.lock, flags);
+	if (--i915_engine_stats_ref == 0)
+		static_branch_disable(&i915_engine_stats_key);
+	mutex_unlock(&i915_engine_stats_mutex);
 }
 
 int intel_enable_engines_stats(struct drm_i915_private *dev_priv)
diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.h b/drivers/gpu/drm/i915/intel_ringbuffer.h
index fd5d838ca7b5..74add0a18e7d 100644
--- a/drivers/gpu/drm/i915/intel_ringbuffer.h
+++ b/drivers/gpu/drm/i915/intel_ringbuffer.h
@@ -754,48 +754,54 @@ void intel_engines_reset_default_submission(struct drm_i915_private *i915);
 struct intel_engine_cs *
 intel_engine_lookup_user(struct drm_i915_private *i915, u8 class, u8 instance);
 
+DECLARE_STATIC_KEY_FALSE(i915_engine_stats_key);
+
 static inline void intel_engine_context_in(struct intel_engine_cs *engine)
 {
 	unsigned long flags;
 
-	if (READ_ONCE(engine->stats.enabled) == 0)
-		return;
+	if (static_branch_unlikely(&i915_engine_stats_key)) {
+		if (READ_ONCE(engine->stats.enabled) == 0)
+			return;
 
-	spin_lock_irqsave(&engine->stats.lock, flags);
+		spin_lock_irqsave(&engine->stats.lock, flags);
 
-	if (engine->stats.enabled > 0) {
-		if (engine->stats.ref++ == 0)
-			engine->stats.start = ktime_get();
-		GEM_BUG_ON(engine->stats.ref == 0);
-	}
+			if (engine->stats.enabled > 0) {
+				if (engine->stats.ref++ == 0)
+					engine->stats.start = ktime_get();
+				GEM_BUG_ON(engine->stats.ref == 0);
+			}
 
-	spin_unlock_irqrestore(&engine->stats.lock, flags);
+		spin_unlock_irqrestore(&engine->stats.lock, flags);
+	}
 }
 
 static inline void intel_engine_context_out(struct intel_engine_cs *engine)
 {
 	unsigned long flags;
 
-	if (READ_ONCE(engine->stats.enabled) == 0)
-		return;
+	if (static_branch_unlikely(&i915_engine_stats_key)) {
+		if (READ_ONCE(engine->stats.enabled) == 0)
+			return;
 
-	spin_lock_irqsave(&engine->stats.lock, flags);
+		spin_lock_irqsave(&engine->stats.lock, flags);
 
-	if (engine->stats.enabled > 0) {
-		/*
-		 * After turning on engine stats, context out might be the
-		 * first event which then needs to be ignored (ref == 0).
-		 */
-		if (engine->stats.ref && --engine->stats.ref == 0) {
-			ktime_t last = ktime_sub(ktime_get(),
-						 engine->stats.start);
+		if (engine->stats.enabled > 0) {
+			/*
+			 * After turning on engine stats, context out might be
+			 * the first event which then needs to be ignored.
+			 */
+			if (engine->stats.ref && --engine->stats.ref == 0) {
+				ktime_t last = ktime_sub(ktime_get(),
+							 engine->stats.start);
 
-			engine->stats.total = ktime_add(engine->stats.total,
-							last);
+				engine->stats.total =
+					ktime_add(engine->stats.total, last);
+			}
 		}
-	}
 
-	spin_unlock_irqrestore(&engine->stats.lock, flags);
+		spin_unlock_irqrestore(&engine->stats.lock, flags);
+	}
 }
 
 int intel_enable_engine_stats(struct intel_engine_cs *engine);
-- 
2.9.4