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

Pull drm fixes from Dave Airlie:
 "AST, i915, radeon and msm fixes, all over the place.

  All fixing build issues, regressions, oopses or failure to detect
  cards"

* 'drm-fixes' of git://people.freedesktop.org/~airlied/linux:
  drm/ast: AST2000 cannot be detected correctly
  drm/ast: open key before detect chips
  drm/msm: don't crash if no msm.vram param
  drm/msm/hdmi: fix build break on non-CCF platforms
  drm/msm: Change nested function to static function
  drm/radeon/dpm: set the thermal type properly for special configs
  drm/radeon: reduce memory footprint for debugging
  drm/radeon: add connector quirk for fujitsu board
  drm/radeon: fix semaphore value init
  drm/radeon: only use me/pfp sync on evergreen+
  drm/i915: Wait for vblank before enabling the TV encoder
  drm/i915: Evict CS TLBs between batches
  drm/i915: Fix irq enable tracking in driver load
  drm/i915: Fix EIO/wedged handling in gem fault handler
  drm/i915: Prevent recursive deadlock on releasing a busy userptr

15 files changed, 371 insertions, 262 deletions

diff --git a/drivers/gpu/drm/ast/ast_main.c b/drivers/gpu/drm/ast/ast_main.c
index a2cc6be97983..b792194e0d9c 100644
--- a/drivers/gpu/drm/ast/ast_main.c
+++ b/drivers/gpu/drm/ast/ast_main.c
@@ -67,6 +67,7 @@ static int ast_detect_chip(struct drm_device *dev)
 {
 	struct ast_private *ast = dev->dev_private;
 	uint32_t data, jreg;
+	ast_open_key(ast);
 
 	if (dev->pdev->device == PCI_CHIP_AST1180) {
 		ast->chip = AST1100;
@@ -104,7 +105,7 @@ static int ast_detect_chip(struct drm_device *dev)
 			}
 			ast->vga2_clone = false;
 		} else {
-			ast->chip = 2000;
+			ast->chip = AST2000;
 			DRM_INFO("AST 2000 detected\n");
 		}
 	}
diff --git a/drivers/gpu/drm/i915/i915_dma.c b/drivers/gpu/drm/i915/i915_dma.c
index 2e7f03ad5ee2..9933c26017ed 100644
--- a/drivers/gpu/drm/i915/i915_dma.c
+++ b/drivers/gpu/drm/i915/i915_dma.c
@@ -1336,12 +1336,17 @@ static int i915_load_modeset_init(struct drm_device *dev)
 
 	intel_power_domains_init_hw(dev_priv);
 
+	/*
+	 * We enable some interrupt sources in our postinstall hooks, so mark
+	 * interrupts as enabled _before_ actually enabling them to avoid
+	 * special cases in our ordering checks.
+	 */
+	dev_priv->pm._irqs_disabled = false;
+
 	ret = drm_irq_install(dev, dev->pdev->irq);
 	if (ret)
 		goto cleanup_gem_stolen;
 
-	dev_priv->pm._irqs_disabled = false;
-
 	/* Important: The output setup functions called by modeset_init need
 	 * working irqs for e.g. gmbus and dp aux transfers. */
 	intel_modeset_init(dev);
diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
index 7a830eac5ba3..3524306d8cfb 100644
--- a/drivers/gpu/drm/i915/i915_drv.h
+++ b/drivers/gpu/drm/i915/i915_drv.h
@@ -184,6 +184,7 @@ enum hpd_pin {
 		if ((1 << (domain)) & (mask))
 
 struct drm_i915_private;
+struct i915_mm_struct;
 struct i915_mmu_object;
 
 enum intel_dpll_id {
@@ -1506,9 +1507,8 @@ struct drm_i915_private {
 	struct i915_gtt gtt; /* VM representing the global address space */
 
 	struct i915_gem_mm mm;
-#if defined(CONFIG_MMU_NOTIFIER)
-	DECLARE_HASHTABLE(mmu_notifiers, 7);
-#endif
+	DECLARE_HASHTABLE(mm_structs, 7);
+	struct mutex mm_lock;
 
 	/* Kernel Modesetting */
 
@@ -1814,8 +1814,8 @@ struct drm_i915_gem_object {
 			unsigned workers :4;
 #define I915_GEM_USERPTR_MAX_WORKERS 15
 
-			struct mm_struct *mm;
-			struct i915_mmu_object *mn;
+			struct i915_mm_struct *mm;
+			struct i915_mmu_object *mmu_object;
 			struct work_struct *work;
 		} userptr;
 	};
diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index ba7f5c6bb50d..ad55b06a3cb1 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -1590,10 +1590,13 @@ unlock:
 out:
 	switch (ret) {
 	case -EIO:
-		/* If this -EIO is due to a gpu hang, give the reset code a
-		 * chance to clean up the mess. Otherwise return the proper
-		 * SIGBUS. */
-		if (i915_terminally_wedged(&dev_priv->gpu_error)) {
+		/*
+		 * We eat errors when the gpu is terminally wedged to avoid
+		 * userspace unduly crashing (gl has no provisions for mmaps to
+		 * fail). But any other -EIO isn't ours (e.g. swap in failure)
+		 * and so needs to be reported.
+		 */
+		if (!i915_terminally_wedged(&dev_priv->gpu_error)) {
 			ret = VM_FAULT_SIGBUS;
 			break;
 		}
diff --git a/drivers/gpu/drm/i915/i915_gem_userptr.c b/drivers/gpu/drm/i915/i915_gem_userptr.c
index fe69fc837d9e..d38413997379 100644
--- a/drivers/gpu/drm/i915/i915_gem_userptr.c
+++ b/drivers/gpu/drm/i915/i915_gem_userptr.c
@@ -32,6 +32,15 @@
 #include <linux/mempolicy.h>
 #include <linux/swap.h>
 
+struct i915_mm_struct {
+	struct mm_struct *mm;
+	struct drm_device *dev;
+	struct i915_mmu_notifier *mn;
+	struct hlist_node node;
+	struct kref kref;
+	struct work_struct work;
+};
+
 #if defined(CONFIG_MMU_NOTIFIER)
 #include <linux/interval_tree.h>
 
@@ -41,16 +50,12 @@ struct i915_mmu_notifier {
 	struct mmu_notifier mn;
 	struct rb_root objects;
 	struct list_head linear;
-	struct drm_device *dev;
-	struct mm_struct *mm;
-	struct work_struct work;
-	unsigned long count;
 	unsigned long serial;
 	bool has_linear;
 };
 
 struct i915_mmu_object {
-	struct i915_mmu_notifier *mmu;
+	struct i915_mmu_notifier *mn;
 	struct interval_tree_node it;
 	struct list_head link;
 	struct drm_i915_gem_object *obj;
@@ -96,18 +101,18 @@ static void *invalidate_range__linear(struct i915_mmu_notifier *mn,
 				      unsigned long start,
 				      unsigned long end)
 {
-	struct i915_mmu_object *mmu;
+	struct i915_mmu_object *mo;
 	unsigned long serial;
 
 restart:
 	serial = mn->serial;
-	list_for_each_entry(mmu, &mn->linear, link) {
+	list_for_each_entry(mo, &mn->linear, link) {
 		struct drm_i915_gem_object *obj;
 
-		if (mmu->it.last < start || mmu->it.start > end)
+		if (mo->it.last < start || mo->it.start > end)
 			continue;
 
-		obj = mmu->obj;
+		obj = mo->obj;
 		drm_gem_object_reference(&obj->base);
 		spin_unlock(&mn->lock);
 
@@ -160,130 +165,47 @@ static const struct mmu_notifier_ops i915_gem_userptr_notifier = {
 };
 
 static struct i915_mmu_notifier *
-__i915_mmu_notifier_lookup(struct drm_device *dev, struct mm_struct *mm)
-{
-	struct drm_i915_private *dev_priv = to_i915(dev);
-	struct i915_mmu_notifier *mmu;
-
-	/* Protected by dev->struct_mutex */
-	hash_for_each_possible(dev_priv->mmu_notifiers, mmu, node, (unsigned long)mm)
-		if (mmu->mm == mm)
-			return mmu;
-
-	return NULL;
-}
-
-static struct i915_mmu_notifier *
-i915_mmu_notifier_get(struct drm_device *dev, struct mm_struct *mm)
+i915_mmu_notifier_create(struct mm_struct *mm)
 {
-	struct drm_i915_private *dev_priv = to_i915(dev);
-	struct i915_mmu_notifier *mmu;
+	struct i915_mmu_notifier *mn;
 	int ret;
 
-	lockdep_assert_held(&dev->struct_mutex);
-
-	mmu = __i915_mmu_notifier_lookup(dev, mm);
-	if (mmu)
-		return mmu;
-
-	mmu = kmalloc(sizeof(*mmu), GFP_KERNEL);
-	if (mmu == NULL)
+	mn = kmalloc(sizeof(*mn), GFP_KERNEL);
+	if (mn == NULL)
 		return ERR_PTR(-ENOMEM);
 
-	spin_lock_init(&mmu->lock);
-	mmu->dev = dev;
-	mmu->mn.ops = &i915_gem_userptr_notifier;
-	mmu->mm = mm;
-	mmu->objects = RB_ROOT;
-	mmu->count = 0;
-	mmu->serial = 1;
-	INIT_LIST_HEAD(&mmu->linear);
-	mmu->has_linear = false;
-
-	/* Protected by mmap_sem (write-lock) */
-	ret = __mmu_notifier_register(&mmu->mn, mm);
+	spin_lock_init(&mn->lock);
+	mn->mn.ops = &i915_gem_userptr_notifier;
+	mn->objects = RB_ROOT;
+	mn->serial = 1;
+	INIT_LIST_HEAD(&mn->linear);
+	mn->has_linear = false;
+
+	 /* Protected by mmap_sem (write-lock) */
+	ret = __mmu_notifier_register(&mn->mn, mm);
 	if (ret) {
-		kfree(mmu);
+		kfree(mn);
 		return ERR_PTR(ret);
 	}
 
-	/* Protected by dev->struct_mutex */
-	hash_add(dev_priv->mmu_notifiers, &mmu->node, (unsigned long)mm);
-	return mmu;
+	return mn;
 }
 
-static void
-__i915_mmu_notifier_destroy_worker(struct work_struct *work)
+static void __i915_mmu_notifier_update_serial(struct i915_mmu_notifier *mn)
 {
-	struct i915_mmu_notifier *mmu = container_of(work, typeof(*mmu), work);
-	mmu_notifier_unregister(&mmu->mn, mmu->mm);
-	kfree(mmu);
-}
-
-static void
-__i915_mmu_notifier_destroy(struct i915_mmu_notifier *mmu)
-{
-	lockdep_assert_held(&mmu->dev->struct_mutex);
-
-	/* Protected by dev->struct_mutex */
-	hash_del(&mmu->node);
-
-	/* Our lock ordering is: mmap_sem, mmu_notifier_scru, struct_mutex.
-	 * We enter the function holding struct_mutex, therefore we need
-	 * to drop our mutex prior to calling mmu_notifier_unregister in
-	 * order to prevent lock inversion (and system-wide deadlock)
-	 * between the mmap_sem and struct-mutex. Hence we defer the
-	 * unregistration to a workqueue where we hold no locks.
-	 */
-	INIT_WORK(&mmu->work, __i915_mmu_notifier_destroy_worker);
-	schedule_work(&mmu->work);
-}
-
-static void __i915_mmu_notifier_update_serial(struct i915_mmu_notifier *mmu)
-{
-	if (++mmu->serial == 0)
-		mmu->serial = 1;
-}
-
-static bool i915_mmu_notifier_has_linear(struct i915_mmu_notifier *mmu)
-{
-	struct i915_mmu_object *mn;
-
-	list_for_each_entry(mn, &mmu->linear, link)
-		if (mn->is_linear)
-			return true;
-
-	return false;
-}
-
-static void
-i915_mmu_notifier_del(struct i915_mmu_notifier *mmu,
-		      struct i915_mmu_object *mn)
-{
-	lockdep_assert_held(&mmu->dev->struct_mutex);
-
-	spin_lock(&mmu->lock);
-	list_del(&mn->link);
-	if (mn->is_linear)
-		mmu->has_linear = i915_mmu_notifier_has_linear(mmu);
-	else
-		interval_tree_remove(&mn->it, &mmu->objects);
-	__i915_mmu_notifier_update_serial(mmu);
-	spin_unlock(&mmu->lock);
-
-	/* Protected against _add() by dev->struct_mutex */
-	if (--mmu->count == 0)
-		__i915_mmu_notifier_destroy(mmu);
+	if (++mn->serial == 0)
+		mn->serial = 1;
 }
 
 static int
-i915_mmu_notifier_add(struct i915_mmu_notifier *mmu,
-		      struct i915_mmu_object *mn)
+i915_mmu_notifier_add(struct drm_device *dev,
+		      struct i915_mmu_notifier *mn,
+		      struct i915_mmu_object *mo)
 {
 	struct interval_tree_node *it;
 	int ret;
 
-	ret = i915_mutex_lock_interruptible(mmu->dev);
+	ret = i915_mutex_lock_interruptible(dev);
 	if (ret)
 		return ret;
 
@@ -291,11 +213,11 @@ i915_mmu_notifier_add(struct i915_mmu_notifier *mmu,
 	 * remove the objects from the interval tree) before we do
 	 * the check for overlapping objects.
 	 */
-	i915_gem_retire_requests(mmu->dev);
+	i915_gem_retire_requests(dev);
 
-	spin_lock(&mmu->lock);
-	it = interval_tree_iter_first(&mmu->objects,
-				      mn->it.start, mn->it.last);
+	spin_lock(&mn->lock);
+	it = interval_tree_iter_first(&mn->objects,
+				      mo->it.start, mo->it.last);
 	if (it) {
 		struct drm_i915_gem_object *obj;
 
@@ -312,86 +234,122 @@ i915_mmu_notifier_add(struct i915_mmu_notifier *mmu,
 
 		obj = container_of(it, struct i915_mmu_object, it)->obj;
 		if (!obj->userptr.workers)
-			mmu->has_linear = mn->is_linear = true;
+			mn->has_linear = mo->is_linear = true;
 		else
 			ret = -EAGAIN;
 	} else
-		interval_tree_insert(&mn->it, &mmu->objects);
+		interval_tree_insert(&mo->it, &mn->objects);
 
 	if (ret == 0) {
-		list_add(&mn->link, &mmu->linear);
-		__i915_mmu_notifier_update_serial(mmu);
+		list_add(&mo->link, &mn->linear);
+		__i915_mmu_notifier_update_serial(mn);
 	}
-	spin_unlock(&mmu->lock);
-	mutex_unlock(&mmu->dev->struct_mutex);
+	spin_unlock(&mn->lock);
+	mutex_unlock(&dev->struct_mutex);
 
 	return ret;
 }
 
+static bool i915_mmu_notifier_has_linear(struct i915_mmu_notifier *mn)
+{
+	struct i915_mmu_object *mo;
+
+	list_for_each_entry(mo, &mn->linear, link)
+		if (mo->is_linear)
+			return true;
+
+	return false;
+}
+
+static void
+i915_mmu_notifier_del(struct i915_mmu_notifier *mn,
+		      struct i915_mmu_object *mo)
+{
+	spin_lock(&mn->lock);
+	list_del(&mo->link);
+	if (mo->is_linear)
+		mn->has_linear = i915_mmu_notifier_has_linear(mn);
+	else
+		interval_tree_remove(&mo->it, &mn->objects);
+	__i915_mmu_notifier_update_serial(mn);
+	spin_unlock(&mn->lock);
+}
+
 static void
 i915_gem_userptr_release__mmu_notifier(struct drm_i915_gem_object *obj)
 {
-	struct i915_mmu_object *mn;
+	struct i915_mmu_object *mo;
 
-	mn = obj->userptr.mn;
-	if (mn == NULL)
+	mo = obj->userptr.mmu_object;
+	if (mo == NULL)
 		return;
 
-	i915_mmu_notifier_del(mn->mmu, mn);
-	obj->userptr.mn = NULL;
+	i915_mmu_notifier_del(mo->mn, mo);
+	kfree(mo);
+
+	obj->userptr.mmu_object = NULL;
+}
+
+static struct i915_mmu_notifier *
+i915_mmu_notifier_find(struct i915_mm_struct *mm)
+{
+	if (mm->mn == NULL) {
+		down_write(&mm->mm->mmap_sem);
+		mutex_lock(&to_i915(mm->dev)->mm_lock);
+		if (mm->mn == NULL)
+			mm->mn = i915_mmu_notifier_create(mm->mm);
+		mutex_unlock(&to_i915(mm->dev)->mm_lock);
+		up_write(&mm->mm->mmap_sem);
+	}
+	return mm->mn;
 }
 
 static int
 i915_gem_userptr_init__mmu_notifier(struct drm_i915_gem_object *obj,
 				    unsigned flags)
 {
-	struct i915_mmu_notifier *mmu;
-	struct i915_mmu_object *mn;
+	struct i915_mmu_notifier *mn;
+	struct i915_mmu_object *mo;
 	int ret;
 
 	if (flags & I915_USERPTR_UNSYNCHRONIZED)
 		return capable(CAP_SYS_ADMIN) ? 0 : -EPERM;
 
-	down_write(&obj->userptr.mm->mmap_sem);
-	ret = i915_mutex_lock_interruptible(obj->base.dev);
-	if (ret == 0) {
-		mmu = i915_mmu_notifier_get(obj->base.dev, obj->userptr.mm);
-		if (!IS_ERR(mmu))
-			mmu->count++; /* preemptive add to act as a refcount */
-		else
-			ret = PTR_ERR(mmu);
-		mutex_unlock(&obj->base.dev->struct_mutex);
-	}
-	up_write(&obj->userptr.mm->mmap_sem);
-	if (ret)
-		return ret;
+	if (WARN_ON(obj->userptr.mm == NULL))
+		return -EINVAL;
 
-	mn = kzalloc(sizeof(*mn), GFP_KERNEL);
-	if (mn == NULL) {
-		ret = -ENOMEM;
-		goto destroy_mmu;
-	}
+	mn = i915_mmu_notifier_find(obj->userptr.mm);
+	if (IS_ERR(mn))
+		return PTR_ERR(mn);
 
-	mn->mmu = mmu;
-	mn->it.start = obj->userptr.ptr;
-	mn->it.last = mn->it.start + obj->base.size - 1;
-	mn->obj = obj;
+	mo = kzalloc(sizeof(*mo), GFP_KERNEL);
+	if (mo == NULL)
+		return -ENOMEM;
 
-	ret = i915_mmu_notifier_add(mmu, mn);
-	if (ret)
-		goto free_mn;
+	mo->mn = mn;
+	mo->it.start = obj->userptr.ptr;
+	mo->it.last = mo->it.start + obj->base.size - 1;
+	mo->obj = obj;
 
-	obj->userptr.mn = mn;
+	ret = i915_mmu_notifier_add(obj->base.dev, mn, mo);
+	if (ret) {
+		kfree(mo);
+		return ret;
+	}
+
+	obj->userptr.mmu_object = mo;
 	return 0;
+}
+
+static void
+i915_mmu_notifier_free(struct i915_mmu_notifier *mn,
+		       struct mm_struct *mm)
+{
+	if (mn == NULL)
+		return;
 
-free_mn:
+	mmu_notifier_unregister(&mn->mn, mm);
 	kfree(mn);
-destroy_mmu:
-	mutex_lock(&obj->base.dev->struct_mutex);
-	if (--mmu->count == 0)
-		__i915_mmu_notifier_destroy(mmu);
-	mutex_unlock(&obj->base.dev->struct_mutex);
-	return ret;
 }
 
 #else
@@ -413,15 +371,114 @@ i915_gem_userptr_init__mmu_notifier(struct drm_i915_gem_object *obj,
 
 	return 0;
 }
+
+static void
+i915_mmu_notifier_free(struct i915_mmu_notifier *mn,
+		       struct mm_struct *mm)
+{
+}
+
 #endif
 
+static struct i915_mm_struct *
+__i915_mm_struct_find(struct drm_i915_private *dev_priv, struct mm_struct *real)
+{
+	struct i915_mm_struct *mm;
+
+	/* Protected by dev_priv->mm_lock */
+	hash_for_each_possible(dev_priv->mm_structs, mm, node, (unsigned long)real)
+		if (mm->mm == real)
+			return mm;
+
+	return NULL;
+}
+
+static int
+i915_gem_userptr_init__mm_struct(struct drm_i915_gem_object *obj)
+{
+	struct drm_i915_private *dev_priv = to_i915(obj->base.dev);
+	struct i915_mm_struct *mm;
+	int ret = 0;
+
+	/* During release of the GEM object we hold the struct_mutex. This
+	 * precludes us from calling mmput() at that time as that may be
+	 * the last reference and so call exit_mmap(). exit_mmap() will
+	 * attempt to reap the vma, and if we were holding a GTT mmap
+	 * would then call drm_gem_vm_close() and attempt to reacquire
+	 * the struct mutex. So in order to avoid that recursion, we have
+	 * to defer releasing the mm reference until after we drop the
+	 * struct_mutex, i.e. we need to schedule a worker to do the clean
+	 * up.
+	 */
+	mutex_lock(&dev_priv->mm_lock);
+	mm = __i915_mm_struct_find(dev_priv, current->mm);
+	if (mm == NULL) {
+		mm = kmalloc(sizeof(*mm), GFP_KERNEL);
+		if (mm == NULL) {
+			ret = -ENOMEM;
+			goto out;
+		}
+
+		kref_init(&mm->kref);
+		mm->dev = obj->base.dev;
+
+		mm->mm = current->mm;
+		atomic_inc(&current->mm->mm_count);
+
+		mm->mn = NULL;
+
+		/* Protected by dev_priv->mm_lock */
+		hash_add(dev_priv->mm_structs,
+			 &mm->node, (unsigned long)mm->mm);
+	} else
+		kref_get(&mm->kref);
+
+	obj->userptr.mm = mm;
+out:
+	mutex_unlock(&dev_priv->mm_lock);
+	return ret;
+}
+
+static void
+__i915_mm_struct_free__worker(struct work_struct *work)
+{
+	struct i915_mm_struct *mm = container_of(work, typeof(*mm), work);
+	i915_mmu_notifier_free(mm->mn, mm->mm);
+	mmdrop(mm->mm);
+	kfree(mm);
+}
+
+static void
+__i915_mm_struct_free(struct kref *kref)
+{
+	struct i915_mm_struct *mm = container_of(kref, typeof(*mm), kref);
+
+	/* Protected by dev_priv->mm_lock */
+	hash_del(&mm->node);
+	mutex_unlock(&to_i915(mm->dev)->mm_lock);
+
+	INIT_WORK(&mm->work, __i915_mm_struct_free__worker);
+	schedule_work(&mm->work);
+}
+
+static void
+i915_gem_userptr_release__mm_struct(struct drm_i915_gem_object *obj)
+{
+	if (obj->userptr.mm == NULL)
+		return;
+
+	kref_put_mutex(&obj->userptr.mm->kref,
+		       __i915_mm_struct_free,
+		       &to_i915(obj->base.dev)->mm_lock);
+	obj->userptr.mm = NULL;
+}
+
 struct get_pages_work {
 	struct work_struct work;
 	struct drm_i915_gem_object *obj;
 	struct task_struct *task;
 };
 
-
 #if IS_ENABLED(CONFIG_SWIOTLB)
 #define swiotlb_active() swiotlb_nr_tbl()
 #else
@@ -479,7 +536,7 @@ __i915_gem_userptr_get_pages_worker(struct work_struct *_work)
 	if (pvec == NULL)
 		pvec = drm_malloc_ab(num_pages, sizeof(struct page *));
 	if (pvec != NULL) {
-		struct mm_struct *mm = obj->userptr.mm;
+		struct mm_struct *mm = obj->userptr.mm->mm;
 
 		down_read(&mm->mmap_sem);
 		while (pinned < num_pages) {
@@ -545,7 +602,7 @@ i915_gem_userptr_get_pages(struct drm_i915_gem_object *obj)
 
 	pvec = NULL;
 	pinned = 0;
-	if (obj->userptr.mm == current->mm) {
+	if (obj->userptr.mm->mm == current->mm) {
 		pvec = kmalloc(num_pages*sizeof(struct page *),
 			       GFP_TEMPORARY | __GFP_NOWARN | __GFP_NORETRY);
 		if (pvec == NULL) {
@@ -651,17 +708,13 @@ static void
 i915_gem_userptr_release(struct drm_i915_gem_object *obj)
 {
 	i915_gem_userptr_release__mmu_notifier(obj);
-
-	if (obj->userptr.mm) {
-		mmput(obj->userptr.mm);
-		obj->userptr.mm = NULL;
-	}
+	i915_gem_userptr_release__mm_struct(obj);
 }
 
 static int
 i915_gem_userptr_dmabuf_export(struct drm_i915_gem_object *obj)
 {
-	if (obj->userptr.mn)
+	if (obj->userptr.mmu_object)
 		return 0;
 
 	return i915_gem_userptr_init__mmu_notifier(obj, 0);
@@ -736,7 +789,6 @@ i915_gem_userptr_ioctl(struct drm_device *dev, void *data, struct drm_file *file
 		return -ENODEV;
 	}
 
-	/* Allocate the new object */
 	obj = i915_gem_object_alloc(dev);
 	if (obj == NULL)
 		return -ENOMEM;
@@ -754,8 +806,8 @@ i915_gem_userptr_ioctl(struct drm_device *dev, void *data, struct drm_file *file
 	 * at binding. This means that we need to hook into the mmu_notifier
 	 * in order to detect if the mmu is destroyed.
 	 */
-	ret = -ENOMEM;
-	if ((obj->userptr.mm = get_task_mm(current)))
+	ret = i915_gem_userptr_init__mm_struct(obj);
+	if (ret == 0)
 		ret = i915_gem_userptr_init__mmu_notifier(obj, args->flags);
 	if (ret == 0)
 		ret = drm_gem_handle_create(file, &obj->base, &handle);
@@ -772,9 +824,8 @@ i915_gem_userptr_ioctl(struct drm_device *dev, void *data, struct drm_file *file
 int
 i915_gem_init_userptr(struct drm_device *dev)
 {
-#if defined(CONFIG_MMU_NOTIFIER)
 	struct drm_i915_private *dev_priv = to_i915(dev);
-	hash_init(dev_priv->mmu_notifiers);
-#endif
+	mutex_init(&dev_priv->mm_lock);
+	hash_init(dev_priv->mm_structs);
 	return 0;
 }
diff --git a/drivers/gpu/drm/i915/i915_reg.h b/drivers/gpu/drm/i915/i915_reg.h
index e4d7607da2c4..f29b44c86a2f 100644
--- a/drivers/gpu/drm/i915/i915_reg.h
+++ b/drivers/gpu/drm/i915/i915_reg.h
@@ -334,16 +334,20 @@
 #define GFX_OP_DESTBUFFER_INFO	 ((0x3<<29)|(0x1d<<24)|(0x8e<<16)|1)
 #define GFX_OP_DRAWRECT_INFO     ((0x3<<29)|(0x1d<<24)|(0x80<<16)|(0x3))
 #define GFX_OP_DRAWRECT_INFO_I965  ((0x7900<<16)|0x2)
-#define SRC_COPY_BLT_CMD                ((2<<29)|(0x43<<22)|4)
+
+#define COLOR_BLT_CMD			(2<<29 | 0x40<<22 | (5-2))
+#define SRC_COPY_BLT_CMD		((2<<29)|(0x43<<22)|4)
 #define XY_SRC_COPY_BLT_CMD		((2<<29)|(0x53<<22)|6)
 #define XY_MONO_SRC_COPY_IMM_BLT	((2<<29)|(0x71<<22)|5)
-#define XY_SRC_COPY_BLT_WRITE_ALPHA	(1<<21)
-#define XY_SRC_COPY_BLT_WRITE_RGB	(1<<20)
+#define   BLT_WRITE_A			(2<<20)
+#define   BLT_WRITE_RGB			(1<<20)
+#define   BLT_WRITE_RGBA		(BLT_WRITE_RGB | BLT_WRITE_A)
 #define   BLT_DEPTH_8			(0<<24)
 #define   BLT_DEPTH_16_565		(1<<24)
 #define   BLT_DEPTH_16_1555		(2<<24)
 #define   BLT_DEPTH_32			(3<<24)
-#define   BLT_ROP_GXCOPY		(0xcc<<16)
+#define   BLT_ROP_SRC_COPY		(0xcc<<16)
+#define   BLT_ROP_COLOR_COPY		(0xf0<<16)
 #define XY_SRC_COPY_BLT_SRC_TILED	(1<<15) /* 965+ only */
 #define XY_SRC_COPY_BLT_DST_TILED	(1<<11) /* 965+ only */
 #define CMD_OP_DISPLAYBUFFER_INFO ((0x0<<29)|(0x14<<23)|2)
diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.c b/drivers/gpu/drm/i915/intel_ringbuffer.c
index 16371a444426..2d068edd1adc 100644
--- a/drivers/gpu/drm/i915/intel_ringbuffer.c
+++ b/drivers/gpu/drm/i915/intel_ringbuffer.c
@@ -1363,54 +1363,66 @@ i965_dispatch_execbuffer(struct intel_engine_cs *ring,
 
 /* Just userspace ABI convention to limit the wa batch bo to a resonable size */
 #define I830_BATCH_LIMIT (256*1024)
+#define I830_TLB_ENTRIES (2)
+#define I830_WA_SIZE max(I830_TLB_ENTRIES*4096, I830_BATCH_LIMIT)
 static int
 i830_dispatch_execbuffer(struct intel_engine_cs *ring,
 				u64 offset, u32 len,
 				unsigned flags)
 {
+	u32 cs_offset = ring->scratch.gtt_offset;
 	int ret;
 
-	if (flags & I915_DISPATCH_PINNED) {
-		ret = intel_ring_begin(ring, 4);
-		if (ret)
-			return ret;
+	ret = intel_ring_begin(ring, 6);
+	if (ret)
+		return ret;
 
-		intel_ring_emit(ring, MI_BATCH_BUFFER);
-		intel_ring_emit(ring, offset | (flags & I915_DISPATCH_SECURE ? 0 : MI_BATCH_NON_SECURE));
-		intel_ring_emit(ring, offset + len - 8);
-		intel_ring_emit(ring, MI_NOOP);
-		intel_ring_advance(ring);
-	} else {
-		u32 cs_offset = ring->scratch.gtt_offset;
+	/* Evict the invalid PTE TLBs */
+	intel_ring_emit(ring, COLOR_BLT_CMD | BLT_WRITE_RGBA);
+	intel_ring_emit(ring, BLT_DEPTH_32 | BLT_ROP_COLOR_COPY | 4096);
+	intel_ring_emit(ring, I830_TLB_ENTRIES << 16 | 4); /* load each page */
+	intel_ring_emit(ring, cs_offset);
+	intel_ring_emit(ring, 0xdeadbeef);
+	intel_ring_emit(ring, MI_NOOP);
+	intel_ring_advance(ring);
 
+	if ((flags & I915_DISPATCH_PINNED) == 0) {
 		if (len > I830_BATCH_LIMIT)
 			return -ENOSPC;
 
-		ret = intel_ring_begin(ring, 9+3);
+		ret = intel_ring_begin(ring, 6 + 2);
 		if (ret)
 			return ret;
-		/* Blit the batch (which has now all relocs applied) to the stable batch
-		 * scratch bo area (so that the CS never stumbles over its tlb
-		 * invalidation bug) ... */
-		intel_ring_emit(ring, XY_SRC_COPY_BLT_CMD |
-				XY_SRC_COPY_BLT_WRITE_ALPHA |
-				XY_SRC_COPY_BLT_WRITE_RGB);
-		intel_ring_emit(ring, BLT_DEPTH_32 | BLT_ROP_GXCOPY | 4096);
-		intel_ring_emit(ring, 0);
-		intel_ring_emit(ring, (DIV_ROUND_UP(len, 4096) << 16) | 1024);
+
+		/* Blit the batch (which has now all relocs applied) to the
+		 * stable batch scratch bo area (so that the CS never
+		 * stumbles over its tlb invalidation bug) ...
+		 */
+		intel_ring_emit(ring, SRC_COPY_BLT_CMD | BLT_WRITE_RGBA);
+		intel_ring_emit(ring, BLT_DEPTH_32 | BLT_ROP_SRC_COPY | 4096);
+		intel_ring_emit(ring, DIV_ROUND_UP(len, 4096) << 16 | 1024);
 		intel_ring_emit(ring, cs_offset);
-		intel_ring_emit(ring, 0);
 		intel_ring_emit(ring, 4096);
 		intel_ring_emit(ring, offset);
+
 		intel_ring_emit(ring, MI_FLUSH);
+		intel_ring_emit(ring, MI_NOOP);
+		intel_ring_advance(ring);
 
 		/* ... and execute it. */
-		intel_ring_emit(ring, MI_BATCH_BUFFER);
-		intel_ring_emit(ring, cs_offset | (flags & I915_DISPATCH_SECURE ? 0 : MI_BATCH_NON_SECURE));
-		intel_ring_emit(ring, cs_offset + len - 8);
-		intel_ring_advance(ring);
+		offset = cs_offset;
 	}
 
+	ret = intel_ring_begin(ring, 4);
+	if (ret)
+		return ret;
+
+	intel_ring_emit(ring, MI_BATCH_BUFFER);
+	intel_ring_emit(ring, offset | (flags & I915_DISPATCH_SECURE ? 0 : MI_BATCH_NON_SECURE));
+	intel_ring_emit(ring, offset + len - 8);
+	intel_ring_emit(ring, MI_NOOP);
+	intel_ring_advance(ring);
+
 	return 0;
 }
 
@@ -2200,7 +2212,7 @@ int intel_init_render_ring_buffer(struct drm_device *dev)
 
 	/* Workaround batchbuffer to combat CS tlb bug. */
 	if (HAS_BROKEN_CS_TLB(dev)) {
-		obj = i915_gem_alloc_object(dev, I830_BATCH_LIMIT);
+		obj = i915_gem_alloc_object(dev, I830_WA_SIZE);
 		if (obj == NULL) {
 			DRM_ERROR("Failed to allocate batch bo\n");
 			return -ENOMEM;
diff --git a/drivers/gpu/drm/i915/intel_tv.c b/drivers/gpu/drm/i915/intel_tv.c
index c69d3ce1b3d6..c14341ca3ef9 100644
--- a/drivers/gpu/drm/i915/intel_tv.c
+++ b/drivers/gpu/drm/i915/intel_tv.c
@@ -854,6 +854,10 @@ intel_enable_tv(struct intel_encoder *encoder)
 	struct drm_device *dev = encoder->base.dev;
 	struct drm_i915_private *dev_priv = dev->dev_private;
 
+	/* Prevents vblank waits from timing out in intel_tv_detect_type() */
+	intel_wait_for_vblank(encoder->base.dev,
+			      to_intel_crtc(encoder->base.crtc)->pipe);
+
 	I915_WRITE(TV_CTL, I915_READ(TV_CTL) | TV_ENC_ENABLE);
 }
 
diff --git a/drivers/gpu/drm/msm/hdmi/hdmi.c b/drivers/gpu/drm/msm/hdmi/hdmi.c
index a125a7e32742..c6c9b02e0ada 100644
--- a/drivers/gpu/drm/msm/hdmi/hdmi.c
+++ b/drivers/gpu/drm/msm/hdmi/hdmi.c
@@ -258,28 +258,30 @@ static void set_hdmi_pdev(struct drm_device *dev,
 	priv->hdmi_pdev = pdev;
 }
 
+#ifdef CONFIG_OF
+static int get_gpio(struct device *dev, struct device_node *of_node, const char *name)
+{
+	int gpio = of_get_named_gpio(of_node, name, 0);
+	if (gpio < 0) {
+		char name2[32];
+		snprintf(name2, sizeof(name2), "%s-gpio", name);
+		gpio = of_get_named_gpio(of_node, name2, 0);
+		if (gpio < 0) {
+			dev_err(dev, "failed to get gpio: %s (%d)\n",
+					name, gpio);
+			gpio = -1;
+		}
+	}
+	return gpio;
+}
+#endif
+
 static int hdmi_bind(struct device *dev, struct device *master, void *data)
 {
 	static struct hdmi_platform_config config = {};
 #ifdef CONFIG_OF
 	struct device_node *of_node = dev->of_node;
 
-	int get_gpio(const char *name)
-	{
-		int gpio = of_get_named_gpio(of_node, name, 0);
-		if (gpio < 0) {
-			char name2[32];
-			snprintf(name2, sizeof(name2), "%s-gpio", name);
-			gpio = of_get_named_gpio(of_node, name2, 0);
-			if (gpio < 0) {
-				dev_err(dev, "failed to get gpio: %s (%d)\n",
-						name, gpio);
-				gpio = -1;
-			}
-		}
-		return gpio;
-	}
-
 	if (of_device_is_compatible(of_node, "qcom,hdmi-tx-8074")) {
 		static const char *hpd_reg_names[] = {"hpd-gdsc", "hpd-5v"};
 		static const char *pwr_reg_names[] = {"core-vdda", "core-vcc"};
@@ -312,12 +314,12 @@ static int hdmi_bind(struct device *dev, struct device *master, void *data)
 	}
 
 	config.mmio_name     = "core_physical";
-	config.ddc_clk_gpio  = get_gpio("qcom,hdmi-tx-ddc-clk");
-	config.ddc_data_gpio = get_gpio("qcom,hdmi-tx-ddc-data");
-	config.hpd_gpio      = get_gpio("qcom,hdmi-tx-hpd");
-	config.mux_en_gpio   = get_gpio("qcom,hdmi-tx-mux-en");
-	config.mux_sel_gpio  = get_gpio("qcom,hdmi-tx-mux-sel");
-	config.mux_lpm_gpio  = get_gpio("qcom,hdmi-tx-mux-lpm");
+	config.ddc_clk_gpio  = get_gpio(dev, of_node, "qcom,hdmi-tx-ddc-clk");
+	config.ddc_data_gpio = get_gpio(dev, of_node, "qcom,hdmi-tx-ddc-data");
+	config.hpd_gpio      = get_gpio(dev, of_node, "qcom,hdmi-tx-hpd");
+	config.mux_en_gpio   = get_gpio(dev, of_node, "qcom,hdmi-tx-mux-en");
+	config.mux_sel_gpio  = get_gpio(dev, of_node, "qcom,hdmi-tx-mux-sel");
+	config.mux_lpm_gpio  = get_gpio(dev, of_node, "qcom,hdmi-tx-mux-lpm");
 
 #else
 	static const char *hpd_clk_names[] = {
diff --git a/drivers/gpu/drm/msm/hdmi/hdmi_phy_8960.c b/drivers/gpu/drm/msm/hdmi/hdmi_phy_8960.c
index 902d7685d441..f408b69486a8 100644
--- a/drivers/gpu/drm/msm/hdmi/hdmi_phy_8960.c
+++ b/drivers/gpu/drm/msm/hdmi/hdmi_phy_8960.c
@@ -15,19 +15,25 @@
  * this program.  If not, see <http://www.gnu.org/licenses/>.
  */
 
+#ifdef CONFIG_COMMON_CLK
 #include <linux/clk.h>
 #include <linux/clk-provider.h>
+#endif
 
 #include "hdmi.h"
 
 struct hdmi_phy_8960 {
 	struct hdmi_phy base;
 	struct hdmi *hdmi;
+#ifdef CONFIG_COMMON_CLK
 	struct clk_hw pll_hw;
 	struct clk *pll;
 	unsigned long pixclk;
+#endif
 };
 #define to_hdmi_phy_8960(x) container_of(x, struct hdmi_phy_8960, base)
+
+#ifdef CONFIG_COMMON_CLK
 #define clk_to_phy(x) container_of(x, struct hdmi_phy_8960, pll_hw)
 
 /*
@@ -374,7 +380,7 @@ static struct clk_init_data pll_init = {
 	.parent_names = hdmi_pll_parents,
 	.num_parents = ARRAY_SIZE(hdmi_pll_parents),
 };
-
+#endif
 
 /*
  * HDMI Phy:
@@ -480,12 +486,15 @@ struct hdmi_phy *hdmi_phy_8960_init(struct hdmi *hdmi)
 {
 	struct hdmi_phy_8960 *phy_8960;
 	struct hdmi_phy *phy = NULL;
-	int ret, i;
+	int ret;
+#ifdef CONFIG_COMMON_CLK
+	int i;
 
 	/* sanity check: */
 	for (i = 0; i < (ARRAY_SIZE(freqtbl) - 1); i++)
 		if (WARN_ON(freqtbl[i].rate < freqtbl[i+1].rate))
 			return ERR_PTR(-EINVAL);
+#endif
 
 	phy_8960 = kzalloc(sizeof(*phy_8960), GFP_KERNEL);
 	if (!phy_8960) {
@@ -499,6 +508,7 @@ struct hdmi_phy *hdmi_phy_8960_init(struct hdmi *hdmi)
 
 	phy_8960->hdmi = hdmi;
 
+#ifdef CONFIG_COMMON_CLK
 	phy_8960->pll_hw.init = &pll_init;
 	phy_8960->pll = devm_clk_register(hdmi->dev->dev, &phy_8960->pll_hw);
 	if (IS_ERR(phy_8960->pll)) {
@@ -506,6 +516,7 @@ struct hdmi_phy *hdmi_phy_8960_init(struct hdmi *hdmi)
 		phy_8960->pll = NULL;
 		goto fail;
 	}
+#endif
 
 	return phy;
 
diff --git a/drivers/gpu/drm/msm/msm_drv.c b/drivers/gpu/drm/msm/msm_drv.c
index 26ee80db17af..fcf95680413d 100644
--- a/drivers/gpu/drm/msm/msm_drv.c
+++ b/drivers/gpu/drm/msm/msm_drv.c
@@ -52,7 +52,7 @@ module_param(reglog, bool, 0600);
 #define reglog 0
 #endif
 
-static char *vram;
+static char *vram = "16m";
 MODULE_PARM_DESC(vram, "Configure VRAM size (for devices without IOMMU/GPUMMU");
 module_param(vram, charp, 0);
 
diff --git a/drivers/gpu/drm/radeon/atombios_dp.c b/drivers/gpu/drm/radeon/atombios_dp.c
index b1e11f8434e2..ac14b67621d3 100644
--- a/drivers/gpu/drm/radeon/atombios_dp.c
+++ b/drivers/gpu/drm/radeon/atombios_dp.c
@@ -405,16 +405,13 @@ bool radeon_dp_getdpcd(struct radeon_connector *radeon_connector)
 	u8 msg[DP_DPCD_SIZE];
 	int ret;
 
-	char dpcd_hex_dump[DP_DPCD_SIZE * 3];
-
 	ret = drm_dp_dpcd_read(&radeon_connector->ddc_bus->aux, DP_DPCD_REV, msg,
 			       DP_DPCD_SIZE);
 	if (ret > 0) {
 		memcpy(dig_connector->dpcd, msg, DP_DPCD_SIZE);
 
-		hex_dump_to_buffer(dig_connector->dpcd, sizeof(dig_connector->dpcd),
-				   32, 1, dpcd_hex_dump, sizeof(dpcd_hex_dump), false);
-		DRM_DEBUG_KMS("DPCD: %s\n", dpcd_hex_dump);
+		DRM_DEBUG_KMS("DPCD: %*ph\n", (int)sizeof(dig_connector->dpcd),
+			      dig_connector->dpcd);
 
 		radeon_dp_probe_oui(radeon_connector);
 
diff --git a/drivers/gpu/drm/radeon/r600.c b/drivers/gpu/drm/radeon/r600.c
index e616eb5f6e7a..3cfb50056f7a 100644
--- a/drivers/gpu/drm/radeon/r600.c
+++ b/drivers/gpu/drm/radeon/r600.c
@@ -2769,8 +2769,8 @@ bool r600_semaphore_ring_emit(struct radeon_device *rdev,
 	radeon_ring_write(ring, lower_32_bits(addr));
 	radeon_ring_write(ring, (upper_32_bits(addr) & 0xff) | sel);
 
-	/* PFP_SYNC_ME packet only exists on 7xx+ */
-	if (emit_wait && (rdev->family >= CHIP_RV770)) {
+	/* PFP_SYNC_ME packet only exists on 7xx+, only enable it on eg+ */
+	if (emit_wait && (rdev->family >= CHIP_CEDAR)) {
 		/* Prevent the PFP from running ahead of the semaphore wait */
 		radeon_ring_write(ring, PACKET3(PACKET3_PFP_SYNC_ME, 0));
 		radeon_ring_write(ring, 0x0);
diff --git a/drivers/gpu/drm/radeon/radeon_atombios.c b/drivers/gpu/drm/radeon/radeon_atombios.c
index 92b2d8dd4735..e74c7e387dde 100644
--- a/drivers/gpu/drm/radeon/radeon_atombios.c
+++ b/drivers/gpu/drm/radeon/radeon_atombios.c
@@ -447,6 +447,13 @@ static bool radeon_atom_apply_quirks(struct drm_device *dev,
 		}
 	}
 
+	/* Fujitsu D3003-S2 board lists DVI-I as DVI-I and VGA */
+	if ((dev->pdev->device == 0x9805) &&
+	    (dev->pdev->subsystem_vendor == 0x1734) &&
+	    (dev->pdev->subsystem_device == 0x11bd)) {
+		if (*connector_type == DRM_MODE_CONNECTOR_VGA)
+			return false;
+	}
 
 	return true;
 }
@@ -2281,19 +2288,31 @@ static void radeon_atombios_add_pplib_thermal_controller(struct radeon_device *r
 				 (controller->ucFanParameters &
 				  ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with");
 			rdev->pm.int_thermal_type = THERMAL_TYPE_KV;
-		} else if ((controller->ucType ==
-			    ATOM_PP_THERMALCONTROLLER_EXTERNAL_GPIO) ||
-			   (controller->ucType ==
-			    ATOM_PP_THERMALCONTROLLER_ADT7473_WITH_INTERNAL) ||
-			   (controller->ucType ==
-			    ATOM_PP_THERMALCONTROLLER_EMC2103_WITH_INTERNAL)) {
-			DRM_INFO("Special thermal controller config\n");
+		} else if (controller->ucType ==
+			   ATOM_PP_THERMALCONTROLLER_EXTERNAL_GPIO) {
+			DRM_INFO("External GPIO thermal controller %s fan control\n",
+				 (controller->ucFanParameters &
+				  ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with");
+			rdev->pm.int_thermal_type = THERMAL_TYPE_EXTERNAL_GPIO;
+		} else if (controller->ucType ==
+			   ATOM_PP_THERMALCONTROLLER_ADT7473_WITH_INTERNAL) {
+			DRM_INFO("ADT7473 with internal thermal controller %s fan control\n",
+				 (controller->ucFanParameters &
+				  ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with");
+			rdev->pm.int_thermal_type = THERMAL_TYPE_ADT7473_WITH_INTERNAL;
+		} else if (controller->ucType ==
+			   ATOM_PP_THERMALCONTROLLER_EMC2103_WITH_INTERNAL) {
+			DRM_INFO("EMC2103 with internal thermal controller %s fan control\n",
+				 (controller->ucFanParameters &
+				  ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with");
+			rdev->pm.int_thermal_type = THERMAL_TYPE_EMC2103_WITH_INTERNAL;
 		} else if (controller->ucType < ARRAY_SIZE(pp_lib_thermal_controller_names)) {
 			DRM_INFO("Possible %s thermal controller at 0x%02x %s fan control\n",
 				 pp_lib_thermal_controller_names[controller->ucType],
 				 controller->ucI2cAddress >> 1,
 				 (controller->ucFanParameters &
 				  ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with");
+			rdev->pm.int_thermal_type = THERMAL_TYPE_EXTERNAL;
 			i2c_bus = radeon_lookup_i2c_gpio(rdev, controller->ucI2cLine);
 			rdev->pm.i2c_bus = radeon_i2c_lookup(rdev, &i2c_bus);
 			if (rdev->pm.i2c_bus) {
diff --git a/drivers/gpu/drm/radeon/radeon_semaphore.c b/drivers/gpu/drm/radeon/radeon_semaphore.c
index 56d9fd66d8ae..abd6753a570a 100644
--- a/drivers/gpu/drm/radeon/radeon_semaphore.c
+++ b/drivers/gpu/drm/radeon/radeon_semaphore.c
@@ -34,7 +34,7 @@
 int radeon_semaphore_create(struct radeon_device *rdev,
 			    struct radeon_semaphore **semaphore)
 {
-	uint32_t *cpu_addr;
+	uint64_t *cpu_addr;
 	int i, r;
 
 	*semaphore = kmalloc(sizeof(struct radeon_semaphore), GFP_KERNEL);