[PATCH 26/70] drm/i915: Map the execlists context regs once during pinning

[Chris Wilson <chris@chris-wilson.co.uk>]

When we pin the execlists context on queuing, it the ideal time to map
the register page that we need to update when we submit the request to
the hardware (and keep it around for future requests).

This avoids having to do an atomic kmap on every submission. On the
other hand, it does depend upon correct request construction.

v2: Rebase

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
---
 drivers/gpu/drm/i915/i915_gem.c         |  10 --
 drivers/gpu/drm/i915/intel_lrc.c        | 189 ++++++++++++--------------------
 drivers/gpu/drm/i915/intel_lrc.h        |   2 -
 drivers/gpu/drm/i915/intel_ringbuffer.h |   1 +
 4 files changed, 73 insertions(+), 129 deletions(-)

diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index f94fe2ba4f6f..071800553a43 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -2589,13 +2589,6 @@ void i915_gem_request_free(struct kref *req_ref)
 	struct intel_context *ctx = req->ctx;
 
 	if (ctx) {
-		if (i915.enable_execlists) {
-			struct intel_engine_cs *ring = req->ring;
-
-			if (ctx != ring->default_context)
-				intel_lr_context_unpin(ring, ctx);
-		}
-
 		i915_gem_context_unreference(ctx);
 	}
 
@@ -2699,9 +2692,6 @@ static void i915_gem_reset_ring_cleanup(struct drm_i915_private *dev_priv,
 				execlist_link);
 		list_del(&submit_req->execlist_link);
 
-		if (submit_req->ctx != ring->default_context)
-			intel_lr_context_unpin(ring, submit_req->ctx);
-
 		i915_gem_request_unreference(submit_req);
 	}
 
diff --git a/drivers/gpu/drm/i915/intel_lrc.c b/drivers/gpu/drm/i915/intel_lrc.c
index 8acfcf39e72d..4c985e186e3a 100644
--- a/drivers/gpu/drm/i915/intel_lrc.c
+++ b/drivers/gpu/drm/i915/intel_lrc.c
@@ -203,9 +203,6 @@ enum {
 };
 #define GEN8_CTX_ID_SHIFT 32
 
-static int intel_lr_context_pin(struct intel_engine_cs *ring,
-		struct intel_context *ctx);
-
 /**
  * intel_sanitize_enable_execlists() - sanitize i915.enable_execlists
  * @dev: DRM device.
@@ -318,47 +315,18 @@ static void execlists_elsp_write(struct intel_engine_cs *ring,
 	intel_uncore_forcewake_put(dev_priv, FORCEWAKE_ALL);
 }
 
-static int execlists_update_context(struct drm_i915_gem_object *ctx_obj,
-				    struct drm_i915_gem_object *ring_obj,
-				    u32 tail)
-{
-	struct page *page;
-	uint32_t *reg_state;
-
-	page = i915_gem_object_get_page(ctx_obj, 1);
-	reg_state = kmap_atomic(page);
-
-	reg_state[CTX_RING_TAIL+1] = tail;
-	reg_state[CTX_RING_BUFFER_START+1] = i915_gem_obj_ggtt_offset(ring_obj);
-
-	kunmap_atomic(reg_state);
-
-	return 0;
-}
-
 static void execlists_submit_contexts(struct intel_engine_cs *ring,
 				      struct intel_context *to0, u32 tail0,
 				      struct intel_context *to1, u32 tail1)
 {
 	struct drm_i915_gem_object *ctx_obj0 = to0->engine[ring->id].state;
-	struct intel_ringbuffer *ringbuf0 = to0->engine[ring->id].ringbuf;
 	struct drm_i915_gem_object *ctx_obj1 = NULL;
-	struct intel_ringbuffer *ringbuf1 = NULL;
-
-	BUG_ON(!ctx_obj0);
-	WARN_ON(!i915_gem_obj_is_pinned(ctx_obj0));
-	WARN_ON(!i915_gem_obj_is_pinned(ringbuf0->obj));
 
-	execlists_update_context(ctx_obj0, ringbuf0->obj, tail0);
+	to0->engine[ring->id].ringbuf->regs[CTX_RING_TAIL+1] = tail0;
 
 	if (to1) {
-		ringbuf1 = to1->engine[ring->id].ringbuf;
+		to1->engine[ring->id].ringbuf->regs[CTX_RING_TAIL+1] = tail1;
 		ctx_obj1 = to1->engine[ring->id].state;
-		BUG_ON(!ctx_obj1);
-		WARN_ON(!i915_gem_obj_is_pinned(ctx_obj1));
-		WARN_ON(!i915_gem_obj_is_pinned(ringbuf1->obj));
-
-		execlists_update_context(ctx_obj1, ringbuf1->obj, tail1);
 	}
 
 	execlists_elsp_write(ring, ctx_obj0, ctx_obj1);
@@ -500,29 +468,17 @@ static int execlists_context_queue(struct intel_engine_cs *ring,
 				   struct drm_i915_gem_request *request)
 {
 	struct drm_i915_gem_request *cursor;
-	struct drm_i915_private *dev_priv = ring->dev->dev_private;
 	int num_elements = 0;
 
-	if (to != ring->default_context)
-		intel_lr_context_pin(ring, to);
+	if (WARN_ON(request == NULL))
+		return -ENODEV;
+
+	if (WARN_ON(to->engine[ring->id].pin_count == 0))
+		return -ENODEV;
+
+	i915_gem_request_reference(request);
+	WARN_ON(to != request->ctx);
 
-	if (!request) {
-		/*
-		 * If there isn't a request associated with this submission,
-		 * create one as a temporary holder.
-		 */
-		request = kzalloc(sizeof(*request), GFP_KERNEL);
-		if (request == NULL)
-			return -ENOMEM;
-		request->ring = ring;
-		request->ctx = to;
-		kref_init(&request->ref);
-		request->uniq = dev_priv->request_uniq++;
-		i915_gem_context_reference(request->ctx);
-	} else {
-		i915_gem_request_reference(request);
-		WARN_ON(to != request->ctx);
-	}
 	request->tail = tail;
 
 	spin_lock_irq(&ring->execlist_lock);
@@ -608,16 +564,47 @@ static int execlists_move_to_gpu(struct intel_ringbuffer *ringbuf,
 	return logical_ring_invalidate_all_caches(ringbuf, ctx);
 }
 
+static int intel_lr_context_pin(struct intel_engine_cs *ring,
+				struct intel_context *ctx)
+{
+	struct drm_i915_gem_object *ctx_obj = ctx->engine[ring->id].state;
+	struct intel_ringbuffer *ringbuf = ctx->engine[ring->id].ringbuf;
+	int ret;
+
+	WARN_ON(!mutex_is_locked(&ring->dev->struct_mutex));
+	if (ctx->engine[ring->id].pin_count++)
+		return 0;
+
+	ret = i915_gem_obj_ggtt_pin(ctx_obj, GEN8_LR_CONTEXT_ALIGN, 0);
+	if (ret)
+		goto reset_pin_count;
+
+	ret = intel_pin_and_map_ringbuffer_obj(ring->dev, ringbuf);
+	if (ret)
+		goto unpin_ctx_obj;
+
+	ringbuf->regs = kmap(i915_gem_object_get_page(ctx_obj, 1));
+	ringbuf->regs[CTX_RING_BUFFER_START+1] =
+		i915_gem_obj_ggtt_offset(ringbuf->obj);
+
+	return 0;
+
+unpin_ctx_obj:
+	i915_gem_object_ggtt_unpin(ctx_obj);
+reset_pin_count:
+	ctx->engine[ring->id].pin_count = 0;
+
+	return ret;
+}
+
 int intel_logical_ring_alloc_request_extras(struct drm_i915_gem_request *request,
 					    struct intel_context *ctx)
 {
 	int ret;
 
-	if (ctx != request->ring->default_context) {
-		ret = intel_lr_context_pin(request->ring, ctx);
-		if (ret)
-			return ret;
-	}
+	ret = intel_lr_context_pin(request->ring, ctx);
+	if (ret)
+		return ret;
 
 	request->ringbuf = ctx->engine[request->ring->id].ringbuf;
 	request->ctx     = ctx;
@@ -929,30 +916,42 @@ int intel_execlists_submission(struct drm_device *dev, struct drm_file *file,
 	return 0;
 }
 
+static void intel_lr_context_unpin(struct intel_engine_cs *ring,
+				   struct intel_context *ctx)
+{
+	struct drm_i915_gem_object *ctx_obj = ctx->engine[ring->id].state;
+	struct intel_ringbuffer *ringbuf = ctx->engine[ring->id].ringbuf;
+
+	if (--ctx->engine[ring->id].pin_count)
+		return;
+
+	kunmap(i915_gem_object_get_page(ctx_obj, 1));
+	ringbuf->regs = NULL;
+
+	intel_unpin_ringbuffer_obj(ringbuf);
+	i915_gem_object_ggtt_unpin(ctx_obj);
+}
+
 void intel_execlists_retire_requests(struct intel_engine_cs *ring)
 {
-	struct drm_i915_gem_request *req, *tmp;
-	struct drm_i915_private *dev_priv = ring->dev->dev_private;
-	struct list_head retired_list;
+	struct list_head list;
 
 	WARN_ON(!mutex_is_locked(&ring->dev->struct_mutex));
 	if (list_empty(&ring->execlist_retired_req_list))
 		return;
 
-	INIT_LIST_HEAD(&retired_list);
 	spin_lock_irq(&ring->execlist_lock);
-	list_replace_init(&ring->execlist_retired_req_list, &retired_list);
+	list_replace_init(&ring->execlist_retired_req_list, &list);
 	spin_unlock_irq(&ring->execlist_lock);
 
-	list_for_each_entry_safe(req, tmp, &retired_list, execlist_link) {
-		struct intel_context *ctx = req->ctx;
-		struct drm_i915_gem_object *ctx_obj =
-				ctx->engine[ring->id].state;
+	while (!list_empty(&list)) {
+		struct drm_i915_gem_request *rq;
+
+		rq = list_first_entry(&list, typeof(*rq), execlist_link);
+		list_del(&rq->execlist_link);
 
-		if (ctx_obj && (ctx != ring->default_context))
-			intel_lr_context_unpin(ring, ctx);
-		list_del(&req->execlist_link);
-		i915_gem_request_unreference(req);
+		intel_lr_context_unpin(ring, rq->ctx);
+		i915_gem_request_unreference(rq);
 	}
 }
 
@@ -995,50 +994,6 @@ int logical_ring_flush_all_caches(struct intel_ringbuffer *ringbuf,
 	return 0;
 }
 
-static int intel_lr_context_pin(struct intel_engine_cs *ring,
-		struct intel_context *ctx)
-{
-	struct drm_i915_gem_object *ctx_obj = ctx->engine[ring->id].state;
-	struct intel_ringbuffer *ringbuf = ctx->engine[ring->id].ringbuf;
-	int ret = 0;
-
-	WARN_ON(!mutex_is_locked(&ring->dev->struct_mutex));
-	if (ctx->engine[ring->id].pin_count++ == 0) {
-		ret = i915_gem_obj_ggtt_pin(ctx_obj,
-				GEN8_LR_CONTEXT_ALIGN, 0);
-		if (ret)
-			goto reset_pin_count;
-
-		ret = intel_pin_and_map_ringbuffer_obj(ring->dev, ringbuf);
-		if (ret)
-			goto unpin_ctx_obj;
-	}
-
-	return ret;
-
-unpin_ctx_obj:
-	i915_gem_object_ggtt_unpin(ctx_obj);
-reset_pin_count:
-	ctx->engine[ring->id].pin_count = 0;
-
-	return ret;
-}
-
-void intel_lr_context_unpin(struct intel_engine_cs *ring,
-		struct intel_context *ctx)
-{
-	struct drm_i915_gem_object *ctx_obj = ctx->engine[ring->id].state;
-	struct intel_ringbuffer *ringbuf = ctx->engine[ring->id].ringbuf;
-
-	if (ctx_obj) {
-		WARN_ON(!mutex_is_locked(&ring->dev->struct_mutex));
-		if (--ctx->engine[ring->id].pin_count == 0) {
-			intel_unpin_ringbuffer_obj(ringbuf);
-			i915_gem_object_ggtt_unpin(ctx_obj);
-		}
-	}
-}
-
 static int intel_logical_ring_workarounds_emit(struct intel_engine_cs *ring,
 					       struct intel_context *ctx)
 {
@@ -1967,7 +1922,7 @@ error_unpin_ctx:
 }
 
 void intel_lr_context_reset(struct drm_device *dev,
-			struct intel_context *ctx)
+			    struct intel_context *ctx)
 {
 	struct drm_i915_private *dev_priv = dev->dev_private;
 	struct intel_engine_cs *ring;
diff --git a/drivers/gpu/drm/i915/intel_lrc.h b/drivers/gpu/drm/i915/intel_lrc.h
index 04d3a6d8b207..b6fd4c2e8b6e 100644
--- a/drivers/gpu/drm/i915/intel_lrc.h
+++ b/drivers/gpu/drm/i915/intel_lrc.h
@@ -70,8 +70,6 @@ static inline void intel_logical_ring_emit(struct intel_ringbuffer *ringbuf,
 void intel_lr_context_free(struct intel_context *ctx);
 int intel_lr_context_deferred_create(struct intel_context *ctx,
 				     struct intel_engine_cs *ring);
-void intel_lr_context_unpin(struct intel_engine_cs *ring,
-		struct intel_context *ctx);
 void intel_lr_context_reset(struct drm_device *dev,
 			struct intel_context *ctx);
 
diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.h b/drivers/gpu/drm/i915/intel_ringbuffer.h
index 39f6dfc0ee54..0f0325e88b5a 100644
--- a/drivers/gpu/drm/i915/intel_ringbuffer.h
+++ b/drivers/gpu/drm/i915/intel_ringbuffer.h
@@ -97,6 +97,7 @@ struct intel_ring_hangcheck {
 struct intel_ringbuffer {
 	struct drm_i915_gem_object *obj;
 	void __iomem *virtual_start;
+	uint32_t *regs;
 
 	struct intel_engine_cs *ring;
 
-- 
2.1.4

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/intel-gfx