drm/i915: Use seqlock in engine stats

We can convert engine stats from a spinlock to seqlock to ensure interrupt
processing is never even a tiny bit delayed by parallel readers.

There is a smidgen bit more cost on the write lock side, and an extremely
unlikely chance that readers will have to retry a few times in face of
heavy interrupt load. But it should be extremely unlikely given how
lightweight read side section is compared to the interrupt processing
side, and also compared to the rest of the code paths which can lead into
it. Furthermore, writer is the ones doing the real, latency sensitive
work, while readers are only informative.
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Suggested-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20180426074716.7352-1-tvrtko.ursulin@linux.intel.com

drivers/gpu/drm/i915/intel_engine_cs.c

@@ -306,7 +306,7 @@ intel_engine_setup(struct drm_i915_private *dev_priv,
 	/* Nothing to do here, execute in order of dependencies */
 	engine->schedule = NULL;
 
-	spin_lock_init(&engine->stats.lock);
+	seqlock_init(&engine->stats.lock);
 
 	ATOMIC_INIT_NOTIFIER_HEAD(&engine->context_status_notifier);
 
@@ -1481,7 +1481,7 @@ int intel_enable_engine_stats(struct intel_engine_cs *engine)
 		return -ENODEV;
 
 	tasklet_disable(&execlists->tasklet);
-	spin_lock_irqsave(&engine->stats.lock, flags);
+	write_seqlock_irqsave(&engine->stats.lock, flags);
 
 	if (unlikely(engine->stats.enabled == ~0)) {
 		err = -EBUSY;
@@ -1505,7 +1505,7 @@ int intel_enable_engine_stats(struct intel_engine_cs *engine)
 	}
 
 unlock:
-	spin_unlock_irqrestore(&engine->stats.lock, flags);
+	write_sequnlock_irqrestore(&engine->stats.lock, flags);
 	tasklet_enable(&execlists->tasklet);
 
 	return err;
@@ -1534,12 +1534,13 @@ static ktime_t __intel_engine_get_busy_time(struct intel_engine_cs *engine)
  */
 ktime_t intel_engine_get_busy_time(struct intel_engine_cs *engine)
 {
+	unsigned int seq;
 	ktime_t total;
-	unsigned long flags;
 
-	spin_lock_irqsave(&engine->stats.lock, flags);
+	do {
+		seq = read_seqbegin(&engine->stats.lock);
 		total = __intel_engine_get_busy_time(engine);
-	spin_unlock_irqrestore(&engine->stats.lock, flags);
+	} while (read_seqretry(&engine->stats.lock, seq));
 
 	return total;
 }
@@ -1557,13 +1558,13 @@ void intel_disable_engine_stats(struct intel_engine_cs *engine)
 	if (!intel_engine_supports_stats(engine))
 		return;
 
-	spin_lock_irqsave(&engine->stats.lock, flags);
+	write_seqlock_irqsave(&engine->stats.lock, flags);
 	WARN_ON_ONCE(engine->stats.enabled == 0);
 	if (--engine->stats.enabled == 0) {
 		engine->stats.total = __intel_engine_get_busy_time(engine);
 		engine->stats.active = 0;
 	}
-	spin_unlock_irqrestore(&engine->stats.lock, flags);
+	write_sequnlock_irqrestore(&engine->stats.lock, flags);
 }
 
 #if IS_ENABLED(CONFIG_DRM_I915_SELFTEST)

drivers/gpu/drm/i915/intel_ringbuffer.h

@@ -3,6 +3,7 @@
 #define _INTEL_RINGBUFFER_H_
 
 #include <linux/hashtable.h>
+#include <linux/seqlock.h>
 
 #include "i915_gem_batch_pool.h"
 #include "i915_gem_timeline.h"
@@ -595,7 +596,7 @@ struct intel_engine_cs {
 		/**
 		 * @lock: Lock protecting the below fields.
 		 */
-		spinlock_t lock;
+		seqlock_t lock;
 		/**
 		 * @enabled: Reference count indicating number of listeners.
 		 */
@@ -1064,7 +1065,7 @@ static inline void intel_engine_context_in(struct intel_engine_cs *engine)
 	if (READ_ONCE(engine->stats.enabled) == 0)
 		return;
 
-	spin_lock_irqsave(&engine->stats.lock, flags);
+	write_seqlock_irqsave(&engine->stats.lock, flags);
 
 	if (engine->stats.enabled > 0) {
 		if (engine->stats.active++ == 0)
@@ -1072,7 +1073,7 @@ static inline void intel_engine_context_in(struct intel_engine_cs *engine)
 		GEM_BUG_ON(engine->stats.active == 0);
 	}
 
-	spin_unlock_irqrestore(&engine->stats.lock, flags);
+	write_sequnlock_irqrestore(&engine->stats.lock, flags);
 }
 
 static inline void intel_engine_context_out(struct intel_engine_cs *engine)
@@ -1082,7 +1083,7 @@ static inline void intel_engine_context_out(struct intel_engine_cs *engine)
 	if (READ_ONCE(engine->stats.enabled) == 0)
 		return;
 
-	spin_lock_irqsave(&engine->stats.lock, flags);
+	write_seqlock_irqsave(&engine->stats.lock, flags);
 
 	if (engine->stats.enabled > 0) {
 		ktime_t last;
@@ -1109,7 +1110,7 @@ static inline void intel_engine_context_out(struct intel_engine_cs *engine)
 		}
 	}
 
-	spin_unlock_irqrestore(&engine->stats.lock, flags);
+	write_sequnlock_irqrestore(&engine->stats.lock, flags);
 }
 
 int intel_enable_engine_stats(struct intel_engine_cs *engine);