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

Pull drm fixes from Dave Airlie:
 "Xmas fixes pull:

  core:
      one atomic fix, revert the WARN_ON dumb buffers patch.

  agp:
      fixup Dave J.

  nouveau:
      fix 3.18 regression for old userspace

  tegra fixes:
      vblank and iommu fixes

  amdkfd:
      fix bugs shown by testing with userspace, init apertures once

  msm:
      hdmi fixes and cleanup

  i915:
      misc fixes

  There is also a link ordering fix that I've asked to be cc'ed to you,
  putting iommu before gpu, it fixes an issue with amdkfd when things
  are all in the kernel, but I didn't like sending it via my tree
  without discussion.

  I'll probably be a bit on/off for a few weeks with pulls now, due to
  holidays and LCA, so don't be surprised if stuff gets a bit backed up,
  and things end up a bit large due to lag"

* 'drm-fixes' of git://people.freedesktop.org/~airlied/linux: (28 commits)
  Revert "drm/gem: Warn on illegal use of the dumb buffer interface v2"
  agp: Fix up email address & attributions in AGP MODULE_AUTHOR tags
  nouveau: bring back legacy mmap handler
  drm/msm/hdmi: rework HDMI IRQ handler
  drm/msm/hdmi: enable regulators before clocks to avoid warnings
  drm/msm/mdp5: update irqs on crtc<->encoder link change
  drm/msm: block incoming update on pending updates
  drm/atomic: fix potential null ptr on plane enable
  drm/msm: Deletion of unnecessary checks before the function call "release_firmware"
  drm/msm: Deletion of unnecessary checks before two function calls
  drm/tegra: dc: Select root window for event dispatch
  drm/tegra: gem: Use the proper size for GEM objects
  drm/tegra: gem: Flush buffer objects upon allocation
  drm/tegra: dc: Fix a potential race on page-flip completion
  drm/tegra: dc: Consistently use the same pipe
  drm/irq: Add drm_crtc_vblank_count()
  drm/irq: Add drm_crtc_handle_vblank()
  drm/irq: Add drm_crtc_send_vblank_event()
  drm/i915: Disable PSMI sleep messages on all rings around context switches
  drm/i915: Force the CS stall for invalidate flushes
  ...

drivers/char/agp/ali-agp.c

@@ -417,6 +417,6 @@ static void __exit agp_ali_cleanup(void)
 module_init(agp_ali_init);
 module_exit(agp_ali_cleanup);
 
-MODULE_AUTHOR("Dave Jones <davej@redhat.com>");
+MODULE_AUTHOR("Dave Jones");
 MODULE_LICENSE("GPL and additional rights");
 

drivers/char/agp/amd64-agp.c

@@ -813,6 +813,6 @@ static void __exit agp_amd64_cleanup(void)
 module_init(agp_amd64_mod_init);
 module_exit(agp_amd64_cleanup);
 
-MODULE_AUTHOR("Dave Jones <davej@redhat.com>, Andi Kleen");
+MODULE_AUTHOR("Dave Jones, Andi Kleen");
 module_param(agp_try_unsupported, bool, 0);
 MODULE_LICENSE("GPL");

drivers/char/agp/ati-agp.c

@@ -579,6 +579,6 @@ static void __exit agp_ati_cleanup(void)
 module_init(agp_ati_init);
 module_exit(agp_ati_cleanup);
 
-MODULE_AUTHOR("Dave Jones <davej@redhat.com>");
+MODULE_AUTHOR("Dave Jones");
 MODULE_LICENSE("GPL and additional rights");
 

drivers/char/agp/backend.c

@@ -356,7 +356,7 @@ static __init int agp_setup(char *s)
 __setup("agp=", agp_setup);
 #endif
 
-MODULE_AUTHOR("Dave Jones <davej@redhat.com>");
+MODULE_AUTHOR("Dave Jones, Jeff Hartmann");
 MODULE_DESCRIPTION("AGP GART driver");
 MODULE_LICENSE("GPL and additional rights");
 MODULE_ALIAS_MISCDEV(AGPGART_MINOR);

drivers/char/agp/intel-agp.c

@@ -920,5 +920,5 @@ static void __exit agp_intel_cleanup(void)
 module_init(agp_intel_init);
 module_exit(agp_intel_cleanup);
 
-MODULE_AUTHOR("Dave Jones <davej@redhat.com>");
+MODULE_AUTHOR("Dave Jones, Various @Intel");
 MODULE_LICENSE("GPL and additional rights");

drivers/char/agp/intel-gtt.c

@@ -1438,5 +1438,5 @@ void intel_gmch_remove(void)
 }
 EXPORT_SYMBOL(intel_gmch_remove);
 
-MODULE_AUTHOR("Dave Jones <davej@redhat.com>");
+MODULE_AUTHOR("Dave Jones, Various @Intel");
 MODULE_LICENSE("GPL and additional rights");

drivers/char/agp/nvidia-agp.c

@@ -1,7 +1,7 @@
 /*
  * Nvidia AGPGART routines.
  * Based upon a 2.4 agpgart diff by the folks from NVIDIA, and hacked up
- * to work in 2.5 by Dave Jones <davej@redhat.com>
+ * to work in 2.5 by Dave Jones.
  */
 
 #include <linux/module.h>

drivers/char/agp/via-agp.c

@@ -595,4 +595,4 @@ module_init(agp_via_init);
 module_exit(agp_via_cleanup);
 
 MODULE_LICENSE("GPL");
-MODULE_AUTHOR("Dave Jones <davej@redhat.com>");
+MODULE_AUTHOR("Dave Jones");

drivers/gpu/drm/amd/amdkfd/kfd_chardev.c

@@ -121,13 +121,9 @@ static int kfd_open(struct inode *inode, struct file *filep)
 	if (IS_ERR(process))
 		return PTR_ERR(process);
 
-	process->is_32bit_user_mode = is_32bit_user_mode;
-
 	dev_dbg(kfd_device, "process %d opened, compat mode (32 bit) - %d\n",
 		process->pasid, process->is_32bit_user_mode);
 
-	kfd_init_apertures(process);
-
 	return 0;
 }
 

drivers/gpu/drm/amd/amdkfd/kfd_flat_memory.c

@@ -299,13 +299,13 @@ int kfd_init_apertures(struct kfd_process *process)
 	struct kfd_dev *dev;
 	struct kfd_process_device *pdd;
 
-	mutex_lock(&process->mutex);
-
 	/*Iterating over all devices*/
 	while ((dev = kfd_topology_enum_kfd_devices(id)) != NULL &&
 		id < NUM_OF_SUPPORTED_GPUS) {
 
 		pdd = kfd_get_process_device_data(dev, process, 1);
+		if (!pdd)
+			return -1;
 
 		/*
 		 * For 64 bit process aperture will be statically reserved in
@@ -348,8 +348,6 @@ int kfd_init_apertures(struct kfd_process *process)
 		id++;
 	}
 
-	mutex_unlock(&process->mutex);
-
 	return 0;
 }
 

drivers/gpu/drm/amd/amdkfd/kfd_process.c

@@ -26,6 +26,8 @@
 #include <linux/slab.h>
 #include <linux/amd-iommu.h>
 #include <linux/notifier.h>
+#include <linux/compat.h>
+
 struct mm_struct;
 
 #include "kfd_priv.h"
@@ -285,8 +287,15 @@ static struct kfd_process *create_process(const struct task_struct *thread)
 	if (err != 0)
 		goto err_process_pqm_init;
 
+	/* init process apertures*/
+	process->is_32bit_user_mode = is_compat_task();
+	if (kfd_init_apertures(process) != 0)
+		goto err_init_apretures;
+
 	return process;
 
+err_init_apretures:
+	pqm_uninit(&process->pqm);
 err_process_pqm_init:
 	hash_del_rcu(&process->kfd_processes);
 	synchronize_rcu();

drivers/gpu/drm/amd/amdkfd/kfd_topology.c

@@ -700,8 +700,6 @@ static ssize_t node_show(struct kobject *kobj, struct attribute *attr,
 				dev->node_props.simd_per_cu);
 		sysfs_show_32bit_prop(buffer, "max_slots_scratch_cu",
 				dev->node_props.max_slots_scratch_cu);
-		sysfs_show_32bit_prop(buffer, "engine_id",
-				dev->node_props.engine_id);
 		sysfs_show_32bit_prop(buffer, "vendor_id",
 				dev->node_props.vendor_id);
 		sysfs_show_32bit_prop(buffer, "device_id",
@@ -715,6 +713,12 @@ static ssize_t node_show(struct kobject *kobj, struct attribute *attr,
 						dev->gpu->kgd));
 			sysfs_show_64bit_prop(buffer, "local_mem_size",
 					kfd2kgd->get_vmem_size(dev->gpu->kgd));
+
+			sysfs_show_32bit_prop(buffer, "fw_version",
+					kfd2kgd->get_fw_version(
+							dev->gpu->kgd,
+							KGD_ENGINE_MEC1));
+
 		}
 
 		ret = sysfs_show_32bit_prop(buffer, "max_engine_clk_ccompute",

drivers/gpu/drm/amd/include/kgd_kfd_interface.h

@@ -45,6 +45,17 @@ enum kgd_memory_pool {
 	KGD_POOL_FRAMEBUFFER = 3,
 };
 
+enum kgd_engine_type {
+	KGD_ENGINE_PFP = 1,
+	KGD_ENGINE_ME,
+	KGD_ENGINE_CE,
+	KGD_ENGINE_MEC1,
+	KGD_ENGINE_MEC2,
+	KGD_ENGINE_RLC,
+	KGD_ENGINE_SDMA,
+	KGD_ENGINE_MAX
+};
+
 struct kgd2kfd_shared_resources {
 	/* Bit n == 1 means VMID n is available for KFD. */
 	unsigned int compute_vmid_bitmap;
@@ -137,6 +148,8 @@ struct kgd2kfd_calls {
  *
  * @hqd_destroy: Destructs and preempts the queue assigned to that hqd slot.
  *
+ * @get_fw_version: Returns FW versions from the header
+ *
  * This structure contains function pointers to services that the kgd driver
  * provides to amdkfd driver.
  *
@@ -176,6 +189,8 @@ struct kfd2kgd_calls {
 	int (*hqd_destroy)(struct kgd_dev *kgd, uint32_t reset_type,
 				unsigned int timeout, uint32_t pipe_id,
 				uint32_t queue_id);
+	uint16_t (*get_fw_version)(struct kgd_dev *kgd,
+				enum kgd_engine_type type);
 };
 
 bool kgd2kfd_init(unsigned interface_version,

drivers/gpu/drm/drm_atomic_helper.c

@@ -61,7 +61,7 @@ drm_atomic_helper_plane_changed(struct drm_atomic_state *state,
 	struct drm_crtc_state *crtc_state;
 
 	if (plane->state->crtc) {
-		crtc_state = state->crtc_states[drm_crtc_index(plane->crtc)];
+		crtc_state = state->crtc_states[drm_crtc_index(plane->state->crtc)];
 
 		if (WARN_ON(!crtc_state))
 			return;

drivers/gpu/drm/drm_irq.c

@@ -830,6 +830,8 @@ drm_get_last_vbltimestamp(struct drm_device *dev, int crtc,
  * vblank events since the system was booted, including lost events due to
  * modesetting activity.
  *
+ * This is the legacy version of drm_crtc_vblank_count().
+ *
  * Returns:
  * The software vblank counter.
  */
@@ -843,6 +845,25 @@ u32 drm_vblank_count(struct drm_device *dev, int crtc)
 }
 EXPORT_SYMBOL(drm_vblank_count);
 
+/**
+ * drm_crtc_vblank_count - retrieve "cooked" vblank counter value
+ * @crtc: which counter to retrieve
+ *
+ * Fetches the "cooked" vblank count value that represents the number of
+ * vblank events since the system was booted, including lost events due to
+ * modesetting activity.
+ *
+ * This is the native KMS version of drm_vblank_count().
+ *
+ * Returns:
+ * The software vblank counter.
+ */
+u32 drm_crtc_vblank_count(struct drm_crtc *crtc)
+{
+	return drm_vblank_count(crtc->dev, drm_crtc_index(crtc));
+}
+EXPORT_SYMBOL(drm_crtc_vblank_count);
+
 /**
  * drm_vblank_count_and_time - retrieve "cooked" vblank counter value
  * and the system timestamp corresponding to that vblank counter value.
@@ -904,6 +925,8 @@ static void send_vblank_event(struct drm_device *dev,
  *
  * Updates sequence # and timestamp on event, and sends it to userspace.
  * Caller must hold event lock.
+ *
+ * This is the legacy version of drm_crtc_send_vblank_event().
  */
 void drm_send_vblank_event(struct drm_device *dev, int crtc,
 		struct drm_pending_vblank_event *e)
@@ -922,6 +945,23 @@ void drm_send_vblank_event(struct drm_device *dev, int crtc,
 }
 EXPORT_SYMBOL(drm_send_vblank_event);
 
+/**
+ * drm_crtc_send_vblank_event - helper to send vblank event after pageflip
+ * @crtc: the source CRTC of the vblank event
+ * @e: the event to send
+ *
+ * Updates sequence # and timestamp on event, and sends it to userspace.
+ * Caller must hold event lock.
+ *
+ * This is the native KMS version of drm_send_vblank_event().
+ */
+void drm_crtc_send_vblank_event(struct drm_crtc *crtc,
+				struct drm_pending_vblank_event *e)
+{
+	drm_send_vblank_event(crtc->dev, drm_crtc_index(crtc), e);
+}
+EXPORT_SYMBOL(drm_crtc_send_vblank_event);
+
 /**
  * drm_vblank_enable - enable the vblank interrupt on a CRTC
  * @dev: DRM device
@@ -1594,6 +1634,8 @@ static void drm_handle_vblank_events(struct drm_device *dev, int crtc)
  *
  * Drivers should call this routine in their vblank interrupt handlers to
  * update the vblank counter and send any signals that may be pending.
+ *
+ * This is the legacy version of drm_crtc_handle_vblank().
  */
 bool drm_handle_vblank(struct drm_device *dev, int crtc)
 {
@@ -1670,3 +1712,21 @@ bool drm_handle_vblank(struct drm_device *dev, int crtc)
 	return true;
 }
 EXPORT_SYMBOL(drm_handle_vblank);
+
+/**
+ * drm_crtc_handle_vblank - handle a vblank event
+ * @crtc: where this event occurred
+ *
+ * Drivers should call this routine in their vblank interrupt handlers to
+ * update the vblank counter and send any signals that may be pending.
+ *
+ * This is the native KMS version of drm_handle_vblank().
+ *
+ * Returns:
+ * True if the event was successfully handled, false on failure.
+ */
+bool drm_crtc_handle_vblank(struct drm_crtc *crtc)
+{
+	return drm_handle_vblank(crtc->dev, drm_crtc_index(crtc));
+}
+EXPORT_SYMBOL(drm_crtc_handle_vblank);

drivers/gpu/drm/i915/i915_drv.c

@@ -811,6 +811,8 @@ int i915_reset(struct drm_device *dev)
 	if (!i915.reset)
 		return 0;
 
+	intel_reset_gt_powersave(dev);
+
 	mutex_lock(&dev->struct_mutex);
 
 	i915_gem_reset(dev);
@@ -880,7 +882,7 @@ int i915_reset(struct drm_device *dev)
 		 * of re-init after reset.
 		 */
 		if (INTEL_INFO(dev)->gen > 5)
-			intel_reset_gt_powersave(dev);
+			intel_enable_gt_powersave(dev);
 	} else {
 		mutex_unlock(&dev->struct_mutex);
 	}
@@ -1584,7 +1586,7 @@ static struct drm_driver driver = {
 	.gem_prime_import = i915_gem_prime_import,
 
 	.dumb_create = i915_gem_dumb_create,
-	.dumb_map_offset = i915_gem_dumb_map_offset,
+	.dumb_map_offset = i915_gem_mmap_gtt,
 	.dumb_destroy = drm_gem_dumb_destroy,
 	.ioctls = i915_ioctls,
 	.fops = &i915_driver_fops,

drivers/gpu/drm/i915/i915_drv.h

@@ -2501,9 +2501,8 @@ void i915_vma_move_to_active(struct i915_vma *vma,
 int i915_gem_dumb_create(struct drm_file *file_priv,
 			 struct drm_device *dev,
 			 struct drm_mode_create_dumb *args);
-int i915_gem_dumb_map_offset(struct drm_file *file_priv,
+int i915_gem_mmap_gtt(struct drm_file *file_priv, struct drm_device *dev,
-			     struct drm_device *dev, uint32_t handle,
+		      uint32_t handle, uint64_t *offset);
-			     uint64_t *offset);
 /**
  * Returns true if seq1 is later than seq2.
  */

drivers/gpu/drm/i915/i915_gem.c

@@ -401,7 +401,6 @@ static int
 i915_gem_create(struct drm_file *file,
 		struct drm_device *dev,
 		uint64_t size,
-		bool dumb,
 		uint32_t *handle_p)
 {
 	struct drm_i915_gem_object *obj;
@@ -417,7 +416,6 @@ i915_gem_create(struct drm_file *file,
 	if (obj == NULL)
 		return -ENOMEM;
 
-	obj->base.dumb = dumb;
 	ret = drm_gem_handle_create(file, &obj->base, &handle);
 	/* drop reference from allocate - handle holds it now */
 	drm_gem_object_unreference_unlocked(&obj->base);
@@ -437,7 +435,7 @@ i915_gem_dumb_create(struct drm_file *file,
 	args->pitch = ALIGN(args->width * DIV_ROUND_UP(args->bpp, 8), 64);
 	args->size = args->pitch * args->height;
 	return i915_gem_create(file, dev,
-			       args->size, true, &args->handle);
+			       args->size, &args->handle);
 }
 
 /**
@@ -450,7 +448,7 @@ i915_gem_create_ioctl(struct drm_device *dev, void *data,
 	struct drm_i915_gem_create *args = data;
 
 	return i915_gem_create(file, dev,
-			       args->size, false, &args->handle);
+			       args->size, &args->handle);
 }
 
 static inline int
@@ -1840,10 +1838,10 @@ static void i915_gem_object_free_mmap_offset(struct drm_i915_gem_object *obj)
 	drm_gem_free_mmap_offset(&obj->base);
 }
 
-static int
+int
 i915_gem_mmap_gtt(struct drm_file *file,
 		  struct drm_device *dev,
-		  uint32_t handle, bool dumb,
+		  uint32_t handle,
 		  uint64_t *offset)
 {
 	struct drm_i915_private *dev_priv = dev->dev_private;
@@ -1860,13 +1858,6 @@ i915_gem_mmap_gtt(struct drm_file *file,
 		goto unlock;
 	}
 
-	/*
-	 * We don't allow dumb mmaps on objects created using another
-	 * interface.
-	 */
-	WARN_ONCE(dumb && !(obj->base.dumb || obj->base.import_attach),
-		  "Illegal dumb map of accelerated buffer.\n");
-
 	if (obj->base.size > dev_priv->gtt.mappable_end) {
 		ret = -E2BIG;
 		goto out;
@@ -1891,15 +1882,6 @@ unlock:
 	return ret;
 }
 
-int
-i915_gem_dumb_map_offset(struct drm_file *file,
-			 struct drm_device *dev,
-			 uint32_t handle,
-			 uint64_t *offset)
-{
-	return i915_gem_mmap_gtt(file, dev, handle, true, offset);
-}
-
 /**
  * i915_gem_mmap_gtt_ioctl - prepare an object for GTT mmap'ing
  * @dev: DRM device
@@ -1921,7 +1903,7 @@ i915_gem_mmap_gtt_ioctl(struct drm_device *dev, void *data,
 {
 	struct drm_i915_gem_mmap_gtt *args = data;
 
-	return i915_gem_mmap_gtt(file, dev, args->handle, false, &args->offset);
+	return i915_gem_mmap_gtt(file, dev, args->handle, &args->offset);
 }
 
 static inline int

drivers/gpu/drm/i915/i915_gem_context.c

@@ -473,7 +473,12 @@ mi_set_context(struct intel_engine_cs *ring,
 	       u32 hw_flags)
 {
 	u32 flags = hw_flags | MI_MM_SPACE_GTT;
-	int ret;
+	const int num_rings =
+		/* Use an extended w/a on ivb+ if signalling from other rings */
+		i915_semaphore_is_enabled(ring->dev) ?
+		hweight32(INTEL_INFO(ring->dev)->ring_mask) - 1 :
+		0;
+	int len, i, ret;
 
 	/* w/a: If Flush TLB Invalidation Mode is enabled, driver must do a TLB
 	 * invalidation prior to MI_SET_CONTEXT. On GEN6 we don't set the value
@@ -490,15 +495,31 @@ mi_set_context(struct intel_engine_cs *ring,
 	if (!IS_HASWELL(ring->dev) && INTEL_INFO(ring->dev)->gen < 8)
 		flags |= (MI_SAVE_EXT_STATE_EN | MI_RESTORE_EXT_STATE_EN);
 
-	ret = intel_ring_begin(ring, 6);
+
+	len = 4;
+	if (INTEL_INFO(ring->dev)->gen >= 7)
+		len += 2 + (num_rings ? 4*num_rings + 2 : 0);
+
+	ret = intel_ring_begin(ring, len);
 	if (ret)
 		return ret;
 
 	/* WaProgramMiArbOnOffAroundMiSetContext:ivb,vlv,hsw,bdw,chv */
-	if (INTEL_INFO(ring->dev)->gen >= 7)
+	if (INTEL_INFO(ring->dev)->gen >= 7) {
 		intel_ring_emit(ring, MI_ARB_ON_OFF | MI_ARB_DISABLE);
-	else
+		if (num_rings) {
-		intel_ring_emit(ring, MI_NOOP);
+			struct intel_engine_cs *signaller;
+
+			intel_ring_emit(ring, MI_LOAD_REGISTER_IMM(num_rings));
+			for_each_ring(signaller, to_i915(ring->dev), i) {
+				if (signaller == ring)
+					continue;
+
+				intel_ring_emit(ring, RING_PSMI_CTL(signaller->mmio_base));
+				intel_ring_emit(ring, _MASKED_BIT_ENABLE(GEN6_PSMI_SLEEP_MSG_DISABLE));
+			}
+		}
+	}
 
 	intel_ring_emit(ring, MI_NOOP);
 	intel_ring_emit(ring, MI_SET_CONTEXT);
@@ -510,10 +531,21 @@ mi_set_context(struct intel_engine_cs *ring,
 	 */
 	intel_ring_emit(ring, MI_NOOP);
 
-	if (INTEL_INFO(ring->dev)->gen >= 7)
+	if (INTEL_INFO(ring->dev)->gen >= 7) {
+		if (num_rings) {
+			struct intel_engine_cs *signaller;
+
+			intel_ring_emit(ring, MI_LOAD_REGISTER_IMM(num_rings));
+			for_each_ring(signaller, to_i915(ring->dev), i) {
+				if (signaller == ring)
+					continue;
+
+				intel_ring_emit(ring, RING_PSMI_CTL(signaller->mmio_base));
+				intel_ring_emit(ring, _MASKED_BIT_DISABLE(GEN6_PSMI_SLEEP_MSG_DISABLE));
+			}
+		}
 		intel_ring_emit(ring, MI_ARB_ON_OFF | MI_ARB_ENABLE);
-	else
+	}
-		intel_ring_emit(ring, MI_NOOP);
 
 	intel_ring_advance(ring);
 

drivers/gpu/drm/i915/i915_gem_execbuffer.c

@@ -121,9 +121,6 @@ eb_lookup_vmas(struct eb_vmas *eb,
 			goto err;
 		}
 
-		WARN_ONCE(obj->base.dumb,
-			  "GPU use of dumb buffer is illegal.\n");
-
 		drm_gem_object_reference(&obj->base);
 		list_add_tail(&obj->obj_exec_link, &objects);
 	}

drivers/gpu/drm/i915/i915_irq.c

@@ -281,10 +281,14 @@ void gen6_enable_rps_interrupts(struct drm_device *dev)
 	struct drm_i915_private *dev_priv = dev->dev_private;
 
 	spin_lock_irq(&dev_priv->irq_lock);
+
 	WARN_ON(dev_priv->rps.pm_iir);
 	WARN_ON(I915_READ(gen6_pm_iir(dev_priv)) & dev_priv->pm_rps_events);
 	dev_priv->rps.interrupts_enabled = true;
+	I915_WRITE(gen6_pm_ier(dev_priv), I915_READ(gen6_pm_ier(dev_priv)) |
+				dev_priv->pm_rps_events);
 	gen6_enable_pm_irq(dev_priv, dev_priv->pm_rps_events);
+
 	spin_unlock_irq(&dev_priv->irq_lock);
 }
 
@@ -3307,8 +3311,10 @@ static void gen5_gt_irq_postinstall(struct drm_device *dev)
 	GEN5_IRQ_INIT(GT, dev_priv->gt_irq_mask, gt_irqs);
 
 	if (INTEL_INFO(dev)->gen >= 6) {
-		pm_irqs |= dev_priv->pm_rps_events;
+		/*
-
+		 * RPS interrupts will get enabled/disabled on demand when RPS
+		 * itself is enabled/disabled.
+		 */
 		if (HAS_VEBOX(dev))
 			pm_irqs |= PM_VEBOX_USER_INTERRUPT;
 
@@ -3520,7 +3526,11 @@ static void gen8_gt_irq_postinstall(struct drm_i915_private *dev_priv)
 	dev_priv->pm_irq_mask = 0xffffffff;
 	GEN8_IRQ_INIT_NDX(GT, 0, ~gt_interrupts[0], gt_interrupts[0]);
 	GEN8_IRQ_INIT_NDX(GT, 1, ~gt_interrupts[1], gt_interrupts[1]);
-	GEN8_IRQ_INIT_NDX(GT, 2, dev_priv->pm_irq_mask, dev_priv->pm_rps_events);
+	/*
+	 * RPS interrupts will get enabled/disabled on demand when RPS itself
+	 * is enabled/disabled.
+	 */
+	GEN8_IRQ_INIT_NDX(GT, 2, dev_priv->pm_irq_mask, 0);
 	GEN8_IRQ_INIT_NDX(GT, 3, ~gt_interrupts[3], gt_interrupts[3]);
 }
 
@@ -3609,7 +3619,7 @@ static void vlv_display_irq_uninstall(struct drm_i915_private *dev_priv)
 
 	vlv_display_irq_reset(dev_priv);
 
-	dev_priv->irq_mask = 0;
+	dev_priv->irq_mask = ~0;
 }
 
 static void valleyview_irq_uninstall(struct drm_device *dev)

drivers/gpu/drm/i915/i915_reg.h

@@ -395,6 +395,7 @@
 #define   PIPE_CONTROL_STORE_DATA_INDEX			(1<<21)
 #define   PIPE_CONTROL_CS_STALL				(1<<20)
 #define   PIPE_CONTROL_TLB_INVALIDATE			(1<<18)
+#define   PIPE_CONTROL_MEDIA_STATE_CLEAR		(1<<16)
 #define   PIPE_CONTROL_QW_WRITE				(1<<14)
 #define   PIPE_CONTROL_POST_SYNC_OP_MASK                (3<<14)
 #define   PIPE_CONTROL_DEPTH_STALL			(1<<13)
@@ -1128,6 +1129,7 @@ enum punit_power_well {
 #define GEN6_VERSYNC	(RING_SYNC_1(VEBOX_RING_BASE))
 #define GEN6_VEVSYNC	(RING_SYNC_2(VEBOX_RING_BASE))
 #define GEN6_NOSYNC 0
+#define RING_PSMI_CTL(base)	((base)+0x50)
 #define RING_MAX_IDLE(base)	((base)+0x54)
 #define RING_HWS_PGA(base)	((base)+0x80)
 #define RING_HWS_PGA_GEN6(base)	((base)+0x2080)
@@ -1458,6 +1460,7 @@ enum punit_power_well {
 #define   GEN6_BLITTER_FBC_NOTIFY			(1<<3)
 
 #define GEN6_RC_SLEEP_PSMI_CONTROL	0x2050
+#define   GEN6_PSMI_SLEEP_MSG_DISABLE	(1 << 0)
 #define   GEN8_RC_SEMA_IDLE_MSG_DISABLE	(1 << 12)
 #define   GEN8_FF_DOP_CLOCK_GATE_DISABLE	(1<<10)
 

drivers/gpu/drm/i915/intel_pm.c

@@ -6191,6 +6191,20 @@ void intel_cleanup_gt_powersave(struct drm_device *dev)
 		valleyview_cleanup_gt_powersave(dev);
 }
 
+static void gen6_suspend_rps(struct drm_device *dev)
+{
+	struct drm_i915_private *dev_priv = dev->dev_private;
+
+	flush_delayed_work(&dev_priv->rps.delayed_resume_work);
+
+	/*
+	 * TODO: disable RPS interrupts on GEN9+ too once RPS support
+	 * is added for it.
+	 */
+	if (INTEL_INFO(dev)->gen < 9)
+		gen6_disable_rps_interrupts(dev);
+}
+
 /**
  * intel_suspend_gt_powersave - suspend PM work and helper threads
  * @dev: drm device
@@ -6206,14 +6220,7 @@ void intel_suspend_gt_powersave(struct drm_device *dev)
 	if (INTEL_INFO(dev)->gen < 6)
 		return;
 
-	flush_delayed_work(&dev_priv->rps.delayed_resume_work);
+	gen6_suspend_rps(dev);
-
-	/*
-	 * TODO: disable RPS interrupts on GEN9+ too once RPS support
-	 * is added for it.
-	 */
-	if (INTEL_INFO(dev)->gen < 9)
-		gen6_disable_rps_interrupts(dev);
 
 	/* Force GPU to min freq during suspend */
 	gen6_rps_idle(dev_priv);
@@ -6316,8 +6323,11 @@ void intel_reset_gt_powersave(struct drm_device *dev)
 {
 	struct drm_i915_private *dev_priv = dev->dev_private;
 
+	if (INTEL_INFO(dev)->gen < 6)
+		return;
+
+	gen6_suspend_rps(dev);
 	dev_priv->rps.enabled = false;
-	intel_enable_gt_powersave(dev);
 }
 
 static void ibx_init_clock_gating(struct drm_device *dev)

drivers/gpu/drm/i915/intel_ringbuffer.c

@@ -362,12 +362,15 @@ gen7_render_ring_flush(struct intel_engine_cs *ring,
 		flags |= PIPE_CONTROL_VF_CACHE_INVALIDATE;
 		flags |= PIPE_CONTROL_CONST_CACHE_INVALIDATE;
 		flags |= PIPE_CONTROL_STATE_CACHE_INVALIDATE;
+		flags |= PIPE_CONTROL_MEDIA_STATE_CLEAR;
 		/*
 		 * TLB invalidate requires a post-sync write.
 		 */
 		flags |= PIPE_CONTROL_QW_WRITE;
 		flags |= PIPE_CONTROL_GLOBAL_GTT_IVB;
 
+		flags |= PIPE_CONTROL_STALL_AT_SCOREBOARD;
+
 		/* Workaround: we must issue a pipe_control with CS-stall bit
 		 * set before a pipe_control command that has the state cache
 		 * invalidate bit set. */

drivers/gpu/drm/msm/adreno/adreno_gpu.c

@@ -386,9 +386,7 @@ void adreno_gpu_cleanup(struct adreno_gpu *gpu)
 			msm_gem_put_iova(gpu->memptrs_bo, gpu->base.id);
 		drm_gem_object_unreference(gpu->memptrs_bo);
 	}
-	if (gpu->pm4)
+	release_firmware(gpu->pm4);
-		release_firmware(gpu->pm4);
+	release_firmware(gpu->pfp);
-	if (gpu->pfp)
-		release_firmware(gpu->pfp);
 	msm_gpu_cleanup(&gpu->base);
 }

drivers/gpu/drm/msm/hdmi/hdmi_connector.c

@@ -141,6 +141,15 @@ static int hpd_enable(struct hdmi_connector *hdmi_connector)
 	uint32_t hpd_ctrl;
 	int i, ret;
 
+	for (i = 0; i < config->hpd_reg_cnt; i++) {
+		ret = regulator_enable(hdmi->hpd_regs[i]);
+		if (ret) {
+			dev_err(dev->dev, "failed to enable hpd regulator: %s (%d)\n",
+					config->hpd_reg_names[i], ret);
+			goto fail;
+		}
+	}
+
 	ret = gpio_config(hdmi, true);
 	if (ret) {
 		dev_err(dev->dev, "failed to configure GPIOs: %d\n", ret);
@@ -164,15 +173,6 @@ static int hpd_enable(struct hdmi_connector *hdmi_connector)
 		}
 	}
 
-	for (i = 0; i < config->hpd_reg_cnt; i++) {
-		ret = regulator_enable(hdmi->hpd_regs[i]);
-		if (ret) {
-			dev_err(dev->dev, "failed to enable hpd regulator: %s (%d)\n",
-					config->hpd_reg_names[i], ret);
-			goto fail;
-		}
-	}
-
 	hdmi_set_mode(hdmi, false);
 	phy->funcs->reset(phy);
 	hdmi_set_mode(hdmi, true);
@@ -200,7 +200,7 @@ fail:
 	return ret;
 }
 
-static int hdp_disable(struct hdmi_connector *hdmi_connector)
+static void hdp_disable(struct hdmi_connector *hdmi_connector)
 {
 	struct hdmi *hdmi = hdmi_connector->hdmi;
 	const struct hdmi_platform_config *config = hdmi->config;
@@ -212,28 +212,19 @@ static int hdp_disable(struct hdmi_connector *hdmi_connector)
 
 	hdmi_set_mode(hdmi, false);
 
-	for (i = 0; i < config->hpd_reg_cnt; i++) {
-		ret = regulator_disable(hdmi->hpd_regs[i]);
-		if (ret) {
-			dev_err(dev->dev, "failed to disable hpd regulator: %s (%d)\n",
-					config->hpd_reg_names[i], ret);
-			goto fail;
-		}
-	}
-
 	for (i = 0; i < config->hpd_clk_cnt; i++)
 		clk_disable_unprepare(hdmi->hpd_clks[i]);
 
 	ret = gpio_config(hdmi, false);
-	if (ret) {
+	if (ret)
-		dev_err(dev->dev, "failed to unconfigure GPIOs: %d\n", ret);
+		dev_warn(dev->dev, "failed to unconfigure GPIOs: %d\n", ret);
-		goto fail;
+
+	for (i = 0; i < config->hpd_reg_cnt; i++) {
+		ret = regulator_disable(hdmi->hpd_regs[i]);
+		if (ret)
+			dev_warn(dev->dev, "failed to disable hpd regulator: %s (%d)\n",
+					config->hpd_reg_names[i], ret);
 	}
-
-	return 0;
-
-fail:
-	return ret;
 }
 
 static void
@@ -260,11 +251,11 @@ void hdmi_connector_irq(struct drm_connector *connector)
 			(hpd_int_status & HDMI_HPD_INT_STATUS_INT)) {
 		bool detected = !!(hpd_int_status & HDMI_HPD_INT_STATUS_CABLE_DETECTED);
 
-		DBG("status=%04x, ctrl=%04x", hpd_int_status, hpd_int_ctrl);
+		/* ack & disable (temporarily) HPD events: */
-
-		/* ack the irq: */
 		hdmi_write(hdmi, REG_HDMI_HPD_INT_CTRL,
-				hpd_int_ctrl | HDMI_HPD_INT_CTRL_INT_ACK);
+			HDMI_HPD_INT_CTRL_INT_ACK);
+
+		DBG("status=%04x, ctrl=%04x", hpd_int_status, hpd_int_ctrl);
 
 		/* detect disconnect if we are connected or visa versa: */
 		hpd_int_ctrl = HDMI_HPD_INT_CTRL_INT_EN;

drivers/gpu/drm/msm/mdp/mdp4/mdp4_crtc.c

@@ -331,17 +331,8 @@ static int mdp4_crtc_atomic_check(struct drm_crtc *crtc,
 		struct drm_crtc_state *state)
 {
 	struct mdp4_crtc *mdp4_crtc = to_mdp4_crtc(crtc);
-	struct drm_device *dev = crtc->dev;
-
 	DBG("%s: check", mdp4_crtc->name);
-
-	if (mdp4_crtc->event) {
-		dev_err(dev->dev, "already pending flip!\n");
-		return -EBUSY;
-	}
-
 	// TODO anything else to check?
-
 	return 0;
 }
 
@@ -357,7 +348,7 @@ static void mdp4_crtc_atomic_flush(struct drm_crtc *crtc)
 	struct drm_device *dev = crtc->dev;
 	unsigned long flags;
 
-	DBG("%s: flush", mdp4_crtc->name);
+	DBG("%s: event: %p", mdp4_crtc->name, crtc->state->event);
 
 	WARN_ON(mdp4_crtc->event);
 

drivers/gpu/drm/msm/mdp/mdp5/mdp5_crtc.c

@@ -303,11 +303,6 @@ static int mdp5_crtc_atomic_check(struct drm_crtc *crtc,
 
 	DBG("%s: check", mdp5_crtc->name);
 
-	if (mdp5_crtc->event) {
-		dev_err(dev->dev, "already pending flip!\n");
-		return -EBUSY;
-	}
-
 	/* request a free CTL, if none is already allocated for this CRTC */
 	if (state->enable && !mdp5_crtc->ctl) {
 		mdp5_crtc->ctl = mdp5_ctlm_request(mdp5_kms->ctlm, crtc);
@@ -364,7 +359,7 @@ static void mdp5_crtc_atomic_flush(struct drm_crtc *crtc)
 	struct drm_device *dev = crtc->dev;
 	unsigned long flags;
 
-	DBG("%s: flush", mdp5_crtc->name);
+	DBG("%s: event: %p", mdp5_crtc->name, crtc->state->event);
 
 	WARN_ON(mdp5_crtc->event);
 
@@ -460,10 +455,7 @@ void mdp5_crtc_set_intf(struct drm_crtc *crtc, int intf,
 	/* now that we know what irq's we want: */
 	mdp5_crtc->err.irqmask = intf2err(intf);
 	mdp5_crtc->vblank.irqmask = intf2vblank(intf);
-
+	mdp_irq_update(&mdp5_kms->base);
-	/* when called from modeset_init(), skip the rest until later: */
-	if (!mdp5_kms)
-		return;
 
 	spin_lock_irqsave(&mdp5_kms->resource_lock, flags);
 	intf_sel = mdp5_read(mdp5_kms, REG_MDP5_DISP_INTF_SEL);

drivers/gpu/drm/msm/mdp/mdp5/mdp5_kms.c

@@ -216,17 +216,7 @@ static int modeset_init(struct mdp5_kms *mdp5_kms)
 		goto fail;
 	}
 
-	/* NOTE: the vsync and error irq's are actually associated with
+	encoder->possible_crtcs = (1 << priv->num_crtcs) - 1;;
-	 * the INTF/encoder.. the easiest way to deal with this (ie. what
-	 * we do now) is assume a fixed relationship between crtc's and
-	 * encoders.  I'm not sure if there is ever a need to more freely
-	 * assign crtcs to encoders, but if there is then we need to take
-	 * care of error and vblank irq's that the crtc has registered,
-	 * and also update user-requested vblank_mask.
-	 */
-	encoder->possible_crtcs = BIT(0);
-	mdp5_crtc_set_intf(priv->crtcs[0], 3, INTF_HDMI);
-
 	priv->encoders[priv->num_encoders++] = encoder;
 
 	/* Construct bridge/connector for HDMI: */

drivers/gpu/drm/msm/mdp/mdp_kms.c

@@ -42,7 +42,10 @@ static void update_irq(struct mdp_kms *mdp_kms)
 	mdp_kms->funcs->set_irqmask(mdp_kms, irqmask);
 }
 
-static void update_irq_unlocked(struct mdp_kms *mdp_kms)
+/* if an mdp_irq's irqmask has changed, such as when mdp5 crtc<->encoder
+ * link changes, this must be called to figure out the new global irqmask
+ */
+void mdp_irq_update(struct mdp_kms *mdp_kms)
 {
 	unsigned long flags;
 	spin_lock_irqsave(&list_lock, flags);
@@ -122,7 +125,7 @@ void mdp_irq_register(struct mdp_kms *mdp_kms, struct mdp_irq *irq)
 	spin_unlock_irqrestore(&list_lock, flags);
 
 	if (needs_update)
-		update_irq_unlocked(mdp_kms);
+		mdp_irq_update(mdp_kms);
 }
 
 void mdp_irq_unregister(struct mdp_kms *mdp_kms, struct mdp_irq *irq)
@@ -141,5 +144,5 @@ void mdp_irq_unregister(struct mdp_kms *mdp_kms, struct mdp_irq *irq)
 	spin_unlock_irqrestore(&list_lock, flags);
 
 	if (needs_update)
-		update_irq_unlocked(mdp_kms);
+		mdp_irq_update(mdp_kms);
 }

drivers/gpu/drm/msm/mdp/mdp_kms.h

@@ -75,7 +75,7 @@ void mdp_update_vblank_mask(struct mdp_kms *mdp_kms, uint32_t mask, bool enable)
 void mdp_irq_wait(struct mdp_kms *mdp_kms, uint32_t irqmask);
 void mdp_irq_register(struct mdp_kms *mdp_kms, struct mdp_irq *irq);
 void mdp_irq_unregister(struct mdp_kms *mdp_kms, struct mdp_irq *irq);
-
+void mdp_irq_update(struct mdp_kms *mdp_kms);
 
 /*
  * pixel format helpers:

drivers/gpu/drm/msm/msm_atomic.c

@@ -23,10 +23,41 @@ struct msm_commit {
 	struct drm_atomic_state *state;
 	uint32_t fence;
 	struct msm_fence_cb fence_cb;
+	uint32_t crtc_mask;
 };
 
 static void fence_cb(struct msm_fence_cb *cb);
 
+/* block until specified crtcs are no longer pending update, and
+ * atomically mark them as pending update
+ */
+static int start_atomic(struct msm_drm_private *priv, uint32_t crtc_mask)
+{
+	int ret;
+
+	spin_lock(&priv->pending_crtcs_event.lock);
+	ret = wait_event_interruptible_locked(priv->pending_crtcs_event,
+			!(priv->pending_crtcs & crtc_mask));
+	if (ret == 0) {
+		DBG("start: %08x", crtc_mask);
+		priv->pending_crtcs |= crtc_mask;
+	}
+	spin_unlock(&priv->pending_crtcs_event.lock);
+
+	return ret;
+}
+
+/* clear specified crtcs (no longer pending update)
+ */
+static void end_atomic(struct msm_drm_private *priv, uint32_t crtc_mask)
+{
+	spin_lock(&priv->pending_crtcs_event.lock);
+	DBG("end: %08x", crtc_mask);
+	priv->pending_crtcs &= ~crtc_mask;
+	wake_up_all_locked(&priv->pending_crtcs_event);
+	spin_unlock(&priv->pending_crtcs_event.lock);
+}
+
 static struct msm_commit *new_commit(struct drm_atomic_state *state)
 {
 	struct msm_commit *c = kzalloc(sizeof(*c), GFP_KERNEL);
@@ -58,12 +89,27 @@ static void complete_commit(struct msm_commit *c)
 
 	drm_atomic_helper_commit_post_planes(dev, state);
 
+	/* NOTE: _wait_for_vblanks() only waits for vblank on
+	 * enabled CRTCs.  So we end up faulting when disabling
+	 * due to (potentially) unref'ing the outgoing fb's
+	 * before the vblank when the disable has latched.
+	 *
+	 * But if it did wait on disabled (or newly disabled)
+	 * CRTCs, that would be racy (ie. we could have missed
+	 * the irq.  We need some way to poll for pipe shut
+	 * down.  Or just live with occasionally hitting the
+	 * timeout in the CRTC disable path (which really should
+	 * not be critical path)
+	 */
+
 	drm_atomic_helper_wait_for_vblanks(dev, state);
 
 	drm_atomic_helper_cleanup_planes(dev, state);
 
 	drm_atomic_state_free(state);
 
+	end_atomic(dev->dev_private, c->crtc_mask);
+
 	kfree(c);
 }
 
@@ -97,8 +143,9 @@ static void add_fb(struct msm_commit *c, struct drm_framebuffer *fb)
 int msm_atomic_commit(struct drm_device *dev,
 		struct drm_atomic_state *state, bool async)
 {
-	struct msm_commit *c;
 	int nplanes = dev->mode_config.num_total_plane;
+	int ncrtcs = dev->mode_config.num_crtc;
+	struct msm_commit *c;
 	int i, ret;
 
 	ret = drm_atomic_helper_prepare_planes(dev, state);
@@ -106,6 +153,18 @@ int msm_atomic_commit(struct drm_device *dev,
 		return ret;
 
 	c = new_commit(state);
+	if (!c)
+		return -ENOMEM;
+
+	/*
+	 * Figure out what crtcs we have:
+	 */
+	for (i = 0; i < ncrtcs; i++) {
+		struct drm_crtc *crtc = state->crtcs[i];
+		if (!crtc)
+			continue;
+		c->crtc_mask |= (1 << drm_crtc_index(crtc));
+	}
 
 	/*
 	 * Figure out what fence to wait for:
@@ -121,6 +180,14 @@ int msm_atomic_commit(struct drm_device *dev,
 			add_fb(c, new_state->fb);
 	}
 
+	/*
+	 * Wait for pending updates on any of the same crtc's and then
+	 * mark our set of crtc's as busy:
+	 */
+	ret = start_atomic(dev->dev_private, c->crtc_mask);
+	if (ret)
+		return ret;
+
 	/*
 	 * This is the point of no return - everything below never fails except
 	 * when the hw goes bonghits. Which means we can commit the new state on

drivers/gpu/drm/msm/msm_drv.c

@@ -193,6 +193,7 @@ static int msm_load(struct drm_device *dev, unsigned long flags)
 
 	priv->wq = alloc_ordered_workqueue("msm", 0);
 	init_waitqueue_head(&priv->fence_event);
+	init_waitqueue_head(&priv->pending_crtcs_event);
 
 	INIT_LIST_HEAD(&priv->inactive_list);
 	INIT_LIST_HEAD(&priv->fence_cbs);

drivers/gpu/drm/msm/msm_drv.h

@@ -96,6 +96,10 @@ struct msm_drm_private {
 	/* callbacks deferred until bo is inactive: */
 	struct list_head fence_cbs;
 
+	/* crtcs pending async atomic updates: */
+	uint32_t pending_crtcs;
+	wait_queue_head_t pending_crtcs_event;
+
 	/* registered MMUs: */
 	unsigned int num_mmus;
 	struct msm_mmu *mmus[NUM_DOMAINS];

drivers/gpu/drm/msm/msm_fbdev.c

@@ -190,8 +190,7 @@ fail_unlock:
 fail:
 
 	if (ret) {
-		if (fbi)
+		framebuffer_release(fbi);
-			framebuffer_release(fbi);
 		if (fb) {
 			drm_framebuffer_unregister_private(fb);
 			drm_framebuffer_remove(fb);

drivers/gpu/drm/msm/msm_gem.c

@@ -535,8 +535,7 @@ void msm_gem_free_object(struct drm_gem_object *obj)
 			drm_free_large(msm_obj->pages);
 
 	} else {
-		if (msm_obj->vaddr)
+		vunmap(msm_obj->vaddr);
-			vunmap(msm_obj->vaddr);
 		put_pages(obj);
 	}
 

drivers/gpu/drm/nouveau/nouveau_display.c

@@ -876,7 +876,6 @@ nouveau_display_dumb_create(struct drm_file *file_priv, struct drm_device *dev,
 	if (ret)
 		return ret;
 
-	bo->gem.dumb = true;
 	ret = drm_gem_handle_create(file_priv, &bo->gem, &args->handle);
 	drm_gem_object_unreference_unlocked(&bo->gem);
 	return ret;
@@ -892,14 +891,6 @@ nouveau_display_dumb_map_offset(struct drm_file *file_priv,
 	gem = drm_gem_object_lookup(dev, file_priv, handle);
 	if (gem) {
 		struct nouveau_bo *bo = nouveau_gem_object(gem);
-
-		/*
-		 * We don't allow dumb mmaps on objects created using another
-		 * interface.
-		 */
-		WARN_ONCE(!(gem->dumb || gem->import_attach),
-			  "Illegal dumb map of accelerated buffer.\n");
-
 		*poffset = drm_vma_node_offset_addr(&bo->bo.vma_node);
 		drm_gem_object_unreference_unlocked(gem);
 		return 0;

drivers/gpu/drm/nouveau/nouveau_gem.c

@@ -444,9 +444,6 @@ validate_list(struct nouveau_channel *chan, struct nouveau_cli *cli,
 	list_for_each_entry(nvbo, list, entry) {
 		struct drm_nouveau_gem_pushbuf_bo *b = &pbbo[nvbo->pbbo_index];
 
-		WARN_ONCE(nvbo->gem.dumb,
-			  "GPU use of dumb buffer is illegal.\n");
-
 		ret = nouveau_gem_set_domain(&nvbo->gem, b->read_domains,
 					     b->write_domains,
 					     b->valid_domains);

drivers/gpu/drm/nouveau/nouveau_ttm.c

@@ -28,6 +28,7 @@
 #include "nouveau_ttm.h"
 #include "nouveau_gem.h"
 
+#include "drm_legacy.h"
 static int
 nouveau_vram_manager_init(struct ttm_mem_type_manager *man, unsigned long psize)
 {
@@ -281,7 +282,7 @@ nouveau_ttm_mmap(struct file *filp, struct vm_area_struct *vma)
 	struct nouveau_drm *drm = nouveau_drm(file_priv->minor->dev);
 
 	if (unlikely(vma->vm_pgoff < DRM_FILE_PAGE_OFFSET))
-		return -EINVAL;
+		return drm_legacy_mmap(filp, vma);
 
 	return ttm_bo_mmap(filp, vma, &drm->ttm.bdev);
 }

drivers/gpu/drm/radeon/radeon_gem.c

@@ -394,10 +394,9 @@ int radeon_gem_set_domain_ioctl(struct drm_device *dev, void *data,
 	return r;
 }
 
-static int radeon_mode_mmap(struct drm_file *filp,
+int radeon_mode_dumb_mmap(struct drm_file *filp,
-			    struct drm_device *dev,
+			  struct drm_device *dev,
-			    uint32_t handle, bool dumb,
+			  uint32_t handle, uint64_t *offset_p)
-			    uint64_t *offset_p)
 {
 	struct drm_gem_object *gobj;
 	struct radeon_bo *robj;
@@ -406,14 +405,6 @@ static int radeon_mode_mmap(struct drm_file *filp,
 	if (gobj == NULL) {
 		return -ENOENT;
 	}
-
-	/*
-	 * We don't allow dumb mmaps on objects created using another
-	 * interface.
-	 */
-	WARN_ONCE(dumb && !(gobj->dumb || gobj->import_attach),
-		"Illegal dumb map of GPU buffer.\n");
-
 	robj = gem_to_radeon_bo(gobj);
 	if (radeon_ttm_tt_has_userptr(robj->tbo.ttm)) {
 		drm_gem_object_unreference_unlocked(gobj);
@@ -424,20 +415,12 @@ static int radeon_mode_mmap(struct drm_file *filp,
 	return 0;
 }
 
-int radeon_mode_dumb_mmap(struct drm_file *filp,
-			  struct drm_device *dev,
-			  uint32_t handle, uint64_t *offset_p)
-{
-	return radeon_mode_mmap(filp, dev, handle, true, offset_p);
-}
-
 int radeon_gem_mmap_ioctl(struct drm_device *dev, void *data,
 			  struct drm_file *filp)
 {
 	struct drm_radeon_gem_mmap *args = data;
 
-	return radeon_mode_mmap(filp, dev, args->handle, false,
+	return radeon_mode_dumb_mmap(filp, dev, args->handle, &args->addr_ptr);
-				&args->addr_ptr);
 }
 
 int radeon_gem_busy_ioctl(struct drm_device *dev, void *data,
@@ -763,7 +746,6 @@ int radeon_mode_dumb_create(struct drm_file *file_priv,
 		return -ENOMEM;
 
 	r = drm_gem_handle_create(file_priv, gobj, &handle);
-	gobj->dumb = true;
 	/* drop reference from allocate - handle holds it now */
 	drm_gem_object_unreference_unlocked(gobj);
 	if (r) {

drivers/gpu/drm/radeon/radeon_kfd.c

@@ -28,6 +28,8 @@
 #include "cikd.h"
 #include "cik_reg.h"
 #include "radeon_kfd.h"
+#include "radeon_ucode.h"
+#include <linux/firmware.h>
 
 #define CIK_PIPE_PER_MEC	(4)
 
@@ -49,6 +51,7 @@ static uint64_t get_vmem_size(struct kgd_dev *kgd);
 static uint64_t get_gpu_clock_counter(struct kgd_dev *kgd);
 
 static uint32_t get_max_engine_clock_in_mhz(struct kgd_dev *kgd);
+static uint16_t get_fw_version(struct kgd_dev *kgd, enum kgd_engine_type type);
 
 /*
  * Register access functions
@@ -91,6 +94,7 @@ static const struct kfd2kgd_calls kfd2kgd = {
 	.hqd_load = kgd_hqd_load,
 	.hqd_is_occupies = kgd_hqd_is_occupies,
 	.hqd_destroy = kgd_hqd_destroy,
+	.get_fw_version = get_fw_version
 };
 
 static const struct kgd2kfd_calls *kgd2kfd;
@@ -561,3 +565,52 @@ static int kgd_hqd_destroy(struct kgd_dev *kgd, uint32_t reset_type,
 	release_queue(kgd);
 	return 0;
 }
+
+static uint16_t get_fw_version(struct kgd_dev *kgd, enum kgd_engine_type type)
+{
+	struct radeon_device *rdev = (struct radeon_device *) kgd;
+	const union radeon_firmware_header *hdr;
+
+	BUG_ON(kgd == NULL || rdev->mec_fw == NULL);
+
+	switch (type) {
+	case KGD_ENGINE_PFP:
+		hdr = (const union radeon_firmware_header *) rdev->pfp_fw->data;
+		break;
+
+	case KGD_ENGINE_ME:
+		hdr = (const union radeon_firmware_header *) rdev->me_fw->data;
+		break;
+
+	case KGD_ENGINE_CE:
+		hdr = (const union radeon_firmware_header *) rdev->ce_fw->data;
+		break;
+
+	case KGD_ENGINE_MEC1:
+		hdr = (const union radeon_firmware_header *) rdev->mec_fw->data;
+		break;
+
+	case KGD_ENGINE_MEC2:
+		hdr = (const union radeon_firmware_header *)
+							rdev->mec2_fw->data;
+		break;
+
+	case KGD_ENGINE_RLC:
+		hdr = (const union radeon_firmware_header *) rdev->rlc_fw->data;
+		break;
+
+	case KGD_ENGINE_SDMA:
+		hdr = (const union radeon_firmware_header *)
+							rdev->sdma_fw->data;
+		break;
+
+	default:
+		return 0;
+	}
+
+	if (hdr == NULL)
+		return 0;
+
+	/* Only 12 bit in use*/
+	return hdr->common.ucode_version;
+}

drivers/gpu/drm/radeon/radeon_object.c

@@ -529,9 +529,6 @@ int radeon_bo_list_validate(struct radeon_device *rdev,
 			u32 current_domain =
 				radeon_mem_type_to_domain(bo->tbo.mem.mem_type);
 
-			WARN_ONCE(bo->gem_base.dumb,
-				  "GPU use of dumb buffer is illegal.\n");
-
 			/* Check if this buffer will be moved and don't move it
 			 * if we have moved too many buffers for this IB already.
 			 *

drivers/gpu/drm/tegra/dc.c

@@ -168,7 +168,7 @@ static int tegra_dc_setup_window(struct tegra_dc *dc, unsigned int index,
 				 const struct tegra_dc_window *window)
 {
 	unsigned h_offset, v_offset, h_size, v_size, h_dda, v_dda, bpp;
-	unsigned long value;
+	unsigned long value, flags;
 	bool yuv, planar;
 
 	/*
@@ -181,6 +181,8 @@ static int tegra_dc_setup_window(struct tegra_dc *dc, unsigned int index,
 	else
 		bpp = planar ? 1 : 2;
 
+	spin_lock_irqsave(&dc->lock, flags);
+
 	value = WINDOW_A_SELECT << index;
 	tegra_dc_writel(dc, value, DC_CMD_DISPLAY_WINDOW_HEADER);
 
@@ -273,6 +275,7 @@ static int tegra_dc_setup_window(struct tegra_dc *dc, unsigned int index,
 
 		case TEGRA_BO_TILING_MODE_BLOCK:
 			DRM_ERROR("hardware doesn't support block linear mode\n");
+			spin_unlock_irqrestore(&dc->lock, flags);
 			return -EINVAL;
 		}
 
@@ -331,6 +334,8 @@ static int tegra_dc_setup_window(struct tegra_dc *dc, unsigned int index,
 
 	tegra_dc_window_commit(dc, index);
 
+	spin_unlock_irqrestore(&dc->lock, flags);
+
 	return 0;
 }
 
@@ -338,11 +343,14 @@ static int tegra_window_plane_disable(struct drm_plane *plane)
 {
 	struct tegra_dc *dc = to_tegra_dc(plane->crtc);
 	struct tegra_plane *p = to_tegra_plane(plane);
+	unsigned long flags;
 	u32 value;
 
 	if (!plane->crtc)
 		return 0;
 
+	spin_lock_irqsave(&dc->lock, flags);
+
 	value = WINDOW_A_SELECT << p->index;
 	tegra_dc_writel(dc, value, DC_CMD_DISPLAY_WINDOW_HEADER);
 
@@ -352,6 +360,8 @@ static int tegra_window_plane_disable(struct drm_plane *plane)
 
 	tegra_dc_window_commit(dc, p->index);
 
+	spin_unlock_irqrestore(&dc->lock, flags);
+
 	return 0;
 }
 
@@ -699,14 +709,16 @@ static int tegra_dc_set_base(struct tegra_dc *dc, int x, int y,
 	struct tegra_bo *bo = tegra_fb_get_plane(fb, 0);
 	unsigned int h_offset = 0, v_offset = 0;
 	struct tegra_bo_tiling tiling;
+	unsigned long value, flags;
 	unsigned int format, swap;
-	unsigned long value;
 	int err;
 
 	err = tegra_fb_get_tiling(fb, &tiling);
 	if (err < 0)
 		return err;
 
+	spin_lock_irqsave(&dc->lock, flags);
+
 	tegra_dc_writel(dc, WINDOW_A_SELECT, DC_CMD_DISPLAY_WINDOW_HEADER);
 
 	value = fb->offsets[0] + y * fb->pitches[0] +
@@ -752,6 +764,7 @@ static int tegra_dc_set_base(struct tegra_dc *dc, int x, int y,
 
 		case TEGRA_BO_TILING_MODE_BLOCK:
 			DRM_ERROR("hardware doesn't support block linear mode\n");
+			spin_unlock_irqrestore(&dc->lock, flags);
 			return -EINVAL;
 		}
 
@@ -778,6 +791,8 @@ static int tegra_dc_set_base(struct tegra_dc *dc, int x, int y,
 	tegra_dc_writel(dc, value << 8, DC_CMD_STATE_CONTROL);
 	tegra_dc_writel(dc, value, DC_CMD_STATE_CONTROL);
 
+	spin_unlock_irqrestore(&dc->lock, flags);
+
 	return 0;
 }
 
@@ -814,23 +829,32 @@ static void tegra_dc_finish_page_flip(struct tegra_dc *dc)
 	unsigned long flags, base;
 	struct tegra_bo *bo;
 
-	if (!dc->event)
+	spin_lock_irqsave(&drm->event_lock, flags);
+
+	if (!dc->event) {
+		spin_unlock_irqrestore(&drm->event_lock, flags);
 		return;
+	}
 
 	bo = tegra_fb_get_plane(crtc->primary->fb, 0);
 
+	spin_lock_irqsave(&dc->lock, flags);
+
 	/* check if new start address has been latched */
+	tegra_dc_writel(dc, WINDOW_A_SELECT, DC_CMD_DISPLAY_WINDOW_HEADER);
 	tegra_dc_writel(dc, READ_MUX, DC_CMD_STATE_ACCESS);
 	base = tegra_dc_readl(dc, DC_WINBUF_START_ADDR);
 	tegra_dc_writel(dc, 0, DC_CMD_STATE_ACCESS);
 
+	spin_unlock_irqrestore(&dc->lock, flags);
+
 	if (base == bo->paddr + crtc->primary->fb->offsets[0]) {
-		spin_lock_irqsave(&drm->event_lock, flags);
+		drm_crtc_send_vblank_event(crtc, dc->event);
-		drm_send_vblank_event(drm, dc->pipe, dc->event);
+		drm_crtc_vblank_put(crtc);
-		drm_vblank_put(drm, dc->pipe);
 		dc->event = NULL;
-		spin_unlock_irqrestore(&drm->event_lock, flags);
 	}
+
+	spin_unlock_irqrestore(&drm->event_lock, flags);
 }
 
 void tegra_dc_cancel_page_flip(struct drm_crtc *crtc, struct drm_file *file)
@@ -843,7 +867,7 @@ void tegra_dc_cancel_page_flip(struct drm_crtc *crtc, struct drm_file *file)
 
 	if (dc->event && dc->event->base.file_priv == file) {
 		dc->event->base.destroy(&dc->event->base);
-		drm_vblank_put(drm, dc->pipe);
+		drm_crtc_vblank_put(crtc);
 		dc->event = NULL;
 	}
 
@@ -853,16 +877,16 @@ void tegra_dc_cancel_page_flip(struct drm_crtc *crtc, struct drm_file *file)
 static int tegra_dc_page_flip(struct drm_crtc *crtc, struct drm_framebuffer *fb,
 			      struct drm_pending_vblank_event *event, uint32_t page_flip_flags)
 {
+	unsigned int pipe = drm_crtc_index(crtc);
 	struct tegra_dc *dc = to_tegra_dc(crtc);
-	struct drm_device *drm = crtc->dev;
 
 	if (dc->event)
 		return -EBUSY;
 
 	if (event) {
-		event->pipe = dc->pipe;
+		event->pipe = pipe;
 		dc->event = event;
-		drm_vblank_get(drm, dc->pipe);
+		drm_crtc_vblank_get(crtc);
 	}
 
 	tegra_dc_set_base(dc, 0, 0, fb);
@@ -1127,7 +1151,7 @@ static irqreturn_t tegra_dc_irq(int irq, void *data)
 		/*
 		dev_dbg(dc->dev, "%s(): vertical blank\n", __func__);
 		*/
-		drm_handle_vblank(dc->base.dev, dc->pipe);
+		drm_crtc_handle_vblank(&dc->base);
 		tegra_dc_finish_page_flip(dc);
 	}
 

drivers/gpu/drm/tegra/drm.c

@@ -694,24 +694,28 @@ static const struct file_operations tegra_drm_fops = {
 	.llseek = noop_llseek,
 };
 
-static struct drm_crtc *tegra_crtc_from_pipe(struct drm_device *drm, int pipe)
+static struct drm_crtc *tegra_crtc_from_pipe(struct drm_device *drm,
+					     unsigned int pipe)
 {
 	struct drm_crtc *crtc;
 
 	list_for_each_entry(crtc, &drm->mode_config.crtc_list, head) {
-		struct tegra_dc *dc = to_tegra_dc(crtc);
+		if (pipe == drm_crtc_index(crtc))
-
-		if (dc->pipe == pipe)
 			return crtc;
 	}
 
 	return NULL;
 }
 
-static u32 tegra_drm_get_vblank_counter(struct drm_device *dev, int crtc)
+static u32 tegra_drm_get_vblank_counter(struct drm_device *drm, int pipe)
 {
+	struct drm_crtc *crtc = tegra_crtc_from_pipe(drm, pipe);
+
+	if (!crtc)
+		return 0;
+
 	/* TODO: implement real hardware counter using syncpoints */
-	return drm_vblank_count(dev, crtc);
+	return drm_crtc_vblank_count(crtc);
 }
 
 static int tegra_drm_enable_vblank(struct drm_device *drm, int pipe)

drivers/gpu/drm/tegra/gem.c

@@ -216,32 +216,58 @@ static void tegra_bo_free(struct drm_device *drm, struct tegra_bo *bo)
 	}
 }
 
-static int tegra_bo_get_pages(struct drm_device *drm, struct tegra_bo *bo,
+static int tegra_bo_get_pages(struct drm_device *drm, struct tegra_bo *bo)
-			      size_t size)
 {
+	struct scatterlist *s;
+	struct sg_table *sgt;
+	unsigned int i;
+
 	bo->pages = drm_gem_get_pages(&bo->gem);
 	if (IS_ERR(bo->pages))
 		return PTR_ERR(bo->pages);
 
-	bo->num_pages = size >> PAGE_SHIFT;
+	bo->num_pages = bo->gem.size >> PAGE_SHIFT;
 
-	bo->sgt = drm_prime_pages_to_sg(bo->pages, bo->num_pages);
+	sgt = drm_prime_pages_to_sg(bo->pages, bo->num_pages);
-	if (IS_ERR(bo->sgt)) {
+	if (IS_ERR(sgt))
-		drm_gem_put_pages(&bo->gem, bo->pages, false, false);
+		goto put_pages;
-		return PTR_ERR(bo->sgt);
+
+	/*
+	 * Fake up the SG table so that dma_map_sg() can be used to flush the
+	 * pages associated with it. Note that this relies on the fact that
+	 * the DMA API doesn't hook into IOMMU on Tegra, therefore mapping is
+	 * only cache maintenance.
+	 *
+	 * TODO: Replace this by drm_clflash_sg() once it can be implemented
+	 * without relying on symbols that are not exported.
+	 */
+	for_each_sg(sgt->sgl, s, sgt->nents, i)
+		sg_dma_address(s) = sg_phys(s);
+
+	if (dma_map_sg(drm->dev, sgt->sgl, sgt->nents, DMA_TO_DEVICE) == 0) {
+		sgt = ERR_PTR(-ENOMEM);
+		goto release_sgt;
 	}
 
+	bo->sgt = sgt;
+
 	return 0;
+
+release_sgt:
+	sg_free_table(sgt);
+	kfree(sgt);
+put_pages:
+	drm_gem_put_pages(&bo->gem, bo->pages, false, false);
+	return PTR_ERR(sgt);
 }
 
-static int tegra_bo_alloc(struct drm_device *drm, struct tegra_bo *bo,
+static int tegra_bo_alloc(struct drm_device *drm, struct tegra_bo *bo)
-			  size_t size)
 {
 	struct tegra_drm *tegra = drm->dev_private;
 	int err;
 
 	if (tegra->domain) {
-		err = tegra_bo_get_pages(drm, bo, size);
+		err = tegra_bo_get_pages(drm, bo);
 		if (err < 0)
 			return err;
 
@@ -251,6 +277,8 @@ static int tegra_bo_alloc(struct drm_device *drm, struct tegra_bo *bo,
 			return err;
 		}
 	} else {
+		size_t size = bo->gem.size;
+
 		bo->vaddr = dma_alloc_writecombine(drm->dev, size, &bo->paddr,
 						   GFP_KERNEL | __GFP_NOWARN);
 		if (!bo->vaddr) {
@@ -274,7 +302,7 @@ struct tegra_bo *tegra_bo_create(struct drm_device *drm, size_t size,
 	if (IS_ERR(bo))
 		return bo;
 
-	err = tegra_bo_alloc(drm, bo, size);
+	err = tegra_bo_alloc(drm, bo);
 	if (err < 0)
 		goto release;
 

include/drm/drmP.h

@@ -901,11 +901,15 @@ extern int drm_vblank_init(struct drm_device *dev, int num_crtcs);
 extern int drm_wait_vblank(struct drm_device *dev, void *data,
 			   struct drm_file *filp);
 extern u32 drm_vblank_count(struct drm_device *dev, int crtc);
+extern u32 drm_crtc_vblank_count(struct drm_crtc *crtc);
 extern u32 drm_vblank_count_and_time(struct drm_device *dev, int crtc,
 				     struct timeval *vblanktime);
 extern void drm_send_vblank_event(struct drm_device *dev, int crtc,
 				     struct drm_pending_vblank_event *e);
+extern void drm_crtc_send_vblank_event(struct drm_crtc *crtc,
+				       struct drm_pending_vblank_event *e);
 extern bool drm_handle_vblank(struct drm_device *dev, int crtc);
+extern bool drm_crtc_handle_vblank(struct drm_crtc *crtc);
 extern int drm_vblank_get(struct drm_device *dev, int crtc);
 extern void drm_vblank_put(struct drm_device *dev, int crtc);
 extern int drm_crtc_vblank_get(struct drm_crtc *crtc);

include/drm/drm_gem.h

@@ -119,13 +119,6 @@ struct drm_gem_object {
 	 * simply leave it as NULL.
 	 */
 	struct dma_buf_attachment *import_attach;
-
-	/**
-	 * dumb - created as dumb buffer
-	 * Whether the gem object was created using the dumb buffer interface
-	 * as such it may not be used for GPU rendering.
-	 */
-	bool dumb;
 };
 
 void drm_gem_object_release(struct drm_gem_object *obj);