Merge branch 'drm-fixes' of git://people.freedesktop.org/~airlied/linux

Pull DRM fixes from Dave Airlie:
 "I was going to leave this until post -rc1 but sysfs fixes broke
  hotplug in userspace, so I had to fix it harder, otherwise a set of
  pulls from intel, radeon and vmware,

  The vmware/ttm changes are bit larger but since its early and they are
  unlikely to break anything else I put them in, it lets vmware work
  with dri3"

* 'drm-fixes' of git://people.freedesktop.org/~airlied/linux: (36 commits)
  drm/sysfs: fix hotplug regression since lifetime changes
  drm/exynos: g2d: fix memory leak to userptr
  drm/i915: Fix gen3 self-refresh watermarks
  drm/ttm: Remove set_need_resched from the ttm fault handler
  drm/ttm: Don't move non-existing data
  drm/radeon: hook up backlight functions for CI and KV family.
  drm/i915: Replicate BIOS eDP bpp clamping hack for hsw
  drm/i915: Do not enable package C8 on unsupported hardware
  drm/i915: Hold pc8 lock around toggling pc8.gpu_idle
  drm/i915: encoder->get_config is no longer optional
  drm/i915/tv: add ->get_config callback
  drm/radeon/cik: Add macrotile mode array query
  drm/radeon/cik: Return backend map information to userspace
  drm/vmwgfx: Make vmwgfx dma buffers prime aware
  drm/vmwgfx: Make surfaces prime-aware
  drm/vmwgfx: Hook up the prime ioctls
  drm/ttm: Add a minimal prime implementation for ttm base objects
  drm/vmwgfx: Fix false lockdep warning
  drm/ttm: Allow execbuf util reserves without ticket
  drm/i915: restore the early forcewake cleanup
  ...

54 files changed, 1057 insertions, 324 deletions

diff --git a/arch/x86/kernel/early-quirks.c b/arch/x86/kernel/early-quirks.c
index 96f958d8cd4..bc4a088f902 100644
--- a/arch/x86/kernel/early-quirks.c
+++ b/arch/x86/kernel/early-quirks.c
@@ -330,8 +330,8 @@ static struct pci_device_id intel_stolen_ids[] __initdata = {
 	INTEL_I915GM_IDS(gen3_stolen_size),
 	INTEL_I945G_IDS(gen3_stolen_size),
 	INTEL_I945GM_IDS(gen3_stolen_size),
-	INTEL_VLV_M_IDS(gen3_stolen_size),
-	INTEL_VLV_D_IDS(gen3_stolen_size),
+	INTEL_VLV_M_IDS(gen6_stolen_size),
+	INTEL_VLV_D_IDS(gen6_stolen_size),
 	INTEL_PINEVIEW_IDS(gen3_stolen_size),
 	INTEL_I965G_IDS(gen3_stolen_size),
 	INTEL_G33_IDS(gen3_stolen_size),
diff --git a/drivers/gpu/drm/drm_sysfs.c b/drivers/gpu/drm/drm_sysfs.c
index 1a35ea53106..bd2bca39579 100644
--- a/drivers/gpu/drm/drm_sysfs.c
+++ b/drivers/gpu/drm/drm_sysfs.c
@@ -489,6 +489,11 @@ void drm_sysfs_hotplug_event(struct drm_device *dev)
 }
 EXPORT_SYMBOL(drm_sysfs_hotplug_event);
 
+static void drm_sysfs_release(struct device *dev)
+{
+	kfree(dev);
+}
+
 /**
  * drm_sysfs_device_add - adds a class device to sysfs for a character driver
  * @dev: DRM device to be added
@@ -501,6 +506,7 @@ EXPORT_SYMBOL(drm_sysfs_hotplug_event);
 int drm_sysfs_device_add(struct drm_minor *minor)
 {
 	char *minor_str;
+	int r;
 
 	if (minor->type == DRM_MINOR_CONTROL)
 		minor_str = "controlD%d";
@@ -509,14 +515,34 @@ int drm_sysfs_device_add(struct drm_minor *minor)
         else
                 minor_str = "card%d";
 
-	minor->kdev = device_create(drm_class, minor->dev->dev,
-				    MKDEV(DRM_MAJOR, minor->index),
-				    minor, minor_str, minor->index);
-	if (IS_ERR(minor->kdev)) {
-		DRM_ERROR("device create failed %ld\n", PTR_ERR(minor->kdev));
-		return PTR_ERR(minor->kdev);
+	minor->kdev = kzalloc(sizeof(*minor->kdev), GFP_KERNEL);
+	if (!minor->dev) {
+		r = -ENOMEM;
+		goto error;
 	}
+
+	device_initialize(minor->kdev);
+	minor->kdev->devt = MKDEV(DRM_MAJOR, minor->index);
+	minor->kdev->class = drm_class;
+	minor->kdev->type = &drm_sysfs_device_minor;
+	minor->kdev->parent = minor->dev->dev;
+	minor->kdev->release = drm_sysfs_release;
+	dev_set_drvdata(minor->kdev, minor);
+
+	r = dev_set_name(minor->kdev, minor_str, minor->index);
+	if (r < 0)
+		goto error;
+
+	r = device_add(minor->kdev);
+	if (r < 0)
+		goto error;
+
 	return 0;
+
+error:
+	DRM_ERROR("device create failed %d\n", r);
+	put_device(minor->kdev);
+	return r;
 }
 
 /**
@@ -529,7 +555,7 @@ int drm_sysfs_device_add(struct drm_minor *minor)
 void drm_sysfs_device_remove(struct drm_minor *minor)
 {
 	if (minor->kdev)
-		device_destroy(drm_class, MKDEV(DRM_MAJOR, minor->index));
+		device_unregister(minor->kdev);
 	minor->kdev = NULL;
 }
 
diff --git a/drivers/gpu/drm/exynos/exynos_drm_g2d.c b/drivers/gpu/drm/exynos/exynos_drm_g2d.c
index 3271fd4b172..7bccedca487 100644
--- a/drivers/gpu/drm/exynos/exynos_drm_g2d.c
+++ b/drivers/gpu/drm/exynos/exynos_drm_g2d.c
@@ -383,6 +383,8 @@ out:
 					g2d_userptr->npages,
 					g2d_userptr->vma);
 
+	exynos_gem_put_vma(g2d_userptr->vma);
+
 	if (!g2d_userptr->out_of_list)
 		list_del_init(&g2d_userptr->list);
 
diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
index 8600c315b4c..ccdbecca070 100644
--- a/drivers/gpu/drm/i915/i915_drv.h
+++ b/drivers/gpu/drm/i915/i915_drv.h
@@ -1816,6 +1816,7 @@ struct drm_i915_file_private {
 #define HAS_POWER_WELL(dev)	(IS_HASWELL(dev) || IS_BROADWELL(dev))
 #define HAS_FPGA_DBG_UNCLAIMED(dev)	(INTEL_INFO(dev)->has_fpga_dbg)
 #define HAS_PSR(dev)		(IS_HASWELL(dev) || IS_BROADWELL(dev))
+#define HAS_PC8(dev)		(IS_HASWELL(dev)) /* XXX HSW:ULX */
 
 #define INTEL_PCH_DEVICE_ID_MASK		0xff00
 #define INTEL_PCH_IBX_DEVICE_ID_TYPE		0x3b00
diff --git a/drivers/gpu/drm/i915/intel_bios.c b/drivers/gpu/drm/i915/intel_bios.c
index 6dd622d733b..e4fba39631a 100644
--- a/drivers/gpu/drm/i915/intel_bios.c
+++ b/drivers/gpu/drm/i915/intel_bios.c
@@ -790,7 +790,12 @@ init_vbt_defaults(struct drm_i915_private *dev_priv)
 
 	/* Default to using SSC */
 	dev_priv->vbt.lvds_use_ssc = 1;
-	dev_priv->vbt.lvds_ssc_freq = intel_bios_ssc_frequency(dev, 1);
+	/*
+	 * Core/SandyBridge/IvyBridge use alternative (120MHz) reference
+	 * clock for LVDS.
+	 */
+	dev_priv->vbt.lvds_ssc_freq = intel_bios_ssc_frequency(dev,
+			!HAS_PCH_SPLIT(dev));
 	DRM_DEBUG_KMS("Set default to SSC at %dMHz\n", dev_priv->vbt.lvds_ssc_freq);
 
 	for (port = PORT_A; port < I915_MAX_PORTS; port++) {
diff --git a/drivers/gpu/drm/i915/intel_ddi.c b/drivers/gpu/drm/i915/intel_ddi.c
index 1591576a610..330077bcd0b 100644
--- a/drivers/gpu/drm/i915/intel_ddi.c
+++ b/drivers/gpu/drm/i915/intel_ddi.c
@@ -1406,6 +1406,26 @@ void intel_ddi_get_config(struct intel_encoder *encoder,
 	default:
 		break;
 	}
+
+	if (encoder->type == INTEL_OUTPUT_EDP && dev_priv->vbt.edp_bpp &&
+	    pipe_config->pipe_bpp > dev_priv->vbt.edp_bpp) {
+		/*
+		 * This is a big fat ugly hack.
+		 *
+		 * Some machines in UEFI boot mode provide us a VBT that has 18
+		 * bpp and 1.62 GHz link bandwidth for eDP, which for reasons
+		 * unknown we fail to light up. Yet the same BIOS boots up with
+		 * 24 bpp and 2.7 GHz link. Use the same bpp as the BIOS uses as
+		 * max, not what it tells us to use.
+		 *
+		 * Note: This will still be broken if the eDP panel is not lit
+		 * up by the BIOS, and thus we can't get the mode at module
+		 * load.
+		 */
+		DRM_DEBUG_KMS("pipe has %d bpp for eDP panel, overriding BIOS-provided max %d bpp\n",
+			      pipe_config->pipe_bpp, dev_priv->vbt.edp_bpp);
+		dev_priv->vbt.edp_bpp = pipe_config->pipe_bpp;
+	}
 }
 
 static void intel_ddi_destroy(struct drm_encoder *encoder)
diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c
index 3cddd508d11..7ec8b488bb1 100644
--- a/drivers/gpu/drm/i915/intel_display.c
+++ b/drivers/gpu/drm/i915/intel_display.c
@@ -6518,6 +6518,9 @@ static void __hsw_disable_package_c8(struct drm_i915_private *dev_priv)
 
 void hsw_enable_package_c8(struct drm_i915_private *dev_priv)
 {
+	if (!HAS_PC8(dev_priv->dev))
+		return;
+
 	mutex_lock(&dev_priv->pc8.lock);
 	__hsw_enable_package_c8(dev_priv);
 	mutex_unlock(&dev_priv->pc8.lock);
@@ -6525,6 +6528,9 @@ void hsw_enable_package_c8(struct drm_i915_private *dev_priv)
 
 void hsw_disable_package_c8(struct drm_i915_private *dev_priv)
 {
+	if (!HAS_PC8(dev_priv->dev))
+		return;
+
 	mutex_lock(&dev_priv->pc8.lock);
 	__hsw_disable_package_c8(dev_priv);
 	mutex_unlock(&dev_priv->pc8.lock);
@@ -6562,6 +6568,9 @@ static void hsw_update_package_c8(struct drm_device *dev)
 	struct drm_i915_private *dev_priv = dev->dev_private;
 	bool allow;
 
+	if (!HAS_PC8(dev_priv->dev))
+		return;
+
 	if (!i915_enable_pc8)
 		return;
 
@@ -6585,18 +6594,28 @@ done:
 
 static void hsw_package_c8_gpu_idle(struct drm_i915_private *dev_priv)
 {
+	if (!HAS_PC8(dev_priv->dev))
+		return;
+
+	mutex_lock(&dev_priv->pc8.lock);
 	if (!dev_priv->pc8.gpu_idle) {
 		dev_priv->pc8.gpu_idle = true;
-		hsw_enable_package_c8(dev_priv);
+		__hsw_enable_package_c8(dev_priv);
 	}
+	mutex_unlock(&dev_priv->pc8.lock);
 }
 
 static void hsw_package_c8_gpu_busy(struct drm_i915_private *dev_priv)
 {
+	if (!HAS_PC8(dev_priv->dev))
+		return;
+
+	mutex_lock(&dev_priv->pc8.lock);
 	if (dev_priv->pc8.gpu_idle) {
 		dev_priv->pc8.gpu_idle = false;
-		hsw_disable_package_c8(dev_priv);
+		__hsw_disable_package_c8(dev_priv);
 	}
+	mutex_unlock(&dev_priv->pc8.lock);
 }
 
 #define for_each_power_domain(domain, mask)				\
@@ -7184,7 +7203,9 @@ static void i9xx_update_cursor(struct drm_crtc *crtc, u32 base)
 		intel_crtc->cursor_visible = visible;
 	}
 	/* and commit changes on next vblank */
+	POSTING_READ(CURCNTR(pipe));
 	I915_WRITE(CURBASE(pipe), base);
+	POSTING_READ(CURBASE(pipe));
 }
 
 static void ivb_update_cursor(struct drm_crtc *crtc, u32 base)
@@ -7213,7 +7234,9 @@ static void ivb_update_cursor(struct drm_crtc *crtc, u32 base)
 		intel_crtc->cursor_visible = visible;
 	}
 	/* and commit changes on next vblank */
+	POSTING_READ(CURCNTR_IVB(pipe));
 	I915_WRITE(CURBASE_IVB(pipe), base);
+	POSTING_READ(CURBASE_IVB(pipe));
 }
 
 /* If no-part of the cursor is visible on the framebuffer, then the GPU may hang... */
@@ -9248,8 +9271,7 @@ check_crtc_state(struct drm_device *dev)
 			enum pipe pipe;
 			if (encoder->base.crtc != &crtc->base)
 				continue;
-			if (encoder->get_config &&
-			    encoder->get_hw_state(encoder, &pipe))
+			if (encoder->get_hw_state(encoder, &pipe))
 				encoder->get_config(encoder, &pipe_config);
 		}
 
@@ -10909,8 +10931,7 @@ static void intel_modeset_readout_hw_state(struct drm_device *dev)
 		if (encoder->get_hw_state(encoder, &pipe)) {
 			crtc = to_intel_crtc(dev_priv->pipe_to_crtc_mapping[pipe]);
 			encoder->base.crtc = &crtc->base;
-			if (encoder->get_config)
-				encoder->get_config(encoder, &crtc->config);
+			encoder->get_config(encoder, &crtc->config);
 		} else {
 			encoder->base.crtc = NULL;
 		}
diff --git a/drivers/gpu/drm/i915/intel_dp.c b/drivers/gpu/drm/i915/intel_dp.c
index eb8139da976..0b2e842fef0 100644
--- a/drivers/gpu/drm/i915/intel_dp.c
+++ b/drivers/gpu/drm/i915/intel_dp.c
@@ -1774,7 +1774,7 @@ static void intel_disable_dp(struct intel_encoder *encoder)
 	 * ensure that we have vdd while we switch off the panel. */
 	ironlake_edp_panel_vdd_on(intel_dp);
 	ironlake_edp_backlight_off(intel_dp);
-	intel_dp_sink_dpms(intel_dp, DRM_MODE_DPMS_ON);
+	intel_dp_sink_dpms(intel_dp, DRM_MODE_DPMS_OFF);
 	ironlake_edp_panel_off(intel_dp);
 
 	/* cpu edp my only be disable _after_ the cpu pipe/plane is disabled. */
diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c
index 0a07d7c9caf..caf2ee4e544 100644
--- a/drivers/gpu/drm/i915/intel_pm.c
+++ b/drivers/gpu/drm/i915/intel_pm.c
@@ -1625,7 +1625,7 @@ static void i9xx_update_wm(struct drm_crtc *unused_crtc)
 			&to_intel_crtc(enabled)->config.adjusted_mode;
 		int clock = adjusted_mode->crtc_clock;
 		int htotal = adjusted_mode->htotal;
-		int hdisplay = to_intel_crtc(crtc)->config.pipe_src_w;
+		int hdisplay = to_intel_crtc(enabled)->config.pipe_src_w;
 		int pixel_size = enabled->fb->bits_per_pixel / 8;
 		unsigned long line_time_us;
 		int entries;
@@ -3888,7 +3888,7 @@ static void gen6_enable_rps(struct drm_device *dev)
 
 	I915_WRITE(GEN6_RC_SLEEP, 0);
 	I915_WRITE(GEN6_RC1e_THRESHOLD, 1000);
-	if (INTEL_INFO(dev)->gen <= 6 || IS_IVYBRIDGE(dev))
+	if (IS_IVYBRIDGE(dev))
 		I915_WRITE(GEN6_RC6_THRESHOLD, 125000);
 	else
 		I915_WRITE(GEN6_RC6_THRESHOLD, 50000);
diff --git a/drivers/gpu/drm/i915/intel_tv.c b/drivers/gpu/drm/i915/intel_tv.c
index 18c406246a2..22cf0f4ba24 100644
--- a/drivers/gpu/drm/i915/intel_tv.c
+++ b/drivers/gpu/drm/i915/intel_tv.c
@@ -902,6 +902,13 @@ intel_tv_mode_valid(struct drm_connector *connector,
 }
 
 
+static void
+intel_tv_get_config(struct intel_encoder *encoder,
+		    struct intel_crtc_config *pipe_config)
+{
+	pipe_config->adjusted_mode.crtc_clock = pipe_config->port_clock;
+}
+
 static bool
 intel_tv_compute_config(struct intel_encoder *encoder,
 			struct intel_crtc_config *pipe_config)
@@ -1621,6 +1628,7 @@ intel_tv_init(struct drm_device *dev)
 			 DRM_MODE_ENCODER_TVDAC);
 
 	intel_encoder->compute_config = intel_tv_compute_config;
+	intel_encoder->get_config = intel_tv_get_config;
 	intel_encoder->mode_set = intel_tv_mode_set;
 	intel_encoder->enable = intel_enable_tv;
 	intel_encoder->disable = intel_disable_tv;
diff --git a/drivers/gpu/drm/i915/intel_uncore.c b/drivers/gpu/drm/i915/intel_uncore.c
index f9883ceff94..0b02078a0b8 100644
--- a/drivers/gpu/drm/i915/intel_uncore.c
+++ b/drivers/gpu/drm/i915/intel_uncore.c
@@ -217,6 +217,19 @@ static void gen6_force_wake_work(struct work_struct *work)
 	spin_unlock_irqrestore(&dev_priv->uncore.lock, irqflags);
 }
 
+static void intel_uncore_forcewake_reset(struct drm_device *dev)
+{
+	struct drm_i915_private *dev_priv = dev->dev_private;
+
+	if (IS_VALLEYVIEW(dev)) {
+		vlv_force_wake_reset(dev_priv);
+	} else if (INTEL_INFO(dev)->gen >= 6) {
+		__gen6_gt_force_wake_reset(dev_priv);
+		if (IS_IVYBRIDGE(dev) || IS_HASWELL(dev))
+			__gen6_gt_force_wake_mt_reset(dev_priv);
+	}
+}
+
 void intel_uncore_early_sanitize(struct drm_device *dev)
 {
 	struct drm_i915_private *dev_priv = dev->dev_private;
@@ -234,19 +247,8 @@ void intel_uncore_early_sanitize(struct drm_device *dev)
 		dev_priv->ellc_size = 128;
 		DRM_INFO("Found %zuMB of eLLC\n", dev_priv->ellc_size);
 	}
-}
 
-static void intel_uncore_forcewake_reset(struct drm_device *dev)
-{
-	struct drm_i915_private *dev_priv = dev->dev_private;
-
-	if (IS_VALLEYVIEW(dev)) {
-		vlv_force_wake_reset(dev_priv);
-	} else if (INTEL_INFO(dev)->gen >= 6) {
-		__gen6_gt_force_wake_reset(dev_priv);
-		if (IS_IVYBRIDGE(dev) || IS_HASWELL(dev))
-			__gen6_gt_force_wake_mt_reset(dev_priv);
-	}
+	intel_uncore_forcewake_reset(dev);
 }
 
 void intel_uncore_sanitize(struct drm_device *dev)
diff --git a/drivers/gpu/drm/radeon/atombios_i2c.c b/drivers/gpu/drm/radeon/atombios_i2c.c
index deaf98cdca3..0652ee0a209 100644
--- a/drivers/gpu/drm/radeon/atombios_i2c.c
+++ b/drivers/gpu/drm/radeon/atombios_i2c.c
@@ -56,8 +56,10 @@ static int radeon_process_i2c_ch(struct radeon_i2c_chan *chan,
 			return -EINVAL;
 		}
 		args.ucRegIndex = buf[0];
-		if (num > 1)
-			memcpy(&out, &buf[1], num - 1);
+		if (num > 1) {
+			num--;
+			memcpy(&out, &buf[1], num);
+		}
 		args.lpI2CDataOut = cpu_to_le16(out);
 	} else {
 		if (num > ATOM_MAX_HW_I2C_READ) {
diff --git a/drivers/gpu/drm/radeon/cik.c b/drivers/gpu/drm/radeon/cik.c
index ae92aa041c6..b43a3a3c906 100644
--- a/drivers/gpu/drm/radeon/cik.c
+++ b/drivers/gpu/drm/radeon/cik.c
@@ -1560,17 +1560,17 @@ u32 cik_get_xclk(struct radeon_device *rdev)
  * cik_mm_rdoorbell - read a doorbell dword
  *
  * @rdev: radeon_device pointer
- * @offset: byte offset into the aperture
+ * @index: doorbell index
  *
  * Returns the value in the doorbell aperture at the
- * requested offset (CIK).
+ * requested doorbell index (CIK).
  */
-u32 cik_mm_rdoorbell(struct radeon_device *rdev, u32 offset)
+u32 cik_mm_rdoorbell(struct radeon_device *rdev, u32 index)
 {
-	if (offset < rdev->doorbell.size) {
-		return readl(((void __iomem *)rdev->doorbell.ptr) + offset);
+	if (index < rdev->doorbell.num_doorbells) {
+		return readl(rdev->doorbell.ptr + index);
 	} else {
-		DRM_ERROR("reading beyond doorbell aperture: 0x%08x!\n", offset);
+		DRM_ERROR("reading beyond doorbell aperture: 0x%08x!\n", index);
 		return 0;
 	}
 }
@@ -1579,18 +1579,18 @@ u32 cik_mm_rdoorbell(struct radeon_device *rdev, u32 offset)
  * cik_mm_wdoorbell - write a doorbell dword
  *
  * @rdev: radeon_device pointer
- * @offset: byte offset into the aperture
+ * @index: doorbell index
  * @v: value to write
  *
  * Writes @v to the doorbell aperture at the
- * requested offset (CIK).
+ * requested doorbell index (CIK).
  */
-void cik_mm_wdoorbell(struct radeon_device *rdev, u32 offset, u32 v)
+void cik_mm_wdoorbell(struct radeon_device *rdev, u32 index, u32 v)
 {
-	if (offset < rdev->doorbell.size) {
-		writel(v, ((void __iomem *)rdev->doorbell.ptr) + offset);
+	if (index < rdev->doorbell.num_doorbells) {
+		writel(v, rdev->doorbell.ptr + index);
 	} else {
-		DRM_ERROR("writing beyond doorbell aperture: 0x%08x!\n", offset);
+		DRM_ERROR("writing beyond doorbell aperture: 0x%08x!\n", index);
 	}
 }
 
@@ -2427,6 +2427,7 @@ static void cik_tiling_mode_table_init(struct radeon_device *rdev)
 				gb_tile_moden = 0;
 				break;
 			}
+			rdev->config.cik.macrotile_mode_array[reg_offset] = gb_tile_moden;
 			WREG32(GB_MACROTILE_MODE0 + (reg_offset * 4), gb_tile_moden);
 		}
 	} else if (num_pipe_configs == 4) {
@@ -2773,6 +2774,7 @@ static void cik_tiling_mode_table_init(struct radeon_device *rdev)
 				gb_tile_moden = 0;
 				break;
 			}
+			rdev->config.cik.macrotile_mode_array[reg_offset] = gb_tile_moden;
 			WREG32(GB_MACROTILE_MODE0 + (reg_offset * 4), gb_tile_moden);
 		}
 	} else if (num_pipe_configs == 2) {
@@ -2990,6 +2992,7 @@ static void cik_tiling_mode_table_init(struct radeon_device *rdev)
 				gb_tile_moden = 0;
 				break;
 			}
+			rdev->config.cik.macrotile_mode_array[reg_offset] = gb_tile_moden;
 			WREG32(GB_MACROTILE_MODE0 + (reg_offset * 4), gb_tile_moden);
 		}
 	} else
@@ -3556,17 +3559,24 @@ void cik_fence_compute_ring_emit(struct radeon_device *rdev,
 	radeon_ring_write(ring, 0);
 }
 
-void cik_semaphore_ring_emit(struct radeon_device *rdev,
+bool cik_semaphore_ring_emit(struct radeon_device *rdev,
 			     struct radeon_ring *ring,
 			     struct radeon_semaphore *semaphore,
 			     bool emit_wait)
 {
+/* TODO: figure out why semaphore cause lockups */
+#if 0
 	uint64_t addr = semaphore->gpu_addr;
 	unsigned sel = emit_wait ? PACKET3_SEM_SEL_WAIT : PACKET3_SEM_SEL_SIGNAL;
 
 	radeon_ring_write(ring, PACKET3(PACKET3_MEM_SEMAPHORE, 1));
 	radeon_ring_write(ring, addr & 0xffffffff);
 	radeon_ring_write(ring, (upper_32_bits(addr) & 0xffff) | sel);
+
+	return true;
+#else
+	return false;
+#endif
 }
 
 /**
@@ -3609,13 +3619,8 @@ int cik_copy_cpdma(struct radeon_device *rdev,
 		return r;
 	}
 
-	if (radeon_fence_need_sync(*fence, ring->idx)) {
-		radeon_semaphore_sync_rings(rdev, sem, (*fence)->ring,
-					    ring->idx);
-		radeon_fence_note_sync(*fence, ring->idx);
-	} else {
-		radeon_semaphore_free(rdev, &sem, NULL);
-	}
+	radeon_semaphore_sync_to(sem, *fence);
+	radeon_semaphore_sync_rings(rdev, sem, ring->idx);
 
 	for (i = 0; i < num_loops; i++) {
 		cur_size_in_bytes = size_in_bytes;
@@ -4052,7 +4057,7 @@ void cik_compute_ring_set_wptr(struct radeon_device *rdev,
 			       struct radeon_ring *ring)
 {
 	rdev->wb.wb[ring->wptr_offs/4] = cpu_to_le32(ring->wptr);
-	WDOORBELL32(ring->doorbell_offset, ring->wptr);
+	WDOORBELL32(ring->doorbell_index, ring->wptr);
 }
 
 /**
@@ -4393,10 +4398,6 @@ static int cik_cp_compute_resume(struct radeon_device *rdev)
 			return r;
 		}
 
-		/* doorbell offset */
-		rdev->ring[idx].doorbell_offset =
-			(rdev->ring[idx].doorbell_page_num * PAGE_SIZE) + 0;
-
 		/* init the mqd struct */
 		memset(buf, 0, sizeof(struct bonaire_mqd));
 
@@ -4508,7 +4509,7 @@ static int cik_cp_compute_resume(struct radeon_device *rdev)
 				RREG32(CP_HQD_PQ_DOORBELL_CONTROL);
 			mqd->queue_state.cp_hqd_pq_doorbell_control &= ~DOORBELL_OFFSET_MASK;
 			mqd->queue_state.cp_hqd_pq_doorbell_control |=
-				DOORBELL_OFFSET(rdev->ring[idx].doorbell_offset / 4);
+				DOORBELL_OFFSET(rdev->ring[idx].doorbell_index);
 			mqd->queue_state.cp_hqd_pq_doorbell_control |= DOORBELL_EN;
 			mqd->queue_state.cp_hqd_pq_doorbell_control &=
 				~(DOORBELL_SOURCE | DOORBELL_HIT);
@@ -7839,14 +7840,14 @@ int cik_init(struct radeon_device *rdev)
 	ring = &rdev->ring[CAYMAN_RING_TYPE_CP1_INDEX];
 	ring->ring_obj = NULL;
 	r600_ring_init(rdev, ring, 1024 * 1024);
-	r = radeon_doorbell_get(rdev, &ring->doorbell_page_num);
+	r = radeon_doorbell_get(rdev, &ring->doorbell_index);
 	if (r)
 		return r;
 
 	ring = &rdev->ring[CAYMAN_RING_TYPE_CP2_INDEX];
 	ring->ring_obj = NULL;
 	r600_ring_init(rdev, ring, 1024 * 1024);
-	r = radeon_doorbell_get(rdev, &ring->doorbell_page_num);
+	r = radeon_doorbell_get(rdev, &ring->doorbell_index);
 	if (r)
 		return r;
 
diff --git a/drivers/gpu/drm/radeon/cik_sdma.c b/drivers/gpu/drm/radeon/cik_sdma.c
index 9c9529de20e..0300727a4f7 100644
--- a/drivers/gpu/drm/radeon/cik_sdma.c
+++ b/drivers/gpu/drm/radeon/cik_sdma.c
@@ -130,7 +130,7 @@ void cik_sdma_fence_ring_emit(struct radeon_device *rdev,
  * Add a DMA semaphore packet to the ring wait on or signal
  * other rings (CIK).
  */
-void cik_sdma_semaphore_ring_emit(struct radeon_device *rdev,
+bool cik_sdma_semaphore_ring_emit(struct radeon_device *rdev,
 				  struct radeon_ring *ring,
 				  struct radeon_semaphore *semaphore,
 				  bool emit_wait)
@@ -141,6 +141,8 @@ void cik_sdma_semaphore_ring_emit(struct radeon_device *rdev,
 	radeon_ring_write(ring, SDMA_PACKET(SDMA_OPCODE_SEMAPHORE, 0, extra_bits));
 	radeon_ring_write(ring, addr & 0xfffffff8);
 	radeon_ring_write(ring, upper_32_bits(addr) & 0xffffffff);
+
+	return true;
 }
 
 /**
@@ -443,13 +445,8 @@ int cik_copy_dma(struct radeon_device *rdev,
 		return r;
 	}
 
-	if (radeon_fence_need_sync(*fence, ring->idx)) {
-		radeon_semaphore_sync_rings(rdev, sem, (*fence)->ring,
-					    ring->idx);
-		radeon_fence_note_sync(*fence, ring->idx);
-	} else {
-		radeon_semaphore_free(rdev, &sem, NULL);
-	}
+	radeon_semaphore_sync_to(sem, *fence);
+	radeon_semaphore_sync_rings(rdev, sem, ring->idx);
 
 	for (i = 0; i < num_loops; i++) {
 		cur_size_in_bytes = size_in_bytes;
diff --git a/drivers/gpu/drm/radeon/cypress_dpm.c b/drivers/gpu/drm/radeon/cypress_dpm.c
index 91bb470de0a..920e1e4a52c 100644
--- a/drivers/gpu/drm/radeon/cypress_dpm.c
+++ b/drivers/gpu/drm/radeon/cypress_dpm.c
@@ -299,7 +299,9 @@ void cypress_program_response_times(struct radeon_device *rdev)
 static int cypress_pcie_performance_request(struct radeon_device *rdev,
 					    u8 perf_req, bool advertise)
 {
+#if defined(CONFIG_ACPI)
 	struct evergreen_power_info *eg_pi = evergreen_get_pi(rdev);
+#endif
 	u32 tmp;
 
 	udelay(10);
diff --git a/drivers/gpu/drm/radeon/evergreen_dma.c b/drivers/gpu/drm/radeon/evergreen_dma.c
index 6a0656d00ed..a37b5443638 100644
--- a/drivers/gpu/drm/radeon/evergreen_dma.c
+++ b/drivers/gpu/drm/radeon/evergreen_dma.c
@@ -131,13 +131,8 @@ int evergreen_copy_dma(struct radeon_device *rdev,
 		return r;
 	}
 
-	if (radeon_fence_need_sync(*fence, ring->idx)) {
-		radeon_semaphore_sync_rings(rdev, sem, (*fence)->ring,
-					    ring->idx);
-		radeon_fence_note_sync(*fence, ring->idx);
-	} else {
-		radeon_semaphore_free(rdev, &sem, NULL);
-	}
+	radeon_semaphore_sync_to(sem, *fence);
+	radeon_semaphore_sync_rings(rdev, sem, ring->idx);
 
 	for (i = 0; i < num_loops; i++) {
 		cur_size_in_dw = size_in_dw;
diff --git a/drivers/gpu/drm/radeon/ni_dpm.c b/drivers/gpu/drm/radeon/ni_dpm.c
index f2633902815..cdc003085a7 100644
--- a/drivers/gpu/drm/radeon/ni_dpm.c
+++ b/drivers/gpu/drm/radeon/ni_dpm.c
@@ -3445,9 +3445,9 @@ static int ni_enable_smc_cac(struct radeon_device *rdev,
 static int ni_pcie_performance_request(struct radeon_device *rdev,
 				       u8 perf_req, bool advertise)
 {
+#if defined(CONFIG_ACPI)
 	struct evergreen_power_info *eg_pi = evergreen_get_pi(rdev);
 
-#if defined(CONFIG_ACPI)
 	if ((perf_req == PCIE_PERF_REQ_PECI_GEN1) ||
             (perf_req == PCIE_PERF_REQ_PECI_GEN2)) {
 		if (eg_pi->pcie_performance_request_registered == false)
diff --git a/drivers/gpu/drm/radeon/r100.c b/drivers/gpu/drm/radeon/r100.c
index 784983d7815..10abc4d5a6c 100644
--- a/drivers/gpu/drm/radeon/r100.c
+++ b/drivers/gpu/drm/radeon/r100.c
@@ -869,13 +869,14 @@ void r100_fence_ring_emit(struct radeon_device *rdev,
 	radeon_ring_write(ring, RADEON_SW_INT_FIRE);
 }
 
-void r100_semaphore_ring_emit(struct radeon_device *rdev,
+bool r100_semaphore_ring_emit(struct radeon_device *rdev,
 			      struct radeon_ring *ring,
 			      struct radeon_semaphore *semaphore,
 			      bool emit_wait)
 {
 	/* Unused on older asics, since we don't have semaphores or multiple rings */
 	BUG();
+	return false;
 }
 
 int r100_copy_blit(struct radeon_device *rdev,
diff --git a/drivers/gpu/drm/radeon/r600.c b/drivers/gpu/drm/radeon/r600.c
index 4e609e8a8d2..9ad06732a78 100644
--- a/drivers/gpu/drm/radeon/r600.c
+++ b/drivers/gpu/drm/radeon/r600.c
@@ -2650,7 +2650,7 @@ void r600_fence_ring_emit(struct radeon_device *rdev,
 	}
 }
 
-void r600_semaphore_ring_emit(struct radeon_device *rdev,
+bool r600_semaphore_ring_emit(struct radeon_device *rdev,
 			      struct radeon_ring *ring,
 			      struct radeon_semaphore *semaphore,
 			      bool emit_wait)
@@ -2664,6 +2664,8 @@ void r600_semaphore_ring_emit(struct radeon_device *rdev,
 	radeon_ring_write(ring, PACKET3(PACKET3_MEM_SEMAPHORE, 1));
 	radeon_ring_write(ring, addr & 0xffffffff);
 	radeon_ring_write(ring, (upper_32_bits(addr) & 0xff) | sel);
+
+	return true;
 }
 
 /**
@@ -2706,13 +2708,8 @@ int r600_copy_cpdma(struct radeon_device *rdev,
 		return r;
 	}
 
-	if (radeon_fence_need_sync(*fence, ring->idx)) {
-		radeon_semaphore_sync_rings(rdev, sem, (*fence)->ring,
-					    ring->idx);
-		radeon_fence_note_sync(*fence, ring->idx);
-	} else {
-		radeon_semaphore_free(rdev, &sem, NULL);
-	}
+	radeon_semaphore_sync_to(sem, *fence);
+	radeon_semaphore_sync_rings(rdev, sem, ring->idx);
 
 	radeon_ring_write(ring, PACKET3(PACKET3_SET_CONFIG_REG, 1));
 	radeon_ring_write(ring, (WAIT_UNTIL - PACKET3_SET_CONFIG_REG_OFFSET) >> 2);
diff --git a/drivers/gpu/drm/radeon/r600_dma.c b/drivers/gpu/drm/radeon/r600_dma.c
index 3b317456512..7844d15c139 100644
--- a/drivers/gpu/drm/radeon/r600_dma.c
+++ b/drivers/gpu/drm/radeon/r600_dma.c
@@ -311,7 +311,7 @@ void r600_dma_fence_ring_emit(struct radeon_device *rdev,
  * Add a DMA semaphore packet to the ring wait on or signal
  * other rings (r6xx-SI).
  */
-void r600_dma_semaphore_ring_emit(struct radeon_device *rdev,
+bool r600_dma_semaphore_ring_emit(struct radeon_device *rdev,
 				  struct radeon_ring *ring,
 				  struct radeon_semaphore *semaphore,
 				  bool emit_wait)
@@ -322,6 +322,8 @@ void r600_dma_semaphore_ring_emit(struct radeon_device *rdev,
 	radeon_ring_write(ring, DMA_PACKET(DMA_PACKET_SEMAPHORE, 0, s, 0));
 	radeon_ring_write(ring, addr & 0xfffffffc);
 	radeon_ring_write(ring, upper_32_bits(addr) & 0xff);
+
+	return true;
 }
 
 /**
@@ -462,13 +464,8 @@ int r600_copy_dma(struct radeon_device *rdev,
 		return r;
 	}
 
-	if (radeon_fence_need_sync(*fence, ring->idx)) {
-		radeon_semaphore_sync_rings(rdev, sem, (*fence)->ring,
-					    ring->idx);
-		radeon_fence_note_sync(*fence, ring->idx);
-	} else {
-		radeon_semaphore_free(rdev, &sem, NULL);
-	}
+	radeon_semaphore_sync_to(sem, *fence);
+	radeon_semaphore_sync_rings(rdev, sem, ring->idx);
 
 	for (i = 0; i < num_loops; i++) {
 		cur_size_in_dw = size_in_dw;
diff --git a/drivers/gpu/drm/radeon/radeon.h b/drivers/gpu/drm/radeon/radeon.h
index b9ee9925860..ecf2a3960c0 100644
--- a/drivers/gpu/drm/radeon/radeon.h
+++ b/drivers/gpu/drm/radeon/radeon.h
@@ -348,6 +348,7 @@ int radeon_fence_emit(struct radeon_device *rdev, struct radeon_fence **fence, i
 void radeon_fence_process(struct radeon_device *rdev, int ring);
 bool radeon_fence_signaled(struct radeon_fence *fence);
 int radeon_fence_wait(struct radeon_fence *fence, bool interruptible);
+int radeon_fence_wait_locked(struct radeon_fence *fence);
 int radeon_fence_wait_next_locked(struct radeon_device *rdev, int ring);
 int radeon_fence_wait_empty_locked(struct radeon_device *rdev, int ring);
 int radeon_fence_wait_any(struct radeon_device *rdev,
@@ -548,17 +549,20 @@ struct radeon_semaphore {
 	struct radeon_sa_bo		*sa_bo;
 	signed				waiters;
 	uint64_t			gpu_addr;
+	struct radeon_fence		*sync_to[RADEON_NUM_RINGS];
 };
 
 int radeon_semaphore_create(struct radeon_device *rdev,
 			    struct radeon_semaphore **semaphore);
-void radeon_semaphore_emit_signal(struct radeon_device *rdev, int ring,
+bool radeon_semaphore_emit_signal(struct radeon_device *rdev, int ring,
 				  struct radeon_semaphore *semaphore);
-void radeon_semaphore_emit_wait(struct radeon_device *rdev, int ring,
+bool radeon_semaphore_emit_wait(struct radeon_device *rdev, int ring,
 				struct radeon_semaphore *semaphore);
+void radeon_semaphore_sync_to(struct radeon_semaphore *semaphore,
+			      struct radeon_fence *fence);
 int radeon_semaphore_sync_rings(struct radeon_device *rdev,
 				struct radeon_semaphore *semaphore,
-				int signaler, int waiter);
+				int waiting_ring);
 void radeon_semaphore_free(struct radeon_device *rdev,
 			   struct radeon_semaphore **semaphore,
 			   struct radeon_fence *fence);
@@ -645,13 +649,15 @@ void radeon_scratch_free(struct radeon_device *rdev, uint32_t reg);
 /*
  * GPU doorbell structures, functions & helpers
  */
+#define RADEON_MAX_DOORBELLS 1024	/* Reserve at most 1024 doorbell slots for radeon-owned rings. */
+
 struct radeon_doorbell {
-	u32			num_pages;
-	bool			free[1024];
 	/* doorbell mmio */
-	resource_size_t			base;
-	resource_size_t			size;
-	void __iomem			*ptr;
+	resource_size_t		base;
+	resource_size_t		size;
+	u32 __iomem		*ptr;
+	u32			num_doorbells;	/* Number of doorbells actually reserved for radeon. */
+	unsigned long		used[DIV_ROUND_UP(RADEON_MAX_DOORBELLS, BITS_PER_LONG)];
 };
 
 int radeon_doorbell_get(struct radeon_device *rdev, u32 *page);
@@ -765,7 +771,6 @@ struct radeon_ib {
 	struct radeon_fence		*fence;
 	struct radeon_vm		*vm;
 	bool				is_const_ib;
-	struct radeon_fence		*sync_to[RADEON_NUM_RINGS];
 	struct radeon_semaphore		*semaphore;
 };
 
@@ -799,8 +804,7 @@ struct radeon_ring {
 	u32 pipe;
 	u32 queue;
 	struct radeon_bo	*mqd_obj;
-	u32 doorbell_page_num;
-	u32 doorbell_offset;
+	u32 doorbell_index;
 	unsigned		wptr_offs;
 };
 
@@ -921,7 +925,6 @@ int radeon_ib_get(struct radeon_device *rdev, int ring,
 		  struct radeon_ib *ib, struct radeon_vm *vm,
 		  unsigned size);
 void radeon_ib_free(struct radeon_device *rdev, struct radeon_ib *ib);
-void radeon_ib_sync_to(struct radeon_ib *ib, struct radeon_fence *fence);
 int radeon_ib_schedule(struct radeon_device *rdev, struct radeon_ib *ib,
 		       struct radeon_ib *const_ib);
 int radeon_ib_pool_init(struct radeon_device *rdev);
@@ -1638,7 +1641,7 @@ struct radeon_asic_ring {
 	/* command emmit functions */
 	void (*ib_execute)(struct radeon_device *rdev, struct radeon_ib *ib);
 	void (*emit_fence)(struct radeon_device *rdev, struct radeon_fence *fence);
-	void (*emit_semaphore)(struct radeon_device *rdev, struct radeon_ring *cp,
+	bool (*emit_semaphore)(struct radeon_device *rdev, struct radeon_ring *cp,
 			       struct radeon_semaphore *semaphore, bool emit_wait);
 	void (*vm_flush)(struct radeon_device *rdev, int ridx, struct radeon_vm *vm);
 
@@ -1979,6 +1982,7 @@ struct cik_asic {
 
 	unsigned tile_config;
 	uint32_t tile_mode_array[32];
+	uint32_t macrotile_mode_array[16];
 };
 
 union radeon_asic_config {
@@ -2239,8 +2243,8 @@ void r100_mm_wreg(struct radeon_device *rdev, uint32_t reg, uint32_t v,
 u32 r100_io_rreg(struct radeon_device *rdev, u32 reg);
 void r100_io_wreg(struct radeon_device *rdev, u32 reg, u32 v);
 
-u32 cik_mm_rdoorbell(struct radeon_device *rdev, u32 offset);
-void cik_mm_wdoorbell(struct radeon_device *rdev, u32 offset, u32 v);
+u32 cik_mm_rdoorbell(struct radeon_device *rdev, u32 index);
+void cik_mm_wdoorbell(struct radeon_device *rdev, u32 index, u32 v);
 
 /*
  * Cast helper
@@ -2303,8 +2307,8 @@ void cik_mm_wdoorbell(struct radeon_device *rdev, u32 offset, u32 v);
 #define RREG32_IO(reg) r100_io_rreg(rdev, (reg))
 #define WREG32_IO(reg, v) r100_io_wreg(rdev, (reg), (v))
 
-#define RDOORBELL32(offset) cik_mm_rdoorbell(rdev, (offset))
-#define WDOORBELL32(offset, v) cik_mm_wdoorbell(rdev, (offset), (v))
+#define RDOORBELL32(index) cik_mm_rdoorbell(rdev, (index))
+#define WDOORBELL32(index, v) cik_mm_wdoorbell(rdev, (index), (v))
 
 /*
  * Indirect registers accessor
diff --git a/drivers/gpu/drm/radeon/radeon_asic.c b/drivers/gpu/drm/radeon/radeon_asic.c
index 50853c0cb49..e354ce94cdd 100644
--- a/drivers/gpu/drm/radeon/radeon_asic.c
+++ b/drivers/gpu/drm/radeon/radeon_asic.c
@@ -2015,6 +2015,8 @@ static struct radeon_asic ci_asic = {
 		.bandwidth_update = &dce8_bandwidth_update,
 		.get_vblank_counter = &evergreen_get_vblank_counter,
 		.wait_for_vblank = &dce4_wait_for_vblank,
+		.set_backlight_level = &atombios_set_backlight_level,
+		.get_backlight_level = &atombios_get_backlight_level,
 		.hdmi_enable = &evergreen_hdmi_enable,
 		.hdmi_setmode = &evergreen_hdmi_setmode,
 	},
@@ -2114,6 +2116,8 @@ static struct radeon_asic kv_asic = {
 		.bandwidth_update = &dce8_bandwidth_update,
 		.get_vblank_counter = &evergreen_get_vblank_counter,
 		.wait_for_vblank = &dce4_wait_for_vblank,
+		.set_backlight_level = &atombios_set_backlight_level,
+		.get_backlight_level = &atombios_get_backlight_level,
 		.hdmi_enable = &evergreen_hdmi_enable,
 		.hdmi_setmode = &evergreen_hdmi_setmode,
 	},
diff --git a/drivers/gpu/drm/radeon/radeon_asic.h b/drivers/gpu/drm/radeon/radeon_asic.h
index f2833ee3a61..c9fd97b5807 100644
--- a/drivers/gpu/drm/radeon/radeon_asic.h
+++ b/drivers/gpu/drm/radeon/radeon_asic.h
@@ -80,7 +80,7 @@ int r100_irq_set(struct radeon_device *rdev);
 int r100_irq_process(struct radeon_device *rdev);
 void r100_fence_ring_emit(struct radeon_device *rdev,
 			  struct radeon_fence *fence);
-void r100_semaphore_ring_emit(struct radeon_device *rdev,
+bool r100_semaphore_ring_emit(struct radeon_device *rdev,
 			      struct radeon_ring *cp,
 			      struct radeon_semaphore *semaphore,
 			      bool emit_wait);
@@ -313,13 +313,13 @@ int r600_cs_parse(struct radeon_cs_parser *p);
 int r600_dma_cs_parse(struct radeon_cs_parser *p);
 void r600_fence_ring_emit(struct radeon_device *rdev,
 			  struct radeon_fence *fence);
-void r600_semaphore_ring_emit(struct radeon_device *rdev,
+bool r600_semaphore_ring_emit(struct radeon_device *rdev,
 			      struct radeon_ring *cp,
 			      struct radeon_semaphore *semaphore,
 			      bool emit_wait);
 void r600_dma_fence_ring_emit(struct radeon_device *rdev,
 			      struct radeon_fence *fence);
-void r600_dma_semaphore_ring_emit(struct radeon_device *rdev,
+bool r600_dma_semaphore_ring_emit(struct radeon_device *rdev,
 				  struct radeon_ring *ring,
 				  struct radeon_semaphore *semaphore,
 				  bool emit_wait);
@@ -566,10 +566,6 @@ int sumo_dpm_force_performance_level(struct radeon_device *rdev,
  */
 void cayman_fence_ring_emit(struct radeon_device *rdev,
 			    struct radeon_fence *fence);
-void cayman_uvd_semaphore_emit(struct radeon_device *rdev,
-			       struct radeon_ring *ring,
-			       struct radeon_semaphore *semaphore,
-			       bool emit_wait);
 void cayman_pcie_gart_tlb_flush(struct radeon_device *rdev);
 int cayman_init(struct radeon_device *rdev);
 void cayman_fini(struct radeon_device *rdev);
@@ -697,7 +693,7 @@ void cik_pciep_wreg(struct radeon_device *rdev, uint32_t reg, uint32_t v);
 int cik_set_uvd_clocks(struct radeon_device *rdev, u32 vclk, u32 dclk);
 void cik_sdma_fence_ring_emit(struct radeon_device *rdev,
 			      struct radeon_fence *fence);
-void cik_sdma_semaphore_ring_emit(struct radeon_device *rdev,
+bool cik_sdma_semaphore_ring_emit(struct radeon_device *rdev,
 				  struct radeon_ring *ring,
 				  struct radeon_semaphore *semaphore,
 				  bool emit_wait);
@@ -717,7 +713,7 @@ void cik_fence_gfx_ring_emit(struct radeon_device *rdev,
 			     struct radeon_fence *fence);
 void cik_fence_compute_ring_emit(struct radeon_device *rdev,
 				 struct radeon_fence *fence);
-void cik_semaphore_ring_emit(struct radeon_device *rdev,
+bool cik_semaphore_ring_emit(struct radeon_device *rdev,
 			     struct radeon_ring *cp,
 			     struct radeon_semaphore *semaphore,
 			     bool emit_wait);
@@ -807,7 +803,7 @@ void uvd_v1_0_stop(struct radeon_device *rdev);
 
 int uvd_v1_0_ring_test(struct radeon_device *rdev, struct radeon_ring *ring);
 int uvd_v1_0_ib_test(struct radeon_device *rdev, struct radeon_ring *ring);
-void uvd_v1_0_semaphore_emit(struct radeon_device *rdev,
+bool uvd_v1_0_semaphore_emit(struct radeon_device *rdev,
 			     struct radeon_ring *ring,
 			     struct radeon_semaphore *semaphore,
 			     bool emit_wait);
@@ -819,7 +815,7 @@ void uvd_v2_2_fence_emit(struct radeon_device *rdev,
 			 struct radeon_fence *fence);
 
 /* uvd v3.1 */
-void uvd_v3_1_semaphore_emit(struct radeon_device *rdev,
+bool uvd_v3_1_semaphore_emit(struct radeon_device *rdev,
 			     struct radeon_ring *ring,
 			     struct radeon_semaphore *semaphore,
 			     bool emit_wait);
diff --git a/drivers/gpu/drm/radeon/radeon_cs.c b/drivers/gpu/drm/radeon/radeon_cs.c
index 26ca223d12d..f41594b2eea 100644
--- a/drivers/gpu/drm/radeon/radeon_cs.c
+++ b/drivers/gpu/drm/radeon/radeon_cs.c
@@ -159,7 +159,8 @@ static void radeon_cs_sync_rings(struct radeon_cs_parser *p)
 		if (!p->relocs[i].robj)
 			continue;
 
-		radeon_ib_sync_to(&p->ib, p->relocs[i].robj->tbo.sync_obj);
+		radeon_semaphore_sync_to(p->ib.semaphore,
+					 p->relocs[i].robj->tbo.sync_obj);
 	}
 }
 
@@ -411,9 +412,9 @@ static int radeon_cs_ib_vm_chunk(struct radeon_device *rdev,
 		goto out;
 	}
 	radeon_cs_sync_rings(parser);
-	radeon_ib_sync_to(&parser->ib, vm->fence);
-	radeon_ib_sync_to(&parser->ib, radeon_vm_grab_id(
-		rdev, vm, parser->ring));
+	radeon_semaphore_sync_to(parser->ib.semaphore, vm->fence);
+	radeon_semaphore_sync_to(parser->ib.semaphore,
+				 radeon_vm_grab_id(rdev, vm, parser->ring));
 
 	if ((rdev->family >= CHIP_TAHITI) &&
 	    (parser->chunk_const_ib_idx != -1)) {
diff --git a/drivers/gpu/drm/radeon/radeon_device.c b/drivers/gpu/drm/radeon/radeon_device.c
index b9234c43f43..39b033b441d 100644
--- a/drivers/gpu/drm/radeon/radeon_device.c
+++ b/drivers/gpu/drm/radeon/radeon_device.c
@@ -251,28 +251,23 @@ void radeon_scratch_free(struct radeon_device *rdev, uint32_t reg)
  */
 int radeon_doorbell_init(struct radeon_device *rdev)
 {
-	int i;
-
 	/* doorbell bar mapping */
 	rdev->doorbell.base = pci_resource_start(rdev->pdev, 2);
 	rdev->doorbell.size = pci_resource_len(rdev->pdev, 2);
 
-	/* limit to 4 MB for now */
-	if (rdev->doorbell.size > (4 * 1024 * 1024))
-		rdev->doorbell.size = 4 * 1024 * 1024;
+	rdev->doorbell.num_doorbells = min_t(u32, rdev->doorbell.size / sizeof(u32), RADEON_MAX_DOORBELLS);
+	if (rdev->doorbell.num_doorbells == 0)
+		return -EINVAL;
 
-	rdev->doorbell.ptr = ioremap(rdev->doorbell.base, rdev->doorbell.size);
+	rdev->doorbell.ptr = ioremap(rdev->doorbell.base, rdev->doorbell.num_doorbells * sizeof(u32));
 	if (rdev->doorbell.ptr == NULL) {
 		return -ENOMEM;
 	}
 	DRM_INFO("doorbell mmio base: 0x%08X\n", (uint32_t)rdev->doorbell.base);
 	DRM_INFO("doorbell mmio size: %u\n", (unsigned)rdev->doorbell.size);
 
-	rdev->doorbell.num_pages = rdev->doorbell.size / PAGE_SIZE;
+	memset(&rdev->doorbell.used, 0, sizeof(rdev->doorbell.used));
 
-	for (i = 0; i < rdev->doorbell.num_pages; i++) {
-		rdev->doorbell.free[i] = true;
-	}
 	return 0;
 }
 
@@ -290,40 +285,38 @@ void radeon_doorbell_fini(struct radeon_device *rdev)
 }
 
 /**
- * radeon_doorbell_get - Allocate a doorbell page
+ * radeon_doorbell_get - Allocate a doorbell entry
  *
  * @rdev: radeon_device pointer
- * @doorbell: doorbell page number
+ * @doorbell: doorbell index
  *
- * Allocate a doorbell page for use by the driver (all asics).
+ * Allocate a doorbell for use by the driver (all asics).
  * Returns 0 on success or -EINVAL on failure.
  */
 int radeon_doorbell_get(struct radeon_device *rdev, u32 *doorbell)
 {
-	int i;
-
-	for (i = 0; i < rdev->doorbell.num_pages; i++) {
-		if (rdev->doorbell.free[i]) {
-			rdev->doorbell.free[i] = false;
-			*doorbell = i;
-			return 0;
-		}
+	unsigned long offset = find_first_zero_bit(rdev->doorbell.used, rdev->doorbell.num_doorbells);
+	if (offset < rdev->doorbell.num_doorbells) {
+		__set_bit(offset, rdev->doorbell.used);
+		*doorbell = offset;
+		return 0;
+	} else {
+		return -EINVAL;
 	}
-	return -EINVAL;
 }
 
 /**
- * radeon_doorbell_free - Free a doorbell page
+ * radeon_doorbell_free - Free a doorbell entry
  *
  * @rdev: radeon_device pointer
- * @doorbell: doorbell page number
+ * @doorbell: doorbell index
  *
- * Free a doorbell page allocated for use by the driver (all asics)
+ * Free a doorbell allocated for use by the driver (all asics)
  */
 void radeon_doorbell_free(struct radeon_device *rdev, u32 doorbell)
 {
-	if (doorbell < rdev->doorbell.num_pages)
-		rdev->doorbell.free[doorbell] = true;
+	if (doorbell < rdev->doorbell.num_doorbells)
+		__clear_bit(doorbell, rdev->doorbell.used);
 }
 
 /*
diff --git a/drivers/gpu/drm/radeon/radeon_drv.c b/drivers/gpu/drm/radeon/radeon_drv.c
index 1aee32213f6..9f5ff28864f 100644
--- a/drivers/gpu/drm/radeon/radeon_drv.c
+++ b/drivers/gpu/drm/radeon/radeon_drv.c
@@ -76,9 +76,10 @@
  *   2.32.0 - new info request for rings working
  *   2.33.0 - Add SI tiling mode array query
  *   2.34.0 - Add CIK tiling mode array query
+ *   2.35.0 - Add CIK macrotile mode array query
  */
 #define KMS_DRIVER_MAJOR	2
-#define KMS_DRIVER_MINOR	34
+#define KMS_DRIVER_MINOR	35
 #define KMS_DRIVER_PATCHLEVEL	0
 int radeon_driver_load_kms(struct drm_device *dev, unsigned long flags);
 int radeon_driver_unload_kms(struct drm_device *dev);
diff --git a/drivers/gpu/drm/radeon/radeon_fence.c b/drivers/gpu/drm/radeon/radeon_fence.c
index 281d14c22a4..d3a86e43c01 100644
--- a/drivers/gpu/drm/radeon/radeon_fence.c
+++ b/drivers/gpu/drm/radeon/radeon_fence.c
@@ -472,6 +472,36 @@ int radeon_fence_wait_any(struct radeon_device *rdev,
 }
 
 /**
+ * radeon_fence_wait_locked - wait for a fence to signal
+ *
+ * @fence: radeon fence object
+ *
+ * Wait for the requested fence to signal (all asics).
+ * Returns 0 if the fence has passed, error for all other cases.
+ */
+int radeon_fence_wait_locked(struct radeon_fence *fence)
+{
+	uint64_t seq[RADEON_NUM_RINGS] = {};
+	int r;
+
+	if (fence == NULL) {
+		WARN(1, "Querying an invalid fence : %p !\n", fence);
+		return -EINVAL;
+	}
+
+	seq[fence->ring] = fence->seq;
+	if (seq[fence->ring] == RADEON_FENCE_SIGNALED_SEQ)
+		return 0;
+
+	r = radeon_fence_wait_seq(fence->rdev, seq, false, false);
+	if (r)
+		return r;
+
+	fence->seq = RADEON_FENCE_SIGNALED_SEQ;
+	return 0;
+}
+
+/**
  * radeon_fence_wait_next_locked - wait for the next fence to signal
  *
  * @rdev: radeon device pointer
diff --git a/drivers/gpu/drm/radeon/radeon_gart.c b/drivers/gpu/drm/radeon/radeon_gart.c
index 8a83b89d470..3044e504f4e 100644
--- a/drivers/gpu/drm/radeon/radeon_gart.c
+++ b/drivers/gpu/drm/radeon/radeon_gart.c
@@ -651,7 +651,7 @@ retry:
 	radeon_asic_vm_set_page(rdev, &ib, vm->pd_gpu_addr,
 				0, pd_entries, 0, 0);
 
-	radeon_ib_sync_to(&ib, vm->fence);
+	radeon_semaphore_sync_to(ib.semaphore, vm->fence);
 	r = radeon_ib_schedule(rdev, &ib, NULL);
 	if (r) {
 		radeon_ib_free(rdev, &ib);
@@ -1209,6 +1209,8 @@ int radeon_vm_bo_update_pte(struct radeon_device *rdev,
 		return -ENOMEM;
 
 	r = radeon_ib_get(rdev, R600_RING_TYPE_DMA_INDEX, &ib, NULL, ndw * 4);
+	if (r)
+		return r;
 	ib.length_dw = 0;
 
 	r = radeon_vm_update_pdes(rdev, vm, &ib, bo_va->soffset, bo_va->eoffset);
@@ -1220,7 +1222,7 @@ int radeon_vm_bo_update_pte(struct radeon_device *rdev,
 	radeon_vm_update_ptes(rdev, vm, &ib, bo_va->soffset, bo_va->eoffset,
 			      addr, radeon_vm_page_flags(bo_va->flags));
 
-	radeon_ib_sync_to(&ib, vm->fence);
+	radeon_semaphore_sync_to(ib.semaphore, vm->fence);
 	r = radeon_ib_schedule(rdev, &ib, NULL);
 	if (r) {
 		radeon_ib_free(rdev, &ib);
diff --git a/drivers/gpu/drm/radeon/radeon_kms.c b/drivers/gpu/drm/radeon/radeon_kms.c
index bb8710531a1..55d0b474bd3 100644
--- a/drivers/gpu/drm/radeon/radeon_kms.c
+++ b/drivers/gpu/drm/radeon/radeon_kms.c
@@ -340,7 +340,7 @@ int radeon_info_ioctl(struct drm_device *dev, void *data, struct drm_file *filp)
 		break;
 	case RADEON_INFO_BACKEND_MAP:
 		if (rdev->family >= CHIP_BONAIRE)
-			return -EINVAL;
+			*value = rdev->config.cik.backend_map;
 		else if (rdev->family >= CHIP_TAHITI)
 			*value = rdev->config.si.backend_map;
 		else if (rdev->family >= CHIP_CAYMAN)
@@ -449,6 +449,15 @@ int radeon_info_ioctl(struct drm_device *dev, void *data, struct drm_file *filp)
 			return -EINVAL;
 		}
 		break;
+	case RADEON_INFO_CIK_MACROTILE_MODE_ARRAY:
+		if (rdev->family >= CHIP_BONAIRE) {
+			value = rdev->config.cik.macrotile_mode_array;
+			value_size = sizeof(uint32_t)*16;
+		} else {
+			DRM_DEBUG_KMS("macrotile mode array is cik+ only!\n");
+			return -EINVAL;
+		}
+		break;
 	case RADEON_INFO_SI_CP_DMA_COMPUTE:
 		*value = 1;
 		break;
diff --git a/drivers/gpu/drm/radeon/radeon_legacy_crtc.c b/drivers/gpu/drm/radeon/radeon_legacy_crtc.c
index 0c7b8c66301..0b158f98d28 100644
--- a/drivers/gpu/drm/radeon/radeon_legacy_crtc.c
+++ b/drivers/gpu/drm/radeon/radeon_legacy_crtc.c
@@ -422,6 +422,7 @@ int radeon_crtc_do_set_base(struct drm_crtc *crtc,
 	/* Pin framebuffer & get tilling informations */
 	obj = radeon_fb->obj;
 	rbo = gem_to_radeon_bo(obj);
+retry:
 	r = radeon_bo_reserve(rbo, false);
 	if (unlikely(r != 0))
 		return r;
@@ -430,6 +431,33 @@ int radeon_crtc_do_set_base(struct drm_crtc *crtc,
 				     &base);
 	if (unlikely(r != 0)) {
 		radeon_bo_unreserve(rbo);
+
+		/* On old GPU like RN50 with little vram pining can fails because
+		 * current fb is taking all space needed. So instead of unpining
+		 * the old buffer after pining the new one, first unpin old one
+		 * and then retry pining new one.
+		 *
+		 * As only master can set mode only master can pin and it is
+		 * unlikely the master client will race with itself especialy
+		 * on those old gpu with single crtc.
+		 *
+		 * We don't shutdown the display controller because new buffer
+		 * will end up in same spot.
+		 */
+		if (!atomic && fb && fb != crtc->fb) {
+			struct radeon_bo *old_rbo;
+			unsigned long nsize, osize;
+
+			old_rbo = gem_to_radeon_bo(to_radeon_framebuffer(fb)->obj);
+			osize = radeon_bo_size(old_rbo);
+			nsize = radeon_bo_size(rbo);
+			if (nsize <= osize && !radeon_bo_reserve(old_rbo, false)) {
+				radeon_bo_unpin(old_rbo);
+				radeon_bo_unreserve(old_rbo);
+				fb = NULL;
+				goto retry;
+			}
+		}
 		return -EINVAL;
 	}
 	radeon_bo_get_tiling_flags(rbo, &tiling_flags, NULL);
diff --git a/drivers/gpu/drm/radeon/radeon_pm.c b/drivers/gpu/drm/radeon/radeon_pm.c
index 866ace070b9..d1385ccc672 100644
--- a/drivers/gpu/drm/radeon/radeon_pm.c
+++ b/drivers/gpu/drm/radeon/radeon_pm.c
@@ -1252,7 +1252,6 @@ int radeon_pm_init(struct radeon_device *rdev)
 	case CHIP_RS780:
 	case CHIP_RS880:
 	case CHIP_CAYMAN:
-	case CHIP_ARUBA:
 	case CHIP_BONAIRE:
 	case CHIP_KABINI:
 	case CHIP_KAVERI:
@@ -1284,6 +1283,7 @@ int radeon_pm_init(struct radeon_device *rdev)
 	case CHIP_BARTS:
 	case CHIP_TURKS:
 	case CHIP_CAICOS:
+	case CHIP_ARUBA:
 	case CHIP_TAHITI:
 	case CHIP_PITCAIRN:
 	case CHIP_VERDE:
diff --git a/drivers/gpu/drm/radeon/radeon_ring.c b/drivers/gpu/drm/radeon/radeon_ring.c
index 18254e1c3e7..9214403ae17 100644
--- a/drivers/gpu/drm/radeon/radeon_ring.c
+++ b/drivers/gpu/drm/radeon/radeon_ring.c
@@ -61,7 +61,7 @@ int radeon_ib_get(struct radeon_device *rdev, int ring,
 		  struct radeon_ib *ib, struct radeon_vm *vm,
 		  unsigned size)
 {
-	int i, r;
+	int r;
 
 	r = radeon_sa_bo_new(rdev, &rdev->ring_tmp_bo, &ib->sa_bo, size, 256, true);
 	if (r) {
@@ -87,8 +87,6 @@ int radeon_ib_get(struct radeon_device *rdev, int ring,
 		ib->gpu_addr = radeon_sa_bo_gpu_addr(ib->sa_bo);
 	}
 	ib->is_const_ib = false;
-	for (i = 0; i < RADEON_NUM_RINGS; ++i)
-		ib->sync_to[i] = NULL;
 
 	return 0;
 }
@@ -109,25 +107,6 @@ void radeon_ib_free(struct radeon_device *rdev, struct radeon_ib *ib)
 }
 
 /**
- * radeon_ib_sync_to - sync to fence before executing the IB
- *
- * @ib: IB object to add fence to
- * @fence: fence to sync to
- *
- * Sync to the fence before executing the IB
- */
-void radeon_ib_sync_to(struct radeon_ib *ib, struct radeon_fence *fence)
-{
-	struct radeon_fence *other;
-
-	if (!fence)
-		return;
-
-	other = ib->sync_to[fence->ring];
-	ib->sync_to[fence->ring] = radeon_fence_later(fence, other);
-}
-
-/**
  * radeon_ib_schedule - schedule an IB (Indirect Buffer) on the ring
  *
  * @rdev: radeon_device pointer
@@ -151,8 +130,7 @@ int radeon_ib_schedule(struct radeon_device *rdev, struct radeon_ib *ib,
 		       struct radeon_ib *const_ib)
 {
 	struct radeon_ring *ring = &rdev->ring[ib->ring];
-	bool need_sync = false;
-	int i, r = 0;
+	int r = 0;
 
 	if (!ib->length_dw || !ring->ready) {
 		/* TODO: Nothings in the ib we should report. */
@@ -166,19 +144,15 @@ int radeon_ib_schedule(struct radeon_device *rdev, struct radeon_ib *ib,
 		dev_err(rdev->dev, "scheduling IB failed (%d).\n", r);
 		return r;
 	}
-	for (i = 0; i < RADEON_NUM_RINGS; ++i) {
-		struct radeon_fence *fence = ib->sync_to[i];
-		if (radeon_fence_need_sync(fence, ib->ring)) {
-			need_sync = true;
-			radeon_semaphore_sync_rings(rdev, ib->semaphore,
-						    fence->ring, ib->ring);
-			radeon_fence_note_sync(fence, ib->ring);
-		}
-	}
-	/* immediately free semaphore when we don't need to sync */
-	if (!need_sync) {
-		radeon_semaphore_free(rdev, &ib->semaphore, NULL);
+
+	/* sync with other rings */
+	r = radeon_semaphore_sync_rings(rdev, ib->semaphore, ib->ring);
+	if (r) {
+		dev_err(rdev->dev, "failed to sync rings (%d)\n", r);
+		radeon_ring_unlock_undo(rdev, ring);
+		return r;
 	}
+
 	/* if we can't remember our last VM flush then flush now! */
 	/* XXX figure out why we have to flush for every IB */
 	if (ib->vm /*&& !ib->vm->last_flush*/) {
diff --git a/drivers/gpu/drm/radeon/radeon_semaphore.c b/drivers/gpu/drm/radeon/radeon_semaphore.c
index 8dcc20f53d7..2b42aa1914f 100644
--- a/drivers/gpu/drm/radeon/radeon_semaphore.c
+++ b/drivers/gpu/drm/radeon/radeon_semaphore.c
@@ -29,12 +29,12 @@
  */
 #include <drm/drmP.h>
 #include "radeon.h"
-
+#include "radeon_trace.h"
 
 int radeon_semaphore_create(struct radeon_device *rdev,
 			    struct radeon_semaphore **semaphore)
 {
-	int r;
+	int i, r;
 
 	*semaphore = kmalloc(sizeof(struct radeon_semaphore), GFP_KERNEL);
 	if (*semaphore == NULL) {
@@ -50,54 +50,121 @@ int radeon_semaphore_create(struct radeon_device *rdev,
 	(*semaphore)->waiters = 0;
 	(*semaphore)->gpu_addr = radeon_sa_bo_gpu_addr((*semaphore)->sa_bo);
 	*((uint64_t*)radeon_sa_bo_cpu_addr((*semaphore)->sa_bo)) = 0;
+
+	for (i = 0; i < RADEON_NUM_RINGS; ++i)
+		(*semaphore)->sync_to[i] = NULL;
+
 	return 0;
 }
 
-void radeon_semaphore_emit_signal(struct radeon_device *rdev, int ring,
+bool radeon_semaphore_emit_signal(struct radeon_device *rdev, int ridx,
 			          struct radeon_semaphore *semaphore)
 {
-	--semaphore->waiters;
-	radeon_semaphore_ring_emit(rdev, ring, &rdev->ring[ring], semaphore, false);
+	struct radeon_ring *ring = &rdev->ring[ridx];
+
+	trace_radeon_semaphore_signale(ridx, semaphore);
+
+	if (radeon_semaphore_ring_emit(rdev, ridx, ring, semaphore, false)) {
+		--semaphore->waiters;
+
+		/* for debugging lockup only, used by sysfs debug files */
+		ring->last_semaphore_signal_addr = semaphore->gpu_addr;
+		return true;
+	}
+	return false;
 }
 
-void radeon_semaphore_emit_wait(struct radeon_device *rdev, int ring,
+bool radeon_semaphore_emit_wait(struct radeon_device *rdev, int ridx,
 			        struct radeon_semaphore *semaphore)
 {
-	++semaphore->waiters;
-	radeon_semaphore_ring_emit(rdev, ring, &rdev->ring[ring], semaphore, true);
+	struct radeon_ring *ring = &rdev->ring[ridx];
+
+	trace_radeon_semaphore_wait(ridx, semaphore);
+
+	if (radeon_semaphore_ring_emit(rdev, ridx, ring, semaphore, true)) {
+		++semaphore->waiters;
+
+		/* for debugging lockup only, used by sysfs debug files */
+		ring->last_semaphore_wait_addr = semaphore->gpu_addr;
+		return true;
+	}
+	return false;
+}
+
+/**
+ * radeon_semaphore_sync_to - use the semaphore to sync to a fence
+ *
+ * @semaphore: semaphore object to add fence to
+ * @fence: fence to sync to
+ *
+ * Sync to the fence using this semaphore object
+ */
+void radeon_semaphore_sync_to(struct radeon_semaphore *semaphore,
+			      struct radeon_fence *fence)
+{
+        struct radeon_fence *other;
+
+        if (!fence)
+                return;
+
+        other = semaphore->sync_to[fence->ring];
+        semaphore->sync_to[fence->ring] = radeon_fence_later(fence, other);
 }
 
-/* caller must hold ring lock */
+/**
+ * radeon_semaphore_sync_rings - sync ring to all registered fences
+ *
+ * @rdev: radeon_device pointer
+ * @semaphore: semaphore object to use for sync
+ * @ring: ring that needs sync
+ *
+ * Ensure that all registered fences are signaled before letting
+ * the ring continue. The caller must hold the ring lock.
+ */
 int radeon_semaphore_sync_rings(struct radeon_device *rdev,
 				struct radeon_semaphore *semaphore,
-				int signaler, int waiter)
+				int ring)
 {
-	int r;
+	int i, r;
 
-	/* no need to signal and wait on the same ring */
-	if (signaler == waiter) {
-		return 0;
-	}
+        for (i = 0; i < RADEON_NUM_RINGS; ++i) {
+		struct radeon_fence *fence = semaphore->sync_to[i];
 
-	/* prevent GPU deadlocks */
-	if (!rdev->ring[signaler].ready) {
-		dev_err(rdev->dev, "Trying to sync to a disabled ring!");
-		return -EINVAL;
-	}
+		/* check if we really need to sync */
+                if (!radeon_fence_need_sync(fence, ring))
+			continue;
 
-	r = radeon_ring_alloc(rdev, &rdev->ring[signaler], 8);
-	if (r) {
-		return r;
-	}
-	radeon_semaphore_emit_signal(rdev, signaler, semaphore);
-	radeon_ring_commit(rdev, &rdev->ring[signaler]);
+		/* prevent GPU deadlocks */
+		if (!rdev->ring[i].ready) {
+			dev_err(rdev->dev, "Syncing to a disabled ring!");
+			return -EINVAL;
+		}
 
-	/* we assume caller has already allocated space on waiters ring */
-	radeon_semaphore_emit_wait(rdev, waiter, semaphore);
+		/* allocate enough space for sync command */
+		r = radeon_ring_alloc(rdev, &rdev->ring[i], 16);
+		if (r) {
+			return r;
+		}
 
-	/* for debugging lockup only, used by sysfs debug files */
-	rdev->ring[signaler].last_semaphore_signal_addr = semaphore->gpu_addr;
-	rdev->ring[waiter].last_semaphore_wait_addr = semaphore->gpu_addr;
+		/* emit the signal semaphore */
+		if (!radeon_semaphore_emit_signal(rdev, i, semaphore)) {
+			/* signaling wasn't successful wait manually */
+			radeon_ring_undo(&rdev->ring[i]);
+			radeon_fence_wait_locked(fence);
+			continue;
+		}
+
+		/* we assume caller has already allocated space on waiters ring */
+		if (!radeon_semaphore_emit_wait(rdev, ring, semaphore)) {
+			/* waiting wasn't successful wait manually */
+			radeon_ring_undo(&rdev->ring[i]);
+			radeon_fence_wait_locked(fence);
+			continue;
+		}
+
+		radeon_ring_commit(rdev, &rdev->ring[i]);
+		radeon_fence_note_sync(fence, ring);
+	}
 
 	return 0;
 }
diff --git a/drivers/gpu/drm/radeon/radeon_trace.h b/drivers/gpu/drm/radeon/radeon_trace.h
index 811bca691b3..9f0e18172b6 100644
--- a/drivers/gpu/drm/radeon/radeon_trace.h
+++ b/drivers/gpu/drm/radeon/radeon_trace.h
@@ -111,6 +111,42 @@ DEFINE_EVENT(radeon_fence_request, radeon_fence_wait_end,
 	    TP_ARGS(dev, seqno)
 );
 
+DECLARE_EVENT_CLASS(radeon_semaphore_request,
+
+	    TP_PROTO(int ring, struct radeon_semaphore *sem),
+
+	    TP_ARGS(ring, sem),
+
+	    TP_STRUCT__entry(
+			     __field(int, ring)
+			     __field(signed, waiters)
+			     __field(uint64_t, gpu_addr)
+			     ),
+
+	    TP_fast_assign(
+			   __entry->ring = ring;
+			   __entry->waiters = sem->waiters;
+			   __entry->gpu_addr = sem->gpu_addr;
+			   ),
+
+	    TP_printk("ring=%u, waiters=%d, addr=%010Lx", __entry->ring,
+		      __entry->waiters, __entry->gpu_addr)
+);
+
+DEFINE_EVENT(radeon_semaphore_request, radeon_semaphore_signale,
+
+	    TP_PROTO(int ring, struct radeon_semaphore *sem),
+
+	    TP_ARGS(ring, sem)
+);
+
+DEFINE_EVENT(radeon_semaphore_request, radeon_semaphore_wait,
+
+	    TP_PROTO(int ring, struct radeon_semaphore *sem),
+
+	    TP_ARGS(ring, sem)
+);
+
 #endif
 
 /* This part must be outside protection */
diff --git a/drivers/gpu/drm/radeon/rv770_dma.c b/drivers/gpu/drm/radeon/rv770_dma.c
index f9b02e3d683..aca8cbe8a33 100644
--- a/drivers/gpu/drm/radeon/rv770_dma.c
+++ b/drivers/gpu/drm/radeon/rv770_dma.c
@@ -66,13 +66,8 @@ int rv770_copy_dma(struct radeon_device *rdev,
 		return r;
 	}
 
-	if (radeon_fence_need_sync(*fence, ring->idx)) {
-		radeon_semaphore_sync_rings(rdev, sem, (*fence)->ring,
-					    ring->idx);
-		radeon_fence_note_sync(*fence, ring->idx);
-	} else {
-		radeon_semaphore_free(rdev, &sem, NULL);
-	}
+	radeon_semaphore_sync_to(sem, *fence);
+	radeon_semaphore_sync_rings(rdev, sem, ring->idx);
 
 	for (i = 0; i < num_loops; i++) {
 		cur_size_in_dw = size_in_dw;
diff --git a/drivers/gpu/drm/radeon/si_dma.c b/drivers/gpu/drm/radeon/si_dma.c
index 8e8f4613353..59be2cfcbb4 100644
--- a/drivers/gpu/drm/radeon/si_dma.c
+++ b/drivers/gpu/drm/radeon/si_dma.c
@@ -195,13 +195,8 @@ int si_copy_dma(struct radeon_device *rdev,
 		return r;
 	}
 
-	if (radeon_fence_need_sync(*fence, ring->idx)) {
-		radeon_semaphore_sync_rings(rdev, sem, (*fence)->ring,
-					    ring->idx);
-		radeon_fence_note_sync(*fence, ring->idx);
-	} else {
-		radeon_semaphore_free(rdev, &sem, NULL);
-	}
+	radeon_semaphore_sync_to(sem, *fence);
+	radeon_semaphore_sync_rings(rdev, sem, ring->idx);
 
 	for (i = 0; i < num_loops; i++) {
 		cur_size_in_bytes = size_in_bytes;
diff --git a/drivers/gpu/drm/radeon/trinity_dpm.c b/drivers/gpu/drm/radeon/trinity_dpm.c
index 9364129ba29..d700698a1f2 100644
--- a/drivers/gpu/drm/radeon/trinity_dpm.c
+++ b/drivers/gpu/drm/radeon/trinity_dpm.c
@@ -1873,9 +1873,9 @@ int trinity_dpm_init(struct radeon_device *rdev)
 	pi->enable_sclk_ds = true;
 	pi->enable_gfx_power_gating = true;
 	pi->enable_gfx_clock_gating = true;
-	pi->enable_mg_clock_gating = true;
-	pi->enable_gfx_dynamic_mgpg = true; /* ??? */
-	pi->override_dynamic_mgpg = true;
+	pi->enable_mg_clock_gating = false;
+	pi->enable_gfx_dynamic_mgpg = false;
+	pi->override_dynamic_mgpg = false;
 	pi->enable_auto_thermal_throttling = true;
 	pi->voltage_drop_in_dce = false; /* need to restructure dpm/modeset interaction */
 	pi->uvd_dpm = true; /* ??? */
diff --git a/drivers/gpu/drm/radeon/uvd_v1_0.c b/drivers/gpu/drm/radeon/uvd_v1_0.c
index 7266805d978..d4a68af1a27 100644
--- a/drivers/gpu/drm/radeon/uvd_v1_0.c
+++ b/drivers/gpu/drm/radeon/uvd_v1_0.c
@@ -357,7 +357,7 @@ int uvd_v1_0_ring_test(struct radeon_device *rdev, struct radeon_ring *ring)
  *
  * Emit a semaphore command (either wait or signal) to the UVD ring.
  */
-void uvd_v1_0_semaphore_emit(struct radeon_device *rdev,
+bool uvd_v1_0_semaphore_emit(struct radeon_device *rdev,
 			     struct radeon_ring *ring,
 			     struct radeon_semaphore *semaphore,
 			     bool emit_wait)
@@ -372,6 +372,8 @@ void uvd_v1_0_semaphore_emit(struct radeon_device *rdev,
 
 	radeon_ring_write(ring, PACKET0(UVD_SEMA_CMD, 0));
 	radeon_ring_write(ring, emit_wait ? 1 : 0);
+
+	return true;
 }
 
 /**
diff --git a/drivers/gpu/drm/radeon/uvd_v3_1.c b/drivers/gpu/drm/radeon/uvd_v3_1.c
index 5b6fa1f62d4..d722db2cf34 100644
--- a/drivers/gpu/drm/radeon/uvd_v3_1.c
+++ b/drivers/gpu/drm/radeon/uvd_v3_1.c
@@ -37,7 +37,7 @@
  *
  * Emit a semaphore command (either wait or signal) to the UVD ring.
  */
-void uvd_v3_1_semaphore_emit(struct radeon_device *rdev,
+bool uvd_v3_1_semaphore_emit(struct radeon_device *rdev,
 			     struct radeon_ring *ring,
 			     struct radeon_semaphore *semaphore,
 			     bool emit_wait)
@@ -52,4 +52,6 @@ void uvd_v3_1_semaphore_emit(struct radeon_device *rdev,
 
 	radeon_ring_write(ring, PACKET0(UVD_SEMA_CMD, 0));
 	radeon_ring_write(ring, 0x80 | (emit_wait ? 1 : 0));
+
+	return true;
 }
diff --git a/drivers/gpu/drm/ttm/ttm_bo.c b/drivers/gpu/drm/ttm/ttm_bo.c
index 8d5a646ebe6..07e02c4bf5a 100644
--- a/drivers/gpu/drm/ttm/ttm_bo.c
+++ b/drivers/gpu/drm/ttm/ttm_bo.c
@@ -151,7 +151,7 @@ static void ttm_bo_release_list(struct kref *list_kref)
 	atomic_dec(&bo->glob->bo_count);
 	if (bo->resv == &bo->ttm_resv)
 		reservation_object_fini(&bo->ttm_resv);
-
+	mutex_destroy(&bo->wu_mutex);
 	if (bo->destroy)
 		bo->destroy(bo);
 	else {
@@ -1123,6 +1123,7 @@ int ttm_bo_init(struct ttm_bo_device *bdev,
 	INIT_LIST_HEAD(&bo->ddestroy);
 	INIT_LIST_HEAD(&bo->swap);
 	INIT_LIST_HEAD(&bo->io_reserve_lru);
+	mutex_init(&bo->wu_mutex);
 	bo->bdev = bdev;
 	bo->glob = bdev->glob;
 	bo->type = type;
@@ -1704,3 +1705,35 @@ void ttm_bo_swapout_all(struct ttm_bo_device *bdev)
 		;
 }
 EXPORT_SYMBOL(ttm_bo_swapout_all);
+
+/**
+ * ttm_bo_wait_unreserved - interruptible wait for a buffer object to become
+ * unreserved
+ *
+ * @bo: Pointer to buffer
+ */
+int ttm_bo_wait_unreserved(struct ttm_buffer_object *bo)
+{
+	int ret;
+
+	/*
+	 * In the absense of a wait_unlocked API,
+	 * Use the bo::wu_mutex to avoid triggering livelocks due to
+	 * concurrent use of this function. Note that this use of
+	 * bo::wu_mutex can go away if we change locking order to
+	 * mmap_sem -> bo::reserve.
+	 */
+	ret = mutex_lock_interruptible(&bo->wu_mutex);
+	if (unlikely(ret != 0))
+		return -ERESTARTSYS;
+	if (!ww_mutex_is_locked(&bo->resv->lock))
+		goto out_unlock;
+	ret = ttm_bo_reserve_nolru(bo, true, false, false, NULL);
+	if (unlikely(ret != 0))
+		goto out_unlock;
+	ww_mutex_unlock(&bo->resv->lock);
+
+out_unlock:
+	mutex_unlock(&bo->wu_mutex);
+	return ret;
+}
diff --git a/drivers/gpu/drm/ttm/ttm_bo_util.c b/drivers/gpu/drm/ttm/ttm_bo_util.c
index 4834c463c38..15b86a94949 100644
--- a/drivers/gpu/drm/ttm/ttm_bo_util.c
+++ b/drivers/gpu/drm/ttm/ttm_bo_util.c
@@ -350,10 +350,13 @@ int ttm_bo_move_memcpy(struct ttm_buffer_object *bo,
 		goto out2;
 
 	/*
-	 * Move nonexistent data. NOP.
+	 * Don't move nonexistent data. Clear destination instead.
 	 */
-	if (old_iomap == NULL && ttm == NULL)
+	if (old_iomap == NULL &&
+	    (ttm == NULL || ttm->state == tt_unpopulated)) {
+		memset_io(new_iomap, 0, new_mem->num_pages*PAGE_SIZE);
 		goto out2;
+	}
 
 	/*
 	 * TTM might be null for moves within the same region.
diff --git a/drivers/gpu/drm/ttm/ttm_bo_vm.c b/drivers/gpu/drm/ttm/ttm_bo_vm.c
index ac617f3ecd0..b249ab9b1eb 100644
--- a/drivers/gpu/drm/ttm/ttm_bo_vm.c
+++ b/drivers/gpu/drm/ttm/ttm_bo_vm.c
@@ -107,13 +107,28 @@ static int ttm_bo_vm_fault(struct vm_area_struct *vma, struct vm_fault *vmf)
 	/*
 	 * Work around locking order reversal in fault / nopfn
 	 * between mmap_sem and bo_reserve: Perform a trylock operation
-	 * for reserve, and if it fails, retry the fault after scheduling.
+	 * for reserve, and if it fails, retry the fault after waiting
+	 * for the buffer to become unreserved.
 	 */
-
-	ret = ttm_bo_reserve(bo, true, true, false, 0);
+	ret = ttm_bo_reserve(bo, true, true, false, NULL);
 	if (unlikely(ret != 0)) {
-		if (ret == -EBUSY)
-			set_need_resched();
+		if (ret != -EBUSY)
+			return VM_FAULT_NOPAGE;
+
+		if (vmf->flags & FAULT_FLAG_ALLOW_RETRY) {
+			if (!(vmf->flags & FAULT_FLAG_RETRY_NOWAIT)) {
+				up_read(&vma->vm_mm->mmap_sem);
+				(void) ttm_bo_wait_unreserved(bo);
+			}
+
+			return VM_FAULT_RETRY;
+		}
+
+		/*
+		 * If we'd want to change locking order to
+		 * mmap_sem -> bo::reserve, we'd use a blocking reserve here
+		 * instead of retrying the fault...
+		 */
 		return VM_FAULT_NOPAGE;
 	}
 
@@ -123,7 +138,6 @@ static int ttm_bo_vm_fault(struct vm_area_struct *vma, struct vm_fault *vmf)
 		case 0:
 			break;
 		case -EBUSY:
-			set_need_resched();
 		case -ERESTARTSYS:
 			retval = VM_FAULT_NOPAGE;
 			goto out_unlock;
diff --git a/drivers/gpu/drm/ttm/ttm_execbuf_util.c b/drivers/gpu/drm/ttm/ttm_execbuf_util.c
index 6c911789ae5..479e9418e3d 100644
--- a/drivers/gpu/drm/ttm/ttm_execbuf_util.c
+++ b/drivers/gpu/drm/ttm/ttm_execbuf_util.c
@@ -32,8 +32,7 @@
 #include <linux/sched.h>
 #include <linux/module.h>
 
-static void ttm_eu_backoff_reservation_locked(struct list_head *list,
-					      struct ww_acquire_ctx *ticket)
+static void ttm_eu_backoff_reservation_locked(struct list_head *list)
 {
 	struct ttm_validate_buffer *entry;
 
@@ -93,8 +92,9 @@ void ttm_eu_backoff_reservation(struct ww_acquire_ctx *ticket,
 	entry = list_first_entry(list, struct ttm_validate_buffer, head);
 	glob = entry->bo->glob;
 	spin_lock(&glob->lru_lock);
-	ttm_eu_backoff_reservation_locked(list, ticket);
-	ww_acquire_fini(ticket);
+	ttm_eu_backoff_reservation_locked(list);
+	if (ticket)
+		ww_acquire_fini(ticket);
 	spin_unlock(&glob->lru_lock);
 }
 EXPORT_SYMBOL(ttm_eu_backoff_reservation);
@@ -130,7 +130,8 @@ int ttm_eu_reserve_buffers(struct ww_acquire_ctx *ticket,
 	entry = list_first_entry(list, struct ttm_validate_buffer, head);
 	glob = entry->bo->glob;
 
-	ww_acquire_init(ticket, &reservation_ww_class);
+	if (ticket)
+		ww_acquire_init(ticket, &reservation_ww_class);
 retry:
 	list_for_each_entry(entry, list, head) {
 		struct ttm_buffer_object *bo = entry->bo;
@@ -139,16 +140,17 @@ retry:
 		if (entry->reserved)
 			continue;
 
-
-		ret = ttm_bo_reserve_nolru(bo, true, false, true, ticket);
+		ret = ttm_bo_reserve_nolru(bo, true, (ticket == NULL), true,
+					   ticket);
 
 		if (ret == -EDEADLK) {
 			/* uh oh, we lost out, drop every reservation and try
 			 * to only reserve this buffer, then start over if
 			 * this succeeds.
 			 */
+			BUG_ON(ticket == NULL);
 			spin_lock(&glob->lru_lock);
-			ttm_eu_backoff_reservation_locked(list, ticket);
+			ttm_eu_backoff_reservation_locked(list);
 			spin_unlock(&glob->lru_lock);
 			ttm_eu_list_ref_sub(list);
 			ret = ww_mutex_lock_slow_interruptible(&bo->resv->lock,
@@ -175,7 +177,8 @@ retry:
 		}
 	}
 
-	ww_acquire_done(ticket);
+	if (ticket)
+		ww_acquire_done(ticket);
 	spin_lock(&glob->lru_lock);
 	ttm_eu_del_from_lru_locked(list);
 	spin_unlock(&glob->lru_lock);
@@ -184,12 +187,14 @@ retry:
 
 err:
 	spin_lock(&glob->lru_lock);
-	ttm_eu_backoff_reservation_locked(list, ticket);
+	ttm_eu_backoff_reservation_locked(list);
 	spin_unlock(&glob->lru_lock);
 	ttm_eu_list_ref_sub(list);
 err_fini:
-	ww_acquire_done(ticket);
-	ww_acquire_fini(ticket);
+	if (ticket) {
+		ww_acquire_done(ticket);
+		ww_acquire_fini(ticket);
+	}
 	return ret;
 }
 EXPORT_SYMBOL(ttm_eu_reserve_buffers);
@@ -224,7 +229,8 @@ void ttm_eu_fence_buffer_objects(struct ww_acquire_ctx *ticket,
 	}
 	spin_unlock(&bdev->fence_lock);
 	spin_unlock(&glob->lru_lock);
-	ww_acquire_fini(ticket);
+	if (ticket)
+		ww_acquire_fini(ticket);
 
 	list_for_each_entry(entry, list, head) {
 		if (entry->old_sync_obj)
diff --git a/drivers/gpu/drm/ttm/ttm_object.c b/drivers/gpu/drm/ttm/ttm_object.c
index a868176c258..6fe7b92a82d 100644
--- a/drivers/gpu/drm/ttm/ttm_object.c
+++ b/drivers/gpu/drm/ttm/ttm_object.c
@@ -1,6 +1,6 @@
 /**************************************************************************
  *
- * Copyright (c) 2009 VMware, Inc., Palo Alto, CA., USA
+ * Copyright (c) 2009-2013 VMware, Inc., Palo Alto, CA., USA
  * All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
@@ -26,6 +26,12 @@
  **************************************************************************/
 /*
  * Authors: Thomas Hellstrom <thellstrom-at-vmware-dot-com>
+ *
+ * While no substantial code is shared, the prime code is inspired by
+ * drm_prime.c, with
+ * Authors:
+ *      Dave Airlie <airlied@redhat.com>
+ *      Rob Clark <rob.clark@linaro.org>
  */
 /** @file ttm_ref_object.c
  *
@@ -34,6 +40,7 @@
  * and release on file close.
  */
 
+
 /**
  * struct ttm_object_file
  *
@@ -84,6 +91,9 @@ struct ttm_object_device {
 	struct drm_open_hash object_hash;
 	atomic_t object_count;
 	struct ttm_mem_global *mem_glob;
+	struct dma_buf_ops ops;
+	void (*dmabuf_release)(struct dma_buf *dma_buf);
+	size_t dma_buf_size;
 };
 
 /**
@@ -116,6 +126,8 @@ struct ttm_ref_object {
 	struct ttm_object_file *tfile;
 };
 
+static void ttm_prime_dmabuf_release(struct dma_buf *dma_buf);
+
 static inline struct ttm_object_file *
 ttm_object_file_ref(struct ttm_object_file *tfile)
 {
@@ -416,9 +428,10 @@ out_err:
 }
 EXPORT_SYMBOL(ttm_object_file_init);
 
-struct ttm_object_device *ttm_object_device_init(struct ttm_mem_global
-						 *mem_glob,
-						 unsigned int hash_order)
+struct ttm_object_device *
+ttm_object_device_init(struct ttm_mem_global *mem_glob,
+		       unsigned int hash_order,
+		       const struct dma_buf_ops *ops)
 {
 	struct ttm_object_device *tdev = kmalloc(sizeof(*tdev), GFP_KERNEL);
 	int ret;
@@ -430,10 +443,17 @@ struct ttm_object_device *ttm_object_device_init(struct ttm_mem_global
 	spin_lock_init(&tdev->object_lock);
 	atomic_set(&tdev->object_count, 0);
 	ret = drm_ht_create(&tdev->object_hash, hash_order);
+	if (ret != 0)
+		goto out_no_object_hash;
 
-	if (likely(ret == 0))
-		return tdev;
+	tdev->ops = *ops;
+	tdev->dmabuf_release = tdev->ops.release;
+	tdev->ops.release = ttm_prime_dmabuf_release;
+	tdev->dma_buf_size = ttm_round_pot(sizeof(struct dma_buf)) +
+		ttm_round_pot(sizeof(struct file));
+	return tdev;
 
+out_no_object_hash:
 	kfree(tdev);
 	return NULL;
 }
@@ -452,3 +472,225 @@ void ttm_object_device_release(struct ttm_object_device **p_tdev)
 	kfree(tdev);
 }
 EXPORT_SYMBOL(ttm_object_device_release);
+
+/**
+ * get_dma_buf_unless_doomed - get a dma_buf reference if possible.
+ *
+ * @dma_buf: Non-refcounted pointer to a struct dma-buf.
+ *
+ * Obtain a file reference from a lookup structure that doesn't refcount
+ * the file, but synchronizes with its release method to make sure it has
+ * not been freed yet. See for example kref_get_unless_zero documentation.
+ * Returns true if refcounting succeeds, false otherwise.
+ *
+ * Nobody really wants this as a public API yet, so let it mature here
+ * for some time...
+ */
+static bool __must_check get_dma_buf_unless_doomed(struct dma_buf *dmabuf)
+{
+	return atomic_long_inc_not_zero(&dmabuf->file->f_count) != 0L;
+}
+
+/**
+ * ttm_prime_refcount_release - refcount release method for a prime object.
+ *
+ * @p_base: Pointer to ttm_base_object pointer.
+ *
+ * This is a wrapper that calls the refcount_release founction of the
+ * underlying object. At the same time it cleans up the prime object.
+ * This function is called when all references to the base object we
+ * derive from are gone.
+ */
+static void ttm_prime_refcount_release(struct ttm_base_object **p_base)
+{
+	struct ttm_base_object *base = *p_base;
+	struct ttm_prime_object *prime;
+
+	*p_base = NULL;
+	prime = container_of(base, struct ttm_prime_object, base);
+	BUG_ON(prime->dma_buf != NULL);
+	mutex_destroy(&prime->mutex);
+	if (prime->refcount_release)
+		prime->refcount_release(&base);
+}
+
+/**
+ * ttm_prime_dmabuf_release - Release method for the dma-bufs we export
+ *
+ * @dma_buf:
+ *
+ * This function first calls the dma_buf release method the driver
+ * provides. Then it cleans up our dma_buf pointer used for lookup,
+ * and finally releases the reference the dma_buf has on our base
+ * object.
+ */
+static void ttm_prime_dmabuf_release(struct dma_buf *dma_buf)
+{
+	struct ttm_prime_object *prime =
+		(struct ttm_prime_object *) dma_buf->priv;
+	struct ttm_base_object *base = &prime->base;
+	struct ttm_object_device *tdev = base->tfile->tdev;
+
+	if (tdev->dmabuf_release)
+		tdev->dmabuf_release(dma_buf);
+	mutex_lock(&prime->mutex);
+	if (prime->dma_buf == dma_buf)
+		prime->dma_buf = NULL;
+	mutex_unlock(&prime->mutex);
+	ttm_mem_global_free(tdev->mem_glob, tdev->dma_buf_size);
+	ttm_base_object_unref(&base);
+}
+
+/**
+ * ttm_prime_fd_to_handle - Get a base object handle from a prime fd
+ *
+ * @tfile: A struct ttm_object_file identifying the caller.
+ * @fd: The prime / dmabuf fd.
+ * @handle: The returned handle.
+ *
+ * This function returns a handle to an object that previously exported
+ * a dma-buf. Note that we don't handle imports yet, because we simply
+ * have no consumers of that implementation.
+ */
+int ttm_prime_fd_to_handle(struct ttm_object_file *tfile,
+			   int fd, u32 *handle)
+{
+	struct ttm_object_device *tdev = tfile->tdev;
+	struct dma_buf *dma_buf;
+	struct ttm_prime_object *prime;
+	struct ttm_base_object *base;
+	int ret;
+
+	dma_buf = dma_buf_get(fd);
+	if (IS_ERR(dma_buf))
+		return PTR_ERR(dma_buf);
+
+	if (dma_buf->ops != &tdev->ops)
+		return -ENOSYS;
+
+	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);
+
+	dma_buf_put(dma_buf);
+
+	return ret;
+}
+EXPORT_SYMBOL_GPL(ttm_prime_fd_to_handle);
+
+/**
+ * ttm_prime_handle_to_fd - Return a dma_buf fd from a ttm prime object
+ *
+ * @tfile: Struct ttm_object_file identifying the caller.
+ * @handle: Handle to the object we're exporting from.
+ * @flags: flags for dma-buf creation. We just pass them on.
+ * @prime_fd: The returned file descriptor.
+ *
+ */
+int ttm_prime_handle_to_fd(struct ttm_object_file *tfile,
+			   uint32_t handle, uint32_t flags,
+			   int *prime_fd)
+{
+	struct ttm_object_device *tdev = tfile->tdev;
+	struct ttm_base_object *base;
+	struct dma_buf *dma_buf;
+	struct ttm_prime_object *prime;
+	int ret;
+
+	base = ttm_base_object_lookup(tfile, handle);
+	if (unlikely(base == NULL ||
+		     base->object_type != ttm_prime_type)) {
+		ret = -ENOENT;
+		goto out_unref;
+	}
+
+	prime = container_of(base, struct ttm_prime_object, base);
+	if (unlikely(!base->shareable)) {
+		ret = -EPERM;
+		goto out_unref;
+	}
+
+	ret = mutex_lock_interruptible(&prime->mutex);
+	if (unlikely(ret != 0)) {
+		ret = -ERESTARTSYS;
+		goto out_unref;
+	}
+
+	dma_buf = prime->dma_buf;
+	if (!dma_buf || !get_dma_buf_unless_doomed(dma_buf)) {
+
+		/*
+		 * Need to create a new dma_buf, with memory accounting.
+		 */
+		ret = ttm_mem_global_alloc(tdev->mem_glob, tdev->dma_buf_size,
+					   false, true);
+		if (unlikely(ret != 0)) {
+			mutex_unlock(&prime->mutex);
+			goto out_unref;
+		}
+
+		dma_buf = dma_buf_export(prime, &tdev->ops,
+					 prime->size, flags);
+		if (IS_ERR(dma_buf)) {
+			ret = PTR_ERR(dma_buf);
+			ttm_mem_global_free(tdev->mem_glob,
+					    tdev->dma_buf_size);
+			mutex_unlock(&prime->mutex);
+			goto out_unref;
+		}
+
+		/*
+		 * dma_buf has taken the base object reference
+		 */
+		base = NULL;
+		prime->dma_buf = dma_buf;
+	}
+	mutex_unlock(&prime->mutex);
+
+	ret = dma_buf_fd(dma_buf, flags);
+	if (ret >= 0) {
+		*prime_fd = ret;
+		ret = 0;
+	} else
+		dma_buf_put(dma_buf);
+
+out_unref:
+	if (base)
+		ttm_base_object_unref(&base);
+	return ret;
+}
+EXPORT_SYMBOL_GPL(ttm_prime_handle_to_fd);
+
+/**
+ * ttm_prime_object_init - Initialize a ttm_prime_object
+ *
+ * @tfile: struct ttm_object_file identifying the caller
+ * @size: The size of the dma_bufs we export.
+ * @prime: The object to be initialized.
+ * @shareable: See ttm_base_object_init
+ * @type: See ttm_base_object_init
+ * @refcount_release: See ttm_base_object_init
+ * @ref_obj_release: See ttm_base_object_init
+ *
+ * Initializes an object which is compatible with the drm_prime model
+ * for data sharing between processes and devices.
+ */
+int ttm_prime_object_init(struct ttm_object_file *tfile, size_t size,
+			  struct ttm_prime_object *prime, bool shareable,
+			  enum ttm_object_type type,
+			  void (*refcount_release) (struct ttm_base_object **),
+			  void (*ref_obj_release) (struct ttm_base_object *,
+						   enum ttm_ref_type ref_type))
+{
+	mutex_init(&prime->mutex);
+	prime->size = PAGE_ALIGN(size);
+	prime->real_type = type;
+	prime->dma_buf = NULL;
+	prime->refcount_release = refcount_release;
+	return ttm_base_object_init(tfile, &prime->base, shareable,
+				    ttm_prime_type,
+				    ttm_prime_refcount_release,
+				    ref_obj_release);
+}
+EXPORT_SYMBOL(ttm_prime_object_init);
diff --git a/drivers/gpu/drm/vmwgfx/Makefile b/drivers/gpu/drm/vmwgfx/Makefile
index 2cc6cd91ac1..9f8b690bcf5 100644
--- a/drivers/gpu/drm/vmwgfx/Makefile
+++ b/drivers/gpu/drm/vmwgfx/Makefile
@@ -6,6 +6,6 @@ vmwgfx-y := vmwgfx_execbuf.o vmwgfx_gmr.o vmwgfx_kms.o vmwgfx_drv.o \
 	    vmwgfx_fifo.o vmwgfx_irq.o vmwgfx_ldu.o vmwgfx_ttm_glue.o \
 	    vmwgfx_overlay.o vmwgfx_marker.o vmwgfx_gmrid_manager.o \
 	    vmwgfx_fence.o vmwgfx_dmabuf.o vmwgfx_scrn.o vmwgfx_context.o \
-	    vmwgfx_surface.o
+	    vmwgfx_surface.o vmwgfx_prime.o
 
 obj-$(CONFIG_DRM_VMWGFX) := vmwgfx.o
diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c
index 20d5485eaf9..c7a549694e5 100644
--- a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c
+++ b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c
@@ -677,7 +677,7 @@ static int vmw_driver_load(struct drm_device *dev, unsigned long chipset)
 	}
 
 	dev_priv->tdev = ttm_object_device_init
-	    (dev_priv->mem_global_ref.object, 12);
+		(dev_priv->mem_global_ref.object, 12, &vmw_prime_dmabuf_ops);
 
 	if (unlikely(dev_priv->tdev == NULL)) {
 		DRM_ERROR("Unable to initialize TTM object management.\n");
@@ -1210,7 +1210,7 @@ static const struct file_operations vmwgfx_driver_fops = {
 
 static struct drm_driver driver = {
 	.driver_features = DRIVER_HAVE_IRQ | DRIVER_IRQ_SHARED |
-	DRIVER_MODESET,
+	DRIVER_MODESET | DRIVER_PRIME,
 	.load = vmw_driver_load,
 	.unload = vmw_driver_unload,
 	.lastclose = vmw_lastclose,
@@ -1235,6 +1235,9 @@ static struct drm_driver driver = {
 	.dumb_map_offset = vmw_dumb_map_offset,
 	.dumb_destroy = vmw_dumb_destroy,
 
+	.prime_fd_to_handle = vmw_prime_fd_to_handle,
+	.prime_handle_to_fd = vmw_prime_handle_to_fd,
+
 	.fops = &vmwgfx_driver_fops,
 	.name = VMWGFX_DRIVER_NAME,
 	.desc = VMWGFX_DRIVER_DESC,
diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.h b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.h
index e401d5dbcb9..db85985c708 100644
--- a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.h
+++ b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.h
@@ -819,6 +819,20 @@ int vmw_overlay_num_free_overlays(struct vmw_private *dev_priv);
 extern const struct ttm_mem_type_manager_func vmw_gmrid_manager_func;
 
 /**
+ * Prime - vmwgfx_prime.c
+ */
+
+extern const struct dma_buf_ops vmw_prime_dmabuf_ops;
+extern int vmw_prime_fd_to_handle(struct drm_device *dev,
+				  struct drm_file *file_priv,
+				  int fd, u32 *handle);
+extern int vmw_prime_handle_to_fd(struct drm_device *dev,
+				  struct drm_file *file_priv,
+				  uint32_t handle, uint32_t flags,
+				  int *prime_fd);
+
+
+/**
  * Inline helper functions
  */
 
diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_prime.c b/drivers/gpu/drm/vmwgfx/vmwgfx_prime.c
new file mode 100644
index 00000000000..31fe32d8d65
--- /dev/null
+++ b/drivers/gpu/drm/vmwgfx/vmwgfx_prime.c
@@ -0,0 +1,137 @@
+/**************************************************************************
+ *
+ * Copyright © 2013 VMware, Inc., Palo Alto, CA., USA
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ **************************************************************************/
+/*
+ * Authors:
+ *     Thomas Hellstrom <thellstrom@vmware.com>
+ *
+ */
+
+#include "vmwgfx_drv.h"
+#include <linux/dma-buf.h>
+#include <drm/ttm/ttm_object.h>
+
+/*
+ * DMA-BUF attach- and mapping methods. No need to implement
+ * these until we have other virtual devices use them.
+ */
+
+static int vmw_prime_map_attach(struct dma_buf *dma_buf,
+				struct device *target_dev,
+				struct dma_buf_attachment *attach)
+{
+	return -ENOSYS;
+}
+
+static void vmw_prime_map_detach(struct dma_buf *dma_buf,
+				 struct dma_buf_attachment *attach)
+{
+}
+
+static struct sg_table *vmw_prime_map_dma_buf(struct dma_buf_attachment *attach,
+					      enum dma_data_direction dir)
+{
+	return ERR_PTR(-ENOSYS);
+}
+
+static void vmw_prime_unmap_dma_buf(struct dma_buf_attachment *attach,
+				    struct sg_table *sgb,
+				    enum dma_data_direction dir)
+{
+}
+
+static void *vmw_prime_dmabuf_vmap(struct dma_buf *dma_buf)
+{
+	return NULL;
+}
+
+static void vmw_prime_dmabuf_vunmap(struct dma_buf *dma_buf, void *vaddr)
+{
+}
+
+static void *vmw_prime_dmabuf_kmap_atomic(struct dma_buf *dma_buf,
+		unsigned long page_num)
+{
+	return NULL;
+}
+
+static void vmw_prime_dmabuf_kunmap_atomic(struct dma_buf *dma_buf,
+		unsigned long page_num, void *addr)
+{
+
+}
+static void *vmw_prime_dmabuf_kmap(struct dma_buf *dma_buf,
+		unsigned long page_num)
+{
+	return NULL;
+}
+
+static void vmw_prime_dmabuf_kunmap(struct dma_buf *dma_buf,
+		unsigned long page_num, void *addr)
+{
+
+}
+
+static int vmw_prime_dmabuf_mmap(struct dma_buf *dma_buf,
+				 struct vm_area_struct *vma)
+{
+	WARN_ONCE(true, "Attempted use of dmabuf mmap. Bad.\n");
+	return -ENOSYS;
+}
+
+const struct dma_buf_ops vmw_prime_dmabuf_ops =  {
+	.attach = vmw_prime_map_attach,
+	.detach = vmw_prime_map_detach,
+	.map_dma_buf = vmw_prime_map_dma_buf,
+	.unmap_dma_buf = vmw_prime_unmap_dma_buf,
+	.release = NULL,
+	.kmap = vmw_prime_dmabuf_kmap,
+	.kmap_atomic = vmw_prime_dmabuf_kmap_atomic,
+	.kunmap = vmw_prime_dmabuf_kunmap,
+	.kunmap_atomic = vmw_prime_dmabuf_kunmap_atomic,
+	.mmap = vmw_prime_dmabuf_mmap,
+	.vmap = vmw_prime_dmabuf_vmap,
+	.vunmap = vmw_prime_dmabuf_vunmap,
+};
+
+int vmw_prime_fd_to_handle(struct drm_device *dev,
+			   struct drm_file *file_priv,
+			   int fd, u32 *handle)
+{
+	struct ttm_object_file *tfile = vmw_fpriv(file_priv)->tfile;
+
+	return ttm_prime_fd_to_handle(tfile, fd, handle);
+}
+
+int vmw_prime_handle_to_fd(struct drm_device *dev,
+			   struct drm_file *file_priv,
+			   uint32_t handle, uint32_t flags,
+			   int *prime_fd)
+{
+	struct ttm_object_file *tfile = vmw_fpriv(file_priv)->tfile;
+
+	return ttm_prime_handle_to_fd(tfile, handle, flags, prime_fd);
+}
diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_resource.c b/drivers/gpu/drm/vmwgfx/vmwgfx_resource.c
index 252501a54de..efe2b74c5eb 100644
--- a/drivers/gpu/drm/vmwgfx/vmwgfx_resource.c
+++ b/drivers/gpu/drm/vmwgfx/vmwgfx_resource.c
@@ -35,7 +35,7 @@
 #define VMW_RES_EVICT_ERR_COUNT 10
 
 struct vmw_user_dma_buffer {
-	struct ttm_base_object base;
+	struct ttm_prime_object prime;
 	struct vmw_dma_buffer dma;
 };
 
@@ -297,7 +297,7 @@ int vmw_user_resource_lookup_handle(struct vmw_private *dev_priv,
 	if (unlikely(base == NULL))
 		return -EINVAL;
 
-	if (unlikely(base->object_type != converter->object_type))
+	if (unlikely(ttm_base_object_type(base) != converter->object_type))
 		goto out_bad_resource;
 
 	res = converter->base_obj_to_res(base);
@@ -387,7 +387,7 @@ static void vmw_user_dmabuf_destroy(struct ttm_buffer_object *bo)
 {
 	struct vmw_user_dma_buffer *vmw_user_bo = vmw_user_dma_buffer(bo);
 
-	ttm_base_object_kfree(vmw_user_bo, base);
+	ttm_prime_object_kfree(vmw_user_bo, prime);
 }
 
 static void vmw_user_dmabuf_release(struct ttm_base_object **p_base)
@@ -401,7 +401,8 @@ static void vmw_user_dmabuf_release(struct ttm_base_object **p_base)
 	if (unlikely(base == NULL))
 		return;
 
-	vmw_user_bo = container_of(base, struct vmw_user_dma_buffer, base);
+	vmw_user_bo = container_of(base, struct vmw_user_dma_buffer,
+				   prime.base);
 	bo = &vmw_user_bo->dma.base;
 	ttm_bo_unref(&bo);
 }
@@ -442,18 +443,19 @@ int vmw_user_dmabuf_alloc(struct vmw_private *dev_priv,
 		return ret;
 
 	tmp = ttm_bo_reference(&user_bo->dma.base);
-	ret = ttm_base_object_init(tfile,
-				   &user_bo->base,
-				   shareable,
-				   ttm_buffer_type,
-				   &vmw_user_dmabuf_release, NULL);
+	ret = ttm_prime_object_init(tfile,
+				    size,
+				    &user_bo->prime,
+				    shareable,
+				    ttm_buffer_type,
+				    &vmw_user_dmabuf_release, NULL);
 	if (unlikely(ret != 0)) {
 		ttm_bo_unref(&tmp);
 		goto out_no_base_object;
 	}
 
 	*p_dma_buf = &user_bo->dma;
-	*handle = user_bo->base.hash.key;
+	*handle = user_bo->prime.base.hash.key;
 
 out_no_base_object:
 	return ret;
@@ -475,8 +477,8 @@ int vmw_user_dmabuf_verify_access(struct ttm_buffer_object *bo,
 		return -EPERM;
 
 	vmw_user_bo = vmw_user_dma_buffer(bo);
-	return (vmw_user_bo->base.tfile == tfile ||
-	vmw_user_bo->base.shareable) ? 0 : -EPERM;
+	return (vmw_user_bo->prime.base.tfile == tfile ||
+		vmw_user_bo->prime.base.shareable) ? 0 : -EPERM;
 }
 
 int vmw_dmabuf_alloc_ioctl(struct drm_device *dev, void *data,
@@ -538,14 +540,15 @@ int vmw_user_dmabuf_lookup(struct ttm_object_file *tfile,
 		return -ESRCH;
 	}
 
-	if (unlikely(base->object_type != ttm_buffer_type)) {
+	if (unlikely(ttm_base_object_type(base) != ttm_buffer_type)) {
 		ttm_base_object_unref(&base);
 		printk(KERN_ERR "Invalid buffer object handle 0x%08lx.\n",
 		       (unsigned long)handle);
 		return -EINVAL;
 	}
 
-	vmw_user_bo = container_of(base, struct vmw_user_dma_buffer, base);
+	vmw_user_bo = container_of(base, struct vmw_user_dma_buffer,
+				   prime.base);
 	(void)ttm_bo_reference(&vmw_user_bo->dma.base);
 	ttm_base_object_unref(&base);
 	*out = &vmw_user_bo->dma;
@@ -562,7 +565,8 @@ int vmw_user_dmabuf_reference(struct ttm_object_file *tfile,
 		return -EINVAL;
 
 	user_bo = container_of(dma_buf, struct vmw_user_dma_buffer, dma);
-	return ttm_ref_object_add(tfile, &user_bo->base, TTM_REF_USAGE, NULL);
+	return ttm_ref_object_add(tfile, &user_bo->prime.base,
+				  TTM_REF_USAGE, NULL);
 }
 
 /*
@@ -807,15 +811,16 @@ int vmw_dumb_create(struct drm_file *file_priv,
 		goto out_no_dmabuf;
 
 	tmp = ttm_bo_reference(&vmw_user_bo->dma.base);
-	ret = ttm_base_object_init(vmw_fpriv(file_priv)->tfile,
-				   &vmw_user_bo->base,
-				   false,
-				   ttm_buffer_type,
-				   &vmw_user_dmabuf_release, NULL);
+	ret = ttm_prime_object_init(vmw_fpriv(file_priv)->tfile,
+				    args->size,
+				    &vmw_user_bo->prime,
+				    false,
+				    ttm_buffer_type,
+				    &vmw_user_dmabuf_release, NULL);
 	if (unlikely(ret != 0))
 		goto out_no_base_object;
 
-	args->handle = vmw_user_bo->base.hash.key;
+	args->handle = vmw_user_bo->prime.base.hash.key;
 
 out_no_base_object:
 	ttm_bo_unref(&tmp);
@@ -994,7 +999,6 @@ void vmw_resource_unreserve(struct vmw_resource *res,
  */
 static int
 vmw_resource_check_buffer(struct vmw_resource *res,
-			  struct ww_acquire_ctx *ticket,
 			  bool interruptible,
 			  struct ttm_validate_buffer *val_buf)
 {
@@ -1011,7 +1015,7 @@ vmw_resource_check_buffer(struct vmw_resource *res,
 	INIT_LIST_HEAD(&val_list);
 	val_buf->bo = ttm_bo_reference(&res->backup->base);
 	list_add_tail(&val_buf->head, &val_list);
-	ret = ttm_eu_reserve_buffers(ticket, &val_list);
+	ret = ttm_eu_reserve_buffers(NULL, &val_list);
 	if (unlikely(ret != 0))
 		goto out_no_reserve;
 
@@ -1029,7 +1033,7 @@ vmw_resource_check_buffer(struct vmw_resource *res,
 	return 0;
 
 out_no_validate:
-	ttm_eu_backoff_reservation(ticket, &val_list);
+	ttm_eu_backoff_reservation(NULL, &val_list);
 out_no_reserve:
 	ttm_bo_unref(&val_buf->bo);
 	if (backup_dirty)
@@ -1074,8 +1078,7 @@ int vmw_resource_reserve(struct vmw_resource *res, bool no_backup)
  * @val_buf:        Backup buffer information.
  */
 static void
-vmw_resource_backoff_reservation(struct ww_acquire_ctx *ticket,
-				 struct ttm_validate_buffer *val_buf)
+vmw_resource_backoff_reservation(struct ttm_validate_buffer *val_buf)
 {
 	struct list_head val_list;
 
@@ -1084,7 +1087,7 @@ vmw_resource_backoff_reservation(struct ww_acquire_ctx *ticket,
 
 	INIT_LIST_HEAD(&val_list);
 	list_add_tail(&val_buf->head, &val_list);
-	ttm_eu_backoff_reservation(ticket, &val_list);
+	ttm_eu_backoff_reservation(NULL, &val_list);
 	ttm_bo_unref(&val_buf->bo);
 }
 
@@ -1099,14 +1102,12 @@ int vmw_resource_do_evict(struct vmw_resource *res, bool interruptible)
 {
 	struct ttm_validate_buffer val_buf;
 	const struct vmw_res_func *func = res->func;
-	struct ww_acquire_ctx ticket;
 	int ret;
 
 	BUG_ON(!func->may_evict);
 
 	val_buf.bo = NULL;
-	ret = vmw_resource_check_buffer(res, &ticket, interruptible,
-					&val_buf);
+	ret = vmw_resource_check_buffer(res, interruptible, &val_buf);
 	if (unlikely(ret != 0))
 		return ret;
 
@@ -1121,7 +1122,7 @@ int vmw_resource_do_evict(struct vmw_resource *res, bool interruptible)
 	res->backup_dirty = true;
 	res->res_dirty = false;
 out_no_unbind:
-	vmw_resource_backoff_reservation(&ticket, &val_buf);
+	vmw_resource_backoff_reservation(&val_buf);
 
 	return ret;
 }
diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_surface.c b/drivers/gpu/drm/vmwgfx/vmwgfx_surface.c
index 58281433974..7de2ea8bd55 100644
--- a/drivers/gpu/drm/vmwgfx/vmwgfx_surface.c
+++ b/drivers/gpu/drm/vmwgfx/vmwgfx_surface.c
@@ -38,7 +38,7 @@
  * @size:           TTM accounting size for the surface.
  */
 struct vmw_user_surface {
-	struct ttm_base_object base;
+	struct ttm_prime_object prime;
 	struct vmw_surface srf;
 	uint32_t size;
 	uint32_t backup_handle;
@@ -580,7 +580,8 @@ static int vmw_surface_init(struct vmw_private *dev_priv,
 static struct vmw_resource *
 vmw_user_surface_base_to_res(struct ttm_base_object *base)
 {
-	return &(container_of(base, struct vmw_user_surface, base)->srf.res);
+	return &(container_of(base, struct vmw_user_surface,
+			      prime.base)->srf.res);
 }
 
 /**
@@ -599,7 +600,7 @@ static void vmw_user_surface_free(struct vmw_resource *res)
 	kfree(srf->offsets);
 	kfree(srf->sizes);
 	kfree(srf->snooper.image);
-	ttm_base_object_kfree(user_srf, base);
+	ttm_prime_object_kfree(user_srf, prime);
 	ttm_mem_global_free(vmw_mem_glob(dev_priv), size);
 }
 
@@ -616,7 +617,7 @@ static void vmw_user_surface_base_release(struct ttm_base_object **p_base)
 {
 	struct ttm_base_object *base = *p_base;
 	struct vmw_user_surface *user_srf =
-	    container_of(base, struct vmw_user_surface, base);
+	    container_of(base, struct vmw_user_surface, prime.base);
 	struct vmw_resource *res = &user_srf->srf.res;
 
 	*p_base = NULL;
@@ -790,8 +791,8 @@ int vmw_surface_define_ioctl(struct drm_device *dev, void *data,
 	}
 	srf->snooper.crtc = NULL;
 
-	user_srf->base.shareable = false;
-	user_srf->base.tfile = NULL;
+	user_srf->prime.base.shareable = false;
+	user_srf->prime.base.tfile = NULL;
 
 	/**
 	 * From this point, the generic resource management functions
@@ -803,9 +804,9 @@ int vmw_surface_define_ioctl(struct drm_device *dev, void *data,
 		goto out_unlock;
 
 	tmp = vmw_resource_reference(&srf->res);
-	ret = ttm_base_object_init(tfile, &user_srf->base,
-				   req->shareable, VMW_RES_SURFACE,
-				   &vmw_user_surface_base_release, NULL);
+	ret = ttm_prime_object_init(tfile, res->backup_size, &user_srf->prime,
+				    req->shareable, VMW_RES_SURFACE,
+				    &vmw_user_surface_base_release, NULL);
 
 	if (unlikely(ret != 0)) {
 		vmw_resource_unreference(&tmp);
@@ -813,7 +814,7 @@ int vmw_surface_define_ioctl(struct drm_device *dev, void *data,
 		goto out_unlock;
 	}
 
-	rep->sid = user_srf->base.hash.key;
+	rep->sid = user_srf->prime.base.hash.key;
 	vmw_resource_unreference(&res);
 
 	ttm_read_unlock(&vmaster->lock);
@@ -823,7 +824,7 @@ out_no_copy:
 out_no_offsets:
 	kfree(srf->sizes);
 out_no_sizes:
-	ttm_base_object_kfree(user_srf, base);
+	ttm_prime_object_kfree(user_srf, prime);
 out_no_user_srf:
 	ttm_mem_global_free(vmw_mem_glob(dev_priv), size);
 out_unlock:
@@ -859,13 +860,14 @@ int vmw_surface_reference_ioctl(struct drm_device *dev, void *data,
 		return -EINVAL;
 	}
 
-	if (unlikely(base->object_type != VMW_RES_SURFACE))
+	if (unlikely(ttm_base_object_type(base) != VMW_RES_SURFACE))
 		goto out_bad_resource;
 
-	user_srf = container_of(base, struct vmw_user_surface, base);
+	user_srf = container_of(base, struct vmw_user_surface, prime.base);
 	srf = &user_srf->srf;
 
-	ret = ttm_ref_object_add(tfile, &user_srf->base, TTM_REF_USAGE, NULL);
+	ret = ttm_ref_object_add(tfile, &user_srf->prime.base,
+				 TTM_REF_USAGE, NULL);
 	if (unlikely(ret != 0)) {
 		DRM_ERROR("Could not add a reference to a surface.\n");
 		goto out_no_reference;
diff --git a/include/drm/ttm/ttm_bo_api.h b/include/drm/ttm/ttm_bo_api.h
index 751eaffbf0d..ee127ec33c6 100644
--- a/include/drm/ttm/ttm_bo_api.h
+++ b/include/drm/ttm/ttm_bo_api.h
@@ -169,6 +169,7 @@ struct ttm_tt;
  * @offset: The current GPU offset, which can have different meanings
  * depending on the memory type. For SYSTEM type memory, it should be 0.
  * @cur_placement: Hint of current placement.
+ * @wu_mutex: Wait unreserved mutex.
  *
  * Base class for TTM buffer object, that deals with data placement and CPU
  * mappings. GPU mappings are really up to the driver, but for simpler GPUs
@@ -250,6 +251,7 @@ struct ttm_buffer_object {
 
 	struct reservation_object *resv;
 	struct reservation_object ttm_resv;
+	struct mutex wu_mutex;
 };
 
 /**
@@ -702,5 +704,5 @@ extern ssize_t ttm_bo_io(struct ttm_bo_device *bdev, struct file *filp,
 			 size_t count, loff_t *f_pos, bool write);
 
 extern void ttm_bo_swapout_all(struct ttm_bo_device *bdev);
-
+extern int ttm_bo_wait_unreserved(struct ttm_buffer_object *bo);
 #endif
diff --git a/include/drm/ttm/ttm_execbuf_util.h b/include/drm/ttm/ttm_execbuf_util.h
index ec8a1d30651..16db7d01a33 100644
--- a/include/drm/ttm/ttm_execbuf_util.h
+++ b/include/drm/ttm/ttm_execbuf_util.h
@@ -70,7 +70,8 @@ extern void ttm_eu_backoff_reservation(struct ww_acquire_ctx *ticket,
 /**
  * function ttm_eu_reserve_buffers
  *
- * @ticket:  [out] ww_acquire_ctx returned by call.
+ * @ticket:  [out] ww_acquire_ctx filled in by call, or NULL if only
+ *           non-blocking reserves should be tried.
  * @list:    thread private list of ttm_validate_buffer structs.
  *
  * Tries to reserve bos pointed to by the list entries for validation.
diff --git a/include/drm/ttm/ttm_object.h b/include/drm/ttm/ttm_object.h
index fc0cf064990..58b029894eb 100644
--- a/include/drm/ttm/ttm_object.h
+++ b/include/drm/ttm/ttm_object.h
@@ -41,6 +41,7 @@
 #include <drm/drm_hashtab.h>
 #include <linux/kref.h>
 #include <linux/rcupdate.h>
+#include <linux/dma-buf.h>
 #include <ttm/ttm_memory.h>
 
 /**
@@ -77,6 +78,7 @@ enum ttm_object_type {
 	ttm_fence_type,
 	ttm_buffer_type,
 	ttm_lock_type,
+	ttm_prime_type,
 	ttm_driver_type0 = 256,
 	ttm_driver_type1,
 	ttm_driver_type2,
@@ -132,6 +134,30 @@ struct ttm_base_object {
 				 enum ttm_ref_type ref_type);
 };
 
+
+/**
+ * struct ttm_prime_object - Modified base object that is prime-aware
+ *
+ * @base: struct ttm_base_object that we derive from
+ * @mutex: Mutex protecting the @dma_buf member.
+ * @size: Size of the dma_buf associated with this object
+ * @real_type: Type of the underlying object. Needed since we're setting
+ * the value of @base::object_type to ttm_prime_type
+ * @dma_buf: Non ref-coutned pointer to a struct dma_buf created from this
+ * object.
+ * @refcount_release: The underlying object's release method. Needed since
+ * we set @base::refcount_release to our own release method.
+ */
+
+struct ttm_prime_object {
+	struct ttm_base_object base;
+	struct mutex mutex;
+	size_t size;
+	enum ttm_object_type real_type;
+	struct dma_buf *dma_buf;
+	void (*refcount_release) (struct ttm_base_object **);
+};
+
 /**
  * ttm_base_object_init
  *
@@ -248,14 +274,18 @@ extern void ttm_object_file_release(struct ttm_object_file **p_tfile);
 /**
  * ttm_object device init - initialize a struct ttm_object_device
  *
+ * @mem_glob: struct ttm_mem_global for memory accounting.
  * @hash_order: Order of hash table used to hash the base objects.
+ * @ops: DMA buf ops for prime objects of this device.
  *
  * This function is typically called on device initialization to prepare
  * data structures needed for ttm base and ref objects.
  */
 
-extern struct ttm_object_device *ttm_object_device_init
-    (struct ttm_mem_global *mem_glob, unsigned int hash_order);
+extern struct ttm_object_device *
+ttm_object_device_init(struct ttm_mem_global *mem_glob,
+		       unsigned int hash_order,
+		       const struct dma_buf_ops *ops);
 
 /**
  * ttm_object_device_release - release data held by a ttm_object_device
@@ -272,4 +302,31 @@ extern void ttm_object_device_release(struct ttm_object_device **p_tdev);
 
 #define ttm_base_object_kfree(__object, __base)\
 	kfree_rcu(__object, __base.rhead)
+
+extern int ttm_prime_object_init(struct ttm_object_file *tfile,
+				 size_t size,
+				 struct ttm_prime_object *prime,
+				 bool shareable,
+				 enum ttm_object_type type,
+				 void (*refcount_release)
+				 (struct ttm_base_object **),
+				 void (*ref_obj_release)
+				 (struct ttm_base_object *,
+				  enum ttm_ref_type ref_type));
+
+static inline enum ttm_object_type
+ttm_base_object_type(struct ttm_base_object *base)
+{
+	return (base->object_type == ttm_prime_type) ?
+		container_of(base, struct ttm_prime_object, base)->real_type :
+		base->object_type;
+}
+extern int ttm_prime_fd_to_handle(struct ttm_object_file *tfile,
+				  int fd, u32 *handle);
+extern int ttm_prime_handle_to_fd(struct ttm_object_file *tfile,
+				  uint32_t handle, uint32_t flags,
+				  int *prime_fd);
+
+#define ttm_prime_object_kfree(__obj, __prime)		\
+	kfree_rcu(__obj, __prime.base.rhead)
 #endif
diff --git a/include/uapi/drm/radeon_drm.h b/include/uapi/drm/radeon_drm.h
index 46d41e8b0dc..2f3f7ea8c77 100644
--- a/include/uapi/drm/radeon_drm.h
+++ b/include/uapi/drm/radeon_drm.h
@@ -981,6 +981,8 @@ struct drm_radeon_cs {
 #define RADEON_INFO_SI_TILE_MODE_ARRAY	0x16
 /* query if CP DMA is supported on the compute ring */
 #define RADEON_INFO_SI_CP_DMA_COMPUTE	0x17
+/* CIK macrotile mode array */
+#define RADEON_INFO_CIK_MACROTILE_MODE_ARRAY	0x18
 
 
 struct drm_radeon_info {