drm/i915: convert some gem structures to per-ring V2

The active list and request list move into the ringbuffer structure,
so each can track its active objects in the order they are in that
ring.  The flushing list does not, as it doesn't matter which ring
caused data to end up in the render cache.  Objects gain a pointer to
the ring they are active on (if any).
Signed-off-by: NZou Nan hai <nanhai.zou@intel.com>
Signed-off-by: NXiang Hai hao <haihao.xiang@intel.com>
Signed-off-by: NEric Anholt <eric@anholt.net>

drivers/gpu/drm/i915/i915_debugfs.c

@@ -77,7 +77,7 @@ static int i915_gem_object_list_info(struct seq_file *m, void *data)
 	case ACTIVE_LIST:
 		seq_printf(m, "Active:\n");
 		lock = &dev_priv->mm.active_list_lock;
-		head = &dev_priv->mm.active_list;
+		head = &dev_priv->render_ring.active_list;
 		break;
 	case INACTIVE_LIST:
 		seq_printf(m, "Inactive:\n");
@@ -129,7 +129,8 @@ static int i915_gem_request_info(struct seq_file *m, void *data)
 	struct drm_i915_gem_request *gem_request;
 
 	seq_printf(m, "Request:\n");
-	list_for_each_entry(gem_request, &dev_priv->mm.request_list, list) {
+	list_for_each_entry(gem_request, &dev_priv->render_ring.request_list,
+			list) {
 		seq_printf(m, "    %d @ %d\n",
 			   gem_request->seqno,
 			   (int) (jiffies - gem_request->emitted_jiffies));
@@ -145,7 +146,7 @@ static int i915_gem_seqno_info(struct seq_file *m, void *data)
 
 	if (dev_priv->hw_status_page != NULL) {
 		seq_printf(m, "Current sequence: %d\n",
-			   i915_get_gem_seqno(dev));
+			   i915_get_gem_seqno(dev,  &dev_priv->render_ring));
 	} else {
 		seq_printf(m, "Current sequence: hws uninitialized\n");
 	}
@@ -197,7 +198,7 @@ static int i915_interrupt_info(struct seq_file *m, void *data)
 		   atomic_read(&dev_priv->irq_received));
 	if (dev_priv->hw_status_page != NULL) {
 		seq_printf(m, "Current sequence:    %d\n",
-			   i915_get_gem_seqno(dev));
+			   i915_get_gem_seqno(dev,  &dev_priv->render_ring));
 	} else {
 		seq_printf(m, "Current sequence:    hws uninitialized\n");
 	}
@@ -287,7 +288,8 @@ static int i915_batchbuffer_info(struct seq_file *m, void *data)
 
 	spin_lock(&dev_priv->mm.active_list_lock);
 
-	list_for_each_entry(obj_priv, &dev_priv->mm.active_list, list) {
+	list_for_each_entry(obj_priv, &dev_priv->render_ring.active_list,
+			list) {
 		obj = &obj_priv->base;
 		if (obj->read_domains & I915_GEM_DOMAIN_COMMAND) {
 		    ret = i915_gem_object_get_pages(obj, 0);

drivers/gpu/drm/i915/i915_dma.c

@@ -82,7 +82,8 @@ static void i915_free_hws(struct drm_device *dev)
 		dev_priv->status_page_dmah = NULL;
 	}
 
-	if (dev_priv->status_gfx_addr) {
+	if (dev_priv->render_ring.status_page.gfx_addr) {
+		dev_priv->render_ring.status_page.gfx_addr = 0;
 		dev_priv->status_gfx_addr = 0;
 		drm_core_ioremapfree(&dev_priv->hws_map, dev);
 	}
@@ -835,9 +836,9 @@ static int i915_set_status_page(struct drm_device *dev, void *data,
 	I915_WRITE(HWS_PGA, ring->status_page.gfx_addr);
 
 	DRM_DEBUG_DRIVER("load hws HWS_PGA with gfx mem 0x%x\n",
-				dev_priv->status_gfx_addr);
+			dev_priv->status_gfx_addr);
 	DRM_DEBUG_DRIVER("load hws at %p\n",
-				dev_priv->hw_status_page);
+			dev_priv->hw_status_page);
 	return 0;
 }
 
@@ -1510,7 +1511,6 @@ int i915_driver_load(struct drm_device *dev, unsigned long flags)
 	resource_size_t base, size;
 	int ret = 0, mmio_bar;
 	uint32_t agp_size, prealloc_size, prealloc_start;
-
 	/* i915 has 4 more counters */
 	dev->counters += 4;
 	dev->types[6] = _DRM_STAT_IRQ;

drivers/gpu/drm/i915/i915_drv.c

@@ -340,7 +340,7 @@ int i965_reset(struct drm_device *dev, u8 flags)
 	/*
 	 * Clear request list
 	 */
-	i915_gem_retire_requests(dev);
+	i915_gem_retire_requests(dev, &dev_priv->render_ring);
 
 	if (need_display)
 		i915_save_display(dev);

drivers/gpu/drm/i915/i915_drv.h

@@ -505,18 +505,7 @@ typedef struct drm_i915_private {
 		 */
 		struct list_head shrink_list;
 
-		/**
-		 * List of objects currently involved in rendering from the
-		 * ringbuffer.
-		 *
-		 * Includes buffers having the contents of their GPU caches
-		 * flushed, not necessarily primitives.  last_rendering_seqno
-		 * represents when the rendering involved will be completed.
-		 *
-		 * A reference is held on the buffer while on this list.
-		 */
 		spinlock_t active_list_lock;
-		struct list_head active_list;
 
 		/**
 		 * List of objects which are not in the ringbuffer but which
@@ -553,12 +542,6 @@ typedef struct drm_i915_private {
 		/** LRU list of objects with fence regs on them. */
 		struct list_head fence_list;
 
-		/**
-		 * List of breadcrumbs associated with GPU requests currently
-		 * outstanding.
-		 */
-		struct list_head request_list;
-
 		/**
 		 * We leave the user IRQ off as much as possible,
 		 * but this means that requests will finish and never
@@ -683,6 +666,9 @@ struct drm_i915_gem_object {
 	 */
 	uint32_t gtt_offset;
 
+	/* Which ring is refering to is this object */
+	struct intel_ring_buffer *ring;
+
 	/**
 	 * Fake offset for use by mmap(2)
 	 */
@@ -756,6 +742,9 @@ struct drm_i915_gem_object {
  * an emission time with seqnos for tracking how far ahead of the GPU we are.
  */
 struct drm_i915_gem_request {
+	/** On Which ring this request was generated */
+	struct intel_ring_buffer *ring;
+
 	/** GEM sequence number associated with this request. */
 	uint32_t seqno;
 
@@ -916,11 +905,13 @@ void i915_gem_object_unpin(struct drm_gem_object *obj);
 int i915_gem_object_unbind(struct drm_gem_object *obj);
 void i915_gem_release_mmap(struct drm_gem_object *obj);
 void i915_gem_lastclose(struct drm_device *dev);
-uint32_t i915_get_gem_seqno(struct drm_device *dev);
+uint32_t i915_get_gem_seqno(struct drm_device *dev,
+		struct intel_ring_buffer *ring);
 bool i915_seqno_passed(uint32_t seq1, uint32_t seq2);
 int i915_gem_object_get_fence_reg(struct drm_gem_object *obj);
 int i915_gem_object_put_fence_reg(struct drm_gem_object *obj);
-void i915_gem_retire_requests(struct drm_device *dev);
+void i915_gem_retire_requests(struct drm_device *dev,
+		 struct intel_ring_buffer *ring);
 void i915_gem_retire_work_handler(struct work_struct *work);
 void i915_gem_clflush_object(struct drm_gem_object *obj);
 int i915_gem_object_set_domain(struct drm_gem_object *obj,
@@ -931,9 +922,13 @@ void i915_gem_cleanup_ringbuffer(struct drm_device *dev);
 int i915_gem_do_init(struct drm_device *dev, unsigned long start,
 		     unsigned long end);
 int i915_gem_idle(struct drm_device *dev);
-uint32_t i915_add_request(struct drm_device *dev, struct drm_file *file_priv,
-			  uint32_t flush_domains);
-int i915_do_wait_request(struct drm_device *dev, uint32_t seqno, int interruptible);
+uint32_t i915_add_request(struct drm_device *dev,
+		struct drm_file *file_priv,
+		uint32_t flush_domains,
+		struct intel_ring_buffer *ring);
+int i915_do_wait_request(struct drm_device *dev,
+		uint32_t seqno, int interruptible,
+		struct intel_ring_buffer *ring);
 int i915_gem_fault(struct vm_area_struct *vma, struct vm_fault *vmf);
 int i915_gem_object_set_to_gtt_domain(struct drm_gem_object *obj,
 				      int write);

drivers/gpu/drm/i915/i915_irq.c

@@ -331,6 +331,7 @@ irqreturn_t ironlake_irq_handler(struct drm_device *dev)
 	int ret = IRQ_NONE;
 	u32 de_iir, gt_iir, de_ier, pch_iir;
 	struct drm_i915_master_private *master_priv;
+	struct intel_ring_buffer *render_ring = &dev_priv->render_ring;
 
 	/* disable master interrupt before clearing iir  */
 	de_ier = I915_READ(DEIER);
@@ -354,10 +355,10 @@ irqreturn_t ironlake_irq_handler(struct drm_device *dev)
 	}
 
 	if (gt_iir & GT_PIPE_NOTIFY) {
-		u32 seqno = i915_get_gem_seqno(dev);
-		dev_priv->mm.irq_gem_seqno = seqno;
+		u32 seqno = render_ring->get_gem_seqno(dev, render_ring);
+		render_ring->irq_gem_seqno = seqno;
 		trace_i915_gem_request_complete(dev, seqno);
-		DRM_WAKEUP(&dev_priv->irq_queue);
+		DRM_WAKEUP(&dev_priv->render_ring.irq_queue);
 		dev_priv->hangcheck_count = 0;
 		mod_timer(&dev_priv->hangcheck_timer, jiffies + DRM_I915_HANGCHECK_PERIOD);
 	}
@@ -588,7 +589,7 @@ static void i915_capture_error_state(struct drm_device *dev)
 		return;
 	}
 
-	error->seqno = i915_get_gem_seqno(dev);
+	error->seqno = i915_get_gem_seqno(dev, &dev_priv->render_ring);
 	error->eir = I915_READ(EIR);
 	error->pgtbl_er = I915_READ(PGTBL_ER);
 	error->pipeastat = I915_READ(PIPEASTAT);
@@ -616,7 +617,9 @@ static void i915_capture_error_state(struct drm_device *dev)
 	batchbuffer[0] = NULL;
 	batchbuffer[1] = NULL;
 	count = 0;
-	list_for_each_entry(obj_priv, &dev_priv->mm.active_list, list) {
+	list_for_each_entry(obj_priv,
+			&dev_priv->render_ring.active_list, list) {
+
 		struct drm_gem_object *obj = &obj_priv->base;
 
 		if (batchbuffer[0] == NULL &&
@@ -653,7 +656,8 @@ static void i915_capture_error_state(struct drm_device *dev)
 
 	if (error->active_bo) {
 		int i = 0;
-		list_for_each_entry(obj_priv, &dev_priv->mm.active_list, list) {
+		list_for_each_entry(obj_priv,
+				&dev_priv->render_ring.active_list, list) {
 			struct drm_gem_object *obj = &obj_priv->base;
 
 			error->active_bo[i].size = obj->size;
@@ -831,7 +835,7 @@ static void i915_handle_error(struct drm_device *dev, bool wedged)
 		/*
 		 * Wakeup waiting processes so they don't hang
 		 */
-		DRM_WAKEUP(&dev_priv->irq_queue);
+		DRM_WAKEUP(&dev_priv->render_ring.irq_queue);
 	}
 
 	queue_work(dev_priv->wq, &dev_priv->error_work);
@@ -850,6 +854,7 @@ irqreturn_t i915_driver_irq_handler(DRM_IRQ_ARGS)
 	unsigned long irqflags;
 	int irq_received;
 	int ret = IRQ_NONE;
+	struct intel_ring_buffer *render_ring = &dev_priv->render_ring;
 
 	atomic_inc(&dev_priv->irq_received);
 
@@ -930,10 +935,11 @@ irqreturn_t i915_driver_irq_handler(DRM_IRQ_ARGS)
 		}
 
 		if (iir & I915_USER_INTERRUPT) {
-			u32 seqno = i915_get_gem_seqno(dev);
-			dev_priv->mm.irq_gem_seqno = seqno;
+			u32 seqno =
+				render_ring->get_gem_seqno(dev, render_ring);
+			render_ring->irq_gem_seqno = seqno;
 			trace_i915_gem_request_complete(dev, seqno);
-			DRM_WAKEUP(&dev_priv->irq_queue);
+			DRM_WAKEUP(&dev_priv->render_ring.irq_queue);
 			dev_priv->hangcheck_count = 0;
 			mod_timer(&dev_priv->hangcheck_timer, jiffies + DRM_I915_HANGCHECK_PERIOD);
 		}
@@ -1038,7 +1044,7 @@ static int i915_wait_irq(struct drm_device * dev, int irq_nr)
 		master_priv->sarea_priv->perf_boxes |= I915_BOX_WAIT;
 
 	render_ring->user_irq_get(dev, render_ring);
-	DRM_WAIT_ON(ret, dev_priv->irq_queue, 3 * DRM_HZ,
+	DRM_WAIT_ON(ret, dev_priv->render_ring.irq_queue, 3 * DRM_HZ,
 		    READ_BREADCRUMB(dev_priv) >= irq_nr);
 	render_ring->user_irq_put(dev, render_ring);
 
@@ -1205,9 +1211,12 @@ int i915_vblank_swap(struct drm_device *dev, void *data,
 	return -EINVAL;
 }
 
-struct drm_i915_gem_request *i915_get_tail_request(struct drm_device *dev) {
+struct drm_i915_gem_request *
+i915_get_tail_request(struct drm_device *dev)
+{
 	drm_i915_private_t *dev_priv = dev->dev_private;
-	return list_entry(dev_priv->mm.request_list.prev, struct drm_i915_gem_request, list);
+	return list_entry(dev_priv->render_ring.request_list.prev,
+			struct drm_i915_gem_request, list);
 }
 
 /**
@@ -1232,8 +1241,10 @@ void i915_hangcheck_elapsed(unsigned long data)
 		acthd = I915_READ(ACTHD_I965);
 
 	/* If all work is done then ACTHD clearly hasn't advanced. */
-	if (list_empty(&dev_priv->mm.request_list) ||
-		       i915_seqno_passed(i915_get_gem_seqno(dev), i915_get_tail_request(dev)->seqno)) {
+	if (list_empty(&dev_priv->render_ring.request_list) ||
+		i915_seqno_passed(i915_get_gem_seqno(dev,
+				&dev_priv->render_ring),
+			i915_get_tail_request(dev)->seqno)) {
 		dev_priv->hangcheck_count = 0;
 		return;
 	}
@@ -1300,7 +1311,7 @@ static int ironlake_irq_postinstall(struct drm_device *dev)
 	(void) I915_READ(DEIER);
 
 	/* user interrupt should be enabled, but masked initial */
-	dev_priv->gt_irq_mask_reg = 0xffffffff;
+	dev_priv->gt_irq_mask_reg = ~render_mask;
 	dev_priv->gt_irq_enable_reg = render_mask;
 
 	I915_WRITE(GTIIR, I915_READ(GTIIR));
@@ -1363,7 +1374,7 @@ int i915_driver_irq_postinstall(struct drm_device *dev)
 	u32 enable_mask = I915_INTERRUPT_ENABLE_FIX | I915_INTERRUPT_ENABLE_VAR;
 	u32 error_mask;
 
-	DRM_INIT_WAITQUEUE(&dev_priv->irq_queue);
+	DRM_INIT_WAITQUEUE(&dev_priv->render_ring.irq_queue);
 
 	dev_priv->vblank_pipe = DRM_I915_VBLANK_PIPE_A | DRM_I915_VBLANK_PIPE_B;
 

drivers/gpu/drm/i915/intel_overlay.c

@@ -212,6 +212,7 @@ static int intel_overlay_on(struct intel_overlay *overlay)
 {
 	struct drm_device *dev = overlay->dev;
 	int ret;
+	drm_i915_private_t *dev_priv = dev->dev_private;
 
 	BUG_ON(overlay->active);
 
@@ -225,11 +226,13 @@ static int intel_overlay_on(struct intel_overlay *overlay)
 	OUT_RING(MI_NOOP);
 	ADVANCE_LP_RING();
 
-	overlay->last_flip_req = i915_add_request(dev, NULL, 0);
+	overlay->last_flip_req =
+		i915_add_request(dev, NULL, 0, &dev_priv->render_ring);
 	if (overlay->last_flip_req == 0)
 		return -ENOMEM;
 
-	ret = i915_do_wait_request(dev, overlay->last_flip_req, 1);
+	ret = i915_do_wait_request(dev,
+			overlay->last_flip_req, 1, &dev_priv->render_ring);
 	if (ret != 0)
 		return ret;
 
@@ -262,7 +265,8 @@ static void intel_overlay_continue(struct intel_overlay *overlay,
 	OUT_RING(flip_addr);
         ADVANCE_LP_RING();
 
-	overlay->last_flip_req = i915_add_request(dev, NULL, 0);
+	overlay->last_flip_req =
+		i915_add_request(dev, NULL, 0, &dev_priv->render_ring);
 }
 
 static int intel_overlay_wait_flip(struct intel_overlay *overlay)
@@ -273,7 +277,8 @@ static int intel_overlay_wait_flip(struct intel_overlay *overlay)
 	u32 tmp;
 
 	if (overlay->last_flip_req != 0) {
-		ret = i915_do_wait_request(dev, overlay->last_flip_req, 1);
+		ret = i915_do_wait_request(dev, overlay->last_flip_req,
+				1, &dev_priv->render_ring);
 		if (ret == 0) {
 			overlay->last_flip_req = 0;
 
@@ -292,11 +297,13 @@ static int intel_overlay_wait_flip(struct intel_overlay *overlay)
         OUT_RING(MI_NOOP);
         ADVANCE_LP_RING();
 
-	overlay->last_flip_req = i915_add_request(dev, NULL, 0);
+	overlay->last_flip_req =
+		i915_add_request(dev, NULL, 0, &dev_priv->render_ring);
 	if (overlay->last_flip_req == 0)
 		return -ENOMEM;
 
-	ret = i915_do_wait_request(dev, overlay->last_flip_req, 1);
+	ret = i915_do_wait_request(dev, overlay->last_flip_req,
+			1, &dev_priv->render_ring);
 	if (ret != 0)
 		return ret;
 
@@ -310,6 +317,7 @@ static int intel_overlay_off(struct intel_overlay *overlay)
 {
 	u32 flip_addr = overlay->flip_addr;
 	struct drm_device *dev = overlay->dev;
+	drm_i915_private_t *dev_priv = dev->dev_private;
 	int ret;
 
 	BUG_ON(!overlay->active);
@@ -330,11 +338,13 @@ static int intel_overlay_off(struct intel_overlay *overlay)
         OUT_RING(MI_NOOP);
         ADVANCE_LP_RING();
 
-	overlay->last_flip_req = i915_add_request(dev, NULL, 0);
+	overlay->last_flip_req =
+		i915_add_request(dev, NULL, 0, &dev_priv->render_ring);
 	if (overlay->last_flip_req == 0)
 		return -ENOMEM;
 
-	ret = i915_do_wait_request(dev, overlay->last_flip_req, 1);
+	ret = i915_do_wait_request(dev, overlay->last_flip_req,
+			1, &dev_priv->render_ring);
 	if (ret != 0)
 		return ret;
 
@@ -348,11 +358,13 @@ static int intel_overlay_off(struct intel_overlay *overlay)
         OUT_RING(MI_NOOP);
 	ADVANCE_LP_RING();
 
-	overlay->last_flip_req = i915_add_request(dev, NULL, 0);
+	overlay->last_flip_req =
+		i915_add_request(dev, NULL, 0, &dev_priv->render_ring);
 	if (overlay->last_flip_req == 0)
 		return -ENOMEM;
 
-	ret = i915_do_wait_request(dev, overlay->last_flip_req, 1);
+	ret = i915_do_wait_request(dev, overlay->last_flip_req,
+			1, &dev_priv->render_ring);
 	if (ret != 0)
 		return ret;
 
@@ -385,6 +397,7 @@ int intel_overlay_recover_from_interrupt(struct intel_overlay *overlay,
 {
 	struct drm_device *dev = overlay->dev;
 	struct drm_gem_object *obj;
+	drm_i915_private_t *dev_priv = dev->dev_private;
 	u32 flip_addr;
 	int ret;
 
@@ -392,12 +405,14 @@ int intel_overlay_recover_from_interrupt(struct intel_overlay *overlay,
 		return -EIO;
 
 	if (overlay->last_flip_req == 0) {
-		overlay->last_flip_req = i915_add_request(dev, NULL, 0);
+		overlay->last_flip_req =
+			i915_add_request(dev, NULL, 0, &dev_priv->render_ring);
 		if (overlay->last_flip_req == 0)
 			return -ENOMEM;
 	}
 
-	ret = i915_do_wait_request(dev, overlay->last_flip_req, interruptible);
+	ret = i915_do_wait_request(dev, overlay->last_flip_req,
+			interruptible, &dev_priv->render_ring);
 	if (ret != 0)
 		return ret;
 
@@ -421,12 +436,13 @@ int intel_overlay_recover_from_interrupt(struct intel_overlay *overlay,
 			OUT_RING(MI_NOOP);
 			ADVANCE_LP_RING();
 
-			overlay->last_flip_req = i915_add_request(dev, NULL, 0);
+			overlay->last_flip_req = i915_add_request(dev, NULL,
+					0, &dev_priv->render_ring);
 			if (overlay->last_flip_req == 0)
 				return -ENOMEM;
 
 			ret = i915_do_wait_request(dev, overlay->last_flip_req,
-					interruptible);
+					interruptible, &dev_priv->render_ring);
 			if (ret != 0)
 				return ret;