Merge "Merge branch 'android-4.4@b834e92' into branch 'msm-4.4'"

12 files changed, 119 insertions, 106 deletions

diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
index fb9f647bb5cd..5044f2257e89 100644
--- a/drivers/gpu/drm/i915/i915_drv.h
+++ b/drivers/gpu/drm/i915/i915_drv.h
@@ -1159,7 +1159,7 @@ struct intel_gen6_power_mgmt {
 	struct intel_rps_client semaphores, mmioflips;
 
 	/* manual wa residency calculations */
-	struct intel_rps_ei up_ei, down_ei;
+	struct intel_rps_ei ei;
 
 	/*
 	 * Protects RPS/RC6 register access and PCU communication.
diff --git a/drivers/gpu/drm/i915/i915_irq.c b/drivers/gpu/drm/i915/i915_irq.c
index 0f42a2782afc..b7b0a38acd67 100644
--- a/drivers/gpu/drm/i915/i915_irq.c
+++ b/drivers/gpu/drm/i915/i915_irq.c
@@ -994,68 +994,51 @@ static void vlv_c0_read(struct drm_i915_private *dev_priv,
 	ei->media_c0 = I915_READ(VLV_MEDIA_C0_COUNT);
 }
 
-static bool vlv_c0_above(struct drm_i915_private *dev_priv,
-			 const struct intel_rps_ei *old,
-			 const struct intel_rps_ei *now,
-			 int threshold)
-{
-	u64 time, c0;
-	unsigned int mul = 100;
-
-	if (old->cz_clock == 0)
-		return false;
-
-	if (I915_READ(VLV_COUNTER_CONTROL) & VLV_COUNT_RANGE_HIGH)
-		mul <<= 8;
-
-	time = now->cz_clock - old->cz_clock;
-	time *= threshold * dev_priv->czclk_freq;
-
-	/* Workload can be split between render + media, e.g. SwapBuffers
-	 * being blitted in X after being rendered in mesa. To account for
-	 * this we need to combine both engines into our activity counter.
-	 */
-	c0 = now->render_c0 - old->render_c0;
-	c0 += now->media_c0 - old->media_c0;
-	c0 *= mul * VLV_CZ_CLOCK_TO_MILLI_SEC;
-
-	return c0 >= time;
-}
-
 void gen6_rps_reset_ei(struct drm_i915_private *dev_priv)
 {
-	vlv_c0_read(dev_priv, &dev_priv->rps.down_ei);
-	dev_priv->rps.up_ei = dev_priv->rps.down_ei;
+	memset(&dev_priv->rps.ei, 0, sizeof(dev_priv->rps.ei));
 }
 
 static u32 vlv_wa_c0_ei(struct drm_i915_private *dev_priv, u32 pm_iir)
 {
+	const struct intel_rps_ei *prev = &dev_priv->rps.ei;
 	struct intel_rps_ei now;
 	u32 events = 0;
 
-	if ((pm_iir & (GEN6_PM_RP_DOWN_EI_EXPIRED | GEN6_PM_RP_UP_EI_EXPIRED)) == 0)
+	if ((pm_iir & GEN6_PM_RP_UP_EI_EXPIRED) == 0)
 		return 0;
 
 	vlv_c0_read(dev_priv, &now);
 	if (now.cz_clock == 0)
 		return 0;
 
-	if (pm_iir & GEN6_PM_RP_DOWN_EI_EXPIRED) {
-		if (!vlv_c0_above(dev_priv,
-				  &dev_priv->rps.down_ei, &now,
-				  dev_priv->rps.down_threshold))
-			events |= GEN6_PM_RP_DOWN_THRESHOLD;
-		dev_priv->rps.down_ei = now;
-	}
+	if (prev->cz_clock) {
+		u64 time, c0;
+		unsigned int mul;
 
-	if (pm_iir & GEN6_PM_RP_UP_EI_EXPIRED) {
-		if (vlv_c0_above(dev_priv,
-				 &dev_priv->rps.up_ei, &now,
-				 dev_priv->rps.up_threshold))
-			events |= GEN6_PM_RP_UP_THRESHOLD;
-		dev_priv->rps.up_ei = now;
+		mul = VLV_CZ_CLOCK_TO_MILLI_SEC * 100; /* scale to threshold% */
+		if (I915_READ(VLV_COUNTER_CONTROL) & VLV_COUNT_RANGE_HIGH)
+			mul <<= 8;
+
+		time = now.cz_clock - prev->cz_clock;
+		time *= dev_priv->czclk_freq;
+
+		/* Workload can be split between render + media,
+		 * e.g. SwapBuffers being blitted in X after being rendered in
+		 * mesa. To account for this we need to combine both engines
+		 * into our activity counter.
+		 */
+		c0 = now.render_c0 - prev->render_c0;
+		c0 += now.media_c0 - prev->media_c0;
+		c0 *= mul;
+
+		if (c0 > time * dev_priv->rps.up_threshold)
+			events = GEN6_PM_RP_UP_THRESHOLD;
+		else if (c0 < time * dev_priv->rps.down_threshold)
+			events = GEN6_PM_RP_DOWN_THRESHOLD;
 	}
 
+	dev_priv->rps.ei = now;
 	return events;
 }
 
@@ -4390,7 +4373,7 @@ void intel_irq_init(struct drm_i915_private *dev_priv)
 	/* Let's track the enabled rps events */
 	if (IS_VALLEYVIEW(dev_priv) && !IS_CHERRYVIEW(dev_priv))
 		/* WaGsvRC0ResidencyMethod:vlv */
-		dev_priv->pm_rps_events = GEN6_PM_RP_DOWN_EI_EXPIRED | GEN6_PM_RP_UP_EI_EXPIRED;
+		dev_priv->pm_rps_events = GEN6_PM_RP_UP_EI_EXPIRED;
 	else
 		dev_priv->pm_rps_events = GEN6_PM_RPS_EVENTS;
 
diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c
index e7c18519274a..fd4690ed93c0 100644
--- a/drivers/gpu/drm/i915/intel_pm.c
+++ b/drivers/gpu/drm/i915/intel_pm.c
@@ -4376,6 +4376,12 @@ static void gen6_set_rps_thresholds(struct drm_i915_private *dev_priv, u8 val)
 		break;
 	}
 
+	/* When byt can survive without system hang with dynamic
+	 * sw freq adjustments, this restriction can be lifted.
+	 */
+	if (IS_VALLEYVIEW(dev_priv))
+		goto skip_hw_write;
+
 	I915_WRITE(GEN6_RP_UP_EI,
 		GT_INTERVAL_FROM_US(dev_priv, ei_up));
 	I915_WRITE(GEN6_RP_UP_THRESHOLD,
@@ -4394,6 +4400,7 @@ static void gen6_set_rps_thresholds(struct drm_i915_private *dev_priv, u8 val)
 		    GEN6_RP_UP_BUSY_AVG |
 		    GEN6_RP_DOWN_IDLE_AVG);
 
+skip_hw_write:
 	dev_priv->rps.power = new_power;
 	dev_priv->rps.up_threshold = threshold_up;
 	dev_priv->rps.down_threshold = threshold_down;
@@ -4404,8 +4411,9 @@ static u32 gen6_rps_pm_mask(struct drm_i915_private *dev_priv, u8 val)
 {
 	u32 mask = 0;
 
+	/* We use UP_EI_EXPIRED interupts for both up/down in manual mode */
 	if (val > dev_priv->rps.min_freq_softlimit)
-		mask |= GEN6_PM_RP_DOWN_EI_EXPIRED | GEN6_PM_RP_DOWN_THRESHOLD | GEN6_PM_RP_DOWN_TIMEOUT;
+		mask |= GEN6_PM_RP_UP_EI_EXPIRED | GEN6_PM_RP_DOWN_THRESHOLD | GEN6_PM_RP_DOWN_TIMEOUT;
 	if (val < dev_priv->rps.max_freq_softlimit)
 		mask |= GEN6_PM_RP_UP_EI_EXPIRED | GEN6_PM_RP_UP_THRESHOLD;
 
@@ -4509,7 +4517,7 @@ void gen6_rps_busy(struct drm_i915_private *dev_priv)
 {
 	mutex_lock(&dev_priv->rps.hw_lock);
 	if (dev_priv->rps.enabled) {
-		if (dev_priv->pm_rps_events & (GEN6_PM_RP_DOWN_EI_EXPIRED | GEN6_PM_RP_UP_EI_EXPIRED))
+		if (dev_priv->pm_rps_events & GEN6_PM_RP_UP_EI_EXPIRED)
 			gen6_rps_reset_ei(dev_priv);
 		I915_WRITE(GEN6_PMINTRMSK,
 			   gen6_rps_pm_mask(dev_priv, dev_priv->rps.cur_freq));
diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/device/base.c b/drivers/gpu/drm/nouveau/nvkm/engine/device/base.c
index ece9f4102c0e..7f8acb3ebfcd 100644
--- a/drivers/gpu/drm/nouveau/nvkm/engine/device/base.c
+++ b/drivers/gpu/drm/nouveau/nvkm/engine/device/base.c
@@ -714,7 +714,7 @@ nv4a_chipset = {
 	.i2c = nv04_i2c_new,
 	.imem = nv40_instmem_new,
 	.mc = nv44_mc_new,
-	.mmu = nv44_mmu_new,
+	.mmu = nv04_mmu_new,
 	.pci = nv40_pci_new,
 	.therm = nv40_therm_new,
 	.timer = nv41_timer_new,
diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/mpeg/nv31.c b/drivers/gpu/drm/nouveau/nvkm/engine/mpeg/nv31.c
index d4d8942b1347..e55f8302d08a 100644
--- a/drivers/gpu/drm/nouveau/nvkm/engine/mpeg/nv31.c
+++ b/drivers/gpu/drm/nouveau/nvkm/engine/mpeg/nv31.c
@@ -198,7 +198,7 @@ nv31_mpeg_intr(struct nvkm_engine *engine)
 		}
 
 		if (type == 0x00000010) {
-			if (!nv31_mpeg_mthd(mpeg, mthd, data))
+			if (nv31_mpeg_mthd(mpeg, mthd, data))
 				show &= ~0x01000000;
 		}
 	}
diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/mpeg/nv44.c b/drivers/gpu/drm/nouveau/nvkm/engine/mpeg/nv44.c
index d433cfa4a8ab..36af0a8927fc 100644
--- a/drivers/gpu/drm/nouveau/nvkm/engine/mpeg/nv44.c
+++ b/drivers/gpu/drm/nouveau/nvkm/engine/mpeg/nv44.c
@@ -172,7 +172,7 @@ nv44_mpeg_intr(struct nvkm_engine *engine)
 		}
 
 		if (type == 0x00000010) {
-			if (!nv44_mpeg_mthd(subdev->device, mthd, data))
+			if (nv44_mpeg_mthd(subdev->device, mthd, data))
 				show &= ~0x01000000;
 		}
 	}
diff --git a/drivers/gpu/drm/radeon/radeon_ttm.c b/drivers/gpu/drm/radeon/radeon_ttm.c
index 35310336dd0a..d684e2b79d2b 100644
--- a/drivers/gpu/drm/radeon/radeon_ttm.c
+++ b/drivers/gpu/drm/radeon/radeon_ttm.c
@@ -213,8 +213,8 @@ static void radeon_evict_flags(struct ttm_buffer_object *bo,
 			rbo->placement.num_busy_placement = 0;
 			for (i = 0; i < rbo->placement.num_placement; i++) {
 				if (rbo->placements[i].flags & TTM_PL_FLAG_VRAM) {
-					if (rbo->placements[0].fpfn < fpfn)
-						rbo->placements[0].fpfn = fpfn;
+					if (rbo->placements[i].fpfn < fpfn)
+						rbo->placements[i].fpfn = fpfn;
 				} else {
 					rbo->placement.busy_placement =
 						&rbo->placements[i];
diff --git a/drivers/gpu/drm/ttm/ttm_object.c b/drivers/gpu/drm/ttm/ttm_object.c
index 4f5fa8d65fe9..144367c0c28f 100644
--- a/drivers/gpu/drm/ttm/ttm_object.c
+++ b/drivers/gpu/drm/ttm/ttm_object.c
@@ -179,7 +179,7 @@ int ttm_base_object_init(struct ttm_object_file *tfile,
 	if (unlikely(ret != 0))
 		goto out_err0;
 
-	ret = ttm_ref_object_add(tfile, base, TTM_REF_USAGE, NULL);
+	ret = ttm_ref_object_add(tfile, base, TTM_REF_USAGE, NULL, false);
 	if (unlikely(ret != 0))
 		goto out_err1;
 
@@ -318,7 +318,8 @@ EXPORT_SYMBOL(ttm_ref_object_exists);
 
 int ttm_ref_object_add(struct ttm_object_file *tfile,
 		       struct ttm_base_object *base,
-		       enum ttm_ref_type ref_type, bool *existed)
+		       enum ttm_ref_type ref_type, bool *existed,
+		       bool require_existed)
 {
 	struct drm_open_hash *ht = &tfile->ref_hash[ref_type];
 	struct ttm_ref_object *ref;
@@ -345,6 +346,9 @@ int ttm_ref_object_add(struct ttm_object_file *tfile,
 		}
 
 		rcu_read_unlock();
+		if (require_existed)
+			return -EPERM;
+
 		ret = ttm_mem_global_alloc(mem_glob, sizeof(*ref),
 					   false, false);
 		if (unlikely(ret != 0))
@@ -635,7 +639,7 @@ int ttm_prime_fd_to_handle(struct ttm_object_file *tfile,
 	prime = (struct ttm_prime_object *) dma_buf->priv;
 	base = &prime->base;
 	*handle = base->hash.key;
-	ret = ttm_ref_object_add(tfile, base, TTM_REF_USAGE, NULL);
+	ret = ttm_ref_object_add(tfile, base, TTM_REF_USAGE, NULL, false);
 
 	dma_buf_put(dma_buf);
 
diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_fence.c b/drivers/gpu/drm/vmwgfx/vmwgfx_fence.c
index 8e689b439890..6c649f7b5929 100644
--- a/drivers/gpu/drm/vmwgfx/vmwgfx_fence.c
+++ b/drivers/gpu/drm/vmwgfx/vmwgfx_fence.c
@@ -539,7 +539,7 @@ int vmw_fence_create(struct vmw_fence_manager *fman,
 		     struct vmw_fence_obj **p_fence)
 {
 	struct vmw_fence_obj *fence;
-	int ret;
+ 	int ret;
 
 	fence = kzalloc(sizeof(*fence), GFP_KERNEL);
 	if (unlikely(fence == NULL))
@@ -702,6 +702,41 @@ void vmw_fence_fifo_up(struct vmw_fence_manager *fman)
 }
 
 
+/**
+ * vmw_fence_obj_lookup - Look up a user-space fence object
+ *
+ * @tfile: A struct ttm_object_file identifying the caller.
+ * @handle: A handle identifying the fence object.
+ * @return: A struct vmw_user_fence base ttm object on success or
+ * an error pointer on failure.
+ *
+ * The fence object is looked up and type-checked. The caller needs
+ * to have opened the fence object first, but since that happens on
+ * creation and fence objects aren't shareable, that's not an
+ * issue currently.
+ */
+static struct ttm_base_object *
+vmw_fence_obj_lookup(struct ttm_object_file *tfile, u32 handle)
+{
+	struct ttm_base_object *base = ttm_base_object_lookup(tfile, handle);
+
+	if (!base) {
+		pr_err("Invalid fence object handle 0x%08lx.\n",
+		       (unsigned long)handle);
+		return ERR_PTR(-EINVAL);
+	}
+
+	if (base->refcount_release != vmw_user_fence_base_release) {
+		pr_err("Invalid fence object handle 0x%08lx.\n",
+		       (unsigned long)handle);
+		ttm_base_object_unref(&base);
+		return ERR_PTR(-EINVAL);
+	}
+
+	return base;
+}
+
+
 int vmw_fence_obj_wait_ioctl(struct drm_device *dev, void *data,
 			     struct drm_file *file_priv)
 {
@@ -727,13 +762,9 @@ int vmw_fence_obj_wait_ioctl(struct drm_device *dev, void *data,
 		arg->kernel_cookie = jiffies + wait_timeout;
 	}
 
-	base = ttm_base_object_lookup(tfile, arg->handle);
-	if (unlikely(base == NULL)) {
-		printk(KERN_ERR "Wait invalid fence object handle "
-		       "0x%08lx.\n",
-		       (unsigned long)arg->handle);
-		return -EINVAL;
-	}
+	base = vmw_fence_obj_lookup(tfile, arg->handle);
+	if (IS_ERR(base))
+		return PTR_ERR(base);
 
 	fence = &(container_of(base, struct vmw_user_fence, base)->fence);
 
@@ -772,13 +803,9 @@ int vmw_fence_obj_signaled_ioctl(struct drm_device *dev, void *data,
 	struct ttm_object_file *tfile = vmw_fpriv(file_priv)->tfile;
 	struct vmw_private *dev_priv = vmw_priv(dev);
 
-	base = ttm_base_object_lookup(tfile, arg->handle);
-	if (unlikely(base == NULL)) {
-		printk(KERN_ERR "Fence signaled invalid fence object handle "
-		       "0x%08lx.\n",
-		       (unsigned long)arg->handle);
-		return -EINVAL;
-	}
+	base = vmw_fence_obj_lookup(tfile, arg->handle);
+	if (IS_ERR(base))
+		return PTR_ERR(base);
 
 	fence = &(container_of(base, struct vmw_user_fence, base)->fence);
 	fman = fman_from_fence(fence);
@@ -1093,6 +1120,7 @@ int vmw_fence_event_ioctl(struct drm_device *dev, void *data,
 		(struct drm_vmw_fence_event_arg *) data;
 	struct vmw_fence_obj *fence = NULL;
 	struct vmw_fpriv *vmw_fp = vmw_fpriv(file_priv);
+	struct ttm_object_file *tfile = vmw_fp->tfile;
 	struct drm_vmw_fence_rep __user *user_fence_rep =
 		(struct drm_vmw_fence_rep __user *)(unsigned long)
 		arg->fence_rep;
@@ -1106,24 +1134,18 @@ int vmw_fence_event_ioctl(struct drm_device *dev, void *data,
 	 */
 	if (arg->handle) {
 		struct ttm_base_object *base =
-			ttm_base_object_lookup_for_ref(dev_priv->tdev,
-						       arg->handle);
-
-		if (unlikely(base == NULL)) {
-			DRM_ERROR("Fence event invalid fence object handle "
-				  "0x%08lx.\n",
-				  (unsigned long)arg->handle);
-			return -EINVAL;
-		}
+			vmw_fence_obj_lookup(tfile, arg->handle);
+
+		if (IS_ERR(base))
+			return PTR_ERR(base);
+
 		fence = &(container_of(base, struct vmw_user_fence,
 				       base)->fence);
 		(void) vmw_fence_obj_reference(fence);
 
 		if (user_fence_rep != NULL) {
-			bool existed;
-
 			ret = ttm_ref_object_add(vmw_fp->tfile, base,
-						 TTM_REF_USAGE, &existed);
+						 TTM_REF_USAGE, NULL, false);
 			if (unlikely(ret != 0)) {
 				DRM_ERROR("Failed to reference a fence "
 					  "object.\n");
@@ -1166,8 +1188,7 @@ int vmw_fence_event_ioctl(struct drm_device *dev, void *data,
 	return 0;
 out_no_create:
 	if (user_fence_rep != NULL)
-		ttm_ref_object_base_unref(vmw_fpriv(file_priv)->tfile,
-					  handle, TTM_REF_USAGE);
+		ttm_ref_object_base_unref(tfile, handle, TTM_REF_USAGE);
 out_no_ref_obj:
 	vmw_fence_obj_unreference(&fence);
 	return ret;
diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_ioctl.c b/drivers/gpu/drm/vmwgfx/vmwgfx_ioctl.c
index b8c6a03c8c54..5ec24fd801cd 100644
--- a/drivers/gpu/drm/vmwgfx/vmwgfx_ioctl.c
+++ b/drivers/gpu/drm/vmwgfx/vmwgfx_ioctl.c
@@ -114,8 +114,6 @@ int vmw_getparam_ioctl(struct drm_device *dev, void *data,
 		param->value = dev_priv->has_dx;
 		break;
 	default:
-		DRM_ERROR("Illegal vmwgfx get param request: %d\n",
-			  param->param);
 		return -EINVAL;
 	}
 
@@ -186,7 +184,7 @@ int vmw_get_cap_3d_ioctl(struct drm_device *dev, void *data,
 	bool gb_objects = !!(dev_priv->capabilities & SVGA_CAP_GBOBJECTS);
 	struct vmw_fpriv *vmw_fp = vmw_fpriv(file_priv);
 
-	if (unlikely(arg->pad64 != 0)) {
+	if (unlikely(arg->pad64 != 0 || arg->max_size == 0)) {
 		DRM_ERROR("Illegal GET_3D_CAP argument.\n");
 		return -EINVAL;
 	}
diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_resource.c b/drivers/gpu/drm/vmwgfx/vmwgfx_resource.c
index e57667ca7557..dbca128a9aa6 100644
--- a/drivers/gpu/drm/vmwgfx/vmwgfx_resource.c
+++ b/drivers/gpu/drm/vmwgfx/vmwgfx_resource.c
@@ -591,7 +591,7 @@ static int vmw_user_dmabuf_synccpu_grab(struct vmw_user_dma_buffer *user_bo,
 		return ret;
 
 	ret = ttm_ref_object_add(tfile, &user_bo->prime.base,
-				 TTM_REF_SYNCCPU_WRITE, &existed);
+				 TTM_REF_SYNCCPU_WRITE, &existed, false);
 	if (ret != 0 || existed)
 		ttm_bo_synccpu_write_release(&user_bo->dma.base);
 
@@ -775,7 +775,7 @@ int vmw_user_dmabuf_reference(struct ttm_object_file *tfile,
 
 	*handle = user_bo->prime.base.hash.key;
 	return ttm_ref_object_add(tfile, &user_bo->prime.base,
-				  TTM_REF_USAGE, NULL);
+				  TTM_REF_USAGE, NULL, false);
 }
 
 /*
diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_surface.c b/drivers/gpu/drm/vmwgfx/vmwgfx_surface.c
index 7d620e82e000..c9c04ccccdd9 100644
--- a/drivers/gpu/drm/vmwgfx/vmwgfx_surface.c
+++ b/drivers/gpu/drm/vmwgfx/vmwgfx_surface.c
@@ -715,11 +715,14 @@ int vmw_surface_define_ioctl(struct drm_device *dev, void *data,
 			128;
 
 	num_sizes = 0;
-	for (i = 0; i < DRM_VMW_MAX_SURFACE_FACES; ++i)
+	for (i = 0; i < DRM_VMW_MAX_SURFACE_FACES; ++i) {
+		if (req->mip_levels[i] > DRM_VMW_MAX_MIP_LEVELS)
+			return -EINVAL;
 		num_sizes += req->mip_levels[i];
+	}
 
-	if (num_sizes > DRM_VMW_MAX_SURFACE_FACES *
-	    DRM_VMW_MAX_MIP_LEVELS)
+	if (num_sizes > DRM_VMW_MAX_SURFACE_FACES * DRM_VMW_MAX_MIP_LEVELS ||
+	    num_sizes == 0)
 		return -EINVAL;
 
 	size = vmw_user_surface_size + 128 +
@@ -904,17 +907,16 @@ vmw_surface_handle_reference(struct vmw_private *dev_priv,
 	uint32_t handle;
 	struct ttm_base_object *base;
 	int ret;
+	bool require_exist = false;
 
 	if (handle_type == DRM_VMW_HANDLE_PRIME) {
 		ret = ttm_prime_fd_to_handle(tfile, u_handle, &handle);
 		if (unlikely(ret != 0))
 			return ret;
 	} else {
-		if (unlikely(drm_is_render_client(file_priv))) {
-			DRM_ERROR("Render client refused legacy "
-				  "surface reference.\n");
-			return -EACCES;
-		}
+		if (unlikely(drm_is_render_client(file_priv)))
+			require_exist = true;
+
 		if (ACCESS_ONCE(vmw_fpriv(file_priv)->locked_master)) {
 			DRM_ERROR("Locked master refused legacy "
 				  "surface reference.\n");
@@ -942,17 +944,14 @@ vmw_surface_handle_reference(struct vmw_private *dev_priv,
 
 		/*
 		 * Make sure the surface creator has the same
-		 * authenticating master.
+		 * authenticating master, or is already registered with us.
 		 */
 		if (drm_is_primary_client(file_priv) &&
-		    user_srf->master != file_priv->master) {
-			DRM_ERROR("Trying to reference surface outside of"
-				  " master domain.\n");
-			ret = -EACCES;
-			goto out_bad_resource;
-		}
+		    user_srf->master != file_priv->master)
+			require_exist = true;
 
-		ret = ttm_ref_object_add(tfile, base, TTM_REF_USAGE, NULL);
+		ret = ttm_ref_object_add(tfile, base, TTM_REF_USAGE, NULL,
+					 require_exist);
 		if (unlikely(ret != 0)) {
 			DRM_ERROR("Could not add a reference to a surface.\n");
 			goto out_bad_resource;