1 files changed, 437 insertions, 202 deletions

diff --git a/drivers/gpu/drm/i915/i915_gem_context.c b/drivers/gpu/drm/i915/i915_gem_context.c
index 403309c2a7d..a5ddf3bce9c 100644
--- a/drivers/gpu/drm/i915/i915_gem_context.c
+++ b/drivers/gpu/drm/i915/i915_gem_context.c
@@ -73,7 +73,7 @@
  *
  * There are two confusing terms used above:
  *  The "current context" means the context which is currently running on the
- *  GPU. The GPU has loaded it's state already and has stored away the gtt
+ *  GPU. The GPU has loaded its state already and has stored away the gtt
  *  offset of the BO. The GPU is not actively referencing the data at this
  *  offset, but it will on the next context switch. The only way to avoid this
  *  is to do a GPU reset.
@@ -93,11 +93,60 @@
  * I've seen in a spec to date, and that was a workaround for a non-shipping
  * part. It should be safe to decrease this, but it's more future proof as is.
  */
-#define CONTEXT_ALIGN (64<<10)
+#define GEN6_CONTEXT_ALIGN (64<<10)
+#define GEN7_CONTEXT_ALIGN 4096
 
-static struct i915_hw_context *
-i915_gem_context_get(struct drm_i915_file_private *file_priv, u32 id);
-static int do_switch(struct i915_hw_context *to);
+static void do_ppgtt_cleanup(struct i915_hw_ppgtt *ppgtt)
+{
+	struct drm_device *dev = ppgtt->base.dev;
+	struct drm_i915_private *dev_priv = dev->dev_private;
+	struct i915_address_space *vm = &ppgtt->base;
+
+	if (ppgtt == dev_priv->mm.aliasing_ppgtt ||
+	    (list_empty(&vm->active_list) && list_empty(&vm->inactive_list))) {
+		ppgtt->base.cleanup(&ppgtt->base);
+		return;
+	}
+
+	/*
+	 * Make sure vmas are unbound before we take down the drm_mm
+	 *
+	 * FIXME: Proper refcounting should take care of this, this shouldn't be
+	 * needed at all.
+	 */
+	if (!list_empty(&vm->active_list)) {
+		struct i915_vma *vma;
+
+		list_for_each_entry(vma, &vm->active_list, mm_list)
+			if (WARN_ON(list_empty(&vma->vma_link) ||
+				    list_is_singular(&vma->vma_link)))
+				break;
+
+		i915_gem_evict_vm(&ppgtt->base, true);
+	} else {
+		i915_gem_retire_requests(dev);
+		i915_gem_evict_vm(&ppgtt->base, false);
+	}
+
+	ppgtt->base.cleanup(&ppgtt->base);
+}
+
+static void ppgtt_release(struct kref *kref)
+{
+	struct i915_hw_ppgtt *ppgtt =
+		container_of(kref, struct i915_hw_ppgtt, ref);
+
+	do_ppgtt_cleanup(ppgtt);
+	kfree(ppgtt);
+}
+
+static size_t get_context_alignment(struct drm_device *dev)
+{
+	if (IS_GEN6(dev))
+		return GEN6_CONTEXT_ALIGN;
+
+	return GEN7_CONTEXT_ALIGN;
+}
 
 static int get_context_size(struct drm_device *dev)
 {
@@ -117,6 +166,9 @@ static int get_context_size(struct drm_device *dev)
 		else
 			ret = GEN7_CXT_TOTAL_SIZE(reg) * 64;
 		break;
+	case 8:
+		ret = GEN8_CXT_TOTAL_SIZE;
+		break;
 	default:
 		BUG();
 	}
@@ -126,19 +178,52 @@ static int get_context_size(struct drm_device *dev)
 
 void i915_gem_context_free(struct kref *ctx_ref)
 {
-	struct i915_hw_context *ctx = container_of(ctx_ref,
+	struct intel_context *ctx = container_of(ctx_ref,
 						   typeof(*ctx), ref);
+	struct i915_hw_ppgtt *ppgtt = NULL;
 
-	drm_gem_object_unreference(&ctx->obj->base);
+	if (ctx->obj) {
+		/* We refcount even the aliasing PPGTT to keep the code symmetric */
+		if (USES_PPGTT(ctx->obj->base.dev))
+			ppgtt = ctx_to_ppgtt(ctx);
+
+		/* XXX: Free up the object before tearing down the address space, in
+		 * case we're bound in the PPGTT */
+		drm_gem_object_unreference(&ctx->obj->base);
+	}
+
+	if (ppgtt)
+		kref_put(&ppgtt->ref, ppgtt_release);
+	list_del(&ctx->link);
 	kfree(ctx);
 }
 
-static struct i915_hw_context *
-create_hw_context(struct drm_device *dev,
+static struct i915_hw_ppgtt *
+create_vm_for_ctx(struct drm_device *dev, struct intel_context *ctx)
+{
+	struct i915_hw_ppgtt *ppgtt;
+	int ret;
+
+	ppgtt = kzalloc(sizeof(*ppgtt), GFP_KERNEL);
+	if (!ppgtt)
+		return ERR_PTR(-ENOMEM);
+
+	ret = i915_gem_init_ppgtt(dev, ppgtt);
+	if (ret) {
+		kfree(ppgtt);
+		return ERR_PTR(ret);
+	}
+
+	ppgtt->ctx = ctx;
+	return ppgtt;
+}
+
+static struct intel_context *
+__create_hw_context(struct drm_device *dev,
 		  struct drm_i915_file_private *file_priv)
 {
 	struct drm_i915_private *dev_priv = dev->dev_private;
-	struct i915_hw_context *ctx;
+	struct intel_context *ctx;
 	int ret;
 
 	ctx = kzalloc(sizeof(*ctx), GFP_KERNEL);
@@ -146,38 +231,47 @@ create_hw_context(struct drm_device *dev,
 		return ERR_PTR(-ENOMEM);
 
 	kref_init(&ctx->ref);
-	ctx->obj = i915_gem_alloc_object(dev, dev_priv->hw_context_size);
-	if (ctx->obj == NULL) {
-		kfree(ctx);
-		DRM_DEBUG_DRIVER("Context object allocated failed\n");
-		return ERR_PTR(-ENOMEM);
-	}
+	list_add_tail(&ctx->link, &dev_priv->context_list);
 
-	if (INTEL_INFO(dev)->gen >= 7) {
-		ret = i915_gem_object_set_cache_level(ctx->obj,
-						      I915_CACHE_L3_LLC);
-		/* Failure shouldn't ever happen this early */
-		if (WARN_ON(ret))
+	if (dev_priv->hw_context_size) {
+		ctx->obj = i915_gem_alloc_object(dev, dev_priv->hw_context_size);
+		if (ctx->obj == NULL) {
+			ret = -ENOMEM;
 			goto err_out;
-	}
+		}
 
-	/* The ring associated with the context object is handled by the normal
-	 * object tracking code. We give an initial ring value simple to pass an
-	 * assertion in the context switch code.
-	 */
-	ctx->ring = &dev_priv->ring[RCS];
+		/*
+		 * Try to make the context utilize L3 as well as LLC.
+		 *
+		 * On VLV we don't have L3 controls in the PTEs so we
+		 * shouldn't touch the cache level, especially as that
+		 * would make the object snooped which might have a
+		 * negative performance impact.
+		 */
+		if (INTEL_INFO(dev)->gen >= 7 && !IS_VALLEYVIEW(dev)) {
+			ret = i915_gem_object_set_cache_level(ctx->obj,
+							      I915_CACHE_L3_LLC);
+			/* Failure shouldn't ever happen this early */
+			if (WARN_ON(ret))
+				goto err_out;
+		}
+	}
 
 	/* Default context will never have a file_priv */
-	if (file_priv == NULL)
-		return ctx;
-
-	ret = idr_alloc(&file_priv->context_idr, ctx, DEFAULT_CONTEXT_ID + 1, 0,
-			GFP_KERNEL);
-	if (ret < 0)
-		goto err_out;
+	if (file_priv != NULL) {
+		ret = idr_alloc(&file_priv->context_idr, ctx,
+				DEFAULT_CONTEXT_ID, 0, GFP_KERNEL);
+		if (ret < 0)
+			goto err_out;
+	} else
+		ret = DEFAULT_CONTEXT_ID;
 
 	ctx->file_priv = file_priv;
 	ctx->id = ret;
+	/* NB: Mark all slices as needing a remap so that when the context first
+	 * loads it will restore whatever remap state already exists. If there
+	 * is no remap info, it will be a NOP. */
+	ctx->remap_slice = (1 << NUM_L3_SLICES(dev)) - 1;
 
 	return ctx;
 
@@ -186,163 +280,269 @@ err_out:
 	return ERR_PTR(ret);
 }
 
-static inline bool is_default_context(struct i915_hw_context *ctx)
-{
-	return (ctx == ctx->ring->default_context);
-}
-
 /**
  * The default context needs to exist per ring that uses contexts. It stores the
  * context state of the GPU for applications that don't utilize HW contexts, as
  * well as an idle case.
  */
-static int create_default_context(struct drm_i915_private *dev_priv)
+static struct intel_context *
+i915_gem_create_context(struct drm_device *dev,
+			struct drm_i915_file_private *file_priv,
+			bool create_vm)
 {
-	struct i915_hw_context *ctx;
-	int ret;
+	const bool is_global_default_ctx = file_priv == NULL;
+	struct drm_i915_private *dev_priv = dev->dev_private;
+	struct intel_context *ctx;
+	int ret = 0;
 
-	BUG_ON(!mutex_is_locked(&dev_priv->dev->struct_mutex));
+	BUG_ON(!mutex_is_locked(&dev->struct_mutex));
 
-	ctx = create_hw_context(dev_priv->dev, NULL);
+	ctx = __create_hw_context(dev, file_priv);
 	if (IS_ERR(ctx))
-		return PTR_ERR(ctx);
+		return ctx;
 
-	/* We may need to do things with the shrinker which require us to
-	 * immediately switch back to the default context. This can cause a
-	 * problem as pinning the default context also requires GTT space which
-	 * may not be available. To avoid this we always pin the
-	 * default context.
-	 */
-	dev_priv->ring[RCS].default_context = ctx;
-	ret = i915_gem_obj_ggtt_pin(ctx->obj, CONTEXT_ALIGN, false, false);
-	if (ret) {
-		DRM_DEBUG_DRIVER("Couldn't pin %d\n", ret);
-		goto err_destroy;
+	if (is_global_default_ctx && ctx->obj) {
+		/* We may need to do things with the shrinker which
+		 * require us to immediately switch back to the default
+		 * context. This can cause a problem as pinning the
+		 * default context also requires GTT space which may not
+		 * be available. To avoid this we always pin the default
+		 * context.
+		 */
+		ret = i915_gem_obj_ggtt_pin(ctx->obj,
+					    get_context_alignment(dev), 0);
+		if (ret) {
+			DRM_DEBUG_DRIVER("Couldn't pin %d\n", ret);
+			goto err_destroy;
+		}
 	}
 
-	ret = do_switch(ctx);
-	if (ret) {
-		DRM_DEBUG_DRIVER("Switch failed %d\n", ret);
-		goto err_unpin;
-	}
+	if (create_vm) {
+		struct i915_hw_ppgtt *ppgtt = create_vm_for_ctx(dev, ctx);
+
+		if (IS_ERR_OR_NULL(ppgtt)) {
+			DRM_DEBUG_DRIVER("PPGTT setup failed (%ld)\n",
+					 PTR_ERR(ppgtt));
+			ret = PTR_ERR(ppgtt);
+			goto err_unpin;
+		} else
+			ctx->vm = &ppgtt->base;
+
+		/* This case is reserved for the global default context and
+		 * should only happen once. */
+		if (is_global_default_ctx) {
+			if (WARN_ON(dev_priv->mm.aliasing_ppgtt)) {
+				ret = -EEXIST;
+				goto err_unpin;
+			}
+
+			dev_priv->mm.aliasing_ppgtt = ppgtt;
+		}
+	} else if (USES_PPGTT(dev)) {
+		/* For platforms which only have aliasing PPGTT, we fake the
+		 * address space and refcounting. */
+		ctx->vm = &dev_priv->mm.aliasing_ppgtt->base;
+		kref_get(&dev_priv->mm.aliasing_ppgtt->ref);
+	} else
+		ctx->vm = &dev_priv->gtt.base;
 
-	DRM_DEBUG_DRIVER("Default HW context loaded\n");
-	return 0;
+	return ctx;
 
 err_unpin:
-	i915_gem_object_unpin(ctx->obj);
+	if (is_global_default_ctx && ctx->obj)
+		i915_gem_object_ggtt_unpin(ctx->obj);
 err_destroy:
 	i915_gem_context_unreference(ctx);
-	return ret;
+	return ERR_PTR(ret);
 }
 
-void i915_gem_context_init(struct drm_device *dev)
+void i915_gem_context_reset(struct drm_device *dev)
 {
 	struct drm_i915_private *dev_priv = dev->dev_private;
+	int i;
+
+	/* Prevent the hardware from restoring the last context (which hung) on
+	 * the next switch */
+	for (i = 0; i < I915_NUM_RINGS; i++) {
+		struct intel_engine_cs *ring = &dev_priv->ring[i];
+		struct intel_context *dctx = ring->default_context;
+
+		/* Do a fake switch to the default context */
+		if (ring->last_context == dctx)
+			continue;
+
+		if (!ring->last_context)
+			continue;
+
+		if (dctx->obj && i == RCS) {
+			WARN_ON(i915_gem_obj_ggtt_pin(dctx->obj,
+						      get_context_alignment(dev), 0));
+			/* Fake a finish/inactive */
+			dctx->obj->base.write_domain = 0;
+			dctx->obj->active = 0;
+		}
 
-	if (!HAS_HW_CONTEXTS(dev)) {
-		dev_priv->hw_contexts_disabled = true;
-		DRM_DEBUG_DRIVER("Disabling HW Contexts; old hardware\n");
-		return;
+		i915_gem_context_unreference(ring->last_context);
+		i915_gem_context_reference(dctx);
+		ring->last_context = dctx;
 	}
+}
 
-	/* If called from reset, or thaw... we've been here already */
-	if (dev_priv->hw_contexts_disabled ||
-	    dev_priv->ring[RCS].default_context)
-		return;
+int i915_gem_context_init(struct drm_device *dev)
+{
+	struct drm_i915_private *dev_priv = dev->dev_private;
+	struct intel_context *ctx;
+	int i;
 
-	dev_priv->hw_context_size = round_up(get_context_size(dev), 4096);
+	/* Init should only be called once per module load. Eventually the
+	 * restriction on the context_disabled check can be loosened. */
+	if (WARN_ON(dev_priv->ring[RCS].default_context))
+		return 0;
 
-	if (dev_priv->hw_context_size > (1<<20)) {
-		dev_priv->hw_contexts_disabled = true;
-		DRM_DEBUG_DRIVER("Disabling HW Contexts; invalid size\n");
-		return;
+	if (HAS_HW_CONTEXTS(dev)) {
+		dev_priv->hw_context_size = round_up(get_context_size(dev), 4096);
+		if (dev_priv->hw_context_size > (1<<20)) {
+			DRM_DEBUG_DRIVER("Disabling HW Contexts; invalid size %d\n",
+					 dev_priv->hw_context_size);
+			dev_priv->hw_context_size = 0;
+		}
 	}
 
-	if (create_default_context(dev_priv)) {
-		dev_priv->hw_contexts_disabled = true;
-		DRM_DEBUG_DRIVER("Disabling HW Contexts; create failed\n");
-		return;
+	ctx = i915_gem_create_context(dev, NULL, USES_PPGTT(dev));
+	if (IS_ERR(ctx)) {
+		DRM_ERROR("Failed to create default global context (error %ld)\n",
+			  PTR_ERR(ctx));
+		return PTR_ERR(ctx);
 	}
 
-	DRM_DEBUG_DRIVER("HW context support initialized\n");
+	/* NB: RCS will hold a ref for all rings */
+	for (i = 0; i < I915_NUM_RINGS; i++)
+		dev_priv->ring[i].default_context = ctx;
+
+	DRM_DEBUG_DRIVER("%s context support initialized\n", dev_priv->hw_context_size ? "HW" : "fake");
+	return 0;
 }
 
 void i915_gem_context_fini(struct drm_device *dev)
 {
 	struct drm_i915_private *dev_priv = dev->dev_private;
-	struct i915_hw_context *dctx = dev_priv->ring[RCS].default_context;
+	struct intel_context *dctx = dev_priv->ring[RCS].default_context;
+	int i;
+
+	if (dctx->obj) {
+		/* The only known way to stop the gpu from accessing the hw context is
+		 * to reset it. Do this as the very last operation to avoid confusing
+		 * other code, leading to spurious errors. */
+		intel_gpu_reset(dev);
+
+		/* When default context is created and switched to, base object refcount
+		 * will be 2 (+1 from object creation and +1 from do_switch()).
+		 * i915_gem_context_fini() will be called after gpu_idle() has switched
+		 * to default context. So we need to unreference the base object once
+		 * to offset the do_switch part, so that i915_gem_context_unreference()
+		 * can then free the base object correctly. */
+		WARN_ON(!dev_priv->ring[RCS].last_context);
+		if (dev_priv->ring[RCS].last_context == dctx) {
+			/* Fake switch to NULL context */
+			WARN_ON(dctx->obj->active);
+			i915_gem_object_ggtt_unpin(dctx->obj);
+			i915_gem_context_unreference(dctx);
+			dev_priv->ring[RCS].last_context = NULL;
+		}
 
-	if (dev_priv->hw_contexts_disabled)
-		return;
+		i915_gem_object_ggtt_unpin(dctx->obj);
+	}
+
+	for (i = 0; i < I915_NUM_RINGS; i++) {
+		struct intel_engine_cs *ring = &dev_priv->ring[i];
 
-	/* The only known way to stop the gpu from accessing the hw context is
-	 * to reset it. Do this as the very last operation to avoid confusing
-	 * other code, leading to spurious errors. */
-	intel_gpu_reset(dev);
+		if (ring->last_context)
+			i915_gem_context_unreference(ring->last_context);
 
-	i915_gem_object_unpin(dctx->obj);
+		ring->default_context = NULL;
+		ring->last_context = NULL;
+	}
 
-	/* When default context is created and switched to, base object refcount
-	 * will be 2 (+1 from object creation and +1 from do_switch()).
-	 * i915_gem_context_fini() will be called after gpu_idle() has switched
-	 * to default context. So we need to unreference the base object once
-	 * to offset the do_switch part, so that i915_gem_context_unreference()
-	 * can then free the base object correctly. */
-	drm_gem_object_unreference(&dctx->obj->base);
 	i915_gem_context_unreference(dctx);
 }
 
-static int context_idr_cleanup(int id, void *p, void *data)
+int i915_gem_context_enable(struct drm_i915_private *dev_priv)
 {
-	struct i915_hw_context *ctx = p;
+	struct intel_engine_cs *ring;
+	int ret, i;
+
+	/* This is the only place the aliasing PPGTT gets enabled, which means
+	 * it has to happen before we bail on reset */
+	if (dev_priv->mm.aliasing_ppgtt) {
+		struct i915_hw_ppgtt *ppgtt = dev_priv->mm.aliasing_ppgtt;
+		ppgtt->enable(ppgtt);
+	}
 
-	BUG_ON(id == DEFAULT_CONTEXT_ID);
+	/* FIXME: We should make this work, even in reset */
+	if (i915_reset_in_progress(&dev_priv->gpu_error))
+		return 0;
+
+	BUG_ON(!dev_priv->ring[RCS].default_context);
+
+	for_each_ring(ring, dev_priv, i) {
+		ret = i915_switch_context(ring, ring->default_context);
+		if (ret)
+			return ret;
+	}
+
+	return 0;
+}
+
+static int context_idr_cleanup(int id, void *p, void *data)
+{
+	struct intel_context *ctx = p;
 
 	i915_gem_context_unreference(ctx);
 	return 0;
 }
 
-struct i915_ctx_hang_stats *
-i915_gem_context_get_hang_stats(struct drm_device *dev,
-				struct drm_file *file,
-				u32 id)
+int i915_gem_context_open(struct drm_device *dev, struct drm_file *file)
 {
-	struct drm_i915_private *dev_priv = dev->dev_private;
 	struct drm_i915_file_private *file_priv = file->driver_priv;
-	struct i915_hw_context *ctx;
+	struct intel_context *ctx;
 
-	if (id == DEFAULT_CONTEXT_ID)
-		return &file_priv->hang_stats;
+	idr_init(&file_priv->context_idr);
 
-	ctx = NULL;
-	if (!dev_priv->hw_contexts_disabled)
-		ctx = i915_gem_context_get(file->driver_priv, id);
-	if (ctx == NULL)
-		return ERR_PTR(-ENOENT);
+	mutex_lock(&dev->struct_mutex);
+	ctx = i915_gem_create_context(dev, file_priv, USES_FULL_PPGTT(dev));
+	mutex_unlock(&dev->struct_mutex);
 
-	return &ctx->hang_stats;
+	if (IS_ERR(ctx)) {
+		idr_destroy(&file_priv->context_idr);
+		return PTR_ERR(ctx);
+	}
+
+	return 0;
 }
 
 void i915_gem_context_close(struct drm_device *dev, struct drm_file *file)
 {
 	struct drm_i915_file_private *file_priv = file->driver_priv;
 
-	mutex_lock(&dev->struct_mutex);
 	idr_for_each(&file_priv->context_idr, context_idr_cleanup, NULL);
 	idr_destroy(&file_priv->context_idr);
-	mutex_unlock(&dev->struct_mutex);
 }
 
-static struct i915_hw_context *
+struct intel_context *
 i915_gem_context_get(struct drm_i915_file_private *file_priv, u32 id)
 {
-	return (struct i915_hw_context *)idr_find(&file_priv->context_idr, id);
+	struct intel_context *ctx;
+
+	ctx = (struct intel_context *)idr_find(&file_priv->context_idr, id);
+	if (!ctx)
+		return ERR_PTR(-ENOENT);
+
+	return ctx;
 }
 
 static inline int
-mi_set_context(struct intel_ring_buffer *ring,
-	       struct i915_hw_context *new_context,
+mi_set_context(struct intel_engine_cs *ring,
+	       struct intel_context *new_context,
 	       u32 hw_flags)
 {
 	int ret;
@@ -352,7 +552,7 @@ mi_set_context(struct intel_ring_buffer *ring,
 	 * explicitly, so we rely on the value at ring init, stored in
 	 * itlb_before_ctx_switch.
 	 */
-	if (IS_GEN6(ring->dev) && ring->itlb_before_ctx_switch) {
+	if (IS_GEN6(ring->dev)) {
 		ret = ring->flush(ring, I915_GEM_GPU_DOMAINS, 0);
 		if (ret)
 			return ret;
@@ -362,8 +562,8 @@ mi_set_context(struct intel_ring_buffer *ring,
 	if (ret)
 		return ret;
 
-	/* WaProgramMiArbOnOffAroundMiSetContext:ivb,vlv,hsw */
-	if (IS_GEN7(ring->dev))
+	/* WaProgramMiArbOnOffAroundMiSetContext:ivb,vlv,hsw,bdw,chv */
+	if (INTEL_INFO(ring->dev)->gen >= 7)
 		intel_ring_emit(ring, MI_ARB_ON_OFF | MI_ARB_DISABLE);
 	else
 		intel_ring_emit(ring, MI_NOOP);
@@ -375,10 +575,13 @@ mi_set_context(struct intel_ring_buffer *ring,
 			MI_SAVE_EXT_STATE_EN |
 			MI_RESTORE_EXT_STATE_EN |
 			hw_flags);
-	/* w/a: MI_SET_CONTEXT must always be followed by MI_NOOP */
+	/*
+	 * w/a: MI_SET_CONTEXT must always be followed by MI_NOOP
+	 * WaMiSetContext_Hang:snb,ivb,vlv
+	 */
 	intel_ring_emit(ring, MI_NOOP);
 
-	if (IS_GEN7(ring->dev))
+	if (INTEL_INFO(ring->dev)->gen >= 7)
 		intel_ring_emit(ring, MI_ARB_ON_OFF | MI_ARB_ENABLE);
 	else
 		intel_ring_emit(ring, MI_NOOP);
@@ -388,45 +591,86 @@ mi_set_context(struct intel_ring_buffer *ring,
 	return ret;
 }
 
-static int do_switch(struct i915_hw_context *to)
+static int do_switch(struct intel_engine_cs *ring,
+		     struct intel_context *to)
 {
-	struct intel_ring_buffer *ring = to->ring;
-	struct i915_hw_context *from = ring->last_context;
+	struct drm_i915_private *dev_priv = ring->dev->dev_private;
+	struct intel_context *from = ring->last_context;
+	struct i915_hw_ppgtt *ppgtt = ctx_to_ppgtt(to);
 	u32 hw_flags = 0;
-	int ret;
+	bool uninitialized = false;
+	int ret, i;
 
-	BUG_ON(from != NULL && from->obj != NULL && from->obj->pin_count == 0);
+	if (from != NULL && ring == &dev_priv->ring[RCS]) {
+		BUG_ON(from->obj == NULL);
+		BUG_ON(!i915_gem_obj_is_pinned(from->obj));
+	}
 
-	if (from == to)
+	if (from == to && from->last_ring == ring && !to->remap_slice)
 		return 0;
 
-	ret = i915_gem_obj_ggtt_pin(to->obj, CONTEXT_ALIGN, false, false);
-	if (ret)
-		return ret;
+	/* Trying to pin first makes error handling easier. */
+	if (ring == &dev_priv->ring[RCS]) {
+		ret = i915_gem_obj_ggtt_pin(to->obj,
+					    get_context_alignment(ring->dev), 0);
+		if (ret)
+			return ret;
+	}
+
+	/*
+	 * Pin can switch back to the default context if we end up calling into
+	 * evict_everything - as a last ditch gtt defrag effort that also
+	 * switches to the default context. Hence we need to reload from here.
+	 */
+	from = ring->last_context;
+
+	if (USES_FULL_PPGTT(ring->dev)) {
+		ret = ppgtt->switch_mm(ppgtt, ring, false);
+		if (ret)
+			goto unpin_out;
+	}
+
+	if (ring != &dev_priv->ring[RCS]) {
+		if (from)
+			i915_gem_context_unreference(from);
+		goto done;
+	}
 
-	/* Clear this page out of any CPU caches for coherent swap-in/out. Note
+	/*
+	 * Clear this page out of any CPU caches for coherent swap-in/out. Note
 	 * that thanks to write = false in this call and us not setting any gpu
 	 * write domains when putting a context object onto the active list
 	 * (when switching away from it), this won't block.
-	 * XXX: We need a real interface to do this instead of trickery. */
+	 *
+	 * XXX: We need a real interface to do this instead of trickery.
+	 */
 	ret = i915_gem_object_set_to_gtt_domain(to->obj, false);
-	if (ret) {
-		i915_gem_object_unpin(to->obj);
-		return ret;
-	}
+	if (ret)
+		goto unpin_out;
 
-	if (!to->obj->has_global_gtt_mapping)
-		i915_gem_gtt_bind_object(to->obj, to->obj->cache_level);
+	if (!to->obj->has_global_gtt_mapping) {
+		struct i915_vma *vma = i915_gem_obj_to_vma(to->obj,
+							   &dev_priv->gtt.base);
+		vma->bind_vma(vma, to->obj->cache_level, GLOBAL_BIND);
+	}
 
-	if (!to->is_initialized || is_default_context(to))
+	if (!to->is_initialized || i915_gem_context_is_default(to))
 		hw_flags |= MI_RESTORE_INHIBIT;
-	else if (WARN_ON_ONCE(from == to)) /* not yet expected */
-		hw_flags |= MI_FORCE_RESTORE;
 
 	ret = mi_set_context(ring, to, hw_flags);
-	if (ret) {
-		i915_gem_object_unpin(to->obj);
-		return ret;
+	if (ret)
+		goto unpin_out;
+
+	for (i = 0; i < MAX_L3_SLICES; i++) {
+		if (!(to->remap_slice & (1<<i)))
+			continue;
+
+		ret = i915_gem_l3_remap(ring, i);
+		/* If it failed, try again next round */
+		if (ret)
+			DRM_DEBUG_DRIVER("L3 remapping failed\n");
+		else
+			to->remap_slice &= ~(1<<i);
 	}
 
 	/* The backing object for the context is done after switching to the
@@ -436,11 +680,8 @@ static int do_switch(struct i915_hw_context *to)
 	 * MI_SET_CONTEXT instead of when the next seqno has completed.
 	 */
 	if (from != NULL) {
-		struct drm_i915_private *dev_priv = from->obj->base.dev->dev_private;
-		struct i915_address_space *ggtt = &dev_priv->gtt.base;
 		from->obj->base.read_domains = I915_GEM_DOMAIN_INSTRUCTION;
-		list_move_tail(&i915_gem_obj_to_vma(from->obj, ggtt)->mm_list, &ggtt->active_list);
-		i915_gem_object_move_to_active(from->obj, ring);
+		i915_vma_move_to_active(i915_gem_obj_to_ggtt(from->obj), ring);
 		/* As long as MI_SET_CONTEXT is serializing, ie. it flushes the
 		 * whole damn pipeline, we don't need to explicitly mark the
 		 * object dirty. The only exception is that the context must be
@@ -451,90 +692,84 @@ static int do_switch(struct i915_hw_context *to)
 		from->obj->dirty = 1;
 		BUG_ON(from->obj->ring != ring);
 
-		ret = i915_add_request(ring, NULL);
-		if (ret) {
-			/* Too late, we've already scheduled a context switch.
-			 * Try to undo the change so that the hw state is
-			 * consistent with out tracking. In case of emergency,
-			 * scream.
-			 */
-			WARN_ON(mi_set_context(ring, from, MI_RESTORE_INHIBIT));
-			return ret;
-		}
-
-		i915_gem_object_unpin(from->obj);
+		/* obj is kept alive until the next request by its active ref */
+		i915_gem_object_ggtt_unpin(from->obj);
 		i915_gem_context_unreference(from);
 	}
 
+	uninitialized = !to->is_initialized && from == NULL;
+	to->is_initialized = true;
+
+done:
 	i915_gem_context_reference(to);
 	ring->last_context = to;
-	to->is_initialized = true;
+	to->last_ring = ring;
+
+	if (uninitialized) {
+		ret = i915_gem_render_state_init(ring);
+		if (ret)
+			DRM_ERROR("init render state: %d\n", ret);
+	}
 
 	return 0;
+
+unpin_out:
+	if (ring->id == RCS)
+		i915_gem_object_ggtt_unpin(to->obj);
+	return ret;
 }
 
 /**
  * i915_switch_context() - perform a GPU context switch.
  * @ring: ring for which we'll execute the context switch
- * @file_priv: file_priv associated with the context, may be NULL
- * @id: context id number
- * @seqno: sequence number by which the new context will be switched to
- * @flags:
+ * @to: the context to switch to
  *
  * The context life cycle is simple. The context refcount is incremented and
  * decremented by 1 and create and destroy. If the context is in use by the GPU,
  * it will have a refoucnt > 1. This allows us to destroy the context abstract
  * object while letting the normal object tracking destroy the backing BO.
  */
-int i915_switch_context(struct intel_ring_buffer *ring,
-			struct drm_file *file,
-			int to_id)
+int i915_switch_context(struct intel_engine_cs *ring,
+			struct intel_context *to)
 {
 	struct drm_i915_private *dev_priv = ring->dev->dev_private;
-	struct i915_hw_context *to;
-
-	if (dev_priv->hw_contexts_disabled)
-		return 0;
 
 	WARN_ON(!mutex_is_locked(&dev_priv->dev->struct_mutex));
 
-	if (ring != &dev_priv->ring[RCS])
+	if (to->obj == NULL) { /* We have the fake context */
+		if (to != ring->last_context) {
+			i915_gem_context_reference(to);
+			if (ring->last_context)
+				i915_gem_context_unreference(ring->last_context);
+			ring->last_context = to;
+		}
 		return 0;
-
-	if (to_id == DEFAULT_CONTEXT_ID) {
-		to = ring->default_context;
-	} else {
-		if (file == NULL)
-			return -EINVAL;
-
-		to = i915_gem_context_get(file->driver_priv, to_id);
-		if (to == NULL)
-			return -ENOENT;
 	}
 
-	return do_switch(to);
+	return do_switch(ring, to);
+}
+
+static bool hw_context_enabled(struct drm_device *dev)
+{
+	return to_i915(dev)->hw_context_size;
 }
 
 int i915_gem_context_create_ioctl(struct drm_device *dev, void *data,
 				  struct drm_file *file)
 {
-	struct drm_i915_private *dev_priv = dev->dev_private;
 	struct drm_i915_gem_context_create *args = data;
 	struct drm_i915_file_private *file_priv = file->driver_priv;
-	struct i915_hw_context *ctx;
+	struct intel_context *ctx;
 	int ret;
 
-	if (!(dev->driver->driver_features & DRIVER_GEM))
-		return -ENODEV;
-
-	if (dev_priv->hw_contexts_disabled)
+	if (!hw_context_enabled(dev))
 		return -ENODEV;
 
 	ret = i915_mutex_lock_interruptible(dev);
 	if (ret)
 		return ret;
 
-	ctx = create_hw_context(dev, file_priv);
+	ctx = i915_gem_create_context(dev, file_priv, USES_FULL_PPGTT(dev));
 	mutex_unlock(&dev->struct_mutex);
 	if (IS_ERR(ctx))
 		return PTR_ERR(ctx);
@@ -550,20 +785,20 @@ int i915_gem_context_destroy_ioctl(struct drm_device *dev, void *data,
 {
 	struct drm_i915_gem_context_destroy *args = data;
 	struct drm_i915_file_private *file_priv = file->driver_priv;
-	struct i915_hw_context *ctx;
+	struct intel_context *ctx;
 	int ret;
 
-	if (!(dev->driver->driver_features & DRIVER_GEM))
-		return -ENODEV;
+	if (args->ctx_id == DEFAULT_CONTEXT_ID)
+		return -ENOENT;
 
 	ret = i915_mutex_lock_interruptible(dev);
 	if (ret)
 		return ret;
 
 	ctx = i915_gem_context_get(file_priv, args->ctx_id);
-	if (!ctx) {
+	if (IS_ERR(ctx)) {
 		mutex_unlock(&dev->struct_mutex);
-		return -ENOENT;
+		return PTR_ERR(ctx);
 	}
 
 	idr_remove(&ctx->file_priv->context_idr, ctx->id);