drm/i915: Refine tracepoints

A lot of minor tweaks to fix the tracepoints, improve the outputting for
ftrace, and to generally make the tracepoints useful again. It is a start
and enough to begin identifying performance issues and gaps in our
coverage.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>

11 files changed, 330 insertions, 359 deletions

diff --git a/drivers/gpu/drm/i915/i915_dma.c b/drivers/gpu/drm/i915/i915_dma.c
index 126e1747fb0c..c79efbc15c5e 100644
--- a/drivers/gpu/drm/i915/i915_dma.c
+++ b/drivers/gpu/drm/i915/i915_dma.c
@@ -2004,7 +2004,6 @@ int i915_driver_load(struct drm_device *dev, unsigned long flags)
 
 	spin_lock_init(&dev_priv->irq_lock);
 	spin_lock_init(&dev_priv->error_lock);
-	dev_priv->trace_irq_seqno = 0;
 
 	ret = drm_vblank_init(dev, I915_NUM_PIPE);
 	if (ret)
diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
index fb5979774c09..bdfda0b8c604 100644
--- a/drivers/gpu/drm/i915/i915_drv.h
+++ b/drivers/gpu/drm/i915/i915_drv.h
@@ -76,10 +76,7 @@ enum plane {
 #define DRIVER_PATCHLEVEL	0
 
 #define WATCH_COHERENCY	0
-#define WATCH_EXEC	0
-#define WATCH_RELOC	0
 #define WATCH_LISTS	0
-#define WATCH_PWRITE	0
 
 #define I915_GEM_PHYS_CURSOR_0 1
 #define I915_GEM_PHYS_CURSOR_1 2
@@ -289,7 +286,6 @@ typedef struct drm_i915_private {
 	int page_flipping;
 
 	atomic_t irq_received;
-	u32 trace_irq_seqno;
 
 	/* protects the irq masks */
 	spinlock_t irq_lock;
@@ -1001,7 +997,6 @@ extern int i915_irq_emit(struct drm_device *dev, void *data,
 			 struct drm_file *file_priv);
 extern int i915_irq_wait(struct drm_device *dev, void *data,
 			 struct drm_file *file_priv);
-void i915_trace_irq_get(struct drm_device *dev, u32 seqno);
 
 extern irqreturn_t i915_driver_irq_handler(DRM_IRQ_ARGS);
 extern void i915_driver_irq_preinstall(struct drm_device * dev);
@@ -1095,8 +1090,7 @@ int i915_gem_get_aperture_ioctl(struct drm_device *dev, void *data,
 				struct drm_file *file_priv);
 void i915_gem_load(struct drm_device *dev);
 int i915_gem_init_object(struct drm_gem_object *obj);
-int __must_check i915_gem_flush_ring(struct drm_device *dev,
-				     struct intel_ring_buffer *ring,
+int __must_check i915_gem_flush_ring(struct intel_ring_buffer *ring,
 				     uint32_t invalidate_domains,
 				     uint32_t flush_domains);
 struct drm_i915_gem_object *i915_gem_alloc_object(struct drm_device *dev,
@@ -1127,10 +1121,9 @@ i915_seqno_passed(uint32_t seq1, uint32_t seq2)
 }
 
 static inline u32
-i915_gem_next_request_seqno(struct drm_device *dev,
-			    struct intel_ring_buffer *ring)
+i915_gem_next_request_seqno(struct intel_ring_buffer *ring)
 {
-	drm_i915_private_t *dev_priv = dev->dev_private;
+	drm_i915_private_t *dev_priv = ring->dev->dev_private;
 	return ring->outstanding_lazy_request = dev_priv->next_seqno;
 }
 
@@ -1155,14 +1148,12 @@ void i915_gem_do_init(struct drm_device *dev,
 		      unsigned long end);
 int __must_check i915_gpu_idle(struct drm_device *dev);
 int __must_check i915_gem_idle(struct drm_device *dev);
-int __must_check i915_add_request(struct drm_device *dev,
-				  struct drm_file *file_priv,
-				  struct drm_i915_gem_request *request,
-				  struct intel_ring_buffer *ring);
-int __must_check i915_do_wait_request(struct drm_device *dev,
-				      uint32_t seqno,
-				      bool interruptible,
-				      struct intel_ring_buffer *ring);
+int __must_check i915_add_request(struct intel_ring_buffer *ring,
+				  struct drm_file *file,
+				  struct drm_i915_gem_request *request);
+int __must_check i915_wait_request(struct intel_ring_buffer *ring,
+				   uint32_t seqno,
+				   bool interruptible);
 int i915_gem_fault(struct vm_area_struct *vma, struct vm_fault *vmf);
 int __must_check
 i915_gem_object_set_to_gtt_domain(struct drm_i915_gem_object *obj,
@@ -1311,7 +1302,7 @@ extern void intel_display_print_error_state(struct seq_file *m,
 #define __i915_read(x, y) \
 static inline u##x i915_read##x(struct drm_i915_private *dev_priv, u32 reg) { \
 	u##x val = read##y(dev_priv->regs + reg); \
-	trace_i915_reg_rw('R', reg, val, sizeof(val)); \
+	trace_i915_reg_rw(false, reg, val, sizeof(val)); \
 	return val; \
 }
 __i915_read(8, b)
@@ -1322,7 +1313,7 @@ __i915_read(64, q)
 
 #define __i915_write(x, y) \
 static inline void i915_write##x(struct drm_i915_private *dev_priv, u32 reg, u##x val) { \
-	trace_i915_reg_rw('W', reg, val, sizeof(val)); \
+	trace_i915_reg_rw(true, reg, val, sizeof(val)); \
 	write##y(val, dev_priv->regs + reg); \
 }
 __i915_write(8, b)
@@ -1371,25 +1362,4 @@ static inline u32 i915_safe_read(struct drm_i915_private *dev_priv, u32 reg)
 	return val;
 }
 
-static inline void
-i915_write(struct drm_i915_private *dev_priv, u32 reg, u64 val, int len)
-{
-       /* Trace down the write operation before the real write */
-       trace_i915_reg_rw('W', reg, val, len);
-       switch (len) {
-       case 8:
-               writeq(val, dev_priv->regs + reg);
-               break;
-       case 4:
-               writel(val, dev_priv->regs + reg);
-               break;
-       case 2:
-               writew(val, dev_priv->regs + reg);
-               break;
-       case 1:
-               writeb(val, dev_priv->regs + reg);
-               break;
-       }
-}
-
 #endif
diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index a41c0e716805..f0f8c6ff684f 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -518,6 +518,8 @@ i915_gem_pread_ioctl(struct drm_device *dev, void *data,
 		goto out;
 	}
 
+	trace_i915_gem_object_pread(obj, args->offset, args->size);
+
 	ret = i915_gem_object_set_cpu_read_domain_range(obj,
 							args->offset,
 							args->size);
@@ -959,6 +961,8 @@ i915_gem_pwrite_ioctl(struct drm_device *dev, void *data,
 		goto out;
 	}
 
+	trace_i915_gem_object_pwrite(obj, args->offset, args->size);
+
 	/* We can only do the GTT pwrite on untiled buffers, as otherwise
 	 * it would end up going through the fenced access, and we'll get
 	 * different detiling behavior between reading and writing.
@@ -1175,6 +1179,8 @@ int i915_gem_fault(struct vm_area_struct *vma, struct vm_fault *vmf)
 	if (ret)
 		goto out;
 
+	trace_i915_gem_object_fault(obj, page_offset, true, write);
+
 	/* Now bind it into the GTT if needed */
 	if (!obj->map_and_fenceable) {
 		ret = i915_gem_object_unbind(obj);
@@ -1668,9 +1674,8 @@ i915_gem_object_is_purgeable(struct drm_i915_gem_object *obj)
 }
 
 static void
-i915_gem_process_flushing_list(struct drm_device *dev,
-			       uint32_t flush_domains,
-			       struct intel_ring_buffer *ring)
+i915_gem_process_flushing_list(struct intel_ring_buffer *ring,
+			       uint32_t flush_domains)
 {
 	struct drm_i915_gem_object *obj, *next;
 
@@ -1683,7 +1688,7 @@ i915_gem_process_flushing_list(struct drm_device *dev,
 			obj->base.write_domain = 0;
 			list_del_init(&obj->gpu_write_list);
 			i915_gem_object_move_to_active(obj, ring,
-						       i915_gem_next_request_seqno(dev, ring));
+						       i915_gem_next_request_seqno(ring));
 
 			trace_i915_gem_object_change_domain(obj,
 							    obj->base.read_domains,
@@ -1693,27 +1698,22 @@ i915_gem_process_flushing_list(struct drm_device *dev,
 }
 
 int
-i915_add_request(struct drm_device *dev,
+i915_add_request(struct intel_ring_buffer *ring,
 		 struct drm_file *file,
-		 struct drm_i915_gem_request *request,
-		 struct intel_ring_buffer *ring)
+		 struct drm_i915_gem_request *request)
 {
-	drm_i915_private_t *dev_priv = dev->dev_private;
-	struct drm_i915_file_private *file_priv = NULL;
+	drm_i915_private_t *dev_priv = ring->dev->dev_private;
 	uint32_t seqno;
 	int was_empty;
 	int ret;
 
 	BUG_ON(request == NULL);
 
-	if (file != NULL)
-		file_priv = file->driver_priv;
-
 	ret = ring->add_request(ring, &seqno);
 	if (ret)
 	    return ret;
 
-	ring->outstanding_lazy_request = false;
+	trace_i915_gem_request_add(ring, seqno);
 
 	request->seqno = seqno;
 	request->ring = ring;
@@ -1721,7 +1721,9 @@ i915_add_request(struct drm_device *dev,
 	was_empty = list_empty(&ring->request_list);
 	list_add_tail(&request->list, &ring->request_list);
 
-	if (file_priv) {
+	if (file) {
+		struct drm_i915_file_private *file_priv = file->driver_priv;
+
 		spin_lock(&file_priv->mm.lock);
 		request->file_priv = file_priv;
 		list_add_tail(&request->client_list,
@@ -1729,6 +1731,8 @@ i915_add_request(struct drm_device *dev,
 		spin_unlock(&file_priv->mm.lock);
 	}
 
+	ring->outstanding_lazy_request = false;
+
 	if (!dev_priv->mm.suspended) {
 		mod_timer(&dev_priv->hangcheck_timer,
 			  jiffies + msecs_to_jiffies(DRM_I915_HANGCHECK_PERIOD));
@@ -1845,18 +1849,15 @@ void i915_gem_reset(struct drm_device *dev)
  * This function clears the request list as sequence numbers are passed.
  */
 static void
-i915_gem_retire_requests_ring(struct drm_device *dev,
-			      struct intel_ring_buffer *ring)
+i915_gem_retire_requests_ring(struct intel_ring_buffer *ring)
 {
-	drm_i915_private_t *dev_priv = dev->dev_private;
 	uint32_t seqno;
 	int i;
 
-	if (!ring->status_page.page_addr ||
-	    list_empty(&ring->request_list))
+	if (list_empty(&ring->request_list))
 		return;
 
-	WARN_ON(i915_verify_lists(dev));
+	WARN_ON(i915_verify_lists(ring->dev));
 
 	seqno = ring->get_seqno(ring);
 
@@ -1874,7 +1875,7 @@ i915_gem_retire_requests_ring(struct drm_device *dev,
 		if (!i915_seqno_passed(seqno, request->seqno))
 			break;
 
-		trace_i915_gem_request_retire(dev, request->seqno);
+		trace_i915_gem_request_retire(ring, request->seqno);
 
 		list_del(&request->list);
 		i915_gem_request_remove_from_client(request);
@@ -1900,13 +1901,13 @@ i915_gem_retire_requests_ring(struct drm_device *dev,
 			i915_gem_object_move_to_inactive(obj);
 	}
 
-	if (unlikely (dev_priv->trace_irq_seqno &&
-		      i915_seqno_passed(dev_priv->trace_irq_seqno, seqno))) {
+	if (unlikely(ring->trace_irq_seqno &&
+		     i915_seqno_passed(seqno, ring->trace_irq_seqno))) {
 		ring->irq_put(ring);
-		dev_priv->trace_irq_seqno = 0;
+		ring->trace_irq_seqno = 0;
 	}
 
-	WARN_ON(i915_verify_lists(dev));
+	WARN_ON(i915_verify_lists(ring->dev));
 }
 
 void
@@ -1930,7 +1931,7 @@ i915_gem_retire_requests(struct drm_device *dev)
 	}
 
 	for (i = 0; i < I915_NUM_RINGS; i++)
-		i915_gem_retire_requests_ring(dev, &dev_priv->ring[i]);
+		i915_gem_retire_requests_ring(&dev_priv->ring[i]);
 }
 
 static void
@@ -1964,11 +1965,11 @@ i915_gem_retire_work_handler(struct work_struct *work)
 			struct drm_i915_gem_request *request;
 			int ret;
 
-			ret = i915_gem_flush_ring(dev, ring, 0,
-						  I915_GEM_GPU_DOMAINS);
+			ret = i915_gem_flush_ring(ring,
+						  0, I915_GEM_GPU_DOMAINS);
 			request = kzalloc(sizeof(*request), GFP_KERNEL);
 			if (ret || request == NULL ||
-			    i915_add_request(dev, NULL, request, ring))
+			    i915_add_request(ring, NULL, request))
 			    kfree(request);
 		}
 
@@ -1981,11 +1982,16 @@ i915_gem_retire_work_handler(struct work_struct *work)
 	mutex_unlock(&dev->struct_mutex);
 }
 
+/**
+ * Waits for a sequence number to be signaled, and cleans up the
+ * request and object lists appropriately for that event.
+ */
 int
-i915_do_wait_request(struct drm_device *dev, uint32_t seqno,
-		     bool interruptible, struct intel_ring_buffer *ring)
+i915_wait_request(struct intel_ring_buffer *ring,
+		  uint32_t seqno,
+		  bool interruptible)
 {
-	drm_i915_private_t *dev_priv = dev->dev_private;
+	drm_i915_private_t *dev_priv = ring->dev->dev_private;
 	u32 ier;
 	int ret = 0;
 
@@ -2011,7 +2017,7 @@ i915_do_wait_request(struct drm_device *dev, uint32_t seqno,
 		if (request == NULL)
 			return -ENOMEM;
 
-		ret = i915_add_request(dev, NULL, request, ring);
+		ret = i915_add_request(ring, NULL, request);
 		if (ret) {
 			kfree(request);
 			return ret;
@@ -2021,18 +2027,18 @@ i915_do_wait_request(struct drm_device *dev, uint32_t seqno,
 	}
 
 	if (!i915_seqno_passed(ring->get_seqno(ring), seqno)) {
-		if (HAS_PCH_SPLIT(dev))
+		if (HAS_PCH_SPLIT(ring->dev))
 			ier = I915_READ(DEIER) | I915_READ(GTIER);
 		else
 			ier = I915_READ(IER);
 		if (!ier) {
 			DRM_ERROR("something (likely vbetool) disabled "
 				  "interrupts, re-enabling\n");
-			i915_driver_irq_preinstall(dev);
-			i915_driver_irq_postinstall(dev);
+			i915_driver_irq_preinstall(ring->dev);
+			i915_driver_irq_postinstall(ring->dev);
 		}
 
-		trace_i915_gem_request_wait_begin(dev, seqno);
+		trace_i915_gem_request_wait_begin(ring, seqno);
 
 		ring->waiting_seqno = seqno;
 		if (ring->irq_get(ring)) {
@@ -2052,7 +2058,7 @@ i915_do_wait_request(struct drm_device *dev, uint32_t seqno,
 			ret = -EBUSY;
 		ring->waiting_seqno = 0;
 
-		trace_i915_gem_request_wait_end(dev, seqno);
+		trace_i915_gem_request_wait_end(ring, seqno);
 	}
 	if (atomic_read(&dev_priv->mm.wedged))
 		ret = -EAGAIN;
@@ -2068,23 +2074,12 @@ i915_do_wait_request(struct drm_device *dev, uint32_t seqno,
 	 * a separate wait queue to handle that.
 	 */
 	if (ret == 0)
-		i915_gem_retire_requests_ring(dev, ring);
+		i915_gem_retire_requests_ring(ring);
 
 	return ret;
 }
 
 /**
- * Waits for a sequence number to be signaled, and cleans up the
- * request and object lists appropriately for that event.
- */
-static int
-i915_wait_request(struct drm_device *dev, uint32_t seqno,
-		  struct intel_ring_buffer *ring)
-{
-	return i915_do_wait_request(dev, seqno, 1, ring);
-}
-
-/**
  * Ensures that all rendering to the object has completed and the object is
  * safe to unbind from the GTT or access from the CPU.
  */
@@ -2092,7 +2087,6 @@ int
 i915_gem_object_wait_rendering(struct drm_i915_gem_object *obj,
 			       bool interruptible)
 {
-	struct drm_device *dev = obj->base.dev;
 	int ret;
 
 	/* This function only exists to support waiting for existing rendering,
@@ -2104,10 +2098,9 @@ i915_gem_object_wait_rendering(struct drm_i915_gem_object *obj,
 	 * it.
 	 */
 	if (obj->active) {
-		ret = i915_do_wait_request(dev,
-					   obj->last_rendering_seqno,
-					   interruptible,
-					   obj->ring);
+		ret = i915_wait_request(obj->ring,
+					obj->last_rendering_seqno,
+					interruptible);
 		if (ret)
 			return ret;
 	}
@@ -2157,6 +2150,8 @@ i915_gem_object_unbind(struct drm_i915_gem_object *obj)
 	if (ret == -ERESTARTSYS)
 		return ret;
 
+	trace_i915_gem_object_unbind(obj);
+
 	i915_gem_gtt_unbind_object(obj);
 	i915_gem_object_put_pages_gtt(obj);
 
@@ -2172,29 +2167,27 @@ i915_gem_object_unbind(struct drm_i915_gem_object *obj)
 	if (i915_gem_object_is_purgeable(obj))
 		i915_gem_object_truncate(obj);
 
-	trace_i915_gem_object_unbind(obj);
-
 	return ret;
 }
 
 int
-i915_gem_flush_ring(struct drm_device *dev,
-		    struct intel_ring_buffer *ring,
+i915_gem_flush_ring(struct intel_ring_buffer *ring,
 		    uint32_t invalidate_domains,
 		    uint32_t flush_domains)
 {
 	int ret;
 
+	trace_i915_gem_ring_flush(ring, invalidate_domains, flush_domains);
+
 	ret = ring->flush(ring, invalidate_domains, flush_domains);
 	if (ret)
 		return ret;
 
-	i915_gem_process_flushing_list(dev, flush_domains, ring);
+	i915_gem_process_flushing_list(ring, flush_domains);
 	return 0;
 }
 
-static int i915_ring_idle(struct drm_device *dev,
-			  struct intel_ring_buffer *ring)
+static int i915_ring_idle(struct intel_ring_buffer *ring)
 {
 	int ret;
 
@@ -2202,15 +2195,15 @@ static int i915_ring_idle(struct drm_device *dev,
 		return 0;
 
 	if (!list_empty(&ring->gpu_write_list)) {
-		ret = i915_gem_flush_ring(dev, ring,
+		ret = i915_gem_flush_ring(ring,
 				    I915_GEM_GPU_DOMAINS, I915_GEM_GPU_DOMAINS);
 		if (ret)
 			return ret;
 	}
 
-	return i915_wait_request(dev,
-				 i915_gem_next_request_seqno(dev, ring),
-				 ring);
+	return i915_wait_request(ring,
+				 i915_gem_next_request_seqno(ring),
+				 true);
 }
 
 int
@@ -2227,7 +2220,7 @@ i915_gpu_idle(struct drm_device *dev)
 
 	/* Flush everything onto the inactive list. */
 	for (i = 0; i < I915_NUM_RINGS; i++) {
-		ret = i915_ring_idle(dev, &dev_priv->ring[i]);
+		ret = i915_ring_idle(&dev_priv->ring[i]);
 		if (ret)
 			return ret;
 	}
@@ -2418,8 +2411,7 @@ i915_gem_object_flush_fence(struct drm_i915_gem_object *obj,
 
 	if (obj->fenced_gpu_access) {
 		if (obj->base.write_domain & I915_GEM_GPU_DOMAINS) {
-			ret = i915_gem_flush_ring(obj->base.dev,
-						  obj->last_fenced_ring,
+			ret = i915_gem_flush_ring(obj->last_fenced_ring,
 						  0, obj->base.write_domain);
 			if (ret)
 				return ret;
@@ -2431,10 +2423,10 @@ i915_gem_object_flush_fence(struct drm_i915_gem_object *obj,
 	if (obj->last_fenced_seqno && pipelined != obj->last_fenced_ring) {
 		if (!ring_passed_seqno(obj->last_fenced_ring,
 				       obj->last_fenced_seqno)) {
-			ret = i915_do_wait_request(obj->base.dev,
-						   obj->last_fenced_seqno,
-						   interruptible,
-						   obj->last_fenced_ring);
+			ret = i915_wait_request(obj->last_fenced_ring,
+						obj->last_fenced_seqno,
+						interruptible);
+
 			if (ret)
 				return ret;
 		}
@@ -2560,10 +2552,9 @@ i915_gem_object_get_fence(struct drm_i915_gem_object *obj,
 			if (reg->setup_seqno) {
 				if (!ring_passed_seqno(obj->last_fenced_ring,
 						       reg->setup_seqno)) {
-					ret = i915_do_wait_request(obj->base.dev,
-								   reg->setup_seqno,
-								   interruptible,
-								   obj->last_fenced_ring);
+					ret = i915_wait_request(obj->last_fenced_ring,
+								reg->setup_seqno,
+								interruptible);
 					if (ret)
 						return ret;
 				}
@@ -2580,7 +2571,7 @@ i915_gem_object_get_fence(struct drm_i915_gem_object *obj,
 		} else if (obj->tiling_changed) {
 			if (obj->fenced_gpu_access) {
 				if (obj->base.write_domain & I915_GEM_GPU_DOMAINS) {
-					ret = i915_gem_flush_ring(obj->base.dev, obj->ring,
+					ret = i915_gem_flush_ring(obj->ring,
 								  0, obj->base.write_domain);
 					if (ret)
 						return ret;
@@ -2597,7 +2588,7 @@ i915_gem_object_get_fence(struct drm_i915_gem_object *obj,
 		if (obj->tiling_changed) {
 			if (pipelined) {
 				reg->setup_seqno =
-					i915_gem_next_request_seqno(dev, pipelined);
+					i915_gem_next_request_seqno(pipelined);
 				obj->last_fenced_seqno = reg->setup_seqno;
 				obj->last_fenced_ring = pipelined;
 			}
@@ -2637,7 +2628,7 @@ i915_gem_object_get_fence(struct drm_i915_gem_object *obj,
 		old->fence_reg = I915_FENCE_REG_NONE;
 		old->last_fenced_ring = pipelined;
 		old->last_fenced_seqno =
-			pipelined ? i915_gem_next_request_seqno(dev, pipelined) : 0;
+			pipelined ? i915_gem_next_request_seqno(pipelined) : 0;
 
 		drm_gem_object_unreference(&old->base);
 	} else if (obj->last_fenced_seqno == 0)
@@ -2649,7 +2640,7 @@ i915_gem_object_get_fence(struct drm_i915_gem_object *obj,
 	obj->last_fenced_ring = pipelined;
 
 	reg->setup_seqno =
-		pipelined ? i915_gem_next_request_seqno(dev, pipelined) : 0;
+		pipelined ? i915_gem_next_request_seqno(pipelined) : 0;
 	obj->last_fenced_seqno = reg->setup_seqno;
 
 update:
@@ -2846,7 +2837,7 @@ i915_gem_object_bind_to_gtt(struct drm_i915_gem_object *obj,
 
 	obj->map_and_fenceable = mappable && fenceable;
 
-	trace_i915_gem_object_bind(obj, obj->gtt_offset, map_and_fenceable);
+	trace_i915_gem_object_bind(obj, map_and_fenceable);
 	return 0;
 }
 
@@ -2869,13 +2860,11 @@ i915_gem_clflush_object(struct drm_i915_gem_object *obj)
 static int
 i915_gem_object_flush_gpu_write_domain(struct drm_i915_gem_object *obj)
 {
-	struct drm_device *dev = obj->base.dev;
-
 	if ((obj->base.write_domain & I915_GEM_GPU_DOMAINS) == 0)
 		return 0;
 
 	/* Queue the GPU write cache flushing we need. */
-	return i915_gem_flush_ring(dev, obj->ring, 0, obj->base.write_domain);
+	return i915_gem_flush_ring(obj->ring, 0, obj->base.write_domain);
 }
 
 /** Flushes the GTT write domain for the object if it's dirty. */
@@ -3024,8 +3013,7 @@ i915_gem_object_flush_gpu(struct drm_i915_gem_object *obj,
 		return 0;
 
 	if (obj->base.write_domain & I915_GEM_GPU_DOMAINS) {
-		ret = i915_gem_flush_ring(obj->base.dev, obj->ring,
-					  0, obj->base.write_domain);
+		ret = i915_gem_flush_ring(obj->ring, 0, obj->base.write_domain);
 		if (ret)
 			return ret;
 	}
@@ -3442,7 +3430,7 @@ i915_gem_busy_ioctl(struct drm_device *dev, void *data,
 		 * flush earlier is beneficial.
 		 */
 		if (obj->base.write_domain & I915_GEM_GPU_DOMAINS) {
-			ret = i915_gem_flush_ring(dev, obj->ring,
+			ret = i915_gem_flush_ring(obj->ring,
 						  0, obj->base.write_domain);
 		} else if (obj->ring->outstanding_lazy_request ==
 			   obj->last_rendering_seqno) {
@@ -3453,9 +3441,7 @@ i915_gem_busy_ioctl(struct drm_device *dev, void *data,
 			 */
 			request = kzalloc(sizeof(*request), GFP_KERNEL);
 			if (request)
-				ret = i915_add_request(dev,
-						       NULL, request,
-						       obj->ring);
+				ret = i915_add_request(obj->ring, NULL,request);
 			else
 				ret = -ENOMEM;
 		}
@@ -3465,7 +3451,7 @@ i915_gem_busy_ioctl(struct drm_device *dev, void *data,
 		 * are actually unmasked, and our working set ends up being
 		 * larger than required.
 		 */
-		i915_gem_retire_requests_ring(dev, obj->ring);
+		i915_gem_retire_requests_ring(obj->ring);
 
 		args->busy = obj->active;
 	}
@@ -3595,6 +3581,8 @@ static void i915_gem_free_object_tail(struct drm_i915_gem_object *obj)
 	kfree(obj->page_cpu_valid);
 	kfree(obj->bit_17);
 	kfree(obj);
+
+	trace_i915_gem_object_destroy(obj);
 }
 
 void i915_gem_free_object(struct drm_gem_object *gem_obj)
@@ -3602,8 +3590,6 @@ void i915_gem_free_object(struct drm_gem_object *gem_obj)
 	struct drm_i915_gem_object *obj = to_intel_bo(gem_obj);
 	struct drm_device *dev = obj->base.dev;
 
-	trace_i915_gem_object_destroy(obj);
-
 	while (obj->pin_count > 0)
 		i915_gem_object_unpin(obj);
 
diff --git a/drivers/gpu/drm/i915/i915_gem_debug.c b/drivers/gpu/drm/i915/i915_gem_debug.c
index 29d014c48ca2..8da1899bd24f 100644
--- a/drivers/gpu/drm/i915/i915_gem_debug.c
+++ b/drivers/gpu/drm/i915/i915_gem_debug.c
@@ -134,51 +134,6 @@ i915_verify_lists(struct drm_device *dev)
 }
 #endif /* WATCH_INACTIVE */
 
-
-#if WATCH_EXEC | WATCH_PWRITE
-static void
-i915_gem_dump_page(struct page *page, uint32_t start, uint32_t end,
-		   uint32_t bias, uint32_t mark)
-{
-	uint32_t *mem = kmap_atomic(page, KM_USER0);
-	int i;
-	for (i = start; i < end; i += 4)
-		DRM_INFO("%08x: %08x%s\n",
-			  (int) (bias + i), mem[i / 4],
-			  (bias + i == mark) ? " ********" : "");
-	kunmap_atomic(mem, KM_USER0);
-	/* give syslog time to catch up */
-	msleep(1);
-}
-
-void
-i915_gem_dump_object(struct drm_i915_gem_object *obj, int len,
-		     const char *where, uint32_t mark)
-{
-	int page;
-
-	DRM_INFO("%s: object at offset %08x\n", where, obj->gtt_offset);
-	for (page = 0; page < (len + PAGE_SIZE-1) / PAGE_SIZE; page++) {
-		int page_len, chunk, chunk_len;
-
-		page_len = len - page * PAGE_SIZE;
-		if (page_len > PAGE_SIZE)
-			page_len = PAGE_SIZE;
-
-		for (chunk = 0; chunk < page_len; chunk += 128) {
-			chunk_len = page_len - chunk;
-			if (chunk_len > 128)
-				chunk_len = 128;
-			i915_gem_dump_page(obj->pages[page],
-					   chunk, chunk + chunk_len,
-					   obj->gtt_offset +
-					   page * PAGE_SIZE,
-					   mark);
-		}
-	}
-}
-#endif
-
 #if WATCH_COHERENCY
 void
 i915_gem_object_check_coherency(struct drm_i915_gem_object *obj, int handle)
diff --git a/drivers/gpu/drm/i915/i915_gem_evict.c b/drivers/gpu/drm/i915/i915_gem_evict.c
index 3d39005540aa..da05a2692a75 100644
--- a/drivers/gpu/drm/i915/i915_gem_evict.c
+++ b/drivers/gpu/drm/i915/i915_gem_evict.c
@@ -30,6 +30,7 @@
 #include "drm.h"
 #include "i915_drv.h"
 #include "i915_drm.h"
+#include "i915_trace.h"
 
 static bool
 mark_free(struct drm_i915_gem_object *obj, struct list_head *unwind)
@@ -63,6 +64,8 @@ i915_gem_evict_something(struct drm_device *dev, int min_size,
 			return 0;
 	}
 
+	trace_i915_gem_evict(dev, min_size, alignment, mappable);
+
 	/*
 	 * The goal is to evict objects and amalgamate space in LRU order.
 	 * The oldest idle objects reside on the inactive list, which is in
@@ -189,6 +192,8 @@ i915_gem_evict_everything(struct drm_device *dev, bool purgeable_only)
 	if (lists_empty)
 		return -ENOSPC;
 
+	trace_i915_gem_evict_everything(dev, purgeable_only);
+
 	/* Flush everything (on to the inactive lists) and evict */
 	ret = i915_gpu_idle(dev);
 	if (ret)
diff --git a/drivers/gpu/drm/i915/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/i915_gem_execbuffer.c
index b0a0238c36d1..84fa24e6cca8 100644
--- a/drivers/gpu/drm/i915/i915_gem_execbuffer.c
+++ b/drivers/gpu/drm/i915/i915_gem_execbuffer.c
@@ -282,21 +282,6 @@ i915_gem_execbuffer_relocate_entry(struct drm_i915_gem_object *obj,
 
 	target_offset = to_intel_bo(target_obj)->gtt_offset;
 
-#if WATCH_RELOC
-	DRM_INFO("%s: obj %p offset %08x target %d "
-		 "read %08x write %08x gtt %08x "
-		 "presumed %08x delta %08x\n",
-		 __func__,
-		 obj,
-		 (int) reloc->offset,
-		 (int) reloc->target_handle,
-		 (int) reloc->read_domains,
-		 (int) reloc->write_domain,
-		 (int) target_offset,
-		 (int) reloc->presumed_offset,
-		 reloc->delta);
-#endif
-
 	/* The target buffer should have appeared before us in the
 	 * exec_object list, so it should have a GTT space bound by now.
 	 */
@@ -747,8 +732,7 @@ i915_gem_execbuffer_flush(struct drm_device *dev,
 	if ((flush_domains | invalidate_domains) & I915_GEM_GPU_DOMAINS) {
 		for (i = 0; i < I915_NUM_RINGS; i++)
 			if (flush_rings & (1 << i)) {
-				ret = i915_gem_flush_ring(dev,
-							  &dev_priv->ring[i],
+				ret = i915_gem_flush_ring(&dev_priv->ring[i],
 							  invalidate_domains,
 							  flush_domains);
 				if (ret)
@@ -787,7 +771,7 @@ i915_gem_execbuffer_sync_rings(struct drm_i915_gem_object *obj,
 		if (request == NULL)
 			return -ENOMEM;
 
-		ret = i915_add_request(obj->base.dev, NULL, request, from);
+		ret = i915_add_request(from, NULL, request);
 		if (ret) {
 			kfree(request);
 			return ret;
@@ -815,12 +799,6 @@ i915_gem_execbuffer_move_to_gpu(struct intel_ring_buffer *ring,
 		i915_gem_object_set_to_gpu_domain(obj, ring, &cd);
 
 	if (cd.invalidate_domains | cd.flush_domains) {
-#if WATCH_EXEC
-		DRM_INFO("%s: invalidate_domains %08x flush_domains %08x\n",
-			  __func__,
-			 cd.invalidate_domains,
-			 cd.flush_domains);
-#endif
 		ret = i915_gem_execbuffer_flush(ring->dev,
 						cd.invalidate_domains,
 						cd.flush_domains,
@@ -924,6 +902,10 @@ i915_gem_execbuffer_move_to_active(struct list_head *objects,
 	struct drm_i915_gem_object *obj;
 
 	list_for_each_entry(obj, objects, exec_list) {
+		  u32 old_read = obj->base.read_domains;
+		  u32 old_write = obj->base.write_domain;
+
+
 		obj->base.read_domains = obj->base.pending_read_domains;
 		obj->base.write_domain = obj->base.pending_write_domain;
 		obj->fenced_gpu_access = obj->pending_fenced_gpu_access;
@@ -937,9 +919,7 @@ i915_gem_execbuffer_move_to_active(struct list_head *objects,
 			intel_mark_busy(ring->dev, obj);
 		}
 
-		trace_i915_gem_object_change_domain(obj,
-						    obj->base.read_domains,
-						    obj->base.write_domain);
+		trace_i915_gem_object_change_domain(obj, old_read, old_write);
 	}
 }
 
@@ -961,14 +941,14 @@ i915_gem_execbuffer_retire_commands(struct drm_device *dev,
 	if (INTEL_INFO(dev)->gen >= 4)
 		invalidate |= I915_GEM_DOMAIN_SAMPLER;
 	if (ring->flush(ring, invalidate, 0)) {
-		i915_gem_next_request_seqno(dev, ring);
+		i915_gem_next_request_seqno(ring);
 		return;
 	}
 
 	/* Add a breadcrumb for the completion of the batch buffer */
 	request = kzalloc(sizeof(*request), GFP_KERNEL);
-	if (request == NULL || i915_add_request(dev, file, request, ring)) {
-		i915_gem_next_request_seqno(dev, ring);
+	if (request == NULL || i915_add_request(ring, file, request)) {
+		i915_gem_next_request_seqno(ring);
 		kfree(request);
 	}
 }
@@ -998,10 +978,6 @@ i915_gem_do_execbuffer(struct drm_device *dev, void *data,
 	if (ret)
 		return ret;
 
-#if WATCH_EXEC
-	DRM_INFO("buffers_ptr %d buffer_count %d len %08x\n",
-		  (int) args->buffers_ptr, args->buffer_count, args->batch_len);
-#endif
 	switch (args->flags & I915_EXEC_RING_MASK) {
 	case I915_EXEC_DEFAULT:
 	case I915_EXEC_RENDER:
@@ -1172,7 +1148,7 @@ i915_gem_do_execbuffer(struct drm_device *dev, void *data,
 	if (ret)
 		goto err;
 
-	seqno = i915_gem_next_request_seqno(dev, ring);
+	seqno = i915_gem_next_request_seqno(ring);
 	for (i = 0; i < ARRAY_SIZE(ring->sync_seqno); i++) {
 		if (seqno < ring->sync_seqno[i]) {
 			/* The GPU can not handle its semaphore value wrapping,
@@ -1187,6 +1163,8 @@ i915_gem_do_execbuffer(struct drm_device *dev, void *data,
 		}
 	}
 
+	trace_i915_gem_ring_dispatch(ring, seqno);
+
 	exec_start = batch_obj->gtt_offset + args->batch_start_offset;
 	exec_len = args->batch_len;
 	if (cliprects) {
@@ -1243,11 +1221,6 @@ i915_gem_execbuffer(struct drm_device *dev, void *data,
 	struct drm_i915_gem_exec_object2 *exec2_list = NULL;
 	int ret, i;
 
-#if WATCH_EXEC
-	DRM_INFO("buffers_ptr %d buffer_count %d len %08x\n",
-		  (int) args->buffers_ptr, args->buffer_count, args->batch_len);
-#endif
-
 	if (args->buffer_count < 1) {
 		DRM_ERROR("execbuf with %d buffers\n", args->buffer_count);
 		return -EINVAL;
@@ -1328,11 +1301,6 @@ i915_gem_execbuffer2(struct drm_device *dev, void *data,
 	struct drm_i915_gem_exec_object2 *exec2_list = NULL;
 	int ret;
 
-#if WATCH_EXEC
-	DRM_INFO("buffers_ptr %d buffer_count %d len %08x\n",
-		  (int) args->buffers_ptr, args->buffer_count, args->batch_len);
-#endif
-
 	if (args->buffer_count < 1) {
 		DRM_ERROR("execbuf2 with %d buffers\n", args->buffer_count);
 		return -EINVAL;
diff --git a/drivers/gpu/drm/i915/i915_irq.c b/drivers/gpu/drm/i915/i915_irq.c
index 95304472f0d0..15d6269027e7 100644
--- a/drivers/gpu/drm/i915/i915_irq.c
+++ b/drivers/gpu/drm/i915/i915_irq.c
@@ -365,7 +365,7 @@ static void notify_ring(struct drm_device *dev,
 		return;
 
 	seqno = ring->get_seqno(ring);
-	trace_i915_gem_request_complete(dev, seqno);
+	trace_i915_gem_request_complete(ring, seqno);
 
 	ring->irq_seqno = seqno;
 	wake_up_all(&ring->irq_queue);
@@ -1273,16 +1273,6 @@ static int i915_emit_irq(struct drm_device * dev)
 	return dev_priv->counter;
 }
 
-void i915_trace_irq_get(struct drm_device *dev, u32 seqno)
-{
-	drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private;
-	struct intel_ring_buffer *ring = LP_RING(dev_priv);
-
-	if (dev_priv->trace_irq_seqno == 0 &&
-	    ring->irq_get(ring))
-		dev_priv->trace_irq_seqno = seqno;
-}
-
 static int i915_wait_irq(struct drm_device * dev, int irq_nr)
 {
 	drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private;
diff --git a/drivers/gpu/drm/i915/i915_trace.h b/drivers/gpu/drm/i915/i915_trace.h
index 7f0fc3ed61aa..d623fefbfaca 100644
--- a/drivers/gpu/drm/i915/i915_trace.h
+++ b/drivers/gpu/drm/i915/i915_trace.h
@@ -7,6 +7,7 @@
 
 #include <drm/drmP.h>
 #include "i915_drv.h"
+#include "intel_ringbuffer.h"
 
 #undef TRACE_SYSTEM
 #define TRACE_SYSTEM i915
@@ -16,9 +17,7 @@
 /* object tracking */
 
 TRACE_EVENT(i915_gem_object_create,
-
 	    TP_PROTO(struct drm_i915_gem_object *obj),
-
 	    TP_ARGS(obj),
 
 	    TP_STRUCT__entry(
@@ -35,33 +34,51 @@ TRACE_EVENT(i915_gem_object_create,
 );
 
 TRACE_EVENT(i915_gem_object_bind,
-
-	    TP_PROTO(struct drm_i915_gem_object *obj, u32 gtt_offset, bool mappable),
-
-	    TP_ARGS(obj, gtt_offset, mappable),
+	    TP_PROTO(struct drm_i915_gem_object *obj, bool mappable),
+	    TP_ARGS(obj, mappable),
 
 	    TP_STRUCT__entry(
 			     __field(struct drm_i915_gem_object *, obj)
-			     __field(u32, gtt_offset)
+			     __field(u32, offset)
+			     __field(u32, size)
 			     __field(bool, mappable)
 			     ),
 
 	    TP_fast_assign(
 			   __entry->obj = obj;
-			   __entry->gtt_offset = gtt_offset;
+			   __entry->offset = obj->gtt_space->start;
+			   __entry->size = obj->gtt_space->size;
 			   __entry->mappable = mappable;
 			   ),
 
-	    TP_printk("obj=%p, gtt_offset=%08x%s",
-		      __entry->obj, __entry->gtt_offset,
+	    TP_printk("obj=%p, offset=%08x size=%x%s",
+		      __entry->obj, __entry->offset, __entry->size,
 		      __entry->mappable ? ", mappable" : "")
 );
 
-TRACE_EVENT(i915_gem_object_change_domain,
+TRACE_EVENT(i915_gem_object_unbind,
+	    TP_PROTO(struct drm_i915_gem_object *obj),
+	    TP_ARGS(obj),
+
+	    TP_STRUCT__entry(
+			     __field(struct drm_i915_gem_object *, obj)
+			     __field(u32, offset)
+			     __field(u32, size)
+			     ),
 
-	    TP_PROTO(struct drm_i915_gem_object *obj, uint32_t old_read_domains, uint32_t old_write_domain),
+	    TP_fast_assign(
+			   __entry->obj = obj;
+			   __entry->offset = obj->gtt_space->start;
+			   __entry->size = obj->gtt_space->size;
+			   ),
 
-	    TP_ARGS(obj, old_read_domains, old_write_domain),
+	    TP_printk("obj=%p, offset=%08x size=%x",
+		      __entry->obj, __entry->offset, __entry->size)
+);
+
+TRACE_EVENT(i915_gem_object_change_domain,
+	    TP_PROTO(struct drm_i915_gem_object *obj, u32 old_read, u32 old_write),
+	    TP_ARGS(obj, old_read, old_write),
 
 	    TP_STRUCT__entry(
 			     __field(struct drm_i915_gem_object *, obj)
@@ -71,177 +88,264 @@ TRACE_EVENT(i915_gem_object_change_domain,
 
 	    TP_fast_assign(
 			   __entry->obj = obj;
-			   __entry->read_domains = obj->base.read_domains | (old_read_domains << 16);
-			   __entry->write_domain = obj->base.write_domain | (old_write_domain << 16);
+			   __entry->read_domains = obj->base.read_domains | (old_read << 16);
+			   __entry->write_domain = obj->base.write_domain | (old_write << 16);
 			   ),
 
-	    TP_printk("obj=%p, read=%04x, write=%04x",
+	    TP_printk("obj=%p, read=%02x=>%02x, write=%02x=>%02x",
 		      __entry->obj,
-		      __entry->read_domains, __entry->write_domain)
+		      __entry->read_domains >> 16,
+		      __entry->read_domains & 0xffff,
+		      __entry->write_domain >> 16,
+		      __entry->write_domain & 0xffff)
 );
 
-DECLARE_EVENT_CLASS(i915_gem_object,
+TRACE_EVENT(i915_gem_object_pwrite,
+	    TP_PROTO(struct drm_i915_gem_object *obj, u32 offset, u32 len),
+	    TP_ARGS(obj, offset, len),
 
-	    TP_PROTO(struct drm_i915_gem_object *obj),
+	    TP_STRUCT__entry(
+			     __field(struct drm_i915_gem_object *, obj)
+			     __field(u32, offset)
+			     __field(u32, len)
+			     ),
 
-	    TP_ARGS(obj),
+	    TP_fast_assign(
+			   __entry->obj = obj;
+			   __entry->offset = offset;
+			   __entry->len = len;
+			   ),
+
+	    TP_printk("obj=%p, offset=%u, len=%u",
+		      __entry->obj, __entry->offset, __entry->len)
+);
+
+TRACE_EVENT(i915_gem_object_pread,
+	    TP_PROTO(struct drm_i915_gem_object *obj, u32 offset, u32 len),
+	    TP_ARGS(obj, offset, len),
 
 	    TP_STRUCT__entry(
 			     __field(struct drm_i915_gem_object *, obj)
+			     __field(u32, offset)
+			     __field(u32, len)
 			     ),
 
 	    TP_fast_assign(
 			   __entry->obj = obj;
+			   __entry->offset = offset;
+			   __entry->len = len;
 			   ),
 
-	    TP_printk("obj=%p", __entry->obj)
+	    TP_printk("obj=%p, offset=%u, len=%u",
+		      __entry->obj, __entry->offset, __entry->len)
 );
 
-DEFINE_EVENT(i915_gem_object, i915_gem_object_clflush,
+TRACE_EVENT(i915_gem_object_fault,
+	    TP_PROTO(struct drm_i915_gem_object *obj, u32 index, bool gtt, bool write),
+	    TP_ARGS(obj, index, gtt, write),
+
+	    TP_STRUCT__entry(
+			     __field(struct drm_i915_gem_object *, obj)
+			     __field(u32, index)
+			     __field(bool, gtt)
+			     __field(bool, write)
+			     ),
+
+	    TP_fast_assign(
+			   __entry->obj = obj;
+			   __entry->index = index;
+			   __entry->gtt = gtt;
+			   __entry->write = write;
+			   ),
 
+	    TP_printk("obj=%p, %s index=%u %s",
+		      __entry->obj,
+		      __entry->gtt ? "GTT" : "CPU",
+		      __entry->index,
+		      __entry->write ? ", writable" : "")
+);
+
+DECLARE_EVENT_CLASS(i915_gem_object,
 	    TP_PROTO(struct drm_i915_gem_object *obj),
+	    TP_ARGS(obj),
 
-	    TP_ARGS(obj)
+	    TP_STRUCT__entry(
+			     __field(struct drm_i915_gem_object *, obj)
+			     ),
+
+	    TP_fast_assign(
+			   __entry->obj = obj;
+			   ),
+
+	    TP_printk("obj=%p", __entry->obj)
 );
 
-DEFINE_EVENT(i915_gem_object, i915_gem_object_unbind,
+DEFINE_EVENT(i915_gem_object, i915_gem_object_clflush,
+	     TP_PROTO(struct drm_i915_gem_object *obj),
+	     TP_ARGS(obj)
+);
 
+DEFINE_EVENT(i915_gem_object, i915_gem_object_destroy,
 	    TP_PROTO(struct drm_i915_gem_object *obj),
-
 	    TP_ARGS(obj)
 );
 
-DEFINE_EVENT(i915_gem_object, i915_gem_object_destroy,
+TRACE_EVENT(i915_gem_evict,
+	    TP_PROTO(struct drm_device *dev, u32 size, u32 align, bool mappable),
+	    TP_ARGS(dev, size, align, mappable),
 
-	    TP_PROTO(struct drm_i915_gem_object *obj),
+	    TP_STRUCT__entry(
+			     __field(u32, dev)
+			     __field(u32, size)
+			     __field(u32, align)
+			     __field(bool, mappable)
+			    ),
 
-	    TP_ARGS(obj)
+	    TP_fast_assign(
+			   __entry->dev = dev->primary->index;
+			   __entry->size = size;
+			   __entry->align = align;
+			   __entry->mappable = mappable;
+			  ),
+
+	    TP_printk("dev=%d, size=%d, align=%d %s",
+		      __entry->dev, __entry->size, __entry->align,
+		      __entry->mappable ? ", mappable" : "")
 );
 
-/* batch tracing */
+TRACE_EVENT(i915_gem_evict_everything,
+	    TP_PROTO(struct drm_device *dev, bool purgeable),
+	    TP_ARGS(dev, purgeable),
 
-TRACE_EVENT(i915_gem_request_submit,
+	    TP_STRUCT__entry(
+			     __field(u32, dev)
+			     __field(bool, purgeable)
+			    ),
+
+	    TP_fast_assign(
+			   __entry->dev = dev->primary->index;
+			   __entry->purgeable = purgeable;
+			  ),
 
-	    TP_PROTO(struct drm_device *dev, u32 seqno),
+	    TP_printk("dev=%d%s",
+		      __entry->dev,
+		      __entry->purgeable ? ", purgeable only" : "")
+);
 
-	    TP_ARGS(dev, seqno),
+TRACE_EVENT(i915_gem_ring_dispatch,
+	    TP_PROTO(struct intel_ring_buffer *ring, u32 seqno),
+	    TP_ARGS(ring, seqno),
 
 	    TP_STRUCT__entry(
 			     __field(u32, dev)
+			     __field(u32, ring)
 			     __field(u32, seqno)
 			     ),
 
 	    TP_fast_assign(
-			   __entry->dev = dev->primary->index;
+			   __entry->dev = ring->dev->primary->index;
+			   __entry->ring = ring->id;
 			   __entry->seqno = seqno;
-			   i915_trace_irq_get(dev, seqno);
+			   i915_trace_irq_get(ring, seqno);
 			   ),
 
-	    TP_printk("dev=%u, seqno=%u", __entry->dev, __entry->seqno)
+	    TP_printk("dev=%u, ring=%u, seqno=%u",
+		      __entry->dev, __entry->ring, __entry->seqno)
 );
 
-TRACE_EVENT(i915_gem_request_flush,
-
-	    TP_PROTO(struct drm_device *dev, u32 seqno,
-		     u32 flush_domains, u32 invalidate_domains),
-
-	    TP_ARGS(dev, seqno, flush_domains, invalidate_domains),
+TRACE_EVENT(i915_gem_ring_flush,
+	    TP_PROTO(struct intel_ring_buffer *ring, u32 invalidate, u32 flush),
+	    TP_ARGS(ring, invalidate, flush),
 
 	    TP_STRUCT__entry(
 			     __field(u32, dev)
-			     __field(u32, seqno)
-			     __field(u32, flush_domains)
-			     __field(u32, invalidate_domains)
+			     __field(u32, ring)
+			     __field(u32, invalidate)
+			     __field(u32, flush)
 			     ),
 
 	    TP_fast_assign(
-			   __entry->dev = dev->primary->index;
-			   __entry->seqno = seqno;
-			   __entry->flush_domains = flush_domains;
-			   __entry->invalidate_domains = invalidate_domains;
+			   __entry->dev = ring->dev->primary->index;
+			   __entry->ring = ring->id;
+			   __entry->invalidate = invalidate;
+			   __entry->flush = flush;
 			   ),
 
-	    TP_printk("dev=%u, seqno=%u, flush=%04x, invalidate=%04x",
-		      __entry->dev, __entry->seqno,
-		      __entry->flush_domains, __entry->invalidate_domains)
+	    TP_printk("dev=%u, ring=%x, invalidate=%04x, flush=%04x",
+		      __entry->dev, __entry->ring,
+		      __entry->invalidate, __entry->flush)
 );
 
 DECLARE_EVENT_CLASS(i915_gem_request,
-
-	    TP_PROTO(struct drm_device *dev, u32 seqno),
-
-	    TP_ARGS(dev, seqno),
+	    TP_PROTO(struct intel_ring_buffer *ring, u32 seqno),
+	    TP_ARGS(ring, seqno),
 
 	    TP_STRUCT__entry(
 			     __field(u32, dev)
+			     __field(u32, ring)
 			     __field(u32, seqno)
 			     ),
 
 	    TP_fast_assign(
-			   __entry->dev = dev->primary->index;
+			   __entry->dev = ring->dev->primary->index;
+			   __entry->ring = ring->id;
 			   __entry->seqno = seqno;
 			   ),
 
-	    TP_printk("dev=%u, seqno=%u", __entry->dev, __entry->seqno)
+	    TP_printk("dev=%u, ring=%u, seqno=%u",
+		      __entry->dev, __entry->ring, __entry->seqno)
 );
 
-DEFINE_EVENT(i915_gem_request, i915_gem_request_complete,
-
-	    TP_PROTO(struct drm_device *dev, u32 seqno),
+DEFINE_EVENT(i915_gem_request, i915_gem_request_add,
+	    TP_PROTO(struct intel_ring_buffer *ring, u32 seqno),
+	    TP_ARGS(ring, seqno)
+);
 
-	    TP_ARGS(dev, seqno)
+DEFINE_EVENT(i915_gem_request, i915_gem_request_complete,
+	    TP_PROTO(struct intel_ring_buffer *ring, u32 seqno),
+	    TP_ARGS(ring, seqno)
 );
 
 DEFINE_EVENT(i915_gem_request, i915_gem_request_retire,
-
-	    TP_PROTO(struct drm_device *dev, u32 seqno),
-
-	    TP_ARGS(dev, seqno)
+	    TP_PROTO(struct intel_ring_buffer *ring, u32 seqno),
+	    TP_ARGS(ring, seqno)
 );
 
 DEFINE_EVENT(i915_gem_request, i915_gem_request_wait_begin,
-
-	    TP_PROTO(struct drm_device *dev, u32 seqno),
-
-	    TP_ARGS(dev, seqno)
+	    TP_PROTO(struct intel_ring_buffer *ring, u32 seqno),
+	    TP_ARGS(ring, seqno)
 );
 
 DEFINE_EVENT(i915_gem_request, i915_gem_request_wait_end,
-
-	    TP_PROTO(struct drm_device *dev, u32 seqno),
-
-	    TP_ARGS(dev, seqno)
+	    TP_PROTO(struct intel_ring_buffer *ring, u32 seqno),
+	    TP_ARGS(ring, seqno)
 );
 
 DECLARE_EVENT_CLASS(i915_ring,
-
-	    TP_PROTO(struct drm_device *dev),
-
-	    TP_ARGS(dev),
+	    TP_PROTO(struct intel_ring_buffer *ring),
+	    TP_ARGS(ring),
 
 	    TP_STRUCT__entry(
 			     __field(u32, dev)
+			     __field(u32, ring)
 			     ),
 
 	    TP_fast_assign(
-			   __entry->dev = dev->primary->index;
+			   __entry->dev = ring->dev->primary->index;
+			   __entry->ring = ring->id;
 			   ),
 
-	    TP_printk("dev=%u", __entry->dev)
+	    TP_printk("dev=%u, ring=%u", __entry->dev, __entry->ring)
 );
 
 DEFINE_EVENT(i915_ring, i915_ring_wait_begin,
-
-	    TP_PROTO(struct drm_device *dev),
-
-	    TP_ARGS(dev)
+	    TP_PROTO(struct intel_ring_buffer *ring),
+	    TP_ARGS(ring)
 );
 
 DEFINE_EVENT(i915_ring, i915_ring_wait_end,
-
-	    TP_PROTO(struct drm_device *dev),
-
-	    TP_ARGS(dev)
+	    TP_PROTO(struct intel_ring_buffer *ring),
+	    TP_ARGS(ring)
 );
 
 TRACE_EVENT(i915_flip_request,
@@ -281,26 +385,29 @@ TRACE_EVENT(i915_flip_complete,
 );
 
 TRACE_EVENT(i915_reg_rw,
-           TP_PROTO(int cmd, uint32_t reg, uint64_t val, int len),
+           TP_PROTO(bool write, u32 reg, u64 val, int len),
 
-           TP_ARGS(cmd, reg, val, len),
+           TP_ARGS(write, reg, val, len),
 
            TP_STRUCT__entry(
-                   __field(int, cmd)
-                   __field(uint32_t, reg)
-                   __field(uint64_t, val)
-                   __field(int, len)
+                   __field(u64, val)
+                   __field(u32, reg)
+                   __field(u16, write)
+                   __field(u16, len)
                    ),
 
            TP_fast_assign(
-                   __entry->cmd = cmd;
+                   __entry->val = (u64)val;
                    __entry->reg = reg;
-                   __entry->val = (uint64_t)val;
+                   __entry->write = write;
                    __entry->len = len;
                    ),
 
-           TP_printk("cmd=%c, reg=0x%x, val=0x%llx, len=%d",
-                     __entry->cmd, __entry->reg, __entry->val, __entry->len)
+           TP_printk("%s reg=0x%x, len=%d, val=(0x%x, 0x%x)",
+                     __entry->write ? "write" : "read",
+		     __entry->reg, __entry->len,
+		     (u32)(__entry->val & 0xffffffff),
+		     (u32)(__entry->val >> 32))
 );
 
 #endif /* _I915_TRACE_H_ */
diff --git a/drivers/gpu/drm/i915/intel_overlay.c b/drivers/gpu/drm/i915/intel_overlay.c
index 3fbb98b948d6..d2fdfd589c85 100644
--- a/drivers/gpu/drm/i915/intel_overlay.c
+++ b/drivers/gpu/drm/i915/intel_overlay.c
@@ -221,16 +221,15 @@ static int intel_overlay_do_wait_request(struct intel_overlay *overlay,
 	int ret;
 
 	BUG_ON(overlay->last_flip_req);
-	ret = i915_add_request(dev, NULL, request, LP_RING(dev_priv));
+	ret = i915_add_request(LP_RING(dev_priv), NULL, request);
 	if (ret) {
 	    kfree(request);
 	    return ret;
 	}
 	overlay->last_flip_req = request->seqno;
 	overlay->flip_tail = tail;
-	ret = i915_do_wait_request(dev,
-				   overlay->last_flip_req, true,
-				   LP_RING(dev_priv));
+	ret = i915_wait_request(LP_RING(dev_priv),
+				overlay->last_flip_req, true);
 	if (ret)
 		return ret;
 
@@ -364,7 +363,7 @@ static int intel_overlay_continue(struct intel_overlay *overlay,
 	OUT_RING(flip_addr);
         ADVANCE_LP_RING();
 
-	ret = i915_add_request(dev, NULL, request, LP_RING(dev_priv));
+	ret = i915_add_request(LP_RING(dev_priv), NULL, request);
 	if (ret) {
 		kfree(request);
 		return ret;
@@ -453,8 +452,8 @@ static int intel_overlay_recover_from_interrupt(struct intel_overlay *overlay,
 	if (overlay->last_flip_req == 0)
 		return 0;
 
-	ret = i915_do_wait_request(dev, overlay->last_flip_req,
-				   interruptible, LP_RING(dev_priv));
+	ret = i915_wait_request(LP_RING(dev_priv),
+				overlay->last_flip_req, interruptible);
 	if (ret)
 		return ret;
 
diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.c b/drivers/gpu/drm/i915/intel_ringbuffer.c
index 235d9c4b40ae..ec7175e0dcd8 100644
--- a/drivers/gpu/drm/i915/intel_ringbuffer.c
+++ b/drivers/gpu/drm/i915/intel_ringbuffer.c
@@ -62,18 +62,9 @@ render_ring_flush(struct intel_ring_buffer *ring,
 		  u32	flush_domains)
 {
 	struct drm_device *dev = ring->dev;
-	drm_i915_private_t *dev_priv = dev->dev_private;
 	u32 cmd;
 	int ret;
 
-#if WATCH_EXEC
-	DRM_INFO("%s: invalidate %08x flush %08x\n", __func__,
-		  invalidate_domains, flush_domains);
-#endif
-
-	trace_i915_gem_request_flush(dev, dev_priv->next_seqno,
-				     invalidate_domains, flush_domains);
-
 	if ((invalidate_domains | flush_domains) & I915_GEM_GPU_DOMAINS) {
 		/*
 		 * read/write caches:
@@ -122,9 +113,6 @@ render_ring_flush(struct intel_ring_buffer *ring,
 		    (IS_G4X(dev) || IS_GEN5(dev)))
 			cmd |= MI_INVALIDATE_ISP;
 
-#if WATCH_EXEC
-		DRM_INFO("%s: queue flush %08x to ring\n", __func__, cmd);
-#endif
 		ret = intel_ring_begin(ring, 2);
 		if (ret)
 			return ret;
@@ -714,11 +702,8 @@ render_ring_dispatch_execbuffer(struct intel_ring_buffer *ring,
 				u32 offset, u32 len)
 {
 	struct drm_device *dev = ring->dev;
-	drm_i915_private_t *dev_priv = dev->dev_private;
 	int ret;
 
-	trace_i915_gem_request_submit(dev, dev_priv->next_seqno + 1);
-
 	if (IS_I830(dev) || IS_845G(dev)) {
 		ret = intel_ring_begin(ring, 4);
 		if (ret)
@@ -953,13 +938,13 @@ int intel_wait_ring_buffer(struct intel_ring_buffer *ring, int n)
 			return 0;
 	}
 
-	trace_i915_ring_wait_begin (dev);
+	trace_i915_ring_wait_begin(ring);
 	end = jiffies + 3 * HZ;
 	do {
 		ring->head = I915_READ_HEAD(ring);
 		ring->space = ring_space(ring);
 		if (ring->space >= n) {
-			trace_i915_ring_wait_end(dev);
+			trace_i915_ring_wait_end(ring);
 			return 0;
 		}
 
@@ -973,7 +958,7 @@ int intel_wait_ring_buffer(struct intel_ring_buffer *ring, int n)
 		if (atomic_read(&dev_priv->mm.wedged))
 			return -EAGAIN;
 	} while (!time_after(jiffies, end));
-	trace_i915_ring_wait_end (dev);
+	trace_i915_ring_wait_end(ring);
 	return -EBUSY;
 }
 
diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.h b/drivers/gpu/drm/i915/intel_ringbuffer.h
index 5e14b09f67ce..bd6a5fbfa929 100644
--- a/drivers/gpu/drm/i915/intel_ringbuffer.h
+++ b/drivers/gpu/drm/i915/intel_ringbuffer.h
@@ -58,6 +58,7 @@ struct  intel_ring_buffer {
 	u32		irq_refcount;
 	u32		irq_mask;
 	u32		irq_seqno;		/* last seq seem at irq time */
+	u32		trace_irq_seqno;
 	u32		waiting_seqno;
 	u32		sync_seqno[I915_NUM_RINGS-1];
 	bool __must_check (*irq_get)(struct intel_ring_buffer *ring);
@@ -186,6 +187,12 @@ int intel_init_blt_ring_buffer(struct drm_device *dev);
 u32 intel_ring_get_active_head(struct intel_ring_buffer *ring);
 void intel_ring_setup_status_page(struct intel_ring_buffer *ring);
 
+static inline void i915_trace_irq_get(struct intel_ring_buffer *ring, u32 seqno)
+{
+	if (ring->trace_irq_seqno == 0 && ring->irq_get(ring))
+		ring->trace_irq_seqno = seqno;
+}
+
 /* DRI warts */
 int intel_render_ring_init_dri(struct drm_device *dev, u64 start, u32 size);