drm/i915: Combine seqno + tracking into a global timeline struct

Our timelines are more than just a seqno. They also provide an ordered
list of requests to be executed. Due to the restriction of handling
individual address spaces, we are limited to a timeline per address
space but we use a fence context per engine within.

Our first step to introducing independent timelines per context (i.e. to
allow each context to have a queue of requests to execute that have a
defined set of dependencies on other requests) is to provide a timeline
abstraction for the global execution queue.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20161028125858.23563-23-chris@chris-wilson.co.uk

drivers/gpu/drm/i915/Makefile

@@ -42,6 +42,7 @@ i915-y += i915_cmd_parser.o \
 	  i915_gem_shrinker.o \
 	  i915_gem_stolen.o \
 	  i915_gem_tiling.o \
+	  i915_gem_timeline.o \
 	  i915_gem_userptr.o \
 	  i915_trace_points.o \
 	  intel_breadcrumbs.o \

drivers/gpu/drm/i915/i915_debugfs.c

@@ -552,7 +552,7 @@ static int i915_gem_pageflip_info(struct seq_file *m, void *data)
 				seq_printf(m, "Flip queued on %s at seqno %x, next seqno %x [current breadcrumb %x], completed? %d\n",
 					   engine->name,
 					   i915_gem_request_get_seqno(work->flip_queued_req),
-					   dev_priv->next_seqno,
+					   dev_priv->gt.global_timeline.next_seqno,
 					   intel_engine_get_seqno(engine),
 					   i915_gem_request_completed(work->flip_queued_req));
 			} else
@@ -662,13 +662,13 @@ static int i915_gem_request_info(struct seq_file *m, void *data)
 		int count;
 
 		count = 0;
-		list_for_each_entry(req, &engine->request_list, link)
+		list_for_each_entry(req, &engine->timeline->requests, link)
 			count++;
 		if (count == 0)
 			continue;
 
 		seq_printf(m, "%s requests: %d\n", engine->name, count);
-		list_for_each_entry(req, &engine->request_list, link)
+		list_for_each_entry(req, &engine->timeline->requests, link)
 			print_request(m, req, "    ");
 
 		any++;
@@ -1052,15 +1052,8 @@ static int
 i915_next_seqno_get(void *data, u64 *val)
 {
 	struct drm_i915_private *dev_priv = data;
-	int ret;
-
-	ret = mutex_lock_interruptible(&dev_priv->drm.struct_mutex);
-	if (ret)
-		return ret;
-
-	*val = dev_priv->next_seqno;
-	mutex_unlock(&dev_priv->drm.struct_mutex);
 
+	*val = READ_ONCE(dev_priv->gt.global_timeline.next_seqno);
 	return 0;
 }
 
@@ -1075,7 +1068,7 @@ i915_next_seqno_set(void *data, u64 val)
 	if (ret)
 		return ret;
 
-	ret = i915_gem_set_seqno(dev, val);
+	ret = i915_gem_set_global_seqno(dev, val);
 	mutex_unlock(&dev->struct_mutex);
 
 	return ret;
@@ -1364,7 +1357,7 @@ static int i915_hangcheck_info(struct seq_file *m, void *unused)
 		seq_printf(m, "\tseqno = %x [current %x, last %x]\n",
 			   engine->hangcheck.seqno,
 			   seqno[id],
-			   engine->last_submitted_seqno);
+			   engine->timeline->last_submitted_seqno);
 		seq_printf(m, "\twaiters? %s, fake irq active? %s\n",
 			   yesno(intel_engine_has_waiter(engine)),
 			   yesno(test_bit(engine->id,
@@ -3181,22 +3174,22 @@ static int i915_engine_info(struct seq_file *m, void *unused)
 		seq_printf(m, "%s\n", engine->name);
 		seq_printf(m, "\tcurrent seqno %x, last %x, hangcheck %x [score %d]\n",
 			   intel_engine_get_seqno(engine),
-			   engine->last_submitted_seqno,
+			   engine->timeline->last_submitted_seqno,
 			   engine->hangcheck.seqno,
 			   engine->hangcheck.score);
 
 		rcu_read_lock();
 
 		seq_printf(m, "\tRequests:\n");
 
-		rq = list_first_entry(&engine->request_list,
-				struct drm_i915_gem_request, link);
-		if (&rq->link != &engine->request_list)
+		rq = list_first_entry(&engine->timeline->requests,
+				      struct drm_i915_gem_request, link);
+		if (&rq->link != &engine->timeline->requests)
 			print_request(m, rq, "\t\tfirst  ");
 
-		rq = list_last_entry(&engine->request_list,
-				struct drm_i915_gem_request, link);
-		if (&rq->link != &engine->request_list)
+		rq = list_last_entry(&engine->timeline->requests,
+				     struct drm_i915_gem_request, link);
+		if (&rq->link != &engine->timeline->requests)
 			print_request(m, rq, "\t\tlast   ");
 
 		rq = i915_gem_find_active_request(engine);

drivers/gpu/drm/i915/i915_drv.c

@@ -831,7 +831,9 @@ static int i915_driver_init_early(struct drm_i915_private *dev_priv,
 	intel_init_display_hooks(dev_priv);
 	intel_init_clock_gating_hooks(dev_priv);
 	intel_init_audio_hooks(dev_priv);
-	i915_gem_load_init(&dev_priv->drm);
+	ret = i915_gem_load_init(&dev_priv->drm);
+	if (ret < 0)
+		goto err_gvt;
 
 	intel_display_crc_init(dev_priv);
 
@@ -841,6 +843,8 @@ static int i915_driver_init_early(struct drm_i915_private *dev_priv,
 
 	return 0;
 
+err_gvt:
+	intel_gvt_cleanup(dev_priv);
 err_workqueues:
 	i915_workqueues_cleanup(dev_priv);
 	return ret;

drivers/gpu/drm/i915/i915_drv.h

@@ -63,6 +63,7 @@
 #include "i915_gem_gtt.h"
 #include "i915_gem_render_state.h"
 #include "i915_gem_request.h"
+#include "i915_gem_timeline.h"
 
 #include "intel_gvt.h"
 
@@ -1830,7 +1831,6 @@ struct drm_i915_private {
 	struct i915_gem_context *kernel_context;
 	struct intel_engine_cs *engine[I915_NUM_ENGINES];
 	struct i915_vma *semaphore;
-	u32 next_seqno;
 
 	struct drm_dma_handle *status_page_dmah;
 	struct resource mch_res;
@@ -2090,6 +2090,9 @@ struct drm_i915_private {
 		void (*resume)(struct drm_i915_private *);
 		void (*cleanup_engine)(struct intel_engine_cs *engine);
 
+		struct list_head timelines;
+		struct i915_gem_timeline global_timeline;
+
 		/**
 		 * Is the GPU currently considered idle, or busy executing
 		 * userspace requests? Whilst idle, we allow runtime power
@@ -3175,7 +3178,7 @@ int i915_gem_get_aperture_ioctl(struct drm_device *dev, void *data,
 				struct drm_file *file_priv);
 int i915_gem_wait_ioctl(struct drm_device *dev, void *data,
 			struct drm_file *file_priv);
-void i915_gem_load_init(struct drm_device *dev);
+int i915_gem_load_init(struct drm_device *dev);
 void i915_gem_load_cleanup(struct drm_device *dev);
 void i915_gem_load_init_fences(struct drm_i915_private *dev_priv);
 int i915_gem_freeze(struct drm_i915_private *dev_priv);
@@ -3347,7 +3350,7 @@ void i915_gem_track_fb(struct drm_i915_gem_object *old,
 		       struct drm_i915_gem_object *new,
 		       unsigned frontbuffer_bits);
 
-int __must_check i915_gem_set_seqno(struct drm_device *dev, u32 seqno);
+int __must_check i915_gem_set_global_seqno(struct drm_device *dev, u32 seqno);
 
 struct drm_i915_gem_request *
 i915_gem_find_active_request(struct intel_engine_cs *engine);

drivers/gpu/drm/i915/i915_gem.c

@@ -371,7 +371,7 @@ i915_gem_object_wait_fence(struct dma_fence *fence,
 	if (flags & I915_WAIT_LOCKED && i915_gem_request_completed(rq))
 		i915_gem_request_retire_upto(rq);
 
-	if (rps && rq->fence.seqno == rq->engine->last_submitted_seqno) {
+	if (rps && rq->fence.seqno == rq->timeline->last_submitted_seqno) {
 		/* The GPU is now idle and this client has stalled.
 		 * Since no other client has submitted a request in the
 		 * meantime, assume that this client is the only one
@@ -2563,7 +2563,7 @@ i915_gem_find_active_request(struct intel_engine_cs *engine)
 	 * extra delay for a recent interrupt is pointless. Hence, we do
 	 * not need an engine->irq_seqno_barrier() before the seqno reads.
 	 */
-	list_for_each_entry(request, &engine->request_list, link) {
+	list_for_each_entry(request, &engine->timeline->requests, link) {
 		if (i915_gem_request_completed(request))
 			continue;
 
@@ -2632,7 +2632,7 @@ static void i915_gem_reset_engine(struct intel_engine_cs *engine)
 	if (i915_gem_context_is_default(incomplete_ctx))
 		return;
 
-	list_for_each_entry_continue(request, &engine->request_list, link)
+	list_for_each_entry_continue(request, &engine->timeline->requests, link)
 		if (request->ctx == incomplete_ctx)
 			reset_request(request);
 }
@@ -2671,7 +2671,8 @@ static void i915_gem_cleanup_engine(struct intel_engine_cs *engine)
 	 * (lockless) lookup doesn't try and wait upon the request as we
 	 * reset it.
 	 */
-	intel_engine_init_seqno(engine, engine->last_submitted_seqno);
+	intel_engine_init_global_seqno(engine,
+				       engine->timeline->last_submitted_seqno);
 
 	/*
 	 * Clear the execlists queue up before freeing the requests, as those
@@ -2979,18 +2980,26 @@ int i915_vma_unbind(struct i915_vma *vma)
 	return 0;
 }
 
-int i915_gem_wait_for_idle(struct drm_i915_private *dev_priv,
-			   unsigned int flags)
+static int wait_for_timeline(struct i915_gem_timeline *tl, unsigned int flags)
 {
-	struct intel_engine_cs *engine;
-	enum intel_engine_id id;
-	int ret;
+	int ret, i;
 
-	for_each_engine(engine, dev_priv, id) {
-		if (engine->last_context == NULL)
-			continue;
+	for (i = 0; i < ARRAY_SIZE(tl->engine); i++) {
+		ret = i915_gem_active_wait(&tl->engine[i].last_request, flags);
+		if (ret)
+			return ret;
+	}
 
-		ret = intel_engine_idle(engine, flags);
+	return 0;
+}
+
+int i915_gem_wait_for_idle(struct drm_i915_private *i915, unsigned int flags)
+{
+	struct i915_gem_timeline *tl;
+	int ret;
+
+	list_for_each_entry(tl, &i915->gt.timelines, link) {
+		ret = wait_for_timeline(tl, flags);
 		if (ret)
 			return ret;
 	}
@@ -4680,28 +4689,52 @@ i915_gem_load_init_fences(struct drm_i915_private *dev_priv)
 	i915_gem_detect_bit_6_swizzle(dev);
 }
 
-void
+int
 i915_gem_load_init(struct drm_device *dev)
 {
 	struct drm_i915_private *dev_priv = to_i915(dev);
+	int err;
 
 	dev_priv->objects =
 		kmem_cache_create("i915_gem_object",
 				  sizeof(struct drm_i915_gem_object), 0,
 				  SLAB_HWCACHE_ALIGN,
 				  NULL);
+	if (!dev_priv->objects) {
+		err = -ENOMEM;
+		goto err_out;
+	}
+
 	dev_priv->vmas =
 		kmem_cache_create("i915_gem_vma",
 				  sizeof(struct i915_vma), 0,
 				  SLAB_HWCACHE_ALIGN,
 				  NULL);
+	if (!dev_priv->vmas) {
+		err = -ENOMEM;
+		goto err_objects;
+	}
+
 	dev_priv->requests =
 		kmem_cache_create("i915_gem_request",
 				  sizeof(struct drm_i915_gem_request), 0,
 				  SLAB_HWCACHE_ALIGN |
 				  SLAB_RECLAIM_ACCOUNT |
 				  SLAB_DESTROY_BY_RCU,
 				  NULL);
+	if (!dev_priv->requests) {
+		err = -ENOMEM;
+		goto err_vmas;
+	}
+
+	mutex_lock(&dev_priv->drm.struct_mutex);
+	INIT_LIST_HEAD(&dev_priv->gt.timelines);
+	err = i915_gem_timeline_init(dev_priv,
+				     &dev_priv->gt.global_timeline,
+				     "[execution]");
+	mutex_unlock(&dev_priv->drm.struct_mutex);
+	if (err)
+		goto err_requests;
 
 	INIT_LIST_HEAD(&dev_priv->context_list);
 	INIT_WORK(&dev_priv->mm.free_work, __i915_gem_free_work);
@@ -4726,6 +4759,17 @@ i915_gem_load_init(struct drm_device *dev)
 	atomic_set(&dev_priv->mm.bsd_engine_dispatch_index, 0);
 
 	spin_lock_init(&dev_priv->fb_tracking.lock);
+
+	return 0;
+
+err_requests:
+	kmem_cache_destroy(dev_priv->requests);
+err_vmas:
+	kmem_cache_destroy(dev_priv->vmas);
+err_objects:
+	kmem_cache_destroy(dev_priv->objects);
+err_out:
+	return err;
 }
 
 void i915_gem_load_cleanup(struct drm_device *dev)

drivers/gpu/drm/i915/i915_gem.h

@@ -31,4 +31,6 @@
 #define GEM_BUG_ON(expr)
 #endif
 
+#define I915_NUM_ENGINES 5
+
 #endif /* __I915_GEM_H__ */