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

Pull drm fixes from Dave Airlie:
 "Just back from LCA + some days off, had some fixes from the past 2 weeks,

  Some amdkfd code removal for a feature that wasn't ready, otherwise
  just one fix for core helper sleeping, exynos, i915, and radeon fixes.

  I thought I had some sti fixes but they were already in, and it
  confused me for a few mins this morning"

* 'drm-fixes' of git://people.freedesktop.org/~airlied/linux:
  drm: fb helper should avoid sleeping in panic context
  drm/exynos: fix warning of vblank reference count
  drm/exynos: remove unnecessary runtime pm operations
  drm/exynos: fix reset codes for memory mapped hdmi phy
  drm/radeon: use rv515_ring_start on r5xx
  drm/radeon: add si dpm quirk list
  drm/radeon: don't print error on -ERESTARTSYS
  drm/i915: Fix mutex->owner inspection race under DEBUG_MUTEXES
  drm/i915: Ban Haswell from using RCS flips
  drm/i915: vlv: sanitize RPS interrupt mask during GPU idling
  drm/i915: fix HW lockup due to missing RPS IRQ workaround on GEN6
  drm/i915: gen9: fix RPS interrupt routing to CPU vs. GT
  drm/exynos: remove the redundant machine checking code
  drm/radeon: add a dpm quirk list
  drm/amdkfd: Fix sparse warning (different address space)
  drm/radeon: fix VM flush on CIK (v3)
  drm/radeon: fix VM flush on SI (v3)
  drm/radeon: fix VM flush on cayman/aruba (v3)
  drm/amdkfd: Drop interrupt SW ring buffer

drivers/gpu/drm/amd/amdkfd/Makefile

@@ -8,7 +8,6 @@ amdkfd-y	:= kfd_module.o kfd_device.o kfd_chardev.o kfd_topology.o \
 		kfd_pasid.o kfd_doorbell.o kfd_flat_memory.o \
 		kfd_process.o kfd_queue.o kfd_mqd_manager.o \
 		kfd_kernel_queue.o kfd_packet_manager.o \
-		kfd_process_queue_manager.o kfd_device_queue_manager.o \
-		kfd_interrupt.o
+		kfd_process_queue_manager.o kfd_device_queue_manager.o
 
 obj-$(CONFIG_HSA_AMD)	+= amdkfd.o

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

@@ -192,13 +192,6 @@ bool kgd2kfd_device_init(struct kfd_dev *kfd,
 		goto kfd_topology_add_device_error;
 	}
 
-	if (kfd_interrupt_init(kfd)) {
-		dev_err(kfd_device,
-			"Error initializing interrupts for device (%x:%x)\n",
-			kfd->pdev->vendor, kfd->pdev->device);
-		goto kfd_interrupt_error;
-	}
-
 	if (!device_iommu_pasid_init(kfd)) {
 		dev_err(kfd_device,
 			"Error initializing iommuv2 for device (%x:%x)\n",
@@ -237,8 +230,6 @@ bool kgd2kfd_device_init(struct kfd_dev *kfd,
 device_queue_manager_error:
 	amd_iommu_free_device(kfd->pdev);
 device_iommu_pasid_error:
-	kfd_interrupt_exit(kfd);
-kfd_interrupt_error:
 	kfd_topology_remove_device(kfd);
 kfd_topology_add_device_error:
 	kfd2kgd->fini_sa_manager(kfd->kgd);
@@ -254,7 +245,6 @@ void kgd2kfd_device_exit(struct kfd_dev *kfd)
 	if (kfd->init_complete) {
 		device_queue_manager_uninit(kfd->dqm);
 		amd_iommu_free_device(kfd->pdev);
-		kfd_interrupt_exit(kfd);
 		kfd_topology_remove_device(kfd);
 	}
 
@@ -296,13 +286,5 @@ int kgd2kfd_resume(struct kfd_dev *kfd)
 /* This is called directly from KGD at ISR. */
 void kgd2kfd_interrupt(struct kfd_dev *kfd, const void *ih_ring_entry)
 {
-	if (kfd->init_complete) {
-		spin_lock(&kfd->interrupt_lock);
-
-		if (kfd->interrupts_active
-		    && enqueue_ih_ring_entry(kfd, ih_ring_entry))
-			schedule_work(&kfd->interrupt_work);
-
-		spin_unlock(&kfd->interrupt_lock);
-	}
+	/* Process interrupts / schedule work as necessary */
 }

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

@@ -280,7 +280,7 @@ static int create_compute_queue_nocpsch(struct device_queue_manager *dqm,
 			q->queue);
 
 	retval = mqd->load_mqd(mqd, q->mqd, q->pipe,
-			q->queue, q->properties.write_ptr);
+			q->queue, (uint32_t __user *) q->properties.write_ptr);
 	if (retval != 0) {
 		deallocate_hqd(dqm, q);
 		mqd->uninit_mqd(mqd, q->mqd, q->mqd_mem_obj);

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

-/*
- * Copyright 2014 Advanced Micro Devices, Inc.
- *
- * 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, sublicense,
- * 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 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 NONINFRINGEMENT.  IN NO EVENT SHALL
- * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) 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.
- */
-
-/*
- * KFD Interrupts.
- *
- * AMD GPUs deliver interrupts by pushing an interrupt description onto the
- * interrupt ring and then sending an interrupt. KGD receives the interrupt
- * in ISR and sends us a pointer to each new entry on the interrupt ring.
- *
- * We generally can't process interrupt-signaled events from ISR, so we call
- * out to each interrupt client module (currently only the scheduler) to ask if
- * each interrupt is interesting. If they return true, then it requires further
- * processing so we copy it to an internal interrupt ring and call each
- * interrupt client again from a work-queue.
- *
- * There's no acknowledgment for the interrupts we use. The hardware simply
- * queues a new interrupt each time without waiting.
- *
- * The fixed-size internal queue means that it's possible for us to lose
- * interrupts because we have no back-pressure to the hardware.
- */
-
-#include <linux/slab.h>
-#include <linux/device.h>
-#include "kfd_priv.h"
-
-#define KFD_INTERRUPT_RING_SIZE 256
-
-static void interrupt_wq(struct work_struct *);
-
-int kfd_interrupt_init(struct kfd_dev *kfd)
-{
-	void *interrupt_ring = kmalloc_array(KFD_INTERRUPT_RING_SIZE,
-					kfd->device_info->ih_ring_entry_size,
-					GFP_KERNEL);
-	if (!interrupt_ring)
-		return -ENOMEM;
-
-	kfd->interrupt_ring = interrupt_ring;
-	kfd->interrupt_ring_size =
-		KFD_INTERRUPT_RING_SIZE * kfd->device_info->ih_ring_entry_size;
-	atomic_set(&kfd->interrupt_ring_wptr, 0);
-	atomic_set(&kfd->interrupt_ring_rptr, 0);
-
-	spin_lock_init(&kfd->interrupt_lock);
-
-	INIT_WORK(&kfd->interrupt_work, interrupt_wq);
-
-	kfd->interrupts_active = true;
-
-	/*
-	 * After this function returns, the interrupt will be enabled. This
-	 * barrier ensures that the interrupt running on a different processor
-	 * sees all the above writes.
-	 */
-	smp_wmb();
-
-	return 0;
-}
-
-void kfd_interrupt_exit(struct kfd_dev *kfd)
-{
-	/*
-	 * Stop the interrupt handler from writing to the ring and scheduling
-	 * workqueue items. The spinlock ensures that any interrupt running
-	 * after we have unlocked sees interrupts_active = false.
-	 */
-	unsigned long flags;
-
-	spin_lock_irqsave(&kfd->interrupt_lock, flags);
-	kfd->interrupts_active = false;
-	spin_unlock_irqrestore(&kfd->interrupt_lock, flags);
-
-	/*
-	 * Flush_scheduled_work ensures that there are no outstanding
-	 * work-queue items that will access interrupt_ring. New work items
-	 * can't be created because we stopped interrupt handling above.
-	 */
-	flush_scheduled_work();
-
-	kfree(kfd->interrupt_ring);
-}
-
-/*
- * This assumes that it can't be called concurrently with itself
- * but only with dequeue_ih_ring_entry.
- */
-bool enqueue_ih_ring_entry(struct kfd_dev *kfd,	const void *ih_ring_entry)
-{
-	unsigned int rptr = atomic_read(&kfd->interrupt_ring_rptr);
-	unsigned int wptr = atomic_read(&kfd->interrupt_ring_wptr);
-
-	if ((rptr - wptr) % kfd->interrupt_ring_size ==
-					kfd->device_info->ih_ring_entry_size) {
-		/* This is very bad, the system is likely to hang. */
-		dev_err_ratelimited(kfd_chardev(),
-			"Interrupt ring overflow, dropping interrupt.\n");
-		return false;
-	}
-
-	memcpy(kfd->interrupt_ring + wptr, ih_ring_entry,
-			kfd->device_info->ih_ring_entry_size);
-
-	wptr = (wptr + kfd->device_info->ih_ring_entry_size) %
-			kfd->interrupt_ring_size;
-	smp_wmb(); /* Ensure memcpy'd data is visible before wptr update. */
-	atomic_set(&kfd->interrupt_ring_wptr, wptr);
-
-	return true;
-}
-
-/*
- * This assumes that it can't be called concurrently with itself
- * but only with enqueue_ih_ring_entry.
- */
-static bool dequeue_ih_ring_entry(struct kfd_dev *kfd, void *ih_ring_entry)
-{
-	/*
-	 * Assume that wait queues have an implicit barrier, i.e. anything that
-	 * happened in the ISR before it queued work is visible.
-	 */
-
-	unsigned int wptr = atomic_read(&kfd->interrupt_ring_wptr);
-	unsigned int rptr = atomic_read(&kfd->interrupt_ring_rptr);
-
-	if (rptr == wptr)
-		return false;
-
-	memcpy(ih_ring_entry, kfd->interrupt_ring + rptr,
-			kfd->device_info->ih_ring_entry_size);
-
-	rptr = (rptr + kfd->device_info->ih_ring_entry_size) %
-			kfd->interrupt_ring_size;
-
-	/*
-	 * Ensure the rptr write update is not visible until
-	 * memcpy has finished reading.
-	 */
-	smp_mb();
-	atomic_set(&kfd->interrupt_ring_rptr, rptr);
-
-	return true;
-}
-
-static void interrupt_wq(struct work_struct *work)
-{
-	struct kfd_dev *dev = container_of(work, struct kfd_dev,
-						interrupt_work);
-
-	uint32_t ih_ring_entry[DIV_ROUND_UP(
-				dev->device_info->ih_ring_entry_size,
-				sizeof(uint32_t))];
-
-	while (dequeue_ih_ring_entry(dev, ih_ring_entry))
-		;
-}

drivers/gpu/drm/amd/amdkfd/kfd_priv.h

@@ -135,22 +135,10 @@ struct kfd_dev {
 
 	struct kgd2kfd_shared_resources shared_resources;
 
-	void *interrupt_ring;
-	size_t interrupt_ring_size;
-	atomic_t interrupt_ring_rptr;
-	atomic_t interrupt_ring_wptr;
-	struct work_struct interrupt_work;
-	spinlock_t interrupt_lock;
-
 	/* QCM Device instance */
 	struct device_queue_manager *dqm;
 
 	bool init_complete;
-	/*
-	 * Interrupts of interest to KFD are copied
-	 * from the HW ring into a SW ring.
-	 */
-	bool interrupts_active;
 };
 
 /* KGD2KFD callbacks */
@@ -531,10 +519,7 @@ struct kfd_dev *kfd_device_by_pci_dev(const struct pci_dev *pdev);
 struct kfd_dev *kfd_topology_enum_kfd_devices(uint8_t idx);
 
 /* Interrupts */
-int kfd_interrupt_init(struct kfd_dev *dev);
-void kfd_interrupt_exit(struct kfd_dev *dev);
 void kgd2kfd_interrupt(struct kfd_dev *kfd, const void *ih_ring_entry);
-bool enqueue_ih_ring_entry(struct kfd_dev *kfd,	const void *ih_ring_entry);
 
 /* Power Management */
 void kgd2kfd_suspend(struct kfd_dev *kfd);

drivers/gpu/drm/drm_fb_helper.c

@@ -741,7 +741,9 @@ int drm_fb_helper_setcmap(struct fb_cmap *cmap, struct fb_info *info)
 	int i, j, rc = 0;
 	int start;
 
-	drm_modeset_lock_all(dev);
+	if (__drm_modeset_lock_all(dev, !!oops_in_progress)) {
+		return -EBUSY;
+	}
 	if (!drm_fb_helper_is_bound(fb_helper)) {
 		drm_modeset_unlock_all(dev);
 		return -EBUSY;
@@ -915,7 +917,9 @@ int drm_fb_helper_pan_display(struct fb_var_screeninfo *var,
 	int ret = 0;
 	int i;
 
-	drm_modeset_lock_all(dev);
+	if (__drm_modeset_lock_all(dev, !!oops_in_progress)) {
+		return -EBUSY;
+	}
 	if (!drm_fb_helper_is_bound(fb_helper)) {
 		drm_modeset_unlock_all(dev);
 		return -EBUSY;

drivers/gpu/drm/exynos/exynos_drm_drv.c

@@ -645,18 +645,6 @@ static int exynos_drm_init(void)
 	if (!is_exynos)
 		return -ENODEV;
 
-	/*
-	 * Register device object only in case of Exynos SoC.
-	 *
-	 * Below codes resolves temporarily infinite loop issue incurred
-	 * by Exynos drm driver when using multi-platform kernel.
-	 * So these codes will be replaced with more generic way later.
-	 */
-	if (!of_machine_is_compatible("samsung,exynos3") &&
-			!of_machine_is_compatible("samsung,exynos4") &&
-			!of_machine_is_compatible("samsung,exynos5"))
-		return -ENODEV;
-
 	exynos_drm_pdev = platform_device_register_simple("exynos-drm", -1,
 								NULL, 0);
 	if (IS_ERR(exynos_drm_pdev))

drivers/gpu/drm/exynos/exynos_hdmi.c

@@ -1669,7 +1669,6 @@ static void hdmi_mode_apply(struct hdmi_context *hdata)
 
 static void hdmiphy_conf_reset(struct hdmi_context *hdata)
 {
-	u8 buffer[2];
 	u32 reg;
 
 	clk_disable_unprepare(hdata->res.sclk_hdmi);
@@ -1677,11 +1676,8 @@ static void hdmiphy_conf_reset(struct hdmi_context *hdata)
 	clk_prepare_enable(hdata->res.sclk_hdmi);
 
 	/* operation mode */
-	buffer[0] = 0x1f;
-	buffer[1] = 0x00;
-
-	if (hdata->hdmiphy_port)
-		i2c_master_send(hdata->hdmiphy_port, buffer, 2);
+	hdmiphy_reg_writeb(hdata, HDMIPHY_MODE_SET_DONE,
+				HDMI_PHY_ENABLE_MODE_SET);
 
 	if (hdata->type == HDMI_TYPE13)
 		reg = HDMI_V13_PHY_RSTOUT;

drivers/gpu/drm/exynos/exynos_mixer.c

@@ -1026,6 +1026,7 @@ static void mixer_win_disable(struct exynos_drm_manager *mgr, int zpos)
 static void mixer_wait_for_vblank(struct exynos_drm_manager *mgr)
 {
 	struct mixer_context *mixer_ctx = mgr_to_mixer(mgr);
+	int err;
 
 	mutex_lock(&mixer_ctx->mixer_mutex);
 	if (!mixer_ctx->powered) {
@@ -1034,7 +1035,11 @@ static void mixer_wait_for_vblank(struct exynos_drm_manager *mgr)
 	}
 	mutex_unlock(&mixer_ctx->mixer_mutex);
 
-	drm_vblank_get(mgr->crtc->dev, mixer_ctx->pipe);
+	err = drm_vblank_get(mgr->crtc->dev, mixer_ctx->pipe);
+	if (err < 0) {
+		DRM_DEBUG_KMS("failed to acquire vblank counter\n");
+		return;
+	}
 
 	atomic_set(&mixer_ctx->wait_vsync_event, 1);
 
@@ -1262,8 +1267,6 @@ static int mixer_bind(struct device *dev, struct device *manager, void *data)
 		return ret;
 	}
 
-	pm_runtime_enable(dev);
-
 	return 0;
 }
 
@@ -1272,8 +1275,6 @@ static void mixer_unbind(struct device *dev, struct device *master, void *data)
 	struct mixer_context *ctx = dev_get_drvdata(dev);
 
 	mixer_mgr_remove(&ctx->manager);
-
-	pm_runtime_disable(dev);
 }
 
 static const struct component_ops mixer_component_ops = {

drivers/gpu/drm/i915/i915_gem.c

@@ -5155,7 +5155,7 @@ static bool mutex_is_locked_by(struct mutex *mutex, struct task_struct *task)
 	if (!mutex_is_locked(mutex))
 		return false;
 
-#if defined(CONFIG_SMP) || defined(CONFIG_DEBUG_MUTEXES)
+#if defined(CONFIG_SMP) && !defined(CONFIG_DEBUG_MUTEXES)
 	return mutex->owner == task;
 #else
 	/* Since UP may be pre-empted, we cannot assume that we own the lock */

drivers/gpu/drm/i915/i915_irq.c

@@ -292,6 +292,23 @@ void gen6_enable_rps_interrupts(struct drm_device *dev)
 	spin_unlock_irq(&dev_priv->irq_lock);
 }
 
+u32 gen6_sanitize_rps_pm_mask(struct drm_i915_private *dev_priv, u32 mask)
+{
+	/*
+	 * SNB,IVB can while VLV,CHV may hard hang on looping batchbuffer
+	 * if GEN6_PM_UP_EI_EXPIRED is masked.
+	 *
+	 * TODO: verify if this can be reproduced on VLV,CHV.
+	 */
+	if (INTEL_INFO(dev_priv)->gen <= 7 && !IS_HASWELL(dev_priv))
+		mask &= ~GEN6_PM_RP_UP_EI_EXPIRED;
+
+	if (INTEL_INFO(dev_priv)->gen >= 8)
+		mask &= ~GEN8_PMINTR_REDIRECT_TO_NON_DISP;
+
+	return mask;
+}
+
 void gen6_disable_rps_interrupts(struct drm_device *dev)
 {
 	struct drm_i915_private *dev_priv = dev->dev_private;
@@ -304,8 +321,7 @@ void gen6_disable_rps_interrupts(struct drm_device *dev)
 
 	spin_lock_irq(&dev_priv->irq_lock);
 
-	I915_WRITE(GEN6_PMINTRMSK, INTEL_INFO(dev_priv)->gen >= 8 ?
-		   ~GEN8_PMINTR_REDIRECT_TO_NON_DISP : ~0);
+	I915_WRITE(GEN6_PMINTRMSK, gen6_sanitize_rps_pm_mask(dev_priv, ~0));
 
 	__gen6_disable_pm_irq(dev_priv, dev_priv->pm_rps_events);
 	I915_WRITE(gen6_pm_ier(dev_priv), I915_READ(gen6_pm_ier(dev_priv)) &

drivers/gpu/drm/i915/intel_display.c

@@ -9815,7 +9815,7 @@ static int intel_crtc_page_flip(struct drm_crtc *crtc,
 		if (obj->tiling_mode != work->old_fb_obj->tiling_mode)
 			/* vlv: DISPLAY_FLIP fails to change tiling */
 			ring = NULL;
-	} else if (IS_IVYBRIDGE(dev)) {
+	} else if (IS_IVYBRIDGE(dev) || IS_HASWELL(dev)) {
 		ring = &dev_priv->ring[BCS];
 	} else if (INTEL_INFO(dev)->gen >= 7) {
 		ring = obj->ring;

drivers/gpu/drm/i915/intel_drv.h

@@ -794,6 +794,7 @@ void gen6_disable_pm_irq(struct drm_i915_private *dev_priv, uint32_t mask);
 void gen6_reset_rps_interrupts(struct drm_device *dev);
 void gen6_enable_rps_interrupts(struct drm_device *dev);
 void gen6_disable_rps_interrupts(struct drm_device *dev);
+u32 gen6_sanitize_rps_pm_mask(struct drm_i915_private *dev_priv, u32 mask);
 void intel_runtime_pm_disable_interrupts(struct drm_i915_private *dev_priv);
 void intel_runtime_pm_enable_interrupts(struct drm_i915_private *dev_priv);
 static inline bool intel_irqs_enabled(struct drm_i915_private *dev_priv)

drivers/gpu/drm/i915/intel_pm.c

@@ -4363,16 +4363,7 @@ static u32 gen6_rps_pm_mask(struct drm_i915_private *dev_priv, u8 val)
 	mask |= dev_priv->pm_rps_events & (GEN6_PM_RP_DOWN_EI_EXPIRED | GEN6_PM_RP_UP_EI_EXPIRED);
 	mask &= dev_priv->pm_rps_events;
 
-	/* IVB and SNB hard hangs on looping batchbuffer
-	 * if GEN6_PM_UP_EI_EXPIRED is masked.
-	 */
-	if (INTEL_INFO(dev_priv->dev)->gen <= 7 && !IS_HASWELL(dev_priv->dev))
-		mask |= GEN6_PM_RP_UP_EI_EXPIRED;
-
-	if (IS_GEN8(dev_priv->dev))
-		mask |= GEN8_PMINTR_REDIRECT_TO_NON_DISP;
-
-	return ~mask;
+	return gen6_sanitize_rps_pm_mask(dev_priv, ~mask);
 }
 
 /* gen6_set_rps is called to update the frequency request, but should also be
@@ -4441,7 +4432,8 @@ static void vlv_set_rps_idle(struct drm_i915_private *dev_priv)
 		return;
 
 	/* Mask turbo interrupt so that they will not come in between */
-	I915_WRITE(GEN6_PMINTRMSK, 0xffffffff);
+	I915_WRITE(GEN6_PMINTRMSK,
+		   gen6_sanitize_rps_pm_mask(dev_priv, ~0));
 
 	vlv_force_gfx_clock(dev_priv, true);
 

drivers/gpu/drm/radeon/cik.c

@@ -6033,6 +6033,17 @@ void cik_vm_flush(struct radeon_device *rdev, struct radeon_ring *ring,
 	radeon_ring_write(ring, 0);
 	radeon_ring_write(ring, 1 << vm_id);
 
+	/* wait for the invalidate to complete */
+	radeon_ring_write(ring, PACKET3(PACKET3_WAIT_REG_MEM, 5));
+	radeon_ring_write(ring, (WAIT_REG_MEM_OPERATION(0) | /* wait */
+				 WAIT_REG_MEM_FUNCTION(0) |  /* always */
+				 WAIT_REG_MEM_ENGINE(0))); /* me */
+	radeon_ring_write(ring, VM_INVALIDATE_REQUEST >> 2);
+	radeon_ring_write(ring, 0);
+	radeon_ring_write(ring, 0); /* ref */
+	radeon_ring_write(ring, 0); /* mask */
+	radeon_ring_write(ring, 0x20); /* poll interval */
+
 	/* compute doesn't have PFP */
 	if (usepfp) {
 		/* sync PFP to ME, otherwise we might get invalid PFP reads */

drivers/gpu/drm/radeon/cik_sdma.c

@@ -903,6 +903,9 @@ void cik_sdma_vm_pad_ib(struct radeon_ib *ib)
 void cik_dma_vm_flush(struct radeon_device *rdev, struct radeon_ring *ring,
 		      unsigned vm_id, uint64_t pd_addr)
 {
+	u32 extra_bits = (SDMA_POLL_REG_MEM_EXTRA_OP(0) |
+			  SDMA_POLL_REG_MEM_EXTRA_FUNC(0)); /* always */
+
 	radeon_ring_write(ring, SDMA_PACKET(SDMA_OPCODE_SRBM_WRITE, 0, 0xf000));
 	if (vm_id < 8) {
 		radeon_ring_write(ring, (VM_CONTEXT0_PAGE_TABLE_BASE_ADDR + (vm_id << 2)) >> 2);
@@ -943,5 +946,12 @@ void cik_dma_vm_flush(struct radeon_device *rdev, struct radeon_ring *ring,
 	radeon_ring_write(ring, SDMA_PACKET(SDMA_OPCODE_SRBM_WRITE, 0, 0xf000));
 	radeon_ring_write(ring, VM_INVALIDATE_REQUEST >> 2);
 	radeon_ring_write(ring, 1 << vm_id);
+
+	radeon_ring_write(ring, SDMA_PACKET(SDMA_OPCODE_POLL_REG_MEM, 0, extra_bits));
+	radeon_ring_write(ring, VM_INVALIDATE_REQUEST >> 2);
+	radeon_ring_write(ring, 0);
+	radeon_ring_write(ring, 0); /* reference */
+	radeon_ring_write(ring, 0); /* mask */
+	radeon_ring_write(ring, (0xfff << 16) | 10); /* retry count, poll interval */
 }
 

drivers/gpu/drm/radeon/ni.c

@@ -2516,6 +2516,16 @@ void cayman_vm_flush(struct radeon_device *rdev, struct radeon_ring *ring,
 	radeon_ring_write(ring, PACKET0(VM_INVALIDATE_REQUEST, 0));
 	radeon_ring_write(ring, 1 << vm_id);
 
+	/* wait for the invalidate to complete */
+	radeon_ring_write(ring, PACKET3(PACKET3_WAIT_REG_MEM, 5));
+	radeon_ring_write(ring, (WAIT_REG_MEM_FUNCTION(0) |  /* always */
+				 WAIT_REG_MEM_ENGINE(0))); /* me */
+	radeon_ring_write(ring, VM_INVALIDATE_REQUEST >> 2);
+	radeon_ring_write(ring, 0);
+	radeon_ring_write(ring, 0); /* ref */
+	radeon_ring_write(ring, 0); /* mask */
+	radeon_ring_write(ring, 0x20); /* poll interval */
+
 	/* sync PFP to ME, otherwise we might get invalid PFP reads */
 	radeon_ring_write(ring, PACKET3(PACKET3_PFP_SYNC_ME, 0));
 	radeon_ring_write(ring, 0x0);

drivers/gpu/drm/radeon/ni_dma.c

@@ -463,5 +463,11 @@ void cayman_dma_vm_flush(struct radeon_device *rdev, struct radeon_ring *ring,
 	radeon_ring_write(ring, DMA_PACKET(DMA_PACKET_SRBM_WRITE, 0, 0, 0));
 	radeon_ring_write(ring, (0xf << 16) | (VM_INVALIDATE_REQUEST >> 2));
 	radeon_ring_write(ring, 1 << vm_id);
+
+	/* wait for invalidate to complete */
+	radeon_ring_write(ring, DMA_SRBM_READ_PACKET);
+	radeon_ring_write(ring, (0xff << 20) | (VM_INVALIDATE_REQUEST >> 2));
+	radeon_ring_write(ring, 0); /* mask */
+	radeon_ring_write(ring, 0); /* value */
 }
 

drivers/gpu/drm/radeon/nid.h

@@ -1133,6 +1133,23 @@
 #define	PACKET3_MEM_SEMAPHORE				0x39
 #define	PACKET3_MPEG_INDEX				0x3A
 #define	PACKET3_WAIT_REG_MEM				0x3C
+#define		WAIT_REG_MEM_FUNCTION(x)                ((x) << 0)
+                /* 0 - always
+		 * 1 - <
+		 * 2 - <=
+		 * 3 - ==
+		 * 4 - !=
+		 * 5 - >=
+		 * 6 - >
+		 */
+#define		WAIT_REG_MEM_MEM_SPACE(x)               ((x) << 4)
+                /* 0 - reg
+		 * 1 - mem
+		 */
+#define		WAIT_REG_MEM_ENGINE(x)                  ((x) << 8)
+                /* 0 - me
+		 * 1 - pfp
+		 */
 #define	PACKET3_MEM_WRITE				0x3D
 #define	PACKET3_PFP_SYNC_ME				0x42
 #define	PACKET3_SURFACE_SYNC				0x43
@@ -1272,6 +1289,13 @@
 					 (1 << 21) |			\
 					 (((n) & 0xFFFFF) << 0))
 
+#define DMA_SRBM_POLL_PACKET		((9 << 28) |			\
+					 (1 << 27) |			\
+					 (1 << 26))
+
+#define DMA_SRBM_READ_PACKET		((9 << 28) |			\
+					 (1 << 27))
+
 /* async DMA Packet types */
 #define	DMA_PACKET_WRITE				  0x2
 #define	DMA_PACKET_COPY					  0x3

drivers/gpu/drm/radeon/radeon_asic.c

@@ -333,6 +333,20 @@ static struct radeon_asic_ring r300_gfx_ring = {
 	.set_wptr = &r100_gfx_set_wptr,
 };
 
+static struct radeon_asic_ring rv515_gfx_ring = {
+	.ib_execute = &r100_ring_ib_execute,
+	.emit_fence = &r300_fence_ring_emit,
+	.emit_semaphore = &r100_semaphore_ring_emit,
+	.cs_parse = &r300_cs_parse,
+	.ring_start = &rv515_ring_start,
+	.ring_test = &r100_ring_test,
+	.ib_test = &r100_ib_test,
+	.is_lockup = &r100_gpu_is_lockup,
+	.get_rptr = &r100_gfx_get_rptr,
+	.get_wptr = &r100_gfx_get_wptr,
+	.set_wptr = &r100_gfx_set_wptr,
+};
+
 static struct radeon_asic r300_asic = {
 	.init = &r300_init,
 	.fini = &r300_fini,
@@ -748,7 +762,7 @@ static struct radeon_asic rv515_asic = {
 		.set_page = &rv370_pcie_gart_set_page,
 	},
 	.ring = {
-		[RADEON_RING_TYPE_GFX_INDEX] = &r300_gfx_ring
+		[RADEON_RING_TYPE_GFX_INDEX] = &rv515_gfx_ring
 	},
 	.irq = {
 		.set = &rs600_irq_set,
@@ -814,7 +828,7 @@ static struct radeon_asic r520_asic = {
 		.set_page = &rv370_pcie_gart_set_page,
 	},
 	.ring = {
-		[RADEON_RING_TYPE_GFX_INDEX] = &r300_gfx_ring
+		[RADEON_RING_TYPE_GFX_INDEX] = &rv515_gfx_ring
 	},
 	.irq = {
 		.set = &rs600_irq_set,

drivers/gpu/drm/radeon/radeon_gem.c

@@ -576,7 +576,7 @@ static void radeon_gem_va_update_vm(struct radeon_device *rdev,
 error_free:
 	drm_free_large(vm_bos);
 
-	if (r)
+	if (r && r != -ERESTARTSYS)
 		DRM_ERROR("Couldn't update BO_VA (%d)\n", r);
 }
 

drivers/gpu/drm/radeon/radeon_pm.c

@@ -1287,8 +1287,39 @@ static int radeon_pm_init_dpm(struct radeon_device *rdev)
 	return ret;
 }
 
+struct radeon_dpm_quirk {
+	u32 chip_vendor;
+	u32 chip_device;
+	u32 subsys_vendor;
+	u32 subsys_device;
+};
+
+/* cards with dpm stability problems */
+static struct radeon_dpm_quirk radeon_dpm_quirk_list[] = {
+	/* TURKS - https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1386534 */
+	{ PCI_VENDOR_ID_ATI, 0x6759, 0x1682, 0x3195 },
+	/* TURKS - https://bugzilla.kernel.org/show_bug.cgi?id=83731 */
+	{ PCI_VENDOR_ID_ATI, 0x6840, 0x1179, 0xfb81 },
+	{ 0, 0, 0, 0 },
+};
+
 int radeon_pm_init(struct radeon_device *rdev)
 {
+	struct radeon_dpm_quirk *p = radeon_dpm_quirk_list;
+	bool disable_dpm = false;
+
+	/* Apply dpm quirks */
+	while (p && p->chip_device != 0) {
+		if (rdev->pdev->vendor == p->chip_vendor &&
+		    rdev->pdev->device == p->chip_device &&
+		    rdev->pdev->subsystem_vendor == p->subsys_vendor &&
+		    rdev->pdev->subsystem_device == p->subsys_device) {
+			disable_dpm = true;
+			break;
+		}
+		++p;
+	}
+
 	/* enable dpm on rv6xx+ */
 	switch (rdev->family) {
 	case CHIP_RV610:
@@ -1344,6 +1375,8 @@ int radeon_pm_init(struct radeon_device *rdev)
 			 (!(rdev->flags & RADEON_IS_IGP)) &&
 			 (!rdev->smc_fw))
 			rdev->pm.pm_method = PM_METHOD_PROFILE;
+		else if (disable_dpm && (radeon_dpm == -1))
+			rdev->pm.pm_method = PM_METHOD_PROFILE;
 		else if (radeon_dpm == 0)
 			rdev->pm.pm_method = PM_METHOD_PROFILE;
 		else

drivers/gpu/drm/radeon/si.c

@@ -5057,6 +5057,16 @@ void si_vm_flush(struct radeon_device *rdev, struct radeon_ring *ring,
 	radeon_ring_write(ring, 0);
 	radeon_ring_write(ring, 1 << vm_id);
 
+	/* wait for the invalidate to complete */
+	radeon_ring_write(ring, PACKET3(PACKET3_WAIT_REG_MEM, 5));
+	radeon_ring_write(ring, (WAIT_REG_MEM_FUNCTION(0) |  /* always */
+				 WAIT_REG_MEM_ENGINE(0))); /* me */
+	radeon_ring_write(ring, VM_INVALIDATE_REQUEST >> 2);
+	radeon_ring_write(ring, 0);
+	radeon_ring_write(ring, 0); /* ref */
+	radeon_ring_write(ring, 0); /* mask */
+	radeon_ring_write(ring, 0x20); /* poll interval */
+
 	/* sync PFP to ME, otherwise we might get invalid PFP reads */
 	radeon_ring_write(ring, PACKET3(PACKET3_PFP_SYNC_ME, 0));
 	radeon_ring_write(ring, 0x0);

drivers/gpu/drm/radeon/si_dma.c

@@ -206,6 +206,14 @@ void si_dma_vm_flush(struct radeon_device *rdev, struct radeon_ring *ring,
 	radeon_ring_write(ring, DMA_PACKET(DMA_PACKET_SRBM_WRITE, 0, 0, 0, 0));
 	radeon_ring_write(ring, (0xf << 16) | (VM_INVALIDATE_REQUEST >> 2));
 	radeon_ring_write(ring, 1 << vm_id);
+
+	/* wait for invalidate to complete */
+	radeon_ring_write(ring, DMA_PACKET(DMA_PACKET_POLL_REG_MEM, 0, 0, 0, 0));
+	radeon_ring_write(ring, VM_INVALIDATE_REQUEST);
+	radeon_ring_write(ring, 0xff << 16); /* retry */
+	radeon_ring_write(ring, 1 << vm_id); /* mask */
+	radeon_ring_write(ring, 0); /* value */
+	radeon_ring_write(ring, (0 << 28) | 0x20); /* func(always) | poll interval */
 }
 
 /**

drivers/gpu/drm/radeon/si_dpm.c

@@ -2908,6 +2908,22 @@ static int si_init_smc_spll_table(struct radeon_device *rdev)
 	return ret;
 }
 
+struct si_dpm_quirk {
+	u32 chip_vendor;
+	u32 chip_device;
+	u32 subsys_vendor;
+	u32 subsys_device;
+	u32 max_sclk;
+	u32 max_mclk;
+};
+
+/* cards with dpm stability problems */
+static struct si_dpm_quirk si_dpm_quirk_list[] = {
+	/* PITCAIRN - https://bugs.freedesktop.org/show_bug.cgi?id=76490 */
+	{ PCI_VENDOR_ID_ATI, 0x6810, 0x1462, 0x3036, 0, 120000 },
+	{ 0, 0, 0, 0 },
+};
+
 static void si_apply_state_adjust_rules(struct radeon_device *rdev,
 					struct radeon_ps *rps)
 {
@@ -2918,7 +2934,22 @@ static void si_apply_state_adjust_rules(struct radeon_device *rdev,
 	u32 mclk, sclk;
 	u16 vddc, vddci;
 	u32 max_sclk_vddc, max_mclk_vddci, max_mclk_vddc;
+	u32 max_sclk = 0, max_mclk = 0;
 	int i;
+	struct si_dpm_quirk *p = si_dpm_quirk_list;
+
+	/* Apply dpm quirks */
+	while (p && p->chip_device != 0) {
+		if (rdev->pdev->vendor == p->chip_vendor &&
+		    rdev->pdev->device == p->chip_device &&
+		    rdev->pdev->subsystem_vendor == p->subsys_vendor &&
+		    rdev->pdev->subsystem_device == p->subsys_device) {
+			max_sclk = p->max_sclk;
+			max_mclk = p->max_mclk;
+			break;
+		}
+		++p;
+	}
 
 	if ((rdev->pm.dpm.new_active_crtc_count > 1) ||
 	    ni_dpm_vblank_too_short(rdev))
@@ -2972,6 +3003,14 @@ static void si_apply_state_adjust_rules(struct radeon_device *rdev,
 			if (ps->performance_levels[i].mclk > max_mclk_vddc)
 				ps->performance_levels[i].mclk = max_mclk_vddc;
 		}
+		if (max_mclk) {
+			if (ps->performance_levels[i].mclk > max_mclk)
+				ps->performance_levels[i].mclk = max_mclk;
+		}
+		if (max_sclk) {
+			if (ps->performance_levels[i].sclk > max_sclk)
+				ps->performance_levels[i].sclk = max_sclk;
+		}
 	}
 
 	/* XXX validate the min clocks required for display */

drivers/gpu/drm/radeon/sid.h

@@ -1632,6 +1632,23 @@
 #define	PACKET3_MPEG_INDEX				0x3A
 #define	PACKET3_COPY_DW					0x3B
 #define	PACKET3_WAIT_REG_MEM				0x3C
+#define		WAIT_REG_MEM_FUNCTION(x)                ((x) << 0)
+                /* 0 - always
+		 * 1 - <
+		 * 2 - <=
+		 * 3 - ==
+		 * 4 - !=
+		 * 5 - >=
+		 * 6 - >
+		 */
+#define		WAIT_REG_MEM_MEM_SPACE(x)               ((x) << 4)
+                /* 0 - reg
+		 * 1 - mem
+		 */
+#define		WAIT_REG_MEM_ENGINE(x)                  ((x) << 8)
+                /* 0 - me
+		 * 1 - pfp
+		 */
 #define	PACKET3_MEM_WRITE				0x3D
 #define	PACKET3_COPY_DATA				0x40
 #define	PACKET3_CP_DMA					0x41
@@ -1835,6 +1852,7 @@
 #define	DMA_PACKET_TRAP					  0x7
 #define	DMA_PACKET_SRBM_WRITE				  0x9
 #define	DMA_PACKET_CONSTANT_FILL			  0xd
+#define	DMA_PACKET_POLL_REG_MEM				  0xe
 #define	DMA_PACKET_NOP					  0xf
 
 #define VCE_STATUS					0x20004