drm/i915: Don't emit semaphore wait if wrap happened

If wrap just happened we need to prevent emitting waits for
pre wrap values. Detect this and emit no-ops instead.

v2: Use olr > seqno to detect wrap instead of *seqno == 0
as suggested by Chris Wilson.

v3: Use last used seqno to detect the wraparound. From
Chris Wilson

v4: Fixed unnecessary last_seqno assigment

References: https://bugs.freedesktop.org/show_bug.cgi?id=57967
Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drivers/gpu/drm/i915/i915_drv.h

@@ -671,7 +671,7 @@ typedef struct drm_i915_private {
 
 	struct pci_dev *bridge_dev;
 	struct intel_ring_buffer ring[I915_NUM_RINGS];
-	uint32_t next_seqno;
+	uint32_t last_seqno, next_seqno;
 
 	drm_dma_handle_t *status_page_dmah;
 	struct resource mch_res;

drivers/gpu/drm/i915/i915_gem.c

@@ -1969,7 +1969,7 @@ i915_gem_get_seqno(struct drm_device *dev, u32 *seqno)
 		dev_priv->next_seqno = 1;
 	}
 
-	*seqno = dev_priv->next_seqno++;
+	*seqno = dev_priv->last_seqno = dev_priv->next_seqno++;
 	return 0;
 }
 

drivers/gpu/drm/i915/intel_ringbuffer.c

@@ -596,6 +596,13 @@ gen6_add_request(struct intel_ring_buffer *ring)
 	return 0;
 }
 
+static inline bool i915_gem_has_seqno_wrapped(struct drm_device *dev,
+					      u32 seqno)
+{
+	struct drm_i915_private *dev_priv = dev->dev_private;
+	return dev_priv->last_seqno < seqno;
+}
+
 /**
  * intel_ring_sync - sync the waiter to the signaller on seqno
  *
@@ -626,11 +633,20 @@ gen6_ring_sync(struct intel_ring_buffer *waiter,
 	if (ret)
 		return ret;
 
-	intel_ring_emit(waiter,
-			dw1 | signaller->semaphore_register[waiter->id]);
-	intel_ring_emit(waiter, seqno);
-	intel_ring_emit(waiter, 0);
-	intel_ring_emit(waiter, MI_NOOP);
+	/* If seqno wrap happened, omit the wait with no-ops */
+	if (likely(!i915_gem_has_seqno_wrapped(waiter->dev, seqno))) {
+		intel_ring_emit(waiter,
+				dw1 |
+				signaller->semaphore_register[waiter->id]);
+		intel_ring_emit(waiter, seqno);
+		intel_ring_emit(waiter, 0);
+		intel_ring_emit(waiter, MI_NOOP);
+	} else {
+		intel_ring_emit(waiter, MI_NOOP);
+		intel_ring_emit(waiter, MI_NOOP);
+		intel_ring_emit(waiter, MI_NOOP);
+		intel_ring_emit(waiter, MI_NOOP);
+	}
 	intel_ring_advance(waiter);
 
 	return 0;